1 ======= 1 =======
2 Locking 2 Locking
3 ======= 3 =======
4 4
5 The text below describes the locking rules for 5 The text below describes the locking rules for VFS-related methods.
6 It is (believed to be) up-to-date. *Please*, i 6 It is (believed to be) up-to-date. *Please*, if you change anything in
7 prototypes or locking protocols - update this 7 prototypes or locking protocols - update this file. And update the relevant
8 instances in the tree, don't leave that to mai 8 instances in the tree, don't leave that to maintainers of filesystems/devices/
9 etc. At the very least, put the list of dubiou 9 etc. At the very least, put the list of dubious cases in the end of this file.
10 Don't turn it into log - maintainers of out-of 10 Don't turn it into log - maintainers of out-of-the-tree code are supposed to
11 be able to use diff(1). 11 be able to use diff(1).
12 12
13 Thing currently missing here: socket operation 13 Thing currently missing here: socket operations. Alexey?
14 14
15 dentry_operations 15 dentry_operations
16 ================= 16 =================
17 17
18 prototypes:: 18 prototypes::
19 19
20 int (*d_revalidate)(struct dentry *, u 20 int (*d_revalidate)(struct dentry *, unsigned int);
21 int (*d_weak_revalidate)(struct dentry 21 int (*d_weak_revalidate)(struct dentry *, unsigned int);
22 int (*d_hash)(const struct dentry *, s 22 int (*d_hash)(const struct dentry *, struct qstr *);
23 int (*d_compare)(const struct dentry * 23 int (*d_compare)(const struct dentry *,
24 unsigned int, const ch 24 unsigned int, const char *, const struct qstr *);
25 int (*d_delete)(struct dentry *); 25 int (*d_delete)(struct dentry *);
26 int (*d_init)(struct dentry *); 26 int (*d_init)(struct dentry *);
27 void (*d_release)(struct dentry *); 27 void (*d_release)(struct dentry *);
28 void (*d_iput)(struct dentry *, struct 28 void (*d_iput)(struct dentry *, struct inode *);
29 char *(*d_dname)((struct dentry *dentr 29 char *(*d_dname)((struct dentry *dentry, char *buffer, int buflen);
30 struct vfsmount *(*d_automount)(struct 30 struct vfsmount *(*d_automount)(struct path *path);
31 int (*d_manage)(const struct path *, b 31 int (*d_manage)(const struct path *, bool);
32 struct dentry *(*d_real)(struct dentry 32 struct dentry *(*d_real)(struct dentry *, enum d_real_type type);
33 33
34 locking rules: 34 locking rules:
35 35
36 ================== =========== ======== 36 ================== =========== ======== ============== ========
37 ops rename_lock ->d_lock 37 ops rename_lock ->d_lock may block rcu-walk
38 ================== =========== ======== 38 ================== =========== ======== ============== ========
39 d_revalidate: no no 39 d_revalidate: no no yes (ref-walk) maybe
40 d_weak_revalidate: no no 40 d_weak_revalidate: no no yes no
41 d_hash no no 41 d_hash no no no maybe
42 d_compare: yes no 42 d_compare: yes no no maybe
43 d_delete: no yes 43 d_delete: no yes no no
44 d_init: no no 44 d_init: no no yes no
45 d_release: no no 45 d_release: no no yes no
46 d_prune: no yes 46 d_prune: no yes no no
47 d_iput: no no 47 d_iput: no no yes no
48 d_dname: no no 48 d_dname: no no no no
49 d_automount: no no 49 d_automount: no no yes no
50 d_manage: no no 50 d_manage: no no yes (ref-walk) maybe
51 d_real no no 51 d_real no no yes no
52 ================== =========== ======== 52 ================== =========== ======== ============== ========
53 53
54 inode_operations 54 inode_operations
55 ================ 55 ================
56 56
57 prototypes:: 57 prototypes::
58 58
59 int (*create) (struct mnt_idmap *, str 59 int (*create) (struct mnt_idmap *, struct inode *,struct dentry *,umode_t, bool);
60 struct dentry * (*lookup) (struct inod 60 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
61 int (*link) (struct dentry *,struct in 61 int (*link) (struct dentry *,struct inode *,struct dentry *);
62 int (*unlink) (struct inode *,struct d 62 int (*unlink) (struct inode *,struct dentry *);
63 int (*symlink) (struct mnt_idmap *, st 63 int (*symlink) (struct mnt_idmap *, struct inode *,struct dentry *,const char *);
64 int (*mkdir) (struct mnt_idmap *, stru 64 int (*mkdir) (struct mnt_idmap *, struct inode *,struct dentry *,umode_t);
65 int (*rmdir) (struct inode *,struct de 65 int (*rmdir) (struct inode *,struct dentry *);
66 int (*mknod) (struct mnt_idmap *, stru 66 int (*mknod) (struct mnt_idmap *, struct inode *,struct dentry *,umode_t,dev_t);
67 int (*rename) (struct mnt_idmap *, str 67 int (*rename) (struct mnt_idmap *, struct inode *, struct dentry *,
68 struct inode *, struct 68 struct inode *, struct dentry *, unsigned int);
69 int (*readlink) (struct dentry *, char 69 int (*readlink) (struct dentry *, char __user *,int);
70 const char *(*get_link) (struct dentry 70 const char *(*get_link) (struct dentry *, struct inode *, struct delayed_call *);
71 void (*truncate) (struct inode *); 71 void (*truncate) (struct inode *);
72 int (*permission) (struct mnt_idmap *, 72 int (*permission) (struct mnt_idmap *, struct inode *, int, unsigned int);
73 struct posix_acl * (*get_inode_acl)(st 73 struct posix_acl * (*get_inode_acl)(struct inode *, int, bool);
74 int (*setattr) (struct mnt_idmap *, st 74 int (*setattr) (struct mnt_idmap *, struct dentry *, struct iattr *);
75 int (*getattr) (struct mnt_idmap *, co 75 int (*getattr) (struct mnt_idmap *, const struct path *, struct kstat *, u32, unsigned int);
76 ssize_t (*listxattr) (struct dentry *, 76 ssize_t (*listxattr) (struct dentry *, char *, size_t);
77 int (*fiemap)(struct inode *, struct f 77 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start, u64 len);
78 void (*update_time)(struct inode *, st 78 void (*update_time)(struct inode *, struct timespec *, int);
79 int (*atomic_open)(struct inode *, str 79 int (*atomic_open)(struct inode *, struct dentry *,
80 struct file *, 80 struct file *, unsigned open_flag,
81 umode_t create 81 umode_t create_mode);
82 int (*tmpfile) (struct mnt_idmap *, st 82 int (*tmpfile) (struct mnt_idmap *, struct inode *,
83 struct file *, umode_t 83 struct file *, umode_t);
84 int (*fileattr_set)(struct mnt_idmap * 84 int (*fileattr_set)(struct mnt_idmap *idmap,
85 struct dentry *den 85 struct dentry *dentry, struct fileattr *fa);
86 int (*fileattr_get)(struct dentry *den 86 int (*fileattr_get)(struct dentry *dentry, struct fileattr *fa);
