Commit | Line | Data |
---|---|---|
b2441318 | 1 | /* SPDX-License-Identifier: GPL-2.0 */ |
1da177e4 LT |
2 | #ifndef _LINUX_FS_H |
3 | #define _LINUX_FS_H | |
4 | ||
1da177e4 | 5 | #include <linux/linkage.h> |
5dd43ce2 | 6 | #include <linux/wait_bit.h> |
1da177e4 LT |
7 | #include <linux/kdev_t.h> |
8 | #include <linux/dcache.h> | |
3f8206d4 | 9 | #include <linux/path.h> |
1da177e4 LT |
10 | #include <linux/stat.h> |
11 | #include <linux/cache.h> | |
1da177e4 | 12 | #include <linux/list.h> |
bc3b14cb | 13 | #include <linux/list_lru.h> |
4f5e65a1 | 14 | #include <linux/llist.h> |
1da177e4 | 15 | #include <linux/radix-tree.h> |
b93b0163 | 16 | #include <linux/xarray.h> |
6b2dbba8 | 17 | #include <linux/rbtree.h> |
1da177e4 | 18 | #include <linux/init.h> |
914e2637 | 19 | #include <linux/pid.h> |
187f1882 | 20 | #include <linux/bug.h> |
1b1dcc1b | 21 | #include <linux/mutex.h> |
c8c06efa | 22 | #include <linux/rwsem.h> |
baabda26 | 23 | #include <linux/mm_types.h> |
3bd858ab | 24 | #include <linux/capability.h> |
6188e10d | 25 | #include <linux/semaphore.h> |
c75b1d94 | 26 | #include <linux/fcntl.h> |
ceb5bdc2 | 27 | #include <linux/rculist_bl.h> |
07b8ce1e | 28 | #include <linux/atomic.h> |
83aeeada | 29 | #include <linux/shrinker.h> |
c1aab02d | 30 | #include <linux/migrate_mode.h> |
92361636 | 31 | #include <linux/uidgid.h> |
5accdf82 | 32 | #include <linux/lockdep.h> |
c2b1ad80 | 33 | #include <linux/percpu-rwsem.h> |
853b39a7 | 34 | #include <linux/workqueue.h> |
fceef393 | 35 | #include <linux/delayed_call.h> |
85787090 | 36 | #include <linux/uuid.h> |
5660e13d | 37 | #include <linux/errseq.h> |
d9a08a9e | 38 | #include <linux/ioprio.h> |
bbe7449e | 39 | #include <linux/fs_types.h> |
f1fffbd4 RV |
40 | #include <linux/build_bug.h> |
41 | #include <linux/stddef.h> | |
a6435940 | 42 | #include <linux/mount.h> |
e6c9a714 | 43 | #include <linux/cred.h> |
a793d79e | 44 | #include <linux/mnt_idmapping.h> |
8b9f3ac5 | 45 | #include <linux/slab.h> |
0e4a8621 | 46 | #include <linux/maple_tree.h> |
fe3944fb | 47 | #include <linux/rw_hint.h> |
1da177e4 | 48 | |
1da177e4 | 49 | #include <asm/byteorder.h> |
607ca46e | 50 | #include <uapi/linux/fs.h> |
1da177e4 | 51 | |
b83ae6d4 | 52 | struct backing_dev_info; |
52ebea74 | 53 | struct bdi_writeback; |
2f8b5444 | 54 | struct bio; |
5a72e899 | 55 | struct io_comp_batch; |
a5694255 | 56 | struct export_operations; |
10c5db28 | 57 | struct fiemap_extent_info; |
a885c8c4 | 58 | struct hd_geometry; |
1da177e4 | 59 | struct iovec; |
92198f7e | 60 | struct kiocb; |
57cc7215 | 61 | struct kobject; |
1da177e4 LT |
62 | struct pipe_inode_info; |
63 | struct poll_table_struct; | |
64 | struct kstatfs; | |
65 | struct vm_area_struct; | |
66 | struct vfsmount; | |
745ca247 | 67 | struct cred; |
a509bc1a | 68 | struct swap_info_struct; |
55985dd7 | 69 | struct seq_file; |
7b7a8665 | 70 | struct workqueue_struct; |
3b93f911 | 71 | struct iov_iter; |
3e7807d5 | 72 | struct fscrypt_inode_info; |
0b81d077 | 73 | struct fscrypt_operations; |
5585f2af EB |
74 | struct fsverity_info; |
75 | struct fsverity_operations; | |
07a3b8d0 AG |
76 | struct fsnotify_mark_connector; |
77 | struct fsnotify_sb_info; | |
f3a09c92 | 78 | struct fs_context; |
d7167b14 | 79 | struct fs_parameter_spec; |
4c5b4799 | 80 | struct fileattr; |
6f7db389 | 81 | struct iomap_ops; |
1da177e4 | 82 | |
74bf17cf | 83 | extern void __init inode_init(void); |
1da177e4 | 84 | extern void __init inode_init_early(void); |
4248b0da MG |
85 | extern void __init files_init(void); |
86 | extern void __init files_maxfiles_init(void); | |
1da177e4 | 87 | |
518de9b3 | 88 | extern unsigned long get_max_files(void); |
9b80a184 | 89 | extern unsigned int sysctl_nr_open; |
dded4f4d | 90 | |
ddef7ed2 CH |
91 | typedef __kernel_rwf_t rwf_t; |
92 | ||
1da177e4 LT |
93 | struct buffer_head; |
94 | typedef int (get_block_t)(struct inode *inode, sector_t iblock, | |
95 | struct buffer_head *bh_result, int create); | |
187372a3 | 96 | typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset, |
7b7a8665 | 97 | ssize_t bytes, void *private); |
1da177e4 | 98 | |
bbc1096a DH |
99 | #define MAY_EXEC 0x00000001 |
100 | #define MAY_WRITE 0x00000002 | |
101 | #define MAY_READ 0x00000004 | |
102 | #define MAY_APPEND 0x00000008 | |
103 | #define MAY_ACCESS 0x00000010 | |
104 | #define MAY_OPEN 0x00000020 | |
105 | #define MAY_CHDIR 0x00000040 | |
106 | /* called from RCU mode, don't block */ | |
107 | #define MAY_NOT_BLOCK 0x00000080 | |
108 | ||
109 | /* | |
110 | * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond | |
75abe329 | 111 | * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open() |
bbc1096a DH |
112 | */ |
113 | ||
114 | /* file is open for reading */ | |
0a4f544d | 115 | #define FMODE_READ ((__force fmode_t)(1 << 0)) |
bbc1096a | 116 | /* file is open for writing */ |
0a4f544d | 117 | #define FMODE_WRITE ((__force fmode_t)(1 << 1)) |
bbc1096a | 118 | /* file is seekable */ |
0a4f544d | 119 | #define FMODE_LSEEK ((__force fmode_t)(1 << 2)) |
bbc1096a | 120 | /* file can be accessed using pread */ |
0a4f544d | 121 | #define FMODE_PREAD ((__force fmode_t)(1 << 3)) |
bbc1096a | 122 | /* file can be accessed using pwrite */ |
0a4f544d | 123 | #define FMODE_PWRITE ((__force fmode_t)(1 << 4)) |
bbc1096a | 124 | /* File is opened for execution with sys_execve / sys_uselib */ |
0a4f544d | 125 | #define FMODE_EXEC ((__force fmode_t)(1 << 5)) |
ddd65e19 | 126 | /* File writes are restricted (block device specific) */ |
1b0aabcc | 127 | #define FMODE_WRITE_RESTRICTED ((__force fmode_t)(1 << 6)) |
c34fc6f2 PS |
128 | /* File supports atomic writes */ |
129 | #define FMODE_CAN_ATOMIC_WRITE ((__force fmode_t)(1 << 7)) | |
0a4f544d | 130 | |
c34fc6f2 | 131 | /* FMODE_* bit 8 */ |
0a4f544d | 132 | |
bbc1096a | 133 | /* 32bit hashes as llseek() offset (for directories) */ |
0a4f544d | 134 | #define FMODE_32BITHASH ((__force fmode_t)(1 << 9)) |
bbc1096a | 135 | /* 64bit hashes as llseek() offset (for directories) */ |
0a4f544d | 136 | #define FMODE_64BITHASH ((__force fmode_t)(1 << 10)) |
bbc1096a DH |
137 | |
138 | /* | |
139 | * Don't update ctime and mtime. | |
140 | * | |
141 | * Currently a special hack for the XFS open_by_handle ioctl, but we'll | |
142 | * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon. | |
143 | */ | |
0a4f544d | 144 | #define FMODE_NOCMTIME ((__force fmode_t)(1 << 11)) |
bbc1096a DH |
145 | |
146 | /* Expect random access pattern */ | |
0a4f544d | 147 | #define FMODE_RANDOM ((__force fmode_t)(1 << 12)) |
bbc1096a | 148 | |
641bb439 | 149 | /* FMODE_* bit 13 */ |
bbc1096a DH |
150 | |
151 | /* File is opened with O_PATH; almost nothing can be done with it */ | |
0a4f544d | 152 | #define FMODE_PATH ((__force fmode_t)(1 << 14)) |
bbc1096a | 153 | |
2be7d348 | 154 | /* File needs atomic accesses to f_pos */ |
0a4f544d | 155 | #define FMODE_ATOMIC_POS ((__force fmode_t)(1 << 15)) |
83f936c7 | 156 | /* Write access to underlying fs */ |
0a4f544d | 157 | #define FMODE_WRITER ((__force fmode_t)(1 << 16)) |
7f7f25e8 | 158 | /* Has read method(s) */ |
0a4f544d | 159 | #define FMODE_CAN_READ ((__force fmode_t)(1 << 17)) |
7f7f25e8 | 160 | /* Has write method(s) */ |
0a4f544d | 161 | #define FMODE_CAN_WRITE ((__force fmode_t)(1 << 18)) |
9c225f26 | 162 | |
0a4f544d CB |
163 | #define FMODE_OPENED ((__force fmode_t)(1 << 19)) |
164 | #define FMODE_CREATED ((__force fmode_t)(1 << 20)) | |
f5d11409 | 165 | |
10dce8af | 166 | /* File is stream-like */ |
0a4f544d | 167 | #define FMODE_STREAM ((__force fmode_t)(1 << 21)) |
10dce8af | 168 | |
a2ad63da | 169 | /* File supports DIRECT IO */ |
0a4f544d | 170 | #define FMODE_CAN_ODIRECT ((__force fmode_t)(1 << 22)) |
a2ad63da | 171 | |
0a4f544d | 172 | #define FMODE_NOREUSE ((__force fmode_t)(1 << 23)) |
17e81022 | 173 | |
0a4f544d | 174 | /* FMODE_* bit 24 */ |
d8aeb44a | 175 | |
62d53c4a | 176 | /* File is embedded in backing_file object */ |
0a4f544d | 177 | #define FMODE_BACKING ((__force fmode_t)(1 << 25)) |
62d53c4a | 178 | |
bbc1096a | 179 | /* File was opened by fanotify and shouldn't generate fanotify events */ |
0a4f544d | 180 | #define FMODE_NONOTIFY ((__force fmode_t)(1 << 26)) |
bbc1096a | 181 | |
91f9943e | 182 | /* File is capable of returning -EAGAIN if I/O will block */ |
0a4f544d | 183 | #define FMODE_NOWAIT ((__force fmode_t)(1 << 27)) |
a07b2000 AV |
184 | |
185 | /* File represents mount that needs unmounting */ | |
0a4f544d | 186 | #define FMODE_NEED_UNMOUNT ((__force fmode_t)(1 << 28)) |
b745fafa | 187 | |
d3b1084d | 188 | /* File does not contribute to nr_files count */ |
0a4f544d | 189 | #define FMODE_NOACCOUNT ((__force fmode_t)(1 << 29)) |
80175539 | 190 | |
1da177e4 LT |
191 | /* |
192 | * Attribute flags. These should be or-ed together to figure out what | |
193 | * has been changed! | |
194 | */ | |
9767d749 MS |
195 | #define ATTR_MODE (1 << 0) |
196 | #define ATTR_UID (1 << 1) | |
197 | #define ATTR_GID (1 << 2) | |
198 | #define ATTR_SIZE (1 << 3) | |
199 | #define ATTR_ATIME (1 << 4) | |
200 | #define ATTR_MTIME (1 << 5) | |
201 | #define ATTR_CTIME (1 << 6) | |
202 | #define ATTR_ATIME_SET (1 << 7) | |
203 | #define ATTR_MTIME_SET (1 << 8) | |
204 | #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */ | |
9767d749 MS |
205 | #define ATTR_KILL_SUID (1 << 11) |
206 | #define ATTR_KILL_SGID (1 << 12) | |
207 | #define ATTR_FILE (1 << 13) | |
208 | #define ATTR_KILL_PRIV (1 << 14) | |
209 | #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */ | |
210 | #define ATTR_TIMES_SET (1 << 16) | |
f2b20f6e | 211 | #define ATTR_TOUCH (1 << 17) |
7e8ae848 | 212 | #define ATTR_DELEG (1 << 18) /* Delegated attrs. Don't break write delegations */ |
1da177e4 | 213 | |
787fb6bc MS |
214 | /* |
215 | * Whiteout is represented by a char device. The following constants define the | |
216 | * mode and device number to use. | |
217 | */ | |
218 | #define WHITEOUT_MODE 0 | |
219 | #define WHITEOUT_DEV 0 | |
220 | ||
1da177e4 LT |
221 | /* |
222 | * This is the Inode Attributes structure, used for notify_change(). It | |
223 | * uses the above definitions as flags, to know which values have changed. | |
224 | * Also, in this manner, a Filesystem can look at only the values it cares | |
225 | * about. Basically, these are the attributes that the VFS layer can | |
226 | * request to change from the FS layer. | |
227 | * | |
228 | * Derek Atkins <warlord@MIT.EDU> 94-10-20 | |
229 | */ | |
230 | struct iattr { | |
231 | unsigned int ia_valid; | |
232 | umode_t ia_mode; | |
45c31150 CB |
233 | /* |
234 | * The two anonymous unions wrap structures with the same member. | |
235 | * | |
236 | * Filesystems raising FS_ALLOW_IDMAP need to use ia_vfs{g,u}id which | |
237 | * are a dedicated type requiring the filesystem to use the dedicated | |
238 | * helpers. Other filesystem can continue to use ia_{g,u}id until they | |
239 | * have been ported. | |
b27c82e1 CB |
240 | * |
241 | * They always contain the same value. In other words FS_ALLOW_IDMAP | |
242 | * pass down the same value on idmapped mounts as they would on regular | |
243 | * mounts. | |
45c31150 CB |
244 | */ |
245 | union { | |
246 | kuid_t ia_uid; | |
247 | vfsuid_t ia_vfsuid; | |
248 | }; | |
249 | union { | |
250 | kgid_t ia_gid; | |
251 | vfsgid_t ia_vfsgid; | |
252 | }; | |
1da177e4 | 253 | loff_t ia_size; |
95582b00 DD |
254 | struct timespec64 ia_atime; |
255 | struct timespec64 ia_mtime; | |
256 | struct timespec64 ia_ctime; | |
cc4e69de MS |
257 | |
258 | /* | |
25985edc | 259 | * Not an attribute, but an auxiliary info for filesystems wanting to |
cc4e69de MS |
260 | * implement an ftruncate() like method. NOTE: filesystem should |
261 | * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL). | |
262 | */ | |
263 | struct file *ia_file; | |
1da177e4 LT |
264 | }; |
265 | ||
1da177e4 LT |
266 | /* |
267 | * Includes for diskquotas. | |
268 | */ | |
269 | #include <linux/quota.h> | |
270 | ||
69c433ed MS |
271 | /* |
272 | * Maximum number of layers of fs stack. Needs to be limited to | |
273 | * prevent kernel stack overflow | |
274 | */ | |
275 | #define FILESYSTEM_MAX_STACK_DEPTH 2 | |
276 | ||
994fc28c ZB |
277 | /** |
278 | * enum positive_aop_returns - aop return codes with specific semantics | |
279 | * | |
280 | * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has | |
281 | * completed, that the page is still locked, and | |
282 | * should be considered active. The VM uses this hint | |
283 | * to return the page to the active list -- it won't | |
284 | * be a candidate for writeback again in the near | |
285 | * future. Other callers must be careful to unlock | |
286 | * the page if they get this return. Returned by | |
287 | * writepage(); | |
288 | * | |
289 | * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has | |
290 | * unlocked it and the page might have been truncated. | |
291 | * The caller should back up to acquiring a new page and | |
292 | * trying again. The aop will be taking reasonable | |
293 | * precautions not to livelock. If the caller held a page | |
294 | * reference, it should drop it before retrying. Returned | |
7e0a1265 | 295 | * by read_folio(). |
994fc28c ZB |
296 | * |
297 | * address_space_operation functions return these large constants to indicate | |
298 | * special semantics to the caller. These are much larger than the bytes in a | |
299 | * page to allow for functions that return the number of bytes operated on in a | |
300 | * given page. | |
301 | */ | |
302 | ||
303 | enum positive_aop_returns { | |
304 | AOP_WRITEPAGE_ACTIVATE = 0x80000, | |
305 | AOP_TRUNCATED_PAGE = 0x80001, | |
306 | }; | |
307 | ||
1da177e4 LT |
308 | /* |
309 | * oh the beauties of C type declarations. | |
310 | */ | |
311 | struct page; | |
312 | struct address_space; | |
313 | struct writeback_control; | |
8151b4c8 | 314 | struct readahead_control; |
1da177e4 | 315 | |
ce71bfea JA |
316 | /* Match RWF_* bits to IOCB bits */ |
317 | #define IOCB_HIPRI (__force int) RWF_HIPRI | |
318 | #define IOCB_DSYNC (__force int) RWF_DSYNC | |
319 | #define IOCB_SYNC (__force int) RWF_SYNC | |
320 | #define IOCB_NOWAIT (__force int) RWF_NOWAIT | |
321 | #define IOCB_APPEND (__force int) RWF_APPEND | |
c34fc6f2 | 322 | #define IOCB_ATOMIC (__force int) RWF_ATOMIC |
ce71bfea JA |
323 | |
324 | /* non-RWF related bits - start at 16 */ | |
325 | #define IOCB_EVENTFD (1 << 16) | |
326 | #define IOCB_DIRECT (1 << 17) | |
327 | #define IOCB_WRITE (1 << 18) | |
dd3e6d50 | 328 | /* iocb->ki_waitq is valid */ |
ce71bfea JA |
329 | #define IOCB_WAITQ (1 << 19) |
330 | #define IOCB_NOIO (1 << 20) | |
6c7ef543 JA |
331 | /* can use bio alloc cache */ |
332 | #define IOCB_ALLOC_CACHE (1 << 21) | |
9cf3516c JA |
333 | /* |
334 | * IOCB_DIO_CALLER_COMP can be set by the iocb owner, to indicate that the | |
335 | * iocb completion can be passed back to the owner for execution from a safe | |
336 | * context rather than needing to be punted through a workqueue. If this | |
337 | * flag is set, the bio completion handling may set iocb->dio_complete to a | |
338 | * handler function and iocb->private to context information for that handler. | |
339 | * The issuer should call the handler with that context information from task | |
340 | * context to complete the processing of the iocb. Note that while this | |
341 | * provides a task context for the dio_complete() callback, it should only be | |
342 | * used on the completion side for non-IO generating completions. It's fine to | |
343 | * call blocking functions from this callback, but they should not wait for | |
344 | * unrelated IO (like cache flushing, new IO generation, etc). | |
345 | */ | |
346 | #define IOCB_DIO_CALLER_COMP (1 << 22) | |
b820de74 BVA |
347 | /* kiocb is a read or write operation submitted by fs/aio.c. */ |
348 | #define IOCB_AIO_RW (1 << 23) | |
e2e40f2c | 349 | |
f6c73a11 RHI |
350 | /* for use in trace events */ |
351 | #define TRACE_IOCB_STRINGS \ | |
352 | { IOCB_HIPRI, "HIPRI" }, \ | |
353 | { IOCB_DSYNC, "DSYNC" }, \ | |
354 | { IOCB_SYNC, "SYNC" }, \ | |
355 | { IOCB_NOWAIT, "NOWAIT" }, \ | |
356 | { IOCB_APPEND, "APPEND" }, \ | |
c34fc6f2 | 357 | { IOCB_ATOMIC, "ATOMIC"}, \ |
f6c73a11 RHI |
358 | { IOCB_EVENTFD, "EVENTFD"}, \ |
359 | { IOCB_DIRECT, "DIRECT" }, \ | |
360 | { IOCB_WRITE, "WRITE" }, \ | |
361 | { IOCB_WAITQ, "WAITQ" }, \ | |
362 | { IOCB_NOIO, "NOIO" }, \ | |
9cf3516c JA |
363 | { IOCB_ALLOC_CACHE, "ALLOC_CACHE" }, \ |
364 | { IOCB_DIO_CALLER_COMP, "CALLER_COMP" } | |
f6c73a11 | 365 | |
e2e40f2c CH |
366 | struct kiocb { |
367 | struct file *ki_filp; | |
368 | loff_t ki_pos; | |
6b19b766 | 369 | void (*ki_complete)(struct kiocb *iocb, long ret); |
e2e40f2c CH |
370 | void *private; |
371 | int ki_flags; | |
d9a08a9e | 372 | u16 ki_ioprio; /* See linux/ioprio.h */ |
9cf3516c JA |
373 | union { |
374 | /* | |
375 | * Only used for async buffered reads, where it denotes the | |
376 | * page waitqueue associated with completing the read. Valid | |
377 | * IFF IOCB_WAITQ is set. | |
378 | */ | |
379 | struct wait_page_queue *ki_waitq; | |
380 | /* | |
381 | * Can be used for O_DIRECT IO, where the completion handling | |
382 | * is punted back to the issuer of the IO. May only be set | |
383 | * if IOCB_DIO_CALLER_COMP is set by the issuer, and the issuer | |
384 | * must then check for presence of this handler when ki_complete | |
385 | * is invoked. The data passed in to this handler must be | |
386 | * assigned to ->private when dio_complete is assigned. | |
387 | */ | |
388 | ssize_t (*dio_complete)(void *data); | |
389 | }; | |
84c4e1f8 | 390 | }; |
e2e40f2c CH |
391 | |
392 | static inline bool is_sync_kiocb(struct kiocb *kiocb) | |
393 | { | |
394 | return kiocb->ki_complete == NULL; | |
395 | } | |
396 | ||
1da177e4 LT |
397 | struct address_space_operations { |
398 | int (*writepage)(struct page *page, struct writeback_control *wbc); | |
5efe7448 | 399 | int (*read_folio)(struct file *, struct folio *); |
1da177e4 LT |
400 | |
401 | /* Write back some dirty pages from this mapping. */ | |
402 | int (*writepages)(struct address_space *, struct writeback_control *); | |
403 | ||
3a3bae50 | 404 | /* Mark a folio dirty. Return true if this dirtied it */ |
6f31a5a2 | 405 | bool (*dirty_folio)(struct address_space *, struct folio *); |
1da177e4 | 406 | |
8151b4c8 | 407 | void (*readahead)(struct readahead_control *); |
1da177e4 | 408 | |
afddba49 | 409 | int (*write_begin)(struct file *, struct address_space *mapping, |
9d6b0cd7 | 410 | loff_t pos, unsigned len, |
1da86618 | 411 | struct folio **foliop, void **fsdata); |
afddba49 NP |
412 | int (*write_end)(struct file *, struct address_space *mapping, |
413 | loff_t pos, unsigned len, unsigned copied, | |
a225800f | 414 | struct folio *folio, void *fsdata); |
afddba49 | 415 | |
1da177e4 LT |
416 | /* Unfortunately this kludge is needed for FIBMAP. Don't use it */ |
417 | sector_t (*bmap)(struct address_space *, sector_t); | |
128d1f82 | 418 | void (*invalidate_folio) (struct folio *, size_t offset, size_t len); |
fa29000b | 419 | bool (*release_folio)(struct folio *, gfp_t); |
d2329aa0 | 420 | void (*free_folio)(struct folio *folio); |
c8b8e32d | 421 | ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter); |
b969c4ab | 422 | /* |
5490da4f | 423 | * migrate the contents of a folio to the specified target. If |
ef277c73 | 424 | * migrate_mode is MIGRATE_ASYNC, it must not block. |
b969c4ab | 425 | */ |
5490da4f MWO |
426 | int (*migrate_folio)(struct address_space *, struct folio *dst, |
427 | struct folio *src, enum migrate_mode); | |
072acba6 | 428 | int (*launder_folio)(struct folio *); |
2e7e80f7 MWO |
429 | bool (*is_partially_uptodate) (struct folio *, size_t from, |
430 | size_t count); | |
520f301c | 431 | void (*is_dirty_writeback) (struct folio *, bool *dirty, bool *wb); |
af7628d6 | 432 | int (*error_remove_folio)(struct address_space *, struct folio *); |
62c230bc MG |
433 | |
434 | /* swapfile support */ | |
a509bc1a MG |
435 | int (*swap_activate)(struct swap_info_struct *sis, struct file *file, |
436 | sector_t *span); | |
437 | void (*swap_deactivate)(struct file *file); | |
e1209d3a | 438 | int (*swap_rw)(struct kiocb *iocb, struct iov_iter *iter); |
1da177e4 LT |
439 | }; |
440 | ||
7dcda1c9 JA |
441 | extern const struct address_space_operations empty_aops; |
442 | ||
eb797a8e MW |
443 | /** |
444 | * struct address_space - Contents of a cacheable, mappable object. | |
445 | * @host: Owner, either the inode or the block_device. | |
446 | * @i_pages: Cached pages. | |
730633f0 JK |
447 | * @invalidate_lock: Guards coherency between page cache contents and |
448 | * file offset->disk block mappings in the filesystem during invalidates. | |
449 | * It is also used to block modification of page cache contents through | |
450 | * memory mappings. | |
eb797a8e | 451 | * @gfp_mask: Memory allocation flags to use for allocating pages. |
e8e17ee9 | 452 | * @i_mmap_writable: Number of VM_SHARED, VM_MAYWRITE mappings. |
09d91cda | 453 | * @nr_thps: Number of THPs in the pagecache (non-shmem only). |
eb797a8e MW |
454 | * @i_mmap: Tree of private and shared mappings. |
455 | * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable. | |
456 | * @nrpages: Number of page entries, protected by the i_pages lock. | |
eb797a8e MW |
457 | * @writeback_index: Writeback starts here. |
458 | * @a_ops: Methods. | |
459 | * @flags: Error bits and flags (AS_*). | |
460 | * @wb_err: The most recent error which has occurred. | |
600f111e MWO |
461 | * @i_private_lock: For use by the owner of the address_space. |
462 | * @i_private_list: For use by the owner of the address_space. | |
463 | * @i_private_data: For use by the owner of the address_space. | |
eb797a8e | 464 | */ |
1da177e4 | 465 | struct address_space { |
eb797a8e MW |
466 | struct inode *host; |
467 | struct xarray i_pages; | |
730633f0 | 468 | struct rw_semaphore invalidate_lock; |
