]>
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; |
3e7807d5 | 70 | struct fscrypt_inode_info; |
0b81d077 | 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) | |
bbc1096a DH |
122 | /* 32bit hashes as llseek() offset (for directories) */ |
123 | #define FMODE_32BITHASH ((__force fmode_t)0x200) | |
124 | /* 64bit hashes as llseek() offset (for directories) */ | |
125 | #define FMODE_64BITHASH ((__force fmode_t)0x400) | |
126 | ||
127 | /* | |
128 | * Don't update ctime and mtime. | |
129 | * | |
130 | * Currently a special hack for the XFS open_by_handle ioctl, but we'll | |
131 | * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon. | |
132 | */ | |
133 | #define FMODE_NOCMTIME ((__force fmode_t)0x800) | |
134 | ||
135 | /* Expect random access pattern */ | |
136 | #define FMODE_RANDOM ((__force fmode_t)0x1000) | |
137 | ||
bbcd53c9 | 138 | /* File is huge (eg. /dev/mem): treat loff_t as unsigned */ |
bbc1096a DH |
139 | #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000) |
140 | ||
141 | /* File is opened with O_PATH; almost nothing can be done with it */ | |
142 | #define FMODE_PATH ((__force fmode_t)0x4000) | |
143 | ||
2be7d348 LT |
144 | /* File needs atomic accesses to f_pos */ |
145 | #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000) | |
83f936c7 AV |
146 | /* Write access to underlying fs */ |
147 | #define FMODE_WRITER ((__force fmode_t)0x10000) | |
7f7f25e8 AV |
148 | /* Has read method(s) */ |
149 | #define FMODE_CAN_READ ((__force fmode_t)0x20000) | |
150 | /* Has write method(s) */ | |
151 | #define FMODE_CAN_WRITE ((__force fmode_t)0x40000) | |
9c225f26 | 152 | |
f5d11409 | 153 | #define FMODE_OPENED ((__force fmode_t)0x80000) |
73a09dd9 | 154 | #define FMODE_CREATED ((__force fmode_t)0x100000) |
f5d11409 | 155 | |
10dce8af KS |
156 | /* File is stream-like */ |
157 | #define FMODE_STREAM ((__force fmode_t)0x200000) | |
158 | ||
a2ad63da N |
159 | /* File supports DIRECT IO */ |
160 | #define FMODE_CAN_ODIRECT ((__force fmode_t)0x400000) | |
161 | ||
17e81022 YZ |
162 | #define FMODE_NOREUSE ((__force fmode_t)0x800000) |
163 | ||
d8aeb44a JA |
164 | /* File supports non-exclusive O_DIRECT writes from multiple threads */ |
165 | #define FMODE_DIO_PARALLEL_WRITE ((__force fmode_t)0x1000000) | |
166 | ||
62d53c4a AG |
167 | /* File is embedded in backing_file object */ |
168 | #define FMODE_BACKING ((__force fmode_t)0x2000000) | |
169 | ||
bbc1096a | 170 | /* File was opened by fanotify and shouldn't generate fanotify events */ |
75069f2b | 171 | #define FMODE_NONOTIFY ((__force fmode_t)0x4000000) |
bbc1096a | 172 | |
91f9943e | 173 | /* File is capable of returning -EAGAIN if I/O will block */ |
a07b2000 AV |
174 | #define FMODE_NOWAIT ((__force fmode_t)0x8000000) |
175 | ||
176 | /* File represents mount that needs unmounting */ | |
177 | #define FMODE_NEED_UNMOUNT ((__force fmode_t)0x10000000) | |
b745fafa | 178 | |
d3b1084d | 179 | /* File does not contribute to nr_files count */ |
a07b2000 | 180 | #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000) |
d3b1084d | 181 | |
c2a25ec0 JA |
182 | /* File supports async buffered reads */ |
183 | #define FMODE_BUF_RASYNC ((__force fmode_t)0x40000000) | |
184 | ||
80175539 SR |
185 | /* File supports async nowait buffered writes */ |
186 | #define FMODE_BUF_WASYNC ((__force fmode_t)0x80000000) | |
187 | ||
1da177e4 LT |
188 | /* |
189 | * Attribute flags. These should be or-ed together to figure out what | |
190 | * has been changed! | |
191 | */ | |
9767d749 MS |
192 | #define ATTR_MODE (1 << 0) |
193 | #define ATTR_UID (1 << 1) | |
194 | #define ATTR_GID (1 << 2) | |
195 | #define ATTR_SIZE (1 << 3) | |
196 | #define ATTR_ATIME (1 << 4) | |
197 | #define ATTR_MTIME (1 << 5) | |
198 | #define ATTR_CTIME (1 << 6) | |
199 | #define ATTR_ATIME_SET (1 << 7) | |
200 | #define ATTR_MTIME_SET (1 << 8) | |
201 | #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */ | |
9767d749 MS |
202 | #define ATTR_KILL_SUID (1 << 11) |
203 | #define ATTR_KILL_SGID (1 << 12) | |
204 | #define ATTR_FILE (1 << 13) | |
205 | #define ATTR_KILL_PRIV (1 << 14) | |
206 | #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */ | |
207 | #define ATTR_TIMES_SET (1 << 16) | |
f2b20f6e | 208 | #define ATTR_TOUCH (1 << 17) |
1da177e4 | 209 | |
787fb6bc MS |
210 | /* |
211 | * Whiteout is represented by a char device. The following constants define the | |
212 | * mode and device number to use. | |
213 | */ | |
214 | #define WHITEOUT_MODE 0 | |
215 | #define WHITEOUT_DEV 0 | |
216 | ||
1da177e4 LT |
217 | /* |
218 | * This is the Inode Attributes structure, used for notify_change(). It | |
219 | * uses the above definitions as flags, to know which values have changed. | |
220 | * Also, in this manner, a Filesystem can look at only the values it cares | |
221 | * about. Basically, these are the attributes that the VFS layer can | |
222 | * request to change from the FS layer. | |
223 | * | |
224 | * Derek Atkins <warlord@MIT.EDU> 94-10-20 | |
225 | */ | |
226 | struct iattr { | |
227 | unsigned int ia_valid; | |
228 | umode_t ia_mode; | |
45c31150 CB |
229 | /* |
230 | * The two anonymous unions wrap structures with the same member. | |
231 | * | |
232 | * Filesystems raising FS_ALLOW_IDMAP need to use ia_vfs{g,u}id which | |
233 | * are a dedicated type requiring the filesystem to use the dedicated | |
234 | * helpers. Other filesystem can continue to use ia_{g,u}id until they | |
235 | * have been ported. | |
b27c82e1 CB |
236 | * |
237 | * They always contain the same value. In other words FS_ALLOW_IDMAP | |
238 | * pass down the same value on idmapped mounts as they would on regular | |
239 | * mounts. | |
45c31150 CB |
240 | */ |
241 | union { | |
242 | kuid_t ia_uid; | |
243 | vfsuid_t ia_vfsuid; | |
244 | }; | |
245 | union { | |
246 | kgid_t ia_gid; | |
247 | vfsgid_t ia_vfsgid; | |
248 | }; | |
1da177e4 | 249 | loff_t ia_size; |
95582b00 DD |
250 | struct timespec64 ia_atime; |
251 | struct timespec64 ia_mtime; | |
252 | struct timespec64 ia_ctime; | |
cc4e69de MS |
253 | |
254 | /* | |
25985edc | 255 | * Not an attribute, but an auxiliary info for filesystems wanting to |
cc4e69de MS |
256 | * implement an ftruncate() like method. NOTE: filesystem should |
257 | * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL). | |
258 | */ | |
259 | struct file *ia_file; | |
1da177e4 LT |
260 | }; |
261 | ||
1da177e4 LT |
262 | /* |
263 | * Includes for diskquotas. | |
264 | */ | |
265 | #include <linux/quota.h> | |
266 | ||
69c433ed MS |
267 | /* |
268 | * Maximum number of layers of fs stack. Needs to be limited to | |
269 | * prevent kernel stack overflow | |
270 | */ | |
271 | #define FILESYSTEM_MAX_STACK_DEPTH 2 | |
272 | ||
994fc28c ZB |
273 | /** |
274 | * enum positive_aop_returns - aop return codes with specific semantics | |
275 | * | |
276 | * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has | |
277 | * completed, that the page is still locked, and | |
278 | * should be considered active. The VM uses this hint | |
279 | * to return the page to the active list -- it won't | |
280 | * be a candidate for writeback again in the near | |
281 | * future. Other callers must be careful to unlock | |
282 | * the page if they get this return. Returned by | |
283 | * writepage(); | |
284 | * | |
285 | * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has | |
286 | * unlocked it and the page might have been truncated. | |
287 | * The caller should back up to acquiring a new page and | |
288 | * trying again. The aop will be taking reasonable | |
289 | * precautions not to livelock. If the caller held a page | |
290 | * reference, it should drop it before retrying. Returned | |
7e0a1265 | 291 | * by read_folio(). |
994fc28c ZB |
292 | * |
293 | * address_space_operation functions return these large constants to indicate | |
294 | * special semantics to the caller. These are much larger than the bytes in a | |
295 | * page to allow for functions that return the number of bytes operated on in a | |
296 | * given page. | |
297 | */ | |
298 | ||
299 | enum positive_aop_returns { | |
300 | AOP_WRITEPAGE_ACTIVATE = 0x80000, | |
301 | AOP_TRUNCATED_PAGE = 0x80001, | |
302 | }; | |
303 | ||
1da177e4 LT |
304 | /* |
305 | * oh the beauties of C type declarations. | |
306 | */ | |
307 | struct page; | |
308 | struct address_space; | |
309 | struct writeback_control; | |
8151b4c8 | 310 | struct readahead_control; |
1da177e4 | 311 | |
c75b1d94 JA |
312 | /* |
313 | * Write life time hint values. | |
69448867 | 314 | * Stored in struct inode as u8. |
c75b1d94 JA |
315 | */ |
316 | enum rw_hint { | |
317 | WRITE_LIFE_NOT_SET = 0, | |
318 | WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE, | |
319 | WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT, | |
320 | WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM, | |
321 | WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG, | |
322 | WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME, | |
323 | }; | |
324 | ||
ce71bfea JA |
325 | /* Match RWF_* bits to IOCB bits */ |
326 | #define IOCB_HIPRI (__force int) RWF_HIPRI | |
327 | #define IOCB_DSYNC (__force int) RWF_DSYNC | |
328 | #define IOCB_SYNC (__force int) RWF_SYNC | |
329 | #define IOCB_NOWAIT (__force int) RWF_NOWAIT | |
330 | #define IOCB_APPEND (__force int) RWF_APPEND | |
331 | ||
332 | /* non-RWF related bits - start at 16 */ | |
333 | #define IOCB_EVENTFD (1 << 16) | |
334 | #define IOCB_DIRECT (1 << 17) | |
335 | #define IOCB_WRITE (1 << 18) | |
dd3e6d50 | 336 | /* iocb->ki_waitq is valid */ |
ce71bfea JA |
337 | #define IOCB_WAITQ (1 << 19) |
338 | #define IOCB_NOIO (1 << 20) | |
6c7ef543 JA |
339 | /* can use bio alloc cache */ |
340 | #define IOCB_ALLOC_CACHE (1 << 21) | |
9cf3516c JA |
341 | /* |
342 | * IOCB_DIO_CALLER_COMP can be set by the iocb owner, to indicate that the | |
343 | * iocb completion can be passed back to the owner for execution from a safe | |
344 | * context rather than needing to be punted through a workqueue. If this | |
345 | * flag is set, the bio completion handling may set iocb->dio_complete to a | |
346 | * handler function and iocb->private to context information for that handler. | |
347 | * The issuer should call the handler with that context information from task | |
348 | * context to complete the processing of the iocb. Note that while this | |
349 | * provides a task context for the dio_complete() callback, it should only be | |
350 | * used on the completion side for non-IO generating completions. It's fine to | |
351 | * call blocking functions from this callback, but they should not wait for | |
352 | * unrelated IO (like cache flushing, new IO generation, etc). | |
353 | */ | |
354 | #define IOCB_DIO_CALLER_COMP (1 << 22) | |
e2e40f2c | 355 | |
f6c73a11 RHI |
356 | /* for use in trace events */ |
357 | #define TRACE_IOCB_STRINGS \ | |
358 | { IOCB_HIPRI, "HIPRI" }, \ | |
359 | { IOCB_DSYNC, "DSYNC" }, \ | |
360 | { IOCB_SYNC, "SYNC" }, \ | |
361 | { IOCB_NOWAIT, "NOWAIT" }, \ | |
362 | { IOCB_APPEND, "APPEND" }, \ | |
363 | { IOCB_EVENTFD, "EVENTFD"}, \ | |
364 | { IOCB_DIRECT, "DIRECT" }, \ | |
365 | { IOCB_WRITE, "WRITE" }, \ | |
366 | { IOCB_WAITQ, "WAITQ" }, \ | |
367 | { IOCB_NOIO, "NOIO" }, \ | |
9cf3516c JA |
368 | { IOCB_ALLOC_CACHE, "ALLOC_CACHE" }, \ |
369 | { IOCB_DIO_CALLER_COMP, "CALLER_COMP" } | |
f6c73a11 | 370 | |
e2e40f2c CH |
371 | struct kiocb { |
372 | struct file *ki_filp; | |
373 | loff_t ki_pos; | |
6b19b766 | 374 | void (*ki_complete)(struct kiocb *iocb, long ret); |
e2e40f2c CH |
375 | void *private; |
376 | int ki_flags; | |
d9a08a9e | 377 | u16 ki_ioprio; /* See linux/ioprio.h */ |
9cf3516c JA |
378 | union { |
379 | /* | |
380 | * Only used for async buffered reads, where it denotes the | |
381 | * page waitqueue associated with completing the read. Valid | |
382 | * IFF IOCB_WAITQ is set. | |
383 | */ | |
384 | struct wait_page_queue *ki_waitq; | |
385 | /* | |
386 | * Can be used for O_DIRECT IO, where the completion handling | |
387 | * is punted back to the issuer of the IO. May only be set | |
388 | * if IOCB_DIO_CALLER_COMP is set by the issuer, and the issuer | |
389 | * must then check for presence of this handler when ki_complete | |
390 | * is invoked. The data passed in to this handler must be | |
391 | * assigned to ->private when dio_complete is assigned. | |
392 | */ | |
393 | ssize_t (*dio_complete)(void *data); | |
394 | }; | |
84c4e1f8 | 395 | }; |
e2e40f2c CH |
396 | |
397 | static inline bool is_sync_kiocb(struct kiocb *kiocb) | |
398 | { | |
399 | return kiocb->ki_complete == NULL; | |
400 | } | |
401 | ||
1da177e4 LT |
402 | struct address_space_operations { |
403 | int (*writepage)(struct page *page, struct writeback_control *wbc); | |
5efe7448 | 404 | int (*read_folio)(struct file *, struct folio *); |
1da177e4 LT |
405 | |
406 | /* Write back some dirty pages from this mapping. */ | |
407 | int (*writepages)(struct address_space *, struct writeback_control *); | |
408 | ||
3a3bae50 | 409 | /* Mark a folio dirty. Return true if this dirtied it */ |
6f31a5a2 | 410 | bool (*dirty_folio)(struct address_space *, struct folio *); |
1da177e4 | 411 | |
8151b4c8 | 412 | void (*readahead)(struct readahead_control *); |
1da177e4 | 413 | |
afddba49 | 414 | int (*write_begin)(struct file *, struct address_space *mapping, |
9d6b0cd7 | 415 | loff_t pos, unsigned len, |
afddba49 NP |
416 | struct page **pagep, void **fsdata); |
417 | int (*write_end)(struct file *, struct address_space *mapping, | |
418 | loff_t pos, unsigned len, unsigned copied, | |
419 | struct page *page, void *fsdata); | |
420 | ||
1da177e4 LT |
421 | /* Unfortunately this kludge is needed for FIBMAP. Don't use it */ |
422 | sector_t (*bmap)(struct address_space *, sector_t); | |
128d1f82 | 423 | void (*invalidate_folio) (struct folio *, size_t offset, size_t len); |
fa29000b | 424 | bool (*release_folio)(struct folio *, gfp_t); |
d2329aa0 | 425 | void (*free_folio)(struct folio *folio); |
c8b8e32d | 426 | ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter); |
b969c4ab | 427 | /* |
5490da4f | 428 | * migrate the contents of a folio to the specified target. If |
ef277c73 | 429 | * migrate_mode is MIGRATE_ASYNC, it must not block. |
b969c4ab | 430 | */ |
5490da4f MWO |
431 | int (*migrate_folio)(struct address_space *, struct folio *dst, |
432 | struct folio *src, enum migrate_mode); | |
072acba6 | 433 | int (*launder_folio)(struct folio *); |
2e7e80f7 MWO |
434 | bool (*is_partially_uptodate) (struct folio *, size_t from, |
435 | size_t count); | |
520f301c | 436 | void (*is_dirty_writeback) (struct folio *, bool *dirty, bool *wb); |
af7628d6 | 437 | int (*error_remove_folio)(struct address_space *, struct folio *); |
62c230bc MG |
438 | |
439 | /* swapfile support */ | |
a509bc1a MG |
440 | int (*swap_activate)(struct swap_info_struct *sis, struct file *file, |
441 | sector_t *span); | |
442 | void (*swap_deactivate)(struct file *file); | |
e1209d3a | 443 | int (*swap_rw)(struct kiocb *iocb, struct iov_iter *iter); |
1da177e4 LT |
444 | }; |
445 | ||
7dcda1c9 JA |
446 | extern const struct address_space_operations empty_aops; |
447 | ||
eb797a8e MW |
448 | /** |
449 | * struct address_space - Contents of a cacheable, mappable object. | |
450 | * @host: Owner, either the inode or the block_device. | |
451 | * @i_pages: Cached pages. | |
730633f0 JK |
452 | * @invalidate_lock: Guards coherency between page cache contents and |
453 | * file offset->disk block mappings in the filesystem during invalidates. | |
454 | * It is also used to block modification of page cache contents through | |
455 | * memory mappings. | |
eb797a8e | 456 | * @gfp_mask: Memory allocation flags to use for allocating pages. |
e8e17ee9 | 457 | * @i_mmap_writable: Number of VM_SHARED, VM_MAYWRITE mappings. |
09d91cda | 458 | * @nr_thps: Number of THPs in the pagecache (non-shmem only). |
eb797a8e MW |
459 | * @i_mmap: Tree of private and shared mappings. |
460 | * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable. | |
461 | * @nrpages: Number of page entries, protected by the i_pages lock. | |
eb797a8e MW |
462 | * @writeback_index: Writeback starts here. |
463 | * @a_ops: Methods. | |
464 | * @flags: Error bits and flags (AS_*). | |
465 | * @wb_err: The most recent error which has occurred. | |
600f111e MWO |
466 | * @i_private_lock: For use by the owner of the address_space. |
467 | * @i_private_list: For use by the owner of the address_space. | |
468 | * @i_private_data: For use by the owner of the address_space. | |
eb797a8e | 469 | */ |
1da177e4 | 470 | struct address_space { |
eb797a8e MW |
471 | struct inode *host; |
472 | struct xarray i_pages; | |
730633f0 | 473 | struct rw_semaphore invalidate_lock; |
eb797a8e MW |
