Loading...
1 The text below describes the locking rules for VFS-related methods.
2It is (believed to be) up-to-date. *Please*, if you change anything in
3prototypes or locking protocols - update this file. And update the relevant
4instances in the tree, don't leave that to maintainers of filesystems/devices/
5etc. At the very least, put the list of dubious cases in the end of this file.
6Don't turn it into log - maintainers of out-of-the-tree code are supposed to
7be able to use diff(1).
8 Thing currently missing here: socket operations. Alexey?
9
10--------------------------- dentry_operations --------------------------
11prototypes:
12 int (*d_revalidate)(struct dentry *, struct nameidata *);
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
25locking rules:
26 rename_lock ->d_lock may block rcu-walk
27d_revalidate: no no yes (ref-walk) maybe
28d_hash no no no maybe
29d_compare: yes no no maybe
30d_delete: no yes no no
31d_release: no no yes no
32d_iput: no no yes no
33d_dname: no no no no
34d_automount: no no yes no
35d_manage: no no yes (ref-walk) maybe
36
37--------------------------- inode_operations ---------------------------
38prototypes:
39 int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
40 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameid
41ata *);
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 *,int);
46 int (*rmdir) (struct inode *,struct dentry *);
47 int (*mknod) (struct inode *,struct dentry *,int,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 void (*truncate_range)(struct inode *, loff_t, loff_t);
63 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start, u64 len);
64
65locking rules:
66 all may block
67 i_mutex(inode)
68lookup: yes
69create: yes
70link: yes (both)
71mknod: yes
72symlink: yes
73mkdir: yes
74unlink: yes (both)
75rmdir: yes (both) (see below)
76rename: yes (all) (see below)
77readlink: no
78follow_link: no
79put_link: no
80truncate: yes (see below)
81setattr: yes
82permission: no (may not block if called in rcu-walk mode)
83get_acl: no
84getattr: no
85setxattr: yes
86getxattr: no
87listxattr: no
88removexattr: yes
89truncate_range: yes
90fiemap: no
91 Additionally, ->rmdir(), ->unlink() and ->rename() have ->i_mutex on
92victim.
93 cross-directory ->rename() has (per-superblock) ->s_vfs_rename_sem.
94 ->truncate() is never called directly - it's a callback, not a
95method. It's called by vmtruncate() - deprecated library function used by
96->setattr(). Locking information above applies to that call (i.e. is
97inherited from ->setattr() - vmtruncate() is used when ATTR_SIZE had been
98passed).
99
100See Documentation/filesystems/directory-locking for more detailed discussion
101of the locking scheme for directory operations.
102
103--------------------------- super_operations ---------------------------
104prototypes:
105 struct inode *(*alloc_inode)(struct super_block *sb);
106 void (*destroy_inode)(struct inode *);
107 void (*dirty_inode) (struct inode *, int flags);
108 int (*write_inode) (struct inode *, struct writeback_control *wbc);
109 int (*drop_inode) (struct inode *);
110 void (*evict_inode) (struct inode *);
111 void (*put_super) (struct super_block *);
112 void (*write_super) (struct super_block *);
113 int (*sync_fs)(struct super_block *sb, int wait);
114 int (*freeze_fs) (struct super_block *);
115 int (*unfreeze_fs) (struct super_block *);
116 int (*statfs) (struct dentry *, struct kstatfs *);
117 int (*remount_fs) (struct super_block *, int *, char *);
118 void (*umount_begin) (struct super_block *);
119 int (*show_options)(struct seq_file *, struct vfsmount *);
120 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
121 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
122 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
123
124locking rules:
125 All may block [not true, see below]
126 s_umount
127alloc_inode:
128destroy_inode:
129dirty_inode:
130write_inode:
131drop_inode: !!!inode->i_lock!!!
132evict_inode:
133put_super: write
134write_super: read
135sync_fs: read
136freeze_fs: read
137unfreeze_fs: read
138statfs: maybe(read) (see below)
139remount_fs: write
140umount_begin: no
141show_options: no (namespace_sem)
142quota_read: no (see below)
143quota_write: no (see below)
144bdev_try_to_free_page: no (see below)
145
146->statfs() has s_umount (shared) when called by ustat(2) (native or
147compat), but that's an accident of bad API; s_umount is used to pin
148the superblock down when we only have dev_t given us by userland to
149identify the superblock. Everything else (statfs(), fstatfs(), etc.)
150doesn't hold it when calling ->statfs() - superblock is pinned down
151by resolving the pathname passed to syscall.
152->quota_read() and ->quota_write() functions are both guaranteed to
153be the only ones operating on the quota file by the quota code (via
154dqio_sem) (unless an admin really wants to screw up something and
155writes to quota files with quotas on). For other details about locking
156see also dquot_operations section.
157->bdev_try_to_free_page is called from the ->releasepage handler of
158the block device inode. See there for more details.