eb797a8e MW |
469 | gfp_t gfp_mask; |
470 | atomic_t i_mmap_writable; | |
09d91cda SL |
471 | #ifdef CONFIG_READ_ONLY_THP_FOR_FS |
472 | /* number of thp, only for non-shmem files */ | |
473 | atomic_t nr_thps; | |
474 | #endif | |
eb797a8e | 475 | struct rb_root_cached i_mmap; |
eb797a8e | 476 | unsigned long nrpages; |
eb797a8e MW |
477 | pgoff_t writeback_index; |
478 | const struct address_space_operations *a_ops; | |
479 | unsigned long flags; | |
5660e13d | 480 | errseq_t wb_err; |
600f111e MWO |
481 | spinlock_t i_private_lock; |
482 | struct list_head i_private_list; | |
d3b1a9a7 | 483 | struct rw_semaphore i_mmap_rwsem; |
600f111e | 484 | void * i_private_data; |
3859a271 | 485 | } __attribute__((aligned(sizeof(long)))) __randomize_layout; |
1da177e4 LT |
486 | /* |
487 | * On most architectures that alignment is already the case; but | |
25985edc | 488 | * must be enforced here for CRIS, to let the least significant bit |
1da177e4 LT |
489 | * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON. |
490 | */ | |
1da177e4 | 491 | |
ff9c745b MW |
492 | /* XArray tags, for tagging dirty and writeback pages in the pagecache. */ |
493 | #define PAGECACHE_TAG_DIRTY XA_MARK_0 | |
494 | #define PAGECACHE_TAG_WRITEBACK XA_MARK_1 | |
495 | #define PAGECACHE_TAG_TOWRITE XA_MARK_2 | |
496 | ||
1da177e4 | 497 | /* |
ff9c745b | 498 | * Returns true if any of the pages in the mapping are marked with the tag. |
1da177e4 | 499 | */ |
ff9c745b MW |
500 | static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag) |
501 | { | |
502 | return xa_marked(&mapping->i_pages, tag); | |
503 | } | |
1da177e4 | 504 | |
8b28f621 DB |
505 | static inline void i_mmap_lock_write(struct address_space *mapping) |
506 | { | |
c8c06efa | 507 | down_write(&mapping->i_mmap_rwsem); |
8b28f621 DB |
508 | } |
509 | ||
c0d0381a MK |
510 | static inline int i_mmap_trylock_write(struct address_space *mapping) |
511 | { | |
512 | return down_write_trylock(&mapping->i_mmap_rwsem); | |
513 | } | |
514 | ||
8b28f621 DB |
515 | static inline void i_mmap_unlock_write(struct address_space *mapping) |
516 | { | |
c8c06efa | 517 | up_write(&mapping->i_mmap_rwsem); |
8b28f621 DB |
518 | } |
519 | ||
6d4675e6 MK |
520 | static inline int i_mmap_trylock_read(struct address_space *mapping) |
521 | { | |
522 | return down_read_trylock(&mapping->i_mmap_rwsem); | |
523 | } | |
524 | ||
3dec0ba0 DB |
525 | static inline void i_mmap_lock_read(struct address_space *mapping) |
526 | { | |
527 | down_read(&mapping->i_mmap_rwsem); | |
528 | } | |
529 | ||
530 | static inline void i_mmap_unlock_read(struct address_space *mapping) | |
531 | { | |
532 | up_read(&mapping->i_mmap_rwsem); | |
533 | } | |
534 | ||
34ae204f MK |
535 | static inline void i_mmap_assert_locked(struct address_space *mapping) |
536 | { | |
537 | lockdep_assert_held(&mapping->i_mmap_rwsem); | |
538 | } | |
539 | ||
540 | static inline void i_mmap_assert_write_locked(struct address_space *mapping) | |
541 | { | |
542 | lockdep_assert_held_write(&mapping->i_mmap_rwsem); | |
543 | } | |
544 | ||
1da177e4 LT |
545 | /* |
546 | * Might pages of this file be mapped into userspace? | |
547 | */ | |
548 | static inline int mapping_mapped(struct address_space *mapping) | |
549 | { | |
f808c13f | 550 | return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root); |
1da177e4 LT |
551 | } |
552 | ||
553 | /* | |
554 | * Might pages of this file have been modified in userspace? | |
e8e17ee9 | 555 | * Note that i_mmap_writable counts all VM_SHARED, VM_MAYWRITE vmas: do_mmap |
1da177e4 LT |
556 | * marks vma as VM_SHARED if it is shared, and the file was opened for |
557 | * writing i.e. vma may be mprotected writable even if now readonly. | |
4bb5f5d9 DH |
558 | * |
559 | * If i_mmap_writable is negative, no new writable mappings are allowed. You | |
560 | * can only deny writable mappings, if none exists right now. | |
1da177e4 LT |
561 | */ |
562 | static inline int mapping_writably_mapped(struct address_space *mapping) | |
563 | { | |
4bb5f5d9 DH |
564 | return atomic_read(&mapping->i_mmap_writable) > 0; |
565 | } | |
566 | ||
567 | static inline int mapping_map_writable(struct address_space *mapping) | |
568 | { | |
569 | return atomic_inc_unless_negative(&mapping->i_mmap_writable) ? | |
570 | 0 : -EPERM; | |
571 | } | |
572 | ||
573 | static inline void mapping_unmap_writable(struct address_space *mapping) | |
574 | { | |
575 | atomic_dec(&mapping->i_mmap_writable); | |
576 | } | |
577 | ||
578 | static inline int mapping_deny_writable(struct address_space *mapping) | |
579 | { | |
580 | return atomic_dec_unless_positive(&mapping->i_mmap_writable) ? | |
581 | 0 : -EBUSY; | |
582 | } | |
583 | ||
584 | static inline void mapping_allow_writable(struct address_space *mapping) | |
585 | { | |
586 | atomic_inc(&mapping->i_mmap_writable); | |
1da177e4 LT |
587 | } |
588 | ||
589 | /* | |
590 | * Use sequence counter to get consistent i_size on 32-bit processors. | |
591 | */ | |
592 | #if BITS_PER_LONG==32 && defined(CONFIG_SMP) | |
593 | #include <linux/seqlock.h> | |
594 | #define __NEED_I_SIZE_ORDERED | |
595 | #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount) | |
596 | #else | |
597 | #define i_size_ordered_init(inode) do { } while (0) | |
598 | #endif | |
599 | ||
f19d4a8f AV |
600 | struct posix_acl; |
601 | #define ACL_NOT_CACHED ((void *)(-1)) | |
332f606b MS |
602 | /* |
603 | * ACL_DONT_CACHE is for stacked filesystems, that rely on underlying fs to | |
cac2f8b8 CB |
604 | * cache the ACL. This also means that ->get_inode_acl() can be called in RCU |
605 | * mode with the LOOKUP_RCU flag. | |
332f606b | 606 | */ |
2a3a2a3f | 607 | #define ACL_DONT_CACHE ((void *)(-3)) |
f19d4a8f | 608 | |
b8a7a3a6 AG |
609 | static inline struct posix_acl * |
610 | uncached_acl_sentinel(struct task_struct *task) | |
611 | { | |
612 | return (void *)task + 1; | |
613 | } | |
614 | ||
615 | static inline bool | |
616 | is_uncached_acl(struct posix_acl *acl) | |
617 | { | |
618 | return (long)acl & 1; | |
619 | } | |
620 | ||
3ddcd056 LT |
621 | #define IOP_FASTPERM 0x0001 |
622 | #define IOP_LOOKUP 0x0002 | |
623 | #define IOP_NOFOLLOW 0x0004 | |
d0a5b995 | 624 | #define IOP_XATTR 0x0008 |
76fca90e | 625 | #define IOP_DEFAULT_READLINK 0x0010 |
3ddcd056 LT |
626 | |
627 | /* | |
628 | * Keep mostly read-only and often accessed (especially for | |
629 | * the RCU path lookup and 'stat' data) fields at the beginning | |
630 | * of the 'struct inode' | |
631 | */ | |
1da177e4 | 632 | struct inode { |
44a7d7a8 | 633 | umode_t i_mode; |
3ddcd056 | 634 | unsigned short i_opflags; |
92361636 EB |
635 | kuid_t i_uid; |
636 | kgid_t i_gid; | |
3ddcd056 LT |
637 | unsigned int i_flags; |
638 | ||
639 | #ifdef CONFIG_FS_POSIX_ACL | |
640 | struct posix_acl *i_acl; | |
641 | struct posix_acl *i_default_acl; | |
642 | #endif | |
643 | ||
44a7d7a8 NP |
644 | const struct inode_operations *i_op; |
645 | struct super_block *i_sb; | |
3ddcd056 | 646 | struct address_space *i_mapping; |
44a7d7a8 | 647 | |
13e12d14 LT |
648 | #ifdef CONFIG_SECURITY |
649 | void *i_security; | |
650 | #endif | |
44a7d7a8 | 651 | |
3ddcd056 LT |
652 | /* Stat data, not accessed from path walking */ |
653 | unsigned long i_ino; | |
a78ef704 MS |
654 | /* |
655 | * Filesystems may only read i_nlink directly. They shall use the | |
656 | * following functions for modification: | |
657 | * | |
658 | * (set|clear|inc|drop)_nlink | |
659 | * inode_(inc|dec)_link_count | |
660 | */ | |
661 | union { | |
662 | const unsigned int i_nlink; | |
663 | unsigned int __i_nlink; | |
664 | }; | |
3ddcd056 | 665 | dev_t i_rdev; |
2f9d3df8 | 666 | loff_t i_size; |
3aa63a56 JL |
667 | time64_t i_atime_sec; |
668 | time64_t i_mtime_sec; | |
669 | time64_t i_ctime_sec; | |
670 | u32 i_atime_nsec; | |
671 | u32 i_mtime_nsec; | |
672 | u32 i_ctime_nsec; | |
673 | u32 i_generation; | |
6cdbb0ef TT |
674 | spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */ |
675 | unsigned short i_bytes; | |
69448867 | 676 | u8 i_blkbits; |
fe3944fb | 677 | enum rw_hint i_write_hint; |
3ddcd056 LT |
678 | blkcnt_t i_blocks; |
679 | ||
680 | #ifdef __NEED_I_SIZE_ORDERED | |
681 | seqcount_t i_size_seqcount; | |
682 | #endif | |
683 | ||
684 | /* Misc */ | |
2b111edb CB |
685 | u32 i_state; |
686 | /* 32-bit hole */ | |
9902af79 | 687 | struct rw_semaphore i_rwsem; |
13e12d14 | 688 | |
44a7d7a8 | 689 | unsigned long dirtied_when; /* jiffies of first dirtying */ |
a2f48706 | 690 | unsigned long dirtied_time_when; |
44a7d7a8 | 691 | |
1da177e4 | 692 | struct hlist_node i_hash; |
c7f54084 | 693 | struct list_head i_io_list; /* backing dev IO list */ |
52ebea74 TH |
694 | #ifdef CONFIG_CGROUP_WRITEBACK |
695 | struct bdi_writeback *i_wb; /* the associated cgroup wb */ | |
2a814908 TH |
696 | |
697 | /* foreign inode detection, see wbc_detach_inode() */ | |
698 | int i_wb_frn_winner; | |
699 | u16 i_wb_frn_avg_time; | |
700 | u16 i_wb_frn_history; | |
52ebea74 | 701 | #endif |
7ccf19a8 | 702 | struct list_head i_lru; /* inode LRU list */ |
1da177e4 | 703 | struct list_head i_sb_list; |
6c60d2b5 | 704 | struct list_head i_wb_list; /* backing dev writeback list */ |
fa0d7e3d | 705 | union { |
b3d9b7a3 | 706 | struct hlist_head i_dentry; |
fa0d7e3d NP |
707 | struct rcu_head i_rcu; |
708 | }; | |
f02a9ad1 | 709 | atomic64_t i_version; |
8019ad13 | 710 | atomic64_t i_sequence; /* see futex */ |
2f9d3df8 | 711 | atomic_t i_count; |
bd5fe6c5 | 712 | atomic_t i_dio_count; |
6cdbb0ef | 713 | atomic_t i_writecount; |
387e3746 | 714 | #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING) |
d984ea60 MZ |
715 | atomic_t i_readcount; /* struct files open RO */ |
716 | #endif | |
fdb0da89 AV |
717 | union { |
718 | const struct file_operations *i_fop; /* former ->i_op->default_file_ops */ | |
719 | void (*free_inode)(struct inode *); | |
720 | }; | |
4a075e39 | 721 | struct file_lock_context *i_flctx; |
1da177e4 | 722 | struct address_space i_data; |
1da177e4 | 723 | struct list_head i_devices; |
4c154168 TT |
724 | union { |
725 | struct pipe_inode_info *i_pipe; | |
577c4eb0 | 726 | struct cdev *i_cdev; |
61ba64fc | 727 | char *i_link; |
84e710da | 728 | unsigned i_dir_seq; |
4c154168 | 729 | }; |
1da177e4 | 730 | |
1da177e4 | 731 | |
3be25f49 EP |
732 | #ifdef CONFIG_FSNOTIFY |
733 | __u32 i_fsnotify_mask; /* all events this inode cares about */ | |
3aa63a56 | 734 | /* 32-bit hole reserved for expanding i_fsnotify_mask */ |
08991e83 | 735 | struct fsnotify_mark_connector __rcu *i_fsnotify_marks; |
0eeca283 RL |
736 | #endif |
737 | ||
643fa961 | 738 | #ifdef CONFIG_FS_ENCRYPTION |
3e7807d5 | 739 | struct fscrypt_inode_info *i_crypt_info; |
0b81d077 JK |
740 | #endif |
741 | ||
5585f2af EB |
742 | #ifdef CONFIG_FS_VERITY |
743 | struct fsverity_info *i_verity_info; | |
744 | #endif | |
745 | ||
8e18e294 | 746 | void *i_private; /* fs or device private pointer */ |
3859a271 | 747 | } __randomize_layout; |
1da177e4 | 748 | |
da18ecbf CB |
749 | /* |
750 | * Get bit address from inode->i_state to use with wait_var_event() | |
751 | * infrastructre. | |
752 | */ | |
753 | #define inode_state_wait_address(inode, bit) ((char *)&(inode)->i_state + (bit)) | |
754 | ||
755 | struct wait_queue_head *inode_bit_waitqueue(struct wait_bit_queue_entry *wqe, | |
756 | struct inode *inode, u32 bit); | |
757 | ||
758 | static inline void inode_wake_up_bit(struct inode *inode, u32 bit) | |
759 | { | |
760 | /* Caller is responsible for correct memory barriers. */ | |
761 | wake_up_var(inode_state_wait_address(inode, bit)); | |
762 | } | |
763 | ||
50e17c00 DD |
764 | struct timespec64 timestamp_truncate(struct timespec64 t, struct inode *inode); |
765 | ||
93407472 FF |
766 | static inline unsigned int i_blocksize(const struct inode *node) |
767 | { | |
768 | return (1 << node->i_blkbits); | |
769 | } | |
770 | ||
1d3382cb AV |
771 | static inline int inode_unhashed(struct inode *inode) |
772 | { | |
773 | return hlist_unhashed(&inode->i_hash); | |
774 | } | |
775 | ||
5bef9151 AV |
776 | /* |
777 | * __mark_inode_dirty expects inodes to be hashed. Since we don't | |
778 | * want special inodes in the fileset inode space, we make them | |
779 | * appear hashed, but do not put on any lists. hlist_del() | |
780 | * will work fine and require no locking. | |
781 | */ | |
782 | static inline void inode_fake_hash(struct inode *inode) | |
783 | { | |
784 | hlist_add_fake(&inode->i_hash); | |
785 | } | |
786 | ||
f2eace23 IM |
787 | /* |
788 | * inode->i_mutex nesting subclasses for the lock validator: | |
789 | * | |
790 | * 0: the object of the current VFS operation | |
791 | * 1: parent | |
792 | * 2: child/target | |
40bd22c9 BF |
793 | * 3: xattr |
794 | * 4: second non-directory | |
d1b72cc6 MS |
795 | * 5: second parent (when locking independent directories in rename) |
796 | * | |
797 | * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two | |
40bd22c9 | 798 | * non-directories at once. |
f2eace23 IM |
799 | * |
800 | * The locking order between these classes is | |
d1b72cc6 | 801 | * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory |
f2eace23 IM |
802 | */ |
803 | enum inode_i_mutex_lock_class | |
804 | { | |
805 | I_MUTEX_NORMAL, | |
806 | I_MUTEX_PARENT, | |
807 | I_MUTEX_CHILD, | |
4df46240 | 808 | I_MUTEX_XATTR, |
d1b72cc6 MS |
809 | I_MUTEX_NONDIR2, |
810 | I_MUTEX_PARENT2, | |
f2eace23 IM |
811 | }; |
812 | ||
5955102c AV |
813 | static inline void inode_lock(struct inode *inode) |
814 | { | |
9902af79 | 815 | down_write(&inode->i_rwsem); |
5955102c AV |
816 | } |
817 | ||
818 | static inline void inode_unlock(struct inode *inode) | |
819 | { | |
9902af79 AV |
820 | up_write(&inode->i_rwsem); |
821 | } | |
822 | ||
823 | static inline void inode_lock_shared(struct inode *inode) | |
824 | { | |
825 | down_read(&inode->i_rwsem); | |
826 | } | |
827 | ||
828 | static inline void inode_unlock_shared(struct inode *inode) | |
829 | { | |
830 | up_read(&inode->i_rwsem); | |
5955102c AV |
831 | } |
832 | ||
833 | static inline int inode_trylock(struct inode *inode) | |
834 | { | |
9902af79 AV |
835 | return down_write_trylock(&inode->i_rwsem); |
836 | } | |
837 | ||
838 | static inline int inode_trylock_shared(struct inode *inode) | |
839 | { | |
840 | return down_read_trylock(&inode->i_rwsem); | |
5955102c AV |
841 | } |
842 | ||
843 | static inline int inode_is_locked(struct inode *inode) | |
844 | { | |
9902af79 | 845 | return rwsem_is_locked(&inode->i_rwsem); |
5955102c AV |
846 | } |
847 | ||
848 | static inline void inode_lock_nested(struct inode *inode, unsigned subclass) | |
849 | { | |
9902af79 | 850 | down_write_nested(&inode->i_rwsem, subclass); |
5955102c AV |
851 | } |
852 | ||
01c2e13d DW |
853 | static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass) |
854 | { | |
855 | down_read_nested(&inode->i_rwsem, subclass); | |
856 | } | |
857 | ||
730633f0 JK |
858 | static inline void filemap_invalidate_lock(struct address_space *mapping) |
859 | { | |
860 | down_write(&mapping->invalidate_lock); | |
861 | } | |
862 | ||
863 | static inline void filemap_invalidate_unlock(struct address_space *mapping) | |
864 | { | |
865 | up_write(&mapping->invalidate_lock); | |
866 | } | |
867 | ||
868 | static inline void filemap_invalidate_lock_shared(struct address_space *mapping) | |
869 | { | |
870 | down_read(&mapping->invalidate_lock); | |
871 | } | |
872 | ||
873 | static inline int filemap_invalidate_trylock_shared( | |
874 | struct address_space *mapping) | |
875 | { | |
876 | return down_read_trylock(&mapping->invalidate_lock); | |
877 | } | |
878 | ||
879 | static inline void filemap_invalidate_unlock_shared( | |
880 | struct address_space *mapping) | |
881 | { | |
882 | up_read(&mapping->invalidate_lock); | |
883 | } | |
884 | ||
375e289e BF |
885 | void lock_two_nondirectories(struct inode *, struct inode*); |
886 | void unlock_two_nondirectories(struct inode *, struct inode*); | |
887 | ||
7506ae6a JK |
888 | void filemap_invalidate_lock_two(struct address_space *mapping1, |
889 | struct address_space *mapping2); | |
890 | void filemap_invalidate_unlock_two(struct address_space *mapping1, | |
891 | struct address_space *mapping2); | |
892 | ||
893 | ||
1da177e4 LT |
894 | /* |
895 | * NOTE: in a 32bit arch with a preemptable kernel and | |
896 | * an UP compile the i_size_read/write must be atomic | |
897 | * with respect to the local cpu (unlike with preempt disabled), | |
898 | * but they don't need to be atomic with respect to other cpus like in | |
899 | * true SMP (so they need either to either locally disable irq around | |
900 | * the read or for example on x86 they can be still implemented as a | |
901 | * cmpxchg8b without the need of the lock prefix). For SMP compiles | |
902 | * and 64bit archs it makes no difference if preempt is enabled or not. | |
903 | */ | |
48ed214d | 904 | static inline loff_t i_size_read(const struct inode *inode) |
1da177e4 LT |
905 | { |
906 | #if BITS_PER_LONG==32 && defined(CONFIG_SMP) | |
907 | loff_t i_size; | |
908 | unsigned int seq; | |
909 | ||
910 | do { | |
911 | seq = read_seqcount_begin(&inode->i_size_seqcount); | |
912 | i_size = inode->i_size; | |
913 | } while (read_seqcount_retry(&inode->i_size_seqcount, seq)); | |
914 | return i_size; | |
2496396f | 915 | #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION) |
1da177e4 LT |
916 | loff_t i_size; |
917 | ||
918 | preempt_disable(); | |
919 | i_size = inode->i_size; | |
920 | preempt_enable(); | |
921 | return i_size; | |
922 | #else | |
d8f899d1 BL |
923 | /* Pairs with smp_store_release() in i_size_write() */ |
924 | return smp_load_acquire(&inode->i_size); | |
1da177e4 LT |
925 | #endif |
926 | } | |
927 | ||
7762f5a0 MS |
928 | /* |
929 | * NOTE: unlike i_size_read(), i_size_write() does need locking around it | |
930 | * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount | |
931 | * can be lost, resulting in subsequent i_size_read() calls spinning forever. | |
932 | */ | |
1da177e4 LT |
933 | static inline void i_size_write(struct inode *inode, loff_t i_size) |
934 | { | |
935 | #if BITS_PER_LONG==32 && defined(CONFIG_SMP) | |
74e3d1e1 | 936 | preempt_disable(); |
1da177e4 LT |
937 | write_seqcount_begin(&inode->i_size_seqcount); |
938 | inode->i_size = i_size; | |
939 | write_seqcount_end(&inode->i_size_seqcount); | |
74e3d1e1 | 940 | preempt_enable(); |
2496396f | 941 | #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION) |
1da177e4 LT |
942 | preempt_disable(); |
943 | inode->i_size = i_size; | |
944 | preempt_enable(); | |
945 | #else | |
d8f899d1 BL |
946 | /* |
947 | * Pairs with smp_load_acquire() in i_size_read() to ensure | |
948 | * changes related to inode size (such as page contents) are | |
949 | * visible before we see the changed inode size. | |
950 | */ | |
951 | smp_store_release(&inode->i_size, i_size); | |
1da177e4 LT |
952 | #endif |
953 | } | |
954 | ||
48ed214d | 955 | static inline unsigned iminor(const struct inode *inode) |
1da177e4 LT |
956 | { |
957 | return MINOR(inode->i_rdev); | |
958 | } | |
959 | ||
48ed214d | 960 | static inline unsigned imajor(const struct inode *inode) |
1da177e4 LT |
961 | { |
962 | return MAJOR(inode->i_rdev); | |
963 | } | |
964 | ||
1da177e4 | 965 | struct fown_struct { |
1934b212 | 966 | struct file *file; /* backpointer for security modules */ |
1da177e4 | 967 | rwlock_t lock; /* protects pid, uid, euid fields */ |
609d7fa9 EB |
968 | struct pid *pid; /* pid or -pgrp where SIGIO should be sent */ |
969 | enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */ | |
92361636 | 970 | kuid_t uid, euid; /* uid/euid of process setting the owner */ |
1da177e4 LT |
971 | int signum; /* posix.1b rt signal to be delivered on IO */ |
972 | }; | |
973 | ||
c790fbf2 MWO |
974 | /** |
975 | * struct file_ra_state - Track a file's readahead state. | |
976 | * @start: Where the most recent readahead started. | |
977 | * @size: Number of pages read in the most recent readahead. | |
84dacdbd N |
978 | * @async_size: Numer of pages that were/are not needed immediately |
979 | * and so were/are genuinely "ahead". Start next readahead when | |
980 | * the first of these pages is accessed. | |
981 | * @ra_pages: Maximum size of a readahead request, copied from the bdi. | |
c790fbf2 MWO |
982 | * @mmap_miss: How many mmap accesses missed in the page cache. |
983 | * @prev_pos: The last byte in the most recent read request. | |
84dacdbd N |
984 | * |
985 | * When this structure is passed to ->readahead(), the "most recent" | |
986 | * readahead means the current readahead. | |
1da177e4 LT |
987 | */ |
988 | struct file_ra_state { | |
c790fbf2 MWO |
989 | pgoff_t start; |
990 | unsigned int size; | |
991 | unsigned int async_size; | |
992 | unsigned int ra_pages; | |
993 | unsigned int mmap_miss; | |
994 | loff_t prev_pos; | |
1da177e4 | 995 | }; |
1da177e4 | 996 | |
5ce1110b FW |
997 | /* |
998 | * Check if @index falls in the readahead windows. | |
999 | */ | |
1000 | static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index) | |
1001 | { | |
f9acc8c7 FW |
1002 | return (index >= ra->start && |
1003 | index < ra->start + ra->size); | |
5ce1110b FW |
1004 | } |
1005 | ||
c0390d54 CB |
1006 | /** |
1007 | * struct file - Represents a file | |
1008 | * @f_count: reference count | |
1009 | * @f_lock: Protects f_ep, f_flags. Must not be taken from IRQ context. | |
1010 | * @f_mode: FMODE_* flags often used in hotpaths | |
1011 | * @f_op: file operations | |
1012 | * @f_mapping: Contents of a cacheable, mappable object. | |
1013 | * @private_data: filesystem or driver specific data | |
1014 | * @f_inode: cached inode | |
1015 | * @f_flags: file flags | |
1016 | * @f_iocb_flags: iocb flags | |
1017 | * @f_cred: stashed credentials of creator/opener | |
1018 | * @f_path: path of the file | |
1019 | * @f_pos_lock: lock protecting file position | |
5e9b50de | 1020 | * @f_pipe: specific to pipes |