474 | gfp_t gfp_mask; |
475 | atomic_t i_mmap_writable; | |
09d91cda SL |
476 | #ifdef CONFIG_READ_ONLY_THP_FOR_FS |
477 | /* number of thp, only for non-shmem files */ | |
478 | atomic_t nr_thps; | |
479 | #endif | |
eb797a8e | 480 | struct rb_root_cached i_mmap; |
eb797a8e | 481 | unsigned long nrpages; |
eb797a8e MW |
482 | pgoff_t writeback_index; |
483 | const struct address_space_operations *a_ops; | |
484 | unsigned long flags; | |
aee79d4e | 485 | struct rw_semaphore i_mmap_rwsem; |
5660e13d | 486 | errseq_t wb_err; |
600f111e MWO |
487 | spinlock_t i_private_lock; |
488 | struct list_head i_private_list; | |
489 | void * i_private_data; | |
3859a271 | 490 | } __attribute__((aligned(sizeof(long)))) __randomize_layout; |
1da177e4 LT |
491 | /* |
492 | * On most architectures that alignment is already the case; but | |
25985edc | 493 | * must be enforced here for CRIS, to let the least significant bit |
1da177e4 LT |
494 | * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON. |
495 | */ | |
1da177e4 | 496 | |
ff9c745b MW |
497 | /* XArray tags, for tagging dirty and writeback pages in the pagecache. */ |
498 | #define PAGECACHE_TAG_DIRTY XA_MARK_0 | |
499 | #define PAGECACHE_TAG_WRITEBACK XA_MARK_1 | |
500 | #define PAGECACHE_TAG_TOWRITE XA_MARK_2 | |
501 | ||
1da177e4 | 502 | /* |
ff9c745b | 503 | * Returns true if any of the pages in the mapping are marked with the tag. |
1da177e4 | 504 | */ |
ff9c745b MW |
505 | static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag) |
506 | { | |
507 | return xa_marked(&mapping->i_pages, tag); | |
508 | } | |
1da177e4 | 509 | |
8b28f621 DB |
510 | static inline void i_mmap_lock_write(struct address_space *mapping) |
511 | { | |
c8c06efa | 512 | down_write(&mapping->i_mmap_rwsem); |
8b28f621 DB |
513 | } |
514 | ||
c0d0381a MK |
515 | static inline int i_mmap_trylock_write(struct address_space *mapping) |
516 | { | |
517 | return down_write_trylock(&mapping->i_mmap_rwsem); | |
518 | } | |
519 | ||
8b28f621 DB |
520 | static inline void i_mmap_unlock_write(struct address_space *mapping) |
521 | { | |
c8c06efa | 522 | up_write(&mapping->i_mmap_rwsem); |
8b28f621 DB |
523 | } |
524 | ||
6d4675e6 MK |
525 | static inline int i_mmap_trylock_read(struct address_space *mapping) |
526 | { | |
527 | return down_read_trylock(&mapping->i_mmap_rwsem); | |
528 | } | |
529 | ||
3dec0ba0 DB |
530 | static inline void i_mmap_lock_read(struct address_space *mapping) |
531 | { | |
532 | down_read(&mapping->i_mmap_rwsem); | |
533 | } | |
534 | ||
535 | static inline void i_mmap_unlock_read(struct address_space *mapping) | |
536 | { | |
537 | up_read(&mapping->i_mmap_rwsem); | |
538 | } | |
539 | ||
34ae204f MK |
540 | static inline void i_mmap_assert_locked(struct address_space *mapping) |
541 | { | |
542 | lockdep_assert_held(&mapping->i_mmap_rwsem); | |
543 | } | |
544 | ||
545 | static inline void i_mmap_assert_write_locked(struct address_space *mapping) | |
546 | { | |
547 | lockdep_assert_held_write(&mapping->i_mmap_rwsem); | |
548 | } | |
549 | ||
1da177e4 LT |
550 | /* |
551 | * Might pages of this file be mapped into userspace? | |
552 | */ | |
553 | static inline int mapping_mapped(struct address_space *mapping) | |
554 | { | |
f808c13f | 555 | return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root); |
1da177e4 LT |
556 | } |
557 | ||
558 | /* | |
559 | * Might pages of this file have been modified in userspace? | |
e8e17ee9 | 560 | * Note that i_mmap_writable counts all VM_SHARED, VM_MAYWRITE vmas: do_mmap |
1da177e4 LT |
561 | * marks vma as VM_SHARED if it is shared, and the file was opened for |
562 | * writing i.e. vma may be mprotected writable even if now readonly. | |
4bb5f5d9 DR |
563 | * |
564 | * If i_mmap_writable is negative, no new writable mappings are allowed. You | |
565 | * can only deny writable mappings, if none exists right now. | |
1da177e4 LT |
566 | */ |
567 | static inline int mapping_writably_mapped(struct address_space *mapping) | |
568 | { | |
4bb5f5d9 DR |
569 | return atomic_read(&mapping->i_mmap_writable) > 0; |
570 | } | |
571 | ||
572 | static inline int mapping_map_writable(struct address_space *mapping) | |
573 | { | |
574 | return atomic_inc_unless_negative(&mapping->i_mmap_writable) ? | |
575 | 0 : -EPERM; | |
576 | } | |
577 | ||
578 | static inline void mapping_unmap_writable(struct address_space *mapping) | |
579 | { | |
580 | atomic_dec(&mapping->i_mmap_writable); | |
581 | } | |
582 | ||
583 | static inline int mapping_deny_writable(struct address_space *mapping) | |
584 | { | |
585 | return atomic_dec_unless_positive(&mapping->i_mmap_writable) ? | |
586 | 0 : -EBUSY; | |
587 | } | |
588 | ||
589 | static inline void mapping_allow_writable(struct address_space *mapping) | |
590 | { | |
591 | atomic_inc(&mapping->i_mmap_writable); | |
1da177e4 LT |
592 | } |
593 | ||
594 | /* | |
595 | * Use sequence counter to get consistent i_size on 32-bit processors. | |
596 | */ | |
597 | #if BITS_PER_LONG==32 && defined(CONFIG_SMP) | |
598 | #include <linux/seqlock.h> | |
599 | #define __NEED_I_SIZE_ORDERED | |
600 | #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount) | |
601 | #else | |
602 | #define i_size_ordered_init(inode) do { } while (0) | |
603 | #endif | |
604 | ||
f19d4a8f AV |
605 | struct posix_acl; |
606 | #define ACL_NOT_CACHED ((void *)(-1)) | |
332f606b MS |
607 | /* |
608 | * ACL_DONT_CACHE is for stacked filesystems, that rely on underlying fs to | |
cac2f8b8 CB |
609 | * cache the ACL. This also means that ->get_inode_acl() can be called in RCU |
610 | * mode with the LOOKUP_RCU flag. | |
332f606b | 611 | */ |
2a3a2a3f | 612 | #define ACL_DONT_CACHE ((void *)(-3)) |
f19d4a8f | 613 | |
b8a7a3a6 AG |
614 | static inline struct posix_acl * |
615 | uncached_acl_sentinel(struct task_struct *task) | |
616 | { | |
617 | return (void *)task + 1; | |
618 | } | |
619 | ||
620 | static inline bool | |
621 | is_uncached_acl(struct posix_acl *acl) | |
622 | { | |
623 | return (long)acl & 1; | |
624 | } | |
625 | ||
3ddcd056 LT |
626 | #define IOP_FASTPERM 0x0001 |
627 | #define IOP_LOOKUP 0x0002 | |
628 | #define IOP_NOFOLLOW 0x0004 | |
d0a5b995 | 629 | #define IOP_XATTR 0x0008 |
76fca90e | 630 | #define IOP_DEFAULT_READLINK 0x0010 |
3ddcd056 | 631 | |
9dd813c1 JK |
632 | struct fsnotify_mark_connector; |
633 | ||
3ddcd056 LT |
634 | /* |
635 | * Keep mostly read-only and often accessed (especially for | |
636 | * the RCU path lookup and 'stat' data) fields at the beginning | |
637 | * of the 'struct inode' | |
638 | */ | |
1da177e4 | 639 | struct inode { |
44a7d7a8 | 640 | umode_t i_mode; |
3ddcd056 | 641 | unsigned short i_opflags; |
92361636 EB |
642 | kuid_t i_uid; |
643 | kgid_t i_gid; | |
3ddcd056 LT |
644 | unsigned int i_flags; |
645 | ||
646 | #ifdef CONFIG_FS_POSIX_ACL | |
647 | struct posix_acl *i_acl; | |
648 | struct posix_acl *i_default_acl; | |
649 | #endif | |
650 | ||
44a7d7a8 NP |
651 | const struct inode_operations *i_op; |
652 | struct super_block *i_sb; | |
3ddcd056 | 653 | struct address_space *i_mapping; |
44a7d7a8 | 654 | |
13e12d14 LT |
655 | #ifdef CONFIG_SECURITY |
656 | void *i_security; | |
657 | #endif | |
44a7d7a8 | 658 | |
3ddcd056 LT |
659 | /* Stat data, not accessed from path walking */ |
660 | unsigned long i_ino; | |
a78ef704 MS |
661 | /* |
662 | * Filesystems may only read i_nlink directly. They shall use the | |
663 | * following functions for modification: | |
664 | * | |
665 | * (set|clear|inc|drop)_nlink | |
666 | * inode_(inc|dec)_link_count | |
667 | */ | |
668 | union { | |
669 | const unsigned int i_nlink; | |
670 | unsigned int __i_nlink; | |
671 | }; | |
3ddcd056 | 672 | dev_t i_rdev; |
2f9d3df8 | 673 | loff_t i_size; |
12cd4402 JL |
674 | struct timespec64 __i_atime; |
675 | struct timespec64 __i_mtime; | |
13bc2445 | 676 | struct timespec64 __i_ctime; /* use inode_*_ctime accessors! */ |
6cdbb0ef TT |
677 | spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */ |
678 | unsigned short i_bytes; | |
69448867 AG |
679 | u8 i_blkbits; |
680 | u8 i_write_hint; | |
3ddcd056 LT |
681 | blkcnt_t i_blocks; |
682 | ||
683 | #ifdef __NEED_I_SIZE_ORDERED | |
684 | seqcount_t i_size_seqcount; | |
685 | #endif | |
686 | ||
687 | /* Misc */ | |
688 | unsigned long i_state; | |
9902af79 | 689 | struct rw_semaphore i_rwsem; |
13e12d14 | 690 | |
44a7d7a8 | 691 | unsigned long dirtied_when; /* jiffies of first dirtying */ |
a2f48706 | 692 | unsigned long dirtied_time_when; |
44a7d7a8 | 693 | |
1da177e4 | 694 | struct hlist_node i_hash; |
c7f54084 | 695 | struct list_head i_io_list; /* backing dev IO list */ |
52ebea74 TH |
696 | #ifdef CONFIG_CGROUP_WRITEBACK |
697 | struct bdi_writeback *i_wb; /* the associated cgroup wb */ | |
2a814908 TH |
698 | |
699 | /* foreign inode detection, see wbc_detach_inode() */ | |
700 | int i_wb_frn_winner; | |
701 | u16 i_wb_frn_avg_time; | |
702 | u16 i_wb_frn_history; | |
52ebea74 | 703 | #endif |
7ccf19a8 | 704 | struct list_head i_lru; /* inode LRU list */ |
1da177e4 | 705 | struct list_head i_sb_list; |
6c60d2b5 | 706 | struct list_head i_wb_list; /* backing dev writeback list */ |
fa0d7e3d | 707 | union { |
b3d9b7a3 | 708 | struct hlist_head i_dentry; |
fa0d7e3d NP |
709 | struct rcu_head i_rcu; |
710 | }; | |
f02a9ad1 | 711 | atomic64_t i_version; |
8019ad13 | 712 | atomic64_t i_sequence; /* see futex */ |
2f9d3df8 | 713 | atomic_t i_count; |
bd5fe6c5 | 714 | atomic_t i_dio_count; |
6cdbb0ef | 715 | atomic_t i_writecount; |
387e3746 | 716 | #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING) |
d984ea60 MZ |
717 | atomic_t i_readcount; /* struct files open RO */ |
718 | #endif | |
fdb0da89 AV |
719 | union { |
720 | const struct file_operations *i_fop; /* former ->i_op->default_file_ops */ | |
721 | void (*free_inode)(struct inode *); | |
722 | }; | |
4a075e39 | 723 | struct file_lock_context *i_flctx; |
1da177e4 | 724 | struct address_space i_data; |
1da177e4 | 725 | struct list_head i_devices; |
4c154168 TT |
726 | union { |
727 | struct pipe_inode_info *i_pipe; | |
577c4eb0 | 728 | struct cdev *i_cdev; |
61ba64fc | 729 | char *i_link; |
84e710da | 730 | unsigned i_dir_seq; |
4c154168 | 731 | }; |
1da177e4 LT |
732 | |
733 | __u32 i_generation; | |
734 | ||
3be25f49 EP |
735 | #ifdef CONFIG_FSNOTIFY |
736 | __u32 i_fsnotify_mask; /* all events this inode cares about */ | |
08991e83 | 737 | struct fsnotify_mark_connector __rcu *i_fsnotify_marks; |
0eeca283 RL |
738 | #endif |
739 | ||
643fa961 | 740 | #ifdef CONFIG_FS_ENCRYPTION |
3e7807d5 | 741 | struct fscrypt_inode_info *i_crypt_info; |
0b81d077 JK |
742 | #endif |
743 | ||
5585f2af EB |
744 | #ifdef CONFIG_FS_VERITY |
745 | struct fsverity_info *i_verity_info; | |
746 | #endif | |
747 | ||
8e18e294 | 748 | void *i_private; /* fs or device private pointer */ |
3859a271 | 749 | } __randomize_layout; |
1da177e4 | 750 | |
50e17c00 DD |
751 | struct timespec64 timestamp_truncate(struct timespec64 t, struct inode *inode); |
752 | ||
93407472 FF |
753 | static inline unsigned int i_blocksize(const struct inode *node) |
754 | { | |
755 | return (1 << node->i_blkbits); | |
756 | } | |
757 | ||
1d3382cb AV |
758 | static inline int inode_unhashed(struct inode *inode) |
759 | { | |
760 | return hlist_unhashed(&inode->i_hash); | |
761 | } | |
762 | ||
5bef9151 AV |
763 | /* |
764 | * __mark_inode_dirty expects inodes to be hashed. Since we don't | |
765 | * want special inodes in the fileset inode space, we make them | |
766 | * appear hashed, but do not put on any lists. hlist_del() | |
767 | * will work fine and require no locking. | |
768 | */ | |
769 | static inline void inode_fake_hash(struct inode *inode) | |
770 | { | |
771 | hlist_add_fake(&inode->i_hash); | |
772 | } | |
773 | ||
f2eace23 IM |
774 | /* |
775 | * inode->i_mutex nesting subclasses for the lock validator: | |
776 | * | |
777 | * 0: the object of the current VFS operation | |
778 | * 1: parent | |
779 | * 2: child/target | |
40bd22c9 BF |
780 | * 3: xattr |
781 | * 4: second non-directory | |
d1b72cc6 MS |
782 | * 5: second parent (when locking independent directories in rename) |
783 | * | |
784 | * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two | |
40bd22c9 | 785 | * non-directories at once. |
f2eace23 IM |
786 | * |
787 | * The locking order between these classes is | |
d1b72cc6 | 788 | * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory |
f2eace23 IM |
789 | */ |
790 | enum inode_i_mutex_lock_class | |
791 | { | |
792 | I_MUTEX_NORMAL, | |
793 | I_MUTEX_PARENT, | |
794 | I_MUTEX_CHILD, | |
4df46240 | 795 | I_MUTEX_XATTR, |
d1b72cc6 MS |
796 | I_MUTEX_NONDIR2, |
797 | I_MUTEX_PARENT2, | |
f2eace23 IM |
798 | }; |
799 | ||
5955102c AV |
800 | static inline void inode_lock(struct inode *inode) |
801 | { | |
9902af79 | 802 | down_write(&inode->i_rwsem); |
5955102c AV |
803 | } |
804 | ||
805 | static inline void inode_unlock(struct inode *inode) | |
806 | { | |
9902af79 AV |
807 | up_write(&inode->i_rwsem); |
808 | } | |
809 | ||
810 | static inline void inode_lock_shared(struct inode *inode) | |
811 | { | |
812 | down_read(&inode->i_rwsem); | |
813 | } | |
814 | ||
815 | static inline void inode_unlock_shared(struct inode *inode) | |
816 | { | |
817 | up_read(&inode->i_rwsem); | |
5955102c AV |
818 | } |
819 | ||
820 | static inline int inode_trylock(struct inode *inode) | |
821 | { | |
9902af79 AV |
822 | return down_write_trylock(&inode->i_rwsem); |
823 | } | |
824 | ||
825 | static inline int inode_trylock_shared(struct inode *inode) | |
826 | { | |
827 | return down_read_trylock(&inode->i_rwsem); | |
5955102c AV |
828 | } |
829 | ||
830 | static inline int inode_is_locked(struct inode *inode) | |
831 | { | |
9902af79 | 832 | return rwsem_is_locked(&inode->i_rwsem); |
5955102c AV |
833 | } |
834 | ||
835 | static inline void inode_lock_nested(struct inode *inode, unsigned subclass) | |
836 | { | |
9902af79 | 837 | down_write_nested(&inode->i_rwsem, subclass); |
5955102c AV |
838 | } |
839 | ||
01c2e13d DW |
840 | static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass) |
841 | { | |
842 | down_read_nested(&inode->i_rwsem, subclass); | |
843 | } | |
844 | ||
730633f0 JK |
845 | static inline void filemap_invalidate_lock(struct address_space *mapping) |
846 | { | |
847 | down_write(&mapping->invalidate_lock); | |
848 | } | |
849 | ||
850 | static inline void filemap_invalidate_unlock(struct address_space *mapping) | |
851 | { | |
852 | up_write(&mapping->invalidate_lock); | |
853 | } | |
854 | ||
855 | static inline void filemap_invalidate_lock_shared(struct address_space *mapping) | |
856 | { | |
857 | down_read(&mapping->invalidate_lock); | |
858 | } | |
859 | ||
860 | static inline int filemap_invalidate_trylock_shared( | |
861 | struct address_space *mapping) | |
862 | { | |
863 | return down_read_trylock(&mapping->invalidate_lock); | |
864 | } | |
865 | ||
866 | static inline void filemap_invalidate_unlock_shared( | |
867 | struct address_space *mapping) | |
868 | { | |
869 | up_read(&mapping->invalidate_lock); | |
870 | } | |
871 | ||
375e289e BF |
872 | void lock_two_nondirectories(struct inode *, struct inode*); |
873 | void unlock_two_nondirectories(struct inode *, struct inode*); | |
874 | ||
7506ae6a JK |
875 | void filemap_invalidate_lock_two(struct address_space *mapping1, |
876 | struct address_space *mapping2); | |
877 | void filemap_invalidate_unlock_two(struct address_space *mapping1, | |
878 | struct address_space *mapping2); | |
879 | ||
880 | ||
1da177e4 LT |
881 | /* |
882 | * NOTE: in a 32bit arch with a preemptable kernel and | |
883 | * an UP compile the i_size_read/write must be atomic | |
884 | * with respect to the local cpu (unlike with preempt disabled), | |
885 | * but they don't need to be atomic with respect to other cpus like in | |
886 | * true SMP (so they need either to either locally disable irq around | |
887 | * the read or for example on x86 they can be still implemented as a | |
888 | * cmpxchg8b without the need of the lock prefix). For SMP compiles | |
889 | * and 64bit archs it makes no difference if preempt is enabled or not. | |
890 | */ | |
48ed214d | 891 | static inline loff_t i_size_read(const struct inode *inode) |
1da177e4 LT |
892 | { |
893 | #if BITS_PER_LONG==32 && defined(CONFIG_SMP) | |
894 | loff_t i_size; | |
895 | unsigned int seq; | |
896 | ||
897 | do { | |
898 | seq = read_seqcount_begin(&inode->i_size_seqcount); | |
899 | i_size = inode->i_size; | |
900 | } while (read_seqcount_retry(&inode->i_size_seqcount, seq)); | |
901 | return i_size; | |
2496396f | 902 | #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION) |
1da177e4 LT |
903 | loff_t i_size; |
904 | ||
905 | preempt_disable(); | |
906 | i_size = inode->i_size; | |
907 | preempt_enable(); | |
908 | return i_size; | |
909 | #else | |
