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 *,
16 unsigned int, const char *, const struct qstr *);
17 int (*d_delete)(struct dentry *);
18 int (*d_init)(struct dentry *);
19 void (*d_release)(struct dentry *);
20 void (*d_iput)(struct dentry *, struct inode *);
21 char *(*d_dname)((struct dentry *dentry, char *buffer, int buflen);
22 struct vfsmount *(*d_automount)(struct path *path);
23 int (*d_manage)(const struct path *, bool);
24 struct dentry *(*d_real)(struct dentry *, const struct inode *,
28 rename_lock ->d_lock may block rcu-walk
29 d_revalidate: no no yes (ref-walk) maybe
30 d_weak_revalidate:no no yes no
32 d_compare: yes no no maybe
33 d_delete: no yes no no
35 d_release: no no yes no
39 d_automount: no no yes no
40 d_manage: no no yes (ref-walk) maybe
43 --------------------------- inode_operations ---------------------------
45 int (*create) (struct inode *,struct dentry *,umode_t, bool);
46 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
47 int (*link) (struct dentry *,struct inode *,struct dentry *);
48 int (*unlink) (struct inode *,struct dentry *);
49 int (*symlink) (struct inode *,struct dentry *,const char *);
50 int (*mkdir) (struct inode *,struct dentry *,umode_t);
51 int (*rmdir) (struct inode *,struct dentry *);
52 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
53 int (*rename) (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) (const struct path *, struct kstat *, u32, unsigned int);
62 ssize_t (*listxattr) (struct dentry *, char *, size_t);
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);
80 rmdir: yes (both) (see below)
81 rename: yes (all) (see below)
85 permission: no (may not block if called in rcu-walk mode)
95 Additionally, ->rmdir(), ->unlink() and ->rename() have ->i_mutex on
97 cross-directory ->rename() has (per-superblock) ->s_vfs_rename_sem.
99 See Documentation/filesystems/directory-locking for more detailed discussion
100 of the locking scheme for directory operations.
102 ----------------------- xattr_handler operations -----------------------
104 bool (*list)(struct dentry *dentry);
105 int (*get)(const struct xattr_handler *handler, struct dentry *dentry,
106 struct inode *inode, const char *name, void *buffer,
108 int (*set)(const struct xattr_handler *handler, struct dentry *dentry,
109 struct inode *inode, const char *name, const void *buffer,
110 size_t size, int flags);
119 --------------------------- super_operations ---------------------------
121 struct inode *(*alloc_inode)(struct super_block *sb);
122 void (*destroy_inode)(struct inode *);
123 void (*dirty_inode) (struct inode *, int flags);
124 int (*write_inode) (struct inode *, struct writeback_control *wbc);
125 int (*drop_inode) (struct inode *);
126 void (*evict_inode) (struct inode *);
127 void (*put_super) (struct super_block *);
128 int (*sync_fs)(struct super_block *sb, int wait);
129 int (*freeze_fs) (struct super_block *);
130 int (*unfreeze_fs) (struct super_block *);
131 int (*statfs) (struct dentry *, struct kstatfs *);
132 int (*remount_fs) (struct super_block *, int *, char *);
133 void (*umount_begin) (struct super_block *);
134 int (*show_options)(struct seq_file *, struct dentry *);
135 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
136 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
137 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
140 All may block [not true, see below]
146 drop_inode: !!!inode->i_lock!!!
152 statfs: maybe(read) (see below)
155 show_options: no (namespace_sem)
156 quota_read: no (see below)
157 quota_write: no (see below)
158 bdev_try_to_free_page: no (see below)
160 ->statfs() has s_umount (shared) when called by ustat(2) (native or
161 compat), but that's an accident of bad API; s_umount is used to pin
162 the superblock down when we only have dev_t given us by userland to
163 identify the superblock. Everything else (statfs(), fstatfs(), etc.)
164 doesn't hold it when calling ->statfs() - superblock is pinned down
165 by resolving the pathname passed to syscall.
