1 The text below describes the locking rules for VFS-related methods.
2 It is (believed to be) up-to-date. *Please*, if you change anything in
3 prototypes or locking protocols - update this file. And update the relevant
4 instances in the tree, don't leave that to maintainers of filesystems/devices/
5 etc. At the very least, put the list of dubious cases in the end of this file.
6 Don't turn it into log - maintainers of out-of-the-tree code are supposed to
7 be able to use diff(1).
8 Thing currently missing here: socket operations. Alexey?
10 --------------------------- dentry_operations --------------------------
12 int (*d_revalidate)(struct dentry *, unsigned int);
13 int (*d_weak_revalidate)(struct dentry *, unsigned int);
14 int (*d_hash)(const struct dentry *, struct qstr *);
15 int (*d_compare)(const struct dentry *, const struct dentry *,
16 unsigned int, const char *, const struct qstr *);
17 int (*d_delete)(struct dentry *);
18 void (*d_release)(struct dentry *);
19 void (*d_iput)(struct dentry *, struct inode *);
20 char *(*d_dname)((struct dentry *dentry, char *buffer, int buflen);
21 struct vfsmount *(*d_automount)(struct path *path);
22 int (*d_manage)(struct dentry *, bool);
23 struct dentry *(*d_real)(struct dentry *, const struct inode *,
27 rename_lock ->d_lock may block rcu-walk
28 d_revalidate: no no yes (ref-walk) maybe
29 d_weak_revalidate:no no yes no
31 d_compare: yes no no maybe
32 d_delete: no yes no no
33 d_release: no no yes no
37 d_automount: no no yes no
38 d_manage: no no yes (ref-walk) maybe
41 --------------------------- inode_operations ---------------------------
43 int (*create) (struct inode *,struct dentry *,umode_t, bool);
44 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
45 int (*link) (struct dentry *,struct inode *,struct dentry *);
46 int (*unlink) (struct inode *,struct dentry *);
47 int (*symlink) (struct inode *,struct dentry *,const char *);
48 int (*mkdir) (struct inode *,struct dentry *,umode_t);
49 int (*rmdir) (struct inode *,struct dentry *);
50 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
51 int (*rename) (struct inode *, struct dentry *,
52 struct inode *, struct dentry *);
53 int (*rename2) (struct inode *, struct dentry *,
54 struct inode *, struct dentry *, unsigned int);
55 int (*readlink) (struct dentry *, char __user *,int);
56 const char *(*get_link) (struct dentry *, struct inode *, void **);
57 void (*truncate) (struct inode *);
58 int (*permission) (struct inode *, int, unsigned int);
59 int (*get_acl)(struct inode *, int);
60 int (*setattr) (struct dentry *, struct iattr *);
61 int (*getattr) (struct vfsmount *, struct dentry *, struct kstat *);
62 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
63 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
64 ssize_t (*listxattr) (struct dentry *, char *, size_t);
65 int (*removexattr) (struct dentry *, const char *);
66 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start, u64 len);
67 void (*update_time)(struct inode *, struct timespec *, int);
68 int (*atomic_open)(struct inode *, struct dentry *,
69 struct file *, unsigned open_flag,
70 umode_t create_mode, int *opened);
71 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
83 rmdir: yes (both) (see below)
84 rename: yes (all) (see below)
85 rename2: yes (all) (see below)
89 permission: no (may not block if called in rcu-walk mode)
101 Additionally, ->rmdir(), ->unlink() and ->rename() have ->i_mutex on
103 cross-directory ->rename() and rename2() has (per-superblock)
106 See Documentation/filesystems/directory-locking for more detailed discussion
107 of the locking scheme for directory operations.
109 --------------------------- super_operations ---------------------------
111 struct inode *(*alloc_inode)(struct super_block *sb);
112 void (*destroy_inode)(struct inode *);
113 void (*dirty_inode) (struct inode *, int flags);
114 int (*write_inode) (struct inode *, struct writeback_control *wbc);
115 int (*drop_inode) (struct inode *);
116 void (*evict_inode) (struct inode *);
117 void (*put_super) (struct super_block *);
118 int (*sync_fs)(struct super_block *sb, int wait);
119 int (*freeze_fs) (struct super_block *);
120 int (*unfreeze_fs) (struct super_block *);
121 int (*statfs) (struct dentry *, struct kstatfs *);
122 int (*remount_fs) (struct super_block *, int *, char *);
123 void (*umount_begin) (struct super_block *);
124 int (*show_options)(struct seq_file *, struct dentry *);
125 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
126 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
127 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
130 All may block [not true, see below]
136 drop_inode: !!!inode->i_lock!!!