910 | return inode->i_size; | |
911 | #endif | |
912 | } | |
913 | ||
7762f5a0 MS |
914 | /* |
915 | * NOTE: unlike i_size_read(), i_size_write() does need locking around it | |
916 | * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount | |
917 | * can be lost, resulting in subsequent i_size_read() calls spinning forever. | |
918 | */ | |
1da177e4 LT |
919 | static inline void i_size_write(struct inode *inode, loff_t i_size) |
920 | { | |
921 | #if BITS_PER_LONG==32 && defined(CONFIG_SMP) | |
74e3d1e1 | 922 | preempt_disable(); |
1da177e4 LT |
923 | write_seqcount_begin(&inode->i_size_seqcount); |
924 | inode->i_size = i_size; | |
925 | write_seqcount_end(&inode->i_size_seqcount); | |
74e3d1e1 | 926 | preempt_enable(); |
2496396f | 927 | #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION) |
1da177e4 LT |
928 | preempt_disable(); |
929 | inode->i_size = i_size; | |
930 | preempt_enable(); | |
931 | #else | |
932 | inode->i_size = i_size; | |
933 | #endif | |
934 | } | |
935 | ||
48ed214d | 936 | static inline unsigned iminor(const struct inode *inode) |
1da177e4 LT |
937 | { |
938 | return MINOR(inode->i_rdev); | |
939 | } | |
940 | ||
48ed214d | 941 | static inline unsigned imajor(const struct inode *inode) |
1da177e4 LT |
942 | { |
943 | return MAJOR(inode->i_rdev); | |
944 | } | |
945 | ||
1da177e4 LT |
946 | struct fown_struct { |
947 | rwlock_t lock; /* protects pid, uid, euid fields */ | |
609d7fa9 EB |
948 | struct pid *pid; /* pid or -pgrp where SIGIO should be sent */ |
949 | enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */ | |
92361636 | 950 | kuid_t uid, euid; /* uid/euid of process setting the owner */ |
1da177e4 LT |
951 | int signum; /* posix.1b rt signal to be delivered on IO */ |
952 | }; | |
953 | ||
c790fbf2 MWO |
954 | /** |
955 | * struct file_ra_state - Track a file's readahead state. | |
956 | * @start: Where the most recent readahead started. | |
957 | * @size: Number of pages read in the most recent readahead. | |
84dacdbd N |
958 | * @async_size: Numer of pages that were/are not needed immediately |
959 | * and so were/are genuinely "ahead". Start next readahead when | |
960 | * the first of these pages is accessed. | |
961 | * @ra_pages: Maximum size of a readahead request, copied from the bdi. | |
c790fbf2 MWO |
962 | * @mmap_miss: How many mmap accesses missed in the page cache. |
963 | * @prev_pos: The last byte in the most recent read request. | |
84dacdbd N |
964 | * |
965 | * When this structure is passed to ->readahead(), the "most recent" | |
966 | * readahead means the current readahead. | |
1da177e4 LT |
967 | */ |
968 | struct file_ra_state { | |
c790fbf2 MWO |
969 | pgoff_t start; |
970 | unsigned int size; | |
971 | unsigned int async_size; | |
972 | unsigned int ra_pages; | |
973 | unsigned int mmap_miss; | |
974 | loff_t prev_pos; | |
1da177e4 | 975 | }; |
1da177e4 | 976 | |
5ce1110b FW |
977 | /* |
978 | * Check if @index falls in the readahead windows. | |
979 | */ | |
980 | static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index) | |
981 | { | |
f9acc8c7 FW |
982 | return (index >= ra->start && |
983 | index < ra->start + ra->size); | |
5ce1110b FW |
984 | } |
985 | ||
a7bc2e8d | 986 | /* |
987 | * f_{lock,count,pos_lock} members can be highly contended and share | |
988 | * the same cacheline. f_{lock,mode} are very frequently used together | |
989 | * and so share the same cacheline as well. The read-mostly | |
990 | * f_{path,inode,op} are kept on a separate cacheline. | |
991 | */ | |
1da177e4 | 992 | struct file { |
2f512016 | 993 | union { |
372a34e6 CB |
994 | /* fput() uses task work when closing and freeing file (default). */ |
995 | struct callback_head f_task_work; | |
996 | /* fput() must use workqueue (most kernel threads). */ | |
e87f2c26 | 997 | struct llist_node f_llist; |
164f4064 | 998 | unsigned int f_iocb_flags; |
e87f2c26 | 999 | }; |
ef3d0fd2 AK |
1000 | |
1001 | /* | |
319c1517 | 1002 | * Protects f_ep, f_flags. |
ef3d0fd2 AK |
1003 | * Must not be taken from IRQ context. |
1004 | */ | |
1005 | spinlock_t f_lock; | |
aeb5d727 | 1006 | fmode_t f_mode; |
a7bc2e8d | 1007 | atomic_long_t f_count; |
9c225f26 | 1008 | struct mutex f_pos_lock; |
1da177e4 | 1009 | loff_t f_pos; |
a7bc2e8d | 1010 | unsigned int f_flags; |
1da177e4 | 1011 | struct fown_struct f_owner; |
d76b0d9b | 1012 | const struct cred *f_cred; |
1da177e4 | 1013 | struct file_ra_state f_ra; |
a7bc2e8d | 1014 | struct path f_path; |
1015 | struct inode *f_inode; /* cached value */ | |
1016 | const struct file_operations *f_op; | |
1da177e4 | 1017 | |
2b47c361 | 1018 | u64 f_version; |
50462062 | 1019 | #ifdef CONFIG_SECURITY |
1da177e4 | 1020 | void *f_security; |
50462062 | 1021 | #endif |
1da177e4 LT |
1022 | /* needed for tty driver, and maybe others */ |
1023 | void *private_data; | |
1024 | ||
1025 | #ifdef CONFIG_EPOLL | |
1026 | /* Used by fs/eventpoll.c to link all the hooks to this file */ | |
319c1517 | 1027 | struct hlist_head *f_ep; |
1da177e4 LT |
1028 | #endif /* #ifdef CONFIG_EPOLL */ |
1029 | struct address_space *f_mapping; | |
5660e13d | 1030 | errseq_t f_wb_err; |
735e4ae5 | 1031 | errseq_t f_sb_err; /* for syncfs */ |
3859a271 KC |
1032 | } __randomize_layout |
1033 | __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */ | |
1da177e4 | 1034 | |
990d6c2d AK |
1035 | struct file_handle { |
1036 | __u32 handle_bytes; | |
1037 | int handle_type; | |
1038 | /* file identifier */ | |
9d82973e | 1039 | unsigned char f_handle[]; |
990d6c2d AK |
1040 | }; |
1041 | ||
cb0942b8 AV |
1042 | static inline struct file *get_file(struct file *f) |
1043 | { | |
1044 | atomic_long_inc(&f->f_count); | |
1045 | return f; | |
1046 | } | |
0ede61d8 CB |
1047 | |
1048 | struct file *get_file_rcu(struct file __rcu **f); | |
61d4fb0b | 1049 | struct file *get_file_active(struct file **f); |
0ede61d8 | 1050 | |
516e0cc5 | 1051 | #define file_count(x) atomic_long_read(&(x)->f_count) |
1da177e4 LT |
1052 | |
1053 | #define MAX_NON_LFS ((1UL<<31) - 1) | |
1054 | ||
1055 | /* Page cache limit. The filesystems should put that into their s_maxbytes | |
1056 | limits, otherwise bad things can happen in VM. */ | |
1057 | #if BITS_PER_LONG==32 | |
0cc3b0ec | 1058 | #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT) |
1da177e4 | 1059 | #elif BITS_PER_LONG==64 |
0cc3b0ec | 1060 | #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX) |
1da177e4 LT |
1061 | #endif |
1062 | ||
7ca76311 | 1063 | /* legacy typedef, should eventually be removed */ |
17fa388d | 1064 | typedef void *fl_owner_t; |
1da177e4 | 1065 | |
a7231a97 JL |
1066 | struct file_lock; |
1067 | ||
1da177e4 LT |
1068 | /* The following constant reflects the upper bound of the file/locking space */ |
1069 | #ifndef OFFSET_MAX | |
ea258f15 ZL |
1070 | #define OFFSET_MAX type_max(loff_t) |
1071 | #define OFFT_OFFSET_MAX type_max(off_t) | |
1da177e4 LT |
1072 | #endif |
1073 | ||
bfcd17a6 TP |
1074 | extern void send_sigio(struct fown_struct *fown, int fd, int band); |
1075 | ||
ee296d7c JL |
1076 | static inline struct inode *file_inode(const struct file *f) |
1077 | { | |
1078 | return f->f_inode; | |
1079 | } | |
1080 | ||
d101a125 MS |
1081 | static inline struct dentry *file_dentry(const struct file *file) |
1082 | { | |
fb16043b | 1083 | return d_real(file->f_path.dentry, file_inode(file)); |
d101a125 MS |
1084 | } |
1085 | ||
1da177e4 | 1086 | struct fasync_struct { |
7a107c0f | 1087 | rwlock_t fa_lock; |
989a2979 ED |
1088 | int magic; |
1089 | int fa_fd; | |
1090 | struct fasync_struct *fa_next; /* singly linked list */ | |
1091 | struct file *fa_file; | |
1092 | struct rcu_head fa_rcu; | |
1da177e4 LT |
1093 | }; |
1094 | ||
1095 | #define FASYNC_MAGIC 0x4601 | |
1096 | ||
1097 | /* SMP safe fasync helpers: */ | |
1098 | extern int fasync_helper(int, struct file *, int, struct fasync_struct **); | |
f7347ce4 LT |
1099 | extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *); |
1100 | extern int fasync_remove_entry(struct file *, struct fasync_struct **); | |
1101 | extern struct fasync_struct *fasync_alloc(void); | |
1102 | extern void fasync_free(struct fasync_struct *); | |
1103 | ||
1da177e4 LT |
1104 | /* can be called from interrupts */ |
1105 | extern void kill_fasync(struct fasync_struct **, int, int); | |
1da177e4 | 1106 | |
e0b93edd | 1107 | extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force); |
bccb5c39 | 1108 | extern int f_setown(struct file *filp, int who, int force); |
1da177e4 | 1109 | extern void f_delown(struct file *filp); |
609d7fa9 | 1110 | extern pid_t f_getown(struct file *filp); |
1da177e4 LT |
1111 | extern int send_sigurg(struct fown_struct *fown); |
1112 | ||
e462ec50 DH |
1113 | /* |
1114 | * sb->s_flags. Note that these mirror the equivalent MS_* flags where | |
1115 | * represented in both. | |
1116 | */ | |
f15afbd3 HG |
1117 | #define SB_RDONLY BIT(0) /* Mount read-only */ |
1118 | #define SB_NOSUID BIT(1) /* Ignore suid and sgid bits */ | |
1119 | #define SB_NODEV BIT(2) /* Disallow access to device special files */ | |
1120 | #define SB_NOEXEC BIT(3) /* Disallow program execution */ | |
1121 | #define SB_SYNCHRONOUS BIT(4) /* Writes are synced at once */ | |
1122 | #define SB_MANDLOCK BIT(6) /* Allow mandatory locks on an FS */ | |
1123 | #define SB_DIRSYNC BIT(7) /* Directory modifications are synchronous */ | |
1124 | #define SB_NOATIME BIT(10) /* Do not update access times. */ | |
1125 | #define SB_NODIRATIME BIT(11) /* Do not update directory access times */ | |
1126 | #define SB_SILENT BIT(15) | |
5aa8fd9c | 1127 | #define SB_POSIXACL BIT(16) /* Supports POSIX ACLs */ |
f15afbd3 HG |
1128 | #define SB_INLINECRYPT BIT(17) /* Use blk-crypto for encrypted files */ |
1129 | #define SB_KERNMOUNT BIT(22) /* this is a kern_mount call */ | |
1130 | #define SB_I_VERSION BIT(23) /* Update inode I_version field */ | |
1131 | #define SB_LAZYTIME BIT(25) /* Update the on-disk [acm]times lazily */ | |
e462ec50 DH |
1132 | |
1133 | /* These sb flags are internal to the kernel */ | |
2c18a63b | 1134 | #define SB_DEAD BIT(21) |
5e874914 | 1135 | #define SB_DYING BIT(24) |
f15afbd3 HG |
1136 | #define SB_SUBMOUNT BIT(26) |
1137 | #define SB_FORCE BIT(27) | |
1138 | #define SB_NOSEC BIT(28) | |
1139 | #define SB_BORN BIT(29) | |
1140 | #define SB_ACTIVE BIT(30) | |
1141 | #define SB_NOUSER BIT(31) | |
e462ec50 | 1142 | |
c843843e DR |
1143 | /* These flags relate to encoding and casefolding */ |
1144 | #define SB_ENC_STRICT_MODE_FL (1 << 0) | |
1145 | ||
1146 | #define sb_has_strict_encoding(sb) \ | |
1147 | (sb->s_encoding_flags & SB_ENC_STRICT_MODE_FL) | |
1148 | ||
1da177e4 LT |
1149 | /* |
1150 | * Umount options | |
1151 | */ | |
1152 | ||
1153 | #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */ | |
1154 | #define MNT_DETACH 0x00000002 /* Just detach from the tree */ | |
1155 | #define MNT_EXPIRE 0x00000004 /* Mark for expiry */ | |
db1f05bb MS |
1156 | #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */ |
1157 | #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */ | |
1da177e4 | 1158 | |
46b15caa TH |
1159 | /* sb->s_iflags */ |
1160 | #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */ | |
90f8572b | 1161 | #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */ |
a2982cc9 | 1162 | #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */ |
1cb039f3 | 1163 | #define SB_I_STABLE_WRITES 0x00000008 /* don't modify blks until WB is done */ |
1da177e4 | 1164 | |
8654df4e EB |
1165 | /* sb->s_iflags to limit user namespace mounts */ |
1166 | #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */ | |
57b56ac6 MZ |
1167 | #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020 |
1168 | #define SB_I_UNTRUSTED_MOUNTER 0x00000040 | |
cd708c93 | 1169 | #define SB_I_EVM_UNSUPPORTED 0x00000080 |
1da177e4 | 1170 | |
32b1924b | 1171 | #define SB_I_SKIP_SYNC 0x00000100 /* Skip superblock at global sync */ |
0b3ea092 | 1172 | #define SB_I_PERSB_BDI 0x00000200 /* has a per-sb bdi */ |
a128b054 | 1173 | #define SB_I_TS_EXPIRY_WARNED 0x00000400 /* warned about timestamp range expiry */ |
04b94071 | 1174 | #define SB_I_RETIRED 0x00000800 /* superblock shouldn't be reused */ |
5aa8fd9c | 1175 | #define SB_I_NOUMASK 0x00001000 /* VFS does not apply umask */ |
32b1924b | 1176 | |
5accdf82 JK |
1177 | /* Possible states of 'frozen' field */ |
1178 | enum { | |
1179 | SB_UNFROZEN = 0, /* FS is unfrozen */ | |
1180 | SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */ | |
5accdf82 JK |
1181 | SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */ |
1182 | SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop | |
1183 | * internal threads if needed) */ | |
1184 | SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */ | |
1185 | }; | |
1186 | ||
1187 | #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1) | |
1188 | ||
1189 | struct sb_writers { | |
880b9577 | 1190 | unsigned short frozen; /* Is sb frozen? */ |
7366f8b6 CB |
1191 | int freeze_kcount; /* How many kernel freeze requests? */ |
1192 | int freeze_ucount; /* How many userspace freeze requests? */ | |
8129ed29 | 1193 | struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS]; |
5accdf82 JK |
1194 | }; |
1195 | ||
1da177e4 LT |
1196 | struct super_block { |
1197 | struct list_head s_list; /* Keep this first */ | |
1198 | dev_t s_dev; /* search index; _not_ kdev_t */ | |
270ba5f7 RK |
1199 | unsigned char s_blocksize_bits; |
1200 | unsigned long s_blocksize; | |
42cb56ae | 1201 | loff_t s_maxbytes; /* Max file size */ |
1da177e4 | 1202 | struct file_system_type *s_type; |
ee9b6d61 | 1203 | const struct super_operations *s_op; |
61e225dc | 1204 | const struct dquot_operations *dq_op; |
0d54b217 | 1205 | const struct quotactl_ops *s_qcop; |
39655164 | 1206 | const struct export_operations *s_export_op; |
1da177e4 | 1207 | unsigned long s_flags; |
46b15caa | 1208 | unsigned long s_iflags; /* internal SB_I_* flags */ |
1da177e4 LT |
1209 | unsigned long s_magic; |
1210 | struct dentry *s_root; | |
1211 | struct rw_semaphore s_umount; | |
1da177e4 | 1212 | int s_count; |
1da177e4 | 1213 | atomic_t s_active; |
50462062 | 1214 | #ifdef CONFIG_SECURITY |
1da177e4 | 1215 | void *s_security; |
50462062 | 1216 | #endif |
e346fb6d | 1217 | const struct xattr_handler * const *s_xattr; |
643fa961 | 1218 | #ifdef CONFIG_FS_ENCRYPTION |
0b81d077 | 1219 | const struct fscrypt_operations *s_cop; |
d7e7b9af | 1220 | struct fscrypt_keyring *s_master_keys; /* master crypto keys in use */ |
5585f2af EB |
1221 | #endif |
1222 | #ifdef CONFIG_FS_VERITY | |
1223 | const struct fsverity_operations *s_vop; | |
c843843e | 1224 | #endif |
5298d4bf | 1225 | #if IS_ENABLED(CONFIG_UNICODE) |
c843843e DR |
1226 | struct unicode_map *s_encoding; |
1227 | __u16 s_encoding_flags; | |
bbbc3fb6 | 1228 | #endif |
f1ee6162 | 1229 | struct hlist_bl_head s_roots; /* alternate root dentries for NFS */ |
39f7c4db | 1230 | struct list_head s_mounts; /* list of mounts; _not_ for fs use */ |
f3a60882 CB |
1231 | struct block_device *s_bdev; /* can go away once we use an accessor for @s_bdev_file */ |
1232 | struct file *s_bdev_file; | |
32a88aa1 | 1233 | struct backing_dev_info *s_bdi; |
acaebfd8 | 1234 | struct mtd_info *s_mtd; |
a5166169 | 1235 | struct hlist_node s_instances; |
2c5f648a | 1236 | unsigned int s_quota_types; /* Bitmask of supported quota types */ |
1da177e4 LT |
1237 | struct quota_info s_dquot; /* Diskquota specific options */ |
1238 | ||
5accdf82 | 1239 | struct sb_writers s_writers; |
1da177e4 | 1240 | |
99c228a9 AG |
1241 | /* |
1242 | * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and | |
1243 | * s_fsnotify_marks together for cache efficiency. They are frequently | |
1244 | * accessed and rarely modified. | |
1245 | */ | |
1246 | void *s_fs_info; /* Filesystem private info */ | |
1247 | ||
1248 | /* Granularity of c/m/atime in ns (cannot be worse than a second) */ | |
1249 | u32 s_time_gran; | |
188d20bc DD |
1250 | /* Time limits for c/m/atime in seconds */ |
1251 | time64_t s_time_min; | |
1252 | time64_t s_time_max; | |
99c228a9 AG |
1253 | #ifdef CONFIG_FSNOTIFY |
1254 | __u32 s_fsnotify_mask; | |
1255 | struct fsnotify_mark_connector __rcu *s_fsnotify_marks; | |
1256 | #endif | |
1257 | ||
85787090 CH |
1258 | char s_id[32]; /* Informational name */ |
1259 | uuid_t s_uuid; /* UUID */ | |
1da177e4 | 1260 | |
8de52778 | 1261 | unsigned int s_max_links; |
1da177e4 LT |
1262 | |
1263 | /* | |
1264 | * The next field is for VFS *only*. No filesystems have any business | |
1265 | * even looking at it. You had been warned. | |
1266 | */ | |
a11f3a05 | 1267 | struct mutex s_vfs_rename_mutex; /* Kludge */ |
1da177e4 | 1268 | |
79c0b2df MS |
1269 | /* |
1270 | * Filesystem subtype. If non-empty the filesystem type field | |