c0390d54 | 1021 | * @f_pos: file position |
c0390d54 CB |
1022 | * @f_security: LSM security context of this file |
1023 | * @f_owner: file owner | |
1024 | * @f_wb_err: writeback error | |
1025 | * @f_sb_err: per sb writeback errors | |
1026 | * @f_ep: link of all epoll hooks for this file | |
1027 | * @f_task_work: task work entry point | |
1028 | * @f_llist: work queue entrypoint | |
1029 | * @f_ra: file's readahead state | |
ea566e18 | 1030 | * @f_freeptr: Pointer used by SLAB_TYPESAFE_BY_RCU file cache (don't touch.) |
a7bc2e8d | 1031 | */ |
1da177e4 | 1032 | struct file { |
c0390d54 CB |
1033 | atomic_long_t f_count; |
1034 | spinlock_t f_lock; | |
1035 | fmode_t f_mode; | |
a7bc2e8d | 1036 | const struct file_operations *f_op; |
c0390d54 CB |
1037 | struct address_space *f_mapping; |
1038 | void *private_data; | |
1039 | struct inode *f_inode; | |
1040 | unsigned int f_flags; | |
1041 | unsigned int f_iocb_flags; | |
1042 | const struct cred *f_cred; | |
1043 | /* --- cacheline 1 boundary (64 bytes) --- */ | |
1044 | struct path f_path; | |
2f512016 | 1045 | union { |
5e9b50de CB |
1046 | /* regular files (with FMODE_ATOMIC_POS) and directories */ |
1047 | struct mutex f_pos_lock; | |
1048 | /* pipes */ | |
1049 | u64 f_pipe; | |
e87f2c26 | 1050 | }; |
c0390d54 | 1051 | loff_t f_pos; |
50462062 | 1052 | #ifdef CONFIG_SECURITY |
c0390d54 | 1053 | void *f_security; |
50462062 | 1054 | #endif |
11068e0b | 1055 | /* --- cacheline 2 boundary (128 bytes) --- */ |
c0390d54 CB |
1056 | struct fown_struct *f_owner; |
1057 | errseq_t f_wb_err; | |
1058 | errseq_t f_sb_err; | |
1da177e4 | 1059 | #ifdef CONFIG_EPOLL |
c0390d54 CB |
1060 | struct hlist_head *f_ep; |
1061 | #endif | |
1062 | union { | |
1063 | struct callback_head f_task_work; | |
1064 | struct llist_node f_llist; | |
1065 | struct file_ra_state f_ra; | |
ea566e18 | 1066 | freeptr_t f_freeptr; |
c0390d54 CB |
1067 | }; |
1068 | /* --- cacheline 3 boundary (192 bytes) --- */ | |
3859a271 KC |
1069 | } __randomize_layout |
1070 | __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */ | |
1da177e4 | 1071 | |
990d6c2d AK |
1072 | struct file_handle { |
1073 | __u32 handle_bytes; | |
1074 | int handle_type; | |
1075 | /* file identifier */ | |
68d6f4f3 | 1076 | unsigned char f_handle[] __counted_by(handle_bytes); |
990d6c2d AK |
1077 | }; |
1078 | ||
cb0942b8 AV |
1079 | static inline struct file *get_file(struct file *f) |
1080 | { | |
f6bdc786 KC |
1081 | long prior = atomic_long_fetch_inc_relaxed(&f->f_count); |
1082 | WARN_ONCE(!prior, "struct file::f_count incremented from zero; use-after-free condition present!\n"); | |
cb0942b8 AV |
1083 | return f; |
1084 | } | |
0ede61d8 CB |
1085 | |
1086 | struct file *get_file_rcu(struct file __rcu **f); | |
61d4fb0b | 1087 | struct file *get_file_active(struct file **f); |
0ede61d8 | 1088 | |
516e0cc5 | 1089 | #define file_count(x) atomic_long_read(&(x)->f_count) |
1da177e4 LT |
1090 | |
1091 | #define MAX_NON_LFS ((1UL<<31) - 1) | |
1092 | ||
1093 | /* Page cache limit. The filesystems should put that into their s_maxbytes | |
1094 | limits, otherwise bad things can happen in VM. */ | |
1095 | #if BITS_PER_LONG==32 | |
0cc3b0ec | 1096 | #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT) |
1da177e4 | 1097 | #elif BITS_PER_LONG==64 |
0cc3b0ec | 1098 | #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX) |
1da177e4 LT |
1099 | #endif |
1100 | ||
7ca76311 | 1101 | /* legacy typedef, should eventually be removed */ |
17fa388d | 1102 | typedef void *fl_owner_t; |
1da177e4 | 1103 | |
a7231a97 | 1104 | struct file_lock; |
c69ff407 | 1105 | struct file_lease; |
a7231a97 | 1106 | |
1da177e4 LT |
1107 | /* The following constant reflects the upper bound of the file/locking space */ |
1108 | #ifndef OFFSET_MAX | |
ea258f15 ZL |
1109 | #define OFFSET_MAX type_max(loff_t) |
1110 | #define OFFT_OFFSET_MAX type_max(off_t) | |
1da177e4 LT |
1111 | #endif |
1112 | ||
1934b212 CB |
1113 | int file_f_owner_allocate(struct file *file); |
1114 | static inline struct fown_struct *file_f_owner(const struct file *file) | |
1115 | { | |
1116 | return READ_ONCE(file->f_owner); | |
1117 | } | |
1118 | ||
bfcd17a6 TP |
1119 | extern void send_sigio(struct fown_struct *fown, int fd, int band); |
1120 | ||
ee296d7c JL |
1121 | static inline struct inode *file_inode(const struct file *f) |
1122 | { | |
1123 | return f->f_inode; | |
1124 | } | |
1125 | ||
3058fca1 AG |
1126 | /* |
1127 | * file_dentry() is a relic from the days that overlayfs was using files with a | |
1128 | * "fake" path, meaning, f_path on overlayfs and f_inode on underlying fs. | |
1129 | * In those days, file_dentry() was needed to get the underlying fs dentry that | |
1130 | * matches f_inode. | |
1131 | * Files with "fake" path should not exist nowadays, so use an assertion to make | |
1132 | * sure that file_dentry() was not papering over filesystem bugs. | |
1133 | */ | |
d101a125 MS |
1134 | static inline struct dentry *file_dentry(const struct file *file) |
1135 | { | |
3058fca1 AG |
1136 | struct dentry *dentry = file->f_path.dentry; |
1137 | ||
1138 | WARN_ON_ONCE(d_inode(dentry) != file_inode(file)); | |
1139 | return dentry; | |
d101a125 MS |
1140 | } |
1141 | ||
1da177e4 | 1142 | struct fasync_struct { |
7a107c0f | 1143 | rwlock_t fa_lock; |
989a2979 ED |
1144 | int magic; |
1145 | int fa_fd; | |
1146 | struct fasync_struct *fa_next; /* singly linked list */ | |
1147 | struct file *fa_file; | |
1148 | struct rcu_head fa_rcu; | |
1da177e4 LT |
1149 | }; |
1150 | ||
1151 | #define FASYNC_MAGIC 0x4601 | |
1152 | ||
1153 | /* SMP safe fasync helpers: */ | |
1154 | extern int fasync_helper(int, struct file *, int, struct fasync_struct **); | |
f7347ce4 LT |
1155 | extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *); |
1156 | extern int fasync_remove_entry(struct file *, struct fasync_struct **); | |
1157 | extern struct fasync_struct *fasync_alloc(void); | |
1158 | extern void fasync_free(struct fasync_struct *); | |
1159 | ||
1da177e4 LT |
1160 | /* can be called from interrupts */ |
1161 | extern void kill_fasync(struct fasync_struct **, int, int); | |
1da177e4 | 1162 | |
e0b93edd | 1163 | extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force); |
bccb5c39 | 1164 | extern int f_setown(struct file *filp, int who, int force); |
1da177e4 | 1165 | extern void f_delown(struct file *filp); |
609d7fa9 | 1166 | extern pid_t f_getown(struct file *filp); |
1934b212 | 1167 | extern int send_sigurg(struct file *file); |
1da177e4 | 1168 | |
e462ec50 DH |
1169 | /* |
1170 | * sb->s_flags. Note that these mirror the equivalent MS_* flags where | |
1171 | * represented in both. | |
1172 | */ | |
f15afbd3 HG |
1173 | #define SB_RDONLY BIT(0) /* Mount read-only */ |
1174 | #define SB_NOSUID BIT(1) /* Ignore suid and sgid bits */ | |
1175 | #define SB_NODEV BIT(2) /* Disallow access to device special files */ | |
1176 | #define SB_NOEXEC BIT(3) /* Disallow program execution */ | |
1177 | #define SB_SYNCHRONOUS BIT(4) /* Writes are synced at once */ | |
1178 | #define SB_MANDLOCK BIT(6) /* Allow mandatory locks on an FS */ | |
1179 | #define SB_DIRSYNC BIT(7) /* Directory modifications are synchronous */ | |
1180 | #define SB_NOATIME BIT(10) /* Do not update access times. */ | |
1181 | #define SB_NODIRATIME BIT(11) /* Do not update directory access times */ | |
1182 | #define SB_SILENT BIT(15) | |
5aa8fd9c | 1183 | #define SB_POSIXACL BIT(16) /* Supports POSIX ACLs */ |
f15afbd3 HG |
1184 | #define SB_INLINECRYPT BIT(17) /* Use blk-crypto for encrypted files */ |
1185 | #define SB_KERNMOUNT BIT(22) /* this is a kern_mount call */ | |
1186 | #define SB_I_VERSION BIT(23) /* Update inode I_version field */ | |
1187 | #define SB_LAZYTIME BIT(25) /* Update the on-disk [acm]times lazily */ | |
e462ec50 DH |
1188 | |
1189 | /* These sb flags are internal to the kernel */ | |
2c18a63b | 1190 | #define SB_DEAD BIT(21) |
5e874914 | 1191 | #define SB_DYING BIT(24) |
f15afbd3 HG |
1192 | #define SB_SUBMOUNT BIT(26) |
1193 | #define SB_FORCE BIT(27) | |
1194 | #define SB_NOSEC BIT(28) | |
1195 | #define SB_BORN BIT(29) | |
1196 | #define SB_ACTIVE BIT(30) | |
1197 | #define SB_NOUSER BIT(31) | |
e462ec50 | 1198 | |
c843843e DR |
1199 | /* These flags relate to encoding and casefolding */ |
1200 | #define SB_ENC_STRICT_MODE_FL (1 << 0) | |
1201 | ||
1202 | #define sb_has_strict_encoding(sb) \ | |
1203 | (sb->s_encoding_flags & SB_ENC_STRICT_MODE_FL) | |
1204 | ||
1da177e4 LT |
1205 | /* |
1206 | * Umount options | |
1207 | */ | |
1208 | ||
1209 | #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */ | |
1210 | #define MNT_DETACH 0x00000002 /* Just detach from the tree */ | |
1211 | #define MNT_EXPIRE 0x00000004 /* Mark for expiry */ | |
db1f05bb MS |
1212 | #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */ |
1213 | #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */ | |
1da177e4 | 1214 | |
46b15caa TH |
1215 | /* sb->s_iflags */ |
1216 | #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */ | |
90f8572b | 1217 | #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */ |
a2982cc9 | 1218 | #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */ |
1cb039f3 | 1219 | #define SB_I_STABLE_WRITES 0x00000008 /* don't modify blks until WB is done */ |
1da177e4 | 1220 | |
8654df4e EB |
1221 | /* sb->s_iflags to limit user namespace mounts */ |
1222 | #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */ | |
57b56ac6 MZ |
1223 | #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020 |
1224 | #define SB_I_UNTRUSTED_MOUNTER 0x00000040 | |
1f65e57d | 1225 | #define SB_I_EVM_HMAC_UNSUPPORTED 0x00000080 |
1da177e4 | 1226 | |
32b1924b | 1227 | #define SB_I_SKIP_SYNC 0x00000100 /* Skip superblock at global sync */ |
0b3ea092 | 1228 | #define SB_I_PERSB_BDI 0x00000200 /* has a per-sb bdi */ |
a128b054 | 1229 | #define SB_I_TS_EXPIRY_WARNED 0x00000400 /* warned about timestamp range expiry */ |
04b94071 | 1230 | #define SB_I_RETIRED 0x00000800 /* superblock shouldn't be reused */ |
5aa8fd9c | 1231 | #define SB_I_NOUMASK 0x00001000 /* VFS does not apply umask */ |
32b1924b | 1232 | |
5accdf82 JK |
1233 | /* Possible states of 'frozen' field */ |
1234 | enum { | |
1235 | SB_UNFROZEN = 0, /* FS is unfrozen */ | |
1236 | SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */ | |
5accdf82 JK |
1237 | SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */ |
1238 | SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop | |
1239 | * internal threads if needed) */ | |
1240 | SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */ | |
1241 | }; | |
1242 | ||
1243 | #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1) | |
1244 | ||
1245 | struct sb_writers { | |
880b9577 | 1246 | unsigned short frozen; /* Is sb frozen? */ |
7366f8b6 CB |
1247 | int freeze_kcount; /* How many kernel freeze requests? */ |
1248 | int freeze_ucount; /* How many userspace freeze requests? */ | |
8129ed29 | 1249 | struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS]; |
5accdf82 JK |
1250 | }; |
1251 | ||
1da177e4 LT |
1252 | struct super_block { |
1253 | struct list_head s_list; /* Keep this first */ | |
1254 | dev_t s_dev; /* search index; _not_ kdev_t */ | |
270ba5f7 RK |
1255 | unsigned char s_blocksize_bits; |
1256 | unsigned long s_blocksize; | |
42cb56ae | 1257 | loff_t s_maxbytes; /* Max file size */ |
1da177e4 | 1258 | struct file_system_type *s_type; |
ee9b6d61 | 1259 | const struct super_operations *s_op; |
61e225dc | 1260 | const struct dquot_operations *dq_op; |
0d54b217 | 1261 | const struct quotactl_ops *s_qcop; |
39655164 | 1262 | const struct export_operations *s_export_op; |
1da177e4 | 1263 | unsigned long s_flags; |
46b15caa | 1264 | unsigned long s_iflags; /* internal SB_I_* flags */ |
1da177e4 LT |
1265 | unsigned long s_magic; |
1266 | struct dentry *s_root; | |
1267 | struct rw_semaphore s_umount; | |
1da177e4 | 1268 | int s_count; |
1da177e4 | 1269 | atomic_t s_active; |
50462062 | 1270 | #ifdef CONFIG_SECURITY |
1da177e4 | 1271 | void *s_security; |
50462062 | 1272 | #endif |
e346fb6d | 1273 | const struct xattr_handler * const *s_xattr; |
643fa961 | 1274 | #ifdef CONFIG_FS_ENCRYPTION |
0b81d077 | 1275 | const struct fscrypt_operations *s_cop; |
d7e7b9af | 1276 | struct fscrypt_keyring *s_master_keys; /* master crypto keys in use */ |
5585f2af EB |
1277 | #endif |
1278 | #ifdef CONFIG_FS_VERITY | |
1279 | const struct fsverity_operations *s_vop; | |
c843843e | 1280 | #endif |
5298d4bf | 1281 | #if IS_ENABLED(CONFIG_UNICODE) |
c843843e DR |
1282 | struct unicode_map *s_encoding; |
1283 | __u16 s_encoding_flags; | |
bbbc3fb6 | 1284 | #endif |
f1ee6162 | 1285 | struct hlist_bl_head s_roots; /* alternate root dentries for NFS */ |
39f7c4db | 1286 | struct list_head s_mounts; /* list of mounts; _not_ for fs use */ |
f3a60882 CB |
1287 | struct block_device *s_bdev; /* can go away once we use an accessor for @s_bdev_file */ |
1288 | struct file *s_bdev_file; | |
32a88aa1 | 1289 | struct backing_dev_info *s_bdi; |
acaebfd8 | 1290 | struct mtd_info *s_mtd; |
a5166169 | 1291 | struct hlist_node s_instances; |
2c5f648a | 1292 | unsigned int s_quota_types; /* Bitmask of supported quota types */ |
1da177e4 LT |
1293 | struct quota_info s_dquot; /* Diskquota specific options */ |
1294 | ||
5accdf82 | 1295 | struct sb_writers s_writers; |
1da177e4 | 1296 | |
99c228a9 AG |
1297 | /* |
1298 | * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and | |
07a3b8d0 | 1299 | * s_fsnotify_info together for cache efficiency. They are frequently |
99c228a9 AG |
1300 | * accessed and rarely modified. |
1301 | */ | |
1302 | void *s_fs_info; /* Filesystem private info */ | |
1303 | ||
1304 | /* Granularity of c/m/atime in ns (cannot be worse than a second) */ | |
1305 | u32 s_time_gran; | |
188d20bc DD |
1306 | /* Time limits for c/m/atime in seconds */ |
1307 | time64_t s_time_min; | |
1308 | time64_t s_time_max; | |
99c228a9 | 1309 | #ifdef CONFIG_FSNOTIFY |
71ff58ce | 1310 | u32 s_fsnotify_mask; |
07a3b8d0 | 1311 | struct fsnotify_sb_info *s_fsnotify_info; |
99c228a9 AG |
1312 | #endif |
1313 | ||
ae8c5117 KO |
1314 | /* |
1315 | * q: why are s_id and s_sysfs_name not the same? both are human | |
1316 | * readable strings that identify the filesystem | |
1317 | * a: s_id is allowed to change at runtime; it's used in log messages, | |
1318 | * and we want to when a device starts out as single device (s_id is dev | |
1319 | * name) but then a device is hot added and we have to switch to | |
1320 | * identifying it by UUID | |
1321 | * but s_sysfs_name is a handle for programmatic access, and can't | |
1322 | * change at runtime | |
1323 | */ | |
85787090 CH |
1324 | char s_id[32]; /* Informational name */ |
1325 | uuid_t s_uuid; /* UUID */ | |
a4af51ce | 1326 | u8 s_uuid_len; /* Default 16, possibly smaller for weird filesystems */ |
1da177e4 | 1327 | |
ae8c5117 KO |
1328 | /* if set, fs shows up under sysfs at /sys/fs/$FSTYP/s_sysfs_name */ |
1329 | char s_sysfs_name[UUID_STRING_LEN + 1]; | |
1da177e4 | 1330 | |
8de52778 | 1331 | unsigned int s_max_links; |
1da177e4 LT |
1332 | |
1333 | /* | |
1334 | * The next field is for VFS *only*. No filesystems have any business | |
1335 | * even looking at it. You had been warned. | |
1336 | */ | |
a11f3a05 | 1337 | struct mutex s_vfs_rename_mutex; /* Kludge */ |
1da177e4 | 1338 | |
79c0b2df MS |
1339 | /* |
1340 | * Filesystem subtype. If non-empty the filesystem type field | |
1341 | * in /proc/mounts will be "type.subtype" | |
1342 | */ | |
a0c9a8b8 | 1343 | const char *s_subtype; |
b3b304a2 | 1344 | |
c8aebb0c | 1345 | const struct dentry_operations *s_d_op; /* default d_op for dentries */ |
9fdfdcf1 | 1346 | |
1720f5dd | 1347 | struct shrinker *s_shrink; /* per-sb shrinker handle */ |
4ed5e82f | 1348 | |
7ada4db8 MS |
1349 | /* Number of inodes with nlink == 0 but still referenced */ |
1350 | atomic_long_t s_remove_count; | |
1351 | ||
d7439fb1 | 1352 | /* Read-only state of the superblock is being changed */ |
4ed5e82f | 1353 | int s_readonly_remount; |
7b7a8665 | 1354 | |
735e4ae5 JL |
1355 | /* per-sb errseq_t for reporting writeback errors via syncfs */ |
1356 | errseq_t s_wb_err; | |
1357 | ||
7b7a8665 CH |
1358 | /* AIO completions deferred from interrupt context */ |
1359 | struct workqueue_struct *s_dio_done_wq; | |
215752fc | 1360 | struct hlist_head s_pins; |
f6041567 | 1361 | |
6e4eab57 EB |
1362 | /* |
1363 | * Owning user namespace and default context in which to | |
1364 | * interpret filesystem uids, gids, quotas, device nodes, | |
1365 | * xattrs and security labels. | |
1366 | */ | |
1367 | struct user_namespace *s_user_ns; | |
1368 | ||
f6041567 | 1369 | /* |
7d10f70f WL |
1370 | * The list_lru structure is essentially just a pointer to a table |
1371 | * of per-node lru lists, each of which has its own spinlock. | |
1372 | * There is no need to put them into separate cachelines. | |
f6041567 | 1373 | */ |
7d10f70f WL |
1374 | struct list_lru s_dentry_lru; |
1375 | struct list_lru s_inode_lru; | |
e2fec7c3 | 1376 | struct rcu_head rcu; |
853b39a7 | 1377 | struct work_struct destroy_work; |
69c433ed | 1378 | |
e97fedb9 | 1379 | struct mutex s_sync_lock; /* sync serialisation lock */ |
69c433ed MS |
1380 | |
1381 | /* | |
1382 | * Indicates how deep in a filesystem stack this SB is | |
1383 | */ | |
1384 | int s_stack_depth; | |
74278da9 DC |
1385 | |
1386 | /* s_inode_list_lock protects s_inodes */ | |
1387 | spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp; | |
1388 | struct list_head s_inodes; /* all inodes */ | |
6c60d2b5 DC |
1389 | |
1390 | spinlock_t s_inode_wblist_lock; | |
1391 | struct list_head s_inodes_wb; /* writeback inodes */ | |
3859a271 | 1392 | } __randomize_layout; |
1da177e4 | 1393 | |
a1ec9040 CB |
1394 | static inline struct user_namespace *i_user_ns(const struct inode *inode) |
1395 | { | |
1396 | return inode->i_sb->s_user_ns; | |
1397 | } | |
1398 | ||
81754357 SF |
1399 | /* Helper functions so that in most cases filesystems will |
1400 | * not need to deal directly with kuid_t and kgid_t and can | |
1401 | * instead deal with the raw numeric values that are stored | |
1402 | * in the filesystem. | |
1403 | */ | |
1404 | static inline uid_t i_uid_read(const struct inode *inode) | |
1405 | { | |
a1ec9040 | 1406 | return from_kuid(i_user_ns(inode), inode->i_uid); |
81754357 SF |
1407 | } |
1408 | ||
1409 | static inline gid_t i_gid_read(const struct inode *inode) | |
1410 | { | |
a1ec9040 | 1411 | return from_kgid(i_user_ns(inode), inode->i_gid); |
81754357 SF |
1412 | } |
1413 | ||
1414 | static inline void i_uid_write(struct inode *inode, uid_t uid) | |
1415 | { | |
a1ec9040 | 1416 | inode->i_uid = make_kuid(i_user_ns(inode), uid); |
81754357 SF |
1417 | } |
1418 | ||
1419 | static inline void i_gid_write(struct inode *inode, gid_t gid) | |
1420 | { | |
a1ec9040 | 1421 | inode->i_gid = make_kgid(i_user_ns(inode), gid); |
81754357 SF |
1422 | } |
1423 | ||
234a3113 | 1424 | /** |
e67fe633 CB |
1425 | * i_uid_into_vfsuid - map an inode's i_uid down according to an idmapping |
1426 | * @idmap: idmap of the mount the inode was found from | |
234a3113 CB |
1427 | * @inode: inode to map |
1428 | * | |
e67fe633 | 1429 | * Return: whe inode's i_uid mapped down according to @idmap. |
234a3113 CB |
1430 | * If the inode's i_uid has no mapping INVALID_VFSUID is returned. |
1431 | */ | |
e67fe633 | 1432 | static inline vfsuid_t i_uid_into_vfsuid(struct mnt_idmap *idmap, |
234a3113 CB |
1433 | const struct inode *inode) |
1434 | { | |
4d7ca409 | 1435 | return make_vfsuid(idmap, i_user_ns(inode), inode->i_uid); |
e6c9a714 CB |
1436 | } |
1437 | ||
1f36146a CB |
1438 | /** |
1439 | * i_uid_needs_update - check whether inode's i_uid needs to be updated | |
0dbe12f2 | 1440 | * @idmap: idmap of the mount the inode was found from |
1f36146a CB |
1441 | * @attr: the new attributes of @inode |
1442 | * @inode: the inode to update | |
1443 | * | |
1444 | * Check whether the $inode's i_uid field needs to be updated taking idmapped | |
1445 | * mounts into account if the filesystem supports it. | |
1446 | * | |
1447 | * Return: true if @inode's i_uid field needs to be updated, false if not. | |
1448 | */ | |
0dbe12f2 | 1449 | static inline bool i_uid_needs_update(struct mnt_idmap *idmap, |
1f36146a CB |
1450 | const struct iattr *attr, |
1451 | const struct inode *inode) | |
1452 | { | |
1453 | return ((attr->ia_valid & ATTR_UID) && | |
1454 | !vfsuid_eq(attr->ia_vfsuid, | |
e67fe633 | 1455 | i_uid_into_vfsuid(idmap, inode))); |
1f36146a CB |
1456 | } |
1457 | ||
1458 | /** | |
1459 | * i_uid_update - update @inode's i_uid field | |
0dbe12f2 | 1460 | * @idmap: idmap of the mount the inode was found from |
1f36146a CB |
1461 | * @attr: the new attributes of @inode |
1462 | * @inode: the inode to update | |
1463 | * | |
1464 | * Safely update @inode's i_uid field translating the vfsuid of any idmapped | |
1465 | * mount into the filesystem kuid. | |
1466 | */ | |
0dbe12f2 | 1467 | static inline void i_uid_update(struct mnt_idmap *idmap, |
1f36146a CB |
1468 | const struct iattr *attr, |
1469 | struct inode *inode) | |
1470 | { | |
1471 | if (attr->ia_valid & ATTR_UID) | |
4d7ca409 | 1472 | inode->i_uid = from_vfsuid(idmap, i_user_ns(inode), |
1f36146a CB |
1473 | attr->ia_vfsuid); |
1474 | } | |
1475 | ||
234a3113 | 1476 | /** |
e67fe633 CB |
1477 | * i_gid_into_vfsgid - map an inode's i_gid down according to an idmapping |
1478 | * @idmap: idmap of the mount the inode was found from | |
234a3113 CB |
1479 | * @inode: inode to map |
1480 | * | |
e67fe633 | 1481 | * Return: the inode's i_gid mapped down according to @idmap. |
234a3113 CB |
1482 | * If the inode's i_gid has no mapping INVALID_VFSGID is returned. |
1483 | */ | |
e67fe633 | 1484 | static inline vfsgid_t i_gid_into_vfsgid(struct mnt_idmap *idmap, |
234a3113 CB |
1485 | const struct inode *inode) |
1486 | { | |