87 struct posix_acl * (*get_acl)(struct m 87 struct posix_acl * (*get_acl)(struct mnt_idmap *, struct dentry *, int);
88 struct offset_ctx *(*get_offset_ctx)(s 88 struct offset_ctx *(*get_offset_ctx)(struct inode *inode);
89 89
90 locking rules: 90 locking rules:
91 all may block 91 all may block
92 92
93 ============== ============================== 93 ============== ==================================================
94 ops i_rwsem(inode) 94 ops i_rwsem(inode)
95 ============== ============================== 95 ============== ==================================================
96 lookup: shared 96 lookup: shared
97 create: exclusive 97 create: exclusive
98 link: exclusive (both) 98 link: exclusive (both)
99 mknod: exclusive 99 mknod: exclusive
100 symlink: exclusive 100 symlink: exclusive
101 mkdir: exclusive 101 mkdir: exclusive
102 unlink: exclusive (both) 102 unlink: exclusive (both)
103 rmdir: exclusive (both)(see below) 103 rmdir: exclusive (both)(see below)
104 rename: exclusive (both parents, some 104 rename: exclusive (both parents, some children) (see below)
105 readlink: no 105 readlink: no
106 get_link: no 106 get_link: no
107 setattr: exclusive 107 setattr: exclusive
108 permission: no (may not block if called in 108 permission: no (may not block if called in rcu-walk mode)
109 get_inode_acl: no 109 get_inode_acl: no
110 get_acl: no 110 get_acl: no
111 getattr: no 111 getattr: no
112 listxattr: no 112 listxattr: no
113 fiemap: no 113 fiemap: no
114 update_time: no 114 update_time: no
115 atomic_open: shared (exclusive if O_CREAT i 115 atomic_open: shared (exclusive if O_CREAT is set in open flags)
116 tmpfile: no 116 tmpfile: no
117 fileattr_get: no or exclusive 117 fileattr_get: no or exclusive
118 fileattr_set: exclusive 118 fileattr_set: exclusive
119 get_offset_ctx no 119 get_offset_ctx no
120 ============== ============================== 120 ============== ==================================================
121 121
122 122
123 Additionally, ->rmdir(), ->unlink() an 123 Additionally, ->rmdir(), ->unlink() and ->rename() have ->i_rwsem
124 exclusive on victim. 124 exclusive on victim.
125 cross-directory ->rename() has (per-su 125 cross-directory ->rename() has (per-superblock) ->s_vfs_rename_sem.
126 ->unlink() and ->rename() have ->i_rws 126 ->unlink() and ->rename() have ->i_rwsem exclusive on all non-directories
127 involved. 127 involved.
128 ->rename() has ->i_rwsem exclusive on 128 ->rename() has ->i_rwsem exclusive on any subdirectory that changes parent.
129 129
130 See Documentation/filesystems/directory-lockin 130 See Documentation/filesystems/directory-locking.rst for more detailed discussion
131 of the locking scheme for directory operations 131 of the locking scheme for directory operations.
132 132
133 xattr_handler operations 133 xattr_handler operations
134 ======================== 134 ========================
135 135
136 prototypes:: 136 prototypes::
137 137
138 bool (*list)(struct dentry *dentry); 138 bool (*list)(struct dentry *dentry);
139 int (*get)(const struct xattr_handler 139 int (*get)(const struct xattr_handler *handler, struct dentry *dentry,
140 struct inode *inode, const 140 struct inode *inode, const char *name, void *buffer,
141 size_t size); 141 size_t size);
142 int (*set)(const struct xattr_handler 142 int (*set)(const struct xattr_handler *handler,
143 struct mnt_idmap *idmap, 143 struct mnt_idmap *idmap,
144 struct dentry *dentry, stru 144 struct dentry *dentry, struct inode *inode, const char *name,
145 const void *buffer, size_t 145 const void *buffer, size_t size, int flags);
146 146
147 locking rules: 147 locking rules:
148 all may block 148 all may block
149 149
150 ===== ============== 150 ===== ==============
151 ops i_rwsem(inode) 151 ops i_rwsem(inode)
152 ===== ============== 152 ===== ==============
153 list: no 153 list: no
154 get: no 154 get: no
155 set: exclusive 155 set: exclusive
156 ===== ============== 156 ===== ==============
157 157
158 super_operations 158 super_operations
159 ================ 159 ================
160 160
161 prototypes:: 161 prototypes::
162 162
163 struct inode *(*alloc_inode)(struct su 163 struct inode *(*alloc_inode)(struct super_block *sb);
164 void (*free_inode)(struct inode *); 164 void (*free_inode)(struct inode *);
165 void (*destroy_inode)(struct inode *); 165 void (*destroy_inode)(struct inode *);
166 void (*dirty_inode) (struct inode *, i 166 void (*dirty_inode) (struct inode *, int flags);
167 int (*write_inode) (struct inode *, st 167 int (*write_inode) (struct inode *, struct writeback_control *wbc);
168 int (*drop_inode) (struct inode *); 168 int (*drop_inode) (struct inode *);
169 void (*evict_inode) (struct inode *); 169 void (*evict_inode) (struct inode *);
170 void (*put_super) (struct super_block 170 void (*put_super) (struct super_block *);
171 int (*sync_fs)(struct super_block *sb, 171 int (*sync_fs)(struct super_block *sb, int wait);
172 int (*freeze_fs) (struct super_block * 172 int (*freeze_fs) (struct super_block *);
173 int (*unfreeze_fs) (struct super_block 173 int (*unfreeze_fs) (struct super_block *);
174 int (*statfs) (struct dentry *, struct 174 int (*statfs) (struct dentry *, struct kstatfs *);
175 int (*remount_fs) (struct super_block 175 int (*remount_fs) (struct super_block *, int *, char *);
176 void (*umount_begin) (struct super_blo 176 void (*umount_begin) (struct super_block *);
177 int (*show_options)(struct seq_file *, 177 int (*show_options)(struct seq_file *, struct dentry *);
178 ssize_t (*quota_read)(struct super_blo 178 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
179 ssize_t (*quota_write)(struct super_bl 179 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
180 180
181 locking rules: 181 locking rules:
182 All may block [not true, see below] 182 All may block [not true, see below]
183 183
184 ====================== ============ ====== 184 ====================== ============ ========================
185 ops s_umount note 185 ops s_umount note
186 ====================== ============ ====== 186 ====================== ============ ========================
187 alloc_inode: 187 alloc_inode:
188 free_inode: called 188 free_inode: called from RCU callback
189 destroy_inode: 189 destroy_inode:
190 dirty_inode: 190 dirty_inode:
191 write_inode: 191 write_inode:
192 drop_inode: !!!ino 192 drop_inode: !!!inode->i_lock!!!