1271 | * in /proc/mounts will be "type.subtype" | |
1272 | */ | |
a0c9a8b8 | 1273 | const char *s_subtype; |
b3b304a2 | 1274 | |
c8aebb0c | 1275 | const struct dentry_operations *s_d_op; /* default d_op for dentries */ |
9fdfdcf1 | 1276 | |
1720f5dd | 1277 | struct shrinker *s_shrink; /* per-sb shrinker handle */ |
4ed5e82f | 1278 | |
7ada4db8 MS |
1279 | /* Number of inodes with nlink == 0 but still referenced */ |
1280 | atomic_long_t s_remove_count; | |
1281 | ||
ec44610f AG |
1282 | /* |
1283 | * Number of inode/mount/sb objects that are being watched, note that | |
1284 | * inodes objects are currently double-accounted. | |
1285 | */ | |
1286 | atomic_long_t s_fsnotify_connectors; | |
721fb6fb | 1287 | |
d7439fb1 | 1288 | /* Read-only state of the superblock is being changed */ |
4ed5e82f | 1289 | int s_readonly_remount; |
7b7a8665 | 1290 | |
735e4ae5 JL |
1291 | /* per-sb errseq_t for reporting writeback errors via syncfs */ |
1292 | errseq_t s_wb_err; | |
1293 | ||
7b7a8665 CH |
1294 | /* AIO completions deferred from interrupt context */ |
1295 | struct workqueue_struct *s_dio_done_wq; | |
215752fc | 1296 | struct hlist_head s_pins; |
f6041567 | 1297 | |
6e4eab57 EB |
1298 | /* |
1299 | * Owning user namespace and default context in which to | |
1300 | * interpret filesystem uids, gids, quotas, device nodes, | |
1301 | * xattrs and security labels. | |
1302 | */ | |
1303 | struct user_namespace *s_user_ns; | |
1304 | ||
f6041567 | 1305 | /* |
7d10f70f WL |
1306 | * The list_lru structure is essentially just a pointer to a table |
1307 | * of per-node lru lists, each of which has its own spinlock. | |
1308 | * There is no need to put them into separate cachelines. | |
f6041567 | 1309 | */ |
7d10f70f WL |
1310 | struct list_lru s_dentry_lru; |
1311 | struct list_lru s_inode_lru; | |
e2fec7c3 | 1312 | struct rcu_head rcu; |
853b39a7 | 1313 | struct work_struct destroy_work; |
69c433ed | 1314 | |
e97fedb9 | 1315 | struct mutex s_sync_lock; /* sync serialisation lock */ |
69c433ed MS |
1316 | |
1317 | /* | |
1318 | * Indicates how deep in a filesystem stack this SB is | |
1319 | */ | |
1320 | int s_stack_depth; | |
74278da9 DC |
1321 | |
1322 | /* s_inode_list_lock protects s_inodes */ | |
1323 | spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp; | |
1324 | struct list_head s_inodes; /* all inodes */ | |
6c60d2b5 DC |
1325 | |
1326 | spinlock_t s_inode_wblist_lock; | |
1327 | struct list_head s_inodes_wb; /* writeback inodes */ | |
3859a271 | 1328 | } __randomize_layout; |
1da177e4 | 1329 | |
a1ec9040 CB |
1330 | static inline struct user_namespace *i_user_ns(const struct inode *inode) |
1331 | { | |
1332 | return inode->i_sb->s_user_ns; | |
1333 | } | |
1334 | ||
81754357 SF |
1335 | /* Helper functions so that in most cases filesystems will |
1336 | * not need to deal directly with kuid_t and kgid_t and can | |
1337 | * instead deal with the raw numeric values that are stored | |
1338 | * in the filesystem. | |
1339 | */ | |
1340 | static inline uid_t i_uid_read(const struct inode *inode) | |
1341 | { | |
a1ec9040 | 1342 | return from_kuid(i_user_ns(inode), inode->i_uid); |
81754357 SF |
1343 | } |
1344 | ||
1345 | static inline gid_t i_gid_read(const struct inode *inode) | |
1346 | { | |
a1ec9040 | 1347 | return from_kgid(i_user_ns(inode), inode->i_gid); |
81754357 SF |
1348 | } |
1349 | ||
1350 | static inline void i_uid_write(struct inode *inode, uid_t uid) | |
1351 | { | |
a1ec9040 | 1352 | inode->i_uid = make_kuid(i_user_ns(inode), uid); |
81754357 SF |
1353 | } |
1354 | ||
1355 | static inline void i_gid_write(struct inode *inode, gid_t gid) | |
1356 | { | |
a1ec9040 | 1357 | inode->i_gid = make_kgid(i_user_ns(inode), gid); |
81754357 SF |
1358 | } |
1359 | ||
234a3113 | 1360 | /** |
e67fe633 CB |
1361 | * i_uid_into_vfsuid - map an inode's i_uid down according to an idmapping |
1362 | * @idmap: idmap of the mount the inode was found from | |
234a3113 CB |
1363 | * @inode: inode to map |
1364 | * | |
e67fe633 | 1365 | * Return: whe inode's i_uid mapped down according to @idmap. |
234a3113 CB |
1366 | * If the inode's i_uid has no mapping INVALID_VFSUID is returned. |
1367 | */ | |
e67fe633 | 1368 | static inline vfsuid_t i_uid_into_vfsuid(struct mnt_idmap *idmap, |
234a3113 CB |
1369 | const struct inode *inode) |
1370 | { | |
4d7ca409 | 1371 | return make_vfsuid(idmap, i_user_ns(inode), inode->i_uid); |
e6c9a714 CB |
1372 | } |
1373 | ||
1f36146a CB |
1374 | /** |
1375 | * i_uid_needs_update - check whether inode's i_uid needs to be updated | |
0dbe12f2 | 1376 | * @idmap: idmap of the mount the inode was found from |
1f36146a CB |
1377 | * @attr: the new attributes of @inode |
1378 | * @inode: the inode to update | |
1379 | * | |
1380 | * Check whether the $inode's i_uid field needs to be updated taking idmapped | |
1381 | * mounts into account if the filesystem supports it. | |
1382 | * | |
1383 | * Return: true if @inode's i_uid field needs to be updated, false if not. | |
1384 | */ | |
0dbe12f2 | 1385 | static inline bool i_uid_needs_update(struct mnt_idmap *idmap, |
1f36146a CB |
1386 | const struct iattr *attr, |
1387 | const struct inode *inode) | |
1388 | { | |
1389 | return ((attr->ia_valid & ATTR_UID) && | |
1390 | !vfsuid_eq(attr->ia_vfsuid, | |
e67fe633 | 1391 | i_uid_into_vfsuid(idmap, inode))); |
1f36146a CB |
1392 | } |
1393 | ||
1394 | /** | |
1395 | * i_uid_update - update @inode's i_uid field | |
0dbe12f2 | 1396 | * @idmap: idmap of the mount the inode was found from |
1f36146a CB |
1397 | * @attr: the new attributes of @inode |
1398 | * @inode: the inode to update | |
1399 | * | |
1400 | * Safely update @inode's i_uid field translating the vfsuid of any idmapped | |
1401 | * mount into the filesystem kuid. | |
1402 | */ | |
0dbe12f2 | 1403 | static inline void i_uid_update(struct mnt_idmap *idmap, |
1f36146a CB |
1404 | const struct iattr *attr, |
1405 | struct inode *inode) | |
1406 | { | |
1407 | if (attr->ia_valid & ATTR_UID) | |
4d7ca409 | 1408 | inode->i_uid = from_vfsuid(idmap, i_user_ns(inode), |
1f36146a CB |
1409 | attr->ia_vfsuid); |
1410 | } | |
1411 | ||
234a3113 | 1412 | /** |
e67fe633 CB |
1413 | * i_gid_into_vfsgid - map an inode's i_gid down according to an idmapping |
1414 | * @idmap: idmap of the mount the inode was found from | |
234a3113 CB |
1415 | * @inode: inode to map |
1416 | * | |
e67fe633 | 1417 | * Return: the inode's i_gid mapped down according to @idmap. |
234a3113 CB |
1418 | * If the inode's i_gid has no mapping INVALID_VFSGID is returned. |
1419 | */ | |
e67fe633 | 1420 | static inline vfsgid_t i_gid_into_vfsgid(struct mnt_idmap *idmap, |
234a3113 CB |
1421 | const struct inode *inode) |
1422 | { | |
4d7ca409 | 1423 | return make_vfsgid(idmap, i_user_ns(inode), inode->i_gid); |
e6c9a714 CB |
1424 | } |
1425 | ||
1f36146a CB |
1426 | /** |
1427 | * i_gid_needs_update - check whether inode's i_gid needs to be updated | |
0dbe12f2 | 1428 | * @idmap: idmap of the mount the inode was found from |
1f36146a CB |
1429 | * @attr: the new attributes of @inode |
1430 | * @inode: the inode to update | |
1431 | * | |
1432 | * Check whether the $inode's i_gid field needs to be updated taking idmapped | |
1433 | * mounts into account if the filesystem supports it. | |
1434 | * | |
1435 | * Return: true if @inode's i_gid field needs to be updated, false if not. | |
1436 | */ | |
0dbe12f2 | 1437 | static inline bool i_gid_needs_update(struct mnt_idmap *idmap, |
1f36146a CB |
1438 | const struct iattr *attr, |
1439 | const struct inode *inode) | |
1440 | { | |
1441 | return ((attr->ia_valid & ATTR_GID) && | |
1442 | !vfsgid_eq(attr->ia_vfsgid, | |
e67fe633 | 1443 | i_gid_into_vfsgid(idmap, inode))); |
1f36146a CB |
1444 | } |
1445 | ||
1446 | /** | |
1447 | * i_gid_update - update @inode's i_gid field | |
0dbe12f2 | 1448 | * @idmap: idmap of the mount the inode was found from |
1f36146a CB |
1449 | * @attr: the new attributes of @inode |
1450 | * @inode: the inode to update | |
1451 | * | |
1452 | * Safely update @inode's i_gid field translating the vfsgid of any idmapped | |
1453 | * mount into the filesystem kgid. | |
1454 | */ | |
0dbe12f2 | 1455 | static inline void i_gid_update(struct mnt_idmap *idmap, |
1f36146a CB |
1456 | const struct iattr *attr, |
1457 | struct inode *inode) | |
1458 | { | |
1459 | if (attr->ia_valid & ATTR_GID) | |
4d7ca409 | 1460 | inode->i_gid = from_vfsgid(idmap, i_user_ns(inode), |
1f36146a CB |
1461 | attr->ia_vfsgid); |
1462 | } | |
1463 | ||
db998553 CB |
1464 | /** |
1465 | * inode_fsuid_set - initialize inode's i_uid field with callers fsuid | |
1466 | * @inode: inode to initialize | |
c14329d3 | 1467 | * @idmap: idmap of the mount the inode was found from |
db998553 CB |
1468 | * |
1469 | * Initialize the i_uid field of @inode. If the inode was found/created via | |
c14329d3 | 1470 | * an idmapped mount map the caller's fsuid according to @idmap. |
db998553 CB |
1471 | */ |
1472 | static inline void inode_fsuid_set(struct inode *inode, | |
c14329d3 | 1473 | struct mnt_idmap *idmap) |
db998553 | 1474 | { |
c14329d3 | 1475 | inode->i_uid = mapped_fsuid(idmap, i_user_ns(inode)); |
db998553 CB |
1476 | } |
1477 | ||
1478 | /** | |
1479 | * inode_fsgid_set - initialize inode's i_gid field with callers fsgid | |
1480 | * @inode: inode to initialize | |
c14329d3 | 1481 | * @idmap: idmap of the mount the inode was found from |
db998553 CB |
1482 | * |
1483 | * Initialize the i_gid field of @inode. If the inode was found/created via | |
c14329d3 | 1484 | * an idmapped mount map the caller's fsgid according to @idmap. |
db998553 CB |
1485 | */ |
1486 | static inline void inode_fsgid_set(struct inode *inode, | |
c14329d3 | 1487 | struct mnt_idmap *idmap) |
db998553 | 1488 | { |
c14329d3 | 1489 | inode->i_gid = mapped_fsgid(idmap, i_user_ns(inode)); |
db998553 CB |
1490 | } |
1491 | ||
8e538913 CB |
1492 | /** |
1493 | * fsuidgid_has_mapping() - check whether caller's fsuid/fsgid is mapped | |
1494 | * @sb: the superblock we want a mapping in | |
4609e1f1 | 1495 | * @idmap: idmap of the relevant mount |
8e538913 CB |
1496 | * |
1497 | * Check whether the caller's fsuid and fsgid have a valid mapping in the | |
1498 | * s_user_ns of the superblock @sb. If the caller is on an idmapped mount map | |
4609e1f1 | 1499 | * the caller's fsuid and fsgid according to the @idmap first. |
8e538913 CB |
1500 | * |
1501 | * Return: true if fsuid and fsgid is mapped, false if not. | |
1502 | */ | |
1503 | static inline bool fsuidgid_has_mapping(struct super_block *sb, | |
4609e1f1 | 1504 | struct mnt_idmap *idmap) |
8e538913 | 1505 | { |
476860b3 CB |
1506 | struct user_namespace *fs_userns = sb->s_user_ns; |
1507 | kuid_t kuid; | |
1508 | kgid_t kgid; | |
1509 | ||
c14329d3 | 1510 | kuid = mapped_fsuid(idmap, fs_userns); |
476860b3 CB |
1511 | if (!uid_valid(kuid)) |
1512 | return false; | |
c14329d3 | 1513 | kgid = mapped_fsgid(idmap, fs_userns); |
476860b3 CB |
1514 | if (!gid_valid(kgid)) |
1515 | return false; | |
1516 | return kuid_has_mapping(fs_userns, kuid) && | |
1517 | kgid_has_mapping(fs_userns, kgid); | |
8e538913 CB |
1518 | } |
1519 | ||
9b6304c1 JL |
1520 | struct timespec64 current_time(struct inode *inode); |
1521 | struct timespec64 inode_set_ctime_current(struct inode *inode); | |
1522 | ||
077c212f JL |
1523 | static inline time64_t inode_get_atime_sec(const struct inode *inode) |
1524 | { | |
12cd4402 | 1525 | return inode->__i_atime.tv_sec; |
077c212f JL |
1526 | } |
1527 | ||
1528 | static inline long inode_get_atime_nsec(const struct inode *inode) | |
1529 | { | |
12cd4402 | 1530 | return inode->__i_atime.tv_nsec; |
077c212f JL |
1531 | } |
1532 | ||
1533 | static inline struct timespec64 inode_get_atime(const struct inode *inode) | |
1534 | { | |
12cd4402 | 1535 | return inode->__i_atime; |
077c212f JL |
1536 | } |
1537 | ||
1538 | static inline struct timespec64 inode_set_atime_to_ts(struct inode *inode, | |
1539 | struct timespec64 ts) | |
1540 | { | |
12cd4402 | 1541 | inode->__i_atime = ts; |
077c212f JL |
1542 | return ts; |
1543 | } | |
1544 | ||
1545 | static inline struct timespec64 inode_set_atime(struct inode *inode, | |
1546 | time64_t sec, long nsec) | |
1547 | { | |
1548 | struct timespec64 ts = { .tv_sec = sec, | |
1549 | .tv_nsec = nsec }; | |
1550 | return inode_set_atime_to_ts(inode, ts); | |
1551 | } | |
1552 | ||
1553 | static inline time64_t inode_get_mtime_sec(const struct inode *inode) | |
1554 | { | |
12cd4402 | 1555 | return inode->__i_mtime.tv_sec; |
077c212f JL |
1556 | } |
1557 | ||
1558 | static inline long inode_get_mtime_nsec(const struct inode *inode) | |
1559 | { | |
12cd4402 | 1560 | return inode->__i_mtime.tv_nsec; |
077c212f JL |
1561 | } |
1562 | ||
1563 | static inline struct timespec64 inode_get_mtime(const struct inode *inode) | |
1564 | { | |
12cd4402 | 1565 | return inode->__i_mtime; |
077c212f JL |
1566 | } |
1567 | ||
1568 | static inline struct timespec64 inode_set_mtime_to_ts(struct inode *inode, | |
1569 | struct timespec64 ts) | |
1570 | { | |
12cd4402 | 1571 | inode->__i_mtime = ts; |
077c212f JL |
1572 | return ts; |
1573 | } | |
1574 | ||
1575 | static inline struct timespec64 inode_set_mtime(struct inode *inode, | |
1576 | time64_t sec, long nsec) | |
1577 | { | |
1578 | struct timespec64 ts = { .tv_sec = sec, | |
1579 | .tv_nsec = nsec }; | |
1580 | return inode_set_mtime_to_ts(inode, ts); | |
1581 | } | |
1582 | ||
1583 | static inline time64_t inode_get_ctime_sec(const struct inode *inode) | |
1584 | { | |
1585 | return inode->__i_ctime.tv_sec; | |
1586 | } | |
1587 | ||
1588 | static inline long inode_get_ctime_nsec(const struct inode *inode) | |
1589 | { | |
1590 | return inode->__i_ctime.tv_nsec; | |
1591 | } | |
1592 | ||
9b6304c1 JL |
1593 | static inline struct timespec64 inode_get_ctime(const struct inode *inode) |
1594 | { | |
647aa768 | 1595 | return inode->__i_ctime; |
9b6304c1 JL |
1596 | } |
1597 | ||
9b6304c1 JL |
1598 | static inline struct timespec64 inode_set_ctime_to_ts(struct inode *inode, |
1599 | struct timespec64 ts) | |
1600 | { | |
13bc2445 | 1601 | inode->__i_ctime = ts; |
9b6304c1 JL |
1602 | return ts; |
1603 | } | |
1604 | ||
1605 | /** | |
1606 | * inode_set_ctime - set the ctime in the inode | |
1607 | * @inode: inode in which to set the ctime | |
1608 | * @sec: tv_sec value to set | |
1609 | * @nsec: tv_nsec value to set | |
1610 | * | |
1611 | * Set the ctime in @inode to { @sec, @nsec } | |
1612 | */ | |
1613 | static inline struct timespec64 inode_set_ctime(struct inode *inode, | |
1614 | time64_t sec, long nsec) | |
1615 | { | |
1616 | struct timespec64 ts = { .tv_sec = sec, | |
1617 | .tv_nsec = nsec }; | |
1618 | ||
1619 | return inode_set_ctime_to_ts(inode, ts); | |
1620 | } | |
1da177e4 | 1621 | |
077c212f JL |
1622 | struct timespec64 simple_inode_init_ts(struct inode *inode); |
1623 | ||
1da177e4 LT |
1624 | /* |
1625 | * Snapshotting support. | |
1626 | */ | |
1da177e4 | 1627 | |
9b852342 DW |
1628 | /* |
1629 | * These are internal functions, please use sb_start_{write,pagefault,intwrite} | |
1630 | * instead. | |
1631 | */ | |
1632 | static inline void __sb_end_write(struct super_block *sb, int level) | |
1633 | { | |
1634 | percpu_up_read(sb->s_writers.rw_sem + level-1); | |
1635 | } | |
1636 | ||
1637 | static inline void __sb_start_write(struct super_block *sb, int level) | |
1638 | { | |
1639 | percpu_down_read(sb->s_writers.rw_sem + level - 1); | |
1640 | } | |
1641 | ||
1642 | static inline bool __sb_start_write_trylock(struct super_block *sb, int level) | |
1643 | { | |
1644 | return percpu_down_read_trylock(sb->s_writers.rw_sem + level - 1); | |
1645 | } | |
5accdf82 | 1646 | |
bee9182d | 1647 | #define __sb_writers_acquired(sb, lev) \ |
8129ed29 | 1648 | percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_) |
bee9182d | 1649 | #define __sb_writers_release(sb, lev) \ |
8129ed29 | 1650 | percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_) |
bee9182d | 1651 | |
8802e580 AG |
1652 | /** |
1653 | * __sb_write_started - check if sb freeze level is held | |
1654 | * @sb: the super we write to | |
1655 | * @level: the freeze level | |
1656 | * | |
c39e2ae3 VN |
1657 | * * > 0 - sb freeze level is held |
1658 | * * 0 - sb freeze level is not held | |
1659 | * * < 0 - !CONFIG_LOCKDEP/LOCK_STATE_UNKNOWN | |
8802e580 AG |
1660 | */ |
1661 | static inline int __sb_write_started(const struct super_block *sb, int level) | |
1662 | { | |
1663 | return lockdep_is_held_type(sb->s_writers.rw_sem + level - 1, 1); | |
1664 | } | |
1665 | ||
3d5cd491 AG |
1666 | /** |
1667 | * sb_write_started - check if SB_FREEZE_WRITE is held | |
1668 | * @sb: the super we write to | |
1669 | * | |
1670 | * May be false positive with !CONFIG_LOCKDEP/LOCK_STATE_UNKNOWN. | |
1671 | */ | |
7f8d12ea NA |