142 statfs: maybe(read) (see below)
145 show_options: no (namespace_sem)
146 quota_read: no (see below)
147 quota_write: no (see below)
148 bdev_try_to_free_page: no (see below)
150 ->statfs() has s_umount (shared) when called by ustat(2) (native or
151 compat), but that's an accident of bad API; s_umount is used to pin
152 the superblock down when we only have dev_t given us by userland to
153 identify the superblock. Everything else (statfs(), fstatfs(), etc.)
154 doesn't hold it when calling ->statfs() - superblock is pinned down
155 by resolving the pathname passed to syscall.
156 ->quota_read() and ->quota_write() functions are both guaranteed to
157 be the only ones operating on the quota file by the quota code (via
158 dqio_sem) (unless an admin really wants to screw up something and
159 writes to quota files with quotas on). For other details about locking
160 see also dquot_operations section.
161 ->bdev_try_to_free_page is called from the ->releasepage handler of
162 the block device inode. See there for more details.
164 --------------------------- file_system_type ---------------------------
166 struct dentry *(*mount) (struct file_system_type *, int,
167 const char *, void *);
168 void (*kill_sb) (struct super_block *);
174 ->mount() returns ERR_PTR or the root dentry; its superblock should be locked
176 ->kill_sb() takes a write-locked superblock, does all shutdown work on it,
177 unlocks and drops the reference.
179 --------------------------- address_space_operations --------------------------
181 int (*writepage)(struct page *page, struct writeback_control *wbc);
182 int (*readpage)(struct file *, struct page *);
183 int (*writepages)(struct address_space *, struct writeback_control *);
184 int (*set_page_dirty)(struct page *page);
185 int (*readpages)(struct file *filp, struct address_space *mapping,
186 struct list_head *pages, unsigned nr_pages);
187 int (*write_begin)(struct file *, struct address_space *mapping,
188 loff_t pos, unsigned len, unsigned flags,
189 struct page **pagep, void **fsdata);
190 int (*write_end)(struct file *, struct address_space *mapping,
191 loff_t pos, unsigned len, unsigned copied,
192 struct page *page, void *fsdata);
193 sector_t (*bmap)(struct address_space *, sector_t);
194 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
195 int (*releasepage) (struct page *, int);
196 void (*freepage)(struct page *);
197 int (*direct_IO)(struct kiocb *, struct iov_iter *iter);
198 int (*migratepage)(struct address_space *, struct page *, struct page *);
199 int (*launder_page)(struct page *);
200 int (*is_partially_uptodate)(struct page *, unsigned long, unsigned long);
201 int (*error_remove_page)(struct address_space *, struct page *);
202 int (*swap_activate)(struct file *);
203 int (*swap_deactivate)(struct file *);
206 All except set_page_dirty and freepage may block
208 PageLocked(page) i_mutex
209 writepage: yes, unlocks (see below)
210 readpage: yes, unlocks
214 write_begin: locks the page yes
215 write_end: yes, unlocks yes
221 migratepage: yes (both)
223 is_partially_uptodate: yes
224 error_remove_page: yes
228 ->write_begin(), ->write_end() and ->readpage() may be called from
229 the request handler (/dev/loop).
231 ->readpage() unlocks the page, either synchronously or via I/O
234 ->readpages() populates the pagecache with the passed pages and starts
235 I/O against them. They come unlocked upon I/O completion.
237 ->writepage() is used for two purposes: for "memory cleansing" and for
238 "sync". These are quite different operations and the behaviour may differ
239 depending upon the mode.
241 If writepage is called for sync (wbc->sync_mode != WBC_SYNC_NONE) then
242 it *must* start I/O against the page, even if that would involve
243 blocking on in-progress I/O.
245 If writepage is called for memory cleansing (sync_mode ==
246 WBC_SYNC_NONE) then its role is to get as much writeout underway as
247 possible. So writepage should try to avoid blocking against
248 currently-in-progress I/O.