193 evict_inode: 193 evict_inode:
194 put_super: write 194 put_super: write
195 sync_fs: read 195 sync_fs: read
196 freeze_fs: write 196 freeze_fs: write
197 unfreeze_fs: write 197 unfreeze_fs: write
198 statfs: maybe(read) (see b 198 statfs: maybe(read) (see below)
199 remount_fs: write 199 remount_fs: write
200 umount_begin: no 200 umount_begin: no
201 show_options: no (names 201 show_options: no (namespace_sem)
202 quota_read: no (see b 202 quota_read: no (see below)
203 quota_write: no (see b 203 quota_write: no (see below)
204 ====================== ============ ====== 204 ====================== ============ ========================
205 205
206 ->statfs() has s_umount (shared) when called b 206 ->statfs() has s_umount (shared) when called by ustat(2) (native or
207 compat), but that's an accident of bad API; s_ 207 compat), but that's an accident of bad API; s_umount is used to pin
208 the superblock down when we only have dev_t gi 208 the superblock down when we only have dev_t given us by userland to
209 identify the superblock. Everything else (sta 209 identify the superblock. Everything else (statfs(), fstatfs(), etc.)
210 doesn't hold it when calling ->statfs() - supe 210 doesn't hold it when calling ->statfs() - superblock is pinned down
211 by resolving the pathname passed to syscall. 211 by resolving the pathname passed to syscall.
212 212
213 ->quota_read() and ->quota_write() functions a 213 ->quota_read() and ->quota_write() functions are both guaranteed to
214 be the only ones operating on the quota file b 214 be the only ones operating on the quota file by the quota code (via
215 dqio_sem) (unless an admin really wants to scr 215 dqio_sem) (unless an admin really wants to screw up something and
216 writes to quota files with quotas on). For oth 216 writes to quota files with quotas on). For other details about locking
217 see also dquot_operations section. 217 see also dquot_operations section.
218 218
219 file_system_type 219 file_system_type
220 ================ 220 ================
221 221
222 prototypes:: 222 prototypes::
223 223
224 struct dentry *(*mount) (struct file_s 224 struct dentry *(*mount) (struct file_system_type *, int,
225 const char *, void *); 225 const char *, void *);
226 void (*kill_sb) (struct super_block *) 226 void (*kill_sb) (struct super_block *);
227 227
228 locking rules: 228 locking rules:
229 229
230 ======= ========= 230 ======= =========
231 ops may block 231 ops may block
232 ======= ========= 232 ======= =========
233 mount yes 233 mount yes
234 kill_sb yes 234 kill_sb yes
235 ======= ========= 235 ======= =========
236 236
237 ->mount() returns ERR_PTR or the root dentry; 237 ->mount() returns ERR_PTR or the root dentry; its superblock should be locked
238 on return. 238 on return.
239 239
240 ->kill_sb() takes a write-locked superblock, d 240 ->kill_sb() takes a write-locked superblock, does all shutdown work on it,
241 unlocks and drops the reference. 241 unlocks and drops the reference.
242 242
243 address_space_operations 243 address_space_operations
244 ======================== 244 ========================
245 prototypes:: 245 prototypes::
246 246
247 int (*writepage)(struct page *page, st 247 int (*writepage)(struct page *page, struct writeback_control *wbc);
248 int (*read_folio)(struct file *, struc 248 int (*read_folio)(struct file *, struct folio *);
249 int (*writepages)(struct address_space 249 int (*writepages)(struct address_space *, struct writeback_control *);
250 bool (*dirty_folio)(struct address_spa 250 bool (*dirty_folio)(struct address_space *, struct folio *folio);
251 void (*readahead)(struct readahead_con 251 void (*readahead)(struct readahead_control *);
252 int (*write_begin)(struct file *, stru 252 int (*write_begin)(struct file *, struct address_space *mapping,
253 loff_t pos, un 253 loff_t pos, unsigned len,
254 struct folio * 254 struct folio **foliop, void **fsdata);
255 int (*write_end)(struct file *, struct 255 int (*write_end)(struct file *, struct address_space *mapping,
256 loff_t pos, un 256 loff_t pos, unsigned len, unsigned copied,
257 struct folio * 257 struct folio *folio, void *fsdata);
258 sector_t (*bmap)(struct address_space 258 sector_t (*bmap)(struct address_space *, sector_t);
259 void (*invalidate_folio) (struct folio 259 void (*invalidate_folio) (struct folio *, size_t start, size_t len);
260 bool (*release_folio)(struct folio *, 260 bool (*release_folio)(struct folio *, gfp_t);
261 void (*free_folio)(struct folio *); 261 void (*free_folio)(struct folio *);
262 int (*direct_IO)(struct kiocb *, struc 262 int (*direct_IO)(struct kiocb *, struct iov_iter *iter);
263 int (*migrate_folio)(struct address_sp 263 int (*migrate_folio)(struct address_space *, struct folio *dst,
264 struct folio *src, enu 264 struct folio *src, enum migrate_mode);
265 int (*launder_folio)(struct folio *); 265 int (*launder_folio)(struct folio *);
266 bool (*is_partially_uptodate)(struct f 266 bool (*is_partially_uptodate)(struct folio *, size_t from, size_t count);
267 int (*error_remove_folio)(struct addre 267 int (*error_remove_folio)(struct address_space *, struct folio *);
268 int (*swap_activate)(struct swap_info_ 268 int (*swap_activate)(struct swap_info_struct *sis, struct file *f, sector_t *span)
269 int (*swap_deactivate)(struct file *); 269 int (*swap_deactivate)(struct file *);
270 int (*swap_rw)(struct kiocb *iocb, str 270 int (*swap_rw)(struct kiocb *iocb, struct iov_iter *iter);
271 271
272 locking rules: 272 locking rules:
273 All except dirty_folio and free_folio 273 All except dirty_folio and free_folio may block
274 274
275 ====================== ====================== 275 ====================== ======================== ========= ===============
276 ops folio locked 276 ops folio locked i_rwsem invalidate_lock
277 ====================== ====================== 277 ====================== ======================== ========= ===============
278 writepage: yes, unlocks (see belo 278 writepage: yes, unlocks (see below)
279 read_folio: yes, unlocks 279 read_folio: yes, unlocks shared
280 writepages: 280 writepages:
281 dirty_folio: maybe 281 dirty_folio: maybe
282 readahead: yes, unlocks 282 readahead: yes, unlocks shared
283 write_begin: locks the folio 283 write_begin: locks the folio exclusive
284 write_end: yes, unlocks 284 write_end: yes, unlocks exclusive
285 bmap: 285 bmap:
286 invalidate_folio: yes 286 invalidate_folio: yes exclusive
287 release_folio: yes 287 release_folio: yes
288 free_folio: yes 288 free_folio: yes
289 direct_IO: 289 direct_IO:
290 migrate_folio: yes (both) 290 migrate_folio: yes (both)
291 launder_folio: yes 291 launder_folio: yes
292 is_partially_uptodate: yes 292 is_partially_uptodate: yes
293 error_remove_folio: yes 293 error_remove_folio: yes
294 swap_activate: no 294 swap_activate: no
295 swap_deactivate: no 295 swap_deactivate: no
296 swap_rw: yes, unlocks 296 swap_rw: yes, unlocks
297 ====================== ====================== 297 ====================== ======================== ========= ===============
298 298
299 ->write_begin(), ->write_end() and ->read_foli 299 ->write_begin(), ->write_end() and ->read_folio() may be called from
300 the request handler (/dev/loop). 300 the request handler (/dev/loop).
