1 ==================== 1 ====================
2 Changes since 2.5.0: 2 Changes since 2.5.0:
3 ==================== 3 ====================
4 4
5 --- 5 ---
6 6
7 **recommended** 7 **recommended**
8 8
9 New helpers: sb_bread(), sb_getblk(), sb_find_ 9 New helpers: sb_bread(), sb_getblk(), sb_find_get_block(), set_bh(),
10 sb_set_blocksize() and sb_min_blocksize(). 10 sb_set_blocksize() and sb_min_blocksize().
11 11
12 Use them. 12 Use them.
13 13
14 (sb_find_get_block() replaces 2.4's get_hash_t 14 (sb_find_get_block() replaces 2.4's get_hash_table())
15 15
16 --- 16 ---
17 17
18 **recommended** 18 **recommended**
19 19
20 New methods: ->alloc_inode() and ->destroy_ino 20 New methods: ->alloc_inode() and ->destroy_inode().
21 21
22 Remove inode->u.foo_inode_i 22 Remove inode->u.foo_inode_i
23 23
24 Declare:: 24 Declare::
25 25
26 struct foo_inode_info { 26 struct foo_inode_info {
27 /* fs-private stuff */ 27 /* fs-private stuff */
28 struct inode vfs_inode; 28 struct inode vfs_inode;
29 }; 29 };
30 static inline struct foo_inode_info *F 30 static inline struct foo_inode_info *FOO_I(struct inode *inode)
31 { 31 {
32 return list_entry(inode, struc 32 return list_entry(inode, struct foo_inode_info, vfs_inode);
33 } 33 }
34 34
35 Use FOO_I(inode) instead of &inode->u.foo_inod 35 Use FOO_I(inode) instead of &inode->u.foo_inode_i;
36 36
37 Add foo_alloc_inode() and foo_destroy_inode() 37 Add foo_alloc_inode() and foo_destroy_inode() - the former should allocate
38 foo_inode_info and return the address of ->vfs 38 foo_inode_info and return the address of ->vfs_inode, the latter should free
39 FOO_I(inode) (see in-tree filesystems for exam 39 FOO_I(inode) (see in-tree filesystems for examples).
40 40
41 Make them ->alloc_inode and ->destroy_inode in 41 Make them ->alloc_inode and ->destroy_inode in your super_operations.
42 42
43 Keep in mind that now you need explicit initia 43 Keep in mind that now you need explicit initialization of private data
44 typically between calling iget_locked() and un 44 typically between calling iget_locked() and unlocking the inode.
45 45
46 At some point that will become mandatory. 46 At some point that will become mandatory.
47 47
48 **mandatory** <<
49 <<
50 The foo_inode_info should always be allocated <<
51 than kmem_cache_alloc() or kmalloc() related t <<
52 correctly. <<
53 <<
54 --- 48 ---
55 49
56 **mandatory** 50 **mandatory**
57 51
58 Change of file_system_type method (->read_supe 52 Change of file_system_type method (->read_super to ->get_sb)
59 53
60 ->read_super() is no more. Ditto for DECLARE_ 54 ->read_super() is no more. Ditto for DECLARE_FSTYPE and DECLARE_FSTYPE_DEV.
61 55
62 Turn your foo_read_super() into a function tha 56 Turn your foo_read_super() into a function that would return 0 in case of
63 success and negative number in case of error ( 57 success and negative number in case of error (-EINVAL unless you have more
64 informative error value to report). Call it f 58 informative error value to report). Call it foo_fill_super(). Now declare::
65 59
66 int foo_get_sb(struct file_system_type *fs_t 60 int foo_get_sb(struct file_system_type *fs_type,
67 int flags, const char *dev_name, void 61 int flags, const char *dev_name, void *data, struct vfsmount *mnt)
68 { 62 {
69 return get_sb_bdev(fs_type, flags, dev 63 return get_sb_bdev(fs_type, flags, dev_name, data, foo_fill_super,
70 mnt); 64 mnt);
71 } 65 }
72 66
73 (or similar with s/bdev/nodev/ or s/bdev/singl 67 (or similar with s/bdev/nodev/ or s/bdev/single/, depending on the kind of
74 filesystem). 68 filesystem).
75 69
76 Replace DECLARE_FSTYPE... with explicit initia 70 Replace DECLARE_FSTYPE... with explicit initializer and have ->get_sb set as
77 foo_get_sb. 71 foo_get_sb.
78 72
79 --- 73 ---
80 74
81 **mandatory** 75 **mandatory**
82 76
83 Locking change: ->s_vfs_rename_sem is taken on 77 Locking change: ->s_vfs_rename_sem is taken only by cross-directory renames.
84 Most likely there is no need to change anythin 78 Most likely there is no need to change anything, but if you relied on
85 global exclusion between renames for some inte 79 global exclusion between renames for some internal purpose - you need to
86 change your internal locking. Otherwise exclu 80 change your internal locking. Otherwise exclusion warranties remain the
87 same (i.e. parents and victim are locked, etc. 81 same (i.e. parents and victim are locked, etc.).
88 82
89 --- 83 ---
90 84
91 **informational** 85 **informational**
92 86
93 Now we have the exclusion between ->lookup() a 87 Now we have the exclusion between ->lookup() and directory removal (by
94 ->rmdir() and ->rename()). If you used to nee 88 ->rmdir() and ->rename()). If you used to need that exclusion and do
95 it by internal locking (most of filesystems co 89 it by internal locking (most of filesystems couldn't care less) - you
96 can relax your locking. 90 can relax your locking.
97 91
98 --- 92 ---
99 93
100 **mandatory** 94 **mandatory**
101 95
102 ->lookup(), ->truncate(), ->create(), ->unlink 96 ->lookup(), ->truncate(), ->create(), ->unlink(), ->mknod(), ->mkdir(),
103 ->rmdir(), ->link(), ->lseek(), ->symlink(), - 97 ->rmdir(), ->link(), ->lseek(), ->symlink(), ->rename()
104 and ->readdir() are called without BKL now. G 98 and ->readdir() are called without BKL now. Grab it on entry, drop upon return
105 - that will guarantee the same locking you use 99 - that will guarantee the same locking you used to have. If your method or its
106 parts do not need BKL - better yet, now you ca 100 parts do not need BKL - better yet, now you can shift lock_kernel() and
107 unlock_kernel() so that they would protect exa 101 unlock_kernel() so that they would protect exactly what needs to be
108 protected. 102 protected.
109 103
110 --- 104 ---
111 105
112 **mandatory** 106 **mandatory**
113 107
114 BKL is also moved from around sb operations. B 108 BKL is also moved from around sb operations. BKL should have been shifted into
115 individual fs sb_op functions. If you don't n 109 individual fs sb_op functions. If you don't need it, remove it.
116 110
117 --- 111 ---
118 112
119 **informational** 113 **informational**
120 114
121 check for ->link() target not being a director 115 check for ->link() target not being a directory is done by callers. Feel
122 free to drop it... 116 free to drop it...
123 117
124 --- 118 ---
125 119
126 **informational** 120 **informational**
127 121
128 ->link() callers hold ->i_mutex on the object 122 ->link() callers hold ->i_mutex on the object we are linking to. Some of your
129 problems might be over... 123 problems might be over...
130 124
131 --- 125 ---
132 126
133 **mandatory** 127 **mandatory**
134 128
135 new file_system_type method - kill_sb(superblo 129 new file_system_type method - kill_sb(superblock). If you are converting
136 an existing filesystem, set it according to -> 130 an existing filesystem, set it according to ->fs_flags::
137 131
138 FS_REQUIRES_DEV - kill_b 132 FS_REQUIRES_DEV - kill_block_super
139 FS_LITTER - kill_l 133 FS_LITTER - kill_litter_super
140 neither - kill_a 134 neither - kill_anon_super
141 135
142 FS_LITTER is gone - just remove it from fs_fla 136 FS_LITTER is gone - just remove it from fs_flags.
143 137
144 --- 138 ---
145 139
146 **mandatory** 140 **mandatory**
147 141
148 FS_SINGLE is gone (actually, that had happened 142 FS_SINGLE is gone (actually, that had happened back when ->get_sb()
149 went in - and hadn't been documented ;-/). Ju 143 went in - and hadn't been documented ;-/). Just remove it from fs_flags
150 (and see ->get_sb() entry for other actions). 144 (and see ->get_sb() entry for other actions).