166 ->quota_read() and ->quota_write() functions are both guaranteed to
167 be the only ones operating on the quota file by the quota code (via
168 dqio_sem) (unless an admin really wants to screw up something and
169 writes to quota files with quotas on). For other details about locking
170 see also dquot_operations section.
171 ->bdev_try_to_free_page is called from the ->releasepage handler of
172 the block device inode. See there for more details.
174 --------------------------- file_system_type ---------------------------
176 struct dentry *(*mount) (struct file_system_type *, int,
177 const char *, void *);
178 void (*kill_sb) (struct super_block *);
184 ->mount() returns ERR_PTR or the root dentry; its superblock should be locked
186 ->kill_sb() takes a write-locked superblock, does all shutdown work on it,
187 unlocks and drops the reference.
189 --------------------------- address_space_operations --------------------------
191 int (*writepage)(struct page *page, struct writeback_control *wbc);
192 int (*readpage)(struct file *, struct page *);
193 int (*writepages)(struct address_space *, struct writeback_control *);
194 int (*set_page_dirty)(struct page *page);
195 int (*readpages)(struct file *filp, struct address_space *mapping,
196 struct list_head *pages, unsigned nr_pages);
197 int (*write_begin)(struct file *, struct address_space *mapping,
198 loff_t pos, unsigned len, unsigned flags,
199 struct page **pagep, void **fsdata);
200 int (*write_end)(struct file *, struct address_space *mapping,
201 loff_t pos, unsigned len, unsigned copied,
202 struct page *page, void *fsdata);
203 sector_t (*bmap)(struct address_space *, sector_t);
204 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
205 int (*releasepage) (struct page *, int);
206 void (*freepage)(struct page *);
207 int (*direct_IO)(struct kiocb *, struct iov_iter *iter);
208 bool (*isolate_page) (struct page *, isolate_mode_t);
209 int (*migratepage)(struct address_space *, struct page *, struct page *);
210 void (*putback_page) (struct page *);
211 int (*launder_page)(struct page *);
212 int (*is_partially_uptodate)(struct page *, unsigned long, unsigned long);
213 int (*error_remove_page)(struct address_space *, struct page *);
214 int (*swap_activate)(struct file *);
215 int (*swap_deactivate)(struct file *);
218 All except set_page_dirty and freepage may block
220 PageLocked(page) i_mutex
221 writepage: yes, unlocks (see below)
222 readpage: yes, unlocks
226 write_begin: locks the page yes
227 write_end: yes, unlocks yes
234 migratepage: yes (both)
237 is_partially_uptodate: yes
238 error_remove_page: yes
242 ->write_begin(), ->write_end() and ->readpage() may be called from
243 the request handler (/dev/loop).
245 ->readpage() unlocks the page, either synchronously or via I/O
248 ->readpages() populates the pagecache with the passed pages and starts
249 I/O against them. They come unlocked upon I/O completion.
251 ->writepage() is used for two purposes: for "memory cleansing" and for
252 "sync". These are quite different operations and the behaviour may differ
253 depending upon the mode.
255 If writepage is called for sync (wbc->sync_mode != WBC_SYNC_NONE) then
256 it *must* start I/O against the page, even if that would involve
257 blocking on in-progress I/O.
259 If writepage is called for memory cleansing (sync_mode ==
260 WBC_SYNC_NONE) then its role is to get as much writeout underway as
261 possible. So writepage should try to avoid blocking against
262 currently-in-progress I/O.
264 If the filesystem is not called for "sync" and it determines that it
265 would need to block against in-progress I/O to be able to start new I/O
266 against the page the filesystem should redirty the page with
267 redirty_page_for_writepage(), then unlock the page and return zero.
268 This may also be done to avoid internal deadlocks, but rarely.
270 If the filesystem is called for sync then it must wait on any
271 in-progress I/O and then start new I/O.