1672 | static inline bool sb_write_started(const struct super_block *sb) |
1673 | { | |
8802e580 | 1674 | return __sb_write_started(sb, SB_FREEZE_WRITE); |
7f8d12ea NA |
1675 | } |
1676 | ||
21b32e6a AG |
1677 | /** |
1678 | * sb_write_not_started - check if SB_FREEZE_WRITE is not held | |
1679 | * @sb: the super we write to | |
1680 | * | |
1681 | * May be false positive with !CONFIG_LOCKDEP/LOCK_STATE_UNKNOWN. | |
1682 | */ | |
1683 | static inline bool sb_write_not_started(const struct super_block *sb) | |
1684 | { | |
1685 | return __sb_write_started(sb, SB_FREEZE_WRITE) <= 0; | |
1686 | } | |
1687 | ||
3d5cd491 AG |
1688 | /** |
1689 | * file_write_started - check if SB_FREEZE_WRITE is held | |
1690 | * @file: the file we write to | |
1691 | * | |
1692 | * May be false positive with !CONFIG_LOCKDEP/LOCK_STATE_UNKNOWN. | |
1693 | * May be false positive with !S_ISREG, because file_start_write() has | |
1694 | * no effect on !S_ISREG. | |
1695 | */ | |
1696 | static inline bool file_write_started(const struct file *file) | |
1697 | { | |
1698 | if (!S_ISREG(file_inode(file)->i_mode)) | |
1699 | return true; | |
1700 | return sb_write_started(file_inode(file)->i_sb); | |
1701 | } | |
1702 | ||
21b32e6a AG |
1703 | /** |
1704 | * file_write_not_started - check if SB_FREEZE_WRITE is not held | |
1705 | * @file: the file we write to | |
1706 | * | |
1707 | * May be false positive with !CONFIG_LOCKDEP/LOCK_STATE_UNKNOWN. | |
1708 | * May be false positive with !S_ISREG, because file_start_write() has | |
1709 | * no effect on !S_ISREG. | |
1710 | */ | |
1711 | static inline bool file_write_not_started(const struct file *file) | |
1712 | { | |
1713 | if (!S_ISREG(file_inode(file)->i_mode)) | |
1714 | return true; | |
1715 | return sb_write_not_started(file_inode(file)->i_sb); | |
7f8d12ea NA |
1716 | } |
1717 | ||
5accdf82 JK |
1718 | /** |
1719 | * sb_end_write - drop write access to a superblock | |
1720 | * @sb: the super we wrote to | |
1721 | * | |
1722 | * Decrement number of writers to the filesystem. Wake up possible waiters | |
1723 | * wanting to freeze the filesystem. | |
1724 | */ | |
1725 | static inline void sb_end_write(struct super_block *sb) | |
1726 | { | |
1727 | __sb_end_write(sb, SB_FREEZE_WRITE); | |
1728 | } | |
1729 | ||
1730 | /** | |
1731 | * sb_end_pagefault - drop write access to a superblock from a page fault | |
1732 | * @sb: the super we wrote to | |
1733 | * | |
1734 | * Decrement number of processes handling write page fault to the filesystem. | |
1735 | * Wake up possible waiters wanting to freeze the filesystem. | |
1736 | */ | |
1737 | static inline void sb_end_pagefault(struct super_block *sb) | |
1738 | { | |
1739 | __sb_end_write(sb, SB_FREEZE_PAGEFAULT); | |
1740 | } | |
1741 | ||
1742 | /** | |
1743 | * sb_end_intwrite - drop write access to a superblock for internal fs purposes | |
1744 | * @sb: the super we wrote to | |
1745 | * | |
1746 | * Decrement fs-internal number of writers to the filesystem. Wake up possible | |
1747 | * waiters wanting to freeze the filesystem. | |
1748 | */ | |
1749 | static inline void sb_end_intwrite(struct super_block *sb) | |
1750 | { | |
1751 | __sb_end_write(sb, SB_FREEZE_FS); | |
1752 | } | |
1753 | ||
1754 | /** | |
1755 | * sb_start_write - get write access to a superblock | |
1756 | * @sb: the super we write to | |
1757 | * | |
1758 | * When a process wants to write data or metadata to a file system (i.e. dirty | |
1759 | * a page or an inode), it should embed the operation in a sb_start_write() - | |
1760 | * sb_end_write() pair to get exclusion against file system freezing. This | |
1761 | * function increments number of writers preventing freezing. If the file | |
1762 | * system is already frozen, the function waits until the file system is | |
1763 | * thawed. | |
1764 | * | |
1765 | * Since freeze protection behaves as a lock, users have to preserve | |
1766 | * ordering of freeze protection and other filesystem locks. Generally, | |
1767 | * freeze protection should be the outermost lock. In particular, we have: | |
1768 | * | |
1769 | * sb_start_write | |
1770 | * -> i_mutex (write path, truncate, directory ops, ...) | |
1771 | * -> s_umount (freeze_super, thaw_super) | |
1772 | */ | |
1773 | static inline void sb_start_write(struct super_block *sb) | |
1774 | { | |
8a3c84b6 | 1775 | __sb_start_write(sb, SB_FREEZE_WRITE); |
5accdf82 JK |
1776 | } |
1777 | ||
8a3c84b6 | 1778 | static inline bool sb_start_write_trylock(struct super_block *sb) |
5accdf82 | 1779 | { |
8a3c84b6 | 1780 | return __sb_start_write_trylock(sb, SB_FREEZE_WRITE); |
5accdf82 JK |
1781 | } |
1782 | ||
1783 | /** | |
1784 | * sb_start_pagefault - get write access to a superblock from a page fault | |
1785 | * @sb: the super we write to | |
1786 | * | |
1787 | * When a process starts handling write page fault, it should embed the | |
1788 | * operation into sb_start_pagefault() - sb_end_pagefault() pair to get | |
1789 | * exclusion against file system freezing. This is needed since the page fault | |
1790 | * is going to dirty a page. This function increments number of running page | |
1791 | * faults preventing freezing. If the file system is already frozen, the | |
1792 | * function waits until the file system is thawed. | |
1793 | * | |
1794 | * Since page fault freeze protection behaves as a lock, users have to preserve | |
1795 | * ordering of freeze protection and other filesystem locks. It is advised to | |
c1e8d7c6 | 1796 | * put sb_start_pagefault() close to mmap_lock in lock ordering. Page fault |
5accdf82 JK |
1797 | * handling code implies lock dependency: |
1798 | * | |
c1e8d7c6 | 1799 | * mmap_lock |
5accdf82 JK |
1800 | * -> sb_start_pagefault |
1801 | */ | |
1802 | static inline void sb_start_pagefault(struct super_block *sb) | |
1803 | { | |
8a3c84b6 | 1804 | __sb_start_write(sb, SB_FREEZE_PAGEFAULT); |
5accdf82 JK |
1805 | } |
1806 | ||
39015399 | 1807 | /** |
5accdf82 JK |
1808 | * sb_start_intwrite - get write access to a superblock for internal fs purposes |
1809 | * @sb: the super we write to | |
1810 | * | |
1811 | * This is the third level of protection against filesystem freezing. It is | |
1812 | * free for use by a filesystem. The only requirement is that it must rank | |
1813 | * below sb_start_pagefault. | |
1814 | * | |
1815 | * For example filesystem can call sb_start_intwrite() when starting a | |
1816 | * transaction which somewhat eases handling of freezing for internal sources | |
1817 | * of filesystem changes (internal fs threads, discarding preallocation on file | |
1818 | * close, etc.). | |
1819 | */ | |
1820 | static inline void sb_start_intwrite(struct super_block *sb) | |
1821 | { | |
8a3c84b6 | 1822 | __sb_start_write(sb, SB_FREEZE_FS); |
5accdf82 | 1823 | } |
1da177e4 | 1824 | |
8a3c84b6 | 1825 | static inline bool sb_start_intwrite_trylock(struct super_block *sb) |
0c8e3fe3 | 1826 | { |
8a3c84b6 | 1827 | return __sb_start_write_trylock(sb, SB_FREEZE_FS); |
0c8e3fe3 AG |
1828 | } |
1829 | ||
01beba79 | 1830 | bool inode_owner_or_capable(struct mnt_idmap *idmap, |
21cb47be | 1831 | const struct inode *inode); |
3bd858ab | 1832 | |
1da177e4 LT |
1833 | /* |
1834 | * VFS helper functions.. | |
1835 | */ | |
abf08576 | 1836 | int vfs_create(struct mnt_idmap *, struct inode *, |
6521f891 | 1837 | struct dentry *, umode_t, bool); |
abf08576 | 1838 | int vfs_mkdir(struct mnt_idmap *, struct inode *, |
6521f891 | 1839 | struct dentry *, umode_t); |
abf08576 | 1840 | int vfs_mknod(struct mnt_idmap *, struct inode *, struct dentry *, |
6521f891 | 1841 | umode_t, dev_t); |
abf08576 | 1842 | int vfs_symlink(struct mnt_idmap *, struct inode *, |
6521f891 | 1843 | struct dentry *, const char *); |
abf08576 | 1844 | int vfs_link(struct dentry *, struct mnt_idmap *, struct inode *, |
6521f891 | 1845 | struct dentry *, struct inode **); |
abf08576 CB |
1846 | int vfs_rmdir(struct mnt_idmap *, struct inode *, struct dentry *); |
1847 | int vfs_unlink(struct mnt_idmap *, struct inode *, struct dentry *, | |
6521f891 | 1848 | struct inode **); |
9fe61450 | 1849 | |
92cb01c7 CB |
1850 | /** |
1851 | * struct renamedata - contains all information required for renaming | |
abf08576 | 1852 | * @old_mnt_idmap: idmap of the old mount the inode was found from |
92cb01c7 CB |
1853 | * @old_dir: parent of source |
1854 | * @old_dentry: source | |
abf08576 | 1855 | * @new_mnt_idmap: idmap of the new mount the inode was found from |
92cb01c7 CB |
1856 | * @new_dir: parent of destination |
1857 | * @new_dentry: destination | |
1858 | * @delegated_inode: returns an inode needing a delegation break | |
1859 | * @flags: rename flags | |
1860 | */ | |
9fe61450 | 1861 | struct renamedata { |
abf08576 | 1862 | struct mnt_idmap *old_mnt_idmap; |
9fe61450 CB |
1863 | struct inode *old_dir; |
1864 | struct dentry *old_dentry; | |
abf08576 | 1865 | struct mnt_idmap *new_mnt_idmap; |
9fe61450 CB |
1866 | struct inode *new_dir; |
1867 | struct dentry *new_dentry; | |
1868 | struct inode **delegated_inode; | |
1869 | unsigned int flags; | |
1870 | } __randomize_layout; | |
a3c751a5 | 1871 | |
9fe61450 | 1872 | int vfs_rename(struct renamedata *); |
a3c751a5 | 1873 | |
abf08576 | 1874 | static inline int vfs_whiteout(struct mnt_idmap *idmap, |
6521f891 | 1875 | struct inode *dir, struct dentry *dentry) |
a3c751a5 | 1876 | { |
abf08576 | 1877 | return vfs_mknod(idmap, dir, dentry, S_IFCHR | WHITEOUT_MODE, |
6521f891 | 1878 | WHITEOUT_DEV); |
a3c751a5 | 1879 | } |
1da177e4 | 1880 | |
d56e0ddb AG |
1881 | struct file *kernel_tmpfile_open(struct mnt_idmap *idmap, |
1882 | const struct path *parentpath, | |
1883 | umode_t mode, int open_flag, | |
1884 | const struct cred *cred); | |
cbb0b9d4 AG |
1885 | struct file *kernel_file_open(const struct path *path, int flags, |
1886 | struct inode *inode, const struct cred *cred); | |
af7bd4dc | 1887 | |
8e6c848e AV |
1888 | int vfs_mkobj(struct dentry *, umode_t, |
1889 | int (*f)(struct dentry *, umode_t, void *), | |
1890 | void *); | |
1891 | ||
c04011fe | 1892 | int vfs_fchown(struct file *file, uid_t user, gid_t group); |
9e96c8c0 | 1893 | int vfs_fchmod(struct file *file, umode_t mode); |
fd5ad30c | 1894 | int vfs_utimes(const struct path *path, struct timespec64 *times); |
c04011fe | 1895 | |
9df6702a MS |
1896 | extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg); |
1897 | ||
2952db0f AB |
1898 | #ifdef CONFIG_COMPAT |
1899 | extern long compat_ptr_ioctl(struct file *file, unsigned int cmd, | |
1900 | unsigned long arg); | |
1901 | #else | |
1902 | #define compat_ptr_ioctl NULL | |
1903 | #endif | |
1904 | ||
8c744fb8 CH |
1905 | /* |
1906 | * VFS file helper functions. | |
1907 | */ | |
f2d40141 | 1908 | void inode_init_owner(struct mnt_idmap *idmap, struct inode *inode, |
21cb47be | 1909 | const struct inode *dir, umode_t mode); |
a2982cc9 | 1910 | extern bool may_open_dev(const struct path *path); |
9452e93e | 1911 | umode_t mode_strip_sgid(struct mnt_idmap *idmap, |
2b3416ce | 1912 | const struct inode *dir, umode_t mode); |
c4b929b8 | 1913 | |
1da177e4 LT |
1914 | /* |
1915 | * This is the "filldir" function type, used by readdir() to let | |
1916 | * the kernel specify what kind of dirent layout it wants to have. | |
1917 | * This allows the kernel to read directories into kernel space or | |
1918 | * to have different dirent layouts depending on the binary type. | |
25885a35 | 1919 | * Return 'true' to keep going and 'false' if there are no more entries. |
1da177e4 | 1920 | */ |
ac7576f4 | 1921 | struct dir_context; |
25885a35 | 1922 | typedef bool (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64, |
ac7576f4 MS |
1923 | unsigned); |
1924 | ||
5c0ba4e0 | 1925 | struct dir_context { |
a09acf4b | 1926 | filldir_t actor; |
bb6f619b | 1927 | loff_t pos; |
5c0ba4e0 | 1928 | }; |
bb6f619b | 1929 | |
b4caecd4 CH |
1930 | /* |
1931 | * These flags let !MMU mmap() govern direct device mapping vs immediate | |
1932 | * copying more easily for MAP_PRIVATE, especially for ROM filesystems. | |
1933 | * | |
1934 | * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE) | |
1935 | * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED) | |
1936 | * NOMMU_MAP_READ: Can be mapped for reading | |
1937 | * NOMMU_MAP_WRITE: Can be mapped for writing | |
1938 | * NOMMU_MAP_EXEC: Can be mapped for execution | |
1939 | */ | |
1940 | #define NOMMU_MAP_COPY 0x00000001 | |
1941 | #define NOMMU_MAP_DIRECT 0x00000008 | |
1942 | #define NOMMU_MAP_READ VM_MAYREAD | |
1943 | #define NOMMU_MAP_WRITE VM_MAYWRITE | |
1944 | #define NOMMU_MAP_EXEC VM_MAYEXEC | |
1945 | ||
1946 | #define NOMMU_VMFLAGS \ | |
1947 | (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC) | |
1948 | ||
2e5dfc99 DW |
1949 | /* |
1950 | * These flags control the behavior of the remap_file_range function pointer. | |
1951 | * If it is called with len == 0 that means "remap to end of source file". | |
5c437fa2 | 1952 | * See Documentation/filesystems/vfs.rst for more details about this call. |
2e5dfc99 DW |
1953 | * |
1954 | * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate) | |
eca3654e | 1955 | * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request |
2e5dfc99 DW |
1956 | */ |
1957 | #define REMAP_FILE_DEDUP (1 << 0) | |
eca3654e | 1958 | #define REMAP_FILE_CAN_SHORTEN (1 << 1) |
2e5dfc99 DW |
1959 | |
1960 | /* | |
1961 | * These flags signal that the caller is ok with altering various aspects of | |
1962 | * the behavior of the remap operation. The changes must be made by the | |
1963 | * implementation; the vfs remap helper functions can take advantage of them. | |
1964 | * Flags in this category exist to preserve the quirky behavior of the hoisted | |
1965 | * btrfs clone/dedupe ioctls. | |
2e5dfc99 | 1966 | */ |
eca3654e | 1967 | #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN) |
b4caecd4 | 1968 | |
10bc8e4a AG |
1969 | /* |
1970 | * These flags control the behavior of vfs_copy_file_range(). | |
1971 | * They are not available to the user via syscall. | |
1972 | * | |
1973 | * COPY_FILE_SPLICE: call splice direct instead of fs clone/copy ops | |
1974 | */ | |
1975 | #define COPY_FILE_SPLICE (1 << 0) | |
1976 | ||
293bc982 | 1977 | struct iov_iter; |
ee692a21 | 1978 | struct io_uring_cmd; |
6faddda6 | 1979 | struct offset_ctx; |
293bc982 | 1980 | |
1da177e4 LT |
1981 | struct file_operations { |
1982 | struct module *owner; | |
1983 | loff_t (*llseek) (struct file *, loff_t, int); | |
1984 | ssize_t (*read) (struct file *, char __user *, size_t, loff_t *); | |
1da177e4 | 1985 | ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *); |
293bc982 AV |
1986 | ssize_t (*read_iter) (struct kiocb *, struct iov_iter *); |
1987 | ssize_t (*write_iter) (struct kiocb *, struct iov_iter *); | |
5a72e899 JA |
1988 | int (*iopoll)(struct kiocb *kiocb, struct io_comp_batch *, |
1989 | unsigned int flags); | |
61922694 | 1990 | int (*iterate_shared) (struct file *, struct dir_context *); |
a3f8683b | 1991 | __poll_t (*poll) (struct file *, struct poll_table_struct *); |
1da177e4 LT |
1992 | long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long); |
1993 | long (*compat_ioctl) (struct file *, unsigned int, unsigned long); | |
1994 | int (*mmap) (struct file *, struct vm_area_struct *); | |
1c972597 | 1995 | unsigned long mmap_supported_flags; |
1da177e4 | 1996 | int (*open) (struct inode *, struct file *); |
75e1fcc0 | 1997 | int (*flush) (struct file *, fl_owner_t id); |
1da177e4 | 1998 | int (*release) (struct inode *, struct file *); |
02c24a82 | 1999 | int (*fsync) (struct file *, loff_t, loff_t, int datasync); |
1da177e4 LT |
2000 | int (*fasync) (int, struct file *, int); |
2001 | int (*lock) (struct file *, int, struct file_lock *); | |
1da177e4 LT |
2002 | unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); |
2003 | int (*check_flags)(int); | |
1da177e4 | 2004 | int (*flock) (struct file *, int, struct file_lock *); |
cbb7e577 JA |
2005 | ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int); |
2006 | ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int); | |
2bfc6685 | 2007 | void (*splice_eof)(struct file *file); |
ed5f17f6 | 2008 | int (*setlease)(struct file *, int, struct file_lock **, void **); |
2fe17c10 CH |
2009 | long (*fallocate)(struct file *file, int mode, loff_t offset, |
2010 | loff_t len); | |
a3816ab0 | 2011 | void (*show_fdinfo)(struct seq_file *m, struct file *f); |
b4caecd4 CH |
2012 | #ifndef CONFIG_MMU |
2013 | unsigned (*mmap_capabilities)(struct file *); | |
2014 | #endif | |
04b38d60 CH |
2015 | ssize_t (*copy_file_range)(struct file *, loff_t, struct file *, |
2016 | loff_t, size_t, unsigned int); | |
42ec3d4c DW |
2017 | loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in, |
2018 | struct file *file_out, loff_t pos_out, | |
