f48e0c6b4c4295deeda9358cfe8f706b8feea2c0
[deliverable/linux.git] / Documentation / filesystems / Locking
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?
9
10 --------------------------- dentry_operations --------------------------
11 prototypes:
12 int (*d_revalidate)(struct dentry *, unsigned int);
13 int (*d_hash)(const struct dentry *, const struct inode *,
14 struct qstr *);
15 int (*d_compare)(const struct dentry *, const struct inode *,
16 const struct dentry *, const struct inode *,
17 unsigned int, const char *, const struct qstr *);
18 int (*d_delete)(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)(struct dentry *, bool);
24
25 locking rules:
26 rename_lock ->d_lock may block rcu-walk
27 d_revalidate: no no yes (ref-walk) maybe
28 d_hash no no no maybe
29 d_compare: yes no no maybe
30 d_delete: no yes no no
31 d_release: no no yes no
32 d_prune: no yes no no
33 d_iput: no no yes no
34 d_dname: no no no no
35 d_automount: no no yes no
36 d_manage: no no yes (ref-walk) maybe
37
38 --------------------------- inode_operations ---------------------------
39 prototypes:
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 (*readlink) (struct dentry *, char __user *,int);
51 void * (*follow_link) (struct dentry *, struct nameidata *);
52 void (*put_link) (struct dentry *, struct nameidata *, void *);
53 void (*truncate) (struct inode *);
54 int (*permission) (struct inode *, int, unsigned int);
55 int (*get_acl)(struct inode *, int);
56 int (*setattr) (struct dentry *, struct iattr *);
57 int (*getattr) (struct vfsmount *, struct dentry *, struct kstat *);
58 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
59 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
60 ssize_t (*listxattr) (struct dentry *, char *, size_t);
61 int (*removexattr) (struct dentry *, const char *);
62 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start, u64 len);
63 void (*update_time)(struct inode *, struct timespec *, int);
64 int (*atomic_open)(struct inode *, struct dentry *,
65 struct file *, unsigned open_flag,
66 umode_t create_mode, int *opened);
67
68 locking rules:
69 all may block
70 i_mutex(inode)
71 lookup: yes
72 create: yes
73 link: yes (both)
74 mknod: yes
75 symlink: yes
76 mkdir: yes
77 unlink: yes (both)
78 rmdir: yes (both) (see below)
79 rename: yes (all) (see below)
80 readlink: no
81 follow_link: no
82 put_link: no
83 setattr: yes
84 permission: no (may not block if called in rcu-walk mode)
85 get_acl: no
86 getattr: no
87 setxattr: yes
88 getxattr: no
89 listxattr: no
90 removexattr: yes
91 fiemap: no
92 update_time: no
93 atomic_open: yes
94
95 Additionally, ->rmdir(), ->unlink() and ->rename() have ->i_mutex on
96 victim.
97 cross-directory ->rename() has (per-superblock) ->s_vfs_rename_sem.
98
99 See Documentation/filesystems/directory-locking for more detailed discussion
100 of the locking scheme for directory operations.
101
102 --------------------------- super_operations ---------------------------
103 prototypes:
104 struct inode *(*alloc_inode)(struct super_block *sb);
105 void (*destroy_inode)(struct inode *);
106 void (*dirty_inode) (struct inode *, int flags);
107 int (*write_inode) (struct inode *, struct writeback_control *wbc);
108 int (*drop_inode) (struct inode *);
109 void (*evict_inode) (struct inode *);
110 void (*put_super) (struct super_block *);
111 int (*sync_fs)(struct super_block *sb, int wait);
112 int (*freeze_fs) (struct super_block *);
113 int (*unfreeze_fs) (struct super_block *);
114 int (*statfs) (struct dentry *, struct kstatfs *);
115 int (*remount_fs) (struct super_block *, int *, char *);
116 void (*umount_begin) (struct super_block *);
117 int (*show_options)(struct seq_file *, struct dentry *);
118 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
119 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
120 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
121
122 locking rules:
123 All may block [not true, see below]
124 s_umount
125 alloc_inode:
126 destroy_inode:
127 dirty_inode:
128 write_inode:
129 drop_inode: !!!inode->i_lock!!!