4d7ca409 | 1487 | return make_vfsgid(idmap, i_user_ns(inode), inode->i_gid); |
e6c9a714 CB |
1488 | } |
1489 | ||
1f36146a CB |
1490 | /** |
1491 | * i_gid_needs_update - check whether inode's i_gid needs to be updated | |
0dbe12f2 | 1492 | * @idmap: idmap of the mount the inode was found from |
1f36146a CB |
1493 | * @attr: the new attributes of @inode |
1494 | * @inode: the inode to update | |
1495 | * | |
1496 | * Check whether the $inode's i_gid field needs to be updated taking idmapped | |
1497 | * mounts into account if the filesystem supports it. | |
1498 | * | |
1499 | * Return: true if @inode's i_gid field needs to be updated, false if not. | |
1500 | */ | |
0dbe12f2 | 1501 | static inline bool i_gid_needs_update(struct mnt_idmap *idmap, |
1f36146a CB |
1502 | const struct iattr *attr, |
1503 | const struct inode *inode) | |
1504 | { | |
1505 | return ((attr->ia_valid & ATTR_GID) && | |
1506 | !vfsgid_eq(attr->ia_vfsgid, | |
e67fe633 | 1507 | i_gid_into_vfsgid(idmap, inode))); |
1f36146a CB |
1508 | } |
1509 | ||
1510 | /** | |
1511 | * i_gid_update - update @inode's i_gid field | |
0dbe12f2 | 1512 | * @idmap: idmap of the mount the inode was found from |
1f36146a CB |
1513 | * @attr: the new attributes of @inode |
1514 | * @inode: the inode to update | |
1515 | * | |
1516 | * Safely update @inode's i_gid field translating the vfsgid of any idmapped | |
1517 | * mount into the filesystem kgid. | |
1518 | */ | |
0dbe12f2 | 1519 | static inline void i_gid_update(struct mnt_idmap *idmap, |
1f36146a CB |
1520 | const struct iattr *attr, |
1521 | struct inode *inode) | |
1522 | { | |
1523 | if (attr->ia_valid & ATTR_GID) | |
4d7ca409 | 1524 | inode->i_gid = from_vfsgid(idmap, i_user_ns(inode), |
1f36146a CB |
1525 | attr->ia_vfsgid); |
1526 | } | |
1527 | ||
db998553 CB |
1528 | /** |
1529 | * inode_fsuid_set - initialize inode's i_uid field with callers fsuid | |
1530 | * @inode: inode to initialize | |
c14329d3 | 1531 | * @idmap: idmap of the mount the inode was found from |
db998553 CB |
1532 | * |
1533 | * Initialize the i_uid field of @inode. If the inode was found/created via | |
c14329d3 | 1534 | * an idmapped mount map the caller's fsuid according to @idmap. |
db998553 CB |
1535 | */ |
1536 | static inline void inode_fsuid_set(struct inode *inode, | |
c14329d3 | 1537 | struct mnt_idmap *idmap) |
db998553 | 1538 | { |
c14329d3 | 1539 | inode->i_uid = mapped_fsuid(idmap, i_user_ns(inode)); |
db998553 CB |
1540 | } |
1541 | ||
1542 | /** | |
1543 | * inode_fsgid_set - initialize inode's i_gid field with callers fsgid | |
1544 | * @inode: inode to initialize | |
c14329d3 | 1545 | * @idmap: idmap of the mount the inode was found from |
db998553 CB |
1546 | * |
1547 | * Initialize the i_gid field of @inode. If the inode was found/created via | |
c14329d3 | 1548 | * an idmapped mount map the caller's fsgid according to @idmap. |
db998553 CB |
1549 | */ |
1550 | static inline void inode_fsgid_set(struct inode *inode, | |
c14329d3 | 1551 | struct mnt_idmap *idmap) |
db998553 | 1552 | { |
c14329d3 | 1553 | inode->i_gid = mapped_fsgid(idmap, i_user_ns(inode)); |
db998553 CB |
1554 | } |
1555 | ||
8e538913 CB |
1556 | /** |
1557 | * fsuidgid_has_mapping() - check whether caller's fsuid/fsgid is mapped | |
1558 | * @sb: the superblock we want a mapping in | |
4609e1f1 | 1559 | * @idmap: idmap of the relevant mount |
8e538913 CB |
1560 | * |
1561 | * Check whether the caller's fsuid and fsgid have a valid mapping in the | |
1562 | * s_user_ns of the superblock @sb. If the caller is on an idmapped mount map | |
4609e1f1 | 1563 | * the caller's fsuid and fsgid according to the @idmap first. |
8e538913 CB |
1564 | * |
1565 | * Return: true if fsuid and fsgid is mapped, false if not. | |
1566 | */ | |
1567 | static inline bool fsuidgid_has_mapping(struct super_block *sb, | |
4609e1f1 | 1568 | struct mnt_idmap *idmap) |
8e538913 | 1569 | { |
476860b3 CB |
1570 | struct user_namespace *fs_userns = sb->s_user_ns; |
1571 | kuid_t kuid; | |
1572 | kgid_t kgid; | |
1573 | ||
c14329d3 | 1574 | kuid = mapped_fsuid(idmap, fs_userns); |
476860b3 CB |
1575 | if (!uid_valid(kuid)) |
1576 | return false; | |
c14329d3 | 1577 | kgid = mapped_fsgid(idmap, fs_userns); |
476860b3 CB |
1578 | if (!gid_valid(kgid)) |
1579 | return false; | |
1580 | return kuid_has_mapping(fs_userns, kuid) && | |
1581 | kgid_has_mapping(fs_userns, kgid); | |
8e538913 CB |
1582 | } |
1583 | ||
9b6304c1 JL |
1584 | struct timespec64 current_time(struct inode *inode); |
1585 | struct timespec64 inode_set_ctime_current(struct inode *inode); | |
1586 | ||
077c212f JL |
1587 | static inline time64_t inode_get_atime_sec(const struct inode *inode) |
1588 | { | |
3aa63a56 | 1589 | return inode->i_atime_sec; |
077c212f JL |
1590 | } |
1591 | ||
1592 | static inline long inode_get_atime_nsec(const struct inode *inode) | |
1593 | { | |
3aa63a56 | 1594 | return inode->i_atime_nsec; |
077c212f JL |
1595 | } |
1596 | ||
1597 | static inline struct timespec64 inode_get_atime(const struct inode *inode) | |
1598 | { | |
3aa63a56 JL |
1599 | struct timespec64 ts = { .tv_sec = inode_get_atime_sec(inode), |
1600 | .tv_nsec = inode_get_atime_nsec(inode) }; | |
1601 | ||
1602 | return ts; | |
077c212f JL |
1603 | } |
1604 | ||
1605 | static inline struct timespec64 inode_set_atime_to_ts(struct inode *inode, | |
1606 | struct timespec64 ts) | |
1607 | { | |
3aa63a56 JL |
1608 | inode->i_atime_sec = ts.tv_sec; |
1609 | inode->i_atime_nsec = ts.tv_nsec; | |
077c212f JL |
1610 | return ts; |
1611 | } | |
1612 | ||
1613 | static inline struct timespec64 inode_set_atime(struct inode *inode, | |
1614 | time64_t sec, long nsec) | |
1615 | { | |
1616 | struct timespec64 ts = { .tv_sec = sec, | |
1617 | .tv_nsec = nsec }; | |
3aa63a56 | 1618 | |
077c212f JL |
1619 | return inode_set_atime_to_ts(inode, ts); |
1620 | } | |
1621 | ||
1622 | static inline time64_t inode_get_mtime_sec(const struct inode *inode) | |
1623 | { | |
3aa63a56 | 1624 | return inode->i_mtime_sec; |
077c212f JL |
1625 | } |
1626 | ||
1627 | static inline long inode_get_mtime_nsec(const struct inode *inode) | |
1628 | { | |
3aa63a56 | 1629 | return inode->i_mtime_nsec; |
077c212f JL |
1630 | } |
1631 | ||
1632 | static inline struct timespec64 inode_get_mtime(const struct inode *inode) | |
1633 | { | |
3aa63a56 JL |
1634 | struct timespec64 ts = { .tv_sec = inode_get_mtime_sec(inode), |
1635 | .tv_nsec = inode_get_mtime_nsec(inode) }; | |
1636 | return ts; | |
077c212f JL |
1637 | } |
1638 | ||
1639 | static inline struct timespec64 inode_set_mtime_to_ts(struct inode *inode, | |
1640 | struct timespec64 ts) | |
1641 | { | |
3aa63a56 JL |
1642 | inode->i_mtime_sec = ts.tv_sec; |
1643 | inode->i_mtime_nsec = ts.tv_nsec; | |
077c212f JL |
1644 | return ts; |
1645 | } | |
1646 | ||
1647 | static inline struct timespec64 inode_set_mtime(struct inode *inode, | |
1648 | time64_t sec, long nsec) | |
1649 | { | |
1650 | struct timespec64 ts = { .tv_sec = sec, | |
1651 | .tv_nsec = nsec }; | |
1652 | return inode_set_mtime_to_ts(inode, ts); | |
1653 | } | |
1654 | ||
1655 | static inline time64_t inode_get_ctime_sec(const struct inode *inode) | |
1656 | { | |
3aa63a56 | 1657 | return inode->i_ctime_sec; |
077c212f JL |
1658 | } |
1659 | ||
1660 | static inline long inode_get_ctime_nsec(const struct inode *inode) | |
1661 | { | |
3aa63a56 | 1662 | return inode->i_ctime_nsec; |
077c212f JL |
1663 | } |
1664 | ||
9b6304c1 JL |
1665 | static inline struct timespec64 inode_get_ctime(const struct inode *inode) |
1666 | { | |
3aa63a56 JL |
1667 | struct timespec64 ts = { .tv_sec = inode_get_ctime_sec(inode), |
1668 | .tv_nsec = inode_get_ctime_nsec(inode) }; | |
1669 | ||
1670 | return ts; | |
9b6304c1 JL |
1671 | } |
1672 | ||
9b6304c1 JL |
1673 | static inline struct timespec64 inode_set_ctime_to_ts(struct inode *inode, |
1674 | struct timespec64 ts) | |
1675 | { | |
3aa63a56 JL |
1676 | inode->i_ctime_sec = ts.tv_sec; |
1677 | inode->i_ctime_nsec = ts.tv_nsec; | |
9b6304c1 JL |
1678 | return ts; |
1679 | } | |
1680 | ||
1681 | /** | |
1682 | * inode_set_ctime - set the ctime in the inode | |
1683 | * @inode: inode in which to set the ctime | |
1684 | * @sec: tv_sec value to set | |
1685 | * @nsec: tv_nsec value to set | |
1686 | * | |
1687 | * Set the ctime in @inode to { @sec, @nsec } | |
1688 | */ | |
1689 | static inline struct timespec64 inode_set_ctime(struct inode *inode, | |
1690 | time64_t sec, long nsec) | |
1691 | { | |
1692 | struct timespec64 ts = { .tv_sec = sec, | |
1693 | .tv_nsec = nsec }; | |
1694 | ||
1695 | return inode_set_ctime_to_ts(inode, ts); | |
1696 | } | |
1da177e4 | 1697 | |
077c212f JL |
1698 | struct timespec64 simple_inode_init_ts(struct inode *inode); |
1699 | ||
1da177e4 LT |
1700 | /* |
1701 | * Snapshotting support. | |
1702 | */ | |
1da177e4 | 1703 | |
9b852342 DW |
1704 | /* |
1705 | * These are internal functions, please use sb_start_{write,pagefault,intwrite} | |
1706 | * instead. | |
1707 | */ | |
1708 | static inline void __sb_end_write(struct super_block *sb, int level) | |
1709 | { | |
1710 | percpu_up_read(sb->s_writers.rw_sem + level-1); | |
1711 | } | |
1712 | ||
1713 | static inline void __sb_start_write(struct super_block *sb, int level) | |
1714 | { | |
1715 | percpu_down_read(sb->s_writers.rw_sem + level - 1); | |
1716 | } | |
1717 | ||
1718 | static inline bool __sb_start_write_trylock(struct super_block *sb, int level) | |
1719 | { | |
1720 | return percpu_down_read_trylock(sb->s_writers.rw_sem + level - 1); | |
1721 | } | |
5accdf82 | 1722 | |
bee9182d | 1723 | #define __sb_writers_acquired(sb, lev) \ |
8129ed29 | 1724 | percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_) |
bee9182d | 1725 | #define __sb_writers_release(sb, lev) \ |
c01a5d89 | 1726 | percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], _THIS_IP_) |
bee9182d | 1727 | |
8802e580 AG |
1728 | /** |
1729 | * __sb_write_started - check if sb freeze level is held | |
1730 | * @sb: the super we write to | |
1731 | * @level: the freeze level | |
1732 | * | |
c39e2ae3 VN |
1733 | * * > 0 - sb freeze level is held |
1734 | * * 0 - sb freeze level is not held | |
1735 | * * < 0 - !CONFIG_LOCKDEP/LOCK_STATE_UNKNOWN | |
8802e580 AG |
1736 | */ |
1737 | static inline int __sb_write_started(const struct super_block *sb, int level) | |
1738 | { | |
1739 | return lockdep_is_held_type(sb->s_writers.rw_sem + level - 1, 1); | |
1740 | } | |
1741 | ||
3d5cd491 AG |
1742 | /** |
1743 | * sb_write_started - check if SB_FREEZE_WRITE is held | |
1744 | * @sb: the super we write to | |
1745 | * | |
1746 | * May be false positive with !CONFIG_LOCKDEP/LOCK_STATE_UNKNOWN. | |
1747 | */ | |
7f8d12ea NA |
1748 | static inline bool sb_write_started(const struct super_block *sb) |
1749 | { | |
8802e580 | 1750 | return __sb_write_started(sb, SB_FREEZE_WRITE); |
7f8d12ea NA |
1751 | } |
1752 | ||
21b32e6a AG |
1753 | /** |
1754 | * sb_write_not_started - check if SB_FREEZE_WRITE is not held | |
1755 | * @sb: the super we write to | |
1756 | * | |
1757 | * May be false positive with !CONFIG_LOCKDEP/LOCK_STATE_UNKNOWN. | |
1758 | */ | |
1759 | static inline bool sb_write_not_started(const struct super_block *sb) | |
1760 | { | |
1761 | return __sb_write_started(sb, SB_FREEZE_WRITE) <= 0; | |
1762 | } | |
1763 | ||
3d5cd491 AG |
1764 | /** |
1765 | * file_write_started - check if SB_FREEZE_WRITE is held | |
1766 | * @file: the file we write to | |
1767 | * | |
1768 | * May be false positive with !CONFIG_LOCKDEP/LOCK_STATE_UNKNOWN. | |
1769 | * May be false positive with !S_ISREG, because file_start_write() has | |
1770 | * no effect on !S_ISREG. | |
1771 | */ | |
1772 | static inline bool file_write_started(const struct file *file) | |
1773 | { | |
1774 | if (!S_ISREG(file_inode(file)->i_mode)) | |
1775 | return true; | |
1776 | return sb_write_started(file_inode(file)->i_sb); | |
1777 | } | |
1778 | ||
21b32e6a AG |
1779 | /** |
1780 | * file_write_not_started - check if SB_FREEZE_WRITE is not held | |
1781 | * @file: the file we write to | |
1782 | * | |
1783 | * May be false positive with !CONFIG_LOCKDEP/LOCK_STATE_UNKNOWN. | |
1784 | * May be false positive with !S_ISREG, because file_start_write() has | |
1785 | * no effect on !S_ISREG. | |
1786 | */ | |
1787 | static inline bool file_write_not_started(const struct file *file) | |
1788 | { | |
1789 | if (!S_ISREG(file_inode(file)->i_mode)) | |
1790 | return true; | |
1791 | return sb_write_not_started(file_inode(file)->i_sb); | |
7f8d12ea NA |
1792 | } |
1793 | ||
5accdf82 JK |
1794 | /** |
1795 | * sb_end_write - drop write access to a superblock | |
1796 | * @sb: the super we wrote to | |
1797 | * | |
1798 | * Decrement number of writers to the filesystem. Wake up possible waiters | |
1799 | * wanting to freeze the filesystem. | |
1800 | */ | |
1801 | static inline void sb_end_write(struct super_block *sb) | |
1802 | { | |
1803 | __sb_end_write(sb, SB_FREEZE_WRITE); | |
1804 | } | |
1805 | ||
1806 | /** | |
1807 | * sb_end_pagefault - drop write access to a superblock from a page fault | |
1808 | * @sb: the super we wrote to | |
1809 | * | |
1810 | * Decrement number of processes handling write page fault to the filesystem. | |
1811 | * Wake up possible waiters wanting to freeze the filesystem. | |
1812 | */ | |
1813 | static inline void sb_end_pagefault(struct super_block *sb) | |
1814 | { | |
1815 | __sb_end_write(sb, SB_FREEZE_PAGEFAULT); | |
1816 | } | |
1817 | ||
1818 | /** | |
1819 | * sb_end_intwrite - drop write access to a superblock for internal fs purposes | |
1820 | * @sb: the super we wrote to | |
1821 | * | |
1822 | * Decrement fs-internal number of writers to the filesystem. Wake up possible | |
1823 | * waiters wanting to freeze the filesystem. | |
1824 | */ | |
1825 | static inline void sb_end_intwrite(struct super_block *sb) | |
1826 | { | |
1827 | __sb_end_write(sb, SB_FREEZE_FS); | |
1828 | } | |
1829 | ||
1830 | /** | |
1831 | * sb_start_write - get write access to a superblock | |
1832 | * @sb: the super we write to | |
1833 | * | |
1834 | * When a process wants to write data or metadata to a file system (i.e. dirty | |
1835 | * a page or an inode), it should embed the operation in a sb_start_write() - | |
1836 | * sb_end_write() pair to get exclusion against file system freezing. This | |
1837 | * function increments number of writers preventing freezing. If the file | |
1838 | * system is already frozen, the function waits until the file system is | |
1839 | * thawed. | |
1840 | * | |
1841 | * Since freeze protection behaves as a lock, users have to preserve | |
1842 | * ordering of freeze protection and other filesystem locks. Generally, | |
1843 | * freeze protection should be the outermost lock. In particular, we have: | |
1844 | * | |
1845 | * sb_start_write | |
1846 | * -> i_mutex (write path, truncate, directory ops, ...) | |
1847 | * -> s_umount (freeze_super, thaw_super) | |
1848 | */ | |
1849 | static inline void sb_start_write(struct super_block *sb) | |
1850 | { | |
8a3c84b6 | 1851 | __sb_start_write(sb, SB_FREEZE_WRITE); |
5accdf82 JK |
1852 | } |
1853 | ||
8a3c84b6 | 1854 | static inline bool sb_start_write_trylock(struct super_block *sb) |
5accdf82 | 1855 | { |
8a3c84b6 | 1856 | return __sb_start_write_trylock(sb, SB_FREEZE_WRITE); |
5accdf82 JK |
1857 | } |
1858 | ||
1859 | /** | |
1860 | * sb_start_pagefault - get write access to a superblock from a page fault | |
1861 | * @sb: the super we write to | |
1862 | * | |
1863 | * When a process starts handling write page fault, it should embed the | |
1864 | * operation into sb_start_pagefault() - sb_end_pagefault() pair to get | |
1865 | * exclusion against file system freezing. This is needed since the page fault | |
1866 | * is going to dirty a page. This function increments number of running page | |
1867 | * faults preventing freezing. If the file system is already frozen, the | |
1868 | * function waits until the file system is thawed. | |
1869 | * | |
1870 | * Since page fault freeze protection behaves as a lock, users have to preserve | |
1871 | * ordering of freeze protection and other filesystem locks. It is advised to | |
c1e8d7c6 | 1872 | * put sb_start_pagefault() close to mmap_lock in lock ordering. Page fault |
5accdf82 JK |
1873 | * handling code implies lock dependency: |
1874 | * | |
c1e8d7c6 | 1875 | * mmap_lock |
5accdf82 JK |
1876 | * -> sb_start_pagefault |
1877 | */ | |
1878 | static inline void sb_start_pagefault(struct super_block *sb) | |
1879 | { | |
8a3c84b6 | 1880 | __sb_start_write(sb, SB_FREEZE_PAGEFAULT); |
5accdf82 JK |
1881 | } |
1882 | ||
39015399 | 1883 | /** |
5accdf82 JK |
1884 | * sb_start_intwrite - get write access to a superblock for internal fs purposes |
1885 | * @sb: the super we write to | |
1886 | * | |
1887 | * This is the third level of protection against filesystem freezing. It is | |
1888 | * free for use by a filesystem. The only requirement is that it must rank | |
1889 | * below sb_start_pagefault. | |
1890 | * | |
1891 | * For example filesystem can call sb_start_intwrite() when starting a | |
1892 | * transaction which somewhat eases handling of freezing for internal sources | |
1893 | * of filesystem changes (internal fs threads, discarding preallocation on file | |
1894 | * close, etc.). | |
1895 | */ | |
1896 | static inline void sb_start_intwrite(struct super_block *sb) | |
1897 | { | |
8a3c84b6 | 1898 | __sb_start_write(sb, SB_FREEZE_FS); |
5accdf82 | 1899 | } |
1da177e4 | 1900 | |
8a3c84b6 | 1901 | static inline bool sb_start_intwrite_trylock(struct super_block *sb) |
0c8e3fe3 | 1902 | { |
8a3c84b6 | 1903 | return __sb_start_write_trylock(sb, SB_FREEZE_FS); |
0c8e3fe3 AG |
1904 | } |
1905 | ||
01beba79 | 1906 | bool inode_owner_or_capable(struct mnt_idmap *idmap, |
21cb47be | 1907 | const struct inode *inode); |
3bd858ab | 1908 | |
1da177e4 LT |
1909 | /* |
1910 | * VFS helper functions.. | |
1911 | */ | |
abf08576 | 1912 | int vfs_create(struct mnt_idmap *, struct inode *, |
6521f891 | 1913 | struct dentry *, umode_t, bool); |
abf08576 | 1914 | int vfs_mkdir(struct mnt_idmap *, struct inode *, |
6521f891 | 1915 | struct dentry *, umode_t); |
abf08576 | 1916 | int vfs_mknod(struct mnt_idmap *, struct inode *, struct dentry *, |
6521f891 | 1917 | umode_t, dev_t); |
abf08576 | 1918 | int vfs_symlink(struct mnt_idmap *, struct inode *, |
6521f891 | 1919 | struct dentry *, const char *); |
abf08576 | 1920 | int vfs_link(struct dentry *, struct mnt_idmap *, struct inode *, |
6521f891 | 1921 | struct dentry *, struct inode **); |
abf08576 CB |
1922 | int vfs_rmdir(struct mnt_idmap *, struct inode *, struct dentry *); |
1923 | int vfs_unlink(struct mnt_idmap *, struct inode *, struct dentry *, | |
6521f891 | 1924 | struct inode **); |
9fe61450 | 1925 | |
92cb01c7 CB |
1926 | /** |
1927 | * struct renamedata - contains all information required for renaming | |
abf08576 | 1928 | * @old_mnt_idmap: idmap of the old mount the inode was found from |
92cb01c7 CB |
1929 | * @old_dir: parent of source |
1930 | * @old_dentry: source | |
abf08576 | 1931 | * @new_mnt_idmap: idmap of the new mount the inode was found from |
92cb01c7 CB |
1932 | * @new_dir: parent of destination |
1933 | * @new_dentry: destination | |
1934 | * @delegated_inode: returns an inode needing a delegation break | |
1935 | * @flags: rename flags | |
1936 | */ | |
9fe61450 | 1937 | struct renamedata { |
abf08576 | 1938 | struct mnt_idmap *old_mnt_idmap; |
9fe61450 CB |
1939 | struct inode *old_dir; |
1940 | struct dentry *old_dentry; | |
abf08576 | 1941 | struct mnt_idmap *new_mnt_idmap; |
9fe61450 CB |
1942 | struct inode *new_dir; |
1943 | struct dentry *new_dentry; | |
1944 | struct inode **delegated_inode; | |
1945 | unsigned int flags; | |
1946 | } __randomize_layout; | |
a3c751a5 | 1947 | |
9fe61450 | 1948 | int vfs_rename(struct renamedata *); |
a3c751a5 | 1949 | |
abf08576 | 1950 | static inline int vfs_whiteout(struct mnt_idmap *idmap, |
6521f891 | 1951 | struct inode *dir, struct dentry *dentry) |
a3c751a5 | 1952 | { |
abf08576 | 1953 | return vfs_mknod(idmap, dir, dentry, S_IFCHR | WHITEOUT_MODE, |
6521f891 | 1954 | WHITEOUT_DEV); |
a3c751a5 | 1955 | } |
1da177e4 | 1956 | |
d56e0ddb AG |
1957 | struct file *kernel_tmpfile_open(struct mnt_idmap *idmap, |
1958 | const struct path *parentpath, | |
1959 | umode_t mode, int open_flag, | |
1960 | const struct cred *cred); | |
cbb0b9d4 | 1961 | struct file *kernel_file_open(const struct path *path, int flags, |
af58dc1f | 1962 | const struct cred *cred); |
af7bd4dc | 1963 | |
8e6c848e AV |
1964 | int vfs_mkobj(struct dentry *, umode_t, |
1965 | int (*f)(struct dentry *, umode_t, void *), | |
1966 | void *); | |
1967 | ||
c04011fe | 1968 | int vfs_fchown(struct file *file, uid_t user, gid_t group); |
9e96c8c0 | 1969 | int vfs_fchmod(struct file *file, umode_t mode); |
fd5ad30c | 1970 | int vfs_utimes(const struct path *path, struct timespec64 *times); |
c04011fe | 1971 | |
9df6702a MS |
1972 | extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg); |
1973 | ||
2952db0f AB |
1974 | #ifdef CONFIG_COMPAT |
1975 | extern long compat_ptr_ioctl(struct file *file, unsigned int cmd, | |
1976 | unsigned long arg); | |
1977 | #else | |
1978 | #define compat_ptr_ioctl NULL | |
1979 | #endif | |
1980 | ||
8c744fb8 CH |
1981 | /* |
1982 | * VFS file helper functions. | |
1983 | */ | |
f2d40141 | 1984 | void inode_init_owner(struct mnt_idmap *idmap, struct inode *inode, |
21cb47be | 1985 | const struct inode *dir, umode_t mode); |
a2982cc9 | 1986 | extern bool may_open_dev(const struct path *path); |
9452e93e | 1987 | umode_t mode_strip_sgid(struct mnt_idmap *idmap, |
2b3416ce | 1988 | const struct inode *dir, umode_t mode); |
9b6a14f0 YT |
1989 | bool in_group_or_capable(struct mnt_idmap *idmap, |
1990 | const struct inode *inode, vfsgid_t vfsgid); | |
c4b929b8 | 1991 | |
1da177e4 LT |
1992 | /* |
1993 | * This is the "filldir" function type, used by readdir() to let | |
