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> |
1da177e4 | 46 | |
1da177e4 | 47 | #include <asm/byteorder.h> |
607ca46e | 48 | #include <uapi/linux/fs.h> |
1da177e4 | 49 | |
b83ae6d4 | 50 | struct backing_dev_info; |
52ebea74 | 51 | struct bdi_writeback; |
2f8b5444 | 52 | struct bio; |
5a72e899 | 53 | struct io_comp_batch; |
a5694255 | 54 | struct export_operations; |
10c5db28 | 55 | struct fiemap_extent_info; |
a885c8c4 | 56 | struct hd_geometry; |
1da177e4 | 57 | struct iovec; |
92198f7e | 58 | struct kiocb; |
57cc7215 | 59 | struct kobject; |
1da177e4 LT |
60 | struct pipe_inode_info; |
61 | struct poll_table_struct; | |
62 | struct kstatfs; | |
63 | struct vm_area_struct; | |
64 | struct vfsmount; | |
745ca247 | 65 | struct cred; |
a509bc1a | 66 | struct swap_info_struct; |
55985dd7 | 67 | struct seq_file; |
7b7a8665 | 68 | struct workqueue_struct; |
3b93f911 | 69 | struct iov_iter; |
0b81d077 JK |
70 | struct fscrypt_info; |
71 | struct fscrypt_operations; | |
5585f2af EB |
72 | struct fsverity_info; |
73 | struct fsverity_operations; | |
f3a09c92 | 74 | struct fs_context; |
d7167b14 | 75 | struct fs_parameter_spec; |
4c5b4799 | 76 | struct fileattr; |
6f7db389 | 77 | struct iomap_ops; |
1da177e4 | 78 | |
74bf17cf | 79 | extern void __init inode_init(void); |
1da177e4 | 80 | extern void __init inode_init_early(void); |
4248b0da MG |
81 | extern void __init files_init(void); |
82 | extern void __init files_maxfiles_init(void); | |
1da177e4 | 83 | |
518de9b3 | 84 | extern unsigned long get_max_files(void); |
9b80a184 | 85 | extern unsigned int sysctl_nr_open; |
dded4f4d | 86 | |
ddef7ed2 CH |
87 | typedef __kernel_rwf_t rwf_t; |
88 | ||
1da177e4 LT |
89 | struct buffer_head; |
90 | typedef int (get_block_t)(struct inode *inode, sector_t iblock, | |
91 | struct buffer_head *bh_result, int create); | |
187372a3 | 92 | typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset, |
7b7a8665 | 93 | ssize_t bytes, void *private); |
1da177e4 | 94 | |
bbc1096a DH |
95 | #define MAY_EXEC 0x00000001 |
96 | #define MAY_WRITE 0x00000002 | |
97 | #define MAY_READ 0x00000004 | |
98 | #define MAY_APPEND 0x00000008 | |
99 | #define MAY_ACCESS 0x00000010 | |
100 | #define MAY_OPEN 0x00000020 | |
101 | #define MAY_CHDIR 0x00000040 | |
102 | /* called from RCU mode, don't block */ | |
103 | #define MAY_NOT_BLOCK 0x00000080 | |
104 | ||
105 | /* | |
106 | * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond | |
75abe329 | 107 | * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open() |
bbc1096a DH |
108 | */ |
109 | ||
110 | /* file is open for reading */ | |
111 | #define FMODE_READ ((__force fmode_t)0x1) | |
112 | /* file is open for writing */ | |
113 | #define FMODE_WRITE ((__force fmode_t)0x2) | |
114 | /* file is seekable */ | |
115 | #define FMODE_LSEEK ((__force fmode_t)0x4) | |
116 | /* file can be accessed using pread */ | |
117 | #define FMODE_PREAD ((__force fmode_t)0x8) | |
118 | /* file can be accessed using pwrite */ | |
119 | #define FMODE_PWRITE ((__force fmode_t)0x10) | |
120 | /* File is opened for execution with sys_execve / sys_uselib */ | |
121 | #define FMODE_EXEC ((__force fmode_t)0x20) | |
122 | /* File is opened with O_NDELAY (only set for block devices) */ | |
123 | #define FMODE_NDELAY ((__force fmode_t)0x40) | |
124 | /* File is opened with O_EXCL (only set for block devices) */ | |
125 | #define FMODE_EXCL ((__force fmode_t)0x80) | |
126 | /* File is opened using open(.., 3, ..) and is writeable only for ioctls | |
127 | (specialy hack for floppy.c) */ | |
128 | #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100) | |
129 | /* 32bit hashes as llseek() offset (for directories) */ | |
130 | #define FMODE_32BITHASH ((__force fmode_t)0x200) | |
131 | /* 64bit hashes as llseek() offset (for directories) */ | |
132 | #define FMODE_64BITHASH ((__force fmode_t)0x400) | |
133 | ||
134 | /* | |
135 | * Don't update ctime and mtime. | |
136 | * | |
137 | * Currently a special hack for the XFS open_by_handle ioctl, but we'll | |
138 | * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon. | |
139 | */ | |
140 | #define FMODE_NOCMTIME ((__force fmode_t)0x800) | |
141 | ||
142 | /* Expect random access pattern */ | |
143 | #define FMODE_RANDOM ((__force fmode_t)0x1000) | |
144 | ||
bbcd53c9 | 145 | /* File is huge (eg. /dev/mem): treat loff_t as unsigned */ |
bbc1096a DH |
146 | #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000) |
147 | ||
148 | /* File is opened with O_PATH; almost nothing can be done with it */ | |
149 | #define FMODE_PATH ((__force fmode_t)0x4000) | |
150 | ||
2be7d348 LT |
151 | /* File needs atomic accesses to f_pos */ |
152 | #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000) | |
83f936c7 AV |
153 | /* Write access to underlying fs */ |
154 | #define FMODE_WRITER ((__force fmode_t)0x10000) | |
7f7f25e8 AV |
155 | /* Has read method(s) */ |
156 | #define FMODE_CAN_READ ((__force fmode_t)0x20000) | |
157 | /* Has write method(s) */ | |
158 | #define FMODE_CAN_WRITE ((__force fmode_t)0x40000) | |
9c225f26 | 159 | |
f5d11409 | 160 | #define FMODE_OPENED ((__force fmode_t)0x80000) |
73a09dd9 | 161 | #define FMODE_CREATED ((__force fmode_t)0x100000) |
f5d11409 | 162 | |
10dce8af KS |
163 | /* File is stream-like */ |
164 | #define FMODE_STREAM ((__force fmode_t)0x200000) | |
165 | ||
a2ad63da N |
166 | /* File supports DIRECT IO */ |
167 | #define FMODE_CAN_ODIRECT ((__force fmode_t)0x400000) | |
168 | ||
17e81022 YZ |
169 | #define FMODE_NOREUSE ((__force fmode_t)0x800000) |
170 | ||
d8aeb44a JA |
171 | /* File supports non-exclusive O_DIRECT writes from multiple threads */ |
172 | #define FMODE_DIO_PARALLEL_WRITE ((__force fmode_t)0x1000000) | |
173 | ||
bbc1096a | 174 | /* File was opened by fanotify and shouldn't generate fanotify events */ |
75069f2b | 175 | #define FMODE_NONOTIFY ((__force fmode_t)0x4000000) |
bbc1096a | 176 | |
91f9943e | 177 | /* File is capable of returning -EAGAIN if I/O will block */ |
a07b2000 AV |
178 | #define FMODE_NOWAIT ((__force fmode_t)0x8000000) |
179 | ||
180 | /* File represents mount that needs unmounting */ | |
181 | #define FMODE_NEED_UNMOUNT ((__force fmode_t)0x10000000) | |
b745fafa | 182 | |
d3b1084d | 183 | /* File does not contribute to nr_files count */ |
a07b2000 | 184 | #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000) |
d3b1084d | 185 | |
c2a25ec0 JA |
186 | /* File supports async buffered reads */ |
187 | #define FMODE_BUF_RASYNC ((__force fmode_t)0x40000000) | |
188 | ||
80175539 SR |
189 | /* File supports async nowait buffered writes */ |
190 | #define FMODE_BUF_WASYNC ((__force fmode_t)0x80000000) | |
191 | ||
1da177e4 LT |
192 | /* |
193 | * Attribute flags. These should be or-ed together to figure out what | |
194 | * has been changed! | |
195 | */ | |
9767d749 MS |
196 | #define ATTR_MODE (1 << 0) |
197 | #define ATTR_UID (1 << 1) | |
198 | #define ATTR_GID (1 << 2) | |
199 | #define ATTR_SIZE (1 << 3) | |
200 | #define ATTR_ATIME (1 << 4) | |
201 | #define ATTR_MTIME (1 << 5) | |
202 | #define ATTR_CTIME (1 << 6) | |
203 | #define ATTR_ATIME_SET (1 << 7) | |
204 | #define ATTR_MTIME_SET (1 << 8) | |
205 | #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */ | |
9767d749 MS |
206 | #define ATTR_KILL_SUID (1 << 11) |
207 | #define ATTR_KILL_SGID (1 << 12) | |
208 | #define ATTR_FILE (1 << 13) | |
209 | #define ATTR_KILL_PRIV (1 << 14) | |
210 | #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */ | |
211 | #define ATTR_TIMES_SET (1 << 16) | |
f2b20f6e | 212 | #define ATTR_TOUCH (1 << 17) |
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 | |
c75b1d94 JA |
316 | /* |
317 | * Write life time hint values. | |
69448867 | 318 | * Stored in struct inode as u8. |
c75b1d94 JA |
319 | */ |
320 | enum rw_hint { | |
321 | WRITE_LIFE_NOT_SET = 0, | |
322 | WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE, | |
323 | WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT, | |
324 | WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM, | |
325 | WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG, | |
326 | WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME, | |
327 | }; | |
328 | ||
ce71bfea JA |
329 | /* Match RWF_* bits to IOCB bits */ |
330 | #define IOCB_HIPRI (__force int) RWF_HIPRI | |
331 | #define IOCB_DSYNC (__force int) RWF_DSYNC | |
332 | #define IOCB_SYNC (__force int) RWF_SYNC | |
333 | #define IOCB_NOWAIT (__force int) RWF_NOWAIT | |
334 | #define IOCB_APPEND (__force int) RWF_APPEND | |
335 | ||
336 | /* non-RWF related bits - start at 16 */ | |
337 | #define IOCB_EVENTFD (1 << 16) | |
338 | #define IOCB_DIRECT (1 << 17) | |
339 | #define IOCB_WRITE (1 << 18) | |
dd3e6d50 | 340 | /* iocb->ki_waitq is valid */ |
ce71bfea JA |
341 | #define IOCB_WAITQ (1 << 19) |
342 | #define IOCB_NOIO (1 << 20) | |
6c7ef543 JA |
343 | /* can use bio alloc cache */ |
344 | #define IOCB_ALLOC_CACHE (1 << 21) | |
e2e40f2c | 345 | |
f6c73a11 RHI |
346 | /* for use in trace events */ |
347 | #define TRACE_IOCB_STRINGS \ | |
348 | { IOCB_HIPRI, "HIPRI" }, \ | |
349 | { IOCB_DSYNC, "DSYNC" }, \ | |
350 | { IOCB_SYNC, "SYNC" }, \ | |
351 | { IOCB_NOWAIT, "NOWAIT" }, \ | |
352 | { IOCB_APPEND, "APPEND" }, \ | |
353 | { IOCB_EVENTFD, "EVENTFD"}, \ | |
354 | { IOCB_DIRECT, "DIRECT" }, \ | |
355 | { IOCB_WRITE, "WRITE" }, \ | |
356 | { IOCB_WAITQ, "WAITQ" }, \ | |
357 | { IOCB_NOIO, "NOIO" }, \ | |
358 | { IOCB_ALLOC_CACHE, "ALLOC_CACHE" } | |
359 | ||
e2e40f2c CH |
360 | struct kiocb { |
361 | struct file *ki_filp; | |
362 | loff_t ki_pos; | |
6b19b766 | 363 | void (*ki_complete)(struct kiocb *iocb, long ret); |
e2e40f2c CH |
364 | void *private; |
365 | int ki_flags; | |
d9a08a9e | 366 | u16 ki_ioprio; /* See linux/ioprio.h */ |
3e08773c | 367 | struct wait_page_queue *ki_waitq; /* for async buffered IO */ |
84c4e1f8 | 368 | }; |
e2e40f2c CH |
369 | |
370 | static inline bool is_sync_kiocb(struct kiocb *kiocb) | |
371 | { | |
372 | return kiocb->ki_complete == NULL; | |
373 | } | |
374 | ||
1da177e4 LT |
375 | struct address_space_operations { |
376 | int (*writepage)(struct page *page, struct writeback_control *wbc); | |
5efe7448 | 377 | int (*read_folio)(struct file *, struct folio *); |
1da177e4 LT |
378 | |
379 | /* Write back some dirty pages from this mapping. */ | |
380 | int (*writepages)(struct address_space *, struct writeback_control *); | |
381 | ||
3a3bae50 | 382 | /* Mark a folio dirty. Return true if this dirtied it */ |
6f31a5a2 | 383 | bool (*dirty_folio)(struct address_space *, struct folio *); |
1da177e4 | 384 | |
8151b4c8 | 385 | void (*readahead)(struct readahead_control *); |
1da177e4 | 386 | |
afddba49 | 387 | int (*write_begin)(struct file *, struct address_space *mapping, |
9d6b0cd7 | 388 | loff_t pos, unsigned len, |
afddba49 NP |
389 | struct page **pagep, void **fsdata); |
390 | int (*write_end)(struct file *, struct address_space *mapping, | |
391 | loff_t pos, unsigned len, unsigned copied, | |
392 | struct page *page, void *fsdata); | |
393 | ||
1da177e4 LT |
394 | /* Unfortunately this kludge is needed for FIBMAP. Don't use it */ |
395 | sector_t (*bmap)(struct address_space *, sector_t); | |
128d1f82 | 396 | void (*invalidate_folio) (struct folio *, size_t offset, size_t len); |
fa29000b | 397 | bool (*release_folio)(struct folio *, gfp_t); |
d2329aa0 | 398 | void (*free_folio)(struct folio *folio); |
c8b8e32d | 399 | ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter); |
b969c4ab | 400 | /* |
5490da4f | 401 | * migrate the contents of a folio to the specified target. If |
ef277c73 | 402 | * migrate_mode is MIGRATE_ASYNC, it must not block. |
b969c4ab | 403 | */ |
5490da4f MWO |
404 | int (*migrate_folio)(struct address_space *, struct folio *dst, |
405 | struct folio *src, enum migrate_mode); | |
072acba6 | 406 | int (*launder_folio)(struct folio *); |
2e7e80f7 MWO |
407 | bool (*is_partially_uptodate) (struct folio *, size_t from, |
408 | size_t count); | |
520f301c | 409 | void (*is_dirty_writeback) (struct folio *, bool *dirty, bool *wb); |
25718736 | 410 | int (*error_remove_page)(struct address_space *, struct page *); |
62c230bc MG |
411 | |
412 | /* swapfile support */ | |
a509bc1a MG |
413 | int (*swap_activate)(struct swap_info_struct *sis, struct file *file, |
414 | sector_t *span); | |
415 | void (*swap_deactivate)(struct file *file); | |
e1209d3a | 416 | int (*swap_rw)(struct kiocb *iocb, struct iov_iter *iter); |
1da177e4 LT |
417 | }; |
418 | ||
7dcda1c9 JA |
419 | extern const struct address_space_operations empty_aops; |
420 | ||
eb797a8e MW |
421 | /** |
422 | * struct address_space - Contents of a cacheable, mappable object. | |
423 | * @host: Owner, either the inode or the block_device. | |
424 | * @i_pages: Cached pages. | |
730633f0 JK |
425 | * @invalidate_lock: Guards coherency between page cache contents and |
426 | * file offset->disk block mappings in the filesystem during invalidates. | |
427 | * It is also used to block modification of page cache contents through | |
428 | * memory mappings. | |
eb797a8e MW |
429 | * @gfp_mask: Memory allocation flags to use for allocating pages. |
430 | * @i_mmap_writable: Number of VM_SHARED mappings. | |
09d91cda | 431 | * @nr_thps: Number of THPs in the pagecache (non-shmem only). |
eb797a8e MW |
432 | * @i_mmap: Tree of private and shared mappings. |
433 | * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable. | |
434 | * @nrpages: Number of page entries, protected by the i_pages lock. | |
eb797a8e MW |
435 | * @writeback_index: Writeback starts here. |
436 | * @a_ops: Methods. | |
437 | * @flags: Error bits and flags (AS_*). | |
438 | * @wb_err: The most recent error which has occurred. | |
439 | * @private_lock: For use by the owner of the address_space. | |
440 | * @private_list: For use by the owner of the address_space. | |
441 | * @private_data: For use by the owner of the address_space. | |
442 | */ | |
1da177e4 | 443 | struct address_space { |
eb797a8e MW |
444 | struct inode *host; |
445 | struct xarray i_pages; | |
730633f0 | 446 | struct rw_semaphore invalidate_lock; |
eb797a8e MW |
447 | gfp_t gfp_mask; |
448 | atomic_t i_mmap_writable; | |
09d91cda SL |
449 | #ifdef CONFIG_READ_ONLY_THP_FOR_FS |
450 | /* number of thp, only for non-shmem files */ | |
451 | atomic_t nr_thps; | |
452 | #endif | |
eb797a8e MW |
453 | struct rb_root_cached i_mmap; |
454 | struct rw_semaphore i_mmap_rwsem; | |
455 | unsigned long nrpages; | |
eb797a8e MW |
456 | pgoff_t writeback_index; |
457 | const struct address_space_operations *a_ops; | |
458 | unsigned long flags; | |
5660e13d | 459 | errseq_t wb_err; |
eb797a8e MW |
460 | spinlock_t private_lock; |
461 | struct list_head private_list; | |
462 | void *private_data; | |
3859a271 | 463 | } __attribute__((aligned(sizeof(long)))) __randomize_layout; |
1da177e4 LT |
464 | /* |
465 | * On most architectures that alignment is already the case; but | |
25985edc | 466 | * must be enforced here for CRIS, to let the least significant bit |
1da177e4 LT |
467 | * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON. |
468 | */ | |
1da177e4 | 469 | |
ff9c745b MW |
470 | /* XArray tags, for tagging dirty and writeback pages in the pagecache. */ |
471 | #define PAGECACHE_TAG_DIRTY XA_MARK_0 | |
472 | #define PAGECACHE_TAG_WRITEBACK XA_MARK_1 | |
473 | #define PAGECACHE_TAG_TOWRITE XA_MARK_2 | |
474 | ||
1da177e4 | 475 | /* |
ff9c745b | 476 | * Returns true if any of the pages in the mapping are marked with the tag. |
1da177e4 | 477 | */ |
ff9c745b MW |
478 | static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag) |
479 | { | |
480 | return xa_marked(&mapping->i_pages, tag); | |
481 | } | |
1da177e4 | 482 | |
8b28f621 DB |
483 | static inline void i_mmap_lock_write(struct address_space *mapping) |
484 | { | |
c8c06efa | 485 | down_write(&mapping->i_mmap_rwsem); |
8b28f621 DB |
486 | } |
487 | ||
c0d0381a MK |
488 | static inline int i_mmap_trylock_write(struct address_space *mapping) |
489 | { | |
490 | return down_write_trylock(&mapping->i_mmap_rwsem); | |
491 | } | |
492 | ||
8b28f621 DB |
493 | static inline void i_mmap_unlock_write(struct address_space *mapping) |
494 | { | |
c8c06efa | 495 | up_write(&mapping->i_mmap_rwsem); |
8b28f621 DB |
496 | } |
497 | ||
6d4675e6 MK |
498 | static inline int i_mmap_trylock_read(struct address_space *mapping) |
499 | { | |
500 | return down_read_trylock(&mapping->i_mmap_rwsem); | |
501 | } | |
502 | ||
3dec0ba0 DB |
503 | static inline void i_mmap_lock_read(struct address_space *mapping) |
504 | { | |
505 | down_read(&mapping->i_mmap_rwsem); | |
506 | } | |
507 | ||
508 | static inline void i_mmap_unlock_read(struct address_space *mapping) | |
509 | { | |
510 | up_read(&mapping->i_mmap_rwsem); | |
511 | } | |
512 | ||
34ae204f MK |
513 | static inline void i_mmap_assert_locked(struct address_space *mapping) |
514 | { | |
515 | lockdep_assert_held(&mapping->i_mmap_rwsem); | |
516 | } | |
517 | ||
518 | static inline void i_mmap_assert_write_locked(struct address_space *mapping) | |
519 | { | |
520 | lockdep_assert_held_write(&mapping->i_mmap_rwsem); | |
521 | } | |
522 | ||
1da177e4 LT |
523 | /* |
524 | * Might pages of this file be mapped into userspace? | |
525 | */ | |
526 | static inline int mapping_mapped(struct address_space *mapping) | |
527 | { | |
f808c13f | 528 | return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root); |
1da177e4 LT |
529 | } |
530 | ||
531 | /* | |
532 | * Might pages of this file have been modified in userspace? | |
45e55300 | 533 | * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap |
1da177e4 LT |
534 | * marks vma as VM_SHARED if it is shared, and the file was opened for |
535 | * writing i.e. vma may be mprotected writable even if now readonly. | |
4bb5f5d9 DH |
536 | * |
537 | * If i_mmap_writable is negative, no new writable mappings are allowed. You | |
538 | * can only deny writable mappings, if none exists right now. | |
1da177e4 LT |
539 | */ |
540 | static inline int mapping_writably_mapped(struct address_space *mapping) | |
541 | { | |
4bb5f5d9 DH |
542 | return atomic_read(&mapping->i_mmap_writable) > 0; |
543 | } | |
544 | ||
545 | static inline int mapping_map_writable(struct address_space *mapping) | |