273 The filesystem should unlock the page synchronously, before returning to the
274 caller, unless ->writepage() returns special WRITEPAGE_ACTIVATE
275 value. WRITEPAGE_ACTIVATE means that page cannot really be written out
276 currently, and VM should stop calling ->writepage() on this page for some
277 time. VM does this by moving page to the head of the active list, hence the
280 Unless the filesystem is going to redirty_page_for_writepage(), unlock the page
281 and return zero, writepage *must* run set_page_writeback() against the page,
282 followed by unlocking it. Once set_page_writeback() has been run against the
283 page, write I/O can be submitted and the write I/O completion handler must run
284 end_page_writeback() once the I/O is complete. If no I/O is submitted, the
285 filesystem must run end_page_writeback() against the page before returning from
288 That is: after 2.5.12, pages which are under writeout are *not* locked. Note,
289 if the filesystem needs the page to be locked during writeout, that is ok, too,
290 the page is allowed to be unlocked at any point in time between the calls to
291 set_page_writeback() and end_page_writeback().
293 Note, failure to run either redirty_page_for_writepage() or the combination of
294 set_page_writeback()/end_page_writeback() on a page submitted to writepage
295 will leave the page itself marked clean but it will be tagged as dirty in the
296 radix tree. This incoherency can lead to all sorts of hard-to-debug problems
297 in the filesystem like having dirty inodes at umount and losing written data.
299 ->writepages() is used for periodic writeback and for syscall-initiated
300 sync operations. The address_space should start I/O against at least
301 *nr_to_write pages. *nr_to_write must be decremented for each page which is
302 written. The address_space implementation may write more (or less) pages
303 than *nr_to_write asks for, but it should try to be reasonably close. If
304 nr_to_write is NULL, all dirty pages must be written.
306 writepages should _only_ write pages which are present on
309 ->set_page_dirty() is called from various places in the kernel
310 when the target page is marked as needing writeback. It may be called
311 under spinlock (it cannot block) and is sometimes called with the page
314 ->bmap() is currently used by legacy ioctl() (FIBMAP) provided by some
315 filesystems and by the swapper. The latter will eventually go away. Please,
316 keep it that way and don't breed new callers.
318 ->invalidatepage() is called when the filesystem must attempt to drop
319 some or all of the buffers from the page when it is being truncated. It
320 returns zero on success. If ->invalidatepage is zero, the kernel uses
321 block_invalidatepage() instead.
323 ->releasepage() is called when the kernel is about to try to drop the
324 buffers from the page in preparation for freeing it. It returns zero to
325 indicate that the buffers are (or may be) freeable. If ->releasepage is zero,
326 the kernel assumes that the fs has no private interest in the buffers.
328 ->freepage() is called when the kernel is done dropping the page
331 ->launder_page() may be called prior to releasing a page if
332 it is still found to be dirty. It returns zero if the page was successfully
333 cleaned, or an error value if not. Note that in order to prevent the page
334 getting mapped back in and redirtied, it needs to be kept locked
335 across the entire operation.
337 ->swap_activate will be called with a non-zero argument on
338 files backing (non block device backed) swapfiles. A return value
339 of zero indicates success, in which case this file can be used for
340 backing swapspace. The swapspace operations will be proxied to the
341 address space operations.
343 ->swap_deactivate() will be called in the sys_swapoff()
344 path after ->swap_activate() returned success.
346 ----------------------- file_lock_operations ------------------------------
348 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
349 void (*fl_release_private)(struct file_lock *);
353 inode->i_lock may block
355 fl_release_private: maybe maybe[1]
357 [1]: ->fl_release_private for flock or POSIX locks is currently allowed
358 to block. Leases however can still be freed while the i_lock is held and
359 so fl_release_private called on a lease should not block.