151 145
152 --- 146 ---
153 147
154 **mandatory** 148 **mandatory**
155 149
156 ->setattr() is called without BKL now. Caller 150 ->setattr() is called without BKL now. Caller _always_ holds ->i_mutex, so
157 watch for ->i_mutex-grabbing code that might b 151 watch for ->i_mutex-grabbing code that might be used by your ->setattr().
158 Callers of notify_change() need ->i_mutex now. 152 Callers of notify_change() need ->i_mutex now.
159 153
160 --- 154 ---
161 155
162 **recommended** 156 **recommended**
163 157
164 New super_block field ``struct export_operatio 158 New super_block field ``struct export_operations *s_export_op`` for
165 explicit support for exporting, e.g. via NFS. 159 explicit support for exporting, e.g. via NFS. The structure is fully
166 documented at its declaration in include/linux 160 documented at its declaration in include/linux/fs.h, and in
167 Documentation/filesystems/nfs/exporting.rst. 161 Documentation/filesystems/nfs/exporting.rst.
168 162
169 Briefly it allows for the definition of decode 163 Briefly it allows for the definition of decode_fh and encode_fh operations
170 to encode and decode filehandles, and allows t 164 to encode and decode filehandles, and allows the filesystem to use
171 a standard helper function for decode_fh, and 165 a standard helper function for decode_fh, and provide file-system specific
172 support for this helper, particularly get_pare 166 support for this helper, particularly get_parent.
173 167
174 It is planned that this will be required for e 168 It is planned that this will be required for exporting once the code
175 settles down a bit. 169 settles down a bit.
176 170
177 **mandatory** 171 **mandatory**
178 172
179 s_export_op is now required for exporting a fi 173 s_export_op is now required for exporting a filesystem.
180 isofs, ext2, ext3, reiserfs, fat !! 174 isofs, ext2, ext3, resierfs, fat
181 can be used as examples of very different file 175 can be used as examples of very different filesystems.
182 176
183 --- 177 ---
184 178
185 **mandatory** 179 **mandatory**
186 180
187 iget4() and the read_inode2 callback have been 181 iget4() and the read_inode2 callback have been superseded by iget5_locked()
188 which has the following prototype:: 182 which has the following prototype::
189 183
190 struct inode *iget5_locked(struct super_bl 184 struct inode *iget5_locked(struct super_block *sb, unsigned long ino,
191 int (*test)(st 185 int (*test)(struct inode *, void *),
192 int (*set)(str 186 int (*set)(struct inode *, void *),
193 void *data); 187 void *data);
194 188
195 'test' is an additional function that can be u 189 'test' is an additional function that can be used when the inode
196 number is not sufficient to identify the actua 190 number is not sufficient to identify the actual file object. 'set'
197 should be a non-blocking function that initial 191 should be a non-blocking function that initializes those parts of a
198 newly created inode to allow the test function 192 newly created inode to allow the test function to succeed. 'data' is
199 passed as an opaque value to both test and set 193 passed as an opaque value to both test and set functions.
200 194
201 When the inode has been created by iget5_locke 195 When the inode has been created by iget5_locked(), it will be returned with the
202 I_NEW flag set and will still be locked. The 196 I_NEW flag set and will still be locked. The filesystem then needs to finalize
203 the initialization. Once the inode is initiali 197 the initialization. Once the inode is initialized it must be unlocked by
204 calling unlock_new_inode(). 198 calling unlock_new_inode().
205 199
206 The filesystem is responsible for setting (and 200 The filesystem is responsible for setting (and possibly testing) i_ino
207 when appropriate. There is also a simpler iget 201 when appropriate. There is also a simpler iget_locked function that
208 just takes the superblock and inode number as 202 just takes the superblock and inode number as arguments and does the
209 test and set for you. 203 test and set for you.
210 204
211 e.g.:: 205 e.g.::
212 206
213 inode = iget_locked(sb, ino); 207 inode = iget_locked(sb, ino);
214 if (inode->i_state & I_NEW) { 208 if (inode->i_state & I_NEW) {
215 err = read_inode_from_disk(ino 209 err = read_inode_from_disk(inode);
216 if (err < 0) { 210 if (err < 0) {
217 iget_failed(inode); 211 iget_failed(inode);
218 return err; 212 return err;
219 } 213 }
220 unlock_new_inode(inode); 214 unlock_new_inode(inode);
221 } 215 }
222 216
223 Note that if the process of setting up a new i 217 Note that if the process of setting up a new inode fails, then iget_failed()
224 should be called on the inode to render it dea 218 should be called on the inode to render it dead, and an appropriate error
225 should be passed back to the caller. 219 should be passed back to the caller.
226 220
227 --- 221 ---
228 222
229 **recommended** 223 **recommended**
230 224
231 ->getattr() finally getting used. See instanc 225 ->getattr() finally getting used. See instances in nfs, minix, etc.
232 226
233 --- 227 ---
234 228
235 **mandatory** 229 **mandatory**
236 230
237 ->revalidate() is gone. If your filesystem ha 231 ->revalidate() is gone. If your filesystem had it - provide ->getattr()
238 and let it call whatever you had as ->revlidat 232 and let it call whatever you had as ->revlidate() + (for symlinks that
239 had ->revalidate()) add calls in ->follow_link 233 had ->revalidate()) add calls in ->follow_link()/->readlink().
240 234
241 --- 235 ---
242 236
243 **mandatory** 237 **mandatory**
244 238
245 ->d_parent changes are not protected by BKL an 239 ->d_parent changes are not protected by BKL anymore. Read access is safe
246 if at least one of the following is true: 240 if at least one of the following is true:
247 241
248 * filesystem has no cross-directory re 242 * filesystem has no cross-directory rename()
249 * we know that parent had been locked 243 * we know that parent had been locked (e.g. we are looking at
250 ->d_parent of ->lookup() argument). 244 ->d_parent of ->lookup() argument).
251 * we are called from ->rename(). 245 * we are called from ->rename().
252 * the child's ->d_lock is held 246 * the child's ->d_lock is held
253 247
254 Audit your code and add locking if needed. No 248 Audit your code and add locking if needed. Notice that any place that is
255 not protected by the conditions above is risky 249 not protected by the conditions above is risky even in the old tree - you
256 had been relying on BKL and that's prone to sc 250 had been relying on BKL and that's prone to screwups. Old tree had quite
257 a few holes of that kind - unprotected access 251 a few holes of that kind - unprotected access to ->d_parent leading to
258 anything from oops to silent memory corruption 252 anything from oops to silent memory corruption.
259 253
260 --- 254 ---
261 255
262 **mandatory** 256 **mandatory**
263 257
264 FS_NOMOUNT is gone. If you use it - just set 258 FS_NOMOUNT is gone. If you use it - just set SB_NOUSER in flags
265 (see rootfs for one kind of solution and bdev/ 259 (see rootfs for one kind of solution and bdev/socket/pipe for another).
266 260
267 --- 261 ---
268 262
269 **recommended** 263 **recommended**
270 264
271 Use bdev_read_only(bdev) instead of is_read_on 265 Use bdev_read_only(bdev) instead of is_read_only(kdev). The latter
272 is still alive, but only because of the mess i 266 is still alive, but only because of the mess in drivers/s390/block/dasd.c.