130 evict_inode:
131 put_super: write
132 sync_fs: read
133 freeze_fs: write
134 unfreeze_fs: write
135 statfs: maybe(read) (see below)
136 remount_fs: write
137 umount_begin: no
138 show_options: no (namespace_sem)
139 quota_read: no (see below)
140 quota_write: no (see below)
141 bdev_try_to_free_page: no (see below)
142
143 ->statfs() has s_umount (shared) when called by ustat(2) (native or
144 compat), but that's an accident of bad API; s_umount is used to pin
145 the superblock down when we only have dev_t given us by userland to
146 identify the superblock. Everything else (statfs(), fstatfs(), etc.)
147 doesn't hold it when calling ->statfs() - superblock is pinned down
148 by resolving the pathname passed to syscall.
149 ->quota_read() and ->quota_write() functions are both guaranteed to
150 be the only ones operating on the quota file by the quota code (via
151 dqio_sem) (unless an admin really wants to screw up something and
152 writes to quota files with quotas on). For other details about locking
153 see also dquot_operations section.
154 ->bdev_try_to_free_page is called from the ->releasepage handler of
155 the block device inode. See there for more details.
156
157 --------------------------- file_system_type ---------------------------
158 prototypes:
159 int (*get_sb) (struct file_system_type *, int,
160 const char *, void *, struct vfsmount *);
161 struct dentry *(*mount) (struct file_system_type *, int,
162 const char *, void *);
163 void (*kill_sb) (struct super_block *);
164 locking rules:
165 may block
166 mount yes
167 kill_sb yes
168
169 ->mount() returns ERR_PTR or the root dentry; its superblock should be locked
170 on return.
171 ->kill_sb() takes a write-locked superblock, does all shutdown work on it,
172 unlocks and drops the reference.
173
174 --------------------------- address_space_operations --------------------------
175 prototypes:
176 int (*writepage)(struct page *page, struct writeback_control *wbc);
177 int (*readpage)(struct file *, struct page *);
178 int (*sync_page)(struct page *);
179 int (*writepages)(struct address_space *, struct writeback_control *);
180 int (*set_page_dirty)(struct page *page);
181 int (*readpages)(struct file *filp, struct address_space *mapping,
182 struct list_head *pages, unsigned nr_pages);
183 int (*write_begin)(struct file *, struct address_space *mapping,
184 loff_t pos, unsigned len, unsigned flags,
185 struct page **pagep, void **fsdata);
186 int (*write_end)(struct file *, struct address_space *mapping,
187 loff_t pos, unsigned len, unsigned copied,
188 struct page *page, void *fsdata);
189 sector_t (*bmap)(struct address_space *, sector_t);
190 int (*invalidatepage) (struct page *, unsigned long);
191 int (*releasepage) (struct page *, int);
192 void (*freepage)(struct page *);
193 int (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
194 loff_t offset, unsigned long nr_segs);
195 int (*get_xip_mem)(struct address_space *, pgoff_t, int, void **,
196 unsigned long *);
197 int (*migratepage)(struct address_space *, struct page *, struct page *);
198 int (*launder_page)(struct page *);
199 int (*is_partially_uptodate)(struct page *, read_descriptor_t *, unsigned long);
200 int (*error_remove_page)(struct address_space *, struct page *);
201 int (*swap_activate)(struct file *);
202 int (*swap_deactivate)(struct file *);
203
204 locking rules:
205 All except set_page_dirty and freepage may block
206
207 PageLocked(page) i_mutex
208 writepage: yes, unlocks (see below)
209 readpage: yes, unlocks
210 sync_page: maybe
211 writepages:
212 set_page_dirty no
213 readpages:
214 write_begin: locks the page yes
215 write_end: yes, unlocks yes
216 bmap:
217 invalidatepage: yes
218 releasepage: yes
219 freepage: yes
220 direct_IO:
221 get_xip_mem: maybe
222 migratepage: yes (both)
223 launder_page: yes
224 is_partially_uptodate: yes
225 error_remove_page: yes
226 swap_activate: no
227 swap_deactivate: no
228
229 ->write_begin(), ->write_end(), ->sync_page() and ->readpage()
230 may be called from the request handler (/dev/loop).
231
232 ->readpage() unlocks the page, either synchronously or via I/O
233 completion.
234
235 ->readpages() populates the pagecache with the passed pages and starts
236 I/O against them. They come unlocked upon I/O completion.
237
238 ->writepage() is used for two purposes: for "memory cleansing" and for
239 "sync". These are quite different operations and the behaviour may differ
240 depending upon the mode.
