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);
25 rename_lock ->d_lock may block rcu-walk
26 d_revalidate: no no yes (ref-walk) maybe
27 d_weak_revalidate:no no yes no
29 d_compare: yes no no maybe
30 d_delete: no yes no no
31 d_release: no no yes no
35 d_automount: no no yes no
36 d_manage: no no yes (ref-walk) maybe
38 --------------------------- inode_operations ---------------------------
40 int (*create) (struct inode *,struct dentry *,umode_t, bool);
41 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
42 int (*link) (struct dentry *,struct inode *,struct dentry *);
43 int (*unlink) (struct inode *,struct dentry *);
44 int (*symlink) (struct inode *,struct dentry *,const char *);
45 int (*mkdir) (struct inode *,struct dentry *,umode_t);
46 int (*rmdir) (struct inode *,struct dentry *);
47 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
48 int (*rename) (struct inode *, struct dentry *,
49 struct inode *, struct dentry *);
50 int (*rename2) (struct inode *, struct dentry *,
51 struct inode *, struct dentry *, unsigned int);
52 int (*readlink) (struct dentry *, char __user *,int);
53 const char *(*get_link) (struct dentry *, struct inode *, void **);
54 void (*truncate) (struct inode *);
55 int (*permission) (struct inode *, int, unsigned int);
56 int (*get_acl)(struct inode *, int);
57 int (*setattr) (struct dentry *, struct iattr *);
58 int (*getattr) (struct vfsmount *, struct dentry *, struct kstat *);
59 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
60 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
61 ssize_t (*listxattr) (struct dentry *, char *, size_t);
62 int (*removexattr) (struct dentry *, const char *);
63 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start, u64 len);
64 void (*update_time)(struct inode *, struct timespec *, int);
65 int (*atomic_open)(struct inode *, struct dentry *,
66 struct file *, unsigned open_flag,
67 umode_t create_mode, int *opened);
68 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
69 int (*dentry_open)(struct dentry *, struct file *, const struct cred *);
81 rmdir: yes (both) (see below)
82 rename: yes (all) (see below)
83 rename2: yes (all) (see below)
87 permission: no (may not block if called in rcu-walk mode)
100 Additionally, ->rmdir(), ->unlink() and ->rename() have ->i_mutex on
102 cross-directory ->rename() and rename2() has (per-superblock)
105 See Documentation/filesystems/directory-locking for more detailed discussion
106 of the locking scheme for directory operations.
108 --------------------------- super_operations ---------------------------
110 struct inode *(*alloc_inode)(struct super_block *sb);
111 void (*destroy_inode)(struct inode *);
112 void (*dirty_inode) (struct inode *, int flags);
113 int (*write_inode) (struct inode *, struct writeback_control *wbc);
114 int (*drop_inode) (struct inode *);
115 void (*evict_inode) (struct inode *);
116 void (*put_super) (struct super_block *);
117 int (*sync_fs)(struct super_block *sb, int wait);
118 int (*freeze_fs) (struct super_block *);
119 int (*unfreeze_fs) (struct super_block *);
120 int (*statfs) (struct dentry *, struct kstatfs *);
121 int (*remount_fs) (struct super_block *, int *, char *);
122 void (*umount_begin) (struct super_block *);
123 int (*show_options)(struct seq_file *, struct dentry *);
124 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
125 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
126 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
129 All may block [not true, see below]
135 drop_inode: !!!inode->i_lock!!!
141 statfs: maybe(read) (see below)
144 show_options: no (namespace_sem)
145 quota_read: no (see below)
146 quota_write: no (see below)
147 bdev_try_to_free_page: no (see below)
149 ->statfs() has s_umount (shared) when called by ustat(2) (native or
150 compat), but that's an accident of bad API; s_umount is used to pin
151 the superblock down when we only have dev_t given us by userland to
152 identify the superblock. Everything else (statfs(), fstatfs(), etc.)
153 doesn't hold it when calling ->statfs() - superblock is pinned down
154 by resolving the pathname passed to syscall.