273 As soon as it gets fixed is_read_only() will d 267 As soon as it gets fixed is_read_only() will die.
274 268
275 --- 269 ---
276 270
277 **mandatory** 271 **mandatory**
278 272
279 ->permission() is called without BKL now. Grab 273 ->permission() is called without BKL now. Grab it on entry, drop upon
280 return - that will guarantee the same locking 274 return - that will guarantee the same locking you used to have. If
281 your method or its parts do not need BKL - bet 275 your method or its parts do not need BKL - better yet, now you can
282 shift lock_kernel() and unlock_kernel() so tha 276 shift lock_kernel() and unlock_kernel() so that they would protect
283 exactly what needs to be protected. 277 exactly what needs to be protected.
284 278
285 --- 279 ---
286 280
287 **mandatory** 281 **mandatory**
288 282
289 ->statfs() is now called without BKL held. BK 283 ->statfs() is now called without BKL held. BKL should have been
290 shifted into individual fs sb_op functions whe 284 shifted into individual fs sb_op functions where it's not clear that
291 it's safe to remove it. If you don't need it, 285 it's safe to remove it. If you don't need it, remove it.
292 286
293 --- 287 ---
294 288
295 **mandatory** 289 **mandatory**
296 290
297 is_read_only() is gone; use bdev_read_only() i 291 is_read_only() is gone; use bdev_read_only() instead.
298 292
299 --- 293 ---
300 294
301 **mandatory** 295 **mandatory**
302 296
303 destroy_buffers() is gone; use invalidate_bdev 297 destroy_buffers() is gone; use invalidate_bdev().
304 298
305 --- 299 ---
306 300
307 **mandatory** 301 **mandatory**
308 302
309 fsync_dev() is gone; use fsync_bdev(). NOTE: 303 fsync_dev() is gone; use fsync_bdev(). NOTE: lvm breakage is
310 deliberate; as soon as struct block_device * i 304 deliberate; as soon as struct block_device * is propagated in a reasonable
311 way by that code fixing will become trivial; u 305 way by that code fixing will become trivial; until then nothing can be
312 done. 306 done.
313 307
314 **mandatory** 308 **mandatory**
315 309
316 block truncatation on error exit from ->write_ 310 block truncatation on error exit from ->write_begin, and ->direct_IO
317 moved from generic methods (block_write_begin, 311 moved from generic methods (block_write_begin, cont_write_begin,
318 nobh_write_begin, blockdev_direct_IO*) to call 312 nobh_write_begin, blockdev_direct_IO*) to callers. Take a look at
319 ext2_write_failed and callers for an example. 313 ext2_write_failed and callers for an example.
320 314
321 **mandatory** 315 **mandatory**
322 316
323 ->truncate is gone. The whole truncate sequen 317 ->truncate is gone. The whole truncate sequence needs to be
324 implemented in ->setattr, which is now mandato 318 implemented in ->setattr, which is now mandatory for filesystems
325 implementing on-disk size changes. Start with 319 implementing on-disk size changes. Start with a copy of the old inode_setattr
326 and vmtruncate, and the reorder the vmtruncate 320 and vmtruncate, and the reorder the vmtruncate + foofs_vmtruncate sequence to
327 be in order of zeroing blocks using block_trun 321 be in order of zeroing blocks using block_truncate_page or similar helpers,
328 size update and on finally on-disk truncation 322 size update and on finally on-disk truncation which should not fail.
329 setattr_prepare (which used to be inode_change 323 setattr_prepare (which used to be inode_change_ok) now includes the size checks
330 for ATTR_SIZE and must be called in the beginn 324 for ATTR_SIZE and must be called in the beginning of ->setattr unconditionally.
331 325
332 **mandatory** 326 **mandatory**
333 327
334 ->clear_inode() and ->delete_inode() are gone; 328 ->clear_inode() and ->delete_inode() are gone; ->evict_inode() should
335 be used instead. It gets called whenever the 329 be used instead. It gets called whenever the inode is evicted, whether it has
336 remaining links or not. Caller does *not* evi 330 remaining links or not. Caller does *not* evict the pagecache or inode-associated
337 metadata buffers; the method has to use trunca 331 metadata buffers; the method has to use truncate_inode_pages_final() to get rid
338 of those. Caller makes sure async writeback ca 332 of those. Caller makes sure async writeback cannot be running for the inode while
339 (or after) ->evict_inode() is called. 333 (or after) ->evict_inode() is called.
340 334
341 ->drop_inode() returns int now; it's called on 335 ->drop_inode() returns int now; it's called on final iput() with
342 inode->i_lock held and it returns true if file 336 inode->i_lock held and it returns true if filesystems wants the inode to be
343 dropped. As before, generic_drop_inode() is s 337 dropped. As before, generic_drop_inode() is still the default and it's been
344 updated appropriately. generic_delete_inode() 338 updated appropriately. generic_delete_inode() is also alive and it consists
345 simply of return 1. Note that all actual evic 339 simply of return 1. Note that all actual eviction work is done by caller after
346 ->drop_inode() returns. 340 ->drop_inode() returns.
347 341
348 As before, clear_inode() must be called exactl 342 As before, clear_inode() must be called exactly once on each call of
349 ->evict_inode() (as it used to be for each cal 343 ->evict_inode() (as it used to be for each call of ->delete_inode()). Unlike
350 before, if you are using inode-associated meta 344 before, if you are using inode-associated metadata buffers (i.e.
351 mark_buffer_dirty_inode()), it's your responsi 345 mark_buffer_dirty_inode()), it's your responsibility to call
352 invalidate_inode_buffers() before clear_inode( 346 invalidate_inode_buffers() before clear_inode().
353 347
354 NOTE: checking i_nlink in the beginning of ->w 348 NOTE: checking i_nlink in the beginning of ->write_inode() and bailing out
355 if it's zero is not *and* *never* *had* *been* 349 if it's zero is not *and* *never* *had* *been* enough. Final unlink() and iput()
356 may happen while the inode is in the middle of 350 may happen while the inode is in the middle of ->write_inode(); e.g. if you blindly
357 free the on-disk inode, you may end up doing t 351 free the on-disk inode, you may end up doing that while ->write_inode() is writing
358 to it. 352 to it.
359 353
360 --- 354 ---
361 355
362 **mandatory** 356 **mandatory**
363 357
364 .d_delete() now only advises the dcache as to 358 .d_delete() now only advises the dcache as to whether or not to cache
365 unreferenced dentries, and is now only called 359 unreferenced dentries, and is now only called when the dentry refcount goes to
366 0. Even on 0 refcount transition, it must be a 360 0. Even on 0 refcount transition, it must be able to tolerate being called 0,
367 1, or more times (eg. constant, idempotent). 361 1, or more times (eg. constant, idempotent).
368 362
369 --- 363 ---
370 364
371 **mandatory** 365 **mandatory**
372 366
373 .d_compare() calling convention and locking ru 367 .d_compare() calling convention and locking rules are significantly
374 changed. Read updated documentation in Documen 368 changed. Read updated documentation in Documentation/filesystems/vfs.rst (and
375 look at examples of other filesystems) for gui 369 look at examples of other filesystems) for guidance.
376 370
377 --- 371 ---
378 372
379 **mandatory** 373 **mandatory**
380 374
381 .d_hash() calling convention and locking rules 375 .d_hash() calling convention and locking rules are significantly
382 changed. Read updated documentation in Documen 376 changed. Read updated documentation in Documentation/filesystems/vfs.rst (and
383 look at examples of other filesystems) for gui 377 look at examples of other filesystems) for guidance.
384 378
385 --- 379 ---
386 380
387 **mandatory** 381 **mandatory**
388 382
389 dcache_lock is gone, replaced by fine grained 383 dcache_lock is gone, replaced by fine grained locks. See fs/dcache.c
390 for details of what locks to replace dcache_lo 384 for details of what locks to replace dcache_lock with in order to protect
391 particular things. Most of the time, a filesys 385 particular things. Most of the time, a filesystem only needs ->d_lock, which
392 protects *all* the dcache state of a given den 386 protects *all* the dcache state of a given dentry.
393 387
394 --- 388 ---
395 389
396 **mandatory** 390 **mandatory**
397 391
398 Filesystems must RCU-free their inodes, if the 392 Filesystems must RCU-free their inodes, if they can have been accessed
399 via rcu-walk path walk (basically, if the file 393 via rcu-walk path walk (basically, if the file can have had a path name in the
400 vfs namespace). 394 vfs namespace).