546 | { | |
547 | return atomic_inc_unless_negative(&mapping->i_mmap_writable) ? | |
548 | 0 : -EPERM; | |
549 | } | |
550 | ||
551 | static inline void mapping_unmap_writable(struct address_space *mapping) | |
552 | { | |
553 | atomic_dec(&mapping->i_mmap_writable); | |
554 | } | |
555 | ||
556 | static inline int mapping_deny_writable(struct address_space *mapping) | |
557 | { | |
558 | return atomic_dec_unless_positive(&mapping->i_mmap_writable) ? | |
559 | 0 : -EBUSY; | |
560 | } | |
561 | ||
562 | static inline void mapping_allow_writable(struct address_space *mapping) | |
563 | { | |
564 | atomic_inc(&mapping->i_mmap_writable); | |
1da177e4 LT |
565 | } |
566 | ||
567 | /* | |
568 | * Use sequence counter to get consistent i_size on 32-bit processors. | |
569 | */ | |
570 | #if BITS_PER_LONG==32 && defined(CONFIG_SMP) | |
571 | #include <linux/seqlock.h> | |
572 | #define __NEED_I_SIZE_ORDERED | |
573 | #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount) | |
574 | #else | |
575 | #define i_size_ordered_init(inode) do { } while (0) | |
576 | #endif | |
577 | ||
f19d4a8f AV |
578 | struct posix_acl; |
579 | #define ACL_NOT_CACHED ((void *)(-1)) | |
332f606b MS |
580 | /* |
581 | * ACL_DONT_CACHE is for stacked filesystems, that rely on underlying fs to | |
cac2f8b8 CB |
582 | * cache the ACL. This also means that ->get_inode_acl() can be called in RCU |
583 | * mode with the LOOKUP_RCU flag. | |
332f606b | 584 | */ |
2a3a2a3f | 585 | #define ACL_DONT_CACHE ((void *)(-3)) |
f19d4a8f | 586 | |
b8a7a3a6 AG |
587 | static inline struct posix_acl * |
588 | uncached_acl_sentinel(struct task_struct *task) | |
589 | { | |
590 | return (void *)task + 1; | |
591 | } | |
592 | ||
593 | static inline bool | |
594 | is_uncached_acl(struct posix_acl *acl) | |
595 | { | |
596 | return (long)acl & 1; | |
597 | } | |
598 | ||
3ddcd056 LT |
599 | #define IOP_FASTPERM 0x0001 |
600 | #define IOP_LOOKUP 0x0002 | |
601 | #define IOP_NOFOLLOW 0x0004 | |
d0a5b995 | 602 | #define IOP_XATTR 0x0008 |
76fca90e | 603 | #define IOP_DEFAULT_READLINK 0x0010 |
3ddcd056 | 604 | |
9dd813c1 JK |
605 | struct fsnotify_mark_connector; |
606 | ||
3ddcd056 LT |
607 | /* |
608 | * Keep mostly read-only and often accessed (especially for | |
609 | * the RCU path lookup and 'stat' data) fields at the beginning | |
610 | * of the 'struct inode' | |
611 | */ | |
1da177e4 | 612 | struct inode { |
44a7d7a8 | 613 | umode_t i_mode; |
3ddcd056 | 614 | unsigned short i_opflags; |
92361636 EB |
615 | kuid_t i_uid; |
616 | kgid_t i_gid; | |
3ddcd056 LT |
617 | unsigned int i_flags; |
618 | ||
619 | #ifdef CONFIG_FS_POSIX_ACL | |
620 | struct posix_acl *i_acl; | |
621 | struct posix_acl *i_default_acl; | |
622 | #endif | |
623 | ||
44a7d7a8 NP |
624 | const struct inode_operations *i_op; |
625 | struct super_block *i_sb; | |
3ddcd056 | 626 | struct address_space *i_mapping; |
44a7d7a8 | 627 | |
13e12d14 LT |
628 | #ifdef CONFIG_SECURITY |
629 | void *i_security; | |
630 | #endif | |
44a7d7a8 | 631 | |
3ddcd056 LT |
632 | /* Stat data, not accessed from path walking */ |
633 | unsigned long i_ino; | |
a78ef704 MS |
634 | /* |
635 | * Filesystems may only read i_nlink directly. They shall use the | |
636 | * following functions for modification: | |
637 | * | |
638 | * (set|clear|inc|drop)_nlink | |
639 | * inode_(inc|dec)_link_count | |
640 | */ | |
641 | union { | |
642 | const unsigned int i_nlink; | |
643 | unsigned int __i_nlink; | |
644 | }; | |
3ddcd056 | 645 | dev_t i_rdev; |
2f9d3df8 | 646 | loff_t i_size; |
95582b00 DD |
647 | struct timespec64 i_atime; |
648 | struct timespec64 i_mtime; | |
649 | struct timespec64 i_ctime; | |
6cdbb0ef TT |
650 | spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */ |
651 | unsigned short i_bytes; | |
69448867 AG |
652 | u8 i_blkbits; |
653 | u8 i_write_hint; | |
3ddcd056 LT |
654 | blkcnt_t i_blocks; |
655 | ||
656 | #ifdef __NEED_I_SIZE_ORDERED | |
657 | seqcount_t i_size_seqcount; | |
658 | #endif | |
659 | ||
660 | /* Misc */ | |
661 | unsigned long i_state; | |
9902af79 | 662 | struct rw_semaphore i_rwsem; |
13e12d14 | 663 | |
44a7d7a8 | 664 | unsigned long dirtied_when; /* jiffies of first dirtying */ |
a2f48706 | 665 | unsigned long dirtied_time_when; |
44a7d7a8 | 666 | |
1da177e4 | 667 | struct hlist_node i_hash; |
c7f54084 | 668 | struct list_head i_io_list; /* backing dev IO list */ |
52ebea74 TH |
669 | #ifdef CONFIG_CGROUP_WRITEBACK |
670 | struct bdi_writeback *i_wb; /* the associated cgroup wb */ | |
2a814908 TH |
671 | |
672 | /* foreign inode detection, see wbc_detach_inode() */ | |
673 | int i_wb_frn_winner; | |
674 | u16 i_wb_frn_avg_time; | |
675 | u16 i_wb_frn_history; | |
52ebea74 | 676 | #endif |
7ccf19a8 | 677 | struct list_head i_lru; /* inode LRU list */ |
1da177e4 | 678 | struct list_head i_sb_list; |
6c60d2b5 | 679 | struct list_head i_wb_list; /* backing dev writeback list */ |
fa0d7e3d | 680 | union { |
b3d9b7a3 | 681 | struct hlist_head i_dentry; |
fa0d7e3d NP |
682 | struct rcu_head i_rcu; |
683 | }; | |
f02a9ad1 | 684 | atomic64_t i_version; |
8019ad13 | 685 | atomic64_t i_sequence; /* see futex */ |
2f9d3df8 | 686 | atomic_t i_count; |
bd5fe6c5 | 687 | atomic_t i_dio_count; |
6cdbb0ef | 688 | atomic_t i_writecount; |
387e3746 | 689 | #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING) |
d984ea60 MZ |
690 | atomic_t i_readcount; /* struct files open RO */ |
691 | #endif | |
fdb0da89 AV |
692 | union { |
693 | const struct file_operations *i_fop; /* former ->i_op->default_file_ops */ | |
694 | void (*free_inode)(struct inode *); | |
695 | }; | |
4a075e39 | 696 | struct file_lock_context *i_flctx; |
1da177e4 | 697 | struct address_space i_data; |
1da177e4 | 698 | struct list_head i_devices; |
4c154168 TT |
699 | union { |
700 | struct pipe_inode_info *i_pipe; | |
577c4eb0 | 701 | struct cdev *i_cdev; |
61ba64fc | 702 | char *i_link; |
84e710da | 703 | unsigned i_dir_seq; |
4c154168 | 704 | }; |
1da177e4 LT |
705 | |
706 | __u32 i_generation; | |
707 | ||
3be25f49 EP |
708 | #ifdef CONFIG_FSNOTIFY |
709 | __u32 i_fsnotify_mask; /* all events this inode cares about */ | |
08991e83 | 710 | struct fsnotify_mark_connector __rcu *i_fsnotify_marks; |
0eeca283 RL |
711 | #endif |
712 | ||
643fa961 | 713 | #ifdef CONFIG_FS_ENCRYPTION |
0b81d077 JK |
714 | struct fscrypt_info *i_crypt_info; |
715 | #endif | |
716 | ||
5585f2af EB |
717 | #ifdef CONFIG_FS_VERITY |
718 | struct fsverity_info *i_verity_info; | |
719 | #endif | |
720 | ||
8e18e294 | 721 | void *i_private; /* fs or device private pointer */ |
3859a271 | 722 | } __randomize_layout; |
1da177e4 | 723 | |
50e17c00 DD |
724 | struct timespec64 timestamp_truncate(struct timespec64 t, struct inode *inode); |
725 | ||
93407472 FF |
726 | static inline unsigned int i_blocksize(const struct inode *node) |
727 | { | |
728 | return (1 << node->i_blkbits); | |
729 | } | |
730 | ||
1d3382cb AV |
731 | static inline int inode_unhashed(struct inode *inode) |
732 | { | |
733 | return hlist_unhashed(&inode->i_hash); | |
734 | } | |
735 | ||
5bef9151 AV |
736 | /* |
737 | * __mark_inode_dirty expects inodes to be hashed. Since we don't | |
738 | * want special inodes in the fileset inode space, we make them | |
739 | * appear hashed, but do not put on any lists. hlist_del() | |
740 | * will work fine and require no locking. | |
741 | */ | |
742 | static inline void inode_fake_hash(struct inode *inode) | |
743 | { | |
744 | hlist_add_fake(&inode->i_hash); | |
745 | } | |
746 | ||
f2eace23 IM |
747 | /* |
748 | * inode->i_mutex nesting subclasses for the lock validator: | |
749 | * | |
750 | * 0: the object of the current VFS operation | |
751 | * 1: parent | |
752 | * 2: child/target | |
40bd22c9 BF |
753 | * 3: xattr |
754 | * 4: second non-directory | |
d1b72cc6 MS |
755 | * 5: second parent (when locking independent directories in rename) |
756 | * | |
757 | * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two | |
40bd22c9 | 758 | * non-directories at once. |
f2eace23 IM |
759 | * |
760 | * The locking order between these classes is | |
d1b72cc6 | 761 | * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory |
f2eace23 IM |
762 | */ |
763 | enum inode_i_mutex_lock_class | |
764 | { | |
765 | I_MUTEX_NORMAL, | |
766 | I_MUTEX_PARENT, | |
767 | I_MUTEX_CHILD, | |
4df46240 | 768 | I_MUTEX_XATTR, |
d1b72cc6 MS |
769 | I_MUTEX_NONDIR2, |
770 | I_MUTEX_PARENT2, | |
f2eace23 IM |
771 | }; |
772 | ||
5955102c AV |
773 | static inline void inode_lock(struct inode *inode) |
774 | { | |
9902af79 | 775 | down_write(&inode->i_rwsem); |
5955102c AV |
776 | } |
777 | ||
778 | static inline void inode_unlock(struct inode *inode) | |
779 | { | |
9902af79 AV |
780 | up_write(&inode->i_rwsem); |
781 | } | |
782 | ||
783 | static inline void inode_lock_shared(struct inode *inode) | |
784 | { | |
785 | down_read(&inode->i_rwsem); | |
786 | } | |
787 | ||
788 | static inline void inode_unlock_shared(struct inode *inode) | |
789 | { | |
790 | up_read(&inode->i_rwsem); | |
5955102c AV |
791 | } |
792 | ||
793 | static inline int inode_trylock(struct inode *inode) | |
794 | { | |
9902af79 AV |
795 | return down_write_trylock(&inode->i_rwsem); |
796 | } | |
797 | ||
798 | static inline int inode_trylock_shared(struct inode *inode) | |
799 | { | |
800 | return down_read_trylock(&inode->i_rwsem); | |
5955102c AV |
801 | } |
802 | ||
803 | static inline int inode_is_locked(struct inode *inode) | |
804 | { | |
9902af79 | 805 | return rwsem_is_locked(&inode->i_rwsem); |
5955102c AV |
806 | } |
807 | ||
808 | static inline void inode_lock_nested(struct inode *inode, unsigned subclass) | |
809 | { | |
9902af79 | 810 | down_write_nested(&inode->i_rwsem, subclass); |
5955102c AV |
811 | } |
812 | ||
01c2e13d DW |
813 | static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass) |
814 | { | |
815 | down_read_nested(&inode->i_rwsem, subclass); | |
816 | } | |
817 | ||
730633f0 JK |
818 | static inline void filemap_invalidate_lock(struct address_space *mapping) |
819 | { | |
820 | down_write(&mapping->invalidate_lock); | |
821 | } | |
822 | ||
823 | static inline void filemap_invalidate_unlock(struct address_space *mapping) | |
824 | { | |
825 | up_write(&mapping->invalidate_lock); | |
826 | } | |
827 | ||
828 | static inline void filemap_invalidate_lock_shared(struct address_space *mapping) | |
829 | { | |
830 | down_read(&mapping->invalidate_lock); | |
831 | } | |
832 | ||
833 | static inline int filemap_invalidate_trylock_shared( | |
834 | struct address_space *mapping) | |
835 | { | |
836 | return down_read_trylock(&mapping->invalidate_lock); | |
837 | } | |
838 | ||
839 | static inline void filemap_invalidate_unlock_shared( | |
840 | struct address_space *mapping) | |
841 | { | |
842 | up_read(&mapping->invalidate_lock); | |
843 | } | |
844 | ||
375e289e BF |
845 | void lock_two_nondirectories(struct inode *, struct inode*); |
846 | void unlock_two_nondirectories(struct inode *, struct inode*); | |
847 | ||
7506ae6a JK |
848 | void filemap_invalidate_lock_two(struct address_space *mapping1, |
849 | struct address_space *mapping2); | |
850 | void filemap_invalidate_unlock_two(struct address_space *mapping1, | |
851 | struct address_space *mapping2); | |
852 | ||
853 | ||
1da177e4 LT |
854 | /* |
855 | * NOTE: in a 32bit arch with a preemptable kernel and | |
856 | * an UP compile the i_size_read/write must be atomic | |
857 | * with respect to the local cpu (unlike with preempt disabled), | |
858 | * but they don't need to be atomic with respect to other cpus like in | |
859 | * true SMP (so they need either to either locally disable irq around | |
860 | * the read or for example on x86 they can be still implemented as a | |
861 | * cmpxchg8b without the need of the lock prefix). For SMP compiles | |
862 | * and 64bit archs it makes no difference if preempt is enabled or not. | |
863 | */ | |
48ed214d | 864 | static inline loff_t i_size_read(const struct inode *inode) |
1da177e4 LT |
865 | { |
866 | #if BITS_PER_LONG==32 && defined(CONFIG_SMP) | |
867 | loff_t i_size; | |
868 | unsigned int seq; | |
869 | ||
870 | do { | |
871 | seq = read_seqcount_begin(&inode->i_size_seqcount); | |
872 | i_size = inode->i_size; | |
873 | } while (read_seqcount_retry(&inode->i_size_seqcount, seq)); | |
874 | return i_size; | |
2496396f | 875 | #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION) |
1da177e4 LT |
876 | loff_t i_size; |
877 | ||
878 | preempt_disable(); | |
879 | i_size = inode->i_size; | |
880 | preempt_enable(); | |
881 | return i_size; | |
882 | #else | |
883 | return inode->i_size; | |
884 | #endif | |
885 | } | |
886 | ||
7762f5a0 MS |
887 | /* |
888 | * NOTE: unlike i_size_read(), i_size_write() does need locking around it | |
889 | * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount | |
890 | * can be lost, resulting in subsequent i_size_read() calls spinning forever. | |
891 | */ | |
1da177e4 LT |
892 | static inline void i_size_write(struct inode *inode, loff_t i_size) |
893 | { | |
894 | #if BITS_PER_LONG==32 && defined(CONFIG_SMP) | |
74e3d1e1 | 895 | preempt_disable(); |
1da177e4 LT |
896 | write_seqcount_begin(&inode->i_size_seqcount); |
897 | inode->i_size = i_size; | |
898 | write_seqcount_end(&inode->i_size_seqcount); | |
74e3d1e1 | 899 | preempt_enable(); |
2496396f | 900 | #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION) |
1da177e4 LT |
901 | preempt_disable(); |
902 | inode->i_size = i_size; | |
903 | preempt_enable(); | |
904 | #else | |
905 | inode->i_size = i_size; | |
906 | #endif | |
907 | } | |
908 | ||
48ed214d | 909 | static inline unsigned iminor(const struct inode *inode) |
1da177e4 LT |
910 | { |
911 | return MINOR(inode->i_rdev); | |
912 | } | |
913 | ||
48ed214d | 914 | static inline unsigned imajor(const struct inode *inode) |
1da177e4 LT |
915 | { |
916 | return MAJOR(inode->i_rdev); | |
917 | } | |
918 | ||
1da177e4 LT |
919 | struct fown_struct { |
920 | rwlock_t lock; /* protects pid, uid, euid fields */ | |
609d7fa9 EB |
921 | struct pid *pid; /* pid or -pgrp where SIGIO should be sent */ |
922 | enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */ | |
92361636 | 923 | kuid_t uid, euid; /* uid/euid of process setting the owner */ |
1da177e4 LT |
924 | int signum; /* posix.1b rt signal to be delivered on IO */ |
925 | }; | |
926 | ||
c790fbf2 MWO |
927 | /** |
928 | * struct file_ra_state - Track a file's readahead state. | |
929 | * @start: Where the most recent readahead started. | |
930 | * @size: Number of pages read in the most recent readahead. | |
84dacdbd N |
931 | * @async_size: Numer of pages that were/are not needed immediately |
932 | * and so were/are genuinely "ahead". Start next readahead when | |
933 | * the first of these pages is accessed. | |
934 | * @ra_pages: Maximum size of a readahead request, copied from the bdi. | |
c790fbf2 MWO |
935 | * @mmap_miss: How many mmap accesses missed in the page cache. |
936 | * @prev_pos: The last byte in the most recent read request. | |
84dacdbd N |
937 | * |
938 | * When this structure is passed to ->readahead(), the "most recent" | |
939 | * readahead means the current readahead. | |
1da177e4 LT |
940 | */ |
941 | struct file_ra_state { | |
c790fbf2 MWO |
942 | pgoff_t start; |
943 | unsigned int size; | |
944 | unsigned int async_size; | |
945 | unsigned int ra_pages; | |
946 | unsigned int mmap_miss; | |
947 | loff_t prev_pos; | |
1da177e4 | 948 | }; |
1da177e4 | 949 | |
5ce1110b FW |
950 | /* |
951 | * Check if @index falls in the readahead windows. | |
952 | */ | |
953 | static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index) | |
954 | { | |
f9acc8c7 FW |
955 | return (index >= ra->start && |
956 | index < ra->start + ra->size); | |
5ce1110b FW |
957 | } |
958 | ||
1da177e4 | 959 | struct file { |
2f512016 | 960 | union { |
e87f2c26 AV |
961 | struct llist_node f_llist; |
962 | struct rcu_head f_rcuhead; | |
164f4064 | 963 | unsigned int f_iocb_flags; |
e87f2c26 | 964 | }; |
0f7fc9e4 | 965 | struct path f_path; |
dd37978c | 966 | struct inode *f_inode; /* cached value */ |
99ac48f5 | 967 | const struct file_operations *f_op; |
ef3d0fd2 AK |
968 | |
969 | /* | |
319c1517 | 970 | * Protects f_ep, f_flags. |
ef3d0fd2 AK |
971 | * Must not be taken from IRQ context. |
972 | */ | |
973 | spinlock_t f_lock; | |
516e0cc5 | 974 | atomic_long_t f_count; |
1da177e4 | 975 | unsigned int f_flags; |
aeb5d727 | 976 | fmode_t f_mode; |
9c225f26 | 977 | struct mutex f_pos_lock; |
1da177e4 LT |
978 | loff_t f_pos; |
979 | struct fown_struct f_owner; | |
d76b0d9b | 980 | const struct cred *f_cred; |
1da177e4 LT |
981 | struct file_ra_state f_ra; |
982 | ||
2b47c361 | 983 | u64 f_version; |
50462062 | 984 | #ifdef CONFIG_SECURITY |
1da177e4 | 985 | void *f_security; |
50462062 | 986 | #endif |
1da177e4 LT |
987 | /* needed for tty driver, and maybe others */ |
988 | void *private_data; | |
989 | ||
990 | #ifdef CONFIG_EPOLL | |
991 | /* Used by fs/eventpoll.c to link all the hooks to this file */ | |
319c1517 | 992 | struct hlist_head *f_ep; |
1da177e4 LT |
993 | #endif /* #ifdef CONFIG_EPOLL */ |
994 | struct address_space *f_mapping; | |
5660e13d | 995 | errseq_t f_wb_err; |
735e4ae5 | 996 | errseq_t f_sb_err; /* for syncfs */ |
3859a271 KC |
997 | } __randomize_layout |
998 | __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */ | |
1da177e4 | 999 | |
990d6c2d AK |
1000 | struct file_handle { |
1001 | __u32 handle_bytes; | |
1002 | int handle_type; | |
1003 | /* file identifier */ | |
9d82973e | 1004 | unsigned char f_handle[]; |
990d6c2d AK |
1005 | }; |
1006 | ||
cb0942b8 AV |
1007 | static inline struct file *get_file(struct file *f) |
1008 | { | |
1009 | atomic_long_inc(&f->f_count); | |
1010 | return f; | |
1011 | } | |
81132a39 | 1012 | #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count) |
516e0cc5 | 1013 | #define file_count(x) atomic_long_read(&(x)->f_count) |
1da177e4 LT |
1014 | |
1015 | #define MAX_NON_LFS ((1UL<<31) - 1) | |
1016 | ||
1017 | /* Page cache limit. The filesystems should put that into their s_maxbytes | |
1018 | limits, otherwise bad things can happen in VM. */ | |
1019 | #if BITS_PER_LONG==32 | |