159
160--------------------------- file_system_type ---------------------------
161prototypes:
162 int (*get_sb) (struct file_system_type *, int,
163 const char *, void *, struct vfsmount *);
164 struct dentry *(*mount) (struct file_system_type *, int,
165 const char *, void *);
166 void (*kill_sb) (struct super_block *);
167locking rules:
168 may block
169mount yes
170kill_sb yes
171
172->mount() returns ERR_PTR or the root dentry; its superblock should be locked
173on return.
174->kill_sb() takes a write-locked superblock, does all shutdown work on it,
175unlocks and drops the reference.
176
177--------------------------- address_space_operations --------------------------
178prototypes:
179 int (*writepage)(struct page *page, struct writeback_control *wbc);
180 int (*readpage)(struct file *, struct page *);
181 int (*sync_page)(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 int (*invalidatepage) (struct page *, unsigned long);
194 int (*releasepage) (struct page *, int);
195 void (*freepage)(struct page *);
196 int (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
197 loff_t offset, unsigned long nr_segs);
198 int (*get_xip_mem)(struct address_space *, pgoff_t, int, void **,
199 unsigned long *);
200 int (*migratepage)(struct address_space *, struct page *, struct page *);
201 int (*launder_page)(struct page *);
202 int (*is_partially_uptodate)(struct page *, read_descriptor_t *, unsigned long);
203 int (*error_remove_page)(struct address_space *, struct page *);
204
205locking rules:
206 All except set_page_dirty and freepage may block
207
208 PageLocked(page) i_mutex
209writepage: yes, unlocks (see below)
210readpage: yes, unlocks
211sync_page: maybe
212writepages:
213set_page_dirty no
214readpages:
215write_begin: locks the page yes
216write_end: yes, unlocks yes
217bmap:
218invalidatepage: yes
219releasepage: yes
220freepage: yes
221direct_IO:
222get_xip_mem: maybe
223migratepage: yes (both)
224launder_page: yes
225is_partially_uptodate: yes
226error_remove_page: yes
227
228 ->write_begin(), ->write_end(), ->sync_page() and ->readpage()
229may be called from the request handler (/dev/loop).
230
231 ->readpage() unlocks the page, either synchronously or via I/O
232completion.
233
234 ->readpages() populates the pagecache with the passed pages and starts
235I/O against them. They come unlocked upon I/O completion.
236
237 ->writepage() is used for two purposes: for "memory cleansing" and for
238"sync". These are quite different operations and the behaviour may differ
239depending upon the mode.
240
241If writepage is called for sync (wbc->sync_mode != WBC_SYNC_NONE) then
242it *must* start I/O against the page, even if that would involve
243blocking on in-progress I/O.
244
245If writepage is called for memory cleansing (sync_mode ==
246WBC_SYNC_NONE) then its role is to get as much writeout underway as
247possible. So writepage should try to avoid blocking against
248currently-in-progress I/O.
249
250If the filesystem is not called for "sync" and it determines that it
251would need to block against in-progress I/O to be able to start new I/O
252against the page the filesystem should redirty the page with
253redirty_page_for_writepage(), then unlock the page and return zero.
254This may also be done to avoid internal deadlocks, but rarely.
255
256If the filesystem is called for sync then it must wait on any
257in-progress I/O and then start new I/O.
258
259The filesystem should unlock the page synchronously, before returning to the
260caller, unless ->writepage() returns special WRITEPAGE_ACTIVATE
261value. WRITEPAGE_ACTIVATE means that page cannot really be written out
262currently, and VM should stop calling ->writepage() on this page for some
263time. VM does this by moving page to the head of the active list, hence the
264name.
265
266Unless the filesystem is going to redirty_page_for_writepage(), unlock the page
267and return zero, writepage *must* run set_page_writeback() against the page,
268followed by unlocking it. Once set_page_writeback() has been run against the
269page, write I/O can be submitted and the write I/O completion handler must run
270end_page_writeback() once the I/O is complete. If no I/O is submitted, the
271filesystem must run end_page_writeback() against the page before returning from
272writepage.
273
274That is: after 2.5.12, pages which are under writeout are *not* locked. Note,
275if the filesystem needs the page to be locked during writeout, that is ok, too,
276the page is allowed to be unlocked at any point in time between the calls to
277set_page_writeback() and end_page_writeback().
278
279Note, failure to run either redirty_page_for_writepage() or the combination of
280set_page_writeback()/end_page_writeback() on a page submitted to writepage
281will leave the page itself marked clean but it will be tagged as dirty in the
282radix tree. This incoherency can lead to all sorts of hard-to-debug problems
283in the filesystem like having dirty inodes at umount and losing written data.
284
285 ->sync_page() locking rules are not well-defined - usually it is called
286with lock on page, but that is not guaranteed. Considering the currently
287existing instances of this method ->sync_page() itself doesn't look
288well-defined...
289
290 ->writepages() is used for periodic writeback and for syscall-initiated
291sync operations. The address_space should start I/O against at least
292*nr_to_write pages. *nr_to_write must be decremented for each page which is
293written. The address_space implementation may write more (or less) pages
294than *nr_to_write asks for, but it should try to be reasonably close. If
295nr_to_write is NULL, all dirty pages must be written.