301 301
302 ->read_folio() unlocks the folio, either synch 302 ->read_folio() unlocks the folio, either synchronously or via I/O
303 completion. 303 completion.
304 304
305 ->readahead() unlocks the folios that I/O is a 305 ->readahead() unlocks the folios that I/O is attempted on like ->read_folio().
306 306
307 ->writepage() is used for two purposes: for "m 307 ->writepage() is used for two purposes: for "memory cleansing" and for
308 "sync". These are quite different operations 308 "sync". These are quite different operations and the behaviour may differ
309 depending upon the mode. 309 depending upon the mode.
310 310
311 If writepage is called for sync (wbc->sync_mod 311 If writepage is called for sync (wbc->sync_mode != WBC_SYNC_NONE) then
312 it *must* start I/O against the page, even if 312 it *must* start I/O against the page, even if that would involve
313 blocking on in-progress I/O. 313 blocking on in-progress I/O.
314 314
315 If writepage is called for memory cleansing (s 315 If writepage is called for memory cleansing (sync_mode ==
316 WBC_SYNC_NONE) then its role is to get as much 316 WBC_SYNC_NONE) then its role is to get as much writeout underway as
317 possible. So writepage should try to avoid bl 317 possible. So writepage should try to avoid blocking against
318 currently-in-progress I/O. 318 currently-in-progress I/O.
319 319
320 If the filesystem is not called for "sync" and 320 If the filesystem is not called for "sync" and it determines that it
321 would need to block against in-progress I/O to 321 would need to block against in-progress I/O to be able to start new I/O
322 against the page the filesystem should redirty 322 against the page the filesystem should redirty the page with
323 redirty_page_for_writepage(), then unlock the 323 redirty_page_for_writepage(), then unlock the page and return zero.
324 This may also be done to avoid internal deadlo 324 This may also be done to avoid internal deadlocks, but rarely.
325 325
326 If the filesystem is called for sync then it m 326 If the filesystem is called for sync then it must wait on any
327 in-progress I/O and then start new I/O. 327 in-progress I/O and then start new I/O.
328 328
329 The filesystem should unlock the page synchron 329 The filesystem should unlock the page synchronously, before returning to the
330 caller, unless ->writepage() returns special W 330 caller, unless ->writepage() returns special WRITEPAGE_ACTIVATE
331 value. WRITEPAGE_ACTIVATE means that page cann 331 value. WRITEPAGE_ACTIVATE means that page cannot really be written out
332 currently, and VM should stop calling ->writep 332 currently, and VM should stop calling ->writepage() on this page for some
333 time. VM does this by moving page to the head 333 time. VM does this by moving page to the head of the active list, hence the
334 name. 334 name.
335 335
336 Unless the filesystem is going to redirty_page 336 Unless the filesystem is going to redirty_page_for_writepage(), unlock the page
337 and return zero, writepage *must* run set_page 337 and return zero, writepage *must* run set_page_writeback() against the page,
338 followed by unlocking it. Once set_page_write 338 followed by unlocking it. Once set_page_writeback() has been run against the
339 page, write I/O can be submitted and the write 339 page, write I/O can be submitted and the write I/O completion handler must run
340 end_page_writeback() once the I/O is complete. 340 end_page_writeback() once the I/O is complete. If no I/O is submitted, the
341 filesystem must run end_page_writeback() again 341 filesystem must run end_page_writeback() against the page before returning from
342 writepage. 342 writepage.
343 343
344 That is: after 2.5.12, pages which are under w 344 That is: after 2.5.12, pages which are under writeout are *not* locked. Note,
345 if the filesystem needs the page to be locked 345 if the filesystem needs the page to be locked during writeout, that is ok, too,
346 the page is allowed to be unlocked at any poin 346 the page is allowed to be unlocked at any point in time between the calls to
347 set_page_writeback() and end_page_writeback(). 347 set_page_writeback() and end_page_writeback().
348 348
349 Note, failure to run either redirty_page_for_w 349 Note, failure to run either redirty_page_for_writepage() or the combination of
350 set_page_writeback()/end_page_writeback() on a 350 set_page_writeback()/end_page_writeback() on a page submitted to writepage
351 will leave the page itself marked clean but it 351 will leave the page itself marked clean but it will be tagged as dirty in the
352 radix tree. This incoherency can lead to all 352 radix tree. This incoherency can lead to all sorts of hard-to-debug problems
353 in the filesystem like having dirty inodes at 353 in the filesystem like having dirty inodes at umount and losing written data.
354 354
355 ->writepages() is used for periodic writeback 355 ->writepages() is used for periodic writeback and for syscall-initiated
356 sync operations. The address_space should sta 356 sync operations. The address_space should start I/O against at least
357 ``*nr_to_write`` pages. ``*nr_to_write`` must 357 ``*nr_to_write`` pages. ``*nr_to_write`` must be decremented for each page
358 which is written. The address_space implement 358 which is written. The address_space implementation may write more (or less)
359 pages than ``*nr_to_write`` asks for, but it s 359 pages than ``*nr_to_write`` asks for, but it should try to be reasonably close.