361 ----------------------- lock_manager_operations ---------------------------
363 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
364 unsigned long (*lm_owner_key)(struct file_lock *);
365 void (*lm_notify)(struct file_lock *); /* unblock callback */
366 int (*lm_grant)(struct file_lock *, struct file_lock *, int);
367 void (*lm_break)(struct file_lock *); /* break_lease callback */
368 int (*lm_change)(struct file_lock **, int);
372 inode->i_lock blocked_lock_lock may block
373 lm_compare_owner: yes[1] maybe no
374 lm_owner_key yes[1] yes no
375 lm_notify: yes yes no
380 [1]: ->lm_compare_owner and ->lm_owner_key are generally called with
381 *an* inode->i_lock held. It may not be the i_lock of the inode
382 associated with either file_lock argument! This is the case with deadlock
383 detection, since the code has to chase down the owners of locks that may
384 be entirely unrelated to the one on which the lock is being acquired.
385 For deadlock detection however, the blocked_lock_lock is also held. The
386 fact that these locks are held ensures that the file_locks do not
387 disappear out from under you while doing the comparison or generating an
390 --------------------------- buffer_head -----------------------------------
392 void (*b_end_io)(struct buffer_head *bh, int uptodate);
395 called from interrupts. In other words, extreme care is needed here.
396 bh is locked, but that's all warranties we have here. Currently only RAID1,
397 highmem, fs/buffer.c, and fs/ntfs/aops.c are providing these. Block devices
398 call this method upon the IO completion.
400 --------------------------- block_device_operations -----------------------
402 int (*open) (struct block_device *, fmode_t);
403 int (*release) (struct gendisk *, fmode_t);
404 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
405 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
406 int (*direct_access) (struct block_device *, sector_t, void **,
408 int (*media_changed) (struct gendisk *);
409 void (*unlock_native_capacity) (struct gendisk *);
410 int (*revalidate_disk) (struct gendisk *);
411 int (*getgeo)(struct block_device *, struct hd_geometry *);
412 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
422 unlock_native_capacity: no
425 swap_slot_free_notify: no (see below)
427 media_changed, unlock_native_capacity and revalidate_disk are called only from
430 swap_slot_free_notify is called with swap_lock and sometimes the page lock
434 --------------------------- file_operations -------------------------------
436 loff_t (*llseek) (struct file *, loff_t, int);
437 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
438 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
439 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
440 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
441 int (*iterate) (struct file *, struct dir_context *);
442 unsigned int (*poll) (struct file *, struct poll_table_struct *);
443 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
444 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
445 int (*mmap) (struct file *, struct vm_area_struct *);
446 int (*open) (struct inode *, struct file *);
447 int (*flush) (struct file *);
448 int (*release) (struct inode *, struct file *);
449 int (*fsync) (struct file *, loff_t start, loff_t end, int datasync);
450 int (*fasync) (int, struct file *, int);
451 int (*lock) (struct file *, int, struct file_lock *);
452 ssize_t (*readv) (struct file *, const struct iovec *, unsigned long,
454 ssize_t (*writev) (struct file *, const struct iovec *, unsigned long,
456 ssize_t (*sendfile) (struct file *, loff_t *, size_t, read_actor_t,
458 ssize_t (*sendpage) (struct file *, struct page *, int, size_t,
460 unsigned long (*get_unmapped_area)(struct file *, unsigned long,
461 unsigned long, unsigned long, unsigned long);
462 int (*check_flags)(int);
463 int (*flock) (struct file *, int, struct file_lock *);
464 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *,
465 size_t, unsigned int);
466 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *,
467 size_t, unsigned int);
468 int (*setlease)(struct file *, long, struct file_lock **, void **);
469 long (*fallocate)(struct file *, int, loff_t, loff_t);
475 ->llseek() locking has moved from llseek to the individual llseek
476 implementations. If your fs is not using generic_file_llseek, you
477 need to acquire and release the appropriate locks in your ->llseek().
478 For many filesystems, it is probably safe to acquire the inode
479 mutex or just to use i_size_read() instead.
480 Note: this does not protect the file->f_pos against concurrent modifications
481 since this is something the userspace has to take care about.
483 ->fasync() is responsible for maintaining the FASYNC bit in filp->f_flags.
484 Most instances call fasync_helper(), which does that maintenance, so it's
485 not normally something one needs to worry about. Return values > 0 will be
486 mapped to zero in the VFS layer.