296
297writepages should _only_ write pages which are present on
298mapping->io_pages.
299
300 ->set_page_dirty() is called from various places in the kernel
301when the target page is marked as needing writeback. It may be called
302under spinlock (it cannot block) and is sometimes called with the page
303not locked.
304
305 ->bmap() is currently used by legacy ioctl() (FIBMAP) provided by some
306filesystems and by the swapper. The latter will eventually go away. Please,
307keep it that way and don't breed new callers.
308
309 ->invalidatepage() is called when the filesystem must attempt to drop
310some or all of the buffers from the page when it is being truncated. It
311returns zero on success. If ->invalidatepage is zero, the kernel uses
312block_invalidatepage() instead.
313
314 ->releasepage() is called when the kernel is about to try to drop the
315buffers from the page in preparation for freeing it. It returns zero to
316indicate that the buffers are (or may be) freeable. If ->releasepage is zero,
317the kernel assumes that the fs has no private interest in the buffers.
318
319 ->freepage() is called when the kernel is done dropping the page
320from the page cache.
321
322 ->launder_page() may be called prior to releasing a page if
323it is still found to be dirty. It returns zero if the page was successfully
324cleaned, or an error value if not. Note that in order to prevent the page
325getting mapped back in and redirtied, it needs to be kept locked
326across the entire operation.
327
328----------------------- file_lock_operations ------------------------------
329prototypes:
330 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
331 void (*fl_release_private)(struct file_lock *);
332
333
334locking rules:
335 file_lock_lock may block
336fl_copy_lock: yes no
337fl_release_private: maybe no
338
339----------------------- lock_manager_operations ---------------------------
340prototypes:
341 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
342 void (*lm_notify)(struct file_lock *); /* unblock callback */
343 int (*lm_grant)(struct file_lock *, struct file_lock *, int);
344 void (*lm_release_private)(struct file_lock *);
345 void (*lm_break)(struct file_lock *); /* break_lease callback */
346 int (*lm_change)(struct file_lock **, int);
347
348locking rules:
349 file_lock_lock may block
350lm_compare_owner: yes no
351lm_notify: yes no
352lm_grant: no no
353lm_release_private: maybe no
354lm_break: yes no
355lm_change yes no
356
357--------------------------- buffer_head -----------------------------------
358prototypes:
359 void (*b_end_io)(struct buffer_head *bh, int uptodate);
360
361locking rules:
362 called from interrupts. In other words, extreme care is needed here.
363bh is locked, but that's all warranties we have here. Currently only RAID1,
364highmem, fs/buffer.c, and fs/ntfs/aops.c are providing these. Block devices
365call this method upon the IO completion.
366
367--------------------------- block_device_operations -----------------------
368prototypes:
369 int (*open) (struct block_device *, fmode_t);
370 int (*release) (struct gendisk *, fmode_t);
371 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
372 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
373 int (*direct_access) (struct block_device *, sector_t, void **, unsigned long *);
374 int (*media_changed) (struct gendisk *);
375 void (*unlock_native_capacity) (struct gendisk *);
376 int (*revalidate_disk) (struct gendisk *);
377 int (*getgeo)(struct block_device *, struct hd_geometry *);
378 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
379
380locking rules:
381 bd_mutex
382open: yes
383release: yes
384ioctl: no
385compat_ioctl: no
386direct_access: no
387media_changed: no
388unlock_native_capacity: no
389revalidate_disk: no
390getgeo: no
391swap_slot_free_notify: no (see below)
392
393media_changed, unlock_native_capacity and revalidate_disk are called only from
394check_disk_change().
395
396swap_slot_free_notify is called with swap_lock and sometimes the page lock
397held.
398
399
400--------------------------- file_operations -------------------------------
401prototypes:
402 loff_t (*llseek) (struct file *, loff_t, int);
403 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
404 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
405 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
406 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
407 int (*readdir) (struct file *, void *, filldir_t);
408 unsigned int (*poll) (struct file *, struct poll_table_struct *);
409 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
410 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
411 int (*mmap) (struct file *, struct vm_area_struct *);
412 int (*open) (struct inode *, struct file *);
413 int (*flush) (struct file *);
414 int (*release) (struct inode *, struct file *);
415 int (*fsync) (struct file *, loff_t start, loff_t end, int datasync);
416 int (*aio_fsync) (struct kiocb *, int datasync);
417 int (*fasync) (int, struct file *, int);
418 int (*lock) (struct file *, int, struct file_lock *);
419 ssize_t (*readv) (struct file *, const struct iovec *, unsigned long,
420 loff_t *);
421 ssize_t (*writev) (struct file *, const struct iovec *, unsigned long,
422 loff_t *);
423 ssize_t (*sendfile) (struct file *, loff_t *, size_t, read_actor_t,
424 void __user *);
425 ssize_t (*sendpage) (struct file *, struct page *, int, size_t,
426 loff_t *, int);
427 unsigned long (*get_unmapped_area)(struct file *, unsigned long,
428 unsigned long, unsigned long, unsigned long);
429 int (*check_flags)(int);
430 int (*flock) (struct file *, int, struct file_lock *);
431 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *,
432 size_t, unsigned int);
433 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *,
434 size_t, unsigned int);
435 int (*setlease)(struct file *, long, struct file_lock **);
436 long (*fallocate)(struct file *, int, loff_t, loff_t);
437};
438
439locking rules:
440 All may block except for ->setlease.
441 No VFS locks held on entry except for ->setlease.
442
443->setlease has the file_list_lock held and must not sleep.