401 395
402 Even though i_dentry and i_rcu share storage i 396 Even though i_dentry and i_rcu share storage in a union, we will
403 initialize the former in inode_init_always(), 397 initialize the former in inode_init_always(), so just leave it alone in
404 the callback. It used to be necessary to clea 398 the callback. It used to be necessary to clean it there, but not anymore
405 (starting at 3.2). 399 (starting at 3.2).
406 400
407 --- 401 ---
408 402
409 **recommended** 403 **recommended**
410 404
411 vfs now tries to do path walking in "rcu-walk 405 vfs now tries to do path walking in "rcu-walk mode", which avoids
412 atomic operations and scalability hazards on d 406 atomic operations and scalability hazards on dentries and inodes (see
413 Documentation/filesystems/path-lookup.txt). d_ 407 Documentation/filesystems/path-lookup.txt). d_hash and d_compare changes
414 (above) are examples of the changes required t 408 (above) are examples of the changes required to support this. For more complex
415 filesystem callbacks, the vfs drops out of rcu 409 filesystem callbacks, the vfs drops out of rcu-walk mode before the fs call, so
416 no changes are required to the filesystem. How 410 no changes are required to the filesystem. However, this is costly and loses
417 the benefits of rcu-walk mode. We will begin t 411 the benefits of rcu-walk mode. We will begin to add filesystem callbacks that
418 are rcu-walk aware, shown below. Filesystems s 412 are rcu-walk aware, shown below. Filesystems should take advantage of this
419 where possible. 413 where possible.
420 414
421 --- 415 ---
422 416
423 **mandatory** 417 **mandatory**
424 418
425 d_revalidate is a callback that is made on eve 419 d_revalidate is a callback that is made on every path element (if
426 the filesystem provides it), which requires dr 420 the filesystem provides it), which requires dropping out of rcu-walk mode. This
427 may now be called in rcu-walk mode (nd->flags 421 may now be called in rcu-walk mode (nd->flags & LOOKUP_RCU). -ECHILD should be
428 returned if the filesystem cannot handle rcu-w 422 returned if the filesystem cannot handle rcu-walk. See
429 Documentation/filesystems/vfs.rst for more det 423 Documentation/filesystems/vfs.rst for more details.
430 424
431 permission is an inode permission check that i 425 permission is an inode permission check that is called on many or all
432 directory inodes on the way down a path walk ( 426 directory inodes on the way down a path walk (to check for exec permission). It
433 must now be rcu-walk aware (mask & MAY_NOT_BLO 427 must now be rcu-walk aware (mask & MAY_NOT_BLOCK). See
434 Documentation/filesystems/vfs.rst for more det 428 Documentation/filesystems/vfs.rst for more details.
435 429
436 --- 430 ---
437 431
438 **mandatory** 432 **mandatory**
439 433
440 In ->fallocate() you must check the mode optio 434 In ->fallocate() you must check the mode option passed in. If your
441 filesystem does not support hole punching (dea 435 filesystem does not support hole punching (deallocating space in the middle of a
442 file) you must return -EOPNOTSUPP if FALLOC_FL 436 file) you must return -EOPNOTSUPP if FALLOC_FL_PUNCH_HOLE is set in mode.
443 Currently you can only have FALLOC_FL_PUNCH_HO 437 Currently you can only have FALLOC_FL_PUNCH_HOLE with FALLOC_FL_KEEP_SIZE set,
444 so the i_size should not change when hole punc 438 so the i_size should not change when hole punching, even when puching the end of
445 a file off. 439 a file off.
446 440
447 --- 441 ---
448 442
449 **mandatory** 443 **mandatory**
450 444
451 ->get_sb() is gone. Switch to use of ->mount( 445 ->get_sb() is gone. Switch to use of ->mount(). Typically it's just
452 a matter of switching from calling ``get_sb_`` 446 a matter of switching from calling ``get_sb_``... to ``mount_``... and changing
453 the function type. If you were doing it manua 447 the function type. If you were doing it manually, just switch from setting
454 ->mnt_root to some pointer to returning that p 448 ->mnt_root to some pointer to returning that pointer. On errors return
455 ERR_PTR(...). 449 ERR_PTR(...).
456 450
457 --- 451 ---
458 452
459 **mandatory** 453 **mandatory**
460 454
461 ->permission() and generic_permission()have lo 455 ->permission() and generic_permission()have lost flags
462 argument; instead of passing IPERM_FLAG_RCU we 456 argument; instead of passing IPERM_FLAG_RCU we add MAY_NOT_BLOCK into mask.
463 457
464 generic_permission() has also lost the check_a 458 generic_permission() has also lost the check_acl argument; ACL checking
465 has been taken to VFS and filesystems need to !! 459 has been taken to VFS and filesystems need to provide a non-NULL ->i_op->get_acl
466 ->i_op->get_inode_acl to read an ACL from disk !! 460 to read an ACL from disk.
467 461
468 --- 462 ---
469 463
470 **mandatory** 464 **mandatory**
471 465
472 If you implement your own ->llseek() you must 466 If you implement your own ->llseek() you must handle SEEK_HOLE and
473 SEEK_DATA. You can handle this by returning - !! 467 SEEK_DATA. You can hanle this by returning -EINVAL, but it would be nicer to
474 support it in some way. The generic handler a 468 support it in some way. The generic handler assumes that the entire file is
475 data and there is a virtual hole at the end of 469 data and there is a virtual hole at the end of the file. So if the provided
476 offset is less than i_size and SEEK_DATA is sp 470 offset is less than i_size and SEEK_DATA is specified, return the same offset.
477 If the above is true for the offset and you ar 471 If the above is true for the offset and you are given SEEK_HOLE, return the end
478 of the file. If the offset is i_size or great 472 of the file. If the offset is i_size or greater return -ENXIO in either case.
479 473
480 **mandatory** 474 **mandatory**
481 475
482 If you have your own ->fsync() you must make s 476 If you have your own ->fsync() you must make sure to call
483 filemap_write_and_wait_range() so that all dir 477 filemap_write_and_wait_range() so that all dirty pages are synced out properly.
484 You must also keep in mind that ->fsync() is n 478 You must also keep in mind that ->fsync() is not called with i_mutex held
485 anymore, so if you require i_mutex locking you 479 anymore, so if you require i_mutex locking you must make sure to take it and
486 release it yourself. 480 release it yourself.
487 481
488 --- 482 ---
489 483
490 **mandatory** 484 **mandatory**
491 485
492 d_alloc_root() is gone, along with a lot of bu 486 d_alloc_root() is gone, along with a lot of bugs caused by code
493 misusing it. Replacement: d_make_root(inode). 487 misusing it. Replacement: d_make_root(inode). On success d_make_root(inode)
494 allocates and returns a new dentry instantiate 488 allocates and returns a new dentry instantiated with the passed in inode.
495 On failure NULL is returned and the passed in 489 On failure NULL is returned and the passed in inode is dropped so the reference
496 to inode is consumed in all cases and failure 490 to inode is consumed in all cases and failure handling need not do any cleanup
497 for the inode. If d_make_root(inode) is passe 491 for the inode. If d_make_root(inode) is passed a NULL inode it returns NULL
498 and also requires no further error handling. T 492 and also requires no further error handling. Typical usage is::
499 493
500 inode = foofs_new_inode(....); 494 inode = foofs_new_inode(....);
501 s->s_root = d_make_root(inode); 495 s->s_root = d_make_root(inode);
502 if (!s->s_root) 496 if (!s->s_root)
503 /* Nothing needed for the inod 497 /* Nothing needed for the inode cleanup */
504 return -ENOMEM; 498 return -ENOMEM;
505 ... 499 ...