2019 | loff_t len, unsigned int remap_flags); | |
45cd0faa | 2020 | int (*fadvise)(struct file *, loff_t, loff_t, int); |
ee692a21 | 2021 | int (*uring_cmd)(struct io_uring_cmd *ioucmd, unsigned int issue_flags); |
de97fcb3 JA |
2022 | int (*uring_cmd_iopoll)(struct io_uring_cmd *, struct io_comp_batch *, |
2023 | unsigned int poll_flags); | |
3859a271 | 2024 | } __randomize_layout; |
1da177e4 | 2025 | |
3e327154 LT |
2026 | /* Wrap a directory iterator that needs exclusive inode access */ |
2027 | int wrap_directory_iterator(struct file *, struct dir_context *, | |
2028 | int (*) (struct file *, struct dir_context *)); | |
2029 | #define WRAP_DIR_ITER(x) \ | |
2030 | static int shared_##x(struct file *file , struct dir_context *ctx) \ | |
2031 | { return wrap_directory_iterator(file, ctx, x); } | |
2032 | ||
1da177e4 | 2033 | struct inode_operations { |
00cd8dd3 | 2034 | struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int); |
fceef393 | 2035 | const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *); |
4609e1f1 | 2036 | int (*permission) (struct mnt_idmap *, struct inode *, int); |
cac2f8b8 | 2037 | struct posix_acl * (*get_inode_acl)(struct inode *, int, bool); |
44a7d7a8 NP |
2038 | |
2039 | int (*readlink) (struct dentry *, char __user *,int); | |
44a7d7a8 | 2040 | |
6c960e68 | 2041 | int (*create) (struct mnt_idmap *, struct inode *,struct dentry *, |
549c7297 | 2042 | umode_t, bool); |
1da177e4 LT |
2043 | int (*link) (struct dentry *,struct inode *,struct dentry *); |
2044 | int (*unlink) (struct inode *,struct dentry *); | |
7a77db95 | 2045 | int (*symlink) (struct mnt_idmap *, struct inode *,struct dentry *, |
549c7297 | 2046 | const char *); |
c54bd91e | 2047 | int (*mkdir) (struct mnt_idmap *, struct inode *,struct dentry *, |
549c7297 | 2048 | umode_t); |
1da177e4 | 2049 | int (*rmdir) (struct inode *,struct dentry *); |
5ebb29be | 2050 | int (*mknod) (struct mnt_idmap *, struct inode *,struct dentry *, |
549c7297 | 2051 | umode_t,dev_t); |
e18275ae | 2052 | int (*rename) (struct mnt_idmap *, struct inode *, struct dentry *, |
520c8b16 | 2053 | struct inode *, struct dentry *, unsigned int); |
c1632a0f | 2054 | int (*setattr) (struct mnt_idmap *, struct dentry *, struct iattr *); |
b74d24f7 | 2055 | int (*getattr) (struct mnt_idmap *, const struct path *, |
549c7297 | 2056 | struct kstat *, u32, unsigned int); |
1da177e4 | 2057 | ssize_t (*listxattr) (struct dentry *, char *, size_t); |
c4b929b8 MF |
2058 | int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start, |
2059 | u64 len); | |
913e9928 | 2060 | int (*update_time)(struct inode *, int); |
d9585277 | 2061 | int (*atomic_open)(struct inode *, struct dentry *, |
30d90494 | 2062 | struct file *, unsigned open_flag, |
44907d79 | 2063 | umode_t create_mode); |
011e2b71 | 2064 | int (*tmpfile) (struct mnt_idmap *, struct inode *, |
863f144f | 2065 | struct file *, umode_t); |
77435322 | 2066 | struct posix_acl *(*get_acl)(struct mnt_idmap *, struct dentry *, |
7420332a | 2067 | int); |
13e83a49 | 2068 | int (*set_acl)(struct mnt_idmap *, struct dentry *, |
549c7297 | 2069 | struct posix_acl *, int); |
8782a9ae | 2070 | int (*fileattr_set)(struct mnt_idmap *idmap, |
4c5b4799 MS |
2071 | struct dentry *dentry, struct fileattr *fa); |
2072 | int (*fileattr_get)(struct dentry *dentry, struct fileattr *fa); | |
6faddda6 | 2073 | struct offset_ctx *(*get_offset_ctx)(struct inode *inode); |
44a7d7a8 | 2074 | } ____cacheline_aligned; |
1da177e4 | 2075 | |
bb7462b6 MS |
2076 | static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio, |
2077 | struct iov_iter *iter) | |
2078 | { | |
2079 | return file->f_op->read_iter(kio, iter); | |
2080 | } | |
2081 | ||
2082 | static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio, | |
2083 | struct iov_iter *iter) | |
2084 | { | |
2085 | return file->f_op->write_iter(kio, iter); | |
2086 | } | |
2087 | ||
f74ac015 MS |
2088 | static inline int call_mmap(struct file *file, struct vm_area_struct *vma) |
2089 | { | |
2090 | return file->f_op->mmap(file, vma); | |
2091 | } | |
2092 | ||
1da177e4 LT |
2093 | extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *); |
2094 | extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *); | |
29732938 ZB |
2095 | extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *, |
2096 | loff_t, size_t, unsigned int); | |
6f7db389 SR |
2097 | int __generic_remap_file_range_prep(struct file *file_in, loff_t pos_in, |
2098 | struct file *file_out, loff_t pos_out, | |
2099 | loff_t *len, unsigned int remap_flags, | |
2100 | const struct iomap_ops *dax_read_ops); | |
2101 | int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in, | |
2102 | struct file *file_out, loff_t pos_out, | |
2103 | loff_t *count, unsigned int remap_flags); | |
42ec3d4c DW |
2104 | extern loff_t do_clone_file_range(struct file *file_in, loff_t pos_in, |
2105 | struct file *file_out, loff_t pos_out, | |
452ce659 | 2106 | loff_t len, unsigned int remap_flags); |
42ec3d4c DW |
2107 | extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in, |
2108 | struct file *file_out, loff_t pos_out, | |
452ce659 | 2109 | loff_t len, unsigned int remap_flags); |
54dbc151 DW |
2110 | extern int vfs_dedupe_file_range(struct file *file, |
2111 | struct file_dedupe_range *same); | |
42ec3d4c DW |
2112 | extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos, |
2113 | struct file *dst_file, loff_t dst_pos, | |
df365836 | 2114 | loff_t len, unsigned int remap_flags); |
f1825366 | 2115 | |
7366f8b6 CB |
2116 | /** |
2117 | * enum freeze_holder - holder of the freeze | |
2118 | * @FREEZE_HOLDER_KERNEL: kernel wants to freeze or thaw filesystem | |
2119 | * @FREEZE_HOLDER_USERSPACE: userspace wants to freeze or thaw filesystem | |
2120 | * @FREEZE_MAY_NEST: whether nesting freeze and thaw requests is allowed | |
2121 | * | |
2122 | * Indicate who the owner of the freeze or thaw request is and whether | |
2123 | * the freeze needs to be exclusive or can nest. | |
2124 | * Without @FREEZE_MAY_NEST, multiple freeze and thaw requests from the | |
2125 | * same holder aren't allowed. It is however allowed to hold a single | |
2126 | * @FREEZE_HOLDER_USERSPACE and a single @FREEZE_HOLDER_KERNEL freeze at | |
2127 | * the same time. This is relied upon by some filesystems during online | |
2128 | * repair or similar. | |
2129 | */ | |
880b9577 DW |
2130 | enum freeze_holder { |
2131 | FREEZE_HOLDER_KERNEL = (1U << 0), | |
2132 | FREEZE_HOLDER_USERSPACE = (1U << 1), | |
7366f8b6 | 2133 | FREEZE_MAY_NEST = (1U << 2), |
880b9577 | 2134 | }; |
1da177e4 | 2135 | |
1da177e4 LT |
2136 | struct super_operations { |
2137 | struct inode *(*alloc_inode)(struct super_block *sb); | |
2138 | void (*destroy_inode)(struct inode *); | |
fdb0da89 | 2139 | void (*free_inode)(struct inode *); |
1da177e4 | 2140 | |
aa385729 | 2141 | void (*dirty_inode) (struct inode *, int flags); |
a9185b41 | 2142 | int (*write_inode) (struct inode *, struct writeback_control *wbc); |
45321ac5 | 2143 | int (*drop_inode) (struct inode *); |
be7ce416 | 2144 | void (*evict_inode) (struct inode *); |
1da177e4 | 2145 | void (*put_super) (struct super_block *); |
1da177e4 | 2146 | int (*sync_fs)(struct super_block *sb, int wait); |
880b9577 | 2147 | int (*freeze_super) (struct super_block *, enum freeze_holder who); |
c4be0c1d | 2148 | int (*freeze_fs) (struct super_block *); |
880b9577 | 2149 | int (*thaw_super) (struct super_block *, enum freeze_holder who); |
c4be0c1d | 2150 | int (*unfreeze_fs) (struct super_block *); |
726c3342 | 2151 | int (*statfs) (struct dentry *, struct kstatfs *); |
1da177e4 | 2152 | int (*remount_fs) (struct super_block *, int *, char *); |
42faad99 | 2153 | void (*umount_begin) (struct super_block *); |
1da177e4 | 2154 | |
34c80b1d | 2155 | int (*show_options)(struct seq_file *, struct dentry *); |
d861c630 | 2156 | int (*show_devname)(struct seq_file *, struct dentry *); |
a6322de6 | 2157 | int (*show_path)(struct seq_file *, struct dentry *); |
64132379 | 2158 | int (*show_stats)(struct seq_file *, struct dentry *); |
0e51a720 | 2159 | #ifdef CONFIG_QUOTA |
1da177e4 LT |
2160 | ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t); |
2161 | ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t); | |
2d0fa467 | 2162 | struct dquot **(*get_dquots)(struct inode *); |
0e51a720 | 2163 | #endif |
4101b624 VD |
2164 | long (*nr_cached_objects)(struct super_block *, |
2165 | struct shrink_control *); | |
2166 | long (*free_cached_objects)(struct super_block *, | |
2167 | struct shrink_control *); | |
87efb390 | 2168 | void (*shutdown)(struct super_block *sb); |
1da177e4 LT |
2169 | }; |
2170 | ||
bbc1096a DH |
2171 | /* |
2172 | * Inode flags - they have no relation to superblock flags now | |
2173 | */ | |
6414e9b0 EB |
2174 | #define S_SYNC (1 << 0) /* Writes are synced at once */ |
2175 | #define S_NOATIME (1 << 1) /* Do not update access times */ | |
2176 | #define S_APPEND (1 << 2) /* Append-only file */ | |
2177 | #define S_IMMUTABLE (1 << 3) /* Immutable file */ | |
2178 | #define S_DEAD (1 << 4) /* removed, but still open directory */ | |
2179 | #define S_NOQUOTA (1 << 5) /* Inode is not counted to quota */ | |
2180 | #define S_DIRSYNC (1 << 6) /* Directory modifications are synchronous */ | |
2181 | #define S_NOCMTIME (1 << 7) /* Do not update file c/mtime */ | |
2182 | #define S_SWAPFILE (1 << 8) /* Do not truncate: swapon got its bmaps */ | |
2183 | #define S_PRIVATE (1 << 9) /* Inode is fs-internal */ | |
2184 | #define S_IMA (1 << 10) /* Inode has an associated IMA struct */ | |
2185 | #define S_AUTOMOUNT (1 << 11) /* Automount/referral quasi-directory */ | |
2186 | #define S_NOSEC (1 << 12) /* no suid or xattr security attributes */ | |
6cd176a5 | 2187 | #ifdef CONFIG_FS_DAX |
6414e9b0 | 2188 | #define S_DAX (1 << 13) /* Direct Access, avoiding the page cache */ |
fbbbad4b | 2189 | #else |
6414e9b0 | 2190 | #define S_DAX 0 /* Make all the DAX code disappear */ |
fbbbad4b | 2191 | #endif |
6414e9b0 EB |
2192 | #define S_ENCRYPTED (1 << 14) /* Encrypted file (using fs/crypto/) */ |
2193 | #define S_CASEFOLD (1 << 15) /* Casefolded file */ | |
2194 | #define S_VERITY (1 << 16) /* Verity file (using fs/verity/) */ | |
1bd9c4e4 | 2195 | #define S_KERNEL_FILE (1 << 17) /* File is in use by the kernel (eg. fs/cachefiles) */ |
bbc1096a DH |
2196 | |
2197 | /* | |
2198 | * Note that nosuid etc flags are inode-specific: setting some file-system | |
2199 | * flags just means all the inodes inherit those flags by default. It might be | |
2200 | * possible to override it selectively if you really wanted to with some | |
2201 | * ioctl() that is not currently implemented. | |
2202 | * | |
e462ec50 | 2203 | * Exception: SB_RDONLY is always applied to the entire file system. |
bbc1096a DH |
2204 | * |
2205 | * Unfortunately, it is possible to change a filesystems flags with it mounted | |
2206 | * with files in use. This means that all of the inodes will not have their | |
2207 | * i_flags updated. Hence, i_flags no longer inherit the superblock mount | |
2208 | * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org | |
2209 | */ | |
2210 | #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg)) | |
2211 | ||
1751e8a6 | 2212 | static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; } |
94e92e7a | 2213 | #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb) |
e462ec50 | 2214 | #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \ |
bbc1096a | 2215 | ((inode)->i_flags & S_SYNC)) |
e462ec50 | 2216 | #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \ |
bbc1096a | 2217 | ((inode)->i_flags & (S_SYNC|S_DIRSYNC))) |
e462ec50 DH |
2218 | #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK) |
2219 | #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME) | |
2220 | #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION) | |
bbc1096a DH |
2221 | |
2222 | #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA) | |
2223 | #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND) | |
2224 | #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE) | |
e4e8b47a MK |
2225 | |
2226 | #ifdef CONFIG_FS_POSIX_ACL | |
e462ec50 | 2227 | #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL) |
e4e8b47a MK |
2228 | #else |
2229 | #define IS_POSIXACL(inode) 0 | |
2230 | #endif | |
bbc1096a DH |
2231 | |
2232 | #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD) | |
2233 | #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME) | |
2234 | #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE) | |
2235 | #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE) | |
2236 | #define IS_IMA(inode) ((inode)->i_flags & S_IMA) | |
2237 | #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT) | |
2238 | #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC) | |
fbbbad4b | 2239 | #define IS_DAX(inode) ((inode)->i_flags & S_DAX) |
2ee6a576 | 2240 | #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED) |
b886ee3e | 2241 | #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD) |
5585f2af | 2242 | #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY) |
bbc1096a | 2243 | |
787fb6bc MS |
2244 | #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \ |
2245 | (inode)->i_rdev == WHITEOUT_DEV) | |
2246 | ||
4609e1f1 | 2247 | static inline bool HAS_UNMAPPED_ID(struct mnt_idmap *idmap, |
ba73d987 | 2248 | struct inode *inode) |
0bd23d09 | 2249 | { |
e67fe633 CB |
2250 | return !vfsuid_valid(i_uid_into_vfsuid(idmap, inode)) || |
2251 | !vfsgid_valid(i_gid_into_vfsgid(idmap, inode)); | |
0bd23d09 EB |
2252 | } |
2253 | ||
c75b1d94 JA |
2254 | static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp) |
2255 | { | |
2256 | *kiocb = (struct kiocb) { | |
2257 | .ki_filp = filp, | |
164f4064 | 2258 | .ki_flags = filp->f_iocb_flags, |
20578bdf | 2259 | .ki_ioprio = get_current_ioprio(), |
c75b1d94 JA |
2260 | }; |
2261 | } | |
2262 | ||
5dcdc43e JX |
2263 | static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src, |
2264 | struct file *filp) | |
2265 | { | |
2266 | *kiocb = (struct kiocb) { | |
2267 | .ki_filp = filp, | |
2268 | .ki_flags = kiocb_src->ki_flags, | |
5dcdc43e JX |
2269 | .ki_ioprio = kiocb_src->ki_ioprio, |
2270 | .ki_pos = kiocb_src->ki_pos, | |
2271 | }; | |
2272 | } | |
2273 | ||
1c0eeaf5 | 2274 | /* |
250df6ed | 2275 | * Inode state bits. Protected by inode->i_lock |
1c0eeaf5 | 2276 | * |
1e9d6333 EB |
2277 | * Four bits determine the dirty state of the inode: I_DIRTY_SYNC, |
2278 | * I_DIRTY_DATASYNC, I_DIRTY_PAGES, and I_DIRTY_TIME. | |
1c0eeaf5 JE |
2279 | * |
2280 | * Four bits define the lifetime of an inode. Initially, inodes are I_NEW, | |
2281 | * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at | |
2282 | * various stages of removing an inode. | |
2283 | * | |
eaff8079 | 2284 | * Two bits are used for locking and completion notification, I_NEW and I_SYNC. |
1c0eeaf5 | 2285 | * |
e7ca2d41 | 2286 | * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on |
1e9d6333 EB |
2287 | * fdatasync() (unless I_DIRTY_DATASYNC is also set). |
2288 | * Timestamp updates are the usual cause. | |
2289 | * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of | |
000cb48e JK |
2290 | * these changes separately from I_DIRTY_SYNC so that we |
2291 | * don't have to write inode on fdatasync() when only | |
1e9d6333 | 2292 | * e.g. the timestamps have changed. |
1c0eeaf5 | 2293 | * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean. |
cbfecb92 | 2294 | * I_DIRTY_TIME The inode itself has dirty timestamps, and the |
1e9d6333 EB |
2295 | * lazytime mount option is enabled. We keep track of this |
2296 | * separately from I_DIRTY_SYNC in order to implement | |
2297 | * lazytime. This gets cleared if I_DIRTY_INODE | |
cbfecb92 LC |
2298 | * (I_DIRTY_SYNC and/or I_DIRTY_DATASYNC) gets set. But |
2299 | * I_DIRTY_TIME can still be set if I_DIRTY_SYNC is already | |
2300 | * in place because writeback might already be in progress | |
2301 | * and we don't want to lose the time update | |
eaff8079 CH |
2302 | * I_NEW Serves as both a mutex and completion notification. |
2303 | * New inodes set I_NEW. If two processes both create | |
2304 | * the same inode, one of them will release its inode and | |
2305 | * wait for I_NEW to be released before returning. | |
2306 | * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can | |
2307 | * also cause waiting on I_NEW, without I_NEW actually | |
2308 | * being set. find_inode() uses this to prevent returning | |
2309 | * nearly-dead inodes. | |
1c0eeaf5 JE |
2310 | * I_WILL_FREE Must be set when calling write_inode_now() if i_count |
2311 | * is zero. I_FREEING must be set when I_WILL_FREE is | |
2312 | * cleared. | |
2313 | * I_FREEING Set when inode is about to be freed but still has dirty | |