444
445->llseek() locking has moved from llseek to the individual llseek
446implementations. If your fs is not using generic_file_llseek, you
447need to acquire and release the appropriate locks in your ->llseek().
448For many filesystems, it is probably safe to acquire the inode
449mutex or just to use i_size_read() instead.
450Note: this does not protect the file->f_pos against concurrent modifications
451since this is something the userspace has to take care about.
452
453->fasync() is responsible for maintaining the FASYNC bit in filp->f_flags.
454Most instances call fasync_helper(), which does that maintenance, so it's
455not normally something one needs to worry about. Return values > 0 will be
456mapped to zero in the VFS layer.
457
458->readdir() and ->ioctl() on directories must be changed. Ideally we would
459move ->readdir() to inode_operations and use a separate method for directory
460->ioctl() or kill the latter completely. One of the problems is that for
461anything that resembles union-mount we won't have a struct file for all
462components. And there are other reasons why the current interface is a mess...
463
464->read on directories probably must go away - we should just enforce -EISDIR
465in sys_read() and friends.
466
467--------------------------- dquot_operations -------------------------------
468prototypes:
469 int (*write_dquot) (struct dquot *);
470 int (*acquire_dquot) (struct dquot *);
471 int (*release_dquot) (struct dquot *);
472 int (*mark_dirty) (struct dquot *);
473 int (*write_info) (struct super_block *, int);
474
475These operations are intended to be more or less wrapping functions that ensure
476a proper locking wrt the filesystem and call the generic quota operations.
477
478What filesystem should expect from the generic quota functions:
479
480 FS recursion Held locks when called
481write_dquot: yes dqonoff_sem or dqptr_sem
482acquire_dquot: yes dqonoff_sem or dqptr_sem
483release_dquot: yes dqonoff_sem or dqptr_sem
484mark_dirty: no -
485write_info: yes dqonoff_sem
486
487FS recursion means calling ->quota_read() and ->quota_write() from superblock
488operations.
489
490More details about quota locking can be found in fs/dquot.c.
491
492--------------------------- vm_operations_struct -----------------------------
493prototypes:
494 void (*open)(struct vm_area_struct*);
495 void (*close)(struct vm_area_struct*);
496 int (*fault)(struct vm_area_struct*, struct vm_fault *);
497 int (*page_mkwrite)(struct vm_area_struct *, struct vm_fault *);
498 int (*access)(struct vm_area_struct *, unsigned long, void*, int, int);
499
500locking rules:
501 mmap_sem PageLocked(page)
502open: yes
503close: yes
504fault: yes can return with page locked
505page_mkwrite: yes can return with page locked
506access: yes
507
508 ->fault() is called when a previously not present pte is about
509to be faulted in. The filesystem must find and return the page associated
510with the passed in "pgoff" in the vm_fault structure. If it is possible that
511the page may be truncated and/or invalidated, then the filesystem must lock
512the page, then ensure it is not already truncated (the page lock will block
513subsequent truncate), and then return with VM_FAULT_LOCKED, and the page
514locked. The VM will unlock the page.
515
516 ->page_mkwrite() is called when a previously read-only pte is
517about to become writeable. The filesystem again must ensure that there are
518no truncate/invalidate races, and then return with the page locked. If
519the page has been truncated, the filesystem should not look up a new page
520like the ->fault() handler, but simply return with VM_FAULT_NOPAGE, which
521will cause the VM to retry the fault.
522
523 ->access() is called when get_user_pages() fails in
524acces_process_vm(), typically used to debug a process through
525/proc/pid/mem or ptrace. This function is needed only for
526VM_IO | VM_PFNMAP VMAs.
527
528================================================================================
529 Dubious stuff
530
531(if you break something or notice that it is broken and do not fix it yourself
532- at least put it here)
1 The text below describes the locking rules for VFS-related methods.
2It is (believed to be) up-to-date. *Please*, if you change anything in
3prototypes or locking protocols - update this file. And update the relevant
4instances in the tree, don't leave that to maintainers of filesystems/devices/
5etc. At the very least, put the list of dubious cases in the end of this file.
6Don't turn it into log - maintainers of out-of-the-tree code are supposed to
7be able to use diff(1).
