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 *, int);
13 int (*d_hash) (struct dentry *, struct qstr *);
14 int (*d_compare) (struct dentry *, struct qstr *, struct qstr *);
15 int (*d_delete)(struct dentry *);
16 void (*d_release)(struct dentry *);
17 void (*d_iput)(struct dentry *, struct inode *);
18 char *(*d_dname)((struct dentry *dentry, char *buffer, int buflen);
22 dcache_lock rename_lock ->d_lock may block
23 d_revalidate: no no no yes
25 d_compare: no yes no no
26 d_delete: yes no yes no
27 d_release: no no no yes
31 --------------------------- inode_operations ---------------------------
33 int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
34 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameid
36 int (*link) (struct dentry *,struct inode *,struct dentry *);
37 int (*unlink) (struct inode *,struct dentry *);
38 int (*symlink) (struct inode *,struct dentry *,const char *);
39 int (*mkdir) (struct inode *,struct dentry *,int);
40 int (*rmdir) (struct inode *,struct dentry *);
41 int (*mknod) (struct inode *,struct dentry *,int,dev_t);
42 int (*rename) (struct inode *, struct dentry *,
43 struct inode *, struct dentry *);
44 int (*readlink) (struct dentry *, char __user *,int);
45 int (*follow_link) (struct dentry *, struct nameidata *);
46 void (*truncate) (struct inode *);
47 int (*permission) (struct inode *, int, struct nameidata *);
48 int (*setattr) (struct dentry *, struct iattr *);
49 int (*getattr) (struct vfsmount *, struct dentry *, struct kstat *);
50 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
51 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
52 ssize_t (*listxattr) (struct dentry *, char *, size_t);
53 int (*removexattr) (struct dentry *, const char *);
56 all may block, none have BKL
65 rmdir: yes (both) (see below)
66 rename: yes (all) (see below)
69 truncate: yes (see below)
77 Additionally, ->rmdir(), ->unlink() and ->rename() have ->i_mutex on
79 cross-directory ->rename() has (per-superblock) ->s_vfs_rename_sem.
80 ->truncate() is never called directly - it's a callback, not a
81 method. It's called by vmtruncate() - library function normally used by
82 ->setattr(). Locking information above applies to that call (i.e. is
83 inherited from ->setattr() - vmtruncate() is used when ATTR_SIZE had been
86 See Documentation/filesystems/directory-locking for more detailed discussion
87 of the locking scheme for directory operations.
89 --------------------------- super_operations ---------------------------
91 struct inode *(*alloc_inode)(struct super_block *sb);
92 void (*destroy_inode)(struct inode *);
93 void (*dirty_inode) (struct inode *);
94 int (*write_inode) (struct inode *, int);
95 int (*drop_inode) (struct inode *);
96 void (*evict_inode) (struct inode *);
97 void (*put_super) (struct super_block *);
98 void (*write_super) (struct super_block *);
99 int (*sync_fs)(struct super_block *sb, int wait);
100 int (*freeze_fs) (struct super_block *);
101 int (*unfreeze_fs) (struct super_block *);
102 int (*statfs) (struct dentry *, struct kstatfs *);
103 int (*remount_fs) (struct super_block *, int *, char *);
104 void (*umount_begin) (struct super_block *);
105 int (*show_options)(struct seq_file *, struct vfsmount *);
106 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
107 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
110 All may block [not true, see below]
115 dirty_inode: (must not sleep)
117 drop_inode: !!!inode_lock!!!
124 statfs: maybe(read) (see below)
127 show_options: no (namespace_sem)
128 quota_read: no (see below)
129 quota_write: no (see below)
131 ->statfs() has s_umount (shared) when called by ustat(2) (native or
132 compat), but that's an accident of bad API; s_umount is used to pin
133 the superblock down when we only have dev_t given us by userland to
134 identify the superblock. Everything else (statfs(), fstatfs(), etc.)
135 doesn't hold it when calling ->statfs() - superblock is pinned down
136 by resolving the pathname passed to syscall.
137 ->quota_read() and ->quota_write() functions are both guaranteed to
138 be the only ones operating on the quota file by the quota code (via
139 dqio_sem) (unless an admin really wants to screw up something and
140 writes to quota files with quotas on). For other details about locking
141 see also dquot_operations section.
143 --------------------------- file_system_type ---------------------------
145 int (*get_sb) (struct file_system_type *, int,
146 const char *, void *, struct vfsmount *);
147 void (*kill_sb) (struct super_block *);
153 ->get_sb() returns error or 0 with locked superblock attached to the vfsmount
154 (exclusive on ->s_umount).
155 ->kill_sb() takes a write-locked superblock, does all shutdown work on it,
156 unlocks and drops the reference.