241
242 If writepage is called for sync (wbc->sync_mode != WBC_SYNC_NONE) then
243 it *must* start I/O against the page, even if that would involve
244 blocking on in-progress I/O.
245
246 If writepage is called for memory cleansing (sync_mode ==
247 WBC_SYNC_NONE) then its role is to get as much writeout underway as
248 possible. So writepage should try to avoid blocking against
249 currently-in-progress I/O.
250
251 If the filesystem is not called for "sync" and it determines that it
252 would need to block against in-progress I/O to be able to start new I/O
253 against the page the filesystem should redirty the page with
254 redirty_page_for_writepage(), then unlock the page and return zero.
255 This may also be done to avoid internal deadlocks, but rarely.
256
257 If the filesystem is called for sync then it must wait on any
258 in-progress I/O and then start new I/O.
259
260 The filesystem should unlock the page synchronously, before returning to the
261 caller, unless ->writepage() returns special WRITEPAGE_ACTIVATE
262 value. WRITEPAGE_ACTIVATE means that page cannot really be written out
263 currently, and VM should stop calling ->writepage() on this page for some
264 time. VM does this by moving page to the head of the active list, hence the
265 name.
266
267 Unless the filesystem is going to redirty_page_for_writepage(), unlock the page
268 and return zero, writepage *must* run set_page_writeback() against the page,
269 followed by unlocking it. Once set_page_writeback() has been run against the
270 page, write I/O can be submitted and the write I/O completion handler must run
271 end_page_writeback() once the I/O is complete. If no I/O is submitted, the
272 filesystem must run end_page_writeback() against the page before returning from
273 writepage.
274
275 That is: after 2.5.12, pages which are under writeout are *not* locked. Note,
276 if the filesystem needs the page to be locked during writeout, that is ok, too,
277 the page is allowed to be unlocked at any point in time between the calls to
278 set_page_writeback() and end_page_writeback().
279
280 Note, failure to run either redirty_page_for_writepage() or the combination of
281 set_page_writeback()/end_page_writeback() on a page submitted to writepage
282 will leave the page itself marked clean but it will be tagged as dirty in the
283 radix tree. This incoherency can lead to all sorts of hard-to-debug problems
284 in the filesystem like having dirty inodes at umount and losing written data.
285
286 ->sync_page() locking rules are not well-defined - usually it is called
287 with lock on page, but that is not guaranteed. Considering the currently
288 existing instances of this method ->sync_page() itself doesn't look
289 well-defined...
290
291 ->writepages() is used for periodic writeback and for syscall-initiated
292 sync operations. The address_space should start I/O against at least
293 *nr_to_write pages. *nr_to_write must be decremented for each page which is
294 written. The address_space implementation may write more (or less) pages
295 than *nr_to_write asks for, but it should try to be reasonably close. If
296 nr_to_write is NULL, all dirty pages must be written.
297
298 writepages should _only_ write pages which are present on
299 mapping->io_pages.
300
301 ->set_page_dirty() is called from various places in the kernel
302 when the target page is marked as needing writeback. It may be called
303 under spinlock (it cannot block) and is sometimes called with the page
304 not locked.
305
306 ->bmap() is currently used by legacy ioctl() (FIBMAP) provided by some
307 filesystems and by the swapper. The latter will eventually go away. Please,
308 keep it that way and don't breed new callers.
309
310 ->invalidatepage() is called when the filesystem must attempt to drop
311 some or all of the buffers from the page when it is being truncated. It
312 returns zero on success. If ->invalidatepage is zero, the kernel uses
313 block_invalidatepage() instead.
314
315 ->releasepage() is called when the kernel is about to try to drop the
316 buffers from the page in preparation for freeing it. It returns zero to
317 indicate that the buffers are (or may be) freeable. If ->releasepage is zero,
318 the kernel assumes that the fs has no private interest in the buffers.
319
320 ->freepage() is called when the kernel is done dropping the page
321 from the page cache.
322
323 ->launder_page() may be called prior to releasing a page if
324 it is still found to be dirty. It returns zero if the page was successfully
325 cleaned, or an error value if not. Note that in order to prevent the page
326 getting mapped back in and redirtied, it needs to be kept locked
327 across the entire operation.