1994 | * the kernel specify what kind of dirent layout it wants to have. | |
1995 | * This allows the kernel to read directories into kernel space or | |
1996 | * to have different dirent layouts depending on the binary type. | |
25885a35 | 1997 | * Return 'true' to keep going and 'false' if there are no more entries. |
1da177e4 | 1998 | */ |
ac7576f4 | 1999 | struct dir_context; |
25885a35 | 2000 | typedef bool (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64, |
ac7576f4 MS |
2001 | unsigned); |
2002 | ||
5c0ba4e0 | 2003 | struct dir_context { |
a09acf4b | 2004 | filldir_t actor; |
bb6f619b | 2005 | loff_t pos; |
5c0ba4e0 | 2006 | }; |
bb6f619b | 2007 | |
b4caecd4 CH |
2008 | /* |
2009 | * These flags let !MMU mmap() govern direct device mapping vs immediate | |
2010 | * copying more easily for MAP_PRIVATE, especially for ROM filesystems. | |
2011 | * | |
2012 | * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE) | |
2013 | * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED) | |
2014 | * NOMMU_MAP_READ: Can be mapped for reading | |
2015 | * NOMMU_MAP_WRITE: Can be mapped for writing | |
2016 | * NOMMU_MAP_EXEC: Can be mapped for execution | |
2017 | */ | |
2018 | #define NOMMU_MAP_COPY 0x00000001 | |
2019 | #define NOMMU_MAP_DIRECT 0x00000008 | |
2020 | #define NOMMU_MAP_READ VM_MAYREAD | |
2021 | #define NOMMU_MAP_WRITE VM_MAYWRITE | |
2022 | #define NOMMU_MAP_EXEC VM_MAYEXEC | |
2023 | ||
2024 | #define NOMMU_VMFLAGS \ | |
2025 | (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC) | |
2026 | ||
2e5dfc99 DW |
2027 | /* |
2028 | * These flags control the behavior of the remap_file_range function pointer. | |
2029 | * If it is called with len == 0 that means "remap to end of source file". | |
5c437fa2 | 2030 | * See Documentation/filesystems/vfs.rst for more details about this call. |
2e5dfc99 DW |
2031 | * |
2032 | * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate) | |
eca3654e | 2033 | * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request |
2e5dfc99 DW |
2034 | */ |
2035 | #define REMAP_FILE_DEDUP (1 << 0) | |
eca3654e | 2036 | #define REMAP_FILE_CAN_SHORTEN (1 << 1) |
2e5dfc99 DW |
2037 | |
2038 | /* | |
2039 | * These flags signal that the caller is ok with altering various aspects of | |
2040 | * the behavior of the remap operation. The changes must be made by the | |
2041 | * implementation; the vfs remap helper functions can take advantage of them. | |
2042 | * Flags in this category exist to preserve the quirky behavior of the hoisted | |
2043 | * btrfs clone/dedupe ioctls. | |
2e5dfc99 | 2044 | */ |
eca3654e | 2045 | #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN) |
b4caecd4 | 2046 | |
10bc8e4a AG |
2047 | /* |
2048 | * These flags control the behavior of vfs_copy_file_range(). | |
2049 | * They are not available to the user via syscall. | |
2050 | * | |
2051 | * COPY_FILE_SPLICE: call splice direct instead of fs clone/copy ops | |
2052 | */ | |
2053 | #define COPY_FILE_SPLICE (1 << 0) | |
2054 | ||
293bc982 | 2055 | struct iov_iter; |
ee692a21 | 2056 | struct io_uring_cmd; |
6faddda6 | 2057 | struct offset_ctx; |
293bc982 | 2058 | |
210a03c9 CB |
2059 | typedef unsigned int __bitwise fop_flags_t; |
2060 | ||
1da177e4 LT |
2061 | struct file_operations { |
2062 | struct module *owner; | |
210a03c9 | 2063 | fop_flags_t fop_flags; |
1da177e4 LT |
2064 | loff_t (*llseek) (struct file *, loff_t, int); |
2065 | ssize_t (*read) (struct file *, char __user *, size_t, loff_t *); | |
1da177e4 | 2066 | ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *); |
293bc982 AV |
2067 | ssize_t (*read_iter) (struct kiocb *, struct iov_iter *); |
2068 | ssize_t (*write_iter) (struct kiocb *, struct iov_iter *); | |
5a72e899 JA |
2069 | int (*iopoll)(struct kiocb *kiocb, struct io_comp_batch *, |
2070 | unsigned int flags); | |
61922694 | 2071 | int (*iterate_shared) (struct file *, struct dir_context *); |
a3f8683b | 2072 | __poll_t (*poll) (struct file *, struct poll_table_struct *); |
1da177e4 LT |
2073 | long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long); |
2074 | long (*compat_ioctl) (struct file *, unsigned int, unsigned long); | |
2075 | int (*mmap) (struct file *, struct vm_area_struct *); | |
2076 | int (*open) (struct inode *, struct file *); | |
75e1fcc0 | 2077 | int (*flush) (struct file *, fl_owner_t id); |
1da177e4 | 2078 | int (*release) (struct inode *, struct file *); |
02c24a82 | 2079 | int (*fsync) (struct file *, loff_t, loff_t, int datasync); |
1da177e4 LT |
2080 | int (*fasync) (int, struct file *, int); |
2081 | int (*lock) (struct file *, int, struct file_lock *); | |
1da177e4 LT |
2082 | unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); |
2083 | int (*check_flags)(int); | |
1da177e4 | 2084 | int (*flock) (struct file *, int, struct file_lock *); |
cbb7e577 JA |
2085 | ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int); |
2086 | ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int); | |
2bfc6685 | 2087 | void (*splice_eof)(struct file *file); |
c69ff407 | 2088 | int (*setlease)(struct file *, int, struct file_lease **, void **); |
2fe17c10 CH |
2089 | long (*fallocate)(struct file *file, int mode, loff_t offset, |
2090 | loff_t len); | |
a3816ab0 | 2091 | void (*show_fdinfo)(struct seq_file *m, struct file *f); |
b4caecd4 CH |
2092 | #ifndef CONFIG_MMU |
2093 | unsigned (*mmap_capabilities)(struct file *); | |
2094 | #endif | |
04b38d60 CH |
2095 | ssize_t (*copy_file_range)(struct file *, loff_t, struct file *, |
2096 | loff_t, size_t, unsigned int); | |
42ec3d4c DW |
2097 | loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in, |
2098 | struct file *file_out, loff_t pos_out, | |
2099 | loff_t len, unsigned int remap_flags); | |
45cd0faa | 2100 | int (*fadvise)(struct file *, loff_t, loff_t, int); |
ee692a21 | 2101 | int (*uring_cmd)(struct io_uring_cmd *ioucmd, unsigned int issue_flags); |
de97fcb3 JA |
2102 | int (*uring_cmd_iopoll)(struct io_uring_cmd *, struct io_comp_batch *, |
2103 | unsigned int poll_flags); | |
3859a271 | 2104 | } __randomize_layout; |
1da177e4 | 2105 | |
210a03c9 CB |
2106 | /* Supports async buffered reads */ |
2107 | #define FOP_BUFFER_RASYNC ((__force fop_flags_t)(1 << 0)) | |
2108 | /* Supports async buffered writes */ | |
2109 | #define FOP_BUFFER_WASYNC ((__force fop_flags_t)(1 << 1)) | |
2110 | /* Supports synchronous page faults for mappings */ | |
2111 | #define FOP_MMAP_SYNC ((__force fop_flags_t)(1 << 2)) | |
2112 | /* Supports non-exclusive O_DIRECT writes from multiple threads */ | |
2113 | #define FOP_DIO_PARALLEL_WRITE ((__force fop_flags_t)(1 << 3)) | |
886b94d2 MWO |
2114 | /* Contains huge pages */ |
2115 | #define FOP_HUGE_PAGES ((__force fop_flags_t)(1 << 4)) | |
641bb439 CB |
2116 | /* Treat loff_t as unsigned (e.g., /dev/mem) */ |
2117 | #define FOP_UNSIGNED_OFFSET ((__force fop_flags_t)(1 << 5)) | |
210a03c9 | 2118 | |
3e327154 LT |
2119 | /* Wrap a directory iterator that needs exclusive inode access */ |
2120 | int wrap_directory_iterator(struct file *, struct dir_context *, | |
2121 | int (*) (struct file *, struct dir_context *)); | |
2122 | #define WRAP_DIR_ITER(x) \ | |
2123 | static int shared_##x(struct file *file , struct dir_context *ctx) \ | |
2124 | { return wrap_directory_iterator(file, ctx, x); } | |
2125 | ||
1da177e4 | 2126 | struct inode_operations { |
00cd8dd3 | 2127 | struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int); |
fceef393 | 2128 | const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *); |
4609e1f1 | 2129 | int (*permission) (struct mnt_idmap *, struct inode *, int); |
cac2f8b8 | 2130 | struct posix_acl * (*get_inode_acl)(struct inode *, int, bool); |
44a7d7a8 NP |
2131 | |
2132 | int (*readlink) (struct dentry *, char __user *,int); | |
44a7d7a8 | 2133 | |
6c960e68 | 2134 | int (*create) (struct mnt_idmap *, struct inode *,struct dentry *, |
549c7297 | 2135 | umode_t, bool); |
1da177e4 LT |
2136 | int (*link) (struct dentry *,struct inode *,struct dentry *); |
2137 | int (*unlink) (struct inode *,struct dentry *); | |
7a77db95 | 2138 | int (*symlink) (struct mnt_idmap *, struct inode *,struct dentry *, |
549c7297 | 2139 | const char *); |
c54bd91e | 2140 | int (*mkdir) (struct mnt_idmap *, struct inode *,struct dentry *, |
549c7297 | 2141 | umode_t); |
1da177e4 | 2142 | int (*rmdir) (struct inode *,struct dentry *); |
5ebb29be | 2143 | int (*mknod) (struct mnt_idmap *, struct inode *,struct dentry *, |
549c7297 | 2144 | umode_t,dev_t); |
e18275ae | 2145 | int (*rename) (struct mnt_idmap *, struct inode *, struct dentry *, |
520c8b16 | 2146 | struct inode *, struct dentry *, unsigned int); |
c1632a0f | 2147 | int (*setattr) (struct mnt_idmap *, struct dentry *, struct iattr *); |
b74d24f7 | 2148 | int (*getattr) (struct mnt_idmap *, const struct path *, |
549c7297 | 2149 | struct kstat *, u32, unsigned int); |
1da177e4 | 2150 | ssize_t (*listxattr) (struct dentry *, char *, size_t); |
c4b929b8 MF |
2151 | int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start, |
2152 | u64 len); | |
913e9928 | 2153 | int (*update_time)(struct inode *, int); |
d9585277 | 2154 | int (*atomic_open)(struct inode *, struct dentry *, |
30d90494 | 2155 | struct file *, unsigned open_flag, |
44907d79 | 2156 | umode_t create_mode); |
011e2b71 | 2157 | int (*tmpfile) (struct mnt_idmap *, struct inode *, |
863f144f | 2158 | struct file *, umode_t); |
77435322 | 2159 | struct posix_acl *(*get_acl)(struct mnt_idmap *, struct dentry *, |
7420332a | 2160 | int); |
13e83a49 | 2161 | int (*set_acl)(struct mnt_idmap *, struct dentry *, |
549c7297 | 2162 | struct posix_acl *, int); |
8782a9ae | 2163 | int (*fileattr_set)(struct mnt_idmap *idmap, |
4c5b4799 MS |
2164 | struct dentry *dentry, struct fileattr *fa); |
2165 | int (*fileattr_get)(struct dentry *dentry, struct fileattr *fa); | |
6faddda6 | 2166 | struct offset_ctx *(*get_offset_ctx)(struct inode *inode); |
44a7d7a8 | 2167 | } ____cacheline_aligned; |
1da177e4 | 2168 | |
f74ac015 MS |
2169 | static inline int call_mmap(struct file *file, struct vm_area_struct *vma) |
2170 | { | |
2171 | return file->f_op->mmap(file, vma); | |
2172 | } | |
2173 | ||
1da177e4 LT |
2174 | extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *); |
2175 | extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *); | |
29732938 ZB |
2176 | extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *, |
2177 | loff_t, size_t, unsigned int); | |
5b9932f6 | 2178 | int remap_verify_area(struct file *file, loff_t pos, loff_t len, bool write); |
6f7db389 SR |
2179 | int __generic_remap_file_range_prep(struct file *file_in, loff_t pos_in, |
2180 | struct file *file_out, loff_t pos_out, | |
2181 | loff_t *len, unsigned int remap_flags, | |
2182 | const struct iomap_ops *dax_read_ops); | |
2183 | int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in, | |
2184 | struct file *file_out, loff_t pos_out, | |
2185 | loff_t *count, unsigned int remap_flags); | |
42ec3d4c DW |
2186 | extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in, |
2187 | struct file *file_out, loff_t pos_out, | |
452ce659 | 2188 | loff_t len, unsigned int remap_flags); |
54dbc151 DW |
2189 | extern int vfs_dedupe_file_range(struct file *file, |
2190 | struct file_dedupe_range *same); | |
42ec3d4c DW |
2191 | extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos, |
2192 | struct file *dst_file, loff_t dst_pos, | |
df365836 | 2193 | loff_t len, unsigned int remap_flags); |
f1825366 | 2194 | |
7366f8b6 CB |
2195 | /** |
2196 | * enum freeze_holder - holder of the freeze | |
2197 | * @FREEZE_HOLDER_KERNEL: kernel wants to freeze or thaw filesystem | |
2198 | * @FREEZE_HOLDER_USERSPACE: userspace wants to freeze or thaw filesystem | |
2199 | * @FREEZE_MAY_NEST: whether nesting freeze and thaw requests is allowed | |
2200 | * | |
2201 | * Indicate who the owner of the freeze or thaw request is and whether | |
2202 | * the freeze needs to be exclusive or can nest. | |
2203 | * Without @FREEZE_MAY_NEST, multiple freeze and thaw requests from the | |
2204 | * same holder aren't allowed. It is however allowed to hold a single | |
2205 | * @FREEZE_HOLDER_USERSPACE and a single @FREEZE_HOLDER_KERNEL freeze at | |
2206 | * the same time. This is relied upon by some filesystems during online | |
2207 | * repair or similar. | |
2208 | */ | |
880b9577 DW |
2209 | enum freeze_holder { |
2210 | FREEZE_HOLDER_KERNEL = (1U << 0), | |
2211 | FREEZE_HOLDER_USERSPACE = (1U << 1), | |
7366f8b6 | 2212 | FREEZE_MAY_NEST = (1U << 2), |
880b9577 | 2213 | }; |
1da177e4 | 2214 | |
1da177e4 LT |
2215 | struct super_operations { |
2216 | struct inode *(*alloc_inode)(struct super_block *sb); | |
2217 | void (*destroy_inode)(struct inode *); | |
fdb0da89 | 2218 | void (*free_inode)(struct inode *); |
1da177e4 | 2219 | |
aa385729 | 2220 | void (*dirty_inode) (struct inode *, int flags); |
a9185b41 | 2221 | int (*write_inode) (struct inode *, struct writeback_control *wbc); |
45321ac5 | 2222 | int (*drop_inode) (struct inode *); |
be7ce416 | 2223 | void (*evict_inode) (struct inode *); |
1da177e4 | 2224 | void (*put_super) (struct super_block *); |
1da177e4 | 2225 | int (*sync_fs)(struct super_block *sb, int wait); |
880b9577 | 2226 | int (*freeze_super) (struct super_block *, enum freeze_holder who); |
c4be0c1d | 2227 | int (*freeze_fs) (struct super_block *); |
880b9577 | 2228 | int (*thaw_super) (struct super_block *, enum freeze_holder who); |
c4be0c1d | 2229 | int (*unfreeze_fs) (struct super_block *); |
726c3342 | 2230 | int (*statfs) (struct dentry *, struct kstatfs *); |
1da177e4 | 2231 | int (*remount_fs) (struct super_block *, int *, char *); |
42faad99 | 2232 | void (*umount_begin) (struct super_block *); |
1da177e4 | 2233 | |
34c80b1d | 2234 | int (*show_options)(struct seq_file *, struct dentry *); |
d861c630 | 2235 | int (*show_devname)(struct seq_file *, struct dentry *); |
a6322de6 | 2236 | int (*show_path)(struct seq_file *, struct dentry *); |
64132379 | 2237 | int (*show_stats)(struct seq_file *, struct dentry *); |
0e51a720 | 2238 | #ifdef CONFIG_QUOTA |
1da177e4 LT |
2239 | ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t); |
2240 | ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t); | |
ccb49011 | 2241 | struct dquot __rcu **(*get_dquots)(struct inode *); |
0e51a720 | 2242 | #endif |
4101b624 VD |
2243 | long (*nr_cached_objects)(struct super_block *, |
2244 | struct shrink_control *); | |
2245 | long (*free_cached_objects)(struct super_block *, | |
2246 | struct shrink_control *); | |
87efb390 | 2247 | void (*shutdown)(struct super_block *sb); |
1da177e4 LT |
2248 | }; |
2249 | ||
bbc1096a DH |
2250 | /* |
2251 | * Inode flags - they have no relation to superblock flags now | |
2252 | */ | |
6414e9b0 EB |
2253 | #define S_SYNC (1 << 0) /* Writes are synced at once */ |
2254 | #define S_NOATIME (1 << 1) /* Do not update access times */ | |
2255 | #define S_APPEND (1 << 2) /* Append-only file */ | |
2256 | #define S_IMMUTABLE (1 << 3) /* Immutable file */ | |
2257 | #define S_DEAD (1 << 4) /* removed, but still open directory */ | |
2258 | #define S_NOQUOTA (1 << 5) /* Inode is not counted to quota */ | |
2259 | #define S_DIRSYNC (1 << 6) /* Directory modifications are synchronous */ | |
2260 | #define S_NOCMTIME (1 << 7) /* Do not update file c/mtime */ | |
2261 | #define S_SWAPFILE (1 << 8) /* Do not truncate: swapon got its bmaps */ | |
2262 | #define S_PRIVATE (1 << 9) /* Inode is fs-internal */ | |
2263 | #define S_IMA (1 << 10) /* Inode has an associated IMA struct */ | |
2264 | #define S_AUTOMOUNT (1 << 11) /* Automount/referral quasi-directory */ | |
2265 | #define S_NOSEC (1 << 12) /* no suid or xattr security attributes */ | |
6cd176a5 | 2266 | #ifdef CONFIG_FS_DAX |
6414e9b0 | 2267 | #define S_DAX (1 << 13) /* Direct Access, avoiding the page cache */ |
fbbbad4b | 2268 | #else |
6414e9b0 | 2269 | #define S_DAX 0 /* Make all the DAX code disappear */ |
fbbbad4b | 2270 | #endif |
6414e9b0 EB |
2271 | #define S_ENCRYPTED (1 << 14) /* Encrypted file (using fs/crypto/) */ |
2272 | #define S_CASEFOLD (1 << 15) /* Casefolded file */ | |
2273 | #define S_VERITY (1 << 16) /* Verity file (using fs/verity/) */ | |
1bd9c4e4 | 2274 | #define S_KERNEL_FILE (1 << 17) /* File is in use by the kernel (eg. fs/cachefiles) */ |
bbc1096a DH |
2275 | |
2276 | /* | |
2277 | * Note that nosuid etc flags are inode-specific: setting some file-system | |
2278 | * flags just means all the inodes inherit those flags by default. It might be | |
2279 | * possible to override it selectively if you really wanted to with some | |
2280 | * ioctl() that is not currently implemented. | |
2281 | * | |
e462ec50 | 2282 | * Exception: SB_RDONLY is always applied to the entire file system. |
bbc1096a DH |
2283 | * |
2284 | * Unfortunately, it is possible to change a filesystems flags with it mounted | |
2285 | * with files in use. This means that all of the inodes will not have their | |
2286 | * i_flags updated. Hence, i_flags no longer inherit the superblock mount | |
2287 | * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org | |
2288 | */ | |
2289 | #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg)) | |
2290 | ||
1751e8a6 | 2291 | static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; } |
94e92e7a | 2292 | #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb) |
e462ec50 | 2293 | #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \ |
bbc1096a | 2294 | ((inode)->i_flags & S_SYNC)) |
e462ec50 | 2295 | #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \ |
bbc1096a | 2296 | ((inode)->i_flags & (S_SYNC|S_DIRSYNC))) |
e462ec50 DH |
2297 | #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK) |
2298 | #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME) | |
2299 | #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION) | |
bbc1096a DH |
2300 | |
2301 | #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA) | |
2302 | #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND) | |
2303 | #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE) | |
e4e8b47a MK |
2304 | |
2305 | #ifdef CONFIG_FS_POSIX_ACL | |
e462ec50 | 2306 | #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL) |
e4e8b47a MK |
2307 | #else |
2308 | #define IS_POSIXACL(inode) 0 | |
2309 | #endif | |
bbc1096a DH |
2310 | |
2311 | #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD) | |
2312 | #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME) | |
e964fc77 AD |
2313 | |
2314 | #ifdef CONFIG_SWAP | |
bbc1096a | 2315 | #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE) |
e964fc77 AD |
2316 | #else |
2317 | #define IS_SWAPFILE(inode) ((void)(inode), 0U) | |
2318 | #endif | |
2319 | ||
bbc1096a DH |
2320 | #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE) |
2321 | #define IS_IMA(inode) ((inode)->i_flags & S_IMA) | |
2322 | #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT) | |
2323 | #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC) | |
fbbbad4b | 2324 | #define IS_DAX(inode) ((inode)->i_flags & S_DAX) |
2ee6a576 | 2325 | #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED) |
b886ee3e | 2326 | #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD) |
5585f2af | 2327 | #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY) |
bbc1096a | 2328 | |
787fb6bc MS |
2329 | #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \ |
2330 | (inode)->i_rdev == WHITEOUT_DEV) | |
2331 | ||
4609e1f1 | 2332 | static inline bool HAS_UNMAPPED_ID(struct mnt_idmap *idmap, |
ba73d987 | 2333 | struct inode *inode) |
0bd23d09 | 2334 | { |
e67fe633 CB |
2335 | return !vfsuid_valid(i_uid_into_vfsuid(idmap, inode)) || |
2336 | !vfsgid_valid(i_gid_into_vfsgid(idmap, inode)); | |
0bd23d09 EB |
2337 | } |
2338 | ||
c75b1d94 JA |
2339 | static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp) |
2340 | { | |
2341 | *kiocb = (struct kiocb) { | |
2342 | .ki_filp = filp, | |
164f4064 | 2343 | .ki_flags = filp->f_iocb_flags, |
20578bdf | 2344 | .ki_ioprio = get_current_ioprio(), |
c75b1d94 JA |
2345 | }; |
2346 | } | |
2347 | ||
5dcdc43e JX |
2348 | static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src, |
2349 | struct file *filp) | |
2350 | { | |
2351 | *kiocb = (struct kiocb) { | |
2352 | .ki_filp = filp, | |
2353 | .ki_flags = kiocb_src->ki_flags, | |
5dcdc43e JX |
2354 | .ki_ioprio = kiocb_src->ki_ioprio, |
2355 | .ki_pos = kiocb_src->ki_pos, | |
2356 | }; | |
2357 | } | |
2358 | ||
1c0eeaf5 | 2359 | /* |
250df6ed | 2360 | * Inode state bits. Protected by inode->i_lock |
1c0eeaf5 | 2361 | * |
1e9d6333 EB |
2362 | * Four bits determine the dirty state of the inode: I_DIRTY_SYNC, |
2363 | * I_DIRTY_DATASYNC, I_DIRTY_PAGES, and I_DIRTY_TIME. | |
1c0eeaf5 JE |
2364 | * |
2365 | * Four bits define the lifetime of an inode. Initially, inodes are I_NEW, | |
2366 | * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at | |
2367 | * various stages of removing an inode. | |
2368 | * | |
eaff8079 | 2369 | * Two bits are used for locking and completion notification, I_NEW and I_SYNC. |
1c0eeaf5 | 2370 | * |
e7ca2d41 | 2371 | * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on |
1e9d6333 EB |
2372 | * fdatasync() (unless I_DIRTY_DATASYNC is also set). |
2373 | * Timestamp updates are the usual cause. | |
2374 | * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of | |
000cb48e JK |
2375 | * these changes separately from I_DIRTY_SYNC so that we |
2376 | * don't have to write inode on fdatasync() when only | |
1e9d6333 | 2377 | * e.g. the timestamps have changed. |
1c0eeaf5 | 2378 | * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean. |
cbfecb92 | 2379 | * I_DIRTY_TIME The inode itself has dirty timestamps, and the |
1e9d6333 EB |
2380 | * lazytime mount option is enabled. We keep track of this |
2381 | * separately from I_DIRTY_SYNC in order to implement | |
2382 | * lazytime. This gets cleared if I_DIRTY_INODE | |