506 500
507 --- 501 ---
508 502
509 **mandatory** 503 **mandatory**
510 504
511 The witch is dead! Well, 2/3 of it, anyway. 505 The witch is dead! Well, 2/3 of it, anyway. ->d_revalidate() and
512 ->lookup() do *not* take struct nameidata anym 506 ->lookup() do *not* take struct nameidata anymore; just the flags.
513 507
514 --- 508 ---
515 509
516 **mandatory** 510 **mandatory**
517 511
518 ->create() doesn't take ``struct nameidata *`` 512 ->create() doesn't take ``struct nameidata *``; unlike the previous
519 two, it gets "is it an O_EXCL or equivalent?" 513 two, it gets "is it an O_EXCL or equivalent?" boolean argument. Note that
520 local filesystems can ignore this argument - t !! 514 local filesystems can ignore tha argument - they are guaranteed that the
521 object doesn't exist. It's remote/distributed 515 object doesn't exist. It's remote/distributed ones that might care...
522 516
523 --- 517 ---
524 518
525 **mandatory** 519 **mandatory**
526 520
527 FS_REVAL_DOT is gone; if you used to have it, 521 FS_REVAL_DOT is gone; if you used to have it, add ->d_weak_revalidate()
528 in your dentry operations instead. 522 in your dentry operations instead.
529 523
530 --- 524 ---
531 525
532 **mandatory** 526 **mandatory**
533 527
534 vfs_readdir() is gone; switch to iterate_dir() 528 vfs_readdir() is gone; switch to iterate_dir() instead
535 529
536 --- 530 ---
537 531
538 **mandatory** 532 **mandatory**
539 533
540 ->readdir() is gone now; switch to ->iterate_s !! 534 ->readdir() is gone now; switch to ->iterate()
541 535
542 **mandatory** 536 **mandatory**
543 537
544 vfs_follow_link has been removed. Filesystems 538 vfs_follow_link has been removed. Filesystems must use nd_set_link
545 from ->follow_link for normal symlinks, or nd_ 539 from ->follow_link for normal symlinks, or nd_jump_link for magic
546 /proc/<pid> style links. 540 /proc/<pid> style links.
547 541
548 --- 542 ---
549 543
550 **mandatory** 544 **mandatory**
551 545
552 iget5_locked()/ilookup5()/ilookup5_nowait() te 546 iget5_locked()/ilookup5()/ilookup5_nowait() test() callback used to be
553 called with both ->i_lock and inode_hash_lock 547 called with both ->i_lock and inode_hash_lock held; the former is *not*
554 taken anymore, so verify that your callbacks d 548 taken anymore, so verify that your callbacks do not rely on it (none
555 of the in-tree instances did). inode_hash_loc 549 of the in-tree instances did). inode_hash_lock is still held,
556 of course, so they are still serialized wrt re 550 of course, so they are still serialized wrt removal from inode hash,
557 as well as wrt set() callback of iget5_locked( 551 as well as wrt set() callback of iget5_locked().
558 552
559 --- 553 ---
560 554
561 **mandatory** 555 **mandatory**
562 556
563 d_materialise_unique() is gone; d_splice_alias 557 d_materialise_unique() is gone; d_splice_alias() does everything you
564 need now. Remember that they have opposite or 558 need now. Remember that they have opposite orders of arguments ;-/
565 559
566 --- 560 ---
567 561
568 **mandatory** 562 **mandatory**
569 563
570 f_dentry is gone; use f_path.dentry, or, bette 564 f_dentry is gone; use f_path.dentry, or, better yet, see if you can avoid
571 it entirely. 565 it entirely.
572 566
573 --- 567 ---
574 568
575 **mandatory** 569 **mandatory**
576 570
577 never call ->read() and ->write() directly; us 571 never call ->read() and ->write() directly; use __vfs_{read,write} or
578 wrappers; instead of checking for ->write or - 572 wrappers; instead of checking for ->write or ->read being NULL, look for
579 FMODE_CAN_{WRITE,READ} in file->f_mode. 573 FMODE_CAN_{WRITE,READ} in file->f_mode.
580 574
581 --- 575 ---
582 576
583 **mandatory** 577 **mandatory**
584 578
585 do _not_ use new_sync_{read,write} for ->read/ 579 do _not_ use new_sync_{read,write} for ->read/->write; leave it NULL
586 instead. 580 instead.
587 581
588 --- 582 ---
589 583
590 **mandatory** 584 **mandatory**
591 ->aio_read/->aio_write are gone. Use 585 ->aio_read/->aio_write are gone. Use ->read_iter/->write_iter.
592 586
593 --- 587 ---
594 588
595 **recommended** 589 **recommended**
596 590
597 for embedded ("fast") symlinks just set inode- 591 for embedded ("fast") symlinks just set inode->i_link to wherever the
598 symlink body is and use simple_follow_link() a 592 symlink body is and use simple_follow_link() as ->follow_link().
599 593
600 --- 594 ---
601 595
602 **mandatory** 596 **mandatory**
603 597
604 calling conventions for ->follow_link() have c 598 calling conventions for ->follow_link() have changed. Instead of returning
605 cookie and using nd_set_link() to store the bo 599 cookie and using nd_set_link() to store the body to traverse, we return
606 the body to traverse and store the cookie usin 600 the body to traverse and store the cookie using explicit void ** argument.
607 nameidata isn't passed at all - nd_jump_link() 601 nameidata isn't passed at all - nd_jump_link() doesn't need it and
608 nd_[gs]et_link() is gone. 602 nd_[gs]et_link() is gone.
609 603
610 --- 604 ---
611 605
612 **mandatory** 606 **mandatory**
613 607
614 calling conventions for ->put_link() have chan 608 calling conventions for ->put_link() have changed. It gets inode instead of
615 dentry, it does not get nameidata at all and 609 dentry, it does not get nameidata at all and it gets called only when cookie
616 is non-NULL. Note that link body isn't availa 610 is non-NULL. Note that link body isn't available anymore, so if you need it,
617 store it as cookie. 611 store it as cookie.
618 612
619 --- 613 ---
620 614
621 **mandatory** 615 **mandatory**
622 616
623 any symlink that might use page_follow_link_li 617 any symlink that might use page_follow_link_light/page_put_link() must
624 have inode_nohighmem(inode) called before anyt 618 have inode_nohighmem(inode) called before anything might start playing with
625 its pagecache. No highmem pages should end up 619 its pagecache. No highmem pages should end up in the pagecache of such
626 symlinks. That includes any preseeding that m 620 symlinks. That includes any preseeding that might be done during symlink
627 creation. page_symlink() will honour the mapp !! 621 creation. __page_symlink() will honour the mapping gfp flags, so once
628 you've done inode_nohighmem() it's safe to use 622 you've done inode_nohighmem() it's safe to use, but if you allocate and
629 insert the page manually, make sure to use the 623 insert the page manually, make sure to use the right gfp flags.
630 624
631 --- 625 ---
632 626
633 **mandatory** 627 **mandatory**
634 628
635 ->follow_link() is replaced with ->get_link(); 629 ->follow_link() is replaced with ->get_link(); same API, except that
636 630
637 * ->get_link() gets inode as a separat 631 * ->get_link() gets inode as a separate argument
638 * ->get_link() may be called in RCU mo 632 * ->get_link() may be called in RCU mode - in that case NULL
639 dentry is passed 633 dentry is passed
640 634
641 --- 635 ---
642 636
643 **mandatory** 637 **mandatory**
644 638
645 ->get_link() gets struct delayed_call ``*done` 639 ->get_link() gets struct delayed_call ``*done`` now, and should do
646 set_delayed_call() where it used to set ``*coo 640 set_delayed_call() where it used to set ``*cookie``.
647 641
648 ->put_link() is gone - just give the destructo 642 ->put_link() is gone - just give the destructor to set_delayed_call()
649 in ->get_link(). 643 in ->get_link().
650 644
651 --- 645 ---
652 646
653 **mandatory** 647 **mandatory**
654 648
655 ->getxattr() and xattr_handler.get() get dentr 649 ->getxattr() and xattr_handler.get() get dentry and inode passed separately.
