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