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 *);
21 dcache_lock rename_lock ->d_lock may block
22 d_revalidate: no no no yes
24 d_compare: no yes no no
25 d_delete: yes no yes no
26 d_release: no no no yes
29 --------------------------- inode_operations ---------------------------
31 int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
32 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameid
34 int (*link) (struct dentry *,struct inode *,struct dentry *);
35 int (*unlink) (struct inode *,struct dentry *);
36 int (*symlink) (struct inode *,struct dentry *,const char *);
37 int (*mkdir) (struct inode *,struct dentry *,int);
38 int (*rmdir) (struct inode *,struct dentry *);
39 int (*mknod) (struct inode *,struct dentry *,int,dev_t);
40 int (*rename) (struct inode *, struct dentry *,
41 struct inode *, struct dentry *);
42 int (*readlink) (struct dentry *, char __user *,int);
43 int (*follow_link) (struct dentry *, struct nameidata *);
44 void (*truncate) (struct inode *);
45 int (*permission) (struct inode *, int, struct nameidata *);
46 int (*setattr) (struct dentry *, struct iattr *);
47 int (*getattr) (struct vfsmount *, struct dentry *, struct kstat *);
48 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
49 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
50 ssize_t (*listxattr) (struct dentry *, char *, size_t);
51 int (*removexattr) (struct dentry *, const char *);
54 all may block, none have BKL
63 rmdir: yes (both) (see below)
64 rename: yes (all) (see below)
67 truncate: yes (see below)
75 Additionally, ->rmdir(), ->unlink() and ->rename() have ->i_sem on
77 cross-directory ->rename() has (per-superblock) ->s_vfs_rename_sem.
78 ->truncate() is never called directly - it's a callback, not a
79 method. It's called by vmtruncate() - library function normally used by
80 ->setattr(). Locking information above applies to that call (i.e. is
81 inherited from ->setattr() - vmtruncate() is used when ATTR_SIZE had been
84 See Documentation/filesystems/directory-locking for more detailed discussion
85 of the locking scheme for directory operations.
87 --------------------------- super_operations ---------------------------
89 struct inode *(*alloc_inode)(struct super_block *sb);
90 void (*destroy_inode)(struct inode *);
91 void (*read_inode) (struct inode *);
92 void (*dirty_inode) (struct inode *);
93 int (*write_inode) (struct inode *, int);
94 void (*put_inode) (struct inode *);
95 void (*drop_inode) (struct inode *);
96 void (*delete_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 void (*write_super_lockfs) (struct super_block *);
101 void (*unlockfs) (struct super_block *);
102 int (*statfs) (struct super_block *, struct kstatfs *);
103 int (*remount_fs) (struct super_block *, int *, char *);
104 void (*clear_inode) (struct inode *);
105 void (*umount_begin) (struct super_block *);
106 int (*show_options)(struct seq_file *, struct vfsmount *);
111 alloc_inode: no no no
113 read_inode: no (see below)
114 dirty_inode: no (must not sleep)
117 drop_inode: no !!!inode_lock!!!
119 put_super: yes yes no
120 write_super: no yes read
122 write_super_lockfs: ?
125 remount_fs: no yes maybe (see below)
127 umount_begin: yes no no
128 show_options: no (vfsmount->sem)
130 ->read_inode() is not a method - it's a callback used in iget().
131 ->remount_fs() will have the s_umount lock if it's already mounted.
132 When called from get_sb_single, it does NOT have the s_umount lock.
134 --------------------------- file_system_type ---------------------------
136 struct super_block *(*get_sb) (struct file_system_type *, int,
137 const char *, void *);
138 void (*kill_sb) (struct super_block *);
144 ->get_sb() returns error or a locked superblock (exclusive on ->s_umount).
145 ->kill_sb() takes a write-locked superblock, does all shutdown work on it,
146 unlocks and drops the reference.
148 --------------------------- address_space_operations --------------------------
150 int (*writepage)(struct page *page, struct writeback_control *wbc);
151 int (*readpage)(struct file *, struct page *);
152 int (*sync_page)(struct page *);
153 int (*writepages)(struct address_space *, struct writeback_control *);
154 int (*set_page_dirty)(struct page *page);
155 int (*readpages)(struct file *filp, struct address_space *mapping,
156 struct list_head *pages, unsigned nr_pages);
157 int (*prepare_write)(struct file *, struct page *, unsigned, unsigned);
158 int (*commit_write)(struct file *, struct page *, unsigned, unsigned);
159 sector_t (*bmap)(struct address_space *, sector_t);
160 int (*invalidatepage) (struct page *, unsigned long);
161 int (*releasepage) (struct page *, int);
162 int (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
163 loff_t offset, unsigned long nr_segs);
166 All except set_page_dirty may block
169 writepage: no yes, unlocks (see below)
170 readpage: no yes, unlocks
175 prepare_write: no yes
178 invalidatepage: no yes
182 ->prepare_write(), ->commit_write(), ->sync_page() and ->readpage()
183 may be called from the request handler (/dev/loop).