656 dentry might be yet to be attached to inode, s 650 dentry might be yet to be attached to inode, so do _not_ use its ->d_inode
657 in the instances. Rationale: !@#!@# security_ 651 in the instances. Rationale: !@#!@# security_d_instantiate() needs to be
658 called before we attach dentry to inode. 652 called before we attach dentry to inode.
659 653
660 --- 654 ---
661 655
662 **mandatory** 656 **mandatory**
663 657
664 symlinks are no longer the only inodes that do 658 symlinks are no longer the only inodes that do *not* have i_bdev/i_cdev/
665 i_pipe/i_link union zeroed out at inode evicti 659 i_pipe/i_link union zeroed out at inode eviction. As the result, you can't
666 assume that non-NULL value in ->i_nlink at ->d 660 assume that non-NULL value in ->i_nlink at ->destroy_inode() implies that
667 it's a symlink. Checking ->i_mode is really n 661 it's a symlink. Checking ->i_mode is really needed now. In-tree we had
668 to fix shmem_destroy_callback() that used to t 662 to fix shmem_destroy_callback() that used to take that kind of shortcut;
669 watch out, since that shortcut is no longer va 663 watch out, since that shortcut is no longer valid.
670 664
671 --- 665 ---
672 666
673 **mandatory** 667 **mandatory**
674 668
675 ->i_mutex is replaced with ->i_rwsem now. ino 669 ->i_mutex is replaced with ->i_rwsem now. inode_lock() et.al. work as
676 they used to - they just take it exclusive. H 670 they used to - they just take it exclusive. However, ->lookup() may be
677 called with parent locked shared. Its instanc 671 called with parent locked shared. Its instances must not
678 672
679 * use d_instantiate) and d_rehash() se 673 * use d_instantiate) and d_rehash() separately - use d_add() or
680 d_splice_alias() instead. 674 d_splice_alias() instead.
681 * use d_rehash() alone - call d_add(ne 675 * use d_rehash() alone - call d_add(new_dentry, NULL) instead.
682 * in the unlikely case when (read-only 676 * in the unlikely case when (read-only) access to filesystem
683 data structures needs exclusion for 677 data structures needs exclusion for some reason, arrange it
684 yourself. None of the in-tree files 678 yourself. None of the in-tree filesystems needed that.
685 * rely on ->d_parent and ->d_name not 679 * rely on ->d_parent and ->d_name not changing after dentry has
686 been fed to d_add() or d_splice_alia 680 been fed to d_add() or d_splice_alias(). Again, none of the
687 in-tree instances relied upon that. 681 in-tree instances relied upon that.
688 682
689 We are guaranteed that lookups of the same nam 683 We are guaranteed that lookups of the same name in the same directory
690 will not happen in parallel ("same" in the sen 684 will not happen in parallel ("same" in the sense of your ->d_compare()).
691 Lookups on different names in the same directo 685 Lookups on different names in the same directory can and do happen in
692 parallel now. 686 parallel now.
693 687
694 --- 688 ---
695 689
696 **mandatory** !! 690 **recommended**
697 691
698 ->iterate_shared() is added. !! 692 ->iterate_shared() is added; it's a parallel variant of ->iterate().
699 Exclusion on struct file level is still provid 693 Exclusion on struct file level is still provided (as well as that
700 between it and lseek on the same struct file), 694 between it and lseek on the same struct file), but if your directory
701 has been opened several times, you can get the 695 has been opened several times, you can get these called in parallel.
702 Exclusion between that method and all director 696 Exclusion between that method and all directory-modifying ones is
703 still provided, of course. 697 still provided, of course.
704 698
705 If you have any per-inode or per-dentry in-cor !! 699 Often enough ->iterate() can serve as ->iterate_shared() without any
706 by ->iterate_shared(), you might need somethin !! 700 changes - it is a read-only operation, after all. If you have any
707 to them. If you do dcache pre-seeding, you'll !! 701 per-inode or per-dentry in-core data structures modified by ->iterate(),
708 d_alloc_parallel() for that; look for in-tree !! 702 you might need something to serialize the access to them. If you
>> 703 do dcache pre-seeding, you'll need to switch to d_alloc_parallel() for
>> 704 that; look for in-tree examples.
>> 705
>> 706 Old method is only used if the new one is absent; eventually it will
>> 707 be removed. Switch while you still can; the old one won't stay.
709 708
710 --- 709 ---
711 710
712 **mandatory** 711 **mandatory**
713 712
714 ->atomic_open() calls without O_CREAT may happ 713 ->atomic_open() calls without O_CREAT may happen in parallel.
715 714
716 --- 715 ---
717 716
718 **mandatory** 717 **mandatory**
719 718
720 ->setxattr() and xattr_handler.set() get dentr 719 ->setxattr() and xattr_handler.set() get dentry and inode passed separately.
721 The xattr_handler.set() gets passed the user n <<
722 is seen from so filesystems can idmap the i_ui <<
723 dentry might be yet to be attached to inode, s 720 dentry might be yet to be attached to inode, so do _not_ use its ->d_inode
724 in the instances. Rationale: !@#!@# security_ 721 in the instances. Rationale: !@#!@# security_d_instantiate() needs to be
725 called before we attach dentry to inode and !@ 722 called before we attach dentry to inode and !@#!@##!@$!$#!@#$!@$!@$ smack
726 ->d_instantiate() uses not just ->getxattr() b 723 ->d_instantiate() uses not just ->getxattr() but ->setxattr() as well.
727 724
728 --- 725 ---
729 726
730 **mandatory** 727 **mandatory**
731 728
732 ->d_compare() doesn't get parent as a separate 729 ->d_compare() doesn't get parent as a separate argument anymore. If you
733 used it for finding the struct super_block inv 730 used it for finding the struct super_block involved, dentry->d_sb will
734 work just as well; if it's something more comp 731 work just as well; if it's something more complicated, use dentry->d_parent.
735 Just be careful not to assume that fetching it 732 Just be careful not to assume that fetching it more than once will yield
736 the same value - in RCU mode it could change u 733 the same value - in RCU mode it could change under you.
737 734
738 --- 735 ---
739 736
740 **mandatory** 737 **mandatory**
741 738
742 ->rename() has an added flags argument. Any f 739 ->rename() has an added flags argument. Any flags not handled by the
743 filesystem should result in EINVAL being retur 740 filesystem should result in EINVAL being returned.
744 741
745 --- 742 ---
746 743
747 744
748 **recommended** 745 **recommended**
749 746
750 ->readlink is optional for symlinks. Don't se 747 ->readlink is optional for symlinks. Don't set, unless filesystem needs
751 to fake something for readlink(2). 748 to fake something for readlink(2).
752 749
753 --- 750 ---
754 751
755 **mandatory** 752 **mandatory**
756 753
757 ->getattr() is now passed a struct path rather 754 ->getattr() is now passed a struct path rather than a vfsmount and
758 dentry separately, and it now has request_mask 755 dentry separately, and it now has request_mask and query_flags arguments
759 to specify the fields and sync type requested 756 to specify the fields and sync type requested by statx. Filesystems not
760 supporting any statx-specific features may ign 757 supporting any statx-specific features may ignore the new arguments.
761 758
762 --- 759 ---
763 760
764 **mandatory** 761 **mandatory**
765 762
766 ->atomic_open() calling conventions have chang 763 ->atomic_open() calling conventions have changed. Gone is ``int *opened``,
767 along with FILE_OPENED/FILE_CREATED. In place 764 along with FILE_OPENED/FILE_CREATED. In place of those we have
768 FMODE_OPENED/FMODE_CREATED, set in file->f_mod 765 FMODE_OPENED/FMODE_CREATED, set in file->f_mode. Additionally, return
769 value for 'called finish_no_open(), open it yo 766 value for 'called finish_no_open(), open it yourself' case has become
770 0, not 1. Since finish_no_open() itself is re 767 0, not 1. Since finish_no_open() itself is returning 0 now, that part
771 does not need any changes in ->atomic_open() i 768 does not need any changes in ->atomic_open() instances.