328
329 ->swap_activate will be called with a non-zero argument on
330 files backing (non block device backed) swapfiles. A return value
331 of zero indicates success, in which case this file can be used for
332 backing swapspace. The swapspace operations will be proxied to the
333 address space operations.
334
335 ->swap_deactivate() will be called in the sys_swapoff()
336 path after ->swap_activate() returned success.
337
338 ----------------------- file_lock_operations ------------------------------
339 prototypes:
340 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
341 void (*fl_release_private)(struct file_lock *);
342
343
344 locking rules:
345 file_lock_lock may block
346 fl_copy_lock: yes no
347 fl_release_private: maybe no
348
349 ----------------------- lock_manager_operations ---------------------------
350 prototypes:
351 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
352 void (*lm_notify)(struct file_lock *); /* unblock callback */
353 int (*lm_grant)(struct file_lock *, struct file_lock *, int);
354 void (*lm_break)(struct file_lock *); /* break_lease callback */
355 int (*lm_change)(struct file_lock **, int);
356
357 locking rules:
358 file_lock_lock may block
359 lm_compare_owner: yes no
360 lm_notify: yes no
361 lm_grant: no no
362 lm_break: yes no
363 lm_change yes no
364
365 --------------------------- buffer_head -----------------------------------
366 prototypes:
367 void (*b_end_io)(struct buffer_head *bh, int uptodate);
368
369 locking rules:
370 called from interrupts. In other words, extreme care is needed here.
371 bh is locked, but that's all warranties we have here. Currently only RAID1,
372 highmem, fs/buffer.c, and fs/ntfs/aops.c are providing these. Block devices
373 call this method upon the IO completion.
374
375 --------------------------- block_device_operations -----------------------
376 prototypes:
377 int (*open) (struct block_device *, fmode_t);
378 int (*release) (struct gendisk *, fmode_t);
379 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
380 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
381 int (*direct_access) (struct block_device *, sector_t, void **, unsigned long *);
382 int (*media_changed) (struct gendisk *);
383 void (*unlock_native_capacity) (struct gendisk *);
384 int (*revalidate_disk) (struct gendisk *);
385 int (*getgeo)(struct block_device *, struct hd_geometry *);
386 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
387
388 locking rules:
389 bd_mutex
390 open: yes
391 release: yes
392 ioctl: no
393 compat_ioctl: no
394 direct_access: no
395 media_changed: no
396 unlock_native_capacity: no
397 revalidate_disk: no
398 getgeo: no
399 swap_slot_free_notify: no (see below)
400
401 media_changed, unlock_native_capacity and revalidate_disk are called only from
402 check_disk_change().
403
404 swap_slot_free_notify is called with swap_lock and sometimes the page lock
405 held.
406
407
408 --------------------------- file_operations -------------------------------
409 prototypes:
410 loff_t (*llseek) (struct file *, loff_t, int);
411 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
412 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
413 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
414 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
415 int (*readdir) (struct file *, void *, filldir_t);
416 unsigned int (*poll) (struct file *, struct poll_table_struct *);
417 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
418 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
419 int (*mmap) (struct file *, struct vm_area_struct *);
420 int (*open) (struct inode *, struct file *);
421 int (*flush) (struct file *);
422 int (*release) (struct inode *, struct file *);
423 int (*fsync) (struct file *, loff_t start, loff_t end, int datasync);
424 int (*aio_fsync) (struct kiocb *, int datasync);
425 int (*fasync) (int, struct file *, int);
426 int (*lock) (struct file *, int, struct file_lock *);
427 ssize_t (*readv) (struct file *, const struct iovec *, unsigned long,
428 loff_t *);
429 ssize_t (*writev) (struct file *, const struct iovec *, unsigned long,
430 loff_t *);
431 ssize_t (*sendfile) (struct file *, loff_t *, size_t, read_actor_t,
432 void __user *);
433 ssize_t (*sendpage) (struct file *, struct page *, int, size_t,
434 loff_t *, int);
435 unsigned long (*get_unmapped_area)(struct file *, unsigned long,
436 unsigned long, unsigned long, unsigned long);
437 int (*check_flags)(int);
438 int (*flock) (struct file *, int, struct file_lock *);
439 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *,
440 size_t, unsigned int);
441 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *,
442 size_t, unsigned int);
443 int (*setlease)(struct file *, long, struct file_lock **);
444 long (*fallocate)(struct file *, int, loff_t, loff_t);
445 };
446
447 locking rules:
448 All may block except for ->setlease.
449 No VFS locks held on entry except for ->setlease.
450
451 ->setlease has the file_list_lock held and must not sleep.