360 If nr_to_write is NULL, all dirty pages must b 360 If nr_to_write is NULL, all dirty pages must be written.
361 361
362 writepages should _only_ write pages which are 362 writepages should _only_ write pages which are present on
363 mapping->io_pages. 363 mapping->io_pages.
364 364
365 ->dirty_folio() is called from various places 365 ->dirty_folio() is called from various places in the kernel when
366 the target folio is marked as needing writebac 366 the target folio is marked as needing writeback. The folio cannot be
367 truncated because either the caller holds the 367 truncated because either the caller holds the folio lock, or the caller
368 has found the folio while holding the page tab 368 has found the folio while holding the page table lock which will block
369 truncation. 369 truncation.
370 370
371 ->bmap() is currently used by legacy ioctl() ( 371 ->bmap() is currently used by legacy ioctl() (FIBMAP) provided by some
372 filesystems and by the swapper. The latter wil 372 filesystems and by the swapper. The latter will eventually go away. Please,
373 keep it that way and don't breed new callers. 373 keep it that way and don't breed new callers.
374 374
375 ->invalidate_folio() is called when the filesy 375 ->invalidate_folio() is called when the filesystem must attempt to drop
376 some or all of the buffers from the page when 376 some or all of the buffers from the page when it is being truncated. It
377 returns zero on success. The filesystem must 377 returns zero on success. The filesystem must exclusively acquire
378 invalidate_lock before invalidating page cache 378 invalidate_lock before invalidating page cache in truncate / hole punch
379 path (and thus calling into ->invalidate_folio 379 path (and thus calling into ->invalidate_folio) to block races between page
380 cache invalidation and page cache filling func 380 cache invalidation and page cache filling functions (fault, read, ...).
381 381
382 ->release_folio() is called when the MM wants 382 ->release_folio() is called when the MM wants to make a change to the
383 folio that would invalidate the filesystem's p 383 folio that would invalidate the filesystem's private data. For example,
384 it may be about to be removed from the address 384 it may be about to be removed from the address_space or split. The folio
385 is locked and not under writeback. It may be 385 is locked and not under writeback. It may be dirty. The gfp parameter
386 is not usually used for allocation, but rather 386 is not usually used for allocation, but rather to indicate what the
387 filesystem may do to attempt to free the priva 387 filesystem may do to attempt to free the private data. The filesystem may
388 return false to indicate that the folio's priv 388 return false to indicate that the folio's private data cannot be freed.
389 If it returns true, it should have already rem 389 If it returns true, it should have already removed the private data from
390 the folio. If a filesystem does not provide a 390 the folio. If a filesystem does not provide a ->release_folio method,
391 the pagecache will assume that private data is 391 the pagecache will assume that private data is buffer_heads and call
392 try_to_free_buffers(). 392 try_to_free_buffers().
393 393
394 ->free_folio() is called when the kernel has d 394 ->free_folio() is called when the kernel has dropped the folio
395 from the page cache. 395 from the page cache.
396 396
397 ->launder_folio() may be called prior to relea 397 ->launder_folio() may be called prior to releasing a folio if
398 it is still found to be dirty. It returns zero 398 it is still found to be dirty. It returns zero if the folio was successfully
399 cleaned, or an error value if not. Note that i 399 cleaned, or an error value if not. Note that in order to prevent the folio
400 getting mapped back in and redirtied, it needs 400 getting mapped back in and redirtied, it needs to be kept locked
401 across the entire operation. 401 across the entire operation.
402 402
403 ->swap_activate() will be called to prepare th 403 ->swap_activate() will be called to prepare the given file for swap. It
404 should perform any validation and preparation 404 should perform any validation and preparation necessary to ensure that
405 writes can be performed with minimal memory al 405 writes can be performed with minimal memory allocation. It should call
406 add_swap_extent(), or the helper iomap_swapfil 406 add_swap_extent(), or the helper iomap_swapfile_activate(), and return
407 the number of extents added. If IO should be 407 the number of extents added. If IO should be submitted through
408 ->swap_rw(), it should set SWP_FS_OPS, otherwi 408 ->swap_rw(), it should set SWP_FS_OPS, otherwise IO will be submitted
409 directly to the block device ``sis->bdev``. 409 directly to the block device ``sis->bdev``.
410 410
411 ->swap_deactivate() will be called in the sys_ 411 ->swap_deactivate() will be called in the sys_swapoff()
412 path after ->swap_activate() returned success. 412 path after ->swap_activate() returned success.
413 413
414 ->swap_rw will be called for swap IO if SWP_FS 414 ->swap_rw will be called for swap IO if SWP_FS_OPS was set by ->swap_activate().
415 415
416 file_lock_operations 416 file_lock_operations
417 ==================== 417 ====================
418 418
419 prototypes:: 419 prototypes::
420 420
421 void (*fl_copy_lock)(struct file_lock 421 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
422 void (*fl_release_private)(struct file 422 void (*fl_release_private)(struct file_lock *);
423 423
424 424
425 locking rules: 425 locking rules:
426 426
427 =================== ============= ====== 427 =================== ============= =========
428 ops inode->i_lock may bl 428 ops inode->i_lock may block
429 =================== ============= ====== 429 =================== ============= =========
430 fl_copy_lock: yes no 430 fl_copy_lock: yes no
431 fl_release_private: maybe maybe[ 431 fl_release_private: maybe maybe[1]_
432 =================== ============= ====== 432 =================== ============= =========
433 433
434 .. [1]: 434 .. [1]:
435 ->fl_release_private for flock or POSIX loc 435 ->fl_release_private for flock or POSIX locks is currently allowed
436 to block. Leases however can still be freed 436 to block. Leases however can still be freed while the i_lock is held and
437 so fl_release_private called on a lease sho 437 so fl_release_private called on a lease should not block.