158 --------------------------- address_space_operations --------------------------
160 int (*writepage)(struct page *page, struct writeback_control *wbc);
161 int (*readpage)(struct file *, struct page *);
162 int (*sync_page)(struct page *);
163 int (*writepages)(struct address_space *, struct writeback_control *);
164 int (*set_page_dirty)(struct page *page);
165 int (*readpages)(struct file *filp, struct address_space *mapping,
166 struct list_head *pages, unsigned nr_pages);
167 int (*write_begin)(struct file *, struct address_space *mapping,
168 loff_t pos, unsigned len, unsigned flags,
169 struct page **pagep, void **fsdata);
170 int (*write_end)(struct file *, struct address_space *mapping,
171 loff_t pos, unsigned len, unsigned copied,
172 struct page *page, void *fsdata);
173 sector_t (*bmap)(struct address_space *, sector_t);
174 int (*invalidatepage) (struct page *, unsigned long);
175 int (*releasepage) (struct page *, int);
176 void (*freepage)(struct page *);
177 int (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
178 loff_t offset, unsigned long nr_segs);
179 int (*launder_page) (struct page *);
182 All except set_page_dirty and freepage may block
184 BKL PageLocked(page) i_mutex
185 writepage: no yes, unlocks (see below)
186 readpage: no yes, unlocks
191 write_begin: no locks the page yes
192 write_end: no yes, unlocks yes
193 perform_write: no n/a yes
195 invalidatepage: no yes
201 ->write_begin(), ->write_end(), ->sync_page() and ->readpage()
202 may be called from the request handler (/dev/loop).
204 ->readpage() unlocks the page, either synchronously or via I/O
207 ->readpages() populates the pagecache with the passed pages and starts
208 I/O against them. They come unlocked upon I/O completion.
210 ->writepage() is used for two purposes: for "memory cleansing" and for
211 "sync". These are quite different operations and the behaviour may differ
212 depending upon the mode.
214 If writepage is called for sync (wbc->sync_mode != WBC_SYNC_NONE) then
215 it *must* start I/O against the page, even if that would involve
216 blocking on in-progress I/O.
218 If writepage is called for memory cleansing (sync_mode ==
219 WBC_SYNC_NONE) then its role is to get as much writeout underway as
220 possible. So writepage should try to avoid blocking against
221 currently-in-progress I/O.
223 If the filesystem is not called for "sync" and it determines that it
224 would need to block against in-progress I/O to be able to start new I/O
225 against the page the filesystem should redirty the page with
226 redirty_page_for_writepage(), then unlock the page and return zero.
227 This may also be done to avoid internal deadlocks, but rarely.
229 If the filesystem is called for sync then it must wait on any
230 in-progress I/O and then start new I/O.
232 The filesystem should unlock the page synchronously, before returning to the
233 caller, unless ->writepage() returns special WRITEPAGE_ACTIVATE
234 value. WRITEPAGE_ACTIVATE means that page cannot really be written out
235 currently, and VM should stop calling ->writepage() on this page for some
236 time. VM does this by moving page to the head of the active list, hence the
239 Unless the filesystem is going to redirty_page_for_writepage(), unlock the page
240 and return zero, writepage *must* run set_page_writeback() against the page,
241 followed by unlocking it. Once set_page_writeback() has been run against the
242 page, write I/O can be submitted and the write I/O completion handler must run
243 end_page_writeback() once the I/O is complete. If no I/O is submitted, the
244 filesystem must run end_page_writeback() against the page before returning from
247 That is: after 2.5.12, pages which are under writeout are *not* locked. Note,
248 if the filesystem needs the page to be locked during writeout, that is ok, too,
249 the page is allowed to be unlocked at any point in time between the calls to
250 set_page_writeback() and end_page_writeback().
252 Note, failure to run either redirty_page_for_writepage() or the combination of
253 set_page_writeback()/end_page_writeback() on a page submitted to writepage
254 will leave the page itself marked clean but it will be tagged as dirty in the
255 radix tree. This incoherency can lead to all sorts of hard-to-debug problems
256 in the filesystem like having dirty inodes at umount and losing written data.
258 ->sync_page() locking rules are not well-defined - usually it is called
259 with lock on page, but that is not guaranteed. Considering the currently
260 existing instances of this method ->sync_page() itself doesn't look
263 ->writepages() is used for periodic writeback and for syscall-initiated
264 sync operations. The address_space should start I/O against at least
265 *nr_to_write pages. *nr_to_write must be decremented for each page which is
266 written. The address_space implementation may write more (or less) pages
267 than *nr_to_write asks for, but it should try to be reasonably close. If
268 nr_to_write is NULL, all dirty pages must be written.