155 ->quota_read() and ->quota_write() functions are both guaranteed to
156 be the only ones operating on the quota file by the quota code (via
157 dqio_sem) (unless an admin really wants to screw up something and
158 writes to quota files with quotas on). For other details about locking
159 see also dquot_operations section.
160 ->bdev_try_to_free_page is called from the ->releasepage handler of
161 the block device inode. See there for more details.
163 --------------------------- file_system_type ---------------------------
165 struct dentry *(*mount) (struct file_system_type *, int,
166 const char *, void *);
167 void (*kill_sb) (struct super_block *);
173 ->mount() returns ERR_PTR or the root dentry; its superblock should be locked
175 ->kill_sb() takes a write-locked superblock, does all shutdown work on it,
176 unlocks and drops the reference.
178 --------------------------- address_space_operations --------------------------
180 int (*writepage)(struct page *page, struct writeback_control *wbc);
181 int (*readpage)(struct file *, struct page *);
182 int (*writepages)(struct address_space *, struct writeback_control *);
183 int (*set_page_dirty)(struct page *page);
184 int (*readpages)(struct file *filp, struct address_space *mapping,
185 struct list_head *pages, unsigned nr_pages);
186 int (*write_begin)(struct file *, struct address_space *mapping,
187 loff_t pos, unsigned len, unsigned flags,
188 struct page **pagep, void **fsdata);
189 int (*write_end)(struct file *, struct address_space *mapping,
190 loff_t pos, unsigned len, unsigned copied,
191 struct page *page, void *fsdata);
192 sector_t (*bmap)(struct address_space *, sector_t);
193 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
194 int (*releasepage) (struct page *, int);
195 void (*freepage)(struct page *);
196 int (*direct_IO)(struct kiocb *, struct iov_iter *iter);
197 int (*migratepage)(struct address_space *, struct page *, struct page *);
198 int (*launder_page)(struct page *);
199 int (*is_partially_uptodate)(struct page *, unsigned long, unsigned long);
200 int (*error_remove_page)(struct address_space *, struct page *);
201 int (*swap_activate)(struct file *);
202 int (*swap_deactivate)(struct file *);
205 All except set_page_dirty and freepage may block
207 PageLocked(page) i_mutex
208 writepage: yes, unlocks (see below)
209 readpage: yes, unlocks
213 write_begin: locks the page yes
214 write_end: yes, unlocks yes
220 migratepage: yes (both)
222 is_partially_uptodate: yes
223 error_remove_page: yes
227 ->write_begin(), ->write_end() and ->readpage() may be called from
228 the request handler (/dev/loop).
230 ->readpage() unlocks the page, either synchronously or via I/O
233 ->readpages() populates the pagecache with the passed pages and starts
234 I/O against them. They come unlocked upon I/O completion.
236 ->writepage() is used for two purposes: for "memory cleansing" and for
237 "sync". These are quite different operations and the behaviour may differ
238 depending upon the mode.
240 If writepage is called for sync (wbc->sync_mode != WBC_SYNC_NONE) then
241 it *must* start I/O against the page, even if that would involve
242 blocking on in-progress I/O.
244 If writepage is called for memory cleansing (sync_mode ==
245 WBC_SYNC_NONE) then its role is to get as much writeout underway as
246 possible. So writepage should try to avoid blocking against
247 currently-in-progress I/O.
249 If the filesystem is not called for "sync" and it determines that it
250 would need to block against in-progress I/O to be able to start new I/O
251 against the page the filesystem should redirty the page with
252 redirty_page_for_writepage(), then unlock the page and return zero.
253 This may also be done to avoid internal deadlocks, but rarely.
255 If the filesystem is called for sync then it must wait on any
256 in-progress I/O and then start new I/O.