250 If the filesystem is not called for "sync" and it determines that it
251 would need to block against in-progress I/O to be able to start new I/O
252 against the page the filesystem should redirty the page with
253 redirty_page_for_writepage(), then unlock the page and return zero.
254 This may also be done to avoid internal deadlocks, but rarely.
256 If the filesystem is called for sync then it must wait on any
257 in-progress I/O and then start new I/O.
259 The filesystem should unlock the page synchronously, before returning to the
260 caller, unless ->writepage() returns special WRITEPAGE_ACTIVATE
261 value. WRITEPAGE_ACTIVATE means that page cannot really be written out
262 currently, and VM should stop calling ->writepage() on this page for some
263 time. VM does this by moving page to the head of the active list, hence the
266 Unless the filesystem is going to redirty_page_for_writepage(), unlock the page
267 and return zero, writepage *must* run set_page_writeback() against the page,
268 followed by unlocking it. Once set_page_writeback() has been run against the
269 page, write I/O can be submitted and the write I/O completion handler must run
270 end_page_writeback() once the I/O is complete. If no I/O is submitted, the
271 filesystem must run end_page_writeback() against the page before returning from
274 That is: after 2.5.12, pages which are under writeout are *not* locked. Note,
275 if the filesystem needs the page to be locked during writeout, that is ok, too,
276 the page is allowed to be unlocked at any point in time between the calls to
277 set_page_writeback() and end_page_writeback().
279 Note, failure to run either redirty_page_for_writepage() or the combination of
280 set_page_writeback()/end_page_writeback() on a page submitted to writepage
281 will leave the page itself marked clean but it will be tagged as dirty in the
282 radix tree. This incoherency can lead to all sorts of hard-to-debug problems
283 in the filesystem like having dirty inodes at umount and losing written data.
285 ->writepages() is used for periodic writeback and for syscall-initiated
286 sync operations. The address_space should start I/O against at least
287 *nr_to_write pages. *nr_to_write must be decremented for each page which is
288 written. The address_space implementation may write more (or less) pages
289 than *nr_to_write asks for, but it should try to be reasonably close. If
290 nr_to_write is NULL, all dirty pages must be written.
292 writepages should _only_ write pages which are present on
295 ->set_page_dirty() is called from various places in the kernel
296 when the target page is marked as needing writeback. It may be called
297 under spinlock (it cannot block) and is sometimes called with the page
300 ->bmap() is currently used by legacy ioctl() (FIBMAP) provided by some
301 filesystems and by the swapper. The latter will eventually go away. Please,
302 keep it that way and don't breed new callers.
304 ->invalidatepage() is called when the filesystem must attempt to drop
305 some or all of the buffers from the page when it is being truncated. It
306 returns zero on success. If ->invalidatepage is zero, the kernel uses
307 block_invalidatepage() instead.
309 ->releasepage() is called when the kernel is about to try to drop the
310 buffers from the page in preparation for freeing it. It returns zero to
311 indicate that the buffers are (or may be) freeable. If ->releasepage is zero,
312 the kernel assumes that the fs has no private interest in the buffers.
314 ->freepage() is called when the kernel is done dropping the page
317 ->launder_page() may be called prior to releasing a page if
318 it is still found to be dirty. It returns zero if the page was successfully
319 cleaned, or an error value if not. Note that in order to prevent the page
320 getting mapped back in and redirtied, it needs to be kept locked
321 across the entire operation.
323 ->swap_activate will be called with a non-zero argument on
324 files backing (non block device backed) swapfiles. A return value
325 of zero indicates success, in which case this file can be used for
326 backing swapspace. The swapspace operations will be proxied to the
327 address space operations.
329 ->swap_deactivate() will be called in the sys_swapoff()
330 path after ->swap_activate() returned success.
332 ----------------------- file_lock_operations ------------------------------
334 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
335 void (*fl_release_private)(struct file_lock *);
339 inode->i_lock may block
341 fl_release_private: maybe maybe[1]
343 [1]: ->fl_release_private for flock or POSIX locks is currently allowed
344 to block. Leases however can still be freed while the i_lock is held and
345 so fl_release_private called on a lease should not block.