185 ->readpage() unlocks the page, either synchronously or via I/O
188 ->readpages() populates the pagecache with the passed pages and starts
189 I/O against them. They come unlocked upon I/O completion.
191 ->writepage() is used for two purposes: for "memory cleansing" and for
192 "sync". These are quite different operations and the behaviour may differ
193 depending upon the mode.
195 If writepage is called for sync (wbc->sync_mode != WBC_SYNC_NONE) then
196 it *must* start I/O against the page, even if that would involve
197 blocking on in-progress I/O.
199 If writepage is called for memory cleansing (sync_mode ==
200 WBC_SYNC_NONE) then its role is to get as much writeout underway as
201 possible. So writepage should try to avoid blocking against
202 currently-in-progress I/O.
204 If the filesystem is not called for "sync" and it determines that it
205 would need to block against in-progress I/O to be able to start new I/O
206 against the page the filesystem should redirty the page with
207 redirty_page_for_writepage(), then unlock the page and return zero.
208 This may also be done to avoid internal deadlocks, but rarely.
210 If the filesytem is called for sync then it must wait on any
211 in-progress I/O and then start new I/O.
213 The filesystem should unlock the page synchronously, before returning
216 Unless the filesystem is going to redirty_page_for_writepage(), unlock the page
217 and return zero, writepage *must* run set_page_writeback() against the page,
218 followed by unlocking it. Once set_page_writeback() has been run against the
219 page, write I/O can be submitted and the write I/O completion handler must run
220 end_page_writeback() once the I/O is complete. If no I/O is submitted, the
221 filesystem must run end_page_writeback() against the page before returning from
224 That is: after 2.5.12, pages which are under writeout are *not* locked. Note,
225 if the filesystem needs the page to be locked during writeout, that is ok, too,
226 the page is allowed to be unlocked at any point in time between the calls to
227 set_page_writeback() and end_page_writeback().
229 Note, failure to run either redirty_page_for_writepage() or the combination of
230 set_page_writeback()/end_page_writeback() on a page submitted to writepage
231 will leave the page itself marked clean but it will be tagged as dirty in the
232 radix tree. This incoherency can lead to all sorts of hard-to-debug problems
233 in the filesystem like having dirty inodes at umount and losing written data.
235 ->sync_page() locking rules are not well-defined - usually it is called
236 with lock on page, but that is not guaranteed. Considering the currently
237 existing instances of this method ->sync_page() itself doesn't look
240 ->writepages() is used for periodic writeback and for syscall-initiated
241 sync operations. The address_space should start I/O against at least
242 *nr_to_write pages. *nr_to_write must be decremented for each page which is
243 written. The address_space implementation may write more (or less) pages
244 than *nr_to_write asks for, but it should try to be reasonably close. If
245 nr_to_write is NULL, all dirty pages must be written.
247 writepages should _only_ write pages which are present on
250 ->set_page_dirty() is called from various places in the kernel
251 when the target page is marked as needing writeback. It may be called
252 under spinlock (it cannot block) and is sometimes called with the page
255 ->bmap() is currently used by legacy ioctl() (FIBMAP) provided by some
256 filesystems and by the swapper. The latter will eventually go away. All
257 instances do not actually need the BKL. Please, keep it that way and don't
260 ->invalidatepage() is called when the filesystem must attempt to drop
261 some or all of the buffers from the page when it is being truncated. It
262 returns zero on success. If ->invalidatepage is zero, the kernel uses
263 block_invalidatepage() instead.
265 ->releasepage() is called when the kernel is about to try to drop the
266 buffers from the page in preparation for freeing it. It returns zero to
267 indicate that the buffers are (or may be) freeable. If ->releasepage is zero,
268 the kernel assumes that the fs has no private interest in the buffers.
270 Note: currently almost all instances of address_space methods are
271 using BKL for internal serialization and that's one of the worst sources
272 of contention. Normally they are calling library functions (in fs/buffer.c)
273 and pass foo_get_block() as a callback (on local block-based filesystems,
274 indeed). BKL is not needed for library stuff and is usually taken by
275 foo_get_block(). It's an overkill, since block bitmaps can be protected by
276 internal fs locking and real critical areas are much smaller than the areas
277 filesystems protect now.
279 ----------------------- file_lock_operations ------------------------------
281 void (*fl_insert)(struct file_lock *); /* lock insertion callback */
282 void (*fl_remove)(struct file_lock *); /* lock removal callback */
283 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
284 void (*fl_release_private)(struct file_lock *);
292 fl_release_private: yes yes
294 ----------------------- lock_manager_operations ---------------------------
296 int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
297 void (*fl_notify)(struct file_lock *); /* unblock callback */
301 fl_compare_owner: yes no
304 Currently only NLM provides instances of this class. None of the
305 them block. If you have out-of-tree instances - please, show up. Locking
306 in that area will change.
307 --------------------------- buffer_head -----------------------------------
309 void (*b_end_io)(struct buffer_head *bh, int uptodate);
312 called from interrupts. In other words, extreme care is needed here.