258 The filesystem should unlock the page synchronously, before returning to the
259 caller, unless ->writepage() returns special WRITEPAGE_ACTIVATE
260 value. WRITEPAGE_ACTIVATE means that page cannot really be written out
261 currently, and VM should stop calling ->writepage() on this page for some
262 time. VM does this by moving page to the head of the active list, hence the
265 Unless the filesystem is going to redirty_page_for_writepage(), unlock the page
266 and return zero, writepage *must* run set_page_writeback() against the page,
267 followed by unlocking it. Once set_page_writeback() has been run against the
268 page, write I/O can be submitted and the write I/O completion handler must run
269 end_page_writeback() once the I/O is complete. If no I/O is submitted, the
270 filesystem must run end_page_writeback() against the page before returning from
273 That is: after 2.5.12, pages which are under writeout are *not* locked. Note,
274 if the filesystem needs the page to be locked during writeout, that is ok, too,
275 the page is allowed to be unlocked at any point in time between the calls to
276 set_page_writeback() and end_page_writeback().
278 Note, failure to run either redirty_page_for_writepage() or the combination of
279 set_page_writeback()/end_page_writeback() on a page submitted to writepage
280 will leave the page itself marked clean but it will be tagged as dirty in the
281 radix tree. This incoherency can lead to all sorts of hard-to-debug problems
282 in the filesystem like having dirty inodes at umount and losing written data.
284 ->writepages() is used for periodic writeback and for syscall-initiated
285 sync operations. The address_space should start I/O against at least
286 *nr_to_write pages. *nr_to_write must be decremented for each page which is
287 written. The address_space implementation may write more (or less) pages
288 than *nr_to_write asks for, but it should try to be reasonably close. If
289 nr_to_write is NULL, all dirty pages must be written.
291 writepages should _only_ write pages which are present on
294 ->set_page_dirty() is called from various places in the kernel
295 when the target page is marked as needing writeback. It may be called
296 under spinlock (it cannot block) and is sometimes called with the page
299 ->bmap() is currently used by legacy ioctl() (FIBMAP) provided by some
300 filesystems and by the swapper. The latter will eventually go away. Please,
301 keep it that way and don't breed new callers.
303 ->invalidatepage() is called when the filesystem must attempt to drop
304 some or all of the buffers from the page when it is being truncated. It
305 returns zero on success. If ->invalidatepage is zero, the kernel uses
306 block_invalidatepage() instead.
308 ->releasepage() is called when the kernel is about to try to drop the
309 buffers from the page in preparation for freeing it. It returns zero to
310 indicate that the buffers are (or may be) freeable. If ->releasepage is zero,
311 the kernel assumes that the fs has no private interest in the buffers.
313 ->freepage() is called when the kernel is done dropping the page
316 ->launder_page() may be called prior to releasing a page if
317 it is still found to be dirty. It returns zero if the page was successfully
318 cleaned, or an error value if not. Note that in order to prevent the page
319 getting mapped back in and redirtied, it needs to be kept locked
320 across the entire operation.
322 ->swap_activate will be called with a non-zero argument on
323 files backing (non block device backed) swapfiles. A return value
324 of zero indicates success, in which case this file can be used for
325 backing swapspace. The swapspace operations will be proxied to the
326 address space operations.
328 ->swap_deactivate() will be called in the sys_swapoff()
329 path after ->swap_activate() returned success.
331 ----------------------- file_lock_operations ------------------------------
333 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
334 void (*fl_release_private)(struct file_lock *);
338 inode->i_lock may block
340 fl_release_private: maybe maybe[1]
342 [1]: ->fl_release_private for flock or POSIX locks is currently allowed
343 to block. Leases however can still be freed while the i_lock is held and
344 so fl_release_private called on a lease should not block.
346 ----------------------- lock_manager_operations ---------------------------
348 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
349 unsigned long (*lm_owner_key)(struct file_lock *);
350 void (*lm_notify)(struct file_lock *); /* unblock callback */
351 int (*lm_grant)(struct file_lock *, struct file_lock *, int);
352 void (*lm_break)(struct file_lock *); /* break_lease callback */
353 int (*lm_change)(struct file_lock **, int);
357 inode->i_lock blocked_lock_lock may block
358 lm_compare_owner: yes[1] maybe no
359 lm_owner_key yes[1] yes no
360 lm_notify: yes yes no
365 [1]: ->lm_compare_owner and ->lm_owner_key are generally called with
366 *an* inode->i_lock held. It may not be the i_lock of the inode
367 associated with either file_lock argument! This is the case with deadlock
368 detection, since the code has to chase down the owners of locks that may
369 be entirely unrelated to the one on which the lock is being acquired.