347 ----------------------- lock_manager_operations ---------------------------
349 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
350 unsigned long (*lm_owner_key)(struct file_lock *);
351 void (*lm_notify)(struct file_lock *); /* unblock callback */
352 int (*lm_grant)(struct file_lock *, struct file_lock *, int);
353 void (*lm_break)(struct file_lock *); /* break_lease callback */
354 int (*lm_change)(struct file_lock **, int);
358 inode->i_lock blocked_lock_lock may block
359 lm_compare_owner: yes[1] maybe no
360 lm_owner_key yes[1] yes no
361 lm_notify: yes yes no
366 [1]: ->lm_compare_owner and ->lm_owner_key are generally called with
367 *an* inode->i_lock held. It may not be the i_lock of the inode
368 associated with either file_lock argument! This is the case with deadlock
369 detection, since the code has to chase down the owners of locks that may
370 be entirely unrelated to the one on which the lock is being acquired.
371 For deadlock detection however, the blocked_lock_lock is also held. The
372 fact that these locks are held ensures that the file_locks do not
373 disappear out from under you while doing the comparison or generating an
376 --------------------------- buffer_head -----------------------------------
378 void (*b_end_io)(struct buffer_head *bh, int uptodate);
381 called from interrupts. In other words, extreme care is needed here.
382 bh is locked, but that's all warranties we have here. Currently only RAID1,
383 highmem, fs/buffer.c, and fs/ntfs/aops.c are providing these. Block devices
384 call this method upon the IO completion.
386 --------------------------- block_device_operations -----------------------
388 int (*open) (struct block_device *, fmode_t);
389 int (*release) (struct gendisk *, fmode_t);
390 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
391 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
392 int (*direct_access) (struct block_device *, sector_t, void __pmem **,
394 int (*media_changed) (struct gendisk *);
395 void (*unlock_native_capacity) (struct gendisk *);
396 int (*revalidate_disk) (struct gendisk *);
397 int (*getgeo)(struct block_device *, struct hd_geometry *);
398 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
408 unlock_native_capacity: no
411 swap_slot_free_notify: no (see below)
413 media_changed, unlock_native_capacity and revalidate_disk are called only from
416 swap_slot_free_notify is called with swap_lock and sometimes the page lock
420 --------------------------- file_operations -------------------------------
422 loff_t (*llseek) (struct file *, loff_t, int);
423 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
424 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
425 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
426 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
427 int (*iterate) (struct file *, struct dir_context *);
428 unsigned int (*poll) (struct file *, struct poll_table_struct *);
429 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
430 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
431 int (*mmap) (struct file *, struct vm_area_struct *);
432 int (*open) (struct inode *, struct file *);
433 int (*flush) (struct file *);
434 int (*release) (struct inode *, struct file *);
435 int (*fsync) (struct file *, loff_t start, loff_t end, int datasync);
436 int (*aio_fsync) (struct kiocb *, int datasync);
437 int (*fasync) (int, struct file *, int);
438 int (*lock) (struct file *, int, struct file_lock *);
439 ssize_t (*readv) (struct file *, const struct iovec *, unsigned long,
441 ssize_t (*writev) (struct file *, const struct iovec *, unsigned long,
443 ssize_t (*sendfile) (struct file *, loff_t *, size_t, read_actor_t,
445 ssize_t (*sendpage) (struct file *, struct page *, int, size_t,
447 unsigned long (*get_unmapped_area)(struct file *, unsigned long,
448 unsigned long, unsigned long, unsigned long);
449 int (*check_flags)(int);
450 int (*flock) (struct file *, int, struct file_lock *);
451 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *,
452 size_t, unsigned int);
453 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *,
454 size_t, unsigned int);
455 int (*setlease)(struct file *, long, struct file_lock **, void **);
456 long (*fallocate)(struct file *, int, loff_t, loff_t);
462 ->llseek() locking has moved from llseek to the individual llseek
463 implementations. If your fs is not using generic_file_llseek, you
464 need to acquire and release the appropriate locks in your ->llseek().
465 For many filesystems, it is probably safe to acquire the inode
466 mutex or just to use i_size_read() instead.
467 Note: this does not protect the file->f_pos against concurrent modifications
468 since this is something the userspace has to take care about.
470 ->fasync() is responsible for maintaining the FASYNC bit in filp->f_flags.