452
453 ->llseek() locking has moved from llseek to the individual llseek
454 implementations. If your fs is not using generic_file_llseek, you
455 need to acquire and release the appropriate locks in your ->llseek().
456 For many filesystems, it is probably safe to acquire the inode
457 mutex or just to use i_size_read() instead.
458 Note: this does not protect the file->f_pos against concurrent modifications
459 since this is something the userspace has to take care about.
460
461 ->fasync() is responsible for maintaining the FASYNC bit in filp->f_flags.
462 Most instances call fasync_helper(), which does that maintenance, so it's
463 not normally something one needs to worry about. Return values > 0 will be
464 mapped to zero in the VFS layer.
465
466 ->readdir() and ->ioctl() on directories must be changed. Ideally we would
467 move ->readdir() to inode_operations and use a separate method for directory
468 ->ioctl() or kill the latter completely. One of the problems is that for
469 anything that resembles union-mount we won't have a struct file for all
470 components. And there are other reasons why the current interface is a mess...
471
472 ->read on directories probably must go away - we should just enforce -EISDIR
473 in sys_read() and friends.
474
475 --------------------------- dquot_operations -------------------------------
476 prototypes:
477 int (*write_dquot) (struct dquot *);
478 int (*acquire_dquot) (struct dquot *);
479 int (*release_dquot) (struct dquot *);
480 int (*mark_dirty) (struct dquot *);
481 int (*write_info) (struct super_block *, int);
482
483 These operations are intended to be more or less wrapping functions that ensure
484 a proper locking wrt the filesystem and call the generic quota operations.
485
486 What filesystem should expect from the generic quota functions:
487
488 FS recursion Held locks when called
489 write_dquot: yes dqonoff_sem or dqptr_sem
490 acquire_dquot: yes dqonoff_sem or dqptr_sem
491 release_dquot: yes dqonoff_sem or dqptr_sem
492 mark_dirty: no -
493 write_info: yes dqonoff_sem
494
495 FS recursion means calling ->quota_read() and ->quota_write() from superblock
496 operations.
497
498 More details about quota locking can be found in fs/dquot.c.
499
500 --------------------------- vm_operations_struct -----------------------------
501 prototypes:
502 void (*open)(struct vm_area_struct*);
503 void (*close)(struct vm_area_struct*);
504 int (*fault)(struct vm_area_struct*, struct vm_fault *);
505 int (*page_mkwrite)(struct vm_area_struct *, struct vm_fault *);
506 int (*access)(struct vm_area_struct *, unsigned long, void*, int, int);
507
508 locking rules:
509 mmap_sem PageLocked(page)
510 open: yes
511 close: yes
512 fault: yes can return with page locked
513 page_mkwrite: yes can return with page locked
514 access: yes
515
516 ->fault() is called when a previously not present pte is about
517 to be faulted in. The filesystem must find and return the page associated
518 with the passed in "pgoff" in the vm_fault structure. If it is possible that
519 the page may be truncated and/or invalidated, then the filesystem must lock
520 the page, then ensure it is not already truncated (the page lock will block
521 subsequent truncate), and then return with VM_FAULT_LOCKED, and the page
522 locked. The VM will unlock the page.
523
524 ->page_mkwrite() is called when a previously read-only pte is
525 about to become writeable. The filesystem again must ensure that there are
526 no truncate/invalidate races, and then return with the page locked. If
527 the page has been truncated, the filesystem should not look up a new page
528 like the ->fault() handler, but simply return with VM_FAULT_NOPAGE, which
529 will cause the VM to retry the fault.
530
531 ->access() is called when get_user_pages() fails in
532 acces_process_vm(), typically used to debug a process through
533 /proc/pid/mem or ptrace. This function is needed only for
534 VM_IO | VM_PFNMAP VMAs.
535
536 ================================================================================
537 Dubious stuff
538
539 (if you break something or notice that it is broken and do not fix it yourself
540 - at least put it here)
This page took 0.061111 seconds and 4 git commands to generate.