8 Thing currently missing here: socket operations. Alexey?
9
10--------------------------- dentry_operations --------------------------
11prototypes:
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);
23
24locking rules:
25 rename_lock ->d_lock may block rcu-walk
26d_revalidate: no no yes (ref-walk) maybe
27d_weak_revalidate:no no yes no
28d_hash no no no maybe
29d_compare: yes no no maybe
30d_delete: no yes no no
31d_release: no no yes no
32d_prune: no yes no no
33d_iput: no no yes no
34d_dname: no no no no
35d_automount: no no yes no
36d_manage: no no yes (ref-walk) maybe
37
38--------------------------- inode_operations ---------------------------
39prototypes:
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 void * (*follow_link) (struct dentry *, struct nameidata *);
54 void (*put_link) (struct dentry *, struct nameidata *, void *);
55 void (*truncate) (struct inode *);
56 int (*permission) (struct inode *, int, unsigned int);
57 int (*get_acl)(struct inode *, int);
58 int (*setattr) (struct dentry *, struct iattr *);
59 int (*getattr) (struct vfsmount *, struct dentry *, struct kstat *);
60 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
61 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
62 ssize_t (*listxattr) (struct dentry *, char *, size_t);
63 int (*removexattr) (struct dentry *, const char *);
64 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start, u64 len);
65 void (*update_time)(struct inode *, struct timespec *, int);
66 int (*atomic_open)(struct inode *, struct dentry *,
67 struct file *, unsigned open_flag,
68 umode_t create_mode, int *opened);
69 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
70
71locking rules:
72 all may block
73 i_mutex(inode)
74lookup: yes
75create: yes
76link: yes (both)
77mknod: yes
78symlink: yes
79mkdir: yes
80unlink: yes (both)
81rmdir: yes (both) (see below)
82rename: yes (all) (see below)
83rename2: yes (all) (see below)
84readlink: no
85follow_link: no
86put_link: no
87setattr: yes
88permission: no (may not block if called in rcu-walk mode)
89get_acl: no
90getattr: no
91setxattr: yes
92getxattr: no
93listxattr: no
94removexattr: yes
95fiemap: no
96update_time: no
97atomic_open: yes
98tmpfile: no
99
100 Additionally, ->rmdir(), ->unlink() and ->rename() have ->i_mutex on
101victim.
102 cross-directory ->rename() and rename2() has (per-superblock)
103->s_vfs_rename_sem.
104
105See Documentation/filesystems/directory-locking for more detailed discussion
106of the locking scheme for directory operations.
107
108--------------------------- super_operations ---------------------------
109prototypes:
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);
127
128locking rules:
129 All may block [not true, see below]
130 s_umount
131alloc_inode:
132destroy_inode:
133dirty_inode:
134write_inode:
135drop_inode: !!!inode->i_lock!!!
136evict_inode:
137put_super: write
138sync_fs: read
139freeze_fs: write
140unfreeze_fs: write
141statfs: maybe(read) (see below)
142remount_fs: write
143umount_begin: no
144show_options: no (namespace_sem)
145quota_read: no (see below)
146quota_write: no (see below)
147bdev_try_to_free_page: no (see below)
148
149->statfs() has s_umount (shared) when called by ustat(2) (native or
150compat), but that's an accident of bad API; s_umount is used to pin
151the superblock down when we only have dev_t given us by userland to
152identify the superblock. Everything else (statfs(), fstatfs(), etc.)
153doesn't hold it when calling ->statfs() - superblock is pinned down
154by resolving the pathname passed to syscall.
155->quota_read() and ->quota_write() functions are both guaranteed to
156be the only ones operating on the quota file by the quota code (via
157dqio_sem) (unless an admin really wants to screw up something and
158writes to quota files with quotas on). For other details about locking
159see also dquot_operations section.
160->bdev_try_to_free_page is called from the ->releasepage handler of
161the block device inode. See there for more details.
162
163--------------------------- file_system_type ---------------------------
164prototypes:
165 int (*get_sb) (struct file_system_type *, int,
166 const char *, void *, struct vfsmount *);
167 struct dentry *(*mount) (struct file_system_type *, int,
168 const char *, void *);
169 void (*kill_sb) (struct super_block *);
170locking rules:
171 may block
172mount yes
173kill_sb yes
174
175->mount() returns ERR_PTR or the root dentry; its superblock should be locked
176on return.
177->kill_sb() takes a write-locked superblock, does all shutdown work on it,
178unlocks and drops the reference.