270 writepages should _only_ write pages which are present on
273 ->set_page_dirty() is called from various places in the kernel
274 when the target page is marked as needing writeback. It may be called
275 under spinlock (it cannot block) and is sometimes called with the page
278 ->bmap() is currently used by legacy ioctl() (FIBMAP) provided by some
279 filesystems and by the swapper. The latter will eventually go away. All
280 instances do not actually need the BKL. Please, keep it that way and don't
283 ->invalidatepage() is called when the filesystem must attempt to drop
284 some or all of the buffers from the page when it is being truncated. It
285 returns zero on success. If ->invalidatepage is zero, the kernel uses
286 block_invalidatepage() instead.
288 ->releasepage() is called when the kernel is about to try to drop the
289 buffers from the page in preparation for freeing it. It returns zero to
290 indicate that the buffers are (or may be) freeable. If ->releasepage is zero,
291 the kernel assumes that the fs has no private interest in the buffers.
293 ->freepage() is called when the kernel is done dropping the page
296 ->launder_page() may be called prior to releasing a page if
297 it is still found to be dirty. It returns zero if the page was successfully
298 cleaned, or an error value if not. Note that in order to prevent the page
299 getting mapped back in and redirtied, it needs to be kept locked
300 across the entire operation.
302 Note: currently almost all instances of address_space methods are
303 using BKL for internal serialization and that's one of the worst sources
304 of contention. Normally they are calling library functions (in fs/buffer.c)
305 and pass foo_get_block() as a callback (on local block-based filesystems,
306 indeed). BKL is not needed for library stuff and is usually taken by
307 foo_get_block(). It's an overkill, since block bitmaps can be protected by
308 internal fs locking and real critical areas are much smaller than the areas
309 filesystems protect now.
311 ----------------------- file_lock_operations ------------------------------
313 void (*fl_insert)(struct file_lock *); /* lock insertion callback */
314 void (*fl_remove)(struct file_lock *); /* lock removal callback */
315 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
316 void (*fl_release_private)(struct file_lock *);
324 fl_release_private: yes yes
326 ----------------------- lock_manager_operations ---------------------------
328 int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
329 void (*fl_notify)(struct file_lock *); /* unblock callback */
330 void (*fl_release_private)(struct file_lock *);
331 void (*fl_break)(struct file_lock *); /* break_lease callback */
335 fl_compare_owner: yes no
337 fl_release_private: yes yes
340 Currently only NFSD and NLM provide instances of this class. None of the
341 them block. If you have out-of-tree instances - please, show up. Locking
342 in that area will change.
343 --------------------------- buffer_head -----------------------------------
345 void (*b_end_io)(struct buffer_head *bh, int uptodate);
348 called from interrupts. In other words, extreme care is needed here.
349 bh is locked, but that's all warranties we have here. Currently only RAID1,
350 highmem, fs/buffer.c, and fs/ntfs/aops.c are providing these. Block devices
351 call this method upon the IO completion.
353 --------------------------- block_device_operations -----------------------
355 int (*open) (struct block_device *, fmode_t);
356 int (*release) (struct gendisk *, fmode_t);
357 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
358 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
359 int (*direct_access) (struct block_device *, sector_t, void **, unsigned long *);
360 int (*media_changed) (struct gendisk *);
361 void (*unlock_native_capacity) (struct gendisk *);
362 int (*revalidate_disk) (struct gendisk *);
363 int (*getgeo)(struct block_device *, struct hd_geometry *);
364 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
374 unlock_native_capacity: no no
375 revalidate_disk: no no
377 swap_slot_free_notify: no no (see below)
379 media_changed, unlock_native_capacity and revalidate_disk are called only from
382 swap_slot_free_notify is called with swap_lock and sometimes the page lock
386 --------------------------- file_operations -------------------------------
388 loff_t (*llseek) (struct file *, loff_t, int);
389 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
390 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
391 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
392 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
393 int (*readdir) (struct file *, void *, filldir_t);
394 unsigned int (*poll) (struct file *, struct poll_table_struct *);
395 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
396 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
397 int (*mmap) (struct file *, struct vm_area_struct *);
398 int (*open) (struct inode *, struct file *);
399 int (*flush) (struct file *);
400 int (*release) (struct inode *, struct file *);
401 int (*fsync) (struct file *, int datasync);
402 int (*aio_fsync) (struct kiocb *, int datasync);
403 int (*fasync) (int, struct file *, int);
404 int (*lock) (struct file *, int, struct file_lock *);
405 ssize_t (*readv) (struct file *, const struct iovec *, unsigned long,
407 ssize_t (*writev) (struct file *, const struct iovec *, unsigned long,
409 ssize_t (*sendfile) (struct file *, loff_t *, size_t, read_actor_t,
411 ssize_t (*sendpage) (struct file *, struct page *, int, size_t,
413 unsigned long (*get_unmapped_area)(struct file *, unsigned long,
414 unsigned long, unsigned long, unsigned long);
415 int (*check_flags)(int);
421 llseek: no (see below)
434 fsync: no (see below)
442 get_unmapped_area: no
445 ->llseek() locking has moved from llseek to the individual llseek
446 implementations. If your fs is not using generic_file_llseek, you
447 need to acquire and release the appropriate locks in your ->llseek().