0cc3b0ec | 1020 | #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT) |
1da177e4 | 1021 | #elif BITS_PER_LONG==64 |
0cc3b0ec | 1022 | #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX) |
1da177e4 LT |
1023 | #endif |
1024 | ||
7ca76311 | 1025 | /* legacy typedef, should eventually be removed */ |
17fa388d | 1026 | typedef void *fl_owner_t; |
1da177e4 | 1027 | |
a7231a97 JL |
1028 | struct file_lock; |
1029 | ||
1da177e4 LT |
1030 | /* The following constant reflects the upper bound of the file/locking space */ |
1031 | #ifndef OFFSET_MAX | |
ea258f15 ZL |
1032 | #define OFFSET_MAX type_max(loff_t) |
1033 | #define OFFT_OFFSET_MAX type_max(off_t) | |
1da177e4 LT |
1034 | #endif |
1035 | ||
bfcd17a6 TP |
1036 | extern void send_sigio(struct fown_struct *fown, int fd, int band); |
1037 | ||
ee296d7c JL |
1038 | static inline struct inode *file_inode(const struct file *f) |
1039 | { | |
1040 | return f->f_inode; | |
1041 | } | |
1042 | ||
d101a125 MS |
1043 | static inline struct dentry *file_dentry(const struct file *file) |
1044 | { | |
fb16043b | 1045 | return d_real(file->f_path.dentry, file_inode(file)); |
d101a125 MS |
1046 | } |
1047 | ||
1da177e4 | 1048 | struct fasync_struct { |
7a107c0f | 1049 | rwlock_t fa_lock; |
989a2979 ED |
1050 | int magic; |
1051 | int fa_fd; | |
1052 | struct fasync_struct *fa_next; /* singly linked list */ | |
1053 | struct file *fa_file; | |
1054 | struct rcu_head fa_rcu; | |
1da177e4 LT |
1055 | }; |
1056 | ||
1057 | #define FASYNC_MAGIC 0x4601 | |
1058 | ||
1059 | /* SMP safe fasync helpers: */ | |
1060 | extern int fasync_helper(int, struct file *, int, struct fasync_struct **); | |
f7347ce4 LT |
1061 | extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *); |
1062 | extern int fasync_remove_entry(struct file *, struct fasync_struct **); | |
1063 | extern struct fasync_struct *fasync_alloc(void); | |
1064 | extern void fasync_free(struct fasync_struct *); | |
1065 | ||
1da177e4 LT |
1066 | /* can be called from interrupts */ |
1067 | extern void kill_fasync(struct fasync_struct **, int, int); | |
1da177e4 | 1068 | |
e0b93edd | 1069 | extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force); |
393cc3f5 | 1070 | extern int f_setown(struct file *filp, unsigned long arg, int force); |
1da177e4 | 1071 | extern void f_delown(struct file *filp); |
609d7fa9 | 1072 | extern pid_t f_getown(struct file *filp); |
1da177e4 LT |
1073 | extern int send_sigurg(struct fown_struct *fown); |
1074 | ||
e462ec50 DH |
1075 | /* |
1076 | * sb->s_flags. Note that these mirror the equivalent MS_* flags where | |
1077 | * represented in both. | |
1078 | */ | |
1079 | #define SB_RDONLY 1 /* Mount read-only */ | |
1080 | #define SB_NOSUID 2 /* Ignore suid and sgid bits */ | |
1081 | #define SB_NODEV 4 /* Disallow access to device special files */ | |
1082 | #define SB_NOEXEC 8 /* Disallow program execution */ | |
1083 | #define SB_SYNCHRONOUS 16 /* Writes are synced at once */ | |
1084 | #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */ | |
1085 | #define SB_DIRSYNC 128 /* Directory modifications are synchronous */ | |
1086 | #define SB_NOATIME 1024 /* Do not update access times. */ | |
1087 | #define SB_NODIRATIME 2048 /* Do not update directory access times */ | |
1088 | #define SB_SILENT 32768 | |
1089 | #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */ | |
457e7a13 | 1090 | #define SB_INLINECRYPT (1<<17) /* Use blk-crypto for encrypted files */ |
e462ec50 DH |
1091 | #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */ |
1092 | #define SB_I_VERSION (1<<23) /* Update inode I_version field */ | |
1093 | #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */ | |
1094 | ||
1095 | /* These sb flags are internal to the kernel */ | |
1096 | #define SB_SUBMOUNT (1<<26) | |
8d0347f6 | 1097 | #define SB_FORCE (1<<27) |
e462ec50 DH |
1098 | #define SB_NOSEC (1<<28) |
1099 | #define SB_BORN (1<<29) | |
1100 | #define SB_ACTIVE (1<<30) | |
1101 | #define SB_NOUSER (1<<31) | |
1102 | ||
c843843e DR |
1103 | /* These flags relate to encoding and casefolding */ |
1104 | #define SB_ENC_STRICT_MODE_FL (1 << 0) | |
1105 | ||
1106 | #define sb_has_strict_encoding(sb) \ | |
1107 | (sb->s_encoding_flags & SB_ENC_STRICT_MODE_FL) | |
1108 | ||
1da177e4 LT |
1109 | /* |
1110 | * Umount options | |
1111 | */ | |
1112 | ||
1113 | #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */ | |
1114 | #define MNT_DETACH 0x00000002 /* Just detach from the tree */ | |
1115 | #define MNT_EXPIRE 0x00000004 /* Mark for expiry */ | |
db1f05bb MS |
1116 | #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */ |
1117 | #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */ | |
1da177e4 | 1118 | |
46b15caa TH |
1119 | /* sb->s_iflags */ |
1120 | #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */ | |
90f8572b | 1121 | #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */ |
a2982cc9 | 1122 | #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */ |
1cb039f3 | 1123 | #define SB_I_STABLE_WRITES 0x00000008 /* don't modify blks until WB is done */ |
1da177e4 | 1124 | |
8654df4e EB |
1125 | /* sb->s_iflags to limit user namespace mounts */ |
1126 | #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */ | |
57b56ac6 MZ |
1127 | #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020 |
1128 | #define SB_I_UNTRUSTED_MOUNTER 0x00000040 | |
1da177e4 | 1129 | |
32b1924b | 1130 | #define SB_I_SKIP_SYNC 0x00000100 /* Skip superblock at global sync */ |
0b3ea092 | 1131 | #define SB_I_PERSB_BDI 0x00000200 /* has a per-sb bdi */ |
a128b054 | 1132 | #define SB_I_TS_EXPIRY_WARNED 0x00000400 /* warned about timestamp range expiry */ |
04b94071 | 1133 | #define SB_I_RETIRED 0x00000800 /* superblock shouldn't be reused */ |
32b1924b | 1134 | |
5accdf82 JK |
1135 | /* Possible states of 'frozen' field */ |
1136 | enum { | |
1137 | SB_UNFROZEN = 0, /* FS is unfrozen */ | |
1138 | SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */ | |
5accdf82 JK |
1139 | SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */ |
1140 | SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop | |
1141 | * internal threads if needed) */ | |
1142 | SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */ | |
1143 | }; | |
1144 | ||
1145 | #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1) | |
1146 | ||
1147 | struct sb_writers { | |
8129ed29 | 1148 | int frozen; /* Is sb frozen? */ |
60b49885 | 1149 | wait_queue_head_t wait_unfrozen; /* wait for thaw */ |
8129ed29 | 1150 | struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS]; |
5accdf82 JK |
1151 | }; |
1152 | ||
1da177e4 LT |
1153 | struct super_block { |
1154 | struct list_head s_list; /* Keep this first */ | |
1155 | dev_t s_dev; /* search index; _not_ kdev_t */ | |
270ba5f7 RK |
1156 | unsigned char s_blocksize_bits; |
1157 | unsigned long s_blocksize; | |
42cb56ae | 1158 | loff_t s_maxbytes; /* Max file size */ |
1da177e4 | 1159 | struct file_system_type *s_type; |
ee9b6d61 | 1160 | const struct super_operations *s_op; |
61e225dc | 1161 | const struct dquot_operations *dq_op; |
0d54b217 | 1162 | const struct quotactl_ops *s_qcop; |
39655164 | 1163 | const struct export_operations *s_export_op; |
1da177e4 | 1164 | unsigned long s_flags; |
46b15caa | 1165 | unsigned long s_iflags; /* internal SB_I_* flags */ |
1da177e4 LT |
1166 | unsigned long s_magic; |
1167 | struct dentry *s_root; | |
1168 | struct rw_semaphore s_umount; | |
1da177e4 | 1169 | int s_count; |
1da177e4 | 1170 | atomic_t s_active; |
50462062 | 1171 | #ifdef CONFIG_SECURITY |
1da177e4 | 1172 | void *s_security; |
50462062 | 1173 | #endif |
bb435453 | 1174 | const struct xattr_handler **s_xattr; |
643fa961 | 1175 | #ifdef CONFIG_FS_ENCRYPTION |
0b81d077 | 1176 | const struct fscrypt_operations *s_cop; |
d7e7b9af | 1177 | struct fscrypt_keyring *s_master_keys; /* master crypto keys in use */ |
5585f2af EB |
1178 | #endif |
1179 | #ifdef CONFIG_FS_VERITY | |
1180 | const struct fsverity_operations *s_vop; | |
c843843e | 1181 | #endif |
5298d4bf | 1182 | #if IS_ENABLED(CONFIG_UNICODE) |
c843843e DR |
1183 | struct unicode_map *s_encoding; |
1184 | __u16 s_encoding_flags; | |
bbbc3fb6 | 1185 | #endif |
f1ee6162 | 1186 | struct hlist_bl_head s_roots; /* alternate root dentries for NFS */ |
39f7c4db | 1187 | struct list_head s_mounts; /* list of mounts; _not_ for fs use */ |
1da177e4 | 1188 | struct block_device *s_bdev; |
32a88aa1 | 1189 | struct backing_dev_info *s_bdi; |
acaebfd8 | 1190 | struct mtd_info *s_mtd; |
a5166169 | 1191 | struct hlist_node s_instances; |
2c5f648a | 1192 | unsigned int s_quota_types; /* Bitmask of supported quota types */ |
1da177e4 LT |
1193 | struct quota_info s_dquot; /* Diskquota specific options */ |
1194 | ||
5accdf82 | 1195 | struct sb_writers s_writers; |
1da177e4 | 1196 | |
99c228a9 AG |
1197 | /* |
1198 | * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and | |
1199 | * s_fsnotify_marks together for cache efficiency. They are frequently | |
1200 | * accessed and rarely modified. | |
1201 | */ | |
1202 | void *s_fs_info; /* Filesystem private info */ | |
1203 | ||
1204 | /* Granularity of c/m/atime in ns (cannot be worse than a second) */ | |
1205 | u32 s_time_gran; | |
188d20bc DD |
1206 | /* Time limits for c/m/atime in seconds */ |
1207 | time64_t s_time_min; | |
1208 | time64_t s_time_max; | |
99c228a9 AG |
1209 | #ifdef CONFIG_FSNOTIFY |
1210 | __u32 s_fsnotify_mask; | |
1211 | struct fsnotify_mark_connector __rcu *s_fsnotify_marks; | |
1212 | #endif | |
1213 | ||
85787090 CH |
1214 | char s_id[32]; /* Informational name */ |
1215 | uuid_t s_uuid; /* UUID */ | |
1da177e4 | 1216 | |
8de52778 | 1217 | unsigned int s_max_links; |
30c40d2c | 1218 | fmode_t s_mode; |
1da177e4 LT |
1219 | |
1220 | /* | |
1221 | * The next field is for VFS *only*. No filesystems have any business | |
1222 | * even looking at it. You had been warned. | |
1223 | */ | |
a11f3a05 | 1224 | struct mutex s_vfs_rename_mutex; /* Kludge */ |
1da177e4 | 1225 | |
79c0b2df MS |
1226 | /* |
1227 | * Filesystem subtype. If non-empty the filesystem type field | |
1228 | * in /proc/mounts will be "type.subtype" | |
1229 | */ | |
a0c9a8b8 | 1230 | const char *s_subtype; |
b3b304a2 | 1231 | |
c8aebb0c | 1232 | const struct dentry_operations *s_d_op; /* default d_op for dentries */ |
9fdfdcf1 | 1233 | |
b0d40c92 | 1234 | struct shrinker s_shrink; /* per-sb shrinker handle */ |
4ed5e82f | 1235 | |
7ada4db8 MS |
1236 | /* Number of inodes with nlink == 0 but still referenced */ |
1237 | atomic_long_t s_remove_count; | |
1238 | ||
ec44610f AG |
1239 | /* |
1240 | * Number of inode/mount/sb objects that are being watched, note that | |
1241 | * inodes objects are currently double-accounted. | |
1242 | */ | |
1243 | atomic_long_t s_fsnotify_connectors; | |
721fb6fb | 1244 | |
4ed5e82f MS |
1245 | /* Being remounted read-only */ |
1246 | int s_readonly_remount; | |
7b7a8665 | 1247 | |
735e4ae5 JL |
1248 | /* per-sb errseq_t for reporting writeback errors via syncfs */ |
1249 | errseq_t s_wb_err; | |
1250 | ||
7b7a8665 CH |
1251 | /* AIO completions deferred from interrupt context */ |
1252 | struct workqueue_struct *s_dio_done_wq; | |
215752fc | 1253 | struct hlist_head s_pins; |
f6041567 | 1254 | |
6e4eab57 EB |
1255 | /* |
1256 | * Owning user namespace and default context in which to | |
1257 | * interpret filesystem uids, gids, quotas, device nodes, | |
1258 | * xattrs and security labels. | |
1259 | */ | |
1260 | struct user_namespace *s_user_ns; | |
1261 | ||
f6041567 | 1262 | /* |
7d10f70f WL |
1263 | * The list_lru structure is essentially just a pointer to a table |
1264 | * of per-node lru lists, each of which has its own spinlock. | |
1265 | * There is no need to put them into separate cachelines. | |
f6041567 | 1266 | */ |
7d10f70f WL |
1267 | struct list_lru s_dentry_lru; |
1268 | struct list_lru s_inode_lru; | |
e2fec7c3 | 1269 | struct rcu_head rcu; |
853b39a7 | 1270 | struct work_struct destroy_work; |
69c433ed | 1271 | |
e97fedb9 | 1272 | struct mutex s_sync_lock; /* sync serialisation lock */ |
69c433ed MS |
1273 | |
1274 | /* | |
1275 | * Indicates how deep in a filesystem stack this SB is | |
1276 | */ | |
1277 | int s_stack_depth; | |
74278da9 DC |
1278 | |
1279 | /* s_inode_list_lock protects s_inodes */ | |
1280 | spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp; | |
1281 | struct list_head s_inodes; /* all inodes */ | |
6c60d2b5 DC |
1282 | |
1283 | spinlock_t s_inode_wblist_lock; | |
1284 | struct list_head s_inodes_wb; /* writeback inodes */ | |
3859a271 | 1285 | } __randomize_layout; |
1da177e4 | 1286 | |
a1ec9040 CB |
1287 | static inline struct user_namespace *i_user_ns(const struct inode *inode) |
1288 | { | |
1289 | return inode->i_sb->s_user_ns; | |
1290 | } | |
1291 | ||
81754357 SF |
1292 | /* Helper functions so that in most cases filesystems will |
1293 | * not need to deal directly with kuid_t and kgid_t and can | |
1294 | * instead deal with the raw numeric values that are stored | |
1295 | * in the filesystem. | |
1296 | */ | |
1297 | static inline uid_t i_uid_read(const struct inode *inode) | |
1298 | { | |
a1ec9040 | 1299 | return from_kuid(i_user_ns(inode), inode->i_uid); |
81754357 SF |
1300 | } |
1301 | ||
1302 | static inline gid_t i_gid_read(const struct inode *inode) | |
1303 | { | |
a1ec9040 | 1304 | return from_kgid(i_user_ns(inode), inode->i_gid); |
81754357 SF |
1305 | } |
1306 | ||
1307 | static inline void i_uid_write(struct inode *inode, uid_t uid) | |
1308 | { | |
a1ec9040 | 1309 | inode->i_uid = make_kuid(i_user_ns(inode), uid); |
81754357 SF |
1310 | } |
1311 | ||
1312 | static inline void i_gid_write(struct inode *inode, gid_t gid) | |
1313 | { | |
a1ec9040 | 1314 | inode->i_gid = make_kgid(i_user_ns(inode), gid); |
81754357 SF |
1315 | } |
1316 | ||
234a3113 | 1317 | /** |
e67fe633 CB |
1318 | * i_uid_into_vfsuid - map an inode's i_uid down according to an idmapping |
1319 | * @idmap: idmap of the mount the inode was found from | |
234a3113 CB |
1320 | * @inode: inode to map |
1321 | * | |
e67fe633 | 1322 | * Return: whe inode's i_uid mapped down according to @idmap. |
234a3113 CB |
1323 | * If the inode's i_uid has no mapping INVALID_VFSUID is returned. |
1324 | */ | |
e67fe633 | 1325 | static inline vfsuid_t i_uid_into_vfsuid(struct mnt_idmap *idmap, |
234a3113 CB |
1326 | const struct inode *inode) |
1327 | { | |
4d7ca409 | 1328 | return make_vfsuid(idmap, i_user_ns(inode), inode->i_uid); |
e6c9a714 CB |
1329 | } |
1330 | ||
1f36146a CB |
1331 | /** |
1332 | * i_uid_needs_update - check whether inode's i_uid needs to be updated | |
0dbe12f2 | 1333 | * @idmap: idmap of the mount the inode was found from |
1f36146a CB |
1334 | * @attr: the new attributes of @inode |
1335 | * @inode: the inode to update | |
1336 | * | |
1337 | * Check whether the $inode's i_uid field needs to be updated taking idmapped | |
1338 | * mounts into account if the filesystem supports it. | |
1339 | * | |
1340 | * Return: true if @inode's i_uid field needs to be updated, false if not. | |
1341 | */ | |
0dbe12f2 | 1342 | static inline bool i_uid_needs_update(struct mnt_idmap *idmap, |
1f36146a CB |
1343 | const struct iattr *attr, |
1344 | const struct inode *inode) | |
1345 | { | |
1346 | return ((attr->ia_valid & ATTR_UID) && | |
1347 | !vfsuid_eq(attr->ia_vfsuid, | |
e67fe633 | 1348 | i_uid_into_vfsuid(idmap, inode))); |
1f36146a CB |
1349 | } |
1350 | ||
1351 | /** | |
1352 | * i_uid_update - update @inode's i_uid field | |
0dbe12f2 | 1353 | * @idmap: idmap of the mount the inode was found from |
1f36146a CB |
1354 | * @attr: the new attributes of @inode |
1355 | * @inode: the inode to update | |
1356 | * | |
1357 | * Safely update @inode's i_uid field translating the vfsuid of any idmapped | |
1358 | * mount into the filesystem kuid. | |
1359 | */ | |
0dbe12f2 | 1360 | static inline void i_uid_update(struct mnt_idmap *idmap, |
1f36146a CB |
1361 | const struct iattr *attr, |
1362 | struct inode *inode) | |
1363 | { | |
1364 | if (attr->ia_valid & ATTR_UID) | |
4d7ca409 | 1365 | inode->i_uid = from_vfsuid(idmap, i_user_ns(inode), |
1f36146a CB |
1366 | attr->ia_vfsuid); |
1367 | } | |
1368 | ||
234a3113 | 1369 | /** |
e67fe633 CB |
1370 | * i_gid_into_vfsgid - map an inode's i_gid down according to an idmapping |
1371 | * @idmap: idmap of the mount the inode was found from | |
234a3113 CB |
1372 | * @inode: inode to map |
1373 | * | |
e67fe633 | 1374 | * Return: the inode's i_gid mapped down according to @idmap. |
234a3113 CB |
1375 | * If the inode's i_gid has no mapping INVALID_VFSGID is returned. |
1376 | */ | |
e67fe633 | 1377 | static inline vfsgid_t i_gid_into_vfsgid(struct mnt_idmap *idmap, |
234a3113 CB |
1378 | const struct inode *inode) |
1379 | { | |
4d7ca409 | 1380 | return make_vfsgid(idmap, i_user_ns(inode), inode->i_gid); |
e6c9a714 CB |
1381 | } |
1382 | ||
1f36146a CB |
1383 | /** |
1384 | * i_gid_needs_update - check whether inode's i_gid needs to be updated | |
0dbe12f2 | 1385 | * @idmap: idmap of the mount the inode was found from |
1f36146a CB |
1386 | * @attr: the new attributes of @inode |
1387 | * @inode: the inode to update | |
1388 | * | |
1389 | * Check whether the $inode's i_gid field needs to be updated taking idmapped | |
1390 | * mounts into account if the filesystem supports it. | |
1391 | * | |
1392 | * Return: true if @inode's i_gid field needs to be updated, false if not. | |
1393 | */ | |
0dbe12f2 | 1394 | static inline bool i_gid_needs_update(struct mnt_idmap *idmap, |
1f36146a CB |
1395 | const struct iattr *attr, |
1396 | const struct inode *inode) | |
1397 | { | |
1398 | return ((attr->ia_valid & ATTR_GID) && | |
1399 | !vfsgid_eq(attr->ia_vfsgid, | |
e67fe633 | 1400 | i_gid_into_vfsgid(idmap, inode))); |
1f36146a CB |
1401 | } |
1402 | ||
1403 | /** | |
1404 | * i_gid_update - update @inode's i_gid field | |
0dbe12f2 | 1405 | * @idmap: idmap of the mount the inode was found from |
1f36146a CB |
1406 | * @attr: the new attributes of @inode |
1407 | * @inode: the inode to update | |
1408 | * | |
1409 | * Safely update @inode's i_gid field translating the vfsgid of any idmapped | |
1410 | * mount into the filesystem kgid. | |
1411 | */ | |