370 For deadlock detection however, the blocked_lock_lock is also held. The
371 fact that these locks are held ensures that the file_locks do not
372 disappear out from under you while doing the comparison or generating an
375 --------------------------- buffer_head -----------------------------------
377 void (*b_end_io)(struct buffer_head *bh, int uptodate);
380 called from interrupts. In other words, extreme care is needed here.
381 bh is locked, but that's all warranties we have here. Currently only RAID1,
382 highmem, fs/buffer.c, and fs/ntfs/aops.c are providing these. Block devices
383 call this method upon the IO completion.
385 --------------------------- block_device_operations -----------------------
387 int (*open) (struct block_device *, fmode_t);
388 int (*release) (struct gendisk *, fmode_t);
389 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
390 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
391 int (*direct_access) (struct block_device *, sector_t, void __pmem **,
393 int (*media_changed) (struct gendisk *);
394 void (*unlock_native_capacity) (struct gendisk *);
395 int (*revalidate_disk) (struct gendisk *);
396 int (*getgeo)(struct block_device *, struct hd_geometry *);
397 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
407 unlock_native_capacity: no
410 swap_slot_free_notify: no (see below)
412 media_changed, unlock_native_capacity and revalidate_disk are called only from
415 swap_slot_free_notify is called with swap_lock and sometimes the page lock
419 --------------------------- file_operations -------------------------------
421 loff_t (*llseek) (struct file *, loff_t, int);
422 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
423 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
424 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
425 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
426 int (*iterate) (struct file *, struct dir_context *);
427 unsigned int (*poll) (struct file *, struct poll_table_struct *);
428 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
429 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
430 int (*mmap) (struct file *, struct vm_area_struct *);
431 int (*open) (struct inode *, struct file *);
432 int (*flush) (struct file *);
433 int (*release) (struct inode *, struct file *);
434 int (*fsync) (struct file *, loff_t start, loff_t end, int datasync);
435 int (*aio_fsync) (struct kiocb *, int datasync);
436 int (*fasync) (int, struct file *, int);
437 int (*lock) (struct file *, int, struct file_lock *);
438 ssize_t (*readv) (struct file *, const struct iovec *, unsigned long,
440 ssize_t (*writev) (struct file *, const struct iovec *, unsigned long,
442 ssize_t (*sendfile) (struct file *, loff_t *, size_t, read_actor_t,
444 ssize_t (*sendpage) (struct file *, struct page *, int, size_t,
446 unsigned long (*get_unmapped_area)(struct file *, unsigned long,
447 unsigned long, unsigned long, unsigned long);
448 int (*check_flags)(int);
449 int (*flock) (struct file *, int, struct file_lock *);
450 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *,
451 size_t, unsigned int);
452 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *,
453 size_t, unsigned int);
454 int (*setlease)(struct file *, long, struct file_lock **, void **);
455 long (*fallocate)(struct file *, int, loff_t, loff_t);
461 ->llseek() locking has moved from llseek to the individual llseek
462 implementations. If your fs is not using generic_file_llseek, you
463 need to acquire and release the appropriate locks in your ->llseek().
464 For many filesystems, it is probably safe to acquire the inode
465 mutex or just to use i_size_read() instead.
466 Note: this does not protect the file->f_pos against concurrent modifications
467 since this is something the userspace has to take care about.
469 ->fasync() is responsible for maintaining the FASYNC bit in filp->f_flags.
470 Most instances call fasync_helper(), which does that maintenance, so it's
471 not normally something one needs to worry about. Return values > 0 will be
472 mapped to zero in the VFS layer.