772 769
773 --- 770 ---
774 771
775 **mandatory** 772 **mandatory**
776 773
777 alloc_file() has become static now; two wrappe 774 alloc_file() has become static now; two wrappers are to be used instead.
778 alloc_file_pseudo(inode, vfsmount, name, flags 775 alloc_file_pseudo(inode, vfsmount, name, flags, ops) is for the cases
779 when dentry needs to be created; that's the ma 776 when dentry needs to be created; that's the majority of old alloc_file()
780 users. Calling conventions: on success a refe 777 users. Calling conventions: on success a reference to new struct file
781 is returned and callers reference to inode is 778 is returned and callers reference to inode is subsumed by that. On
782 failure, ERR_PTR() is returned and no caller's 779 failure, ERR_PTR() is returned and no caller's references are affected,
783 so the caller needs to drop the inode referenc 780 so the caller needs to drop the inode reference it held.
784 alloc_file_clone(file, flags, ops) does not af 781 alloc_file_clone(file, flags, ops) does not affect any caller's references.
785 On success you get a new struct file sharing t 782 On success you get a new struct file sharing the mount/dentry with the
786 original, on failure - ERR_PTR(). 783 original, on failure - ERR_PTR().
787 784
788 --- 785 ---
789 786
790 **mandatory** 787 **mandatory**
791 788
792 ->clone_file_range() and ->dedupe_file_range h 789 ->clone_file_range() and ->dedupe_file_range have been replaced with
793 ->remap_file_range(). See Documentation/files 790 ->remap_file_range(). See Documentation/filesystems/vfs.rst for more
794 information. 791 information.
795 792
796 --- 793 ---
797 794
798 **recommended** 795 **recommended**
799 796
800 ->lookup() instances doing an equivalent of:: 797 ->lookup() instances doing an equivalent of::
801 798
802 if (IS_ERR(inode)) 799 if (IS_ERR(inode))
803 return ERR_CAST(inode); 800 return ERR_CAST(inode);
804 return d_splice_alias(inode, dentry); 801 return d_splice_alias(inode, dentry);
805 802
806 don't need to bother with the check - d_splice 803 don't need to bother with the check - d_splice_alias() will do the
807 right thing when given ERR_PTR(...) as inode. 804 right thing when given ERR_PTR(...) as inode. Moreover, passing NULL
808 inode to d_splice_alias() will also do the rig 805 inode to d_splice_alias() will also do the right thing (equivalent of
809 d_add(dentry, NULL); return NULL;), so that ki 806 d_add(dentry, NULL); return NULL;), so that kind of special cases
810 also doesn't need a separate treatment. 807 also doesn't need a separate treatment.
811 808
812 --- 809 ---
813 810
814 **strongly recommended** 811 **strongly recommended**
815 812
816 take the RCU-delayed parts of ->destroy_inode( 813 take the RCU-delayed parts of ->destroy_inode() into a new method -
817 ->free_inode(). If ->destroy_inode() becomes 814 ->free_inode(). If ->destroy_inode() becomes empty - all the better,
818 just get rid of it. Synchronous work (e.g. th 815 just get rid of it. Synchronous work (e.g. the stuff that can't
819 be done from an RCU callback, or any WARN_ON() 816 be done from an RCU callback, or any WARN_ON() where we want the
820 stack trace) *might* be movable to ->evict_ino 817 stack trace) *might* be movable to ->evict_inode(); however,
821 that goes only for the things that are not nee 818 that goes only for the things that are not needed to balance something
822 done by ->alloc_inode(). IOW, if it's cleanin 819 done by ->alloc_inode(). IOW, if it's cleaning up the stuff that
823 might have accumulated over the life of in-cor 820 might have accumulated over the life of in-core inode, ->evict_inode()
824 might be a fit. 821 might be a fit.
825 822
826 Rules for inode destruction: 823 Rules for inode destruction:
827 824
828 * if ->destroy_inode() is non-NULL, it 825 * if ->destroy_inode() is non-NULL, it gets called
829 * if ->free_inode() is non-NULL, it ge 826 * if ->free_inode() is non-NULL, it gets scheduled by call_rcu()
830 * combination of NULL ->destroy_inode 827 * combination of NULL ->destroy_inode and NULL ->free_inode is
831 treated as NULL/free_inode_nonrcu, t 828 treated as NULL/free_inode_nonrcu, to preserve the compatibility.
832 829
833 Note that the callback (be it via ->free_inode 830 Note that the callback (be it via ->free_inode() or explicit call_rcu()
834 in ->destroy_inode()) is *NOT* ordered wrt sup 831 in ->destroy_inode()) is *NOT* ordered wrt superblock destruction;
835 as the matter of fact, the superblock and all 832 as the matter of fact, the superblock and all associated structures
836 might be already gone. The filesystem driver 833 might be already gone. The filesystem driver is guaranteed to be still
837 there, but that's it. Freeing memory in the c 834 there, but that's it. Freeing memory in the callback is fine; doing
838 more than that is possible, but requires a lot 835 more than that is possible, but requires a lot of care and is best
839 avoided. 836 avoided.
840 837
841 --- 838 ---
842 839
843 **mandatory** 840 **mandatory**
844 841
845 DCACHE_RCUACCESS is gone; having an RCU delay 842 DCACHE_RCUACCESS is gone; having an RCU delay on dentry freeing is the
846 default. DCACHE_NORCU opts out, and only d_al 843 default. DCACHE_NORCU opts out, and only d_alloc_pseudo() has any
847 business doing so. 844 business doing so.
848 845
849 --- 846 ---
850 847
851 **mandatory** 848 **mandatory**
852 849
853 d_alloc_pseudo() is internal-only; uses outsid 850 d_alloc_pseudo() is internal-only; uses outside of alloc_file_pseudo() are
854 very suspect (and won't work in modules). Suc 851 very suspect (and won't work in modules). Such uses are very likely to
855 be misspelled d_alloc_anon(). 852 be misspelled d_alloc_anon().
856 853
857 --- 854 ---
858 855
859 **mandatory** 856 **mandatory**
860 857
861 [should've been added in 2016] stale comment i !! 858 [should've been added in 2016] stale comment in finish_open() nonwithstanding,
862 failure exits in ->atomic_open() instances sho 859 failure exits in ->atomic_open() instances should *NOT* fput() the file,
863 no matter what. Everything is handled by the 860 no matter what. Everything is handled by the caller.