0dbe12f2 | 1412 | static inline void i_gid_update(struct mnt_idmap *idmap, |
1f36146a CB |
1413 | const struct iattr *attr, |
1414 | struct inode *inode) | |
1415 | { | |
1416 | if (attr->ia_valid & ATTR_GID) | |
4d7ca409 | 1417 | inode->i_gid = from_vfsgid(idmap, i_user_ns(inode), |
1f36146a CB |
1418 | attr->ia_vfsgid); |
1419 | } | |
1420 | ||
db998553 CB |
1421 | /** |
1422 | * inode_fsuid_set - initialize inode's i_uid field with callers fsuid | |
1423 | * @inode: inode to initialize | |
c14329d3 | 1424 | * @idmap: idmap of the mount the inode was found from |
db998553 CB |
1425 | * |
1426 | * Initialize the i_uid field of @inode. If the inode was found/created via | |
c14329d3 | 1427 | * an idmapped mount map the caller's fsuid according to @idmap. |
db998553 CB |
1428 | */ |
1429 | static inline void inode_fsuid_set(struct inode *inode, | |
c14329d3 | 1430 | struct mnt_idmap *idmap) |
db998553 | 1431 | { |
c14329d3 | 1432 | inode->i_uid = mapped_fsuid(idmap, i_user_ns(inode)); |
db998553 CB |
1433 | } |
1434 | ||
1435 | /** | |
1436 | * inode_fsgid_set - initialize inode's i_gid field with callers fsgid | |
1437 | * @inode: inode to initialize | |
c14329d3 | 1438 | * @idmap: idmap of the mount the inode was found from |
db998553 CB |
1439 | * |
1440 | * Initialize the i_gid field of @inode. If the inode was found/created via | |
c14329d3 | 1441 | * an idmapped mount map the caller's fsgid according to @idmap. |
db998553 CB |
1442 | */ |
1443 | static inline void inode_fsgid_set(struct inode *inode, | |
c14329d3 | 1444 | struct mnt_idmap *idmap) |
db998553 | 1445 | { |
c14329d3 | 1446 | inode->i_gid = mapped_fsgid(idmap, i_user_ns(inode)); |
db998553 CB |
1447 | } |
1448 | ||
8e538913 CB |
1449 | /** |
1450 | * fsuidgid_has_mapping() - check whether caller's fsuid/fsgid is mapped | |
1451 | * @sb: the superblock we want a mapping in | |
4609e1f1 | 1452 | * @idmap: idmap of the relevant mount |
8e538913 CB |
1453 | * |
1454 | * Check whether the caller's fsuid and fsgid have a valid mapping in the | |
1455 | * s_user_ns of the superblock @sb. If the caller is on an idmapped mount map | |
4609e1f1 | 1456 | * the caller's fsuid and fsgid according to the @idmap first. |
8e538913 CB |
1457 | * |
1458 | * Return: true if fsuid and fsgid is mapped, false if not. | |
1459 | */ | |
1460 | static inline bool fsuidgid_has_mapping(struct super_block *sb, | |
4609e1f1 | 1461 | struct mnt_idmap *idmap) |
8e538913 | 1462 | { |
476860b3 CB |
1463 | struct user_namespace *fs_userns = sb->s_user_ns; |
1464 | kuid_t kuid; | |
1465 | kgid_t kgid; | |
1466 | ||
c14329d3 | 1467 | kuid = mapped_fsuid(idmap, fs_userns); |
476860b3 CB |
1468 | if (!uid_valid(kuid)) |
1469 | return false; | |
c14329d3 | 1470 | kgid = mapped_fsgid(idmap, fs_userns); |
476860b3 CB |
1471 | if (!gid_valid(kgid)) |
1472 | return false; | |
1473 | return kuid_has_mapping(fs_userns, kuid) && | |
1474 | kgid_has_mapping(fs_userns, kgid); | |
8e538913 CB |
1475 | } |
1476 | ||
95582b00 | 1477 | extern struct timespec64 current_time(struct inode *inode); |
1da177e4 LT |
1478 | |
1479 | /* | |
1480 | * Snapshotting support. | |
1481 | */ | |
1da177e4 | 1482 | |
9b852342 DW |
1483 | /* |
1484 | * These are internal functions, please use sb_start_{write,pagefault,intwrite} | |
1485 | * instead. | |
1486 | */ | |
1487 | static inline void __sb_end_write(struct super_block *sb, int level) | |
1488 | { | |
1489 | percpu_up_read(sb->s_writers.rw_sem + level-1); | |
1490 | } | |
1491 | ||
1492 | static inline void __sb_start_write(struct super_block *sb, int level) | |
1493 | { | |
1494 | percpu_down_read(sb->s_writers.rw_sem + level - 1); | |
1495 | } | |
1496 | ||
1497 | static inline bool __sb_start_write_trylock(struct super_block *sb, int level) | |
1498 | { | |
1499 | return percpu_down_read_trylock(sb->s_writers.rw_sem + level - 1); | |
1500 | } | |
5accdf82 | 1501 | |
bee9182d | 1502 | #define __sb_writers_acquired(sb, lev) \ |
8129ed29 | 1503 | percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_) |
bee9182d | 1504 | #define __sb_writers_release(sb, lev) \ |
8129ed29 | 1505 | percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_) |
bee9182d | 1506 | |
7f8d12ea NA |
1507 | static inline bool sb_write_started(const struct super_block *sb) |
1508 | { | |
1509 | return lockdep_is_held_type(sb->s_writers.rw_sem + SB_FREEZE_WRITE - 1, 1); | |
1510 | } | |
1511 | ||
5accdf82 JK |
1512 | /** |
1513 | * sb_end_write - drop write access to a superblock | |
1514 | * @sb: the super we wrote to | |
1515 | * | |
1516 | * Decrement number of writers to the filesystem. Wake up possible waiters | |
1517 | * wanting to freeze the filesystem. | |
1518 | */ | |
1519 | static inline void sb_end_write(struct super_block *sb) | |
1520 | { | |
1521 | __sb_end_write(sb, SB_FREEZE_WRITE); | |
1522 | } | |
1523 | ||
1524 | /** | |
1525 | * sb_end_pagefault - drop write access to a superblock from a page fault | |
1526 | * @sb: the super we wrote to | |
1527 | * | |
1528 | * Decrement number of processes handling write page fault to the filesystem. | |
1529 | * Wake up possible waiters wanting to freeze the filesystem. | |
1530 | */ | |
1531 | static inline void sb_end_pagefault(struct super_block *sb) | |
1532 | { | |
1533 | __sb_end_write(sb, SB_FREEZE_PAGEFAULT); | |
1534 | } | |
1535 | ||
1536 | /** | |
1537 | * sb_end_intwrite - drop write access to a superblock for internal fs purposes | |
1538 | * @sb: the super we wrote to | |
1539 | * | |
1540 | * Decrement fs-internal number of writers to the filesystem. Wake up possible | |
1541 | * waiters wanting to freeze the filesystem. | |
1542 | */ | |
1543 | static inline void sb_end_intwrite(struct super_block *sb) | |
1544 | { | |
1545 | __sb_end_write(sb, SB_FREEZE_FS); | |
1546 | } | |
1547 | ||
1548 | /** | |
1549 | * sb_start_write - get write access to a superblock | |
1550 | * @sb: the super we write to | |
1551 | * | |
1552 | * When a process wants to write data or metadata to a file system (i.e. dirty | |
1553 | * a page or an inode), it should embed the operation in a sb_start_write() - | |
1554 | * sb_end_write() pair to get exclusion against file system freezing. This | |
1555 | * function increments number of writers preventing freezing. If the file | |
1556 | * system is already frozen, the function waits until the file system is | |
1557 | * thawed. | |
1558 | * | |
1559 | * Since freeze protection behaves as a lock, users have to preserve | |
1560 | * ordering of freeze protection and other filesystem locks. Generally, | |
1561 | * freeze protection should be the outermost lock. In particular, we have: | |
1562 | * | |
1563 | * sb_start_write | |
1564 | * -> i_mutex (write path, truncate, directory ops, ...) | |
1565 | * -> s_umount (freeze_super, thaw_super) | |
1566 | */ | |
1567 | static inline void sb_start_write(struct super_block *sb) | |
1568 | { | |
8a3c84b6 | 1569 | __sb_start_write(sb, SB_FREEZE_WRITE); |
5accdf82 JK |
1570 | } |
1571 | ||
8a3c84b6 | 1572 | static inline bool sb_start_write_trylock(struct super_block *sb) |
5accdf82 | 1573 | { |
8a3c84b6 | 1574 | return __sb_start_write_trylock(sb, SB_FREEZE_WRITE); |
5accdf82 JK |
1575 | } |
1576 | ||
1577 | /** | |
1578 | * sb_start_pagefault - get write access to a superblock from a page fault | |
1579 | * @sb: the super we write to | |
1580 | * | |
1581 | * When a process starts handling write page fault, it should embed the | |
1582 | * operation into sb_start_pagefault() - sb_end_pagefault() pair to get | |
1583 | * exclusion against file system freezing. This is needed since the page fault | |
1584 | * is going to dirty a page. This function increments number of running page | |
1585 | * faults preventing freezing. If the file system is already frozen, the | |
1586 | * function waits until the file system is thawed. | |
1587 | * | |
1588 | * Since page fault freeze protection behaves as a lock, users have to preserve | |
1589 | * ordering of freeze protection and other filesystem locks. It is advised to | |
c1e8d7c6 | 1590 | * put sb_start_pagefault() close to mmap_lock in lock ordering. Page fault |
5accdf82 JK |
1591 | * handling code implies lock dependency: |
1592 | * | |
c1e8d7c6 | 1593 | * mmap_lock |
5accdf82 JK |
1594 | * -> sb_start_pagefault |
1595 | */ | |
1596 | static inline void sb_start_pagefault(struct super_block *sb) | |
1597 | { | |
8a3c84b6 | 1598 | __sb_start_write(sb, SB_FREEZE_PAGEFAULT); |
5accdf82 JK |
1599 | } |
1600 | ||
39015399 | 1601 | /** |
5accdf82 JK |
1602 | * sb_start_intwrite - get write access to a superblock for internal fs purposes |
1603 | * @sb: the super we write to | |
1604 | * | |
1605 | * This is the third level of protection against filesystem freezing. It is | |
1606 | * free for use by a filesystem. The only requirement is that it must rank | |
1607 | * below sb_start_pagefault. | |
1608 | * | |
1609 | * For example filesystem can call sb_start_intwrite() when starting a | |
1610 | * transaction which somewhat eases handling of freezing for internal sources | |
1611 | * of filesystem changes (internal fs threads, discarding preallocation on file | |
1612 | * close, etc.). | |
1613 | */ | |
1614 | static inline void sb_start_intwrite(struct super_block *sb) | |
1615 | { | |
8a3c84b6 | 1616 | __sb_start_write(sb, SB_FREEZE_FS); |
5accdf82 | 1617 | } |
1da177e4 | 1618 | |
8a3c84b6 | 1619 | static inline bool sb_start_intwrite_trylock(struct super_block *sb) |
0c8e3fe3 | 1620 | { |
8a3c84b6 | 1621 | return __sb_start_write_trylock(sb, SB_FREEZE_FS); |
0c8e3fe3 AG |
1622 | } |
1623 | ||
01beba79 | 1624 | bool inode_owner_or_capable(struct mnt_idmap *idmap, |
21cb47be | 1625 | const struct inode *inode); |
3bd858ab | 1626 | |
1da177e4 LT |
1627 | /* |
1628 | * VFS helper functions.. | |
1629 | */ | |
abf08576 | 1630 | int vfs_create(struct mnt_idmap *, struct inode *, |
6521f891 | 1631 | struct dentry *, umode_t, bool); |
abf08576 | 1632 | int vfs_mkdir(struct mnt_idmap *, struct inode *, |
6521f891 | 1633 | struct dentry *, umode_t); |
abf08576 | 1634 | int vfs_mknod(struct mnt_idmap *, struct inode *, struct dentry *, |
6521f891 | 1635 | umode_t, dev_t); |
abf08576 | 1636 | int vfs_symlink(struct mnt_idmap *, struct inode *, |
6521f891 | 1637 | struct dentry *, const char *); |
abf08576 | 1638 | int vfs_link(struct dentry *, struct mnt_idmap *, struct inode *, |
6521f891 | 1639 | struct dentry *, struct inode **); |
abf08576 CB |
1640 | int vfs_rmdir(struct mnt_idmap *, struct inode *, struct dentry *); |
1641 | int vfs_unlink(struct mnt_idmap *, struct inode *, struct dentry *, | |
6521f891 | 1642 | struct inode **); |
9fe61450 | 1643 | |
92cb01c7 CB |
1644 | /** |
1645 | * struct renamedata - contains all information required for renaming | |
abf08576 | 1646 | * @old_mnt_idmap: idmap of the old mount the inode was found from |
92cb01c7 CB |
1647 | * @old_dir: parent of source |
1648 | * @old_dentry: source | |
abf08576 | 1649 | * @new_mnt_idmap: idmap of the new mount the inode was found from |
92cb01c7 CB |
1650 | * @new_dir: parent of destination |
1651 | * @new_dentry: destination | |
1652 | * @delegated_inode: returns an inode needing a delegation break | |
1653 | * @flags: rename flags | |
1654 | */ | |
9fe61450 | 1655 | struct renamedata { |
abf08576 | 1656 | struct mnt_idmap *old_mnt_idmap; |
9fe61450 CB |
1657 | struct inode *old_dir; |
1658 | struct dentry *old_dentry; | |
abf08576 | 1659 | struct mnt_idmap *new_mnt_idmap; |
9fe61450 CB |
1660 | struct inode *new_dir; |
1661 | struct dentry *new_dentry; | |
1662 | struct inode **delegated_inode; | |
1663 | unsigned int flags; | |
1664 | } __randomize_layout; | |
a3c751a5 | 1665 | |
9fe61450 | 1666 | int vfs_rename(struct renamedata *); |
a3c751a5 | 1667 | |
abf08576 | 1668 | static inline int vfs_whiteout(struct mnt_idmap *idmap, |
6521f891 | 1669 | struct inode *dir, struct dentry *dentry) |
a3c751a5 | 1670 | { |
abf08576 | 1671 | return vfs_mknod(idmap, dir, dentry, S_IFCHR | WHITEOUT_MODE, |
6521f891 | 1672 | WHITEOUT_DEV); |
a3c751a5 | 1673 | } |
1da177e4 | 1674 | |
abf08576 | 1675 | struct file *vfs_tmpfile_open(struct mnt_idmap *idmap, |
22873dea MS |
1676 | const struct path *parentpath, |
1677 | umode_t mode, int open_flag, const struct cred *cred); | |
af7bd4dc | 1678 | |
8e6c848e AV |
1679 | int vfs_mkobj(struct dentry *, umode_t, |
1680 | int (*f)(struct dentry *, umode_t, void *), | |
1681 | void *); | |
1682 | ||
c04011fe | 1683 | int vfs_fchown(struct file *file, uid_t user, gid_t group); |
9e96c8c0 | 1684 | int vfs_fchmod(struct file *file, umode_t mode); |
fd5ad30c | 1685 | int vfs_utimes(const struct path *path, struct timespec64 *times); |
c04011fe | 1686 | |
9df6702a MS |
1687 | extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg); |
1688 | ||
2952db0f AB |
1689 | #ifdef CONFIG_COMPAT |
1690 | extern long compat_ptr_ioctl(struct file *file, unsigned int cmd, | |
1691 | unsigned long arg); | |
1692 | #else | |
1693 | #define compat_ptr_ioctl NULL | |
1694 | #endif | |
1695 | ||
8c744fb8 CH |
1696 | /* |
1697 | * VFS file helper functions. | |
1698 | */ | |
f2d40141 | 1699 | void inode_init_owner(struct mnt_idmap *idmap, struct inode *inode, |
21cb47be | 1700 | const struct inode *dir, umode_t mode); |
a2982cc9 | 1701 | extern bool may_open_dev(const struct path *path); |
9452e93e | 1702 | umode_t mode_strip_sgid(struct mnt_idmap *idmap, |
2b3416ce | 1703 | const struct inode *dir, umode_t mode); |
c4b929b8 | 1704 | |
1da177e4 LT |
1705 | /* |
1706 | * This is the "filldir" function type, used by readdir() to let | |
1707 | * the kernel specify what kind of dirent layout it wants to have. | |
1708 | * This allows the kernel to read directories into kernel space or | |
1709 | * to have different dirent layouts depending on the binary type. | |
25885a35 | 1710 | * Return 'true' to keep going and 'false' if there are no more entries. |
1da177e4 | 1711 | */ |
ac7576f4 | 1712 | struct dir_context; |
25885a35 | 1713 | typedef bool (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64, |
ac7576f4 MS |
1714 | unsigned); |
1715 | ||
5c0ba4e0 | 1716 | struct dir_context { |
a09acf4b | 1717 | filldir_t actor; |
bb6f619b | 1718 | loff_t pos; |
5c0ba4e0 | 1719 | }; |
bb6f619b | 1720 | |
b4caecd4 CH |
1721 | /* |
1722 | * These flags let !MMU mmap() govern direct device mapping vs immediate | |
1723 | * copying more easily for MAP_PRIVATE, especially for ROM filesystems. | |
1724 | * | |
1725 | * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE) | |
1726 | * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED) | |
1727 | * NOMMU_MAP_READ: Can be mapped for reading | |
1728 | * NOMMU_MAP_WRITE: Can be mapped for writing | |
1729 | * NOMMU_MAP_EXEC: Can be mapped for execution | |
1730 | */ | |
1731 | #define NOMMU_MAP_COPY 0x00000001 | |
1732 | #define NOMMU_MAP_DIRECT 0x00000008 | |
1733 | #define NOMMU_MAP_READ VM_MAYREAD | |
1734 | #define NOMMU_MAP_WRITE VM_MAYWRITE | |
1735 | #define NOMMU_MAP_EXEC VM_MAYEXEC | |
1736 | ||
1737 | #define NOMMU_VMFLAGS \ | |
1738 | (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC) | |
1739 | ||
2e5dfc99 DW |
1740 | /* |
1741 | * These flags control the behavior of the remap_file_range function pointer. | |
1742 | * If it is called with len == 0 that means "remap to end of source file". | |
5c437fa2 | 1743 | * See Documentation/filesystems/vfs.rst for more details about this call. |
2e5dfc99 DW |
1744 | * |
1745 | * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate) | |
eca3654e | 1746 | * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request |
2e5dfc99 DW |
1747 | */ |
1748 | #define REMAP_FILE_DEDUP (1 << 0) | |
eca3654e | 1749 | #define REMAP_FILE_CAN_SHORTEN (1 << 1) |
2e5dfc99 DW |
1750 | |
1751 | /* | |
1752 | * These flags signal that the caller is ok with altering various aspects of | |
1753 | * the behavior of the remap operation. The changes must be made by the | |
1754 | * implementation; the vfs remap helper functions can take advantage of them. | |
1755 | * Flags in this category exist to preserve the quirky behavior of the hoisted | |
1756 | * btrfs clone/dedupe ioctls. | |
2e5dfc99 | 1757 | */ |
eca3654e | 1758 | #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN) |
b4caecd4 | 1759 | |
10bc8e4a AG |
1760 | /* |
1761 | * These flags control the behavior of vfs_copy_file_range(). | |
1762 | * They are not available to the user via syscall. | |
1763 | * | |
1764 | * COPY_FILE_SPLICE: call splice direct instead of fs clone/copy ops | |
1765 | */ | |
1766 | #define COPY_FILE_SPLICE (1 << 0) | |
1767 | ||
293bc982 | 1768 | struct iov_iter; |
ee692a21 | 1769 | struct io_uring_cmd; |
293bc982 | 1770 | |
1da177e4 LT |
1771 | struct file_operations { |
1772 | struct module *owner; | |
1773 | loff_t (*llseek) (struct file *, loff_t, int); | |
1774 | ssize_t (*read) (struct file *, char __user *, size_t, loff_t *); | |
1da177e4 | 1775 | ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *); |
293bc982 AV |
1776 | ssize_t (*read_iter) (struct kiocb *, struct iov_iter *); |
1777 | ssize_t (*write_iter) (struct kiocb *, struct iov_iter *); | |
5a72e899 JA |
1778 | int (*iopoll)(struct kiocb *kiocb, struct io_comp_batch *, |
1779 | unsigned int flags); | |
bb6f619b | 1780 | int (*iterate) (struct file *, struct dir_context *); |
61922694 | 1781 | int (*iterate_shared) (struct file *, struct dir_context *); |
a3f8683b | 1782 | __poll_t (*poll) (struct file *, struct poll_table_struct *); |
1da177e4 LT |
1783 | long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long); |
1784 | long (*compat_ioctl) (struct file *, unsigned int, unsigned long); | |
1785 | int (*mmap) (struct file *, struct vm_area_struct *); | |
1c972597 | 1786 | unsigned long mmap_supported_flags; |
1da177e4 | 1787 | int (*open) (struct inode *, struct file *); |
75e1fcc0 | 1788 | int (*flush) (struct file *, fl_owner_t id); |
1da177e4 | 1789 | int (*release) (struct inode *, struct file *); |
02c24a82 | 1790 | int (*fsync) (struct file *, loff_t, loff_t, int datasync); |
1da177e4 LT |
1791 | int (*fasync) (int, struct file *, int); |
1792 | int (*lock) (struct file *, int, struct file_lock *); | |
1da177e4 LT |
1793 | ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int); |