2314 | * pages or buffers attached or the inode itself is still | |
2315 | * dirty. | |
dbd5768f JK |
2316 | * I_CLEAR Added by clear_inode(). In this state the inode is |
2317 | * clean and can be destroyed. Inode keeps I_FREEING. | |
1c0eeaf5 JE |
2318 | * |
2319 | * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are | |
2320 | * prohibited for many purposes. iget() must wait for | |
2321 | * the inode to be completely released, then create it | |
2322 | * anew. Other functions will just ignore such inodes, | |
eaff8079 | 2323 | * if appropriate. I_NEW is used for waiting. |
1c0eeaf5 | 2324 | * |
169ebd90 JK |
2325 | * I_SYNC Writeback of inode is running. The bit is set during |
2326 | * data writeback, and cleared with a wakeup on the bit | |
2327 | * address once it is done. The bit is also used to pin | |
2328 | * the inode in memory for flusher thread. | |
1c0eeaf5 | 2329 | * |
bd5fe6c5 CH |
2330 | * I_REFERENCED Marks the inode as recently references on the LRU list. |
2331 | * | |
2332 | * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit(). | |
2333 | * | |
682aa8e1 TH |
2334 | * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to |
2335 | * synchronize competing switching instances and to tell | |
b93b0163 | 2336 | * wb stat updates to grab the i_pages lock. See |
a9519def | 2337 | * inode_switch_wbs_work_fn() for details. |
682aa8e1 | 2338 | * |
ad0af710 AG |
2339 | * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper |
2340 | * and work dirs among overlayfs mounts. | |
2341 | * | |
c2b6d621 AV |
2342 | * I_CREATING New object's inode in the middle of setting up. |
2343 | * | |
dae2f8ed IW |
2344 | * I_DONTCACHE Evict inode as soon as it is not used anymore. |
2345 | * | |
5afced3b JK |
2346 | * I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists. |
2347 | * Used to detect that mark_inode_dirty() should not move | |
2348 | * inode between dirty lists. | |
2349 | * | |
08276bda DH |
2350 | * I_PINNING_FSCACHE_WB Inode is pinning an fscache object for writeback. |
2351 | * | |
1c0eeaf5 | 2352 | * Q: What is the difference between I_WILL_FREE and I_FREEING? |
1c0eeaf5 | 2353 | */ |
9e38d86f NP |
2354 | #define I_DIRTY_SYNC (1 << 0) |
2355 | #define I_DIRTY_DATASYNC (1 << 1) | |
2356 | #define I_DIRTY_PAGES (1 << 2) | |
eaff8079 CH |
2357 | #define __I_NEW 3 |
2358 | #define I_NEW (1 << __I_NEW) | |
9e38d86f NP |
2359 | #define I_WILL_FREE (1 << 4) |
2360 | #define I_FREEING (1 << 5) | |
2361 | #define I_CLEAR (1 << 6) | |
eaff8079 | 2362 | #define __I_SYNC 7 |
1c0eeaf5 | 2363 | #define I_SYNC (1 << __I_SYNC) |
9e38d86f | 2364 | #define I_REFERENCED (1 << 8) |
bd5fe6c5 | 2365 | #define __I_DIO_WAKEUP 9 |
ac74d8d6 | 2366 | #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP) |
f4e0c30c | 2367 | #define I_LINKABLE (1 << 10) |
0ae45f63 | 2368 | #define I_DIRTY_TIME (1 << 11) |
682aa8e1 | 2369 | #define I_WB_SWITCH (1 << 13) |
c2b6d621 AV |
2370 | #define I_OVL_INUSE (1 << 14) |
2371 | #define I_CREATING (1 << 15) | |
dae2f8ed | 2372 | #define I_DONTCACHE (1 << 16) |
5afced3b | 2373 | #define I_SYNC_QUEUED (1 << 17) |
c9c4ff12 | 2374 | #define I_PINNING_NETFS_WB (1 << 18) |
1da177e4 | 2375 | |
0e11f644 CH |
2376 | #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC) |
2377 | #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES) | |
0ae45f63 | 2378 | #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME) |
1da177e4 LT |
2379 | |
2380 | extern void __mark_inode_dirty(struct inode *, int); | |
2381 | static inline void mark_inode_dirty(struct inode *inode) | |
2382 | { | |
2383 | __mark_inode_dirty(inode, I_DIRTY); | |
2384 | } | |
2385 | ||
2386 | static inline void mark_inode_dirty_sync(struct inode *inode) | |
2387 | { | |
2388 | __mark_inode_dirty(inode, I_DIRTY_SYNC); | |
2389 | } | |
2390 | ||
ed296c6c EB |
2391 | /* |
2392 | * Returns true if the given inode itself only has dirty timestamps (its pages | |
2393 | * may still be dirty) and isn't currently being allocated or freed. | |
2394 | * Filesystems should call this if when writing an inode when lazytime is | |
2395 | * enabled, they want to opportunistically write the timestamps of other inodes | |
2396 | * located very nearby on-disk, e.g. in the same inode block. This returns true | |
2397 | * if the given inode is in need of such an opportunistic update. Requires | |
2398 | * i_lock, or at least later re-checking under i_lock. | |
2399 | */ | |
2400 | static inline bool inode_is_dirtytime_only(struct inode *inode) | |
2401 | { | |
2402 | return (inode->i_state & (I_DIRTY_TIME | I_NEW | | |
2403 | I_FREEING | I_WILL_FREE)) == I_DIRTY_TIME; | |
2404 | } | |
2405 | ||
7ada4db8 MS |
2406 | extern void inc_nlink(struct inode *inode); |
2407 | extern void drop_nlink(struct inode *inode); | |
2408 | extern void clear_nlink(struct inode *inode); | |
2409 | extern void set_nlink(struct inode *inode, unsigned int nlink); | |
d8c76e6f DH |
2410 | |
2411 | static inline void inode_inc_link_count(struct inode *inode) | |
2412 | { | |
2413 | inc_nlink(inode); | |
66d21730 AD |
2414 | mark_inode_dirty(inode); |
2415 | } | |
2416 | ||
9a53c3a7 DH |
2417 | static inline void inode_dec_link_count(struct inode *inode) |
2418 | { | |
2419 | drop_nlink(inode); | |
66d21730 AD |
2420 | mark_inode_dirty(inode); |
2421 | } | |
2422 | ||
c3b2da31 JB |
2423 | enum file_time_flags { |
2424 | S_ATIME = 1, | |
2425 | S_MTIME = 2, | |
2426 | S_CTIME = 4, | |
2427 | S_VERSION = 8, | |
2428 | }; | |
2429 | ||
c6718543 | 2430 | extern bool atime_needs_update(const struct path *, struct inode *); |
badcf2b7 | 2431 | extern void touch_atime(const struct path *); |
913e9928 | 2432 | int inode_update_time(struct inode *inode, int flags); |
e60feb44 | 2433 | |
1da177e4 LT |
2434 | static inline void file_accessed(struct file *file) |
2435 | { | |
2436 | if (!(file->f_flags & O_NOATIME)) | |
68ac1234 | 2437 | touch_atime(&file->f_path); |
1da177e4 LT |
2438 | } |
2439 | ||
e38f7f53 | 2440 | extern int file_modified(struct file *file); |
66fa3ced | 2441 | int kiocb_modified(struct kiocb *iocb); |
e38f7f53 | 2442 | |
c3765016 | 2443 | int sync_inode_metadata(struct inode *inode, int wait); |
1da177e4 | 2444 | |
1da177e4 LT |
2445 | struct file_system_type { |
2446 | const char *name; | |
2447 | int fs_flags; | |
bbc1096a DH |
2448 | #define FS_REQUIRES_DEV 1 |
2449 | #define FS_BINARY_MOUNTDATA 2 | |
2450 | #define FS_HAS_SUBTYPE 4 | |
0c55cfc4 | 2451 | #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */ |
0b3b094a | 2452 | #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */ |
a6435940 | 2453 | #define FS_ALLOW_IDMAP 32 /* FS has been updated to handle vfs idmappings. */ |
bbc1096a | 2454 | #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */ |
f3a09c92 | 2455 | int (*init_fs_context)(struct fs_context *); |
d7167b14 | 2456 | const struct fs_parameter_spec *parameters; |
c96e41e9 AV |
2457 | struct dentry *(*mount) (struct file_system_type *, int, |
2458 | const char *, void *); | |
1da177e4 LT |
2459 | void (*kill_sb) (struct super_block *); |
2460 | struct module *owner; | |
2461 | struct file_system_type * next; | |
a5166169 | 2462 | struct hlist_head fs_supers; |
d475fd42 | 2463 | |
cf516249 | 2464 | struct lock_class_key s_lock_key; |
897c6ff9 | 2465 | struct lock_class_key s_umount_key; |
51ee049e | 2466 | struct lock_class_key s_vfs_rename_key; |
5accdf82 | 2467 | struct lock_class_key s_writers_key[SB_FREEZE_LEVELS]; |
d475fd42 PZ |
2468 | |
2469 | struct lock_class_key i_lock_key; | |
2470 | struct lock_class_key i_mutex_key; | |
730633f0 | 2471 | struct lock_class_key invalidate_lock_key; |
14358e6d | 2472 | struct lock_class_key i_mutex_dir_key; |
1da177e4 LT |
2473 | }; |
2474 | ||
7f78e035 EB |
2475 | #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME) |
2476 | ||
152a0836 AV |
2477 | extern struct dentry *mount_bdev(struct file_system_type *fs_type, |
2478 | int flags, const char *dev_name, void *data, | |
2479 | int (*fill_super)(struct super_block *, void *, int)); | |
fc14f2fe AV |
2480 | extern struct dentry *mount_single(struct file_system_type *fs_type, |
2481 | int flags, void *data, | |
2482 | int (*fill_super)(struct super_block *, void *, int)); | |
3c26ff6e AV |
2483 | extern struct dentry *mount_nodev(struct file_system_type *fs_type, |
2484 | int flags, void *data, | |
2485 | int (*fill_super)(struct super_block *, void *, int)); | |
ea441d11 | 2486 | extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path); |
04b94071 | 2487 | void retire_super(struct super_block *sb); |
1da177e4 LT |
2488 | void generic_shutdown_super(struct super_block *sb); |
2489 | void kill_block_super(struct super_block *sb); | |
2490 | void kill_anon_super(struct super_block *sb); | |
2491 | void kill_litter_super(struct super_block *sb); | |
2492 | void deactivate_super(struct super_block *sb); | |
74dbbdd7 | 2493 | void deactivate_locked_super(struct super_block *sb); |
1da177e4 | 2494 | int set_anon_super(struct super_block *s, void *data); |
cb50b348 | 2495 | int set_anon_super_fc(struct super_block *s, struct fs_context *fc); |
0ee5dc67 AV |
2496 | int get_anon_bdev(dev_t *); |
2497 | void free_anon_bdev(dev_t); | |
cb50b348 AV |
2498 | struct super_block *sget_fc(struct fs_context *fc, |
2499 | int (*test)(struct super_block *, struct fs_context *), | |
2500 | int (*set)(struct super_block *, struct fs_context *)); | |
1da177e4 LT |
2501 | struct super_block *sget(struct file_system_type *type, |
2502 | int (*test)(struct super_block *,void *), | |
2503 | int (*set)(struct super_block *,void *), | |
9249e17f | 2504 | int flags, void *data); |
69881be3 | 2505 | struct super_block *sget_dev(struct fs_context *fc, dev_t dev); |
1da177e4 LT |
2506 | |
2507 | /* Alas, no aliases. Too much hassle with bringing module.h everywhere */ | |
2508 | #define fops_get(fops) \ | |
2509 | (((fops) && try_module_get((fops)->owner) ? (fops) : NULL)) | |
2510 | #define fops_put(fops) \ | |
2511 | do { if (fops) module_put((fops)->owner); } while(0) | |
e84f9e57 AV |
2512 | /* |
2513 | * This one is to be used *ONLY* from ->open() instances. | |
2514 | * fops must be non-NULL, pinned down *and* module dependencies | |
2515 | * should be sufficient to pin the caller down as well. | |
2516 | */ | |
2517 | #define replace_fops(f, fops) \ | |
2518 | do { \ | |
2519 | struct file *__file = (f); \ | |
2520 | fops_put(__file->f_op); \ | |
2521 | BUG_ON(!(__file->f_op = (fops))); \ | |
2522 | } while(0) | |
1da177e4 LT |
2523 | |
2524 | extern int register_filesystem(struct file_system_type *); | |
2525 | extern int unregister_filesystem(struct file_system_type *); | |
f0bb5aaf | 2526 | extern int vfs_statfs(const struct path *, struct kstatfs *); |
c8b91acc AV |
2527 | extern int user_statfs(const char __user *, struct kstatfs *); |
2528 | extern int fd_statfs(int, struct kstatfs *); | |
880b9577 DW |
2529 | int freeze_super(struct super_block *super, enum freeze_holder who); |
2530 | int thaw_super(struct super_block *super, enum freeze_holder who); | |
fca39346 JK |
2531 | extern __printf(2, 3) |
2532 | int super_setup_bdi_name(struct super_block *sb, char *fmt, ...); | |
2533 | extern int super_setup_bdi(struct super_block *sb); | |
1da177e4 | 2534 | |
ce3b0f8d AV |
2535 | extern int current_umask(void); |
2536 | ||
5a14696c BF |
2537 | extern void ihold(struct inode * inode); |
2538 | extern void iput(struct inode *); | |
541d4c79 JL |
2539 | int inode_update_timestamps(struct inode *inode, int flags); |
2540 | int generic_update_time(struct inode *, int); | |
5a14696c | 2541 | |
f87fd4c2 | 2542 | /* /sys/fs */ |
00d26666 | 2543 | extern struct kobject *fs_kobj; |
f87fd4c2 | 2544 | |
09cbfeaf | 2545 | #define MAX_RW_COUNT (INT_MAX & PAGE_MASK) |
bfcd17a6 | 2546 | |
1da177e4 | 2547 | /* fs/open.c */ |
adb5c247 | 2548 | struct audit_names; |
91a27b2a | 2549 | struct filename { |
adb5c247 JL |
2550 | const char *name; /* pointer to actual string */ |
2551 | const __user char *uptr; /* original userland pointer */ | |
03adc61e | 2552 | atomic_t refcnt; |
1c949843 | 2553 | struct audit_names *aname; |
fd2f7cb5 | 2554 | const char iname[]; |
91a27b2a | 2555 | }; |
f1fffbd4 | 2556 | static_assert(offsetof(struct filename, iname) % sizeof(long) == 0); |
1da177e4 | 2557 | |
4a00c673 | 2558 | static inline struct mnt_idmap *file_mnt_idmap(const struct file *file) |
256c8aed CB |
2559 | { |
2560 | return mnt_idmap(file->f_path.mnt); | |
2561 | } | |
2562 | ||
bb49e9e7 CB |
2563 | /** |
2564 | * is_idmapped_mnt - check whether a mount is mapped | |
2565 | * @mnt: the mount to check | |
2566 | * | |
256c8aed | 2567 | * If @mnt has an non @nop_mnt_idmap attached to it then @mnt is mapped. |
bb49e9e7 CB |
2568 | * |
2569 | * Return: true if mount is mapped, false if not. | |
2570 | */ | |
2571 | static inline bool is_idmapped_mnt(const struct vfsmount *mnt) | |
2572 | { | |
256c8aed | 2573 | return mnt_idmap(mnt) != &nop_mnt_idmap; |
bb49e9e7 CB |
2574 | } |
2575 | ||
7df818b2 | 2576 | extern long vfs_truncate(const struct path *, loff_t); |
abf08576 | 2577 | int do_truncate(struct mnt_idmap *, struct dentry *, loff_t start, |
643fe55a | 2578 | unsigned int time_attrs, struct file *filp); |
72c72bdf | 2579 | extern int vfs_fallocate(struct file *file, int mode, loff_t offset, |
3e63cbb1 | 2580 | loff_t len); |
8e8a1407 | 2581 | extern long do_sys_open(int dfd, const char __user *filename, int flags, |
a218d0fd | 2582 | umode_t mode); |
669abf4e | 2583 | extern struct file *file_open_name(struct filename *, int, umode_t); |
a218d0fd | 2584 | extern struct file *filp_open(const char *, int, umode_t); |
ffb37ca3 | 2585 | extern struct file *file_open_root(const struct path *, |
378c6520 | 2586 | const char *, int, umode_t); |
ffb37ca3 AV |
2587 | static inline struct file *file_open_root_mnt(struct vfsmount *mnt, |
2588 | const char *name, int flags, umode_t mode) | |
2589 | { | |
2590 | return file_open_root(&(struct path){.mnt = mnt, .dentry = mnt->mnt_root}, | |
2591 | name, flags, mode); | |
2592 | } | |
62d53c4a AG |
2593 | struct file *dentry_open(const struct path *path, int flags, |
2594 | const struct cred *creds); | |
2595 | struct file *dentry_create(const struct path *path, int flags, umode_t mode, | |
2596 | const struct cred *cred); | |
def3ae83 | 2597 | struct path *backing_file_user_path(struct file *f); |
62d53c4a AG |
2598 | |
2599 | /* | |
08582d67 AG |
2600 | * When mmapping a file on a stackable filesystem (e.g., overlayfs), the file |
2601 | * stored in ->vm_file is a backing file whose f_inode is on the underlying | |
3efdc78f AV |
2602 | * filesystem. When the mapped file path and inode number are displayed to |
2603 | * user (e.g. via /proc/<pid>/maps), these helpers should be used to get the | |
2604 | * path and inode number to display to the user, which is the path of the fd | |
2605 | * that user has requested to map and the inode number that would be returned | |
2606 | * by fstat() on that same fd. | |
08582d67 | 2607 | */ |
3efdc78f | 2608 | /* Get the path to display in /proc/<pid>/maps */ |
08582d67 | 2609 | static inline const struct path *file_user_path(struct file *f) |
62d53c4a AG |
2610 | { |
2611 | if (unlikely(f->f_mode & FMODE_BACKING)) | |
def3ae83 | 2612 | return backing_file_user_path(f); |
62d53c4a AG |
2613 | return &f->f_path; |
2614 | } | |
3efdc78f AV |
2615 | /* Get the inode whose inode number to display in /proc/<pid>/maps */ |
2616 | static inline const struct inode *file_user_inode(struct file *f) | |
2617 | { | |
2618 | if (unlikely(f->f_mode & FMODE_BACKING)) | |
2619 | return d_inode(backing_file_user_path(f)->dentry); | |
2620 | return file_inode(f); | |
2621 | } | |
62d53c4a | 2622 | |
19f391eb AV |
2623 | static inline struct file *file_clone_open(struct file *file) |
2624 | { | |
2625 | return dentry_open(&file->f_path, file->f_flags, file->f_cred); | |
2626 | } | |
1da177e4 | 2627 | extern int filp_close(struct file *, fl_owner_t id); |
91a27b2a | 2628 | |
51f39a1f | 2629 | extern struct filename *getname_flags(const char __user *, int, int *); |
8228e2c3 | 2630 | extern struct filename *getname_uflags(const char __user *, int); |
91a27b2a | 2631 | extern struct filename *getname(const char __user *); |
c4ad8f98 | 2632 | extern struct filename *getname_kernel(const char *); |
55422d0b | 2633 | extern void putname(struct filename *name); |
91a27b2a | 2634 | |
30d90494 | 2635 | extern int finish_open(struct file *file, struct dentry *dentry, |
be12af3e | 2636 | int (*open)(struct inode *, struct file *)); |
e45198a6 | 2637 | extern int finish_no_open(struct file *file, struct dentry *dentry); |
1da177e4 | 2638 | |
9751b338 MS |
2639 | /* Helper for the simple case when original dentry is used */ |
2640 | static inline int finish_open_simple(struct file *file, int error) | |