438 438
439 lock_manager_operations 439 lock_manager_operations
440 ======================= 440 =======================
441 441
442 prototypes:: 442 prototypes::
443 443
444 void (*lm_notify)(struct file_lock *); 444 void (*lm_notify)(struct file_lock *); /* unblock callback */
445 int (*lm_grant)(struct file_lock *, st 445 int (*lm_grant)(struct file_lock *, struct file_lock *, int);
446 void (*lm_break)(struct file_lock *); 446 void (*lm_break)(struct file_lock *); /* break_lease callback */
447 int (*lm_change)(struct file_lock **, 447 int (*lm_change)(struct file_lock **, int);
448 bool (*lm_breaker_owns_lease)(struct f 448 bool (*lm_breaker_owns_lease)(struct file_lock *);
449 bool (*lm_lock_expirable)(struct file_ 449 bool (*lm_lock_expirable)(struct file_lock *);
450 void (*lm_expire_lock)(void); 450 void (*lm_expire_lock)(void);
451 451
452 locking rules: 452 locking rules:
453 453
454 ====================== ============= ====== 454 ====================== ============= ================= =========
455 ops flc_lock blocke 455 ops flc_lock blocked_lock_lock may block
456 ====================== ============= ====== 456 ====================== ============= ================= =========
457 lm_notify: no yes 457 lm_notify: no yes no
458 lm_grant: no no 458 lm_grant: no no no
459 lm_break: yes no 459 lm_break: yes no no
460 lm_change yes no 460 lm_change yes no no
461 lm_breaker_owns_lease: yes no 461 lm_breaker_owns_lease: yes no no
462 lm_lock_expirable yes no 462 lm_lock_expirable yes no no
463 lm_expire_lock no no 463 lm_expire_lock no no yes
464 ====================== ============= ====== 464 ====================== ============= ================= =========
465 465
466 buffer_head 466 buffer_head
467 =========== 467 ===========
468 468
469 prototypes:: 469 prototypes::
470 470
471 void (*b_end_io)(struct buffer_head *b 471 void (*b_end_io)(struct buffer_head *bh, int uptodate);
472 472
473 locking rules: 473 locking rules:
474 474
475 called from interrupts. In other words, extrem 475 called from interrupts. In other words, extreme care is needed here.
476 bh is locked, but that's all warranties we hav 476 bh is locked, but that's all warranties we have here. Currently only RAID1,
477 highmem, fs/buffer.c, and fs/ntfs/aops.c are p 477 highmem, fs/buffer.c, and fs/ntfs/aops.c are providing these. Block devices
478 call this method upon the IO completion. 478 call this method upon the IO completion.
479 479
480 block_device_operations 480 block_device_operations
481 ======================= 481 =======================
482 prototypes:: 482 prototypes::
483 483
484 int (*open) (struct block_device *, fm 484 int (*open) (struct block_device *, fmode_t);
485 int (*release) (struct gendisk *, fmod 485 int (*release) (struct gendisk *, fmode_t);
486 int (*ioctl) (struct block_device *, f 486 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
487 int (*compat_ioctl) (struct block_devi 487 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
488 int (*direct_access) (struct block_dev 488 int (*direct_access) (struct block_device *, sector_t, void **,
489 unsigned long 489 unsigned long *);
490 void (*unlock_native_capacity) (struct 490 void (*unlock_native_capacity) (struct gendisk *);
491 int (*getgeo)(struct block_device *, s 491 int (*getgeo)(struct block_device *, struct hd_geometry *);
492 void (*swap_slot_free_notify) (struct 492 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
493 493
494 locking rules: 494 locking rules:
495 495
496 ======================= =================== 496 ======================= ===================
497 ops open_mutex 497 ops open_mutex
498 ======================= =================== 498 ======================= ===================
499 open: yes 499 open: yes
500 release: yes 500 release: yes
501 ioctl: no 501 ioctl: no
502 compat_ioctl: no 502 compat_ioctl: no
503 direct_access: no 503 direct_access: no
504 unlock_native_capacity: no 504 unlock_native_capacity: no
505 getgeo: no 505 getgeo: no
506 swap_slot_free_notify: no (see below) 506 swap_slot_free_notify: no (see below)
507 ======================= =================== 507 ======================= ===================
508 508
509 swap_slot_free_notify is called with swap_lock 509 swap_slot_free_notify is called with swap_lock and sometimes the page lock
510 held. 510 held.
511 511
512 512
513 file_operations 513 file_operations
514 =============== 514 ===============
515 515
516 prototypes:: 516 prototypes::
517 517
518 loff_t (*llseek) (struct file *, loff_ 518 loff_t (*llseek) (struct file *, loff_t, int);
519 ssize_t (*read) (struct file *, char _ 519 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
520 ssize_t (*write) (struct file *, const 520 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
521 ssize_t (*read_iter) (struct kiocb *, 521 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
522 ssize_t (*write_iter) (struct kiocb *, 522 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
523 int (*iopoll) (struct kiocb *kiocb, bo 523 int (*iopoll) (struct kiocb *kiocb, bool spin);
524 int (*iterate_shared) (struct file *, 524 int (*iterate_shared) (struct file *, struct dir_context *);
525 __poll_t (*poll) (struct file *, struc 525 __poll_t (*poll) (struct file *, struct poll_table_struct *);
526 long (*unlocked_ioctl) (struct file *, 526 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
527 long (*compat_ioctl) (struct file *, u 527 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
528 int (*mmap) (struct file *, struct vm_ 528 int (*mmap) (struct file *, struct vm_area_struct *);
529 int (*open) (struct inode *, struct fi 529 int (*open) (struct inode *, struct file *);
530 int (*flush) (struct file *); 530 int (*flush) (struct file *);
531 int (*release) (struct inode *, struct 531 int (*release) (struct inode *, struct file *);
532 int (*fsync) (struct file *, loff_t st 532 int (*fsync) (struct file *, loff_t start, loff_t end, int datasync);
533 int (*fasync) (int, struct file *, int 533 int (*fasync) (int, struct file *, int);
534 int (*lock) (struct file *, int, struc 534 int (*lock) (struct file *, int, struct file_lock *);
535 unsigned long (*get_unmapped_area)(str 535 unsigned long (*get_unmapped_area)(struct file *, unsigned long,
536 unsigned long, unsigne 536 unsigned long, unsigned long, unsigned long);
537 int (*check_flags)(int); 537 int (*check_flags)(int);
538 int (*flock) (struct file *, int, stru 538 int (*flock) (struct file *, int, struct file_lock *);
539 ssize_t (*splice_write)(struct pipe_in 539 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *,
540 size_t, unsigned int); 540 size_t, unsigned int);
541 ssize_t (*splice_read)(struct file *, 541 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *,
542 size_t, unsigned int); 542 size_t, unsigned int);
543 int (*setlease)(struct file *, long, s 543 int (*setlease)(struct file *, long, struct file_lock **, void **);
544 long (*fallocate)(struct file *, int, 544 long (*fallocate)(struct file *, int, loff_t, loff_t);
545 void (*show_fdinfo)(struct seq_file *m 545 void (*show_fdinfo)(struct seq_file *m, struct file *f);
546 unsigned (*mmap_capabilities)(struct f 546 unsigned (*mmap_capabilities)(struct file *);
547 ssize_t (*copy_file_range)(struct file 547 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
548 loff_t, size_t, unsign 548 loff_t, size_t, unsigned int);
549 loff_t (*remap_file_range)(struct file 549 loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
550 struct file *file_out, 550 struct file *file_out, loff_t pos_out,
551 loff_t len, unsigned i 551 loff_t len, unsigned int remap_flags);
552 int (*fadvise)(struct file *, loff_t, 552 int (*fadvise)(struct file *, loff_t, loff_t, int);
553 553
554 locking rules: 554 locking rules:
555 All may block. 555 All may block.
556 556
557 ->llseek() locking has moved from llseek to th 557 ->llseek() locking has moved from llseek to the individual llseek
558 implementations. If your fs is not using gene 558 implementations. If your fs is not using generic_file_llseek, you
559 need to acquire and release the appropriate lo 559 need to acquire and release the appropriate locks in your ->llseek().