1794 | unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); | |
1795 | int (*check_flags)(int); | |
1da177e4 | 1796 | int (*flock) (struct file *, int, struct file_lock *); |
cbb7e577 JA |
1797 | ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int); |
1798 | ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int); | |
e6f5c789 | 1799 | int (*setlease)(struct file *, long, struct file_lock **, void **); |
2fe17c10 CH |
1800 | long (*fallocate)(struct file *file, int mode, loff_t offset, |
1801 | loff_t len); | |
a3816ab0 | 1802 | void (*show_fdinfo)(struct seq_file *m, struct file *f); |
b4caecd4 CH |
1803 | #ifndef CONFIG_MMU |
1804 | unsigned (*mmap_capabilities)(struct file *); | |
1805 | #endif | |
04b38d60 CH |
1806 | ssize_t (*copy_file_range)(struct file *, loff_t, struct file *, |
1807 | loff_t, size_t, unsigned int); | |
42ec3d4c DW |
1808 | loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in, |
1809 | struct file *file_out, loff_t pos_out, | |
1810 | loff_t len, unsigned int remap_flags); | |
45cd0faa | 1811 | int (*fadvise)(struct file *, loff_t, loff_t, int); |
ee692a21 | 1812 | int (*uring_cmd)(struct io_uring_cmd *ioucmd, unsigned int issue_flags); |
de97fcb3 JA |
1813 | int (*uring_cmd_iopoll)(struct io_uring_cmd *, struct io_comp_batch *, |
1814 | unsigned int poll_flags); | |
3859a271 | 1815 | } __randomize_layout; |
1da177e4 LT |
1816 | |
1817 | struct inode_operations { | |
00cd8dd3 | 1818 | struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int); |
fceef393 | 1819 | const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *); |
4609e1f1 | 1820 | int (*permission) (struct mnt_idmap *, struct inode *, int); |
cac2f8b8 | 1821 | struct posix_acl * (*get_inode_acl)(struct inode *, int, bool); |
44a7d7a8 NP |
1822 | |
1823 | int (*readlink) (struct dentry *, char __user *,int); | |
44a7d7a8 | 1824 | |
6c960e68 | 1825 | int (*create) (struct mnt_idmap *, struct inode *,struct dentry *, |
549c7297 | 1826 | umode_t, bool); |
1da177e4 LT |
1827 | int (*link) (struct dentry *,struct inode *,struct dentry *); |
1828 | int (*unlink) (struct inode *,struct dentry *); | |
7a77db95 | 1829 | int (*symlink) (struct mnt_idmap *, struct inode *,struct dentry *, |
549c7297 | 1830 | const char *); |
c54bd91e | 1831 | int (*mkdir) (struct mnt_idmap *, struct inode *,struct dentry *, |
549c7297 | 1832 | umode_t); |
1da177e4 | 1833 | int (*rmdir) (struct inode *,struct dentry *); |
5ebb29be | 1834 | int (*mknod) (struct mnt_idmap *, struct inode *,struct dentry *, |
549c7297 | 1835 | umode_t,dev_t); |
e18275ae | 1836 | int (*rename) (struct mnt_idmap *, struct inode *, struct dentry *, |
520c8b16 | 1837 | struct inode *, struct dentry *, unsigned int); |
c1632a0f | 1838 | int (*setattr) (struct mnt_idmap *, struct dentry *, struct iattr *); |
b74d24f7 | 1839 | int (*getattr) (struct mnt_idmap *, const struct path *, |
549c7297 | 1840 | struct kstat *, u32, unsigned int); |
1da177e4 | 1841 | ssize_t (*listxattr) (struct dentry *, char *, size_t); |
c4b929b8 MF |
1842 | int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start, |
1843 | u64 len); | |
95582b00 | 1844 | int (*update_time)(struct inode *, struct timespec64 *, int); |
d9585277 | 1845 | int (*atomic_open)(struct inode *, struct dentry *, |
30d90494 | 1846 | struct file *, unsigned open_flag, |
44907d79 | 1847 | umode_t create_mode); |
011e2b71 | 1848 | int (*tmpfile) (struct mnt_idmap *, struct inode *, |
863f144f | 1849 | struct file *, umode_t); |
77435322 | 1850 | struct posix_acl *(*get_acl)(struct mnt_idmap *, struct dentry *, |
7420332a | 1851 | int); |
13e83a49 | 1852 | int (*set_acl)(struct mnt_idmap *, struct dentry *, |
549c7297 | 1853 | struct posix_acl *, int); |
8782a9ae | 1854 | int (*fileattr_set)(struct mnt_idmap *idmap, |
4c5b4799 MS |
1855 | struct dentry *dentry, struct fileattr *fa); |
1856 | int (*fileattr_get)(struct dentry *dentry, struct fileattr *fa); | |
44a7d7a8 | 1857 | } ____cacheline_aligned; |
1da177e4 | 1858 | |
bb7462b6 MS |
1859 | static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio, |
1860 | struct iov_iter *iter) | |
1861 | { | |
1862 | return file->f_op->read_iter(kio, iter); | |
1863 | } | |
1864 | ||
1865 | static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio, | |
1866 | struct iov_iter *iter) | |
1867 | { | |
1868 | return file->f_op->write_iter(kio, iter); | |
1869 | } | |
1870 | ||
f74ac015 MS |
1871 | static inline int call_mmap(struct file *file, struct vm_area_struct *vma) |
1872 | { | |
1873 | return file->f_op->mmap(file, vma); | |
1874 | } | |
1875 | ||
1da177e4 LT |
1876 | extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *); |
1877 | extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *); | |
29732938 ZB |
1878 | extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *, |
1879 | loff_t, size_t, unsigned int); | |
f16acc9d DC |
1880 | extern ssize_t generic_copy_file_range(struct file *file_in, loff_t pos_in, |
1881 | struct file *file_out, loff_t pos_out, | |
1882 | size_t len, unsigned int flags); | |
6f7db389 SR |
1883 | int __generic_remap_file_range_prep(struct file *file_in, loff_t pos_in, |
1884 | struct file *file_out, loff_t pos_out, | |
1885 | loff_t *len, unsigned int remap_flags, | |
1886 | const struct iomap_ops *dax_read_ops); | |
1887 | int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in, | |
1888 | struct file *file_out, loff_t pos_out, | |
1889 | loff_t *count, unsigned int remap_flags); | |
42ec3d4c DW |
1890 | extern loff_t do_clone_file_range(struct file *file_in, loff_t pos_in, |
1891 | struct file *file_out, loff_t pos_out, | |
452ce659 | 1892 | loff_t len, unsigned int remap_flags); |
42ec3d4c DW |
1893 | extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in, |
1894 | struct file *file_out, loff_t pos_out, | |
452ce659 | 1895 | loff_t len, unsigned int remap_flags); |
54dbc151 DW |
1896 | extern int vfs_dedupe_file_range(struct file *file, |
1897 | struct file_dedupe_range *same); | |
42ec3d4c DW |
1898 | extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos, |
1899 | struct file *dst_file, loff_t dst_pos, | |
df365836 | 1900 | loff_t len, unsigned int remap_flags); |
f1825366 | 1901 | |
1da177e4 | 1902 | |
1da177e4 LT |
1903 | struct super_operations { |
1904 | struct inode *(*alloc_inode)(struct super_block *sb); | |
1905 | void (*destroy_inode)(struct inode *); | |
fdb0da89 | 1906 | void (*free_inode)(struct inode *); |
1da177e4 | 1907 | |
aa385729 | 1908 | void (*dirty_inode) (struct inode *, int flags); |
a9185b41 | 1909 | int (*write_inode) (struct inode *, struct writeback_control *wbc); |
45321ac5 | 1910 | int (*drop_inode) (struct inode *); |
be7ce416 | 1911 | void (*evict_inode) (struct inode *); |
1da177e4 | 1912 | void (*put_super) (struct super_block *); |
1da177e4 | 1913 | int (*sync_fs)(struct super_block *sb, int wait); |
48b6bca6 | 1914 | int (*freeze_super) (struct super_block *); |
c4be0c1d | 1915 | int (*freeze_fs) (struct super_block *); |
48b6bca6 | 1916 | int (*thaw_super) (struct super_block *); |
c4be0c1d | 1917 | int (*unfreeze_fs) (struct super_block *); |
726c3342 | 1918 | int (*statfs) (struct dentry *, struct kstatfs *); |
1da177e4 | 1919 | int (*remount_fs) (struct super_block *, int *, char *); |
42faad99 | 1920 | void (*umount_begin) (struct super_block *); |
1da177e4 | 1921 | |
34c80b1d | 1922 | int (*show_options)(struct seq_file *, struct dentry *); |
d861c630 | 1923 | int (*show_devname)(struct seq_file *, struct dentry *); |
a6322de6 | 1924 | int (*show_path)(struct seq_file *, struct dentry *); |
64132379 | 1925 | int (*show_stats)(struct seq_file *, struct dentry *); |
0e51a720 | 1926 | #ifdef CONFIG_QUOTA |
1da177e4 LT |
1927 | ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t); |
1928 | ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t); | |
2d0fa467 | 1929 | struct dquot **(*get_dquots)(struct inode *); |
0e51a720 | 1930 | #endif |
4101b624 VD |
1931 | long (*nr_cached_objects)(struct super_block *, |
1932 | struct shrink_control *); | |
1933 | long (*free_cached_objects)(struct super_block *, | |
1934 | struct shrink_control *); | |
1da177e4 LT |
1935 | }; |
1936 | ||
bbc1096a DH |
1937 | /* |
1938 | * Inode flags - they have no relation to superblock flags now | |
1939 | */ | |
6414e9b0 EB |
1940 | #define S_SYNC (1 << 0) /* Writes are synced at once */ |
1941 | #define S_NOATIME (1 << 1) /* Do not update access times */ | |
1942 | #define S_APPEND (1 << 2) /* Append-only file */ | |
1943 | #define S_IMMUTABLE (1 << 3) /* Immutable file */ | |
1944 | #define S_DEAD (1 << 4) /* removed, but still open directory */ | |
1945 | #define S_NOQUOTA (1 << 5) /* Inode is not counted to quota */ | |
1946 | #define S_DIRSYNC (1 << 6) /* Directory modifications are synchronous */ | |
1947 | #define S_NOCMTIME (1 << 7) /* Do not update file c/mtime */ | |
1948 | #define S_SWAPFILE (1 << 8) /* Do not truncate: swapon got its bmaps */ | |
1949 | #define S_PRIVATE (1 << 9) /* Inode is fs-internal */ | |
1950 | #define S_IMA (1 << 10) /* Inode has an associated IMA struct */ | |
1951 | #define S_AUTOMOUNT (1 << 11) /* Automount/referral quasi-directory */ | |
1952 | #define S_NOSEC (1 << 12) /* no suid or xattr security attributes */ | |
6cd176a5 | 1953 | #ifdef CONFIG_FS_DAX |
6414e9b0 | 1954 | #define S_DAX (1 << 13) /* Direct Access, avoiding the page cache */ |
fbbbad4b | 1955 | #else |
6414e9b0 | 1956 | #define S_DAX 0 /* Make all the DAX code disappear */ |
fbbbad4b | 1957 | #endif |
6414e9b0 EB |
1958 | #define S_ENCRYPTED (1 << 14) /* Encrypted file (using fs/crypto/) */ |
1959 | #define S_CASEFOLD (1 << 15) /* Casefolded file */ | |
1960 | #define S_VERITY (1 << 16) /* Verity file (using fs/verity/) */ | |
1bd9c4e4 | 1961 | #define S_KERNEL_FILE (1 << 17) /* File is in use by the kernel (eg. fs/cachefiles) */ |
bbc1096a DH |
1962 | |
1963 | /* | |
1964 | * Note that nosuid etc flags are inode-specific: setting some file-system | |
1965 | * flags just means all the inodes inherit those flags by default. It might be | |
1966 | * possible to override it selectively if you really wanted to with some | |
1967 | * ioctl() that is not currently implemented. | |
1968 | * | |
e462ec50 | 1969 | * Exception: SB_RDONLY is always applied to the entire file system. |
bbc1096a DH |
1970 | * |
1971 | * Unfortunately, it is possible to change a filesystems flags with it mounted | |
1972 | * with files in use. This means that all of the inodes will not have their | |
1973 | * i_flags updated. Hence, i_flags no longer inherit the superblock mount | |
1974 | * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org | |
1975 | */ | |
1976 | #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg)) | |
1977 | ||
1751e8a6 | 1978 | static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; } |
94e92e7a | 1979 | #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb) |
e462ec50 | 1980 | #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \ |
bbc1096a | 1981 | ((inode)->i_flags & S_SYNC)) |
e462ec50 | 1982 | #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \ |
bbc1096a | 1983 | ((inode)->i_flags & (S_SYNC|S_DIRSYNC))) |
e462ec50 DH |
1984 | #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK) |
1985 | #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME) | |
1986 | #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION) | |
bbc1096a DH |
1987 | |
1988 | #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA) | |
1989 | #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND) | |
1990 | #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE) | |
e462ec50 | 1991 | #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL) |
bbc1096a DH |
1992 | |
1993 | #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD) | |
1994 | #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME) | |
1995 | #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE) | |
1996 | #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE) | |
1997 | #define IS_IMA(inode) ((inode)->i_flags & S_IMA) | |
1998 | #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT) | |
1999 | #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC) | |
fbbbad4b | 2000 | #define IS_DAX(inode) ((inode)->i_flags & S_DAX) |
2ee6a576 | 2001 | #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED) |
b886ee3e | 2002 | #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD) |
5585f2af | 2003 | #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY) |
bbc1096a | 2004 | |
787fb6bc MS |
2005 | #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \ |
2006 | (inode)->i_rdev == WHITEOUT_DEV) | |
2007 | ||
4609e1f1 | 2008 | static inline bool HAS_UNMAPPED_ID(struct mnt_idmap *idmap, |
ba73d987 | 2009 | struct inode *inode) |
0bd23d09 | 2010 | { |
e67fe633 CB |
2011 | return !vfsuid_valid(i_uid_into_vfsuid(idmap, inode)) || |
2012 | !vfsgid_valid(i_gid_into_vfsgid(idmap, inode)); | |
0bd23d09 EB |
2013 | } |
2014 | ||
c75b1d94 JA |
2015 | static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp) |
2016 | { | |
2017 | *kiocb = (struct kiocb) { | |
2018 | .ki_filp = filp, | |
164f4064 | 2019 | .ki_flags = filp->f_iocb_flags, |
20578bdf | 2020 | .ki_ioprio = get_current_ioprio(), |
c75b1d94 JA |
2021 | }; |
2022 | } | |
2023 | ||
5dcdc43e JX |
2024 | static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src, |
2025 | struct file *filp) | |
2026 | { | |
2027 | *kiocb = (struct kiocb) { | |
2028 | .ki_filp = filp, | |
2029 | .ki_flags = kiocb_src->ki_flags, | |
5dcdc43e JX |
2030 | .ki_ioprio = kiocb_src->ki_ioprio, |
2031 | .ki_pos = kiocb_src->ki_pos, | |
2032 | }; | |
2033 | } | |
2034 | ||
1c0eeaf5 | 2035 | /* |
250df6ed | 2036 | * Inode state bits. Protected by inode->i_lock |
1c0eeaf5 | 2037 | * |
1e9d6333 EB |
2038 | * Four bits determine the dirty state of the inode: I_DIRTY_SYNC, |
2039 | * I_DIRTY_DATASYNC, I_DIRTY_PAGES, and I_DIRTY_TIME. | |
1c0eeaf5 JE |
2040 | * |
2041 | * Four bits define the lifetime of an inode. Initially, inodes are I_NEW, | |
2042 | * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at | |
2043 | * various stages of removing an inode. | |
2044 | * | |
eaff8079 | 2045 | * Two bits are used for locking and completion notification, I_NEW and I_SYNC. |
1c0eeaf5 | 2046 | * |
e7ca2d41 | 2047 | * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on |
1e9d6333 EB |
2048 | * fdatasync() (unless I_DIRTY_DATASYNC is also set). |
2049 | * Timestamp updates are the usual cause. | |
2050 | * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of | |
000cb48e JK |
2051 | * these changes separately from I_DIRTY_SYNC so that we |
2052 | * don't have to write inode on fdatasync() when only | |
1e9d6333 | 2053 | * e.g. the timestamps have changed. |
1c0eeaf5 | 2054 | * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean. |
cbfecb92 | 2055 | * I_DIRTY_TIME The inode itself has dirty timestamps, and the |
1e9d6333 EB |
2056 | * lazytime mount option is enabled. We keep track of this |
2057 | * separately from I_DIRTY_SYNC in order to implement | |
2058 | * lazytime. This gets cleared if I_DIRTY_INODE | |
cbfecb92 LC |
2059 | * (I_DIRTY_SYNC and/or I_DIRTY_DATASYNC) gets set. But |
2060 | * I_DIRTY_TIME can still be set if I_DIRTY_SYNC is already | |
2061 | * in place because writeback might already be in progress | |
2062 | * and we don't want to lose the time update | |
eaff8079 CH |
2063 | * I_NEW Serves as both a mutex and completion notification. |
2064 | * New inodes set I_NEW. If two processes both create | |
2065 | * the same inode, one of them will release its inode and | |
2066 | * wait for I_NEW to be released before returning. | |
2067 | * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can | |
2068 | * also cause waiting on I_NEW, without I_NEW actually | |
2069 | * being set. find_inode() uses this to prevent returning | |
2070 | * nearly-dead inodes. | |
1c0eeaf5 JE |
2071 | * I_WILL_FREE Must be set when calling write_inode_now() if i_count |
2072 | * is zero. I_FREEING must be set when I_WILL_FREE is | |
2073 | * cleared. | |
2074 | * I_FREEING Set when inode is about to be freed but still has dirty | |
2075 | * pages or buffers attached or the inode itself is still | |
2076 | * dirty. | |
dbd5768f JK |
2077 | * I_CLEAR Added by clear_inode(). In this state the inode is |
2078 | * clean and can be destroyed. Inode keeps I_FREEING. | |
1c0eeaf5 JE |
2079 | * |
2080 | * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are | |
2081 | * prohibited for many purposes. iget() must wait for | |
2082 | * the inode to be completely released, then create it | |
2083 | * anew. Other functions will just ignore such inodes, | |
eaff8079 | 2084 | * if appropriate. I_NEW is used for waiting. |
1c0eeaf5 | 2085 | * |
169ebd90 JK |
2086 | * I_SYNC Writeback of inode is running. The bit is set during |
2087 | * data writeback, and cleared with a wakeup on the bit | |
2088 | * address once it is done. The bit is also used to pin | |
2089 | * the inode in memory for flusher thread. | |
1c0eeaf5 | 2090 | * |
bd5fe6c5 CH |
2091 | * I_REFERENCED Marks the inode as recently references on the LRU list. |
2092 | * | |
2093 | * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit(). | |
2094 | * | |
682aa8e1 TH |
2095 | * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to |
2096 | * synchronize competing switching instances and to tell | |
b93b0163 | 2097 | * wb stat updates to grab the i_pages lock. See |
a9519def | 2098 | * inode_switch_wbs_work_fn() for details. |
682aa8e1 | 2099 | * |
ad0af710 AG |
2100 | * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper |
2101 | * and work dirs among overlayfs mounts. | |
2102 | * | |
c2b6d621 AV |
2103 | * I_CREATING New object's inode in the middle of setting up. |
2104 | * | |
dae2f8ed IW |
2105 | * I_DONTCACHE Evict inode as soon as it is not used anymore. |
2106 | * | |
5afced3b JK |
2107 | * I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists. |
2108 | * Used to detect that mark_inode_dirty() should not move | |
2109 | * inode between dirty lists. | |
2110 | * | |
08276bda DH |
2111 | * I_PINNING_FSCACHE_WB Inode is pinning an fscache object for writeback. |