cbfecb92 LC |
2383 | * (I_DIRTY_SYNC and/or I_DIRTY_DATASYNC) gets set. But |
2384 | * I_DIRTY_TIME can still be set if I_DIRTY_SYNC is already | |
2385 | * in place because writeback might already be in progress | |
2386 | * and we don't want to lose the time update | |
eaff8079 CH |
2387 | * I_NEW Serves as both a mutex and completion notification. |
2388 | * New inodes set I_NEW. If two processes both create | |
2389 | * the same inode, one of them will release its inode and | |
2390 | * wait for I_NEW to be released before returning. | |
2391 | * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can | |
2392 | * also cause waiting on I_NEW, without I_NEW actually | |
2393 | * being set. find_inode() uses this to prevent returning | |
2394 | * nearly-dead inodes. | |
1c0eeaf5 JE |
2395 | * I_WILL_FREE Must be set when calling write_inode_now() if i_count |
2396 | * is zero. I_FREEING must be set when I_WILL_FREE is | |
2397 | * cleared. | |
2398 | * I_FREEING Set when inode is about to be freed but still has dirty | |
2399 | * pages or buffers attached or the inode itself is still | |
2400 | * dirty. | |
dbd5768f JK |
2401 | * I_CLEAR Added by clear_inode(). In this state the inode is |
2402 | * clean and can be destroyed. Inode keeps I_FREEING. | |
1c0eeaf5 JE |
2403 | * |
2404 | * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are | |
2405 | * prohibited for many purposes. iget() must wait for | |
2406 | * the inode to be completely released, then create it | |
2407 | * anew. Other functions will just ignore such inodes, | |
eaff8079 | 2408 | * if appropriate. I_NEW is used for waiting. |
1c0eeaf5 | 2409 | * |
169ebd90 JK |
2410 | * I_SYNC Writeback of inode is running. The bit is set during |
2411 | * data writeback, and cleared with a wakeup on the bit | |
2412 | * address once it is done. The bit is also used to pin | |
2413 | * the inode in memory for flusher thread. | |
1c0eeaf5 | 2414 | * |
bd5fe6c5 CH |
2415 | * I_REFERENCED Marks the inode as recently references on the LRU list. |
2416 | * | |
682aa8e1 TH |
2417 | * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to |
2418 | * synchronize competing switching instances and to tell | |
b93b0163 | 2419 | * wb stat updates to grab the i_pages lock. See |
a9519def | 2420 | * inode_switch_wbs_work_fn() for details. |
682aa8e1 | 2421 | * |
ad0af710 AG |
2422 | * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper |
2423 | * and work dirs among overlayfs mounts. | |
2424 | * | |
c2b6d621 AV |
2425 | * I_CREATING New object's inode in the middle of setting up. |
2426 | * | |
dae2f8ed IW |
2427 | * I_DONTCACHE Evict inode as soon as it is not used anymore. |
2428 | * | |
5afced3b JK |
2429 | * I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists. |
2430 | * Used to detect that mark_inode_dirty() should not move | |
2431 | * inode between dirty lists. | |
2432 | * | |
08276bda DH |
2433 | * I_PINNING_FSCACHE_WB Inode is pinning an fscache object for writeback. |
2434 | * | |
2a062983 ZC |
2435 | * I_LRU_ISOLATING Inode is pinned being isolated from LRU without holding |
2436 | * i_count. | |
2437 | * | |
1c0eeaf5 | 2438 | * Q: What is the difference between I_WILL_FREE and I_FREEING? |
2ed634c9 CB |
2439 | * |
2440 | * __I_{SYNC,NEW,LRU_ISOLATING} are used to derive unique addresses to wait | |
2441 | * upon. There's one free address left. | |
1c0eeaf5 | 2442 | */ |
2ed634c9 | 2443 | #define __I_NEW 0 |
eaff8079 | 2444 | #define I_NEW (1 << __I_NEW) |
2ed634c9 | 2445 | #define __I_SYNC 1 |
1c0eeaf5 | 2446 | #define I_SYNC (1 << __I_SYNC) |
2ed634c9 CB |
2447 | #define __I_LRU_ISOLATING 2 |
2448 | #define I_LRU_ISOLATING (1 << __I_LRU_ISOLATING) | |
2449 | ||
2450 | #define I_DIRTY_SYNC (1 << 3) | |
2451 | #define I_DIRTY_DATASYNC (1 << 4) | |
2452 | #define I_DIRTY_PAGES (1 << 5) | |
2453 | #define I_WILL_FREE (1 << 6) | |
2454 | #define I_FREEING (1 << 7) | |
2455 | #define I_CLEAR (1 << 8) | |
2456 | #define I_REFERENCED (1 << 9) | |
f4e0c30c | 2457 | #define I_LINKABLE (1 << 10) |
0ae45f63 | 2458 | #define I_DIRTY_TIME (1 << 11) |
2ed634c9 CB |
2459 | #define I_WB_SWITCH (1 << 12) |
2460 | #define I_OVL_INUSE (1 << 13) | |
2461 | #define I_CREATING (1 << 14) | |
2462 | #define I_DONTCACHE (1 << 15) | |
2463 | #define I_SYNC_QUEUED (1 << 16) | |
2464 | #define I_PINNING_NETFS_WB (1 << 17) | |
1da177e4 | 2465 | |
0e11f644 CH |
2466 | #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC) |
2467 | #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES) | |
0ae45f63 | 2468 | #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME) |
1da177e4 LT |
2469 | |
2470 | extern void __mark_inode_dirty(struct inode *, int); | |
2471 | static inline void mark_inode_dirty(struct inode *inode) | |
2472 | { | |
2473 | __mark_inode_dirty(inode, I_DIRTY); | |
2474 | } | |
2475 | ||
2476 | static inline void mark_inode_dirty_sync(struct inode *inode) | |
2477 | { | |
2478 | __mark_inode_dirty(inode, I_DIRTY_SYNC); | |
2479 | } | |
2480 | ||
ed296c6c EB |
2481 | /* |
2482 | * Returns true if the given inode itself only has dirty timestamps (its pages | |
2483 | * may still be dirty) and isn't currently being allocated or freed. | |
2484 | * Filesystems should call this if when writing an inode when lazytime is | |
2485 | * enabled, they want to opportunistically write the timestamps of other inodes | |
2486 | * located very nearby on-disk, e.g. in the same inode block. This returns true | |
2487 | * if the given inode is in need of such an opportunistic update. Requires | |
2488 | * i_lock, or at least later re-checking under i_lock. | |
2489 | */ | |
2490 | static inline bool inode_is_dirtytime_only(struct inode *inode) | |
2491 | { | |
2492 | return (inode->i_state & (I_DIRTY_TIME | I_NEW | | |
2493 | I_FREEING | I_WILL_FREE)) == I_DIRTY_TIME; | |
2494 | } | |
2495 | ||
7ada4db8 MS |
2496 | extern void inc_nlink(struct inode *inode); |
2497 | extern void drop_nlink(struct inode *inode); | |
2498 | extern void clear_nlink(struct inode *inode); | |
2499 | extern void set_nlink(struct inode *inode, unsigned int nlink); | |
d8c76e6f DH |
2500 | |
2501 | static inline void inode_inc_link_count(struct inode *inode) | |
2502 | { | |
2503 | inc_nlink(inode); | |
66d21730 AD |
2504 | mark_inode_dirty(inode); |
2505 | } | |
2506 | ||
9a53c3a7 DH |
2507 | static inline void inode_dec_link_count(struct inode *inode) |
2508 | { | |
2509 | drop_nlink(inode); | |
66d21730 AD |
2510 | mark_inode_dirty(inode); |
2511 | } | |
2512 | ||
c3b2da31 JB |
2513 | enum file_time_flags { |
2514 | S_ATIME = 1, | |
2515 | S_MTIME = 2, | |
2516 | S_CTIME = 4, | |
2517 | S_VERSION = 8, | |
2518 | }; | |
2519 | ||
c6718543 | 2520 | extern bool atime_needs_update(const struct path *, struct inode *); |
badcf2b7 | 2521 | extern void touch_atime(const struct path *); |
913e9928 | 2522 | int inode_update_time(struct inode *inode, int flags); |
e60feb44 | 2523 | |
1da177e4 LT |
2524 | static inline void file_accessed(struct file *file) |
2525 | { | |
2526 | if (!(file->f_flags & O_NOATIME)) | |
68ac1234 | 2527 | touch_atime(&file->f_path); |
1da177e4 LT |
2528 | } |
2529 | ||
e38f7f53 | 2530 | extern int file_modified(struct file *file); |
66fa3ced | 2531 | int kiocb_modified(struct kiocb *iocb); |
e38f7f53 | 2532 | |
c3765016 | 2533 | int sync_inode_metadata(struct inode *inode, int wait); |
1da177e4 | 2534 | |
1da177e4 LT |
2535 | struct file_system_type { |
2536 | const char *name; | |
2537 | int fs_flags; | |
bbc1096a DH |
2538 | #define FS_REQUIRES_DEV 1 |
2539 | #define FS_BINARY_MOUNTDATA 2 | |
2540 | #define FS_HAS_SUBTYPE 4 | |
0c55cfc4 | 2541 | #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */ |
0b3b094a | 2542 | #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */ |
a6435940 | 2543 | #define FS_ALLOW_IDMAP 32 /* FS has been updated to handle vfs idmappings. */ |
bbc1096a | 2544 | #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */ |
f3a09c92 | 2545 | int (*init_fs_context)(struct fs_context *); |
d7167b14 | 2546 | const struct fs_parameter_spec *parameters; |
c96e41e9 AV |
2547 | struct dentry *(*mount) (struct file_system_type *, int, |
2548 | const char *, void *); | |
1da177e4 LT |
2549 | void (*kill_sb) (struct super_block *); |
2550 | struct module *owner; | |
2551 | struct file_system_type * next; | |
a5166169 | 2552 | struct hlist_head fs_supers; |
d475fd42 | 2553 | |
cf516249 | 2554 | struct lock_class_key s_lock_key; |
897c6ff9 | 2555 | struct lock_class_key s_umount_key; |
51ee049e | 2556 | struct lock_class_key s_vfs_rename_key; |
5accdf82 | 2557 | struct lock_class_key s_writers_key[SB_FREEZE_LEVELS]; |
d475fd42 PZ |
2558 | |
2559 | struct lock_class_key i_lock_key; | |
2560 | struct lock_class_key i_mutex_key; | |
730633f0 | 2561 | struct lock_class_key invalidate_lock_key; |
14358e6d | 2562 | struct lock_class_key i_mutex_dir_key; |
1da177e4 LT |
2563 | }; |
2564 | ||
7f78e035 EB |
2565 | #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME) |
2566 | ||
152a0836 AV |
2567 | extern struct dentry *mount_bdev(struct file_system_type *fs_type, |
2568 | int flags, const char *dev_name, void *data, | |
2569 | int (*fill_super)(struct super_block *, void *, int)); | |
fc14f2fe AV |
2570 | extern struct dentry *mount_single(struct file_system_type *fs_type, |
2571 | int flags, void *data, | |
2572 | int (*fill_super)(struct super_block *, void *, int)); | |
3c26ff6e AV |
2573 | extern struct dentry *mount_nodev(struct file_system_type *fs_type, |
2574 | int flags, void *data, | |
2575 | int (*fill_super)(struct super_block *, void *, int)); | |
ea441d11 | 2576 | extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path); |
04b94071 | 2577 | void retire_super(struct super_block *sb); |
1da177e4 LT |
2578 | void generic_shutdown_super(struct super_block *sb); |
2579 | void kill_block_super(struct super_block *sb); | |
2580 | void kill_anon_super(struct super_block *sb); | |
2581 | void kill_litter_super(struct super_block *sb); | |
2582 | void deactivate_super(struct super_block *sb); | |
74dbbdd7 | 2583 | void deactivate_locked_super(struct super_block *sb); |
1da177e4 | 2584 | int set_anon_super(struct super_block *s, void *data); |
cb50b348 | 2585 | int set_anon_super_fc(struct super_block *s, struct fs_context *fc); |
0ee5dc67 AV |
2586 | int get_anon_bdev(dev_t *); |
2587 | void free_anon_bdev(dev_t); | |
cb50b348 AV |
2588 | struct super_block *sget_fc(struct fs_context *fc, |
2589 | int (*test)(struct super_block *, struct fs_context *), | |
2590 | int (*set)(struct super_block *, struct fs_context *)); | |
1da177e4 LT |
2591 | struct super_block *sget(struct file_system_type *type, |
2592 | int (*test)(struct super_block *,void *), | |
2593 | int (*set)(struct super_block *,void *), | |
9249e17f | 2594 | int flags, void *data); |
69881be3 | 2595 | struct super_block *sget_dev(struct fs_context *fc, dev_t dev); |
1da177e4 LT |
2596 | |
2597 | /* Alas, no aliases. Too much hassle with bringing module.h everywhere */ | |
8447d848 MG |
2598 | #define fops_get(fops) ({ \ |
2599 | const struct file_operations *_fops = (fops); \ | |
2600 | (((_fops) && try_module_get((_fops)->owner) ? (_fops) : NULL)); \ | |
2601 | }) | |
2602 | ||
2603 | #define fops_put(fops) ({ \ | |
2604 | const struct file_operations *_fops = (fops); \ | |
2605 | if (_fops) \ | |
2606 | module_put((_fops)->owner); \ | |
2607 | }) | |
2608 | ||
e84f9e57 AV |
2609 | /* |
2610 | * This one is to be used *ONLY* from ->open() instances. | |
2611 | * fops must be non-NULL, pinned down *and* module dependencies | |
2612 | * should be sufficient to pin the caller down as well. | |
2613 | */ | |
2614 | #define replace_fops(f, fops) \ | |
2615 | do { \ | |
2616 | struct file *__file = (f); \ | |
2617 | fops_put(__file->f_op); \ | |
2618 | BUG_ON(!(__file->f_op = (fops))); \ | |
2619 | } while(0) | |
1da177e4 LT |
2620 | |
2621 | extern int register_filesystem(struct file_system_type *); | |
2622 | extern int unregister_filesystem(struct file_system_type *); | |
f0bb5aaf | 2623 | extern int vfs_statfs(const struct path *, struct kstatfs *); |
c8b91acc AV |
2624 | extern int user_statfs(const char __user *, struct kstatfs *); |
2625 | extern int fd_statfs(int, struct kstatfs *); | |
880b9577 DW |
2626 | int freeze_super(struct super_block *super, enum freeze_holder who); |
2627 | int thaw_super(struct super_block *super, enum freeze_holder who); | |
fca39346 JK |
2628 | extern __printf(2, 3) |
2629 | int super_setup_bdi_name(struct super_block *sb, char *fmt, ...); | |
2630 | extern int super_setup_bdi(struct super_block *sb); | |
1da177e4 | 2631 | |
a4af51ce KO |
2632 | static inline void super_set_uuid(struct super_block *sb, const u8 *uuid, unsigned len) |
2633 | { | |
2634 | if (WARN_ON(len > sizeof(sb->s_uuid))) | |
2635 | len = sizeof(sb->s_uuid); | |
2636 | sb->s_uuid_len = len; | |
2637 | memcpy(&sb->s_uuid, uuid, len); | |
2638 | } | |
2639 | ||
ae8c5117 KO |
2640 | /* set sb sysfs name based on sb->s_bdev */ |
2641 | static inline void super_set_sysfs_name_bdev(struct super_block *sb) | |
2642 | { | |
2643 | snprintf(sb->s_sysfs_name, sizeof(sb->s_sysfs_name), "%pg", sb->s_bdev); | |
2644 | } | |
2645 | ||
2646 | /* set sb sysfs name based on sb->s_uuid */ | |
2647 | static inline void super_set_sysfs_name_uuid(struct super_block *sb) | |
2648 | { | |
2649 | WARN_ON(sb->s_uuid_len != sizeof(sb->s_uuid)); | |
2650 | snprintf(sb->s_sysfs_name, sizeof(sb->s_sysfs_name), "%pU", sb->s_uuid.b); | |
2651 | } | |
2652 | ||
2653 | /* set sb sysfs name based on sb->s_id */ | |
2654 | static inline void super_set_sysfs_name_id(struct super_block *sb) | |
2655 | { | |
2656 | strscpy(sb->s_sysfs_name, sb->s_id, sizeof(sb->s_sysfs_name)); | |
2657 | } | |
2658 | ||
2659 | /* try to use something standard before you use this */ | |
2660 | __printf(2, 3) | |
2661 | static inline void super_set_sysfs_name_generic(struct super_block *sb, const char *fmt, ...) | |
2662 | { | |
2663 | va_list args; | |
2664 | ||
2665 | va_start(args, fmt); | |
2666 | vsnprintf(sb->s_sysfs_name, sizeof(sb->s_sysfs_name), fmt, args); | |
2667 | va_end(args); | |
2668 | } | |
2669 | ||
ce3b0f8d AV |
2670 | extern int current_umask(void); |
2671 | ||
5a14696c BF |
2672 | extern void ihold(struct inode * inode); |
2673 | extern void iput(struct inode *); | |
541d4c79 JL |
2674 | int inode_update_timestamps(struct inode *inode, int flags); |
2675 | int generic_update_time(struct inode *, int); | |
5a14696c | 2676 | |
f87fd4c2 | 2677 | /* /sys/fs */ |
00d26666 | 2678 | extern struct kobject *fs_kobj; |
f87fd4c2 | 2679 | |
09cbfeaf | 2680 | #define MAX_RW_COUNT (INT_MAX & PAGE_MASK) |
bfcd17a6 | 2681 | |
1da177e4 | 2682 | /* fs/open.c */ |
adb5c247 | 2683 | struct audit_names; |
91a27b2a | 2684 | struct filename { |
adb5c247 JL |
2685 | const char *name; /* pointer to actual string */ |
2686 | const __user char *uptr; /* original userland pointer */ | |
03adc61e | 2687 | atomic_t refcnt; |
1c949843 | 2688 | struct audit_names *aname; |
fd2f7cb5 | 2689 | const char iname[]; |
91a27b2a | 2690 | }; |
f1fffbd4 | 2691 | static_assert(offsetof(struct filename, iname) % sizeof(long) == 0); |
1da177e4 | 2692 | |
4a00c673 | 2693 | static inline struct mnt_idmap *file_mnt_idmap(const struct file *file) |
256c8aed CB |
2694 | { |
2695 | return mnt_idmap(file->f_path.mnt); | |
2696 | } | |
2697 | ||
bb49e9e7 CB |
2698 | /** |
2699 | * is_idmapped_mnt - check whether a mount is mapped | |
2700 | * @mnt: the mount to check | |
2701 | * | |
256c8aed | 2702 | * If @mnt has an non @nop_mnt_idmap attached to it then @mnt is mapped. |
bb49e9e7 CB |
2703 | * |
2704 | * Return: true if mount is mapped, false if not. | |
2705 | */ | |
2706 | static inline bool is_idmapped_mnt(const struct vfsmount *mnt) | |
2707 | { | |
256c8aed | 2708 | return mnt_idmap(mnt) != &nop_mnt_idmap; |
bb49e9e7 CB |
2709 | } |
2710 | ||
7df818b2 | 2711 | extern long vfs_truncate(const struct path *, loff_t); |
abf08576 | 2712 | int do_truncate(struct mnt_idmap *, struct dentry *, loff_t start, |
643fe55a | 2713 | unsigned int time_attrs, struct file *filp); |
72c72bdf | 2714 | extern int vfs_fallocate(struct file *file, int mode, loff_t offset, |
3e63cbb1 | 2715 | loff_t len); |
8e8a1407 | 2716 | extern long do_sys_open(int dfd, const char __user *filename, int flags, |
a218d0fd | 2717 | umode_t mode); |
669abf4e | 2718 | extern struct file *file_open_name(struct filename *, int, umode_t); |
a218d0fd | 2719 | extern struct file *filp_open(const char *, int, umode_t); |
ffb37ca3 | 2720 | extern struct file *file_open_root(const struct path *, |
378c6520 | 2721 | const char *, int, umode_t); |
ffb37ca3 AV |
2722 | static inline struct file *file_open_root_mnt(struct vfsmount *mnt, |
2723 | const char *name, int flags, umode_t mode) | |
2724 | { | |
2725 | return file_open_root(&(struct path){.mnt = mnt, .dentry = mnt->mnt_root}, | |
2726 | name, flags, mode); | |
2727 | } | |
62d53c4a AG |
2728 | struct file *dentry_open(const struct path *path, int flags, |
2729 | const struct cred *creds); | |
2730 | struct file *dentry_create(const struct path *path, int flags, umode_t mode, | |
2731 | const struct cred *cred); | |
def3ae83 | 2732 | struct path *backing_file_user_path(struct file *f); |
62d53c4a AG |
2733 | |
2734 | /* | |
08582d67 AG |
2735 | * When mmapping a file on a stackable filesystem (e.g., overlayfs), the file |
2736 | * stored in ->vm_file is a backing file whose f_inode is on the underlying | |
3efdc78f AV |
2737 | * filesystem. When the mapped file path and inode number are displayed to |
2738 | * user (e.g. via /proc/<pid>/maps), these helpers should be used to get the | |
2739 | * path and inode number to display to the user, which is the path of the fd | |
2740 | * that user has requested to map and the inode number that would be returned | |
2741 | * by fstat() on that same fd. | |
08582d67 | 2742 | */ |
3efdc78f | 2743 | /* Get the path to display in /proc/<pid>/maps */ |
08582d67 | 2744 | static inline const struct path *file_user_path(struct file *f) |
62d53c4a AG |
2745 | { |
2746 | if (unlikely(f->f_mode & FMODE_BACKING)) | |
def3ae83 | 2747 | return backing_file_user_path(f); |
62d53c4a AG |
2748 | return &f->f_path; |
2749 | } | |
3efdc78f AV |
2750 | /* Get the inode whose inode number to display in /proc/<pid>/maps */ |
2751 | static inline const struct inode *file_user_inode(struct file *f) | |
2752 | { | |
2753 | if (unlikely(f->f_mode & FMODE_BACKING)) | |
2754 | return d_inode(backing_file_user_path(f)->dentry); | |
2755 | return file_inode(f); | |
2756 | } | |
62d53c4a | 2757 | |
19f391eb AV |
2758 | static inline struct file *file_clone_open(struct file *file) |
2759 | { | |
2760 | return dentry_open(&file->f_path, file->f_flags, file->f_cred); | |
2761 | } | |
1da177e4 | 2762 | extern int filp_close(struct file *, fl_owner_t id); |
91a27b2a | 2763 | |
dff60734 | 2764 | extern struct filename *getname_flags(const char __user *, int); |
8228e2c3 | 2765 | extern struct filename *getname_uflags(const char __user *, int); |
91a27b2a | 2766 | extern struct filename *getname(const char __user *); |
c4ad8f98 | 2767 | extern struct filename *getname_kernel(const char *); |
55422d0b | 2768 | extern void putname(struct filename *name); |
91a27b2a | 2769 | |
30d90494 | 2770 | extern int finish_open(struct file *file, struct dentry *dentry, |
be12af3e | 2771 | int (*open)(struct inode *, struct file *)); |
e45198a6 | 2772 | extern int finish_no_open(struct file *file, struct dentry *dentry); |
1da177e4 | 2773 | |
9751b338 MS |
2774 | /* Helper for the simple case when original dentry is used */ |
2775 | static inline int finish_open_simple(struct file *file, int error) | |
2776 | { | |
2777 | if (error) | |
2778 | return error; | |
2779 | ||
2780 | return finish_open(file, file->f_path.dentry, NULL); | |
2781 | } | |
2782 | ||
1da177e4 LT |
2783 | /* fs/dcache.c */ |
2784 | extern void __init vfs_caches_init_early(void); | |
4248b0da | 2785 | extern void __init vfs_caches_init(void); |
1da177e4 | 2786 | |
b86c089b CL |
2787 | extern struct kmem_cache *names_cachep; |
2788 | ||
a608ca21 | 2789 | #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL) |
3446a8aa | 2790 | #define __putname(name) kmem_cache_free(names_cachep, (void *)(name)) |
1da177e4 | 2791 | |
a212b105 | 2792 | extern struct super_block *blockdev_superblock; |
a212b105 TH |
2793 | static inline bool sb_is_blkdev_sb(struct super_block *sb) |
2794 | { | |
dd0dca22 | 2795 | return IS_ENABLED(CONFIG_BLOCK) && sb == blockdev_superblock; |
47e4491b | 2796 | } |
5c0d6b60 | 2797 | |
b818f09e | 2798 | void emergency_thaw_all(void); |
60b0680f | 2799 | extern int sync_filesystem(struct super_block *); |
9361401e | 2800 | extern const struct file_operations def_blk_fops; |
4b6f5d20 | 2801 | extern const struct file_operations def_chr_fops; |
1da177e4 LT |
2802 | |
2803 | /* fs/char_dev.c */ | |
8a932f73 | 2804 | #define CHRDEV_MAJOR_MAX 512 |
49db08c3 LW |
2805 | /* Marks the bottom of the first segment of free char majors */ |
2806 | #define CHRDEV_MAJOR_DYN_END 234 | |
a5d31a3f LG |
2807 | /* Marks the top and bottom of the second segment of free char majors */ |
2808 | #define CHRDEV_MAJOR_DYN_EXT_START 511 | |
2809 | #define CHRDEV_MAJOR_DYN_EXT_END 384 | |
2810 | ||
1da177e4 LT |
2811 | extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *); |
2812 | extern int register_chrdev_region(dev_t, unsigned, const char *); | |
1905b1bf TH |
2813 | extern int __register_chrdev(unsigned int major, unsigned int baseminor, |
2814 | unsigned int count, const char *name, | |
2815 | const struct file_operations *fops); | |
2816 | extern void __unregister_chrdev(unsigned int major, unsigned int baseminor, | |
2817 | unsigned int count, const char *name); | |
1da177e4 | 2818 | extern void unregister_chrdev_region(dev_t, unsigned); |
68eef3b4 | 2819 | extern void chrdev_show(struct seq_file *,off_t); |