560 For many filesystems, it is probably safe to a 560 For many filesystems, it is probably safe to acquire the inode
561 mutex or just to use i_size_read() instead. 561 mutex or just to use i_size_read() instead.
562 Note: this does not protect the file->f_pos ag 562 Note: this does not protect the file->f_pos against concurrent modifications
563 since this is something the userspace has to t 563 since this is something the userspace has to take care about.
564 564
565 ->iterate_shared() is called with i_rwsem held 565 ->iterate_shared() is called with i_rwsem held for reading, and with the
566 file f_pos_lock held exclusively 566 file f_pos_lock held exclusively
567 567
568 ->fasync() is responsible for maintaining the 568 ->fasync() is responsible for maintaining the FASYNC bit in filp->f_flags.
569 Most instances call fasync_helper(), which doe 569 Most instances call fasync_helper(), which does that maintenance, so it's
570 not normally something one needs to worry abou 570 not normally something one needs to worry about. Return values > 0 will be
571 mapped to zero in the VFS layer. 571 mapped to zero in the VFS layer.
572 572
573 ->readdir() and ->ioctl() on directories must 573 ->readdir() and ->ioctl() on directories must be changed. Ideally we would
574 move ->readdir() to inode_operations and use a 574 move ->readdir() to inode_operations and use a separate method for directory
575 ->ioctl() or kill the latter completely. One o 575 ->ioctl() or kill the latter completely. One of the problems is that for
576 anything that resembles union-mount we won't h 576 anything that resembles union-mount we won't have a struct file for all
577 components. And there are other reasons why th 577 components. And there are other reasons why the current interface is a mess...
578 578
579 ->read on directories probably must go away - 579 ->read on directories probably must go away - we should just enforce -EISDIR
580 in sys_read() and friends. 580 in sys_read() and friends.
581 581
582 ->setlease operations should call generic_setl 582 ->setlease operations should call generic_setlease() before or after setting
583 the lease within the individual filesystem to 583 the lease within the individual filesystem to record the result of the
584 operation 584 operation
585 585
586 ->fallocate implementation must be really care 586 ->fallocate implementation must be really careful to maintain page cache
587 consistency when punching holes or performing 587 consistency when punching holes or performing other operations that invalidate
588 page cache contents. Usually the filesystem ne 588 page cache contents. Usually the filesystem needs to call
589 truncate_inode_pages_range() to invalidate rel 589 truncate_inode_pages_range() to invalidate relevant range of the page cache.
590 However the filesystem usually also needs to u 590 However the filesystem usually also needs to update its internal (and on disk)
591 view of file offset -> disk block mapping. Unt 591 view of file offset -> disk block mapping. Until this update is finished, the
592 filesystem needs to block page faults and read 592 filesystem needs to block page faults and reads from reloading now-stale page
593 cache contents from the disk. Since VFS acquir 593 cache contents from the disk. Since VFS acquires mapping->invalidate_lock in
594 shared mode when loading pages from disk (file 594 shared mode when loading pages from disk (filemap_fault(), filemap_read(),
595 readahead paths), the fallocate implementation 595 readahead paths), the fallocate implementation must take the invalidate_lock to
596 prevent reloading. 596 prevent reloading.
597 597
598 ->copy_file_range and ->remap_file_range imple 598 ->copy_file_range and ->remap_file_range implementations need to serialize
599 against modifications of file data while the o 599 against modifications of file data while the operation is running. For
600 blocking changes through write(2) and similar 600 blocking changes through write(2) and similar operations inode->i_rwsem can be
601 used. To block changes to file contents via a 601 used. To block changes to file contents via a memory mapping during the
602 operation, the filesystem must take mapping->i 602 operation, the filesystem must take mapping->invalidate_lock to coordinate
603 with ->page_mkwrite. 603 with ->page_mkwrite.
604 604
605 dquot_operations 605 dquot_operations
606 ================ 606 ================
607 607
608 prototypes:: 608 prototypes::
609 609
610 int (*write_dquot) (struct dquot *); 610 int (*write_dquot) (struct dquot *);
611 int (*acquire_dquot) (struct dquot *); 611 int (*acquire_dquot) (struct dquot *);
612 int (*release_dquot) (struct dquot *); 612 int (*release_dquot) (struct dquot *);
613 int (*mark_dirty) (struct dquot *); 613 int (*mark_dirty) (struct dquot *);
614 int (*write_info) (struct super_block 614 int (*write_info) (struct super_block *, int);
615 615
616 These operations are intended to be more or le 616 These operations are intended to be more or less wrapping functions that ensure
617 a proper locking wrt the filesystem and call t 617 a proper locking wrt the filesystem and call the generic quota operations.
618 618
619 What filesystem should expect from the generic 619 What filesystem should expect from the generic quota functions:
620 620
621 ============== ============ ============== 621 ============== ============ =========================
622 ops FS recursion Held locks whe 622 ops FS recursion Held locks when called
623 ============== ============ ============== 623 ============== ============ =========================
624 write_dquot: yes dqonoff_sem or 624 write_dquot: yes dqonoff_sem or dqptr_sem
625 acquire_dquot: yes dqonoff_sem or 625 acquire_dquot: yes dqonoff_sem or dqptr_sem
626 release_dquot: yes dqonoff_sem or 626 release_dquot: yes dqonoff_sem or dqptr_sem
627 mark_dirty: no - 627 mark_dirty: no -
628 write_info: yes dqonoff_sem 628 write_info: yes dqonoff_sem
629 ============== ============ ============== 629 ============== ============ =========================
630 630
631 FS recursion means calling ->quota_read() and 631 FS recursion means calling ->quota_read() and ->quota_write() from superblock
632 operations. 632 operations.