488 ->readdir() and ->ioctl() on directories must be changed. Ideally we would
489 move ->readdir() to inode_operations and use a separate method for directory
490 ->ioctl() or kill the latter completely. One of the problems is that for
491 anything that resembles union-mount we won't have a struct file for all
492 components. And there are other reasons why the current interface is a mess...
494 ->read on directories probably must go away - we should just enforce -EISDIR
495 in sys_read() and friends.
497 ->setlease operations should call generic_setlease() before or after setting
498 the lease within the individual filesystem to record the result of the
501 --------------------------- dquot_operations -------------------------------
503 int (*write_dquot) (struct dquot *);
504 int (*acquire_dquot) (struct dquot *);
505 int (*release_dquot) (struct dquot *);
506 int (*mark_dirty) (struct dquot *);
507 int (*write_info) (struct super_block *, int);
509 These operations are intended to be more or less wrapping functions that ensure
510 a proper locking wrt the filesystem and call the generic quota operations.
512 What filesystem should expect from the generic quota functions:
514 FS recursion Held locks when called
515 write_dquot: yes dqonoff_sem or dqptr_sem
516 acquire_dquot: yes dqonoff_sem or dqptr_sem
517 release_dquot: yes dqonoff_sem or dqptr_sem
519 write_info: yes dqonoff_sem
521 FS recursion means calling ->quota_read() and ->quota_write() from superblock
524 More details about quota locking can be found in fs/dquot.c.
526 --------------------------- vm_operations_struct -----------------------------
528 void (*open)(struct vm_area_struct*);
529 void (*close)(struct vm_area_struct*);
530 int (*fault)(struct vm_area_struct*, struct vm_fault *);
531 int (*page_mkwrite)(struct vm_area_struct *, struct vm_fault *);
532 int (*pfn_mkwrite)(struct vm_area_struct *, struct vm_fault *);
533 int (*access)(struct vm_area_struct *, unsigned long, void*, int, int);
536 mmap_sem PageLocked(page)
539 fault: yes can return with page locked
541 page_mkwrite: yes can return with page locked
545 ->fault() is called when a previously not present pte is about
546 to be faulted in. The filesystem must find and return the page associated
547 with the passed in "pgoff" in the vm_fault structure. If it is possible that
548 the page may be truncated and/or invalidated, then the filesystem must lock
549 the page, then ensure it is not already truncated (the page lock will block
550 subsequent truncate), and then return with VM_FAULT_LOCKED, and the page
551 locked. The VM will unlock the page.
553 ->map_pages() is called when VM asks to map easy accessible pages.
554 Filesystem should find and map pages associated with offsets from "start_pgoff"
555 till "end_pgoff". ->map_pages() is called with page table locked and must
556 not block. If it's not possible to reach a page without blocking,
557 filesystem should skip it. Filesystem should use do_set_pte() to setup
558 page table entry. Pointer to entry associated with the page is passed in
559 "pte" field in vm_fault structure. Pointers to entries for other offsets
560 should be calculated relative to "pte".
562 ->page_mkwrite() is called when a previously read-only pte is
563 about to become writeable. The filesystem again must ensure that there are
564 no truncate/invalidate races, and then return with the page locked. If
565 the page has been truncated, the filesystem should not look up a new page
566 like the ->fault() handler, but simply return with VM_FAULT_NOPAGE, which
567 will cause the VM to retry the fault.
569 ->pfn_mkwrite() is the same as page_mkwrite but when the pte is
570 VM_PFNMAP or VM_MIXEDMAP with a page-less entry. Expected return is
571 VM_FAULT_NOPAGE. Or one of the VM_FAULT_ERROR types. The default behavior
572 after this call is to make the pte read-write, unless pfn_mkwrite returns
575 ->access() is called when get_user_pages() fails in
576 access_process_vm(), typically used to debug a process through
577 /proc/pid/mem or ptrace. This function is needed only for
578 VM_IO | VM_PFNMAP VMAs.
580 ================================================================================
583 (if you break something or notice that it is broken and do not fix it yourself
584 - at least put it here)