1da177e4 | 2820 | |
1905b1bf TH |
2821 | static inline int register_chrdev(unsigned int major, const char *name, |
2822 | const struct file_operations *fops) | |
2823 | { | |
2824 | return __register_chrdev(major, 0, 256, name, fops); | |
2825 | } | |
2826 | ||
2827 | static inline void unregister_chrdev(unsigned int major, const char *name) | |
2828 | { | |
2829 | __unregister_chrdev(major, 0, 256, name); | |
2830 | } | |
2831 | ||
1da177e4 LT |
2832 | extern void init_special_inode(struct inode *, umode_t, dev_t); |
2833 | ||
2834 | /* Invalid inode operations -- fs/bad_inode.c */ | |
2835 | extern void make_bad_inode(struct inode *); | |
0e3ef1fe | 2836 | extern bool is_bad_inode(struct inode *); |
1da177e4 | 2837 | |
a823e458 JL |
2838 | extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart, |
2839 | loff_t lend); | |
5660e13d JL |
2840 | extern int __must_check file_check_and_advance_wb_err(struct file *file); |
2841 | extern int __must_check file_write_and_wait_range(struct file *file, | |
2842 | loff_t start, loff_t end); | |
2843 | ||
a823e458 JL |
2844 | static inline int file_write_and_wait(struct file *file) |
2845 | { | |
2846 | return file_write_and_wait_range(file, 0, LLONG_MAX); | |
2847 | } | |
2848 | ||
8018ab05 CH |
2849 | extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end, |
2850 | int datasync); | |
2851 | extern int vfs_fsync(struct file *file, int datasync); | |
e2592217 | 2852 | |
22f96b38 JA |
2853 | extern int sync_file_range(struct file *file, loff_t offset, loff_t nbytes, |
2854 | unsigned int flags); | |
2855 | ||
91b94c5d AV |
2856 | static inline bool iocb_is_dsync(const struct kiocb *iocb) |
2857 | { | |
2858 | return (iocb->ki_flags & IOCB_DSYNC) || | |
2859 | IS_SYNC(iocb->ki_filp->f_mapping->host); | |
2860 | } | |
2861 | ||
e2592217 CH |
2862 | /* |
2863 | * Sync the bytes written if this was a synchronous write. Expect ki_pos | |
2864 | * to already be updated for the write, and will return either the amount | |
2865 | * of bytes passed in, or an error if syncing the file failed. | |
2866 | */ | |
2867 | static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count) | |
2868 | { | |
91b94c5d | 2869 | if (iocb_is_dsync(iocb)) { |
e2592217 CH |
2870 | int ret = vfs_fsync_range(iocb->ki_filp, |
2871 | iocb->ki_pos - count, iocb->ki_pos - 1, | |
2872 | (iocb->ki_flags & IOCB_SYNC) ? 0 : 1); | |
2873 | if (ret) | |
2874 | return ret; | |
2875 | } | |
2876 | ||
2877 | return count; | |
d311d79d | 2878 | } |
e2592217 | 2879 | |
1da177e4 LT |
2880 | extern void emergency_sync(void); |
2881 | extern void emergency_remount(void); | |
30460e1e | 2882 | |
9361401e | 2883 | #ifdef CONFIG_BLOCK |
30460e1e CM |
2884 | extern int bmap(struct inode *inode, sector_t *block); |
2885 | #else | |
2886 | static inline int bmap(struct inode *inode, sector_t *block) | |
2887 | { | |
2888 | return -EINVAL; | |
2889 | } | |
9361401e | 2890 | #endif |
30460e1e | 2891 | |
abf08576 | 2892 | int notify_change(struct mnt_idmap *, struct dentry *, |
2f221d6f | 2893 | struct iattr *, struct inode **); |
4609e1f1 CB |
2894 | int inode_permission(struct mnt_idmap *, struct inode *, int); |
2895 | int generic_permission(struct mnt_idmap *, struct inode *, int); | |
02f92b38 CB |
2896 | static inline int file_permission(struct file *file, int mask) |
2897 | { | |
4609e1f1 | 2898 | return inode_permission(file_mnt_idmap(file), |
47291baa | 2899 | file_inode(file), mask); |
02f92b38 CB |
2900 | } |
2901 | static inline int path_permission(const struct path *path, int mask) | |
2902 | { | |
4609e1f1 | 2903 | return inode_permission(mnt_idmap(path->mnt), |
47291baa | 2904 | d_inode(path->dentry), mask); |
02f92b38 | 2905 | } |
9452e93e | 2906 | int __check_sticky(struct mnt_idmap *idmap, struct inode *dir, |
ba73d987 | 2907 | struct inode *inode); |
1da177e4 | 2908 | |
f696a365 MS |
2909 | static inline bool execute_ok(struct inode *inode) |
2910 | { | |
2911 | return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode); | |
2912 | } | |
2913 | ||
6e3e2c43 AV |
2914 | static inline bool inode_wrong_type(const struct inode *inode, umode_t mode) |
2915 | { | |
2916 | return (inode->i_mode ^ mode) & S_IFMT; | |
2917 | } | |
2918 | ||
f6c05b9e AG |
2919 | /** |
2920 | * file_start_write - get write access to a superblock for regular file io | |
2921 | * @file: the file we want to write to | |
2922 | * | |
2923 | * This is a variant of sb_start_write() which is a noop on non-regualr file. | |
2924 | * Should be matched with a call to file_end_write(). | |
2925 | */ | |
8d71db4f AV |
2926 | static inline void file_start_write(struct file *file) |
2927 | { | |
2928 | if (!S_ISREG(file_inode(file)->i_mode)) | |
2929 | return; | |
8a3c84b6 | 2930 | sb_start_write(file_inode(file)->i_sb); |
8d71db4f AV |
2931 | } |
2932 | ||
5ae98f15 JK |
2933 | static inline bool file_start_write_trylock(struct file *file) |
2934 | { | |
2935 | if (!S_ISREG(file_inode(file)->i_mode)) | |
2936 | return true; | |
8a3c84b6 | 2937 | return sb_start_write_trylock(file_inode(file)->i_sb); |
5ae98f15 JK |
2938 | } |
2939 | ||
f6c05b9e AG |
2940 | /** |
2941 | * file_end_write - drop write access to a superblock of a regular file | |
2942 | * @file: the file we wrote to | |
2943 | * | |
2944 | * Should be matched with a call to file_start_write(). | |
2945 | */ | |
8d71db4f AV |
2946 | static inline void file_end_write(struct file *file) |
2947 | { | |
2948 | if (!S_ISREG(file_inode(file)->i_mode)) | |
2949 | return; | |
f6c05b9e | 2950 | sb_end_write(file_inode(file)->i_sb); |
8d71db4f | 2951 | } |
bfe219d3 | 2952 | |
ed0360bb AG |
2953 | /** |
2954 | * kiocb_start_write - get write access to a superblock for async file io | |
2955 | * @iocb: the io context we want to submit the write with | |
2956 | * | |
2957 | * This is a variant of sb_start_write() for async io submission. | |
2958 | * Should be matched with a call to kiocb_end_write(). | |
2959 | */ | |
2960 | static inline void kiocb_start_write(struct kiocb *iocb) | |
2961 | { | |
2962 | struct inode *inode = file_inode(iocb->ki_filp); | |
2963 | ||
2964 | sb_start_write(inode->i_sb); | |
2965 | /* | |
2966 | * Fool lockdep by telling it the lock got released so that it | |
2967 | * doesn't complain about the held lock when we return to userspace. | |
2968 | */ | |
2969 | __sb_writers_release(inode->i_sb, SB_FREEZE_WRITE); | |
2970 | } | |
2971 | ||
2972 | /** | |
2973 | * kiocb_end_write - drop write access to a superblock after async file io | |
2974 | * @iocb: the io context we sumbitted the write with | |
2975 | * | |
2976 | * Should be matched with a call to kiocb_start_write(). | |
2977 | */ | |
2978 | static inline void kiocb_end_write(struct kiocb *iocb) | |
2979 | { | |
2980 | struct inode *inode = file_inode(iocb->ki_filp); | |
2981 | ||
2982 | /* | |
2983 | * Tell lockdep we inherited freeze protection from submission thread. | |
2984 | */ | |
2985 | __sb_writers_acquired(inode->i_sb, SB_FREEZE_WRITE); | |
2986 | sb_end_write(inode->i_sb); | |
8d71db4f | 2987 | } |
bfe219d3 | 2988 | |
07b8ce1e | 2989 | /* |
592ca09b DH |
2990 | * This is used for regular files where some users -- especially the |
2991 | * currently executed binary in a process, previously handled via | |
2992 | * VM_DENYWRITE -- cannot handle concurrent write (and maybe mmap | |
2993 | * read-write shared) accesses. | |
2994 | * | |
07b8ce1e AV |
2995 | * get_write_access() gets write permission for a file. |
2996 | * put_write_access() releases this write permission. | |
592ca09b DH |
2997 | * deny_write_access() denies write access to a file. |
2998 | * allow_write_access() re-enables write access to a file. | |
2999 | * | |
3000 | * The i_writecount field of an inode can have the following values: | |
3001 | * 0: no write access, no denied write access | |
3002 | * < 0: (-i_writecount) users that denied write access to the file. | |
3003 | * > 0: (i_writecount) users that have write access to the file. | |
07b8ce1e AV |
3004 | * |
3005 | * Normally we operate on that counter with atomic_{inc,dec} and it's safe | |
3006 | * except for the cases where we don't hold i_writecount yet. Then we need to | |
3007 | * use {get,deny}_write_access() - these functions check the sign and refuse | |
3008 | * to do the change if sign is wrong. | |
3009 | */ | |
3010 | static inline int get_write_access(struct inode *inode) | |
3011 | { | |
3012 | return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY; | |
3013 | } | |
3014 | static inline int deny_write_access(struct file *file) | |
3015 | { | |
496ad9aa | 3016 | struct inode *inode = file_inode(file); |
07b8ce1e AV |
3017 | return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY; |
3018 | } | |
1da177e4 LT |
3019 | static inline void put_write_access(struct inode * inode) |
3020 | { | |
3021 | atomic_dec(&inode->i_writecount); | |
3022 | } | |
3023 | static inline void allow_write_access(struct file *file) | |
3024 | { | |
3025 | if (file) | |
496ad9aa | 3026 | atomic_inc(&file_inode(file)->i_writecount); |
1da177e4 | 3027 | } |
f1fe29b4 DH |
3028 | static inline bool inode_is_open_for_write(const struct inode *inode) |
3029 | { | |
3030 | return atomic_read(&inode->i_writecount) > 0; | |
3031 | } | |
3032 | ||
387e3746 | 3033 | #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING) |
a5c96ebf MZ |
3034 | static inline void i_readcount_dec(struct inode *inode) |
3035 | { | |
45e0d4b9 | 3036 | BUG_ON(atomic_dec_return(&inode->i_readcount) < 0); |
a5c96ebf MZ |
3037 | } |
3038 | static inline void i_readcount_inc(struct inode *inode) | |
3039 | { | |
3040 | atomic_inc(&inode->i_readcount); | |
3041 | } | |
3042 | #else | |
3043 | static inline void i_readcount_dec(struct inode *inode) | |
3044 | { | |
3045 | return; | |
3046 | } | |
3047 | static inline void i_readcount_inc(struct inode *inode) | |
3048 | { | |
3049 | return; | |
3050 | } | |
3051 | #endif | |
ed8cae8b | 3052 | extern int do_pipe_flags(int *, int); |
1da177e4 | 3053 | |
bdd1d2d3 | 3054 | extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *); |
61a707c5 | 3055 | ssize_t __kernel_read(struct file *file, void *buf, size_t count, loff_t *pos); |
e13ec939 | 3056 | extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *); |
73e18f7c | 3057 | extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *); |
1da177e4 LT |
3058 | extern struct file * open_exec(const char *); |
3059 | ||
3060 | /* fs/dcache.c -- generic fs support functions */ | |
a6e5787f | 3061 | extern bool is_subdir(struct dentry *, struct dentry *); |
640eb7e7 | 3062 | extern bool path_is_under(const struct path *, const struct path *); |
1da177e4 | 3063 | |
9bf39ab2 MS |
3064 | extern char *file_path(struct file *, char *, int); |
3065 | ||
42c3732f CL |
3066 | /** |
3067 | * is_dot_dotdot - returns true only if @name is "." or ".." | |
3068 | * @name: file name to check | |
3069 | * @len: length of file name, in bytes | |
3070 | */ | |
3071 | static inline bool is_dot_dotdot(const char *name, size_t len) | |
3072 | { | |
3073 | return len && unlikely(name[0] == '.') && | |
3074 | (len == 1 || (len == 2 && name[1] == '.')); | |
3075 | } | |
3076 | ||
1da177e4 LT |
3077 | #include <linux/err.h> |
3078 | ||
3079 | /* needed for stackable file system support */ | |
965c8e59 | 3080 | extern loff_t default_llseek(struct file *file, loff_t offset, int whence); |
1da177e4 | 3081 | |
965c8e59 | 3082 | extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence); |
1da177e4 | 3083 | |
54e34621 | 3084 | extern int inode_init_always(struct super_block *, struct inode *); |
1da177e4 | 3085 | extern void inode_init_once(struct inode *); |
2aa15890 | 3086 | extern void address_space_init_once(struct address_space *mapping); |
1da177e4 LT |
3087 | extern struct inode * igrab(struct inode *); |
3088 | extern ino_t iunique(struct super_block *, ino_t); | |
3089 | extern int inode_needs_sync(struct inode *inode); | |
45321ac5 | 3090 | extern int generic_delete_inode(struct inode *inode); |
9bcb4b73 AV |
3091 | static inline int generic_drop_inode(struct inode *inode) |
3092 | { | |
88149082 | 3093 | return !inode->i_nlink || inode_unhashed(inode); |
9bcb4b73 | 3094 | } |
2c567af4 | 3095 | extern void d_mark_dontcache(struct inode *inode); |
1da177e4 | 3096 | |
88bd5121 AA |
3097 | extern struct inode *ilookup5_nowait(struct super_block *sb, |
3098 | unsigned long hashval, int (*test)(struct inode *, void *), | |
3099 | void *data); | |
1da177e4 LT |
3100 | extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval, |
3101 | int (*test)(struct inode *, void *), void *data); | |
3102 | extern struct inode *ilookup(struct super_block *sb, unsigned long ino); | |
3103 | ||
80ea09a0 MS |
3104 | extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval, |
3105 | int (*test)(struct inode *, void *), | |
3106 | int (*set)(struct inode *, void *), | |
3107 | void *data); | |
7180f8d9 MG |
3108 | struct inode *iget5_locked(struct super_block *, unsigned long, |
3109 | int (*test)(struct inode *, void *), | |
3110 | int (*set)(struct inode *, void *), void *); | |
3111 | struct inode *iget5_locked_rcu(struct super_block *, unsigned long, | |
3112 | int (*test)(struct inode *, void *), | |
3113 | int (*set)(struct inode *, void *), void *); | |
1da177e4 | 3114 | extern struct inode * iget_locked(struct super_block *, unsigned long); |
fe032c42 TT |
3115 | extern struct inode *find_inode_nowait(struct super_block *, |
3116 | unsigned long, | |
3117 | int (*match)(struct inode *, | |
3118 | unsigned long, void *), | |
3119 | void *data); | |
3f19b2ab DH |
3120 | extern struct inode *find_inode_rcu(struct super_block *, unsigned long, |
3121 | int (*)(struct inode *, void *), void *); | |
3122 | extern struct inode *find_inode_by_ino_rcu(struct super_block *, unsigned long); | |
261bca86 AV |
3123 | extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *); |
3124 | extern int insert_inode_locked(struct inode *); | |
e096d0c7 JB |
3125 | #ifdef CONFIG_DEBUG_LOCK_ALLOC |
3126 | extern void lockdep_annotate_inode_mutex_key(struct inode *inode); | |
3127 | #else | |
3128 | static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { }; | |
3129 | #endif | |
1da177e4 | 3130 | extern void unlock_new_inode(struct inode *); |
c2b6d621 | 3131 | extern void discard_new_inode(struct inode *); |
85fe4025 | 3132 | extern unsigned int get_next_ino(void); |
799ea9e9 | 3133 | extern void evict_inodes(struct super_block *sb); |
3e9d80a8 | 3134 | void dump_mapping(const struct address_space *); |
1da177e4 | 3135 | |
e809d5f0 | 3136 | /* |
b6334e2c | 3137 | * Userspace may rely on the inode number being non-zero. For example, glibc |
e809d5f0 CD |
3138 | * simply ignores files with zero i_ino in unlink() and other places. |
3139 | * | |
3140 | * As an additional complication, if userspace was compiled with | |
3141 | * _FILE_OFFSET_BITS=32 on a 64-bit kernel we'll only end up reading out the | |
3142 | * lower 32 bits, so we need to check that those aren't zero explicitly. With | |
3143 | * _FILE_OFFSET_BITS=64, this may cause some harmless false-negatives, but | |
3144 | * better safe than sorry. | |
3145 | */ | |
3146 | static inline bool is_zero_ino(ino_t ino) | |
3147 | { | |
3148 | return (u32)ino == 0; | |
3149 | } | |
3150 | ||
1da177e4 | 3151 | extern void __iget(struct inode * inode); |
b46980fe | 3152 | extern void iget_failed(struct inode *); |
dbd5768f | 3153 | extern void clear_inode(struct inode *); |
2e00c97e | 3154 | extern void __destroy_inode(struct inode *); |
a209dfc7 ED |
3155 | extern struct inode *new_inode_pseudo(struct super_block *sb); |
3156 | extern struct inode *new_inode(struct super_block *sb); | |
ff0c7d15 | 3157 | extern void free_inode_nonrcu(struct inode *inode); |
9452e93e | 3158 | extern int setattr_should_drop_suidgid(struct mnt_idmap *, struct inode *); |
66a67c86 | 3159 | extern int file_remove_privs_flags(struct file *file, unsigned int flags); |
5fa8e0a1 | 3160 | extern int file_remove_privs(struct file *); |
4f704d9a CB |
3161 | int setattr_should_drop_sgid(struct mnt_idmap *idmap, |
3162 | const struct inode *inode); | |
1da177e4 | 3163 | |
8b9f3ac5 MS |
3164 | /* |
3165 | * This must be used for allocating filesystems specific inodes to set | |
3166 | * up the inode reclaim context correctly. | |
3167 | */ | |
a5674119 | 3168 | #define alloc_inode_sb(_sb, _cache, _gfp) kmem_cache_alloc_lru(_cache, &_sb->s_inode_lru, _gfp) |
8b9f3ac5 | 3169 | |
1da177e4 | 3170 | extern void __insert_inode_hash(struct inode *, unsigned long hashval); |
646ec461 CH |
3171 | static inline void insert_inode_hash(struct inode *inode) |
3172 | { | |
1da177e4 LT |
3173 | __insert_inode_hash(inode, inode->i_ino); |
3174 | } | |
f2ee7abf ED |
3175 | |
3176 | extern void __remove_inode_hash(struct inode *); | |
3177 | static inline void remove_inode_hash(struct inode *inode) | |
3178 | { | |
cbedaac6 | 3179 | if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash)) |
f2ee7abf ED |
3180 | __remove_inode_hash(inode); |
3181 | } | |
3182 | ||
646ec461 | 3183 | extern void inode_sb_list_add(struct inode *inode); |
51b8c1fe | 3184 | extern void inode_add_lru(struct inode *inode); |
1da177e4 | 3185 | |
1da177e4 LT |
3186 | extern int sb_set_blocksize(struct super_block *, int); |
3187 | extern int sb_min_blocksize(struct super_block *, int); | |
3188 | ||
3189 | extern int generic_file_mmap(struct file *, struct vm_area_struct *); | |
3190 | extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *); | |
3309dd04 | 3191 | extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *); |
f6f7a25a | 3192 | int generic_write_checks_count(struct kiocb *iocb, loff_t *count); |
02e83f46 DW |
3193 | extern int generic_write_check_limits(struct file *file, loff_t pos, |
3194 | loff_t *count); | |
a3171351 | 3195 | extern int generic_file_rw_checks(struct file *file_in, struct file *file_out); |
87fa0f3e CH |
3196 | ssize_t filemap_read(struct kiocb *iocb, struct iov_iter *to, |
3197 | ssize_t already_read); | |
ed978a81 | 3198 | extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *); |
8174202b | 3199 | extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *); |
8174202b | 3200 | extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *); |
1af5bb49 | 3201 | extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *); |
800ba295 | 3202 | ssize_t generic_perform_write(struct kiocb *, struct iov_iter *); |
44fff0fa CH |
3203 | ssize_t direct_write_fallback(struct kiocb *iocb, struct iov_iter *iter, |
3204 | ssize_t direct_written, ssize_t buffered_written); | |
88e6faef | 3205 | |
18e9710e | 3206 | ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos, |
ddef7ed2 | 3207 | rwf_t flags); |
abbb6589 | 3208 | ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos, |
ddef7ed2 | 3209 | rwf_t flags); |
5dcdc43e JX |
3210 | ssize_t vfs_iocb_iter_read(struct file *file, struct kiocb *iocb, |
3211 | struct iov_iter *iter); | |
3212 | ssize_t vfs_iocb_iter_write(struct file *file, struct kiocb *iocb, | |
3213 | struct iov_iter *iter); | |
dbe4e192 | 3214 | |
88e6faef | 3215 | /* fs/splice.c */ |
07073eb0 DH |
3216 | ssize_t filemap_splice_read(struct file *in, loff_t *ppos, |
3217 | struct pipe_inode_info *pipe, | |
3218 | size_t len, unsigned int flags); | |
69df79a4 DH |
3219 | ssize_t copy_splice_read(struct file *in, loff_t *ppos, |
3220 | struct pipe_inode_info *pipe, | |
3221 | size_t len, unsigned int flags); | |
8d020765 | 3222 | extern ssize_t iter_file_splice_write(struct pipe_inode_info *, |
cbb7e577 | 3223 | struct file *, loff_t *, size_t, unsigned int); |
1c118596 | 3224 | |
88e6faef | 3225 | |
1da177e4 LT |
3226 | extern void |
3227 | file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping); | |
965c8e59 | 3228 | extern loff_t noop_llseek(struct file *file, loff_t offset, int whence); |
868941b1 | 3229 | #define no_llseek NULL |
46a1c2c7 | 3230 | extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize); |
965c8e59 | 3231 | extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence); |
5760495a | 3232 | extern loff_t generic_file_llseek_size(struct file *file, loff_t offset, |
965c8e59 | 3233 | int whence, loff_t maxsize, loff_t eof); |
d688d65a CB |
3234 | loff_t generic_llseek_cookie(struct file *file, loff_t offset, int whence, |
3235 | u64 *cookie); | |
1bf9d14d AV |
3236 | extern loff_t fixed_size_llseek(struct file *file, loff_t offset, |
3237 | int whence, loff_t size); | |
b25472f9 AV |
3238 | extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t); |
3239 | extern loff_t no_seek_end_llseek(struct file *, loff_t, int); | |
87112933 | 3240 | int rw_verify_area(int, struct file *, const loff_t *, size_t); |
1da177e4 LT |
3241 | extern int generic_file_open(struct inode * inode, struct file * filp); |
3242 | extern int nonseekable_open(struct inode * inode, struct file * filp); | |
10dce8af | 3243 | extern int stream_open(struct inode * inode, struct file * filp); |
1da177e4 | 3244 | |
9361401e | 3245 | #ifdef CONFIG_BLOCK |
8a4c1e42 | 3246 | typedef void (dio_submit_t)(struct bio *bio, struct inode *inode, |
facd07b0 | 3247 | loff_t file_offset); |
1da177e4 LT |
3248 | |
3249 | enum { | |
1e431f5c CH |
3250 | /* need locking between buffered and direct access */ |
3251 | DIO_LOCKING = 0x01, | |
3252 | ||
3253 | /* filesystem does not support filling holes */ | |
3254 | DIO_SKIP_HOLES = 0x02, | |
1da177e4 LT |
3255 | }; |
3256 | ||
17f8c842 OS |
3257 | ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode, |
3258 | struct block_device *bdev, struct iov_iter *iter, | |
c8b8e32d | 3259 | get_block_t get_block, |
0aaf08de | 3260 | dio_iodone_t end_io, |
17f8c842 | 3261 | int flags); |
7bb46a67 | 3262 | |
17f8c842 OS |
3263 | static inline ssize_t blockdev_direct_IO(struct kiocb *iocb, |
3264 | struct inode *inode, | |
c8b8e32d | 3265 | struct iov_iter *iter, |