313 bh is locked, but that's all warranties we have here. Currently only RAID1,
314 highmem and fs/buffer.c are providing these. Block devices call this method
315 upon the IO completion.
317 --------------------------- block_device_operations -----------------------
319 int (*open) (struct inode *, struct file *);
320 int (*release) (struct inode *, struct file *);
321 int (*ioctl) (struct inode *, struct file *, unsigned, unsigned long);
322 int (*media_changed) (struct gendisk *);
323 int (*revalidate_disk) (struct gendisk *);
331 revalidate_disk: no no
333 The last two are called only from check_disk_change().
335 --------------------------- file_operations -------------------------------
337 loff_t (*llseek) (struct file *, loff_t, int);
338 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
339 ssize_t (*aio_read) (struct kiocb *, char __user *, size_t, loff_t);
340 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
341 ssize_t (*aio_write) (struct kiocb *, const char __user *, size_t,
343 int (*readdir) (struct file *, void *, filldir_t);
344 unsigned int (*poll) (struct file *, struct poll_table_struct *);
345 int (*ioctl) (struct inode *, struct file *, unsigned int,
347 int (*mmap) (struct file *, struct vm_area_struct *);
348 int (*open) (struct inode *, struct file *);
349 int (*flush) (struct file *);
350 int (*release) (struct inode *, struct file *);
351 int (*fsync) (struct file *, struct dentry *, int datasync);
352 int (*aio_fsync) (struct kiocb *, int datasync);
353 int (*fasync) (int, struct file *, int);
354 int (*lock) (struct file *, int, struct file_lock *);
355 ssize_t (*readv) (struct file *, const struct iovec *, unsigned long,
357 ssize_t (*writev) (struct file *, const struct iovec *, unsigned long,
359 ssize_t (*sendfile) (struct file *, loff_t *, size_t, read_actor_t,
361 ssize_t (*sendpage) (struct file *, struct page *, int, size_t,
363 unsigned long (*get_unmapped_area)(struct file *, unsigned long,
364 unsigned long, unsigned long, unsigned long);
365 int (*check_flags)(int);
366 int (*dir_notify)(struct file *, unsigned long);
370 All except ->poll() may block.
372 llseek: no (see below)
379 ioctl: yes (see below)
381 open: maybe (see below)
384 fsync: no (see below)
386 fasync: yes (see below)
392 get_unmapped_area: no
396 ->llseek() locking has moved from llseek to the individual llseek
397 implementations. If your fs is not using generic_file_llseek, you
398 need to acquire and release the appropriate locks in your ->llseek().
399 For many filesystems, it is probably safe to acquire the inode
400 semaphore. Note some filesystems (i.e. remote ones) provide no
401 protection for i_size so you will need to use the BKL.
403 ->open() locking is in-transit: big lock partially moved into the methods.
404 The only exception is ->open() in the instances of file_operations that never
405 end up in ->i_fop/->proc_fops, i.e. ones that belong to character devices
406 (chrdev_open() takes lock before replacing ->f_op and calling the secondary
407 method. As soon as we fix the handling of module reference counters all
408 instances of ->open() will be called without the BKL.
410 Note: ext2_release() was *the* source of contention on fs-intensive
411 loads and dropping BKL on ->release() helps to get rid of that (we still
412 grab BKL for cases when we close a file that had been opened r/w, but that
413 can and should be done using the internal locking with smaller critical areas).
414 Current worst offender is ext2_get_block()...
416 ->fasync() is a mess. This area needs a big cleanup and that will probably
419 ->readdir() and ->ioctl() on directories must be changed. Ideally we would
420 move ->readdir() to inode_operations and use a separate method for directory
421 ->ioctl() or kill the latter completely. One of the problems is that for
422 anything that resembles union-mount we won't have a struct file for all
423 components. And there are other reasons why the current interface is a mess...
425 ->read on directories probably must go away - we should just enforce -EISDIR
426 in sys_read() and friends.
428 ->fsync() has i_sem on inode.
430 --------------------------- dquot_operations -------------------------------
432 void (*initialize) (struct inode *, short);
433 void (*drop) (struct inode *);
434 int (*alloc_block) (const struct inode *, unsigned long, char);
435 int (*alloc_inode) (const struct inode *, unsigned long);
436 void (*free_block) (const struct inode *, unsigned long);
437 void (*free_inode) (const struct inode *, unsigned long);
438 int (*transfer) (struct dentry *, struct iattr *);
450 --------------------------- vm_operations_struct -----------------------------
452 void (*open)(struct vm_area_struct*);
453 void (*close)(struct vm_area_struct*);
454 struct page *(*nopage)(struct vm_area_struct*, unsigned long, int *);
462 ================================================================================
465 (if you break something or notice that it is broken and do not fix it yourself
466 - at least put it here)
468 ipc/shm.c::shm_delete() - may need BKL.
469 ->read() and ->write() in many drivers are (probably) missing BKL.
470 drivers/sgi/char/graphics.c::sgi_graphics_nopage() - may need BKL.