471 Most instances call fasync_helper(), which does that maintenance, so it's
472 not normally something one needs to worry about. Return values > 0 will be
473 mapped to zero in the VFS layer.
475 ->readdir() and ->ioctl() on directories must be changed. Ideally we would
476 move ->readdir() to inode_operations and use a separate method for directory
477 ->ioctl() or kill the latter completely. One of the problems is that for
478 anything that resembles union-mount we won't have a struct file for all
479 components. And there are other reasons why the current interface is a mess...
481 ->read on directories probably must go away - we should just enforce -EISDIR
482 in sys_read() and friends.
484 ->setlease operations should call generic_setlease() before or after setting
485 the lease within the individual filesystem to record the result of the
488 --------------------------- dquot_operations -------------------------------
490 int (*write_dquot) (struct dquot *);
491 int (*acquire_dquot) (struct dquot *);
492 int (*release_dquot) (struct dquot *);
493 int (*mark_dirty) (struct dquot *);
494 int (*write_info) (struct super_block *, int);
496 These operations are intended to be more or less wrapping functions that ensure
497 a proper locking wrt the filesystem and call the generic quota operations.
499 What filesystem should expect from the generic quota functions:
501 FS recursion Held locks when called
502 write_dquot: yes dqonoff_sem or dqptr_sem
503 acquire_dquot: yes dqonoff_sem or dqptr_sem
504 release_dquot: yes dqonoff_sem or dqptr_sem
506 write_info: yes dqonoff_sem
508 FS recursion means calling ->quota_read() and ->quota_write() from superblock
511 More details about quota locking can be found in fs/dquot.c.
513 --------------------------- vm_operations_struct -----------------------------
515 void (*open)(struct vm_area_struct*);
516 void (*close)(struct vm_area_struct*);
517 int (*fault)(struct vm_area_struct*, struct vm_fault *);
518 int (*page_mkwrite)(struct vm_area_struct *, struct vm_fault *);
519 int (*pfn_mkwrite)(struct vm_area_struct *, struct vm_fault *);
520 int (*access)(struct vm_area_struct *, unsigned long, void*, int, int);
523 mmap_sem PageLocked(page)
526 fault: yes can return with page locked
528 page_mkwrite: yes can return with page locked
532 ->fault() is called when a previously not present pte is about
533 to be faulted in. The filesystem must find and return the page associated
534 with the passed in "pgoff" in the vm_fault structure. If it is possible that
535 the page may be truncated and/or invalidated, then the filesystem must lock
536 the page, then ensure it is not already truncated (the page lock will block
537 subsequent truncate), and then return with VM_FAULT_LOCKED, and the page
538 locked. The VM will unlock the page.
540 ->map_pages() is called when VM asks to map easy accessible pages.
541 Filesystem should find and map pages associated with offsets from "pgoff"
542 till "max_pgoff". ->map_pages() is called with page table locked and must
543 not block. If it's not possible to reach a page without blocking,
544 filesystem should skip it. Filesystem should use do_set_pte() to setup
545 page table entry. Pointer to entry associated with offset "pgoff" is
546 passed in "pte" field in vm_fault structure. Pointers to entries for other
547 offsets should be calculated relative to "pte".
549 ->page_mkwrite() is called when a previously read-only pte is
550 about to become writeable. The filesystem again must ensure that there are
551 no truncate/invalidate races, and then return with the page locked. If
552 the page has been truncated, the filesystem should not look up a new page
553 like the ->fault() handler, but simply return with VM_FAULT_NOPAGE, which
554 will cause the VM to retry the fault.
556 ->pfn_mkwrite() is the same as page_mkwrite but when the pte is
557 VM_PFNMAP or VM_MIXEDMAP with a page-less entry. Expected return is
558 VM_FAULT_NOPAGE. Or one of the VM_FAULT_ERROR types. The default behavior
559 after this call is to make the pte read-write, unless pfn_mkwrite returns
562 ->access() is called when get_user_pages() fails in
563 access_process_vm(), typically used to debug a process through
564 /proc/pid/mem or ptrace. This function is needed only for
565 VM_IO | VM_PFNMAP VMAs.
567 ================================================================================
570 (if you break something or notice that it is broken and do not fix it yourself
571 - at least put it here)