2112 | * | |
1c0eeaf5 | 2113 | * Q: What is the difference between I_WILL_FREE and I_FREEING? |
1c0eeaf5 | 2114 | */ |
9e38d86f NP |
2115 | #define I_DIRTY_SYNC (1 << 0) |
2116 | #define I_DIRTY_DATASYNC (1 << 1) | |
2117 | #define I_DIRTY_PAGES (1 << 2) | |
eaff8079 CH |
2118 | #define __I_NEW 3 |
2119 | #define I_NEW (1 << __I_NEW) | |
9e38d86f NP |
2120 | #define I_WILL_FREE (1 << 4) |
2121 | #define I_FREEING (1 << 5) | |
2122 | #define I_CLEAR (1 << 6) | |
eaff8079 | 2123 | #define __I_SYNC 7 |
1c0eeaf5 | 2124 | #define I_SYNC (1 << __I_SYNC) |
9e38d86f | 2125 | #define I_REFERENCED (1 << 8) |
bd5fe6c5 | 2126 | #define __I_DIO_WAKEUP 9 |
ac74d8d6 | 2127 | #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP) |
f4e0c30c | 2128 | #define I_LINKABLE (1 << 10) |
0ae45f63 | 2129 | #define I_DIRTY_TIME (1 << 11) |
682aa8e1 | 2130 | #define I_WB_SWITCH (1 << 13) |
c2b6d621 AV |
2131 | #define I_OVL_INUSE (1 << 14) |
2132 | #define I_CREATING (1 << 15) | |
dae2f8ed | 2133 | #define I_DONTCACHE (1 << 16) |
5afced3b | 2134 | #define I_SYNC_QUEUED (1 << 17) |
08276bda | 2135 | #define I_PINNING_FSCACHE_WB (1 << 18) |
1da177e4 | 2136 | |
0e11f644 CH |
2137 | #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC) |
2138 | #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES) | |
0ae45f63 | 2139 | #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME) |
1da177e4 LT |
2140 | |
2141 | extern void __mark_inode_dirty(struct inode *, int); | |
2142 | static inline void mark_inode_dirty(struct inode *inode) | |
2143 | { | |
2144 | __mark_inode_dirty(inode, I_DIRTY); | |
2145 | } | |
2146 | ||
2147 | static inline void mark_inode_dirty_sync(struct inode *inode) | |
2148 | { | |
2149 | __mark_inode_dirty(inode, I_DIRTY_SYNC); | |
2150 | } | |
2151 | ||
ed296c6c EB |
2152 | /* |
2153 | * Returns true if the given inode itself only has dirty timestamps (its pages | |
2154 | * may still be dirty) and isn't currently being allocated or freed. | |
2155 | * Filesystems should call this if when writing an inode when lazytime is | |
2156 | * enabled, they want to opportunistically write the timestamps of other inodes | |
2157 | * located very nearby on-disk, e.g. in the same inode block. This returns true | |
2158 | * if the given inode is in need of such an opportunistic update. Requires | |
2159 | * i_lock, or at least later re-checking under i_lock. | |
2160 | */ | |
2161 | static inline bool inode_is_dirtytime_only(struct inode *inode) | |
2162 | { | |
2163 | return (inode->i_state & (I_DIRTY_TIME | I_NEW | | |
2164 | I_FREEING | I_WILL_FREE)) == I_DIRTY_TIME; | |
2165 | } | |
2166 | ||
7ada4db8 MS |
2167 | extern void inc_nlink(struct inode *inode); |
2168 | extern void drop_nlink(struct inode *inode); | |
2169 | extern void clear_nlink(struct inode *inode); | |
2170 | extern void set_nlink(struct inode *inode, unsigned int nlink); | |
d8c76e6f DH |
2171 | |
2172 | static inline void inode_inc_link_count(struct inode *inode) | |
2173 | { | |
2174 | inc_nlink(inode); | |
66d21730 AD |
2175 | mark_inode_dirty(inode); |
2176 | } | |
2177 | ||
9a53c3a7 DH |
2178 | static inline void inode_dec_link_count(struct inode *inode) |
2179 | { | |
2180 | drop_nlink(inode); | |
66d21730 AD |
2181 | mark_inode_dirty(inode); |
2182 | } | |
2183 | ||
c3b2da31 JB |
2184 | enum file_time_flags { |
2185 | S_ATIME = 1, | |
2186 | S_MTIME = 2, | |
2187 | S_CTIME = 4, | |
2188 | S_VERSION = 8, | |
2189 | }; | |
2190 | ||
c6718543 | 2191 | extern bool atime_needs_update(const struct path *, struct inode *); |
badcf2b7 | 2192 | extern void touch_atime(const struct path *); |
e60feb44 JB |
2193 | int inode_update_time(struct inode *inode, struct timespec64 *time, int flags); |
2194 | ||
1da177e4 LT |
2195 | static inline void file_accessed(struct file *file) |
2196 | { | |
2197 | if (!(file->f_flags & O_NOATIME)) | |
68ac1234 | 2198 | touch_atime(&file->f_path); |
1da177e4 LT |
2199 | } |
2200 | ||
e38f7f53 | 2201 | extern int file_modified(struct file *file); |
66fa3ced | 2202 | int kiocb_modified(struct kiocb *iocb); |
e38f7f53 | 2203 | |
c3765016 | 2204 | int sync_inode_metadata(struct inode *inode, int wait); |
1da177e4 | 2205 | |
1da177e4 LT |
2206 | struct file_system_type { |
2207 | const char *name; | |
2208 | int fs_flags; | |
bbc1096a DH |
2209 | #define FS_REQUIRES_DEV 1 |
2210 | #define FS_BINARY_MOUNTDATA 2 | |
2211 | #define FS_HAS_SUBTYPE 4 | |
0c55cfc4 | 2212 | #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */ |
0b3b094a | 2213 | #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */ |
a6435940 | 2214 | #define FS_ALLOW_IDMAP 32 /* FS has been updated to handle vfs idmappings. */ |
bbc1096a | 2215 | #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */ |
f3a09c92 | 2216 | int (*init_fs_context)(struct fs_context *); |
d7167b14 | 2217 | const struct fs_parameter_spec *parameters; |
c96e41e9 AV |
2218 | struct dentry *(*mount) (struct file_system_type *, int, |
2219 | const char *, void *); | |
1da177e4 LT |
2220 | void (*kill_sb) (struct super_block *); |
2221 | struct module *owner; | |
2222 | struct file_system_type * next; | |
a5166169 | 2223 | struct hlist_head fs_supers; |
d475fd42 | 2224 | |
cf516249 | 2225 | struct lock_class_key s_lock_key; |
897c6ff9 | 2226 | struct lock_class_key s_umount_key; |
51ee049e | 2227 | struct lock_class_key s_vfs_rename_key; |
5accdf82 | 2228 | struct lock_class_key s_writers_key[SB_FREEZE_LEVELS]; |
d475fd42 PZ |
2229 | |
2230 | struct lock_class_key i_lock_key; | |
2231 | struct lock_class_key i_mutex_key; | |
730633f0 | 2232 | struct lock_class_key invalidate_lock_key; |
14358e6d | 2233 | struct lock_class_key i_mutex_dir_key; |
1da177e4 LT |
2234 | }; |
2235 | ||
7f78e035 EB |
2236 | #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME) |
2237 | ||
152a0836 AV |
2238 | extern struct dentry *mount_bdev(struct file_system_type *fs_type, |
2239 | int flags, const char *dev_name, void *data, | |
2240 | int (*fill_super)(struct super_block *, void *, int)); | |
fc14f2fe AV |
2241 | extern struct dentry *mount_single(struct file_system_type *fs_type, |
2242 | int flags, void *data, | |
2243 | int (*fill_super)(struct super_block *, void *, int)); | |
3c26ff6e AV |
2244 | extern struct dentry *mount_nodev(struct file_system_type *fs_type, |
2245 | int flags, void *data, | |
2246 | int (*fill_super)(struct super_block *, void *, int)); | |
ea441d11 | 2247 | extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path); |
04b94071 | 2248 | void retire_super(struct super_block *sb); |
1da177e4 LT |
2249 | void generic_shutdown_super(struct super_block *sb); |
2250 | void kill_block_super(struct super_block *sb); | |
2251 | void kill_anon_super(struct super_block *sb); | |
2252 | void kill_litter_super(struct super_block *sb); | |
2253 | void deactivate_super(struct super_block *sb); | |
74dbbdd7 | 2254 | void deactivate_locked_super(struct super_block *sb); |
1da177e4 | 2255 | int set_anon_super(struct super_block *s, void *data); |
cb50b348 | 2256 | int set_anon_super_fc(struct super_block *s, struct fs_context *fc); |
0ee5dc67 AV |
2257 | int get_anon_bdev(dev_t *); |
2258 | void free_anon_bdev(dev_t); | |
cb50b348 AV |
2259 | struct super_block *sget_fc(struct fs_context *fc, |
2260 | int (*test)(struct super_block *, struct fs_context *), | |
2261 | int (*set)(struct super_block *, struct fs_context *)); | |
1da177e4 LT |
2262 | struct super_block *sget(struct file_system_type *type, |
2263 | int (*test)(struct super_block *,void *), | |
2264 | int (*set)(struct super_block *,void *), | |
9249e17f | 2265 | int flags, void *data); |
1da177e4 LT |
2266 | |
2267 | /* Alas, no aliases. Too much hassle with bringing module.h everywhere */ | |
2268 | #define fops_get(fops) \ | |
2269 | (((fops) && try_module_get((fops)->owner) ? (fops) : NULL)) | |
2270 | #define fops_put(fops) \ | |
2271 | do { if (fops) module_put((fops)->owner); } while(0) | |
e84f9e57 AV |
2272 | /* |
2273 | * This one is to be used *ONLY* from ->open() instances. | |
2274 | * fops must be non-NULL, pinned down *and* module dependencies | |
2275 | * should be sufficient to pin the caller down as well. | |
2276 | */ | |
2277 | #define replace_fops(f, fops) \ | |
2278 | do { \ | |
2279 | struct file *__file = (f); \ | |
2280 | fops_put(__file->f_op); \ | |
2281 | BUG_ON(!(__file->f_op = (fops))); \ | |
2282 | } while(0) | |
1da177e4 LT |
2283 | |
2284 | extern int register_filesystem(struct file_system_type *); | |
2285 | extern int unregister_filesystem(struct file_system_type *); | |
f0bb5aaf | 2286 | extern int vfs_statfs(const struct path *, struct kstatfs *); |
c8b91acc AV |
2287 | extern int user_statfs(const char __user *, struct kstatfs *); |
2288 | extern int fd_statfs(int, struct kstatfs *); | |
18e9e510 JB |
2289 | extern int freeze_super(struct super_block *super); |
2290 | extern int thaw_super(struct super_block *super); | |
fca39346 JK |
2291 | extern __printf(2, 3) |
2292 | int super_setup_bdi_name(struct super_block *sb, char *fmt, ...); | |
2293 | extern int super_setup_bdi(struct super_block *sb); | |
1da177e4 | 2294 | |
ce3b0f8d AV |
2295 | extern int current_umask(void); |
2296 | ||
5a14696c BF |
2297 | extern void ihold(struct inode * inode); |
2298 | extern void iput(struct inode *); | |
95582b00 | 2299 | extern int generic_update_time(struct inode *, struct timespec64 *, int); |
5a14696c | 2300 | |
f87fd4c2 | 2301 | /* /sys/fs */ |
00d26666 | 2302 | extern struct kobject *fs_kobj; |
f87fd4c2 | 2303 | |
09cbfeaf | 2304 | #define MAX_RW_COUNT (INT_MAX & PAGE_MASK) |
bfcd17a6 | 2305 | |
1da177e4 | 2306 | /* fs/open.c */ |
adb5c247 | 2307 | struct audit_names; |
91a27b2a | 2308 | struct filename { |
adb5c247 JL |
2309 | const char *name; /* pointer to actual string */ |
2310 | const __user char *uptr; /* original userland pointer */ | |
55422d0b | 2311 | int refcnt; |
1c949843 | 2312 | struct audit_names *aname; |
fd2f7cb5 | 2313 | const char iname[]; |
91a27b2a | 2314 | }; |
f1fffbd4 | 2315 | static_assert(offsetof(struct filename, iname) % sizeof(long) == 0); |
1da177e4 | 2316 | |
256c8aed CB |
2317 | static inline struct mnt_idmap *file_mnt_idmap(struct file *file) |
2318 | { | |
2319 | return mnt_idmap(file->f_path.mnt); | |
2320 | } | |
2321 | ||
bb49e9e7 CB |
2322 | /** |
2323 | * is_idmapped_mnt - check whether a mount is mapped | |
2324 | * @mnt: the mount to check | |
2325 | * | |
256c8aed | 2326 | * If @mnt has an non @nop_mnt_idmap attached to it then @mnt is mapped. |
bb49e9e7 CB |
2327 | * |
2328 | * Return: true if mount is mapped, false if not. | |
2329 | */ | |
2330 | static inline bool is_idmapped_mnt(const struct vfsmount *mnt) | |
2331 | { | |
256c8aed | 2332 | return mnt_idmap(mnt) != &nop_mnt_idmap; |
bb49e9e7 CB |
2333 | } |
2334 | ||
7df818b2 | 2335 | extern long vfs_truncate(const struct path *, loff_t); |
abf08576 | 2336 | int do_truncate(struct mnt_idmap *, struct dentry *, loff_t start, |
643fe55a | 2337 | unsigned int time_attrs, struct file *filp); |
72c72bdf | 2338 | extern int vfs_fallocate(struct file *file, int mode, loff_t offset, |
3e63cbb1 | 2339 | loff_t len); |
8e8a1407 | 2340 | extern long do_sys_open(int dfd, const char __user *filename, int flags, |
a218d0fd | 2341 | umode_t mode); |
669abf4e | 2342 | extern struct file *file_open_name(struct filename *, int, umode_t); |
a218d0fd | 2343 | extern struct file *filp_open(const char *, int, umode_t); |
ffb37ca3 | 2344 | extern struct file *file_open_root(const struct path *, |
378c6520 | 2345 | const char *, int, umode_t); |
ffb37ca3 AV |
2346 | static inline struct file *file_open_root_mnt(struct vfsmount *mnt, |
2347 | const char *name, int flags, umode_t mode) | |
2348 | { | |
2349 | return file_open_root(&(struct path){.mnt = mnt, .dentry = mnt->mnt_root}, | |
2350 | name, flags, mode); | |
2351 | } | |
765927b2 | 2352 | extern struct file * dentry_open(const struct path *, int, const struct cred *); |
fb70bf12 CL |
2353 | extern struct file *dentry_create(const struct path *path, int flags, |
2354 | umode_t mode, const struct cred *cred); | |
2abc77af AV |
2355 | extern struct file * open_with_fake_path(const struct path *, int, |
2356 | struct inode*, const struct cred *); | |
19f391eb AV |
2357 | static inline struct file *file_clone_open(struct file *file) |
2358 | { | |
2359 | return dentry_open(&file->f_path, file->f_flags, file->f_cred); | |
2360 | } | |
1da177e4 | 2361 | extern int filp_close(struct file *, fl_owner_t id); |
91a27b2a | 2362 | |
51f39a1f | 2363 | extern struct filename *getname_flags(const char __user *, int, int *); |
8228e2c3 | 2364 | extern struct filename *getname_uflags(const char __user *, int); |
91a27b2a | 2365 | extern struct filename *getname(const char __user *); |
c4ad8f98 | 2366 | extern struct filename *getname_kernel(const char *); |
55422d0b | 2367 | extern void putname(struct filename *name); |
91a27b2a | 2368 | |
30d90494 | 2369 | extern int finish_open(struct file *file, struct dentry *dentry, |
be12af3e | 2370 | int (*open)(struct inode *, struct file *)); |
e45198a6 | 2371 | extern int finish_no_open(struct file *file, struct dentry *dentry); |
1da177e4 | 2372 | |
9751b338 MS |
2373 | /* Helper for the simple case when original dentry is used */ |
2374 | static inline int finish_open_simple(struct file *file, int error) | |
2375 | { | |
2376 | if (error) | |
2377 | return error; | |
2378 | ||
2379 | return finish_open(file, file->f_path.dentry, NULL); | |
2380 | } | |
2381 | ||
1da177e4 LT |
2382 | /* fs/dcache.c */ |
2383 | extern void __init vfs_caches_init_early(void); | |
4248b0da | 2384 | extern void __init vfs_caches_init(void); |
1da177e4 | 2385 | |
b86c089b CL |
2386 | extern struct kmem_cache *names_cachep; |
2387 | ||
a608ca21 | 2388 | #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL) |
3446a8aa | 2389 | #define __putname(name) kmem_cache_free(names_cachep, (void *)(name)) |
1da177e4 | 2390 | |
a212b105 | 2391 | extern struct super_block *blockdev_superblock; |
a212b105 TH |
2392 | static inline bool sb_is_blkdev_sb(struct super_block *sb) |
2393 | { | |
dd0dca22 | 2394 | return IS_ENABLED(CONFIG_BLOCK) && sb == blockdev_superblock; |
47e4491b | 2395 | } |
5c0d6b60 | 2396 | |
b818f09e | 2397 | void emergency_thaw_all(void); |
60b0680f | 2398 | extern int sync_filesystem(struct super_block *); |
9361401e | 2399 | extern const struct file_operations def_blk_fops; |
4b6f5d20 | 2400 | extern const struct file_operations def_chr_fops; |
1da177e4 LT |
2401 | |
2402 | /* fs/char_dev.c */ | |
8a932f73 | 2403 | #define CHRDEV_MAJOR_MAX 512 |
49db08c3 LW |
2404 | /* Marks the bottom of the first segment of free char majors */ |
2405 | #define CHRDEV_MAJOR_DYN_END 234 | |
a5d31a3f LG |
2406 | /* Marks the top and bottom of the second segment of free char majors */ |
2407 | #define CHRDEV_MAJOR_DYN_EXT_START 511 | |
2408 | #define CHRDEV_MAJOR_DYN_EXT_END 384 | |
2409 | ||
1da177e4 LT |
2410 | extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *); |
2411 | extern int register_chrdev_region(dev_t, unsigned, const char *); | |
1905b1bf TH |
2412 | extern int __register_chrdev(unsigned int major, unsigned int baseminor, |
2413 | unsigned int count, const char *name, | |
2414 | const struct file_operations *fops); | |
2415 | extern void __unregister_chrdev(unsigned int major, unsigned int baseminor, | |
2416 | unsigned int count, const char *name); | |
1da177e4 | 2417 | extern void unregister_chrdev_region(dev_t, unsigned); |
68eef3b4 | 2418 | extern void chrdev_show(struct seq_file *,off_t); |
1da177e4 | 2419 | |
1905b1bf TH |
2420 | static inline int register_chrdev(unsigned int major, const char *name, |
2421 | const struct file_operations *fops) | |
2422 | { | |
2423 | return __register_chrdev(major, 0, 256, name, fops); | |
2424 | } | |
2425 | ||
2426 | static inline void unregister_chrdev(unsigned int major, const char *name) | |
2427 | { | |
2428 | __unregister_chrdev(major, 0, 256, name); | |
2429 | } | |
2430 | ||
1da177e4 LT |
2431 | extern void init_special_inode(struct inode *, umode_t, dev_t); |
2432 | ||
2433 | /* Invalid inode operations -- fs/bad_inode.c */ | |
2434 | extern void make_bad_inode(struct inode *); | |
0e3ef1fe | 2435 | extern bool is_bad_inode(struct inode *); |
1da177e4 | 2436 | |
a823e458 JL |
2437 | extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart, |
2438 | loff_t lend); | |
5660e13d JL |
2439 | extern int __must_check file_check_and_advance_wb_err(struct file *file); |
2440 | extern int __must_check file_write_and_wait_range(struct file *file, | |
2441 | loff_t start, loff_t end); | |
2442 | ||
a823e458 JL |
2443 | static inline int file_write_and_wait(struct file *file) |
2444 | { | |
2445 | return file_write_and_wait_range(file, 0, LLONG_MAX); | |
2446 | } | |
2447 | ||
8018ab05 CH |
2448 | extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end, |
2449 | int datasync); | |
2450 | extern int vfs_fsync(struct file *file, int datasync); | |
e2592217 | 2451 | |
22f96b38 JA |
2452 | extern int sync_file_range(struct file *file, loff_t offset, loff_t nbytes, |
2453 | unsigned int flags); | |
2454 | ||
91b94c5d AV |
2455 | static inline bool iocb_is_dsync(const struct kiocb *iocb) |
2456 | { | |
2457 | return (iocb->ki_flags & IOCB_DSYNC) || | |
2458 | IS_SYNC(iocb->ki_filp->f_mapping->host); | |
2459 | } | |
2460 | ||
e2592217 CH |
2461 | /* |
2462 | * Sync the bytes written if this was a synchronous write. Expect ki_pos | |
2463 | * to already be updated for the write, and will return either the amount | |
2464 | * of bytes passed in, or an error if syncing the file failed. | |
2465 | */ | |
2466 | static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count) | |
2467 | { | |
91b94c5d | 2468 | if (iocb_is_dsync(iocb)) { |
e2592217 CH |
2469 | int ret = vfs_fsync_range(iocb->ki_filp, |
2470 | iocb->ki_pos - count, iocb->ki_pos - 1, | |
2471 | (iocb->ki_flags & IOCB_SYNC) ? 0 : 1); | |
2472 | if (ret) | |
2473 | return ret; | |
2474 | } | |
2475 | ||
2476 | return count; | |
d311d79d | 2477 | } |
e2592217 | 2478 | |
1da177e4 LT |
2479 | extern void emergency_sync(void); |
2480 | extern void emergency_remount(void); | |
30460e1e | 2481 | |