448 For many filesystems, it is probably safe to acquire the inode
449 mutex or just to use i_size_read() instead.
450 Note: this does not protect the file->f_pos against concurrent modifications
451 since this is something the userspace has to take care about.
453 Note: ext2_release() was *the* source of contention on fs-intensive
454 loads and dropping BKL on ->release() helps to get rid of that (we still
455 grab BKL for cases when we close a file that had been opened r/w, but that
456 can and should be done using the internal locking with smaller critical areas).
457 Current worst offender is ext2_get_block()...
459 ->fasync() is called without BKL protection, and is responsible for
460 maintaining the FASYNC bit in filp->f_flags. Most instances call
461 fasync_helper(), which does that maintenance, so it's not normally
462 something one needs to worry about. Return values > 0 will be mapped to
463 zero in the VFS layer.
465 ->readdir() and ->ioctl() on directories must be changed. Ideally we would
466 move ->readdir() to inode_operations and use a separate method for directory
467 ->ioctl() or kill the latter completely. One of the problems is that for
468 anything that resembles union-mount we won't have a struct file for all
469 components. And there are other reasons why the current interface is a mess...
471 ->read on directories probably must go away - we should just enforce -EISDIR
472 in sys_read() and friends.
474 ->fsync() has i_mutex on inode.
476 --------------------------- dquot_operations -------------------------------
478 int (*write_dquot) (struct dquot *);
479 int (*acquire_dquot) (struct dquot *);
480 int (*release_dquot) (struct dquot *);
481 int (*mark_dirty) (struct dquot *);
482 int (*write_info) (struct super_block *, int);
484 These operations are intended to be more or less wrapping functions that ensure
485 a proper locking wrt the filesystem and call the generic quota operations.
487 What filesystem should expect from the generic quota functions:
489 FS recursion Held locks when called
490 write_dquot: yes dqonoff_sem or dqptr_sem
491 acquire_dquot: yes dqonoff_sem or dqptr_sem
492 release_dquot: yes dqonoff_sem or dqptr_sem
494 write_info: yes dqonoff_sem
496 FS recursion means calling ->quota_read() and ->quota_write() from superblock
499 More details about quota locking can be found in fs/dquot.c.
501 --------------------------- vm_operations_struct -----------------------------
503 void (*open)(struct vm_area_struct*);
504 void (*close)(struct vm_area_struct*);
505 int (*fault)(struct vm_area_struct*, struct vm_fault *);
506 int (*page_mkwrite)(struct vm_area_struct *, struct vm_fault *);
507 int (*access)(struct vm_area_struct *, unsigned long, void*, int, int);
510 BKL mmap_sem PageLocked(page)
513 fault: no yes can return with page locked
514 page_mkwrite: no yes can return with page locked
517 ->fault() is called when a previously not present pte is about
518 to be faulted in. The filesystem must find and return the page associated
519 with the passed in "pgoff" in the vm_fault structure. If it is possible that
520 the page may be truncated and/or invalidated, then the filesystem must lock
521 the page, then ensure it is not already truncated (the page lock will block
522 subsequent truncate), and then return with VM_FAULT_LOCKED, and the page
523 locked. The VM will unlock the page.
525 ->page_mkwrite() is called when a previously read-only pte is
526 about to become writeable. The filesystem again must ensure that there are
527 no truncate/invalidate races, and then return with the page locked. If
528 the page has been truncated, the filesystem should not look up a new page
529 like the ->fault() handler, but simply return with VM_FAULT_NOPAGE, which
530 will cause the VM to retry the fault.
532 ->access() is called when get_user_pages() fails in
533 acces_process_vm(), typically used to debug a process through
534 /proc/pid/mem or ptrace. This function is needed only for
535 VM_IO | VM_PFNMAP VMAs.
537 ================================================================================
540 (if you break something or notice that it is broken and do not fix it yourself
541 - at least put it here)
543 ipc/shm.c::shm_delete() - may need BKL.
544 ->read() and ->write() in many drivers are (probably) missing BKL.