17f8c842 | 3266 | get_block_t get_block) |
1da177e4 | 3267 | { |
17f8c842 | 3268 | return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter, |
0aaf08de | 3269 | get_block, NULL, DIO_LOCKING | DIO_SKIP_HOLES); |
1da177e4 | 3270 | } |
9361401e | 3271 | #endif |
1da177e4 | 3272 | |
1c48d441 | 3273 | bool inode_dio_finished(const struct inode *inode); |
1d59d61f | 3274 | void inode_dio_wait(struct inode *inode); |
1c48d441 | 3275 | void inode_dio_wait_interruptible(struct inode *inode); |
fe0f07d0 | 3276 | |
39015399 | 3277 | /** |
fe0f07d0 JA |
3278 | * inode_dio_begin - signal start of a direct I/O requests |
3279 | * @inode: inode the direct I/O happens on | |
3280 | * | |
3281 | * This is called once we've finished processing a direct I/O request, | |
3282 | * and is used to wake up callers waiting for direct I/O to be quiesced. | |
3283 | */ | |
3284 | static inline void inode_dio_begin(struct inode *inode) | |
3285 | { | |
3286 | atomic_inc(&inode->i_dio_count); | |
3287 | } | |
3288 | ||
39015399 | 3289 | /** |
fe0f07d0 JA |
3290 | * inode_dio_end - signal finish of a direct I/O requests |
3291 | * @inode: inode the direct I/O happens on | |
3292 | * | |
3293 | * This is called once we've finished processing a direct I/O request, | |
3294 | * and is used to wake up callers waiting for direct I/O to be quiesced. | |
3295 | */ | |
3296 | static inline void inode_dio_end(struct inode *inode) | |
3297 | { | |
3298 | if (atomic_dec_and_test(&inode->i_dio_count)) | |
1c48d441 | 3299 | wake_up_var(&inode->i_dio_count); |
fe0f07d0 | 3300 | } |
1d59d61f | 3301 | |
5f16f322 TT |
3302 | extern void inode_set_flags(struct inode *inode, unsigned int flags, |
3303 | unsigned int mask); | |
3304 | ||
4b6f5d20 | 3305 | extern const struct file_operations generic_ro_fops; |
1da177e4 LT |
3306 | |
3307 | #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m)) | |
3308 | ||
5d826c84 | 3309 | extern int readlink_copy(char __user *, int, const char *); |
1da177e4 | 3310 | extern int page_readlink(struct dentry *, char __user *, int); |
fceef393 AV |
3311 | extern const char *page_get_link(struct dentry *, struct inode *, |
3312 | struct delayed_call *); | |
3313 | extern void page_put_link(void *); | |
1da177e4 | 3314 | extern int page_symlink(struct inode *inode, const char *symname, int len); |
c5ef1c42 | 3315 | extern const struct inode_operations page_symlink_inode_operations; |
fceef393 | 3316 | extern void kfree_link(void *); |
0d72b928 | 3317 | void generic_fillattr(struct mnt_idmap *, u32, struct inode *, struct kstat *); |
4f911138 | 3318 | void generic_fill_statx_attr(struct inode *inode, struct kstat *stat); |
0f9ca80f PS |
3319 | void generic_fill_statx_atomic_writes(struct kstat *stat, |
3320 | unsigned int unit_min, | |
3321 | unsigned int unit_max); | |
a528d35e DH |
3322 | extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int); |
3323 | extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int); | |
b462707e | 3324 | void __inode_add_bytes(struct inode *inode, loff_t bytes); |
1da177e4 | 3325 | void inode_add_bytes(struct inode *inode, loff_t bytes); |
1c8924eb | 3326 | void __inode_sub_bytes(struct inode *inode, loff_t bytes); |
1da177e4 | 3327 | void inode_sub_bytes(struct inode *inode, loff_t bytes); |
f4a8116a JK |
3328 | static inline loff_t __inode_get_bytes(struct inode *inode) |
3329 | { | |
3330 | return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes; | |
3331 | } | |
1da177e4 LT |
3332 | loff_t inode_get_bytes(struct inode *inode); |
3333 | void inode_set_bytes(struct inode *inode, loff_t bytes); | |
fceef393 AV |
3334 | const char *simple_get_link(struct dentry *, struct inode *, |
3335 | struct delayed_call *); | |
61ba64fc | 3336 | extern const struct inode_operations simple_symlink_inode_operations; |
1da177e4 | 3337 | |
5c0ba4e0 | 3338 | extern int iterate_dir(struct file *, struct dir_context *); |
1da177e4 | 3339 | |
09f1bde4 CH |
3340 | int vfs_fstatat(int dfd, const char __user *filename, struct kstat *stat, |
3341 | int flags); | |
da9aa5d9 | 3342 | int vfs_fstat(int fd, struct kstat *stat); |
a528d35e DH |
3343 | |
3344 | static inline int vfs_stat(const char __user *filename, struct kstat *stat) | |
3345 | { | |
0b2c6693 | 3346 | return vfs_fstatat(AT_FDCWD, filename, stat, 0); |
a528d35e DH |
3347 | } |
3348 | static inline int vfs_lstat(const char __user *name, struct kstat *stat) | |
3349 | { | |
0b2c6693 | 3350 | return vfs_fstatat(AT_FDCWD, name, stat, AT_SYMLINK_NOFOLLOW); |
a528d35e | 3351 | } |
a528d35e | 3352 | |
d60874cd | 3353 | extern const char *vfs_get_link(struct dentry *, struct delayed_call *); |
fd4a0edf | 3354 | extern int vfs_readlink(struct dentry *, char __user *, int); |
1da177e4 | 3355 | |
ee416bcd | 3356 | extern struct file_system_type *get_filesystem(struct file_system_type *fs); |
c18479fe | 3357 | extern void put_filesystem(struct file_system_type *fs); |
1da177e4 | 3358 | extern struct file_system_type *get_fs_type(const char *name); |
1da177e4 | 3359 | extern void drop_super(struct super_block *sb); |
ba6379f7 | 3360 | extern void drop_super_exclusive(struct super_block *sb); |
01a05b33 | 3361 | extern void iterate_supers(void (*)(struct super_block *, void *), void *); |
43e15cdb AV |
3362 | extern void iterate_supers_type(struct file_system_type *, |
3363 | void (*)(struct super_block *, void *), void *); | |
1da177e4 LT |
3364 | |
3365 | extern int dcache_dir_open(struct inode *, struct file *); | |
3366 | extern int dcache_dir_close(struct inode *, struct file *); | |
3367 | extern loff_t dcache_dir_lseek(struct file *, loff_t, int); | |
5f99f4e7 | 3368 | extern int dcache_readdir(struct file *, struct dir_context *); |
c1632a0f | 3369 | extern int simple_setattr(struct mnt_idmap *, struct dentry *, |
549c7297 | 3370 | struct iattr *); |
b74d24f7 | 3371 | extern int simple_getattr(struct mnt_idmap *, const struct path *, |
549c7297 | 3372 | struct kstat *, u32, unsigned int); |
726c3342 | 3373 | extern int simple_statfs(struct dentry *, struct kstatfs *); |
20955e89 | 3374 | extern int simple_open(struct inode *inode, struct file *file); |
1da177e4 LT |
3375 | extern int simple_link(struct dentry *, struct inode *, struct dentry *); |
3376 | extern int simple_unlink(struct inode *, struct dentry *); | |
3377 | extern int simple_rmdir(struct inode *, struct dentry *); | |
0c476792 JL |
3378 | void simple_rename_timestamp(struct inode *old_dir, struct dentry *old_dentry, |
3379 | struct inode *new_dir, struct dentry *new_dentry); | |
6429e463 LB |
3380 | extern int simple_rename_exchange(struct inode *old_dir, struct dentry *old_dentry, |
3381 | struct inode *new_dir, struct dentry *new_dentry); | |
e18275ae | 3382 | extern int simple_rename(struct mnt_idmap *, struct inode *, |
549c7297 CB |
3383 | struct dentry *, struct inode *, struct dentry *, |
3384 | unsigned int); | |
a3d1e7eb AV |
3385 | extern void simple_recursive_removal(struct dentry *, |
3386 | void (*callback)(struct dentry *)); | |
02c24a82 | 3387 | extern int noop_fsync(struct file *, loff_t, loff_t, int); |
f44c7763 | 3388 | extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter); |
1da177e4 | 3389 | extern int simple_empty(struct dentry *); |
afddba49 | 3390 | extern int simple_write_begin(struct file *file, struct address_space *mapping, |
9d6b0cd7 | 3391 | loff_t pos, unsigned len, |
1da86618 | 3392 | struct folio **foliop, void **fsdata); |
c1e3dbe9 | 3393 | extern const struct address_space_operations ram_aops; |
b26d4cd3 | 3394 | extern int always_delete_dentry(const struct dentry *); |
6987843f | 3395 | extern struct inode *alloc_anon_inode(struct super_block *); |
c69ff407 | 3396 | extern int simple_nosetlease(struct file *, int, struct file_lease **, void **); |
b26d4cd3 | 3397 | extern const struct dentry_operations simple_dentry_operations; |
1da177e4 | 3398 | |
00cd8dd3 | 3399 | extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags); |
1da177e4 | 3400 | extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *); |
4b6f5d20 | 3401 | extern const struct file_operations simple_dir_operations; |
c5ef1c42 | 3402 | extern const struct inode_operations simple_dir_inode_operations; |
fbabfd0f EB |
3403 | extern void make_empty_dir_inode(struct inode *inode); |
3404 | extern bool is_empty_dir_inode(struct inode *inode); | |
cda37124 | 3405 | struct tree_descr { const char *name; const struct file_operations *ops; int mode; }; |
1da177e4 | 3406 | struct dentry *d_alloc_name(struct dentry *, const char *); |
cda37124 EB |
3407 | extern int simple_fill_super(struct super_block *, unsigned long, |
3408 | const struct tree_descr *); | |
1f5ce9e9 | 3409 | extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count); |
1da177e4 LT |
3410 | extern void simple_release_fs(struct vfsmount **mount, int *count); |
3411 | ||
93b07113 AM |
3412 | extern ssize_t simple_read_from_buffer(void __user *to, size_t count, |
3413 | loff_t *ppos, const void *from, size_t available); | |
6a727b43 JS |
3414 | extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos, |
3415 | const void __user *from, size_t count); | |
1da177e4 | 3416 | |
6faddda6 | 3417 | struct offset_ctx { |
0e4a8621 CL |
3418 | struct maple_tree mt; |
3419 | unsigned long next_offset; | |
6faddda6 CL |
3420 | }; |
3421 | ||
3422 | void simple_offset_init(struct offset_ctx *octx); | |
3423 | int simple_offset_add(struct offset_ctx *octx, struct dentry *dentry); | |
3424 | void simple_offset_remove(struct offset_ctx *octx, struct dentry *dentry); | |
ecba88a3 | 3425 | int simple_offset_empty(struct dentry *dentry); |
5a1a25be CL |
3426 | int simple_offset_rename(struct inode *old_dir, struct dentry *old_dentry, |
3427 | struct inode *new_dir, struct dentry *new_dentry); | |
6faddda6 CL |
3428 | int simple_offset_rename_exchange(struct inode *old_dir, |
3429 | struct dentry *old_dentry, | |
3430 | struct inode *new_dir, | |
3431 | struct dentry *new_dentry); | |
3432 | void simple_offset_destroy(struct offset_ctx *octx); | |
3433 | ||
3434 | extern const struct file_operations simple_offset_dir_operations; | |
3435 | ||
ac13a829 | 3436 | extern int __generic_file_fsync(struct file *, loff_t, loff_t, int); |
02c24a82 | 3437 | extern int generic_file_fsync(struct file *, loff_t, loff_t, int); |
d5aacad5 | 3438 | |
30ca22c7 PL |
3439 | extern int generic_check_addressable(unsigned, u64); |
3440 | ||
70dfe3f0 | 3441 | extern void generic_set_sb_d_ops(struct super_block *sb); |
6a79a4e1 GKB |
3442 | extern int generic_ci_match(const struct inode *parent, |
3443 | const struct qstr *name, | |
3444 | const struct qstr *folded_name, | |
3445 | const u8 *de_name, u32 de_name_len); | |
c843843e | 3446 | |
2824083d GKB |
3447 | static inline bool sb_has_encoding(const struct super_block *sb) |
3448 | { | |
3449 | #if IS_ENABLED(CONFIG_UNICODE) | |
3450 | return !!sb->s_encoding; | |
3451 | #else | |
3452 | return false; | |
3453 | #endif | |
3454 | } | |
c843843e | 3455 | |
4609e1f1 | 3456 | int may_setattr(struct mnt_idmap *idmap, struct inode *inode, |
7bb698f0 | 3457 | unsigned int ia_valid); |
c1632a0f | 3458 | int setattr_prepare(struct mnt_idmap *, struct dentry *, struct iattr *); |
25d9e2d1 | 3459 | extern int inode_newsize_ok(const struct inode *, loff_t offset); |
c1632a0f | 3460 | void setattr_copy(struct mnt_idmap *, struct inode *inode, |
2f221d6f | 3461 | const struct iattr *attr); |
1da177e4 | 3462 | |
c3b2da31 | 3463 | extern int file_update_time(struct file *file); |
1da177e4 | 3464 | |
f05a3849 | 3465 | static inline bool vma_is_dax(const struct vm_area_struct *vma) |
baabda26 DW |
3466 | { |
3467 | return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host); | |
3468 | } | |
3469 | ||
2bb6d283 DW |
3470 | static inline bool vma_is_fsdax(struct vm_area_struct *vma) |
3471 | { | |
3472 | struct inode *inode; | |
3473 | ||
52650c8b | 3474 | if (!IS_ENABLED(CONFIG_FS_DAX) || !vma->vm_file) |
2bb6d283 DW |
3475 | return false; |
3476 | if (!vma_is_dax(vma)) | |
3477 | return false; | |
3478 | inode = file_inode(vma->vm_file); | |
230f5a89 | 3479 | if (S_ISCHR(inode->i_mode)) |
2bb6d283 DW |
3480 | return false; /* device-dax */ |
3481 | return true; | |
3482 | } | |
3483 | ||
2ba48ce5 AV |
3484 | static inline int iocb_flags(struct file *file) |
3485 | { | |
3486 | int res = 0; | |
3487 | if (file->f_flags & O_APPEND) | |
3488 | res |= IOCB_APPEND; | |
efbe3c24 | 3489 | if (file->f_flags & O_DIRECT) |
2ba48ce5 | 3490 | res |= IOCB_DIRECT; |
91b94c5d | 3491 | if (file->f_flags & O_DSYNC) |
dde0c2e7 CH |
3492 | res |= IOCB_DSYNC; |
3493 | if (file->f_flags & __O_SYNC) | |
3494 | res |= IOCB_SYNC; | |
2ba48ce5 AV |
3495 | return res; |
3496 | } | |
3497 | ||
c34fc6f2 PS |
3498 | static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags, |
3499 | int rw_type) | |
fdd2f5b7 | 3500 | { |
1752f0ad PB |
3501 | int kiocb_flags = 0; |
3502 | ||
ce71bfea JA |
3503 | /* make sure there's no overlap between RWF and private IOCB flags */ |
3504 | BUILD_BUG_ON((__force int) RWF_SUPPORTED & IOCB_EVENTFD); | |
3505 | ||
1752f0ad PB |
3506 | if (!flags) |
3507 | return 0; | |
fdd2f5b7 GR |
3508 | if (unlikely(flags & ~RWF_SUPPORTED)) |
3509 | return -EOPNOTSUPP; | |
73fa7547 RF |
3510 | if (unlikely((flags & RWF_APPEND) && (flags & RWF_NOAPPEND))) |
3511 | return -EINVAL; | |
fdd2f5b7 | 3512 | |
b745fafa | 3513 | if (flags & RWF_NOWAIT) { |
91f9943e | 3514 | if (!(ki->ki_filp->f_mode & FMODE_NOWAIT)) |
b745fafa | 3515 | return -EOPNOTSUPP; |
b745fafa | 3516 | } |
c34fc6f2 PS |
3517 | if (flags & RWF_ATOMIC) { |
3518 | if (rw_type != WRITE) | |
3519 | return -EOPNOTSUPP; | |
3520 | if (!(ki->ki_filp->f_mode & FMODE_CAN_ATOMIC_WRITE)) | |
3521 | return -EOPNOTSUPP; | |
3522 | } | |
ce71bfea | 3523 | kiocb_flags |= (__force int) (flags & RWF_SUPPORTED); |
fdd2f5b7 | 3524 | if (flags & RWF_SYNC) |
ce71bfea | 3525 | kiocb_flags |= IOCB_DSYNC; |
1752f0ad | 3526 | |
73fa7547 RF |
3527 | if ((flags & RWF_NOAPPEND) && (ki->ki_flags & IOCB_APPEND)) { |
3528 | if (IS_APPEND(file_inode(ki->ki_filp))) | |
3529 | return -EPERM; | |
3530 | ki->ki_flags &= ~IOCB_APPEND; | |
3531 | } | |
3532 | ||
1752f0ad | 3533 | ki->ki_flags |= kiocb_flags; |
fdd2f5b7 GR |
3534 | return 0; |
3535 | } | |
3536 | ||
1da177e4 LT |
3537 | /* Transaction based IO helpers */ |
3538 | ||
3539 | /* | |
3540 | * An argresp is stored in an allocated page and holds the | |
3541 | * size of the argument or response, along with its content | |
3542 | */ | |
3543 | struct simple_transaction_argresp { | |
3544 | ssize_t size; | |
5e01fdff | 3545 | char data[]; |
1da177e4 LT |
3546 | }; |
3547 | ||
3548 | #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp)) | |
3549 | ||
3550 | char *simple_transaction_get(struct file *file, const char __user *buf, | |
3551 | size_t size); | |
3552 | ssize_t simple_transaction_read(struct file *file, char __user *buf, | |
3553 | size_t size, loff_t *pos); | |
3554 | int simple_transaction_release(struct inode *inode, struct file *file); | |
3555 | ||
76791ab2 | 3556 | void simple_transaction_set(struct file *file, size_t n); |
1da177e4 | 3557 | |
acaefc25 AB |
3558 | /* |
3559 | * simple attribute files | |
3560 | * | |
3561 | * These attributes behave similar to those in sysfs: | |
3562 | * | |
3563 | * Writing to an attribute immediately sets a value, an open file can be | |
3564 | * written to multiple times. | |
3565 | * | |
3566 | * Reading from an attribute creates a buffer from the value that might get | |
3567 | * read with multiple read calls. When the attribute has been read | |
3568 | * completely, no further read calls are possible until the file is opened | |
3569 | * again. | |
3570 | * | |
3571 | * All attributes contain a text representation of a numeric value | |
3572 | * that are accessed with the get() and set() functions. | |
3573 | */ | |
2e41f274 | 3574 | #define DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, __is_signed) \ |
acaefc25 AB |
3575 | static int __fops ## _open(struct inode *inode, struct file *file) \ |
3576 | { \ | |
3577 | __simple_attr_check_format(__fmt, 0ull); \ | |
3578 | return simple_attr_open(inode, file, __get, __set, __fmt); \ | |
3579 | } \ | |
828c0950 | 3580 | static const struct file_operations __fops = { \ |
acaefc25 AB |
3581 | .owner = THIS_MODULE, \ |
3582 | .open = __fops ## _open, \ | |
74bedc4d | 3583 | .release = simple_attr_release, \ |
acaefc25 | 3584 | .read = simple_attr_read, \ |
2e41f274 | 3585 | .write = (__is_signed) ? simple_attr_write_signed : simple_attr_write, \ |
1ec5584e | 3586 | .llseek = generic_file_llseek, \ |
68be3029 | 3587 | } |
acaefc25 | 3588 | |
2e41f274 AM |
3589 | #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \ |
3590 | DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, false) | |
3591 | ||
3592 | #define DEFINE_SIMPLE_ATTRIBUTE_SIGNED(__fops, __get, __set, __fmt) \ | |
3593 | DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, true) | |
3594 | ||
b9075fa9 JP |
3595 | static inline __printf(1, 2) |
3596 | void __simple_attr_check_format(const char *fmt, ...) | |
acaefc25 AB |
3597 | { |
3598 | /* don't do anything, just let the compiler check the arguments; */ | |
3599 | } | |
3600 | ||
3601 | int simple_attr_open(struct inode *inode, struct file *file, | |
8b88b099 | 3602 | int (*get)(void *, u64 *), int (*set)(void *, u64), |
acaefc25 | 3603 | const char *fmt); |
74bedc4d | 3604 | int simple_attr_release(struct inode *inode, struct file *file); |
acaefc25 AB |
3605 | ssize_t simple_attr_read(struct file *file, char __user *buf, |
3606 | size_t len, loff_t *ppos); | |
3607 | ssize_t simple_attr_write(struct file *file, const char __user *buf, | |
3608 | size_t len, loff_t *ppos); | |
2e41f274 AM |
3609 | ssize_t simple_attr_write_signed(struct file *file, const char __user *buf, |
3610 | size_t len, loff_t *ppos); | |
acaefc25 | 3611 | |
4be28540 | 3612 | struct ctl_table; |
6e7c1770 | 3613 | int __init list_bdev_fs_names(char *buf, size_t size); |
62239ac2 | 3614 | |
3cd90ea4 | 3615 | #define __FMODE_EXEC ((__force int) FMODE_EXEC) |
1a44bc8c NK |
3616 | #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY) |
3617 | ||
6d125529 | 3618 | #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE]) |
ecf081d1 | 3619 | #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \ |
1a44bc8c | 3620 | (flag & __FMODE_NONOTIFY))) |
5300990c | 3621 | |
d37177ba | 3622 | static inline bool is_sxid(umode_t mode) |
69b45732 | 3623 | { |
8d84e39d | 3624 | return mode & (S_ISUID | S_ISGID); |
69b45732 AK |
3625 | } |
3626 | ||
9452e93e | 3627 | static inline int check_sticky(struct mnt_idmap *idmap, |
ba73d987 | 3628 | struct inode *dir, struct inode *inode) |
cbdf35bc MS |
3629 | { |
3630 | if (!(dir->i_mode & S_ISVTX)) | |
3631 | return 0; | |
3632 | ||
9452e93e | 3633 | return __check_sticky(idmap, dir, inode); |
cbdf35bc MS |
3634 | } |
3635 | ||
69b45732 AK |
3636 | static inline void inode_has_no_xattr(struct inode *inode) |
3637 | { | |
e462ec50 | 3638 | if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC)) |
69b45732 AK |
3639 | inode->i_flags |= S_NOSEC; |
3640 | } | |
3641 | ||
a7400222 AV |
3642 | static inline bool is_root_inode(struct inode *inode) |
3643 | { | |
3644 | return inode == inode->i_sb->s_root->d_inode; | |
3645 | } | |
3646 | ||
5f99f4e7 AV |
3647 | static inline bool dir_emit(struct dir_context *ctx, |
3648 | const char *name, int namelen, | |
3649 | u64 ino, unsigned type) | |
3650 | { | |
25885a35 | 3651 | return ctx->actor(ctx, name, namelen, ctx->pos, ino, type); |
5f99f4e7 AV |
3652 | } |
3653 | static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx) | |
3654 | { | |
3655 | return ctx->actor(ctx, ".", 1, ctx->pos, | |
25885a35 | 3656 | file->f_path.dentry->d_inode->i_ino, DT_DIR); |
5f99f4e7 AV |
3657 | } |
3658 | static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx) | |
3659 | { | |
3660 | return ctx->actor(ctx, "..", 2, ctx->pos, | |
f378ec4e | 3661 | d_parent_ino(file->f_path.dentry), DT_DIR); |
5f99f4e7 AV |
3662 | } |
3663 | static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx) | |
3664 | { | |
3665 | if (ctx->pos == 0) { | |
3666 | if (!dir_emit_dot(file, ctx)) | |
3667 | return false; | |
3668 | ctx->pos = 1; | |
3669 | } | |
3670 | if (ctx->pos == 1) { | |
3671 | if (!dir_emit_dotdot(file, ctx)) | |
3672 | return false; | |
3673 | ctx->pos = 2; | |
3674 | } | |
3675 | return true; | |
3676 | } | |
5ded75ec AV |
3677 | static inline bool dir_relax(struct inode *inode) |
3678 | { | |
5955102c AV |
3679 | inode_unlock(inode); |
3680 | inode_lock(inode); | |
5ded75ec AV |
3681 | return !IS_DEADDIR(inode); |
3682 | } | |
5f99f4e7 | 3683 | |
ae05327a AV |
3684 | static inline bool dir_relax_shared(struct inode *inode) |
3685 | { | |
3686 | inode_unlock_shared(inode); | |
3687 | inode_lock_shared(inode); | |
3688 | return !IS_DEADDIR(inode); | |
3689 | } | |
3690 | ||
90f8572b | 3691 | extern bool path_noexec(const struct path *path); |
21fc61c7 | 3692 | extern void inode_nohighmem(struct inode *inode); |
90f8572b | 3693 | |
45cd0faa AG |
3694 | /* mm/fadvise.c */ |
3695 | extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len, | |
3696 | int advice); | |
cf1ea059 JK |
3697 | extern int generic_fadvise(struct file *file, loff_t offset, loff_t len, |
3698 | int advice); | |
45cd0faa | 3699 | |
1bc6d445 CB |
3700 | static inline bool vfs_empty_path(int dfd, const char __user *path) |
3701 | { | |
3702 | char c; | |
3703 | ||
3704 | if (dfd < 0) | |
3705 | return false; | |
3706 | ||
3707 | /* We now allow NULL to be used for empty path. */ | |
3708 | if (!path) | |
3709 | return true; | |
3710 | ||
3711 | if (unlikely(get_user(c, path))) | |
3712 | return false; | |
3713 | ||
3714 | return !c; | |
3715 | } | |
3716 | ||
c34fc6f2 PS |
3717 | bool generic_atomic_write_valid(struct iov_iter *iter, loff_t pos); |
3718 | ||
1da177e4 | 3719 | #endif /* _LINUX_FS_H */ |