179
180--------------------------- address_space_operations --------------------------
181prototypes:
182 int (*writepage)(struct page *page, struct writeback_control *wbc);
183 int (*readpage)(struct file *, struct page *);
184 int (*sync_page)(struct page *);
185 int (*writepages)(struct address_space *, struct writeback_control *);
186 int (*set_page_dirty)(struct page *page);
187 int (*readpages)(struct file *filp, struct address_space *mapping,
188 struct list_head *pages, unsigned nr_pages);
189 int (*write_begin)(struct file *, struct address_space *mapping,
190 loff_t pos, unsigned len, unsigned flags,
191 struct page **pagep, void **fsdata);
192 int (*write_end)(struct file *, struct address_space *mapping,
193 loff_t pos, unsigned len, unsigned copied,
194 struct page *page, void *fsdata);
195 sector_t (*bmap)(struct address_space *, sector_t);
196 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
197 int (*releasepage) (struct page *, int);
198 void (*freepage)(struct page *);
199 int (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
200 loff_t offset, unsigned long nr_segs);
201 int (*get_xip_mem)(struct address_space *, pgoff_t, int, void **,
202 unsigned long *);
203 int (*migratepage)(struct address_space *, struct page *, struct page *);
204 int (*launder_page)(struct page *);
205 int (*is_partially_uptodate)(struct page *, unsigned long, unsigned long);
206 int (*error_remove_page)(struct address_space *, struct page *);
207 int (*swap_activate)(struct file *);
208 int (*swap_deactivate)(struct file *);
209
210locking rules:
211 All except set_page_dirty and freepage may block
212
213 PageLocked(page) i_mutex
214writepage: yes, unlocks (see below)
215readpage: yes, unlocks
216sync_page: maybe
217writepages:
218set_page_dirty no
219readpages:
220write_begin: locks the page yes
221write_end: yes, unlocks yes
222bmap:
223invalidatepage: yes
224releasepage: yes
225freepage: yes
226direct_IO:
227get_xip_mem: maybe
228migratepage: yes (both)
229launder_page: yes
230is_partially_uptodate: yes
231error_remove_page: yes
232swap_activate: no
233swap_deactivate: no
234
235 ->write_begin(), ->write_end(), ->sync_page() and ->readpage()
236may be called from the request handler (/dev/loop).
237
238 ->readpage() unlocks the page, either synchronously or via I/O
239completion.
240
241 ->readpages() populates the pagecache with the passed pages and starts
242I/O against them. They come unlocked upon I/O completion.
243
244 ->writepage() is used for two purposes: for "memory cleansing" and for
245"sync". These are quite different operations and the behaviour may differ
246depending upon the mode.
247
248If writepage is called for sync (wbc->sync_mode != WBC_SYNC_NONE) then
249it *must* start I/O against the page, even if that would involve
250blocking on in-progress I/O.
251
252If writepage is called for memory cleansing (sync_mode ==
253WBC_SYNC_NONE) then its role is to get as much writeout underway as
254possible. So writepage should try to avoid blocking against
255currently-in-progress I/O.
256
257If the filesystem is not called for "sync" and it determines that it
258would need to block against in-progress I/O to be able to start new I/O
259against the page the filesystem should redirty the page with
260redirty_page_for_writepage(), then unlock the page and return zero.
261This may also be done to avoid internal deadlocks, but rarely.
262
263If the filesystem is called for sync then it must wait on any
264in-progress I/O and then start new I/O.
265
266The filesystem should unlock the page synchronously, before returning to the
267caller, unless ->writepage() returns special WRITEPAGE_ACTIVATE
268value. WRITEPAGE_ACTIVATE means that page cannot really be written out
269currently, and VM should stop calling ->writepage() on this page for some
270time. VM does this by moving page to the head of the active list, hence the
271name.
272
273Unless the filesystem is going to redirty_page_for_writepage(), unlock the page
274and return zero, writepage *must* run set_page_writeback() against the page,
275followed by unlocking it. Once set_page_writeback() has been run against the
276page, write I/O can be submitted and the write I/O completion handler must run
277end_page_writeback() once the I/O is complete. If no I/O is submitted, the
278filesystem must run end_page_writeback() against the page before returning from
279writepage.
280
281That is: after 2.5.12, pages which are under writeout are *not* locked. Note,
282if the filesystem needs the page to be locked during writeout, that is ok, too,
283the page is allowed to be unlocked at any point in time between the calls to
284set_page_writeback() and end_page_writeback().
285
286Note, failure to run either redirty_page_for_writepage() or the combination of
287set_page_writeback()/end_page_writeback() on a page submitted to writepage
288will leave the page itself marked clean but it will be tagged as dirty in the
289radix tree. This incoherency can lead to all sorts of hard-to-debug problems
290in the filesystem like having dirty inodes at umount and losing written data.
291
292 ->sync_page() locking rules are not well-defined - usually it is called
293with lock on page, but that is not guaranteed. Considering the currently
294existing instances of this method ->sync_page() itself doesn't look
295well-defined...
296
297 ->writepages() is used for periodic writeback and for syscall-initiated
298sync operations. The address_space should start I/O against at least
299*nr_to_write pages. *nr_to_write must be decremented for each page which is
300written. The address_space implementation may write more (or less) pages
301than *nr_to_write asks for, but it should try to be reasonably close. If
302nr_to_write is NULL, all dirty pages must be written.
303
304writepages should _only_ write pages which are present on
305mapping->io_pages.
306
307 ->set_page_dirty() is called from various places in the kernel
308when the target page is marked as needing writeback. It may be called
309under spinlock (it cannot block) and is sometimes called with the page
310not locked.
311
312 ->bmap() is currently used by legacy ioctl() (FIBMAP) provided by some
313filesystems and by the swapper. The latter will eventually go away. Please,
314keep it that way and don't breed new callers.
315
316 ->invalidatepage() is called when the filesystem must attempt to drop
317some or all of the buffers from the page when it is being truncated. It
318returns zero on success. If ->invalidatepage is zero, the kernel uses
319block_invalidatepage() instead.