9361401e | 2482 | #ifdef CONFIG_BLOCK |
30460e1e CM |
2483 | extern int bmap(struct inode *inode, sector_t *block); |
2484 | #else | |
2485 | static inline int bmap(struct inode *inode, sector_t *block) | |
2486 | { | |
2487 | return -EINVAL; | |
2488 | } | |
9361401e | 2489 | #endif |
30460e1e | 2490 | |
abf08576 | 2491 | int notify_change(struct mnt_idmap *, struct dentry *, |
2f221d6f | 2492 | struct iattr *, struct inode **); |
4609e1f1 CB |
2493 | int inode_permission(struct mnt_idmap *, struct inode *, int); |
2494 | int generic_permission(struct mnt_idmap *, struct inode *, int); | |
02f92b38 CB |
2495 | static inline int file_permission(struct file *file, int mask) |
2496 | { | |
4609e1f1 | 2497 | return inode_permission(file_mnt_idmap(file), |
47291baa | 2498 | file_inode(file), mask); |
02f92b38 CB |
2499 | } |
2500 | static inline int path_permission(const struct path *path, int mask) | |
2501 | { | |
4609e1f1 | 2502 | return inode_permission(mnt_idmap(path->mnt), |
47291baa | 2503 | d_inode(path->dentry), mask); |
02f92b38 | 2504 | } |
9452e93e | 2505 | int __check_sticky(struct mnt_idmap *idmap, struct inode *dir, |
ba73d987 | 2506 | struct inode *inode); |
1da177e4 | 2507 | |
f696a365 MS |
2508 | static inline bool execute_ok(struct inode *inode) |
2509 | { | |
2510 | return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode); | |
2511 | } | |
2512 | ||
6e3e2c43 AV |
2513 | static inline bool inode_wrong_type(const struct inode *inode, umode_t mode) |
2514 | { | |
2515 | return (inode->i_mode ^ mode) & S_IFMT; | |
2516 | } | |
2517 | ||
8d71db4f AV |
2518 | static inline void file_start_write(struct file *file) |
2519 | { | |
2520 | if (!S_ISREG(file_inode(file)->i_mode)) | |
2521 | return; | |
8a3c84b6 | 2522 | sb_start_write(file_inode(file)->i_sb); |
8d71db4f AV |
2523 | } |
2524 | ||
5ae98f15 JK |
2525 | static inline bool file_start_write_trylock(struct file *file) |
2526 | { | |
2527 | if (!S_ISREG(file_inode(file)->i_mode)) | |
2528 | return true; | |
8a3c84b6 | 2529 | return sb_start_write_trylock(file_inode(file)->i_sb); |
5ae98f15 JK |
2530 | } |
2531 | ||
8d71db4f AV |
2532 | static inline void file_end_write(struct file *file) |
2533 | { | |
2534 | if (!S_ISREG(file_inode(file)->i_mode)) | |
2535 | return; | |
2536 | __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE); | |
2537 | } | |
bfe219d3 | 2538 | |
07b8ce1e | 2539 | /* |
592ca09b DH |
2540 | * This is used for regular files where some users -- especially the |
2541 | * currently executed binary in a process, previously handled via | |
2542 | * VM_DENYWRITE -- cannot handle concurrent write (and maybe mmap | |
2543 | * read-write shared) accesses. | |
2544 | * | |
07b8ce1e AV |
2545 | * get_write_access() gets write permission for a file. |
2546 | * put_write_access() releases this write permission. | |
592ca09b DH |
2547 | * deny_write_access() denies write access to a file. |
2548 | * allow_write_access() re-enables write access to a file. | |
2549 | * | |
2550 | * The i_writecount field of an inode can have the following values: | |
2551 | * 0: no write access, no denied write access | |
2552 | * < 0: (-i_writecount) users that denied write access to the file. | |
2553 | * > 0: (i_writecount) users that have write access to the file. | |
07b8ce1e AV |
2554 | * |
2555 | * Normally we operate on that counter with atomic_{inc,dec} and it's safe | |
2556 | * except for the cases where we don't hold i_writecount yet. Then we need to | |
2557 | * use {get,deny}_write_access() - these functions check the sign and refuse | |
2558 | * to do the change if sign is wrong. | |
2559 | */ | |
2560 | static inline int get_write_access(struct inode *inode) | |
2561 | { | |
2562 | return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY; | |
2563 | } | |
2564 | static inline int deny_write_access(struct file *file) | |
2565 | { | |
496ad9aa | 2566 | struct inode *inode = file_inode(file); |
07b8ce1e AV |
2567 | return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY; |
2568 | } | |
1da177e4 LT |
2569 | static inline void put_write_access(struct inode * inode) |
2570 | { | |
2571 | atomic_dec(&inode->i_writecount); | |
2572 | } | |
2573 | static inline void allow_write_access(struct file *file) | |
2574 | { | |
2575 | if (file) | |
496ad9aa | 2576 | atomic_inc(&file_inode(file)->i_writecount); |
1da177e4 | 2577 | } |
f1fe29b4 DH |
2578 | static inline bool inode_is_open_for_write(const struct inode *inode) |
2579 | { | |
2580 | return atomic_read(&inode->i_writecount) > 0; | |
2581 | } | |
2582 | ||
387e3746 | 2583 | #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING) |
a5c96ebf MZ |
2584 | static inline void i_readcount_dec(struct inode *inode) |
2585 | { | |
2586 | BUG_ON(!atomic_read(&inode->i_readcount)); | |
2587 | atomic_dec(&inode->i_readcount); | |
2588 | } | |
2589 | static inline void i_readcount_inc(struct inode *inode) | |
2590 | { | |
2591 | atomic_inc(&inode->i_readcount); | |
2592 | } | |
2593 | #else | |
2594 | static inline void i_readcount_dec(struct inode *inode) | |
2595 | { | |
2596 | return; | |
2597 | } | |
2598 | static inline void i_readcount_inc(struct inode *inode) | |
2599 | { | |
2600 | return; | |
2601 | } | |
2602 | #endif | |
ed8cae8b | 2603 | extern int do_pipe_flags(int *, int); |
1da177e4 | 2604 | |
bdd1d2d3 | 2605 | extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *); |
61a707c5 | 2606 | ssize_t __kernel_read(struct file *file, void *buf, size_t count, loff_t *pos); |
e13ec939 | 2607 | extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *); |
73e18f7c | 2608 | extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *); |
1da177e4 LT |
2609 | extern struct file * open_exec(const char *); |
2610 | ||
2611 | /* fs/dcache.c -- generic fs support functions */ | |
a6e5787f | 2612 | extern bool is_subdir(struct dentry *, struct dentry *); |
640eb7e7 | 2613 | extern bool path_is_under(const struct path *, const struct path *); |
1da177e4 | 2614 | |
9bf39ab2 MS |
2615 | extern char *file_path(struct file *, char *, int); |
2616 | ||
1da177e4 LT |
2617 | #include <linux/err.h> |
2618 | ||
2619 | /* needed for stackable file system support */ | |
965c8e59 | 2620 | extern loff_t default_llseek(struct file *file, loff_t offset, int whence); |
1da177e4 | 2621 | |
965c8e59 | 2622 | extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence); |
1da177e4 | 2623 | |
54e34621 | 2624 | extern int inode_init_always(struct super_block *, struct inode *); |
1da177e4 | 2625 | extern void inode_init_once(struct inode *); |
2aa15890 | 2626 | extern void address_space_init_once(struct address_space *mapping); |
1da177e4 LT |
2627 | extern struct inode * igrab(struct inode *); |
2628 | extern ino_t iunique(struct super_block *, ino_t); | |
2629 | extern int inode_needs_sync(struct inode *inode); | |
45321ac5 | 2630 | extern int generic_delete_inode(struct inode *inode); |
9bcb4b73 AV |
2631 | static inline int generic_drop_inode(struct inode *inode) |
2632 | { | |
88149082 | 2633 | return !inode->i_nlink || inode_unhashed(inode); |
9bcb4b73 | 2634 | } |
2c567af4 | 2635 | extern void d_mark_dontcache(struct inode *inode); |
1da177e4 | 2636 | |
88bd5121 AA |
2637 | extern struct inode *ilookup5_nowait(struct super_block *sb, |
2638 | unsigned long hashval, int (*test)(struct inode *, void *), | |
2639 | void *data); | |
1da177e4 LT |
2640 | extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval, |
2641 | int (*test)(struct inode *, void *), void *data); | |
2642 | extern struct inode *ilookup(struct super_block *sb, unsigned long ino); | |
2643 | ||
80ea09a0 MS |
2644 | extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval, |
2645 | int (*test)(struct inode *, void *), | |
2646 | int (*set)(struct inode *, void *), | |
2647 | void *data); | |
1da177e4 LT |
2648 | extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *); |
2649 | extern struct inode * iget_locked(struct super_block *, unsigned long); | |
fe032c42 TT |
2650 | extern struct inode *find_inode_nowait(struct super_block *, |
2651 | unsigned long, | |
2652 | int (*match)(struct inode *, | |
2653 | unsigned long, void *), | |
2654 | void *data); | |
3f19b2ab DH |
2655 | extern struct inode *find_inode_rcu(struct super_block *, unsigned long, |
2656 | int (*)(struct inode *, void *), void *); | |
2657 | extern struct inode *find_inode_by_ino_rcu(struct super_block *, unsigned long); | |
261bca86 AV |
2658 | extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *); |
2659 | extern int insert_inode_locked(struct inode *); | |
e096d0c7 JB |
2660 | #ifdef CONFIG_DEBUG_LOCK_ALLOC |
2661 | extern void lockdep_annotate_inode_mutex_key(struct inode *inode); | |
2662 | #else | |
2663 | static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { }; | |
2664 | #endif | |
1da177e4 | 2665 | extern void unlock_new_inode(struct inode *); |
c2b6d621 | 2666 | extern void discard_new_inode(struct inode *); |
85fe4025 | 2667 | extern unsigned int get_next_ino(void); |
799ea9e9 | 2668 | extern void evict_inodes(struct super_block *sb); |
3e9d80a8 | 2669 | void dump_mapping(const struct address_space *); |
1da177e4 | 2670 | |
e809d5f0 CD |
2671 | /* |
2672 | * Userspace may rely on the the inode number being non-zero. For example, glibc | |
2673 | * simply ignores files with zero i_ino in unlink() and other places. | |
2674 | * | |
2675 | * As an additional complication, if userspace was compiled with | |
2676 | * _FILE_OFFSET_BITS=32 on a 64-bit kernel we'll only end up reading out the | |
2677 | * lower 32 bits, so we need to check that those aren't zero explicitly. With | |
2678 | * _FILE_OFFSET_BITS=64, this may cause some harmless false-negatives, but | |
2679 | * better safe than sorry. | |
2680 | */ | |
2681 | static inline bool is_zero_ino(ino_t ino) | |
2682 | { | |
2683 | return (u32)ino == 0; | |
2684 | } | |
2685 | ||
1da177e4 | 2686 | extern void __iget(struct inode * inode); |
b46980fe | 2687 | extern void iget_failed(struct inode *); |
dbd5768f | 2688 | extern void clear_inode(struct inode *); |
2e00c97e | 2689 | extern void __destroy_inode(struct inode *); |
a209dfc7 ED |
2690 | extern struct inode *new_inode_pseudo(struct super_block *sb); |
2691 | extern struct inode *new_inode(struct super_block *sb); | |
ff0c7d15 | 2692 | extern void free_inode_nonrcu(struct inode *inode); |
9452e93e | 2693 | extern int setattr_should_drop_suidgid(struct mnt_idmap *, struct inode *); |
5fa8e0a1 | 2694 | extern int file_remove_privs(struct file *); |
4f704d9a CB |
2695 | int setattr_should_drop_sgid(struct mnt_idmap *idmap, |
2696 | const struct inode *inode); | |
1da177e4 | 2697 | |
8b9f3ac5 MS |
2698 | /* |
2699 | * This must be used for allocating filesystems specific inodes to set | |
2700 | * up the inode reclaim context correctly. | |
2701 | */ | |
2702 | static inline void * | |
2703 | alloc_inode_sb(struct super_block *sb, struct kmem_cache *cache, gfp_t gfp) | |
2704 | { | |
2705 | return kmem_cache_alloc_lru(cache, &sb->s_inode_lru, gfp); | |
2706 | } | |
2707 | ||
1da177e4 | 2708 | extern void __insert_inode_hash(struct inode *, unsigned long hashval); |
646ec461 CH |
2709 | static inline void insert_inode_hash(struct inode *inode) |
2710 | { | |
1da177e4 LT |
2711 | __insert_inode_hash(inode, inode->i_ino); |
2712 | } | |
f2ee7abf ED |
2713 | |
2714 | extern void __remove_inode_hash(struct inode *); | |
2715 | static inline void remove_inode_hash(struct inode *inode) | |
2716 | { | |
cbedaac6 | 2717 | if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash)) |
f2ee7abf ED |
2718 | __remove_inode_hash(inode); |
2719 | } | |
2720 | ||
646ec461 | 2721 | extern void inode_sb_list_add(struct inode *inode); |
51b8c1fe | 2722 | extern void inode_add_lru(struct inode *inode); |
1da177e4 | 2723 | |
1da177e4 LT |
2724 | extern int sb_set_blocksize(struct super_block *, int); |
2725 | extern int sb_min_blocksize(struct super_block *, int); | |
2726 | ||
2727 | extern int generic_file_mmap(struct file *, struct vm_area_struct *); | |
2728 | extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *); | |
3309dd04 | 2729 | extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *); |
f6f7a25a | 2730 | int generic_write_checks_count(struct kiocb *iocb, loff_t *count); |
02e83f46 DW |
2731 | extern int generic_write_check_limits(struct file *file, loff_t pos, |
2732 | loff_t *count); | |
a3171351 | 2733 | extern int generic_file_rw_checks(struct file *file_in, struct file *file_out); |
87fa0f3e CH |
2734 | ssize_t filemap_read(struct kiocb *iocb, struct iov_iter *to, |
2735 | ssize_t already_read); | |
ed978a81 | 2736 | extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *); |
8174202b | 2737 | extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *); |
8174202b | 2738 | extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *); |
1af5bb49 | 2739 | extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *); |
800ba295 | 2740 | ssize_t generic_perform_write(struct kiocb *, struct iov_iter *); |
88e6faef | 2741 | |
18e9710e | 2742 | ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos, |
ddef7ed2 | 2743 | rwf_t flags); |
abbb6589 | 2744 | ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos, |
ddef7ed2 | 2745 | rwf_t flags); |
5dcdc43e JX |
2746 | ssize_t vfs_iocb_iter_read(struct file *file, struct kiocb *iocb, |
2747 | struct iov_iter *iter); | |
2748 | ssize_t vfs_iocb_iter_write(struct file *file, struct kiocb *iocb, | |
2749 | struct iov_iter *iter); | |
dbe4e192 | 2750 | |
88e6faef | 2751 | /* fs/splice.c */ |
07073eb0 DH |
2752 | ssize_t filemap_splice_read(struct file *in, loff_t *ppos, |
2753 | struct pipe_inode_info *pipe, | |
2754 | size_t len, unsigned int flags); | |
33b3b041 DH |
2755 | ssize_t direct_splice_read(struct file *in, loff_t *ppos, |
2756 | struct pipe_inode_info *pipe, | |
2757 | size_t len, unsigned int flags); | |
cbb7e577 | 2758 | extern ssize_t generic_file_splice_read(struct file *, loff_t *, |
88e6faef | 2759 | struct pipe_inode_info *, size_t, unsigned int); |
8d020765 | 2760 | extern ssize_t iter_file_splice_write(struct pipe_inode_info *, |
cbb7e577 | 2761 | struct file *, loff_t *, size_t, unsigned int); |
88e6faef | 2762 | extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe, |
cbb7e577 | 2763 | struct file *out, loff_t *, size_t len, unsigned int flags); |
1c118596 MS |
2764 | extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out, |
2765 | loff_t *opos, size_t len, unsigned int flags); | |
2766 | ||
88e6faef | 2767 | |
1da177e4 LT |
2768 | extern void |
2769 | file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping); | |
965c8e59 | 2770 | extern loff_t noop_llseek(struct file *file, loff_t offset, int whence); |
868941b1 | 2771 | #define no_llseek NULL |
46a1c2c7 | 2772 | extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize); |
965c8e59 | 2773 | extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence); |
5760495a | 2774 | extern loff_t generic_file_llseek_size(struct file *file, loff_t offset, |
965c8e59 | 2775 | int whence, loff_t maxsize, loff_t eof); |
1bf9d14d AV |
2776 | extern loff_t fixed_size_llseek(struct file *file, loff_t offset, |
2777 | int whence, loff_t size); | |
b25472f9 AV |
2778 | extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t); |
2779 | extern loff_t no_seek_end_llseek(struct file *, loff_t, int); | |
87112933 | 2780 | int rw_verify_area(int, struct file *, const loff_t *, size_t); |
1da177e4 LT |
2781 | extern int generic_file_open(struct inode * inode, struct file * filp); |
2782 | extern int nonseekable_open(struct inode * inode, struct file * filp); | |
10dce8af | 2783 | extern int stream_open(struct inode * inode, struct file * filp); |
1da177e4 | 2784 | |
9361401e | 2785 | #ifdef CONFIG_BLOCK |
8a4c1e42 | 2786 | typedef void (dio_submit_t)(struct bio *bio, struct inode *inode, |
facd07b0 | 2787 | loff_t file_offset); |
1da177e4 LT |
2788 | |
2789 | enum { | |
1e431f5c CH |
2790 | /* need locking between buffered and direct access */ |
2791 | DIO_LOCKING = 0x01, | |
2792 | ||
2793 | /* filesystem does not support filling holes */ | |
2794 | DIO_SKIP_HOLES = 0x02, | |
1da177e4 LT |
2795 | }; |
2796 | ||
17f8c842 OS |
2797 | ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode, |
2798 | struct block_device *bdev, struct iov_iter *iter, | |
c8b8e32d | 2799 | get_block_t get_block, |
0aaf08de | 2800 | dio_iodone_t end_io, |
17f8c842 | 2801 | int flags); |
7bb46a67 | 2802 | |
17f8c842 OS |
2803 | static inline ssize_t blockdev_direct_IO(struct kiocb *iocb, |
2804 | struct inode *inode, | |
c8b8e32d | 2805 | struct iov_iter *iter, |
17f8c842 | 2806 | get_block_t get_block) |
1da177e4 | 2807 | { |
17f8c842 | 2808 | return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter, |
0aaf08de | 2809 | get_block, NULL, DIO_LOCKING | DIO_SKIP_HOLES); |
1da177e4 | 2810 | } |
9361401e | 2811 | #endif |
1da177e4 | 2812 | |
1d59d61f | 2813 | void inode_dio_wait(struct inode *inode); |
fe0f07d0 | 2814 | |
39015399 | 2815 | /** |
fe0f07d0 JA |
2816 | * inode_dio_begin - signal start of a direct I/O requests |
2817 | * @inode: inode the direct I/O happens on | |
2818 | * | |
2819 | * This is called once we've finished processing a direct I/O request, | |
2820 | * and is used to wake up callers waiting for direct I/O to be quiesced. | |
2821 | */ | |
2822 | static inline void inode_dio_begin(struct inode *inode) | |
2823 | { | |
2824 | atomic_inc(&inode->i_dio_count); | |
2825 | } | |
2826 | ||
39015399 | 2827 | /** |
fe0f07d0 JA |
2828 | * inode_dio_end - signal finish of a direct I/O requests |
2829 | * @inode: inode the direct I/O happens on | |
2830 | * | |
2831 | * This is called once we've finished processing a direct I/O request, | |
2832 | * and is used to wake up callers waiting for direct I/O to be quiesced. | |
2833 | */ | |
2834 | static inline void inode_dio_end(struct inode *inode) | |
2835 | { | |
2836 | if (atomic_dec_and_test(&inode->i_dio_count)) | |
2837 | wake_up_bit(&inode->i_state, __I_DIO_WAKEUP); | |
2838 | } | |
1d59d61f | 2839 | |
a92853b6 KK |
2840 | /* |
2841 | * Warn about a page cache invalidation failure diring a direct I/O write. | |
2842 | */ | |