474 ->readdir() and ->ioctl() on directories must be changed. Ideally we would
475 move ->readdir() to inode_operations and use a separate method for directory
476 ->ioctl() or kill the latter completely. One of the problems is that for
477 anything that resembles union-mount we won't have a struct file for all
478 components. And there are other reasons why the current interface is a mess...
480 ->read on directories probably must go away - we should just enforce -EISDIR
481 in sys_read() and friends.
483 ->setlease operations should call generic_setlease() before or after setting
484 the lease within the individual filesystem to record the result of the
487 --------------------------- dquot_operations -------------------------------
489 int (*write_dquot) (struct dquot *);
490 int (*acquire_dquot) (struct dquot *);
491 int (*release_dquot) (struct dquot *);
492 int (*mark_dirty) (struct dquot *);
493 int (*write_info) (struct super_block *, int);
495 These operations are intended to be more or less wrapping functions that ensure
496 a proper locking wrt the filesystem and call the generic quota operations.
498 What filesystem should expect from the generic quota functions:
500 FS recursion Held locks when called
501 write_dquot: yes dqonoff_sem or dqptr_sem
502 acquire_dquot: yes dqonoff_sem or dqptr_sem
503 release_dquot: yes dqonoff_sem or dqptr_sem
505 write_info: yes dqonoff_sem
507 FS recursion means calling ->quota_read() and ->quota_write() from superblock
510 More details about quota locking can be found in fs/dquot.c.
512 --------------------------- vm_operations_struct -----------------------------
514 void (*open)(struct vm_area_struct*);
515 void (*close)(struct vm_area_struct*);
516 int (*fault)(struct vm_area_struct*, struct vm_fault *);
517 int (*page_mkwrite)(struct vm_area_struct *, struct vm_fault *);
518 int (*pfn_mkwrite)(struct vm_area_struct *, struct vm_fault *);
519 int (*access)(struct vm_area_struct *, unsigned long, void*, int, int);
522 mmap_sem PageLocked(page)
525 fault: yes can return with page locked
527 page_mkwrite: yes can return with page locked
531 ->fault() is called when a previously not present pte is about
532 to be faulted in. The filesystem must find and return the page associated
533 with the passed in "pgoff" in the vm_fault structure. If it is possible that
534 the page may be truncated and/or invalidated, then the filesystem must lock
535 the page, then ensure it is not already truncated (the page lock will block
536 subsequent truncate), and then return with VM_FAULT_LOCKED, and the page
537 locked. The VM will unlock the page.
539 ->map_pages() is called when VM asks to map easy accessible pages.
540 Filesystem should find and map pages associated with offsets from "pgoff"
541 till "max_pgoff". ->map_pages() is called with page table locked and must
542 not block. If it's not possible to reach a page without blocking,
543 filesystem should skip it. Filesystem should use do_set_pte() to setup
544 page table entry. Pointer to entry associated with offset "pgoff" is
545 passed in "pte" field in vm_fault structure. Pointers to entries for other
546 offsets should be calculated relative to "pte".
548 ->page_mkwrite() is called when a previously read-only pte is
549 about to become writeable. The filesystem again must ensure that there are
550 no truncate/invalidate races, and then return with the page locked. If
551 the page has been truncated, the filesystem should not look up a new page
552 like the ->fault() handler, but simply return with VM_FAULT_NOPAGE, which
553 will cause the VM to retry the fault.
555 ->pfn_mkwrite() is the same as page_mkwrite but when the pte is
556 VM_PFNMAP or VM_MIXEDMAP with a page-less entry. Expected return is
557 VM_FAULT_NOPAGE. Or one of the VM_FAULT_ERROR types. The default behavior
558 after this call is to make the pte read-write, unless pfn_mkwrite returns
561 ->access() is called when get_user_pages() fails in
562 access_process_vm(), typically used to debug a process through
563 /proc/pid/mem or ptrace. This function is needed only for
564 VM_IO | VM_PFNMAP VMAs.
566 ================================================================================
569 (if you break something or notice that it is broken and do not fix it yourself
570 - at least put it here)