320
321 ->releasepage() is called when the kernel is about to try to drop the
322buffers from the page in preparation for freeing it. It returns zero to
323indicate that the buffers are (or may be) freeable. If ->releasepage is zero,
324the kernel assumes that the fs has no private interest in the buffers.
325
326 ->freepage() is called when the kernel is done dropping the page
327from the page cache.
328
329 ->launder_page() may be called prior to releasing a page if
330it is still found to be dirty. It returns zero if the page was successfully
331cleaned, or an error value if not. Note that in order to prevent the page
332getting mapped back in and redirtied, it needs to be kept locked
333across the entire operation.
334
335 ->swap_activate will be called with a non-zero argument on
336files backing (non block device backed) swapfiles. A return value
337of zero indicates success, in which case this file can be used for
338backing swapspace. The swapspace operations will be proxied to the
339address space operations.
340
341 ->swap_deactivate() will be called in the sys_swapoff()
342path after ->swap_activate() returned success.
343
344----------------------- file_lock_operations ------------------------------
345prototypes:
346 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
347 void (*fl_release_private)(struct file_lock *);
348
349
350locking rules:
351 inode->i_lock may block
352fl_copy_lock: yes no
353fl_release_private: maybe no
354
355----------------------- lock_manager_operations ---------------------------
356prototypes:
357 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
358 unsigned long (*lm_owner_key)(struct file_lock *);
359 void (*lm_notify)(struct file_lock *); /* unblock callback */
360 int (*lm_grant)(struct file_lock *, struct file_lock *, int);
361 void (*lm_break)(struct file_lock *); /* break_lease callback */
362 int (*lm_change)(struct file_lock **, int);
363
364locking rules:
365
366 inode->i_lock blocked_lock_lock may block
367lm_compare_owner: yes[1] maybe no
368lm_owner_key yes[1] yes no
369lm_notify: yes yes no
370lm_grant: no no no
371lm_break: yes no no
372lm_change yes no no
373
374[1]: ->lm_compare_owner and ->lm_owner_key are generally called with
375*an* inode->i_lock held. It may not be the i_lock of the inode
376associated with either file_lock argument! This is the case with deadlock
377detection, since the code has to chase down the owners of locks that may
378be entirely unrelated to the one on which the lock is being acquired.
379For deadlock detection however, the blocked_lock_lock is also held. The
380fact that these locks are held ensures that the file_locks do not
381disappear out from under you while doing the comparison or generating an
382owner key.
383
384--------------------------- buffer_head -----------------------------------
385prototypes:
386 void (*b_end_io)(struct buffer_head *bh, int uptodate);
387
388locking rules:
389 called from interrupts. In other words, extreme care is needed here.
390bh is locked, but that's all warranties we have here. Currently only RAID1,
391highmem, fs/buffer.c, and fs/ntfs/aops.c are providing these. Block devices
392call this method upon the IO completion.
393
394--------------------------- block_device_operations -----------------------
395prototypes:
396 int (*open) (struct block_device *, fmode_t);
397 int (*release) (struct gendisk *, fmode_t);
398 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
399 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
400 int (*direct_access) (struct block_device *, sector_t, void **, unsigned long *);
401 int (*media_changed) (struct gendisk *);
402 void (*unlock_native_capacity) (struct gendisk *);
403 int (*revalidate_disk) (struct gendisk *);
404 int (*getgeo)(struct block_device *, struct hd_geometry *);
405 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
406
407locking rules:
408 bd_mutex
409open: yes
410release: yes
411ioctl: no
412compat_ioctl: no
413direct_access: no
414media_changed: no
415unlock_native_capacity: no
416revalidate_disk: no
417getgeo: no
418swap_slot_free_notify: no (see below)
419
420media_changed, unlock_native_capacity and revalidate_disk are called only from
421check_disk_change().
422
423swap_slot_free_notify is called with swap_lock and sometimes the page lock
424held.
425
426
427--------------------------- file_operations -------------------------------
428prototypes:
429 loff_t (*llseek) (struct file *, loff_t, int);
430 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
431 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
432 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
433 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
434 int (*iterate) (struct file *, struct dir_context *);
435 unsigned int (*poll) (struct file *, struct poll_table_struct *);
436 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
437 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
438 int (*mmap) (struct file *, struct vm_area_struct *);
439 int (*open) (struct inode *, struct file *);
440 int (*flush) (struct file *);
441 int (*release) (struct inode *, struct file *);
442 int (*fsync) (struct file *, loff_t start, loff_t end, int datasync);
443 int (*aio_fsync) (struct kiocb *, int datasync);
444 int (*fasync) (int, struct file *, int);
445 int (*lock) (struct file *, int, struct file_lock *);
446 ssize_t (*readv) (struct file *, const struct iovec *, unsigned long,
447 loff_t *);
448 ssize_t (*writev) (struct file *, const struct iovec *, unsigned long,
449 loff_t *);
450 ssize_t (*sendfile) (struct file *, loff_t *, size_t, read_actor_t,
451 void __user *);
452 ssize_t (*sendpage) (struct file *, struct page *, int, size_t,
453 loff_t *, int);
454 unsigned long (*get_unmapped_area)(struct file *, unsigned long,
455 unsigned long, unsigned long, unsigned long);
456 int (*check_flags)(int);
457 int (*flock) (struct file *, int, struct file_lock *);
458 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *,
459 size_t, unsigned int);
460 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *,
461 size_t, unsigned int);
462 int (*setlease)(struct file *, long, struct file_lock **);
463 long (*fallocate)(struct file *, int, loff_t, loff_t);
464};
465
466locking rules:
467 All may block except for ->setlease.
468 No VFS locks held on entry except for ->setlease.
469
470->setlease has the file_list_lock held and must not sleep.
471
472->llseek() locking has moved from llseek to the individual llseek
473implementations. If your fs is not using generic_file_llseek, you
474need to acquire and release the appropriate locks in your ->llseek().
475For many filesystems, it is probably safe to acquire the inode
476mutex or just to use i_size_read() instead.
477Note: this does not protect the file->f_pos against concurrent modifications
478since this is something the userspace has to take care about.
479
480->fasync() is responsible for maintaining the FASYNC bit in filp->f_flags.
481Most instances call fasync_helper(), which does that maintenance, so it's
482not normally something one needs to worry about. Return values > 0 will be
483mapped to zero in the VFS layer.
484
485->readdir() and ->ioctl() on directories must be changed. Ideally we would
486move ->readdir() to inode_operations and use a separate method for directory
487->ioctl() or kill the latter completely. One of the problems is that for
488anything that resembles union-mount we won't have a struct file for all
489components. And there are other reasons why the current interface is a mess...
490
491->read on directories probably must go away - we should just enforce -EISDIR
492in sys_read() and friends.
493
494--------------------------- dquot_operations -------------------------------
495prototypes:
496 int (*write_dquot) (struct dquot *);
497 int (*acquire_dquot) (struct dquot *);
498 int (*release_dquot) (struct dquot *);
499 int (*mark_dirty) (struct dquot *);
500 int (*write_info) (struct super_block *, int);
501
502These operations are intended to be more or less wrapping functions that ensure
503a proper locking wrt the filesystem and call the generic quota operations.
504
505What filesystem should expect from the generic quota functions:
506
507 FS recursion Held locks when called
508write_dquot: yes dqonoff_sem or dqptr_sem
509acquire_dquot: yes dqonoff_sem or dqptr_sem
510release_dquot: yes dqonoff_sem or dqptr_sem
511mark_dirty: no -
512write_info: yes dqonoff_sem
513
514FS recursion means calling ->quota_read() and ->quota_write() from superblock
515operations.
516
517More details about quota locking can be found in fs/dquot.c.
518
519--------------------------- vm_operations_struct -----------------------------
520prototypes:
521 void (*open)(struct vm_area_struct*);
522 void (*close)(struct vm_area_struct*);
523 int (*fault)(struct vm_area_struct*, struct vm_fault *);
524 int (*page_mkwrite)(struct vm_area_struct *, struct vm_fault *);
525 int (*access)(struct vm_area_struct *, unsigned long, void*, int, int);
526
527locking rules:
528 mmap_sem PageLocked(page)
529open: yes
530close: yes
531fault: yes can return with page locked
532map_pages: yes
533page_mkwrite: yes can return with page locked
534access: yes
535
536 ->fault() is called when a previously not present pte is about
537to be faulted in. The filesystem must find and return the page associated
538with the passed in "pgoff" in the vm_fault structure. If it is possible that
539the page may be truncated and/or invalidated, then the filesystem must lock
540the page, then ensure it is not already truncated (the page lock will block
541subsequent truncate), and then return with VM_FAULT_LOCKED, and the page
542locked. The VM will unlock the page.
543
544 ->map_pages() is called when VM asks to map easy accessible pages.
545Filesystem should find and map pages associated with offsets from "pgoff"
546till "max_pgoff". ->map_pages() is called with page table locked and must
547not block. If it's not possible to reach a page without blocking,
548filesystem should skip it. Filesystem should use do_set_pte() to setup
549page table entry. Pointer to entry associated with offset "pgoff" is
550passed in "pte" field in vm_fault structure. Pointers to entries for other
551offsets should be calculated relative to "pte".
552
553 ->page_mkwrite() is called when a previously read-only pte is
554about to become writeable. The filesystem again must ensure that there are
555no truncate/invalidate races, and then return with the page locked. If
556the page has been truncated, the filesystem should not look up a new page
557like the ->fault() handler, but simply return with VM_FAULT_NOPAGE, which
558will cause the VM to retry the fault.
559
560 ->access() is called when get_user_pages() fails in
561access_process_vm(), typically used to debug a process through
562/proc/pid/mem or ptrace. This function is needed only for
563VM_IO | VM_PFNMAP VMAs.
564
565================================================================================
566 Dubious stuff
567
568(if you break something or notice that it is broken and do not fix it yourself
569- at least put it here)