2843 | void dio_warn_stale_pagecache(struct file *filp); | |
2844 | ||
5f16f322 TT |
2845 | extern void inode_set_flags(struct inode *inode, unsigned int flags, |
2846 | unsigned int mask); | |
2847 | ||
4b6f5d20 | 2848 | extern const struct file_operations generic_ro_fops; |
1da177e4 LT |
2849 | |
2850 | #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m)) | |
2851 | ||
5d826c84 | 2852 | extern int readlink_copy(char __user *, int, const char *); |
1da177e4 | 2853 | extern int page_readlink(struct dentry *, char __user *, int); |
fceef393 AV |
2854 | extern const char *page_get_link(struct dentry *, struct inode *, |
2855 | struct delayed_call *); | |
2856 | extern void page_put_link(void *); | |
1da177e4 | 2857 | extern int page_symlink(struct inode *inode, const char *symname, int len); |
c5ef1c42 | 2858 | extern const struct inode_operations page_symlink_inode_operations; |
fceef393 | 2859 | extern void kfree_link(void *); |
b74d24f7 | 2860 | void generic_fillattr(struct mnt_idmap *, struct inode *, struct kstat *); |
4f911138 | 2861 | void generic_fill_statx_attr(struct inode *inode, struct kstat *stat); |
a528d35e DH |
2862 | extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int); |
2863 | extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int); | |
b462707e | 2864 | void __inode_add_bytes(struct inode *inode, loff_t bytes); |
1da177e4 | 2865 | void inode_add_bytes(struct inode *inode, loff_t bytes); |
1c8924eb | 2866 | void __inode_sub_bytes(struct inode *inode, loff_t bytes); |
1da177e4 | 2867 | void inode_sub_bytes(struct inode *inode, loff_t bytes); |
f4a8116a JK |
2868 | static inline loff_t __inode_get_bytes(struct inode *inode) |
2869 | { | |
2870 | return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes; | |
2871 | } | |
1da177e4 LT |
2872 | loff_t inode_get_bytes(struct inode *inode); |
2873 | void inode_set_bytes(struct inode *inode, loff_t bytes); | |
fceef393 AV |
2874 | const char *simple_get_link(struct dentry *, struct inode *, |
2875 | struct delayed_call *); | |
61ba64fc | 2876 | extern const struct inode_operations simple_symlink_inode_operations; |
1da177e4 | 2877 | |
5c0ba4e0 | 2878 | extern int iterate_dir(struct file *, struct dir_context *); |
1da177e4 | 2879 | |
09f1bde4 CH |
2880 | int vfs_fstatat(int dfd, const char __user *filename, struct kstat *stat, |
2881 | int flags); | |
da9aa5d9 | 2882 | int vfs_fstat(int fd, struct kstat *stat); |
a528d35e DH |
2883 | |
2884 | static inline int vfs_stat(const char __user *filename, struct kstat *stat) | |
2885 | { | |
0b2c6693 | 2886 | return vfs_fstatat(AT_FDCWD, filename, stat, 0); |
a528d35e DH |
2887 | } |
2888 | static inline int vfs_lstat(const char __user *name, struct kstat *stat) | |
2889 | { | |
0b2c6693 | 2890 | return vfs_fstatat(AT_FDCWD, name, stat, AT_SYMLINK_NOFOLLOW); |
a528d35e | 2891 | } |
a528d35e | 2892 | |
d60874cd | 2893 | extern const char *vfs_get_link(struct dentry *, struct delayed_call *); |
fd4a0edf | 2894 | extern int vfs_readlink(struct dentry *, char __user *, int); |
1da177e4 | 2895 | |
ee416bcd | 2896 | extern struct file_system_type *get_filesystem(struct file_system_type *fs); |
c18479fe | 2897 | extern void put_filesystem(struct file_system_type *fs); |
1da177e4 LT |
2898 | extern struct file_system_type *get_fs_type(const char *name); |
2899 | extern struct super_block *get_super(struct block_device *); | |
4504230a | 2900 | extern struct super_block *get_active_super(struct block_device *bdev); |
1da177e4 | 2901 | extern void drop_super(struct super_block *sb); |
ba6379f7 | 2902 | extern void drop_super_exclusive(struct super_block *sb); |
01a05b33 | 2903 | extern void iterate_supers(void (*)(struct super_block *, void *), void *); |
43e15cdb AV |
2904 | extern void iterate_supers_type(struct file_system_type *, |
2905 | void (*)(struct super_block *, void *), void *); | |
1da177e4 LT |
2906 | |
2907 | extern int dcache_dir_open(struct inode *, struct file *); | |
2908 | extern int dcache_dir_close(struct inode *, struct file *); | |
2909 | extern loff_t dcache_dir_lseek(struct file *, loff_t, int); | |
5f99f4e7 | 2910 | extern int dcache_readdir(struct file *, struct dir_context *); |
c1632a0f | 2911 | extern int simple_setattr(struct mnt_idmap *, struct dentry *, |
549c7297 | 2912 | struct iattr *); |
b74d24f7 | 2913 | extern int simple_getattr(struct mnt_idmap *, const struct path *, |
549c7297 | 2914 | struct kstat *, u32, unsigned int); |
726c3342 | 2915 | extern int simple_statfs(struct dentry *, struct kstatfs *); |
20955e89 | 2916 | extern int simple_open(struct inode *inode, struct file *file); |
1da177e4 LT |
2917 | extern int simple_link(struct dentry *, struct inode *, struct dentry *); |
2918 | extern int simple_unlink(struct inode *, struct dentry *); | |
2919 | extern int simple_rmdir(struct inode *, struct dentry *); | |
6429e463 LB |
2920 | extern int simple_rename_exchange(struct inode *old_dir, struct dentry *old_dentry, |
2921 | struct inode *new_dir, struct dentry *new_dentry); | |
e18275ae | 2922 | extern int simple_rename(struct mnt_idmap *, struct inode *, |
549c7297 CB |
2923 | struct dentry *, struct inode *, struct dentry *, |
2924 | unsigned int); | |
a3d1e7eb AV |
2925 | extern void simple_recursive_removal(struct dentry *, |
2926 | void (*callback)(struct dentry *)); | |
02c24a82 | 2927 | extern int noop_fsync(struct file *, loff_t, loff_t, int); |
f44c7763 | 2928 | extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter); |
1da177e4 | 2929 | extern int simple_empty(struct dentry *); |
afddba49 | 2930 | extern int simple_write_begin(struct file *file, struct address_space *mapping, |
9d6b0cd7 | 2931 | loff_t pos, unsigned len, |
afddba49 | 2932 | struct page **pagep, void **fsdata); |
c1e3dbe9 | 2933 | extern const struct address_space_operations ram_aops; |
b26d4cd3 | 2934 | extern int always_delete_dentry(const struct dentry *); |
6987843f | 2935 | extern struct inode *alloc_anon_inode(struct super_block *); |
e6f5c789 | 2936 | extern int simple_nosetlease(struct file *, long, struct file_lock **, void **); |
b26d4cd3 | 2937 | extern const struct dentry_operations simple_dentry_operations; |
1da177e4 | 2938 | |
00cd8dd3 | 2939 | extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags); |
1da177e4 | 2940 | extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *); |
4b6f5d20 | 2941 | extern const struct file_operations simple_dir_operations; |
c5ef1c42 | 2942 | extern const struct inode_operations simple_dir_inode_operations; |
fbabfd0f EB |
2943 | extern void make_empty_dir_inode(struct inode *inode); |
2944 | extern bool is_empty_dir_inode(struct inode *inode); | |
cda37124 | 2945 | struct tree_descr { const char *name; const struct file_operations *ops; int mode; }; |
1da177e4 | 2946 | struct dentry *d_alloc_name(struct dentry *, const char *); |
cda37124 EB |
2947 | extern int simple_fill_super(struct super_block *, unsigned long, |
2948 | const struct tree_descr *); | |
1f5ce9e9 | 2949 | extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count); |
1da177e4 LT |
2950 | extern void simple_release_fs(struct vfsmount **mount, int *count); |
2951 | ||
93b07113 AM |
2952 | extern ssize_t simple_read_from_buffer(void __user *to, size_t count, |
2953 | loff_t *ppos, const void *from, size_t available); | |
6a727b43 JS |
2954 | extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos, |
2955 | const void __user *from, size_t count); | |
1da177e4 | 2956 | |
ac13a829 | 2957 | extern int __generic_file_fsync(struct file *, loff_t, loff_t, int); |
02c24a82 | 2958 | extern int generic_file_fsync(struct file *, loff_t, loff_t, int); |
d5aacad5 | 2959 | |
30ca22c7 PL |
2960 | extern int generic_check_addressable(unsigned, u64); |
2961 | ||
608af703 | 2962 | extern void generic_set_encrypted_ci_d_ops(struct dentry *dentry); |
c843843e | 2963 | |
4609e1f1 | 2964 | int may_setattr(struct mnt_idmap *idmap, struct inode *inode, |
7bb698f0 | 2965 | unsigned int ia_valid); |
c1632a0f | 2966 | int setattr_prepare(struct mnt_idmap *, struct dentry *, struct iattr *); |
25d9e2d1 | 2967 | extern int inode_newsize_ok(const struct inode *, loff_t offset); |
c1632a0f | 2968 | void setattr_copy(struct mnt_idmap *, struct inode *inode, |
2f221d6f | 2969 | const struct iattr *attr); |
1da177e4 | 2970 | |
c3b2da31 | 2971 | extern int file_update_time(struct file *file); |
1da177e4 | 2972 | |
f05a3849 | 2973 | static inline bool vma_is_dax(const struct vm_area_struct *vma) |
baabda26 DW |
2974 | { |
2975 | return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host); | |
2976 | } | |
2977 | ||
2bb6d283 DW |
2978 | static inline bool vma_is_fsdax(struct vm_area_struct *vma) |
2979 | { | |
2980 | struct inode *inode; | |
2981 | ||
52650c8b | 2982 | if (!IS_ENABLED(CONFIG_FS_DAX) || !vma->vm_file) |
2bb6d283 DW |
2983 | return false; |
2984 | if (!vma_is_dax(vma)) | |
2985 | return false; | |
2986 | inode = file_inode(vma->vm_file); | |
230f5a89 | 2987 | if (S_ISCHR(inode->i_mode)) |
2bb6d283 DW |
2988 | return false; /* device-dax */ |
2989 | return true; | |
2990 | } | |
2991 | ||
2ba48ce5 AV |
2992 | static inline int iocb_flags(struct file *file) |
2993 | { | |
2994 | int res = 0; | |
2995 | if (file->f_flags & O_APPEND) | |
2996 | res |= IOCB_APPEND; | |
efbe3c24 | 2997 | if (file->f_flags & O_DIRECT) |
2ba48ce5 | 2998 | res |= IOCB_DIRECT; |
91b94c5d | 2999 | if (file->f_flags & O_DSYNC) |
dde0c2e7 CH |
3000 | res |= IOCB_DSYNC; |
3001 | if (file->f_flags & __O_SYNC) | |
3002 | res |= IOCB_SYNC; | |
2ba48ce5 AV |
3003 | return res; |
3004 | } | |
3005 | ||
ddef7ed2 | 3006 | static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags) |
fdd2f5b7 | 3007 | { |
1752f0ad PB |
3008 | int kiocb_flags = 0; |
3009 | ||
ce71bfea JA |
3010 | /* make sure there's no overlap between RWF and private IOCB flags */ |
3011 | BUILD_BUG_ON((__force int) RWF_SUPPORTED & IOCB_EVENTFD); | |
3012 | ||
1752f0ad PB |
3013 | if (!flags) |
3014 | return 0; | |
fdd2f5b7 GR |
3015 | if (unlikely(flags & ~RWF_SUPPORTED)) |
3016 | return -EOPNOTSUPP; | |
3017 | ||
b745fafa | 3018 | if (flags & RWF_NOWAIT) { |
91f9943e | 3019 | if (!(ki->ki_filp->f_mode & FMODE_NOWAIT)) |
b745fafa | 3020 | return -EOPNOTSUPP; |
ce71bfea | 3021 | kiocb_flags |= IOCB_NOIO; |
b745fafa | 3022 | } |
ce71bfea | 3023 | kiocb_flags |= (__force int) (flags & RWF_SUPPORTED); |
fdd2f5b7 | 3024 | if (flags & RWF_SYNC) |
ce71bfea | 3025 | kiocb_flags |= IOCB_DSYNC; |
1752f0ad PB |
3026 | |
3027 | ki->ki_flags |= kiocb_flags; | |
fdd2f5b7 GR |
3028 | return 0; |
3029 | } | |
3030 | ||
1da177e4 LT |
3031 | static inline ino_t parent_ino(struct dentry *dentry) |
3032 | { | |
3033 | ino_t res; | |
3034 | ||
b5c84bf6 NP |
3035 | /* |
3036 | * Don't strictly need d_lock here? If the parent ino could change | |
3037 | * then surely we'd have a deeper race in the caller? | |
3038 | */ | |
1da177e4 LT |
3039 | spin_lock(&dentry->d_lock); |
3040 | res = dentry->d_parent->d_inode->i_ino; | |
3041 | spin_unlock(&dentry->d_lock); | |
3042 | return res; | |
3043 | } | |
3044 | ||
1da177e4 LT |
3045 | /* Transaction based IO helpers */ |
3046 | ||
3047 | /* | |
3048 | * An argresp is stored in an allocated page and holds the | |
3049 | * size of the argument or response, along with its content | |
3050 | */ | |
3051 | struct simple_transaction_argresp { | |
3052 | ssize_t size; | |
5e01fdff | 3053 | char data[]; |
1da177e4 LT |
3054 | }; |
3055 | ||
3056 | #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp)) | |
3057 | ||
3058 | char *simple_transaction_get(struct file *file, const char __user *buf, | |
3059 | size_t size); | |
3060 | ssize_t simple_transaction_read(struct file *file, char __user *buf, | |
3061 | size_t size, loff_t *pos); | |
3062 | int simple_transaction_release(struct inode *inode, struct file *file); | |
3063 | ||
76791ab2 | 3064 | void simple_transaction_set(struct file *file, size_t n); |
1da177e4 | 3065 | |
acaefc25 AB |
3066 | /* |
3067 | * simple attribute files | |
3068 | * | |
3069 | * These attributes behave similar to those in sysfs: | |
3070 | * | |
3071 | * Writing to an attribute immediately sets a value, an open file can be | |
3072 | * written to multiple times. | |
3073 | * | |
3074 | * Reading from an attribute creates a buffer from the value that might get | |
3075 | * read with multiple read calls. When the attribute has been read | |
3076 | * completely, no further read calls are possible until the file is opened | |
3077 | * again. | |
3078 | * | |
3079 | * All attributes contain a text representation of a numeric value | |
3080 | * that are accessed with the get() and set() functions. | |
3081 | */ | |
2e41f274 | 3082 | #define DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, __is_signed) \ |
acaefc25 AB |
3083 | static int __fops ## _open(struct inode *inode, struct file *file) \ |
3084 | { \ | |
3085 | __simple_attr_check_format(__fmt, 0ull); \ | |
3086 | return simple_attr_open(inode, file, __get, __set, __fmt); \ | |
3087 | } \ | |
828c0950 | 3088 | static const struct file_operations __fops = { \ |
acaefc25 AB |
3089 | .owner = THIS_MODULE, \ |
3090 | .open = __fops ## _open, \ | |
74bedc4d | 3091 | .release = simple_attr_release, \ |
acaefc25 | 3092 | .read = simple_attr_read, \ |
2e41f274 | 3093 | .write = (__is_signed) ? simple_attr_write_signed : simple_attr_write, \ |
1ec5584e | 3094 | .llseek = generic_file_llseek, \ |
68be3029 | 3095 | } |
acaefc25 | 3096 | |
2e41f274 AM |
3097 | #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \ |
3098 | DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, false) | |
3099 | ||
3100 | #define DEFINE_SIMPLE_ATTRIBUTE_SIGNED(__fops, __get, __set, __fmt) \ | |
3101 | DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, true) | |
3102 | ||
b9075fa9 JP |
3103 | static inline __printf(1, 2) |
3104 | void __simple_attr_check_format(const char *fmt, ...) | |
acaefc25 AB |
3105 | { |
3106 | /* don't do anything, just let the compiler check the arguments; */ | |
3107 | } | |
3108 | ||
3109 | int simple_attr_open(struct inode *inode, struct file *file, | |
8b88b099 | 3110 | int (*get)(void *, u64 *), int (*set)(void *, u64), |
acaefc25 | 3111 | const char *fmt); |
74bedc4d | 3112 | int simple_attr_release(struct inode *inode, struct file *file); |
acaefc25 AB |
3113 | ssize_t simple_attr_read(struct file *file, char __user *buf, |
3114 | size_t len, loff_t *ppos); | |
3115 | ssize_t simple_attr_write(struct file *file, const char __user *buf, | |
3116 | size_t len, loff_t *ppos); | |
2e41f274 AM |
3117 | ssize_t simple_attr_write_signed(struct file *file, const char __user *buf, |
3118 | size_t len, loff_t *ppos); | |
acaefc25 | 3119 | |
4be28540 | 3120 | struct ctl_table; |
6e7c1770 | 3121 | int __init list_bdev_fs_names(char *buf, size_t size); |
62239ac2 | 3122 | |
3cd90ea4 | 3123 | #define __FMODE_EXEC ((__force int) FMODE_EXEC) |
1a44bc8c NK |
3124 | #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY) |
3125 | ||
6d125529 | 3126 | #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE]) |
ecf081d1 | 3127 | #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \ |
1a44bc8c | 3128 | (flag & __FMODE_NONOTIFY))) |
5300990c | 3129 | |
d37177ba | 3130 | static inline bool is_sxid(umode_t mode) |
69b45732 | 3131 | { |
8d84e39d | 3132 | return mode & (S_ISUID | S_ISGID); |
69b45732 AK |
3133 | } |
3134 | ||
9452e93e | 3135 | static inline int check_sticky(struct mnt_idmap *idmap, |
ba73d987 | 3136 | struct inode *dir, struct inode *inode) |
cbdf35bc MS |
3137 | { |
3138 | if (!(dir->i_mode & S_ISVTX)) | |
3139 | return 0; | |
3140 | ||
9452e93e | 3141 | return __check_sticky(idmap, dir, inode); |
cbdf35bc MS |
3142 | } |
3143 | ||
69b45732 AK |
3144 | static inline void inode_has_no_xattr(struct inode *inode) |
3145 | { | |
e462ec50 | 3146 | if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC)) |
69b45732 AK |
3147 | inode->i_flags |= S_NOSEC; |
3148 | } | |
3149 | ||
a7400222 AV |
3150 | static inline bool is_root_inode(struct inode *inode) |
3151 | { | |
3152 | return inode == inode->i_sb->s_root->d_inode; | |
3153 | } | |
3154 | ||
5f99f4e7 AV |
3155 | static inline bool dir_emit(struct dir_context *ctx, |
3156 | const char *name, int namelen, | |
3157 | u64 ino, unsigned type) | |
3158 | { | |
25885a35 | 3159 | return ctx->actor(ctx, name, namelen, ctx->pos, ino, type); |
5f99f4e7 AV |
3160 | } |
3161 | static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx) | |
3162 | { | |
3163 | return ctx->actor(ctx, ".", 1, ctx->pos, | |
25885a35 | 3164 | file->f_path.dentry->d_inode->i_ino, DT_DIR); |
5f99f4e7 AV |
3165 | } |
3166 | static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx) | |
3167 | { | |
3168 | return ctx->actor(ctx, "..", 2, ctx->pos, | |
25885a35 | 3169 | parent_ino(file->f_path.dentry), DT_DIR); |
5f99f4e7 AV |
3170 | } |
3171 | static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx) | |
3172 | { | |
3173 | if (ctx->pos == 0) { | |
3174 | if (!dir_emit_dot(file, ctx)) | |
3175 | return false; | |
3176 | ctx->pos = 1; | |
3177 | } | |
3178 | if (ctx->pos == 1) { | |
3179 | if (!dir_emit_dotdot(file, ctx)) | |
3180 | return false; | |
3181 | ctx->pos = 2; | |
3182 | } | |
3183 | return true; | |
3184 | } | |
5ded75ec AV |
3185 | static inline bool dir_relax(struct inode *inode) |
3186 | { | |
5955102c AV |
3187 | inode_unlock(inode); |
3188 | inode_lock(inode); | |
5ded75ec AV |
3189 | return !IS_DEADDIR(inode); |
3190 | } | |
5f99f4e7 | 3191 | |
ae05327a AV |
3192 | static inline bool dir_relax_shared(struct inode *inode) |
3193 | { | |
3194 | inode_unlock_shared(inode); | |
3195 | inode_lock_shared(inode); | |
3196 | return !IS_DEADDIR(inode); | |
3197 | } | |
3198 | ||
90f8572b | 3199 | extern bool path_noexec(const struct path *path); |
21fc61c7 | 3200 | extern void inode_nohighmem(struct inode *inode); |
90f8572b | 3201 | |
45cd0faa AG |
3202 | /* mm/fadvise.c */ |
3203 | extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len, | |
3204 | int advice); | |
cf1ea059 JK |
3205 | extern int generic_fadvise(struct file *file, loff_t offset, loff_t len, |
3206 | int advice); | |
45cd0faa | 3207 | |
1da177e4 | 3208 | #endif /* _LINUX_FS_H */ |