2641 | { | |
2642 | if (error) | |
2643 | return error; | |
2644 | ||
2645 | return finish_open(file, file->f_path.dentry, NULL); | |
2646 | } | |
2647 | ||
1da177e4 LT |
2648 | /* fs/dcache.c */ |
2649 | extern void __init vfs_caches_init_early(void); | |
4248b0da | 2650 | extern void __init vfs_caches_init(void); |
1da177e4 | 2651 | |
b86c089b CL |
2652 | extern struct kmem_cache *names_cachep; |
2653 | ||
a608ca21 | 2654 | #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL) |
3446a8aa | 2655 | #define __putname(name) kmem_cache_free(names_cachep, (void *)(name)) |
1da177e4 | 2656 | |
a212b105 | 2657 | extern struct super_block *blockdev_superblock; |
a212b105 TH |
2658 | static inline bool sb_is_blkdev_sb(struct super_block *sb) |
2659 | { | |
dd0dca22 | 2660 | return IS_ENABLED(CONFIG_BLOCK) && sb == blockdev_superblock; |
47e4491b | 2661 | } |
5c0d6b60 | 2662 | |
b818f09e | 2663 | void emergency_thaw_all(void); |
60b0680f | 2664 | extern int sync_filesystem(struct super_block *); |
9361401e | 2665 | extern const struct file_operations def_blk_fops; |
4b6f5d20 | 2666 | extern const struct file_operations def_chr_fops; |
1da177e4 LT |
2667 | |
2668 | /* fs/char_dev.c */ | |
8a932f73 | 2669 | #define CHRDEV_MAJOR_MAX 512 |
49db08c3 LW |
2670 | /* Marks the bottom of the first segment of free char majors */ |
2671 | #define CHRDEV_MAJOR_DYN_END 234 | |
a5d31a3f LG |
2672 | /* Marks the top and bottom of the second segment of free char majors */ |
2673 | #define CHRDEV_MAJOR_DYN_EXT_START 511 | |
2674 | #define CHRDEV_MAJOR_DYN_EXT_END 384 | |
2675 | ||
1da177e4 LT |
2676 | extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *); |
2677 | extern int register_chrdev_region(dev_t, unsigned, const char *); | |
1905b1bf TH |
2678 | extern int __register_chrdev(unsigned int major, unsigned int baseminor, |
2679 | unsigned int count, const char *name, | |
2680 | const struct file_operations *fops); | |
2681 | extern void __unregister_chrdev(unsigned int major, unsigned int baseminor, | |
2682 | unsigned int count, const char *name); | |
1da177e4 | 2683 | extern void unregister_chrdev_region(dev_t, unsigned); |
68eef3b4 | 2684 | extern void chrdev_show(struct seq_file *,off_t); |
1da177e4 | 2685 | |
1905b1bf TH |
2686 | static inline int register_chrdev(unsigned int major, const char *name, |
2687 | const struct file_operations *fops) | |
2688 | { | |
2689 | return __register_chrdev(major, 0, 256, name, fops); | |
2690 | } | |
2691 | ||
2692 | static inline void unregister_chrdev(unsigned int major, const char *name) | |
2693 | { | |
2694 | __unregister_chrdev(major, 0, 256, name); | |
2695 | } | |
2696 | ||
1da177e4 LT |
2697 | extern void init_special_inode(struct inode *, umode_t, dev_t); |
2698 | ||
2699 | /* Invalid inode operations -- fs/bad_inode.c */ | |
2700 | extern void make_bad_inode(struct inode *); | |
0e3ef1fe | 2701 | extern bool is_bad_inode(struct inode *); |
1da177e4 | 2702 | |
a823e458 JL |
2703 | extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart, |
2704 | loff_t lend); | |
5660e13d JL |
2705 | extern int __must_check file_check_and_advance_wb_err(struct file *file); |
2706 | extern int __must_check file_write_and_wait_range(struct file *file, | |
2707 | loff_t start, loff_t end); | |
2708 | ||
a823e458 JL |
2709 | static inline int file_write_and_wait(struct file *file) |
2710 | { | |
2711 | return file_write_and_wait_range(file, 0, LLONG_MAX); | |
2712 | } | |
2713 | ||
8018ab05 CH |
2714 | extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end, |
2715 | int datasync); | |
2716 | extern int vfs_fsync(struct file *file, int datasync); | |
e2592217 | 2717 | |
22f96b38 JA |
2718 | extern int sync_file_range(struct file *file, loff_t offset, loff_t nbytes, |
2719 | unsigned int flags); | |
2720 | ||
91b94c5d AV |
2721 | static inline bool iocb_is_dsync(const struct kiocb *iocb) |
2722 | { | |
2723 | return (iocb->ki_flags & IOCB_DSYNC) || | |
2724 | IS_SYNC(iocb->ki_filp->f_mapping->host); | |
2725 | } | |
2726 | ||
e2592217 CH |
2727 | /* |
2728 | * Sync the bytes written if this was a synchronous write. Expect ki_pos | |
2729 | * to already be updated for the write, and will return either the amount | |
2730 | * of bytes passed in, or an error if syncing the file failed. | |
2731 | */ | |
2732 | static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count) | |
2733 | { | |
91b94c5d | 2734 | if (iocb_is_dsync(iocb)) { |
e2592217 CH |
2735 | int ret = vfs_fsync_range(iocb->ki_filp, |
2736 | iocb->ki_pos - count, iocb->ki_pos - 1, | |
2737 | (iocb->ki_flags & IOCB_SYNC) ? 0 : 1); | |
2738 | if (ret) | |
2739 | return ret; | |
2740 | } | |
2741 | ||
2742 | return count; | |
d311d79d | 2743 | } |
e2592217 | 2744 | |
1da177e4 LT |
2745 | extern void emergency_sync(void); |
2746 | extern void emergency_remount(void); | |
30460e1e | 2747 | |
9361401e | 2748 | #ifdef CONFIG_BLOCK |
30460e1e CM |
2749 | extern int bmap(struct inode *inode, sector_t *block); |
2750 | #else | |
2751 | static inline int bmap(struct inode *inode, sector_t *block) | |
2752 | { | |
2753 | return -EINVAL; | |
2754 | } | |
9361401e | 2755 | #endif |
30460e1e | 2756 | |
abf08576 | 2757 | int notify_change(struct mnt_idmap *, struct dentry *, |
2f221d6f | 2758 | struct iattr *, struct inode **); |
4609e1f1 CB |
2759 | int inode_permission(struct mnt_idmap *, struct inode *, int); |
2760 | int generic_permission(struct mnt_idmap *, struct inode *, int); | |
02f92b38 CB |
2761 | static inline int file_permission(struct file *file, int mask) |
2762 | { | |
4609e1f1 | 2763 | return inode_permission(file_mnt_idmap(file), |
47291baa | 2764 | file_inode(file), mask); |
02f92b38 CB |
2765 | } |
2766 | static inline int path_permission(const struct path *path, int mask) | |
2767 | { | |
4609e1f1 | 2768 | return inode_permission(mnt_idmap(path->mnt), |
47291baa | 2769 | d_inode(path->dentry), mask); |
02f92b38 | 2770 | } |
9452e93e | 2771 | int __check_sticky(struct mnt_idmap *idmap, struct inode *dir, |
ba73d987 | 2772 | struct inode *inode); |
1da177e4 | 2773 | |
f696a365 MS |
2774 | static inline bool execute_ok(struct inode *inode) |
2775 | { | |
2776 | return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode); | |
2777 | } | |
2778 | ||
6e3e2c43 AV |
2779 | static inline bool inode_wrong_type(const struct inode *inode, umode_t mode) |
2780 | { | |
2781 | return (inode->i_mode ^ mode) & S_IFMT; | |
2782 | } | |
2783 | ||
f6c05b9e AG |
2784 | /** |
2785 | * file_start_write - get write access to a superblock for regular file io | |
2786 | * @file: the file we want to write to | |
2787 | * | |
2788 | * This is a variant of sb_start_write() which is a noop on non-regualr file. | |
2789 | * Should be matched with a call to file_end_write(). | |
2790 | */ | |
8d71db4f AV |
2791 | static inline void file_start_write(struct file *file) |
2792 | { | |
2793 | if (!S_ISREG(file_inode(file)->i_mode)) | |
2794 | return; | |
8a3c84b6 | 2795 | sb_start_write(file_inode(file)->i_sb); |
8d71db4f AV |
2796 | } |
2797 | ||
5ae98f15 JK |
2798 | static inline bool file_start_write_trylock(struct file *file) |
2799 | { | |
2800 | if (!S_ISREG(file_inode(file)->i_mode)) | |
2801 | return true; | |
8a3c84b6 | 2802 | return sb_start_write_trylock(file_inode(file)->i_sb); |
5ae98f15 JK |
2803 | } |
2804 | ||
f6c05b9e AG |
2805 | /** |
2806 | * file_end_write - drop write access to a superblock of a regular file | |
2807 | * @file: the file we wrote to | |
2808 | * | |
2809 | * Should be matched with a call to file_start_write(). | |
2810 | */ | |
8d71db4f AV |
2811 | static inline void file_end_write(struct file *file) |
2812 | { | |
2813 | if (!S_ISREG(file_inode(file)->i_mode)) | |
2814 | return; | |
f6c05b9e | 2815 | sb_end_write(file_inode(file)->i_sb); |
8d71db4f | 2816 | } |
bfe219d3 | 2817 | |
ed0360bb AG |
2818 | /** |
2819 | * kiocb_start_write - get write access to a superblock for async file io | |
2820 | * @iocb: the io context we want to submit the write with | |
2821 | * | |
2822 | * This is a variant of sb_start_write() for async io submission. | |
2823 | * Should be matched with a call to kiocb_end_write(). | |
2824 | */ | |
2825 | static inline void kiocb_start_write(struct kiocb *iocb) | |
2826 | { | |
2827 | struct inode *inode = file_inode(iocb->ki_filp); | |
2828 | ||
2829 | sb_start_write(inode->i_sb); | |
2830 | /* | |
2831 | * Fool lockdep by telling it the lock got released so that it | |
2832 | * doesn't complain about the held lock when we return to userspace. | |
2833 | */ | |
2834 | __sb_writers_release(inode->i_sb, SB_FREEZE_WRITE); | |
2835 | } | |
2836 | ||
2837 | /** | |
2838 | * kiocb_end_write - drop write access to a superblock after async file io | |
2839 | * @iocb: the io context we sumbitted the write with | |
2840 | * | |
2841 | * Should be matched with a call to kiocb_start_write(). | |
2842 | */ | |
2843 | static inline void kiocb_end_write(struct kiocb *iocb) | |
2844 | { | |
2845 | struct inode *inode = file_inode(iocb->ki_filp); | |
2846 | ||
2847 | /* | |
2848 | * Tell lockdep we inherited freeze protection from submission thread. | |
2849 | */ | |
2850 | __sb_writers_acquired(inode->i_sb, SB_FREEZE_WRITE); | |
2851 | sb_end_write(inode->i_sb); | |
8d71db4f | 2852 | } |
bfe219d3 | 2853 | |
07b8ce1e | 2854 | /* |
592ca09b DH |
2855 | * This is used for regular files where some users -- especially the |
2856 | * currently executed binary in a process, previously handled via | |
2857 | * VM_DENYWRITE -- cannot handle concurrent write (and maybe mmap | |
2858 | * read-write shared) accesses. | |
2859 | * | |
07b8ce1e AV |
2860 | * get_write_access() gets write permission for a file. |
2861 | * put_write_access() releases this write permission. | |
592ca09b DH |
2862 | * deny_write_access() denies write access to a file. |
2863 | * allow_write_access() re-enables write access to a file. | |
2864 | * | |
2865 | * The i_writecount field of an inode can have the following values: | |
2866 | * 0: no write access, no denied write access | |
2867 | * < 0: (-i_writecount) users that denied write access to the file. | |
2868 | * > 0: (i_writecount) users that have write access to the file. | |
07b8ce1e AV |
2869 | * |
2870 | * Normally we operate on that counter with atomic_{inc,dec} and it's safe | |
2871 | * except for the cases where we don't hold i_writecount yet. Then we need to | |
2872 | * use {get,deny}_write_access() - these functions check the sign and refuse | |
2873 | * to do the change if sign is wrong. | |
2874 | */ | |
2875 | static inline int get_write_access(struct inode *inode) | |
2876 | { | |
2877 | return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY; | |
2878 | } | |
2879 | static inline int deny_write_access(struct file *file) | |
2880 | { | |
496ad9aa | 2881 | struct inode *inode = file_inode(file); |
07b8ce1e AV |
2882 | return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY; |
2883 | } | |
1da177e4 LT |
2884 | static inline void put_write_access(struct inode * inode) |
2885 | { | |
2886 | atomic_dec(&inode->i_writecount); | |
2887 | } | |
2888 | static inline void allow_write_access(struct file *file) | |
2889 | { | |
2890 | if (file) | |
496ad9aa | 2891 | atomic_inc(&file_inode(file)->i_writecount); |
1da177e4 | 2892 | } |
f1fe29b4 DH |
2893 | static inline bool inode_is_open_for_write(const struct inode *inode) |
2894 | { | |
2895 | return atomic_read(&inode->i_writecount) > 0; | |
2896 | } | |
2897 | ||
387e3746 | 2898 | #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING) |
a5c96ebf MZ |
2899 | static inline void i_readcount_dec(struct inode *inode) |
2900 | { | |
45e0d4b9 | 2901 | BUG_ON(atomic_dec_return(&inode->i_readcount) < 0); |
a5c96ebf MZ |
2902 | } |
2903 | static inline void i_readcount_inc(struct inode *inode) | |
2904 | { | |
2905 | atomic_inc(&inode->i_readcount); | |
2906 | } | |
2907 | #else | |
2908 | static inline void i_readcount_dec(struct inode *inode) | |
2909 | { | |
2910 | return; | |
2911 | } | |
2912 | static inline void i_readcount_inc(struct inode *inode) | |
2913 | { | |
2914 | return; | |
2915 | } | |
2916 | #endif | |
ed8cae8b | 2917 | extern int do_pipe_flags(int *, int); |
1da177e4 | 2918 | |
bdd1d2d3 | 2919 | extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *); |
61a707c5 | 2920 | ssize_t __kernel_read(struct file *file, void *buf, size_t count, loff_t *pos); |
e13ec939 | 2921 | extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *); |
73e18f7c | 2922 | extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *); |
1da177e4 LT |
2923 | extern struct file * open_exec(const char *); |
2924 | ||
2925 | /* fs/dcache.c -- generic fs support functions */ | |
a6e5787f | 2926 | extern bool is_subdir(struct dentry *, struct dentry *); |
640eb7e7 | 2927 | extern bool path_is_under(const struct path *, const struct path *); |
1da177e4 | 2928 | |
9bf39ab2 MS |
2929 | extern char *file_path(struct file *, char *, int); |
2930 | ||
1da177e4 LT |
2931 | #include <linux/err.h> |
2932 | ||
2933 | /* needed for stackable file system support */ | |
965c8e59 | 2934 | extern loff_t default_llseek(struct file *file, loff_t offset, int whence); |
1da177e4 | 2935 | |
965c8e59 | 2936 | extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence); |
1da177e4 | 2937 | |
54e34621 | 2938 | extern int inode_init_always(struct super_block *, struct inode *); |
1da177e4 | 2939 | extern void inode_init_once(struct inode *); |
2aa15890 | 2940 | extern void address_space_init_once(struct address_space *mapping); |
1da177e4 LT |
2941 | extern struct inode * igrab(struct inode *); |
2942 | extern ino_t iunique(struct super_block *, ino_t); | |
2943 | extern int inode_needs_sync(struct inode *inode); | |
45321ac5 | 2944 | extern int generic_delete_inode(struct inode *inode); |
9bcb4b73 AV |
2945 | static inline int generic_drop_inode(struct inode *inode) |
2946 | { | |
88149082 | 2947 | return !inode->i_nlink || inode_unhashed(inode); |
9bcb4b73 | 2948 | } |
2c567af4 | 2949 | extern void d_mark_dontcache(struct inode *inode); |
1da177e4 | 2950 | |
88bd5121 AA |
2951 | extern struct inode *ilookup5_nowait(struct super_block *sb, |
2952 | unsigned long hashval, int (*test)(struct inode *, void *), | |
2953 | void *data); | |
1da177e4 LT |
2954 | extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval, |
2955 | int (*test)(struct inode *, void *), void *data); | |
2956 | extern struct inode *ilookup(struct super_block *sb, unsigned long ino); | |
2957 | ||
80ea09a0 MS |
2958 | extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval, |
2959 | int (*test)(struct inode *, void *), | |
2960 | int (*set)(struct inode *, void *), | |
2961 | void *data); | |
1da177e4 LT |
2962 | extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *); |
2963 | extern struct inode * iget_locked(struct super_block *, unsigned long); | |
fe032c42 TT |
2964 | extern struct inode *find_inode_nowait(struct super_block *, |
2965 | unsigned long, | |
2966 | int (*match)(struct inode *, | |
2967 | unsigned long, void *), | |
2968 | void *data); | |
3f19b2ab DH |
2969 | extern struct inode *find_inode_rcu(struct super_block *, unsigned long, |
2970 | int (*)(struct inode *, void *), void *); | |
2971 | extern struct inode *find_inode_by_ino_rcu(struct super_block *, unsigned long); | |
261bca86 AV |
2972 | extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *); |
2973 | extern int insert_inode_locked(struct inode *); | |
e096d0c7 JB |
2974 | #ifdef CONFIG_DEBUG_LOCK_ALLOC |
2975 | extern void lockdep_annotate_inode_mutex_key(struct inode *inode); | |
2976 | #else | |
2977 | static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { }; | |
2978 | #endif | |
1da177e4 | 2979 | extern void unlock_new_inode(struct inode *); |
c2b6d621 | 2980 | extern void discard_new_inode(struct inode *); |
85fe4025 | 2981 | extern unsigned int get_next_ino(void); |
799ea9e9 | 2982 | extern void evict_inodes(struct super_block *sb); |
3e9d80a8 | 2983 | void dump_mapping(const struct address_space *); |
1da177e4 | 2984 | |
e809d5f0 | 2985 | /* |
b6334e2c | 2986 | * Userspace may rely on the inode number being non-zero. For example, glibc |
e809d5f0 CD |
2987 | * simply ignores files with zero i_ino in unlink() and other places. |
2988 | * | |
2989 | * As an additional complication, if userspace was compiled with | |
2990 | * _FILE_OFFSET_BITS=32 on a 64-bit kernel we'll only end up reading out the | |
2991 | * lower 32 bits, so we need to check that those aren't zero explicitly. With | |
2992 | * _FILE_OFFSET_BITS=64, this may cause some harmless false-negatives, but | |
2993 | * better safe than sorry. | |
2994 | */ | |
2995 | static inline bool is_zero_ino(ino_t ino) | |
2996 | { | |
2997 | return (u32)ino == 0; | |
2998 | } | |
2