864 <<
865 --- <<
866 <<
867 **mandatory** <<
868 <<
869 clone_private_mount() returns a longterm mount <<
870 its result is kern_unmount() or kern_unmount_a <<
871 <<
872 --- <<
873 <<
874 **mandatory** <<
875 <<
876 zero-length bvec segments are disallowed, they <<
877 passed on to an iterator. <<
878 <<
879 --- <<
880 <<
881 **mandatory** <<
882 <<
883 For bvec based itererators bio_iov_iter_get_pa <<
884 uses the one provided. Anyone issuing kiocb-I/ <<
885 page references stay until I/O has completed, <<
886 been called or returned with non -EIOCBQUEUED <<
887 <<
888 --- <<
889 <<
890 **mandatory** <<
891 <<
892 mnt_want_write_file() can now only be paired w <<
893 whereas previously it could be paired with mnt <<
894 <<
895 --- <<
896 <<
897 **mandatory** <<
898 <<
899 iov_iter_copy_from_user_atomic() is gone; use <<
900 The difference is copy_page_from_iter_atomic() <<
901 you don't need iov_iter_advance() after it. H <<
902 only a part of obtained data, you should do io <<
903 <<
904 --- <<
905 <<
906 **mandatory** <<
907 <<
908 Calling conventions for file_open_root() chang <<
909 instead of passing mount and dentry separately <<
910 pass <mnt, mnt->mnt_root> pair (i.e. the root <<
911 is provided - file_open_root_mnt(). In-tree u <<
912 <<
913 --- <<
914 <<
915 **mandatory** <<
916 <<
917 no_llseek is gone; don't set .llseek to that - <<
918 Checks for "does that file have llseek(2), or <<
919 should be done by looking at FMODE_LSEEK in fi <<
920 <<
921 --- <<
922 <<
923 *mandatory* <<
924 <<
925 filldir_t (readdir callbacks) calling conventi <<
926 returning 0 or -E... it returns bool now. fal <<
927 to) and true - "keep going" (as 0 in old calli <<
928 callers never looked at specific -E... values <<
929 instances require no changes at all, all filld <<
930 converted. <<
931 <<
932 --- <<
933 <<
934 **mandatory** <<
935 <<
936 Calling conventions for ->tmpfile() have chang <<
937 file pointer instead of struct dentry pointer. <<
938 changed to simplify callers. The passed file <<
939 success must be opened before returning (e.g. <<
940 finish_open_simple()). <<
941 <<
942 --- <<
943 <<
944 **mandatory** <<
945 <<
946 Calling convention for ->huge_fault has change <<
947 order instead of an enum page_entry_size, and <<
948 mmap_lock held. All in-tree users have been a <<
949 depend on the mmap_lock being held, but out of <<
950 for themselves. If they do need it, they can <<
951 be called with the mmap_lock held. <<
952 <<
953 --- <<
954 <<
955 **mandatory** <<
956 <<
957 The order of opening block devices and matchin <<
958 changed. <<
959 <<
960 The old logic opened block devices first and t <<
961 suitable superblock to reuse based on the bloc <<
962 <<
963 The new logic tries to find a suitable superbl <<
964 number, and opening the block device afterward <<
965 <<
966 Since opening block devices cannot happen unde <<
967 ordering requirements s_umount is now dropped <<
968 reacquired before calling fill_super(). <<
969 <<
970 In the old logic concurrent mounters would fin <<
971 superblocks for the filesystem type. Since the <<
972 would hold s_umount they would wait until the <<
973 was discarded due to initialization failure. <<
974 <<
975 Since the new logic drops s_umount concurrent <<
976 would spin. Instead they are now made to wait <<
977 mechanism without having to hold s_umount. <<
978 <<
979 --- <<
980 <<
981 **mandatory** <<
982 <<
983 The holder of a block device is now the superb <<
984 <<
985 The holder of a block device used to be the fi <<
986 particularly useful. It wasn't possible to go <<
987 superblock without matching on the device poin <<
988 This mechanism would only work for a single de <<
989 find the owning superblock of any additional d <<
990 <<
991 In the old mechanism reusing or creating a sup <<
992 umount(2) relied on the file_system_type as th <<
993 underdocumented however: <<
994 <<
995 (1) Any concurrent mounter that managed to gra <<
996 existing superblock was made to wait until <<
997 ready or until the superblock was removed <<
998 the filesystem type. If the superblock is <<
999 reuse it. <<
1000 <<
1001 (2) If the mounter came after deactivate_lock <<
1002 the superblock had been removed from the <<
1003 filesystem type the mounter would wait un <<
1004 reuse the block device and allocate a new <<
1005 <<
1006 (3) If the mounter came after deactivate_lock <<
1007 the superblock had been removed from the <<
1008 filesystem type the mounter would reuse t <<
1009 superblock (the bd_holder point may still <<
1010 <<
1011 Because the holder of the block device was th <<
1012 mounter could open the block devices of any s <<
1013 file_system_type without risking seeing EBUSY <<
1014 still in use by another superblock. <<
1015 <<
1016 Making the superblock the owner of the block <<
1017 is now a unique superblock and thus block dev <<
1018 reused by concurrent mounters. So a concurren <<
1019 see EBUSY when trying to open a block device <<
1020 superblock. <<
1021 <<
1022 The new logic thus waits until the superblock <<
1023 ->kill_sb(). Removal of the superblock from t <<
1024 filesystem type is now moved to a later point <<
1025 <<
1026 (1) Any concurrent mounter managing to grab a <<
1027 superblock is made to wait until the supe <<
1028 the superblock and all devices are shutdo <<
1029 superblock is ready the caller will simpl <<
1030 <<
1031 (2) If the mounter comes after deactivate_loc <<
1032 the superblock has been removed from the <<
1033 filesystem type the mounter is made to wa <<
1034 devices are shut down in ->kill_sb() and <<
1035 list of superblocks of the filesystem typ <<
1036 superblock and grab ownership of the bloc <<
1037 the block device will be set to the newly <<
1038 <<
1039 (3) This case is now collapsed into (2) as th <<
1040 of superblocks of the filesystem type unt <<
1041 ->kill_sb(). In other words, if the super <<
1042 superblock of the filesystem type anymore <<
1043 all associated block devices (the bd_hold <<
1044 <<
1045 As this is a VFS level change it has no pract <<
1046 other than that all of them must use one of t <<
1047 kill_anon_super(), or kill_block_super() help <<
1048 <<
1049 --- <<
1050 <<
1051 **mandatory** <<
1052 <<
1053 Lock ordering has been changed so that s_umou <<
1054 All places where s_umount was taken under ope <<
1055 <<
1056 --- <<
1057 <<
1058 **mandatory** <<
1059 <<
1060 export_operations ->encode_fh() no longer has <<
1061 encode FILEID_INO32_GEN* file handles. <<
1062 Filesystems that used the default implementat <<
1063 generic_encode_ino32_fh() explicitly. <<
1064 <<
1065 --- <<
1066 <<
1067 **mandatory** <<
1068 <<
1069 If ->rename() update of .. on cross-directory <<
1070 directory modifications, do *not* lock the su <<
1071 ->rename() - it's done by the caller now [tha <<
1072 28eceeda130f "fs: Lock moved directories"]. <<
1073 <<
1074 --- <<
1075 <<
1076 **mandatory** <<
1077 <<
1078 On same-directory ->rename() the (tautologica <<
1079 by any locks; just don't do it if the old par <<
1080 We really can't lock two subdirectories in sa <<
1081 deadlocks. <<
1082 <<
1083 --- <<
1084 <<
1085 **mandatory** <<
1086 <<
1087 lock_rename() and lock_rename_child() may fai <<
1088 their arguments do not have a common ancestor <<
1089 is returned, with no locks taken. In-tree us <<
1090 would need to do so. <<
1091 <<
1092 --- <<
1093 <<
1094 **mandatory** <<
1095 <<
1096 The list of children anchored in parent dentr <<
1097 Field names got changed (->d_children/->d_sib <<
1098 for anchor/entries resp.), so any affected pl <<
1099 by compiler. <<
1100 <<
1101 --- <<
1102 <<
1103 **mandatory** <<
1104 <<
1105 ->d_delete() instances are now called for den <<
1106 and refcount equal to 0. They are not permit <<
1107 None of in-tree instances did anything of tha <<
1108 <<
1109 --- <<
1110 <<
1111 **mandatory** <<
1112 <<
1113 ->d_prune() instances are now called without <<
1114 ->d_lock on dentry itself is still held; if y <<
1115 of the in-tree instances did), use your own s <<
1116 <<
1117 ->d_iput() and ->d_release() are called with <<
1118 list of parent's children. It is still unhas <<
1119 removed from parent's ->d_children yet. <<
1120 <<
1121 Anyone iterating through the list of children <<
1122 half-killed dentries that might be seen there <<
1123 see them negative, unhashed and with negative <<
1124 of the in-kernel users would've done the righ <<
1125 <<
1126 --- <<
1127 <<
1128 **recommended** <<
1129 <<
1130 Block device freezing and thawing have been m <<
1131 <<
1132 Before this change, get_active_super() would <<
1133 superblock of the main block device, i.e., th <<
1134 device freezing now works for any block devic <<
1135 just the main block device. The get_active_su <<
1136 pointer are gone. <<
1137 <<
1138 --- <<
1139 <<
1140 **mandatory** <<
1141 <<
1142 set_blocksize() takes opened struct file inst <<
1143 and it *must* be opened exclusive. <<
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.