633 633
634 More details about quota locking can be found 634 More details about quota locking can be found in fs/dquot.c.
635 635
636 vm_operations_struct 636 vm_operations_struct
637 ==================== 637 ====================
638 638
639 prototypes:: 639 prototypes::
640 640
641 void (*open)(struct vm_area_struct *); 641 void (*open)(struct vm_area_struct *);
642 void (*close)(struct vm_area_struct *) 642 void (*close)(struct vm_area_struct *);
643 vm_fault_t (*fault)(struct vm_fault *) 643 vm_fault_t (*fault)(struct vm_fault *);
644 vm_fault_t (*huge_fault)(struct vm_fau 644 vm_fault_t (*huge_fault)(struct vm_fault *, unsigned int order);
645 vm_fault_t (*map_pages)(struct vm_faul 645 vm_fault_t (*map_pages)(struct vm_fault *, pgoff_t start, pgoff_t end);
646 vm_fault_t (*page_mkwrite)(struct vm_a 646 vm_fault_t (*page_mkwrite)(struct vm_area_struct *, struct vm_fault *);
647 vm_fault_t (*pfn_mkwrite)(struct vm_ar 647 vm_fault_t (*pfn_mkwrite)(struct vm_area_struct *, struct vm_fault *);
648 int (*access)(struct vm_area_struct *, 648 int (*access)(struct vm_area_struct *, unsigned long, void*, int, int);
649 649
650 locking rules: 650 locking rules:
651 651
652 ============= ========== ============== 652 ============= ========== ===========================
653 ops mmap_lock PageLocked(pag 653 ops mmap_lock PageLocked(page)
654 ============= ========== ============== 654 ============= ========== ===========================
655 open: write 655 open: write
656 close: read/write 656 close: read/write
657 fault: read can return wit 657 fault: read can return with page locked
658 huge_fault: maybe-read 658 huge_fault: maybe-read
659 map_pages: maybe-read 659 map_pages: maybe-read
660 page_mkwrite: read can return wit 660 page_mkwrite: read can return with page locked
661 pfn_mkwrite: read 661 pfn_mkwrite: read
662 access: read 662 access: read
663 ============= ========== ============== 663 ============= ========== ===========================
664 664
665 ->fault() is called when a previously not pres 665 ->fault() is called when a previously not present pte is about to be faulted
666 in. The filesystem must find and return the pa 666 in. The filesystem must find and return the page associated with the passed in
667 "pgoff" in the vm_fault structure. If it is po 667 "pgoff" in the vm_fault structure. If it is possible that the page may be
668 truncated and/or invalidated, then the filesys 668 truncated and/or invalidated, then the filesystem must lock invalidate_lock,
669 then ensure the page is not already truncated 669 then ensure the page is not already truncated (invalidate_lock will block
670 subsequent truncate), and then return with VM_ 670 subsequent truncate), and then return with VM_FAULT_LOCKED, and the page
671 locked. The VM will unlock the page. 671 locked. The VM will unlock the page.
672 672
673 ->huge_fault() is called when there is no PUD 673 ->huge_fault() is called when there is no PUD or PMD entry present. This
674 gives the filesystem the opportunity to instal 674 gives the filesystem the opportunity to install a PUD or PMD sized page.
675 Filesystems can also use the ->fault method to 675 Filesystems can also use the ->fault method to return a PMD sized page,
676 so implementing this function may not be neces 676 so implementing this function may not be necessary. In particular,
677 filesystems should not call filemap_fault() fr 677 filesystems should not call filemap_fault() from ->huge_fault().
678 The mmap_lock may not be held when this method 678 The mmap_lock may not be held when this method is called.
679 679
680 ->map_pages() is called when VM asks to map ea 680 ->map_pages() is called when VM asks to map easy accessible pages.
681 Filesystem should find and map pages associate 681 Filesystem should find and map pages associated with offsets from "start_pgoff"
682 till "end_pgoff". ->map_pages() is called with 682 till "end_pgoff". ->map_pages() is called with the RCU lock held and must
683 not block. If it's not possible to reach a pa 683 not block. If it's not possible to reach a page without blocking,
684 filesystem should skip it. Filesystem should u 684 filesystem should skip it. Filesystem should use set_pte_range() to setup
685 page table entry. Pointer to entry associated 685 page table entry. Pointer to entry associated with the page is passed in
686 "pte" field in vm_fault structure. Pointers to 686 "pte" field in vm_fault structure. Pointers to entries for other offsets
687 should be calculated relative to "pte". 687 should be calculated relative to "pte".
688 688
689 ->page_mkwrite() is called when a previously r 689 ->page_mkwrite() is called when a previously read-only pte is about to become
690 writeable. The filesystem again must ensure th 690 writeable. The filesystem again must ensure that there are no
691 truncate/invalidate races or races with operat 691 truncate/invalidate races or races with operations such as ->remap_file_range
692 or ->copy_file_range, and then return with the 692 or ->copy_file_range, and then return with the page locked. Usually
693 mapping->invalidate_lock is suitable for prope 693 mapping->invalidate_lock is suitable for proper serialization. If the page has
694 been truncated, the filesystem should not look 694 been truncated, the filesystem should not look up a new page like the ->fault()
695 handler, but simply return with VM_FAULT_NOPAG 695 handler, but simply return with VM_FAULT_NOPAGE, which will cause the VM to
696 retry the fault. 696 retry the fault.
697 697
698 ->pfn_mkwrite() is the same as page_mkwrite bu 698 ->pfn_mkwrite() is the same as page_mkwrite but when the pte is
699 VM_PFNMAP or VM_MIXEDMAP with a page-less entr 699 VM_PFNMAP or VM_MIXEDMAP with a page-less entry. Expected return is
700 VM_FAULT_NOPAGE. Or one of the VM_FAULT_ERROR 700 VM_FAULT_NOPAGE. Or one of the VM_FAULT_ERROR types. The default behavior
701 after this call is to make the pte read-write, 701 after this call is to make the pte read-write, unless pfn_mkwrite returns
702 an error. 702 an error.
703 703
704 ->access() is called when get_user_pages() fai 704 ->access() is called when get_user_pages() fails in
705 access_process_vm(), typically used to debug a 705 access_process_vm(), typically used to debug a process through
706 /proc/pid/mem or ptrace. This function is nee 706 /proc/pid/mem or ptrace. This function is needed only for
707 VM_IO | VM_PFNMAP VMAs. 707 VM_IO | VM_PFNMAP VMAs.
708 708
709 ---------------------------------------------- 709 --------------------------------------------------------------------------------
710 710
711 Dubious stuff 711 Dubious stuff
712 712
713 (if you break something or notice that it is b 713 (if you break something or notice that it is broken and do not fix it yourself
714 - at least put it here) 714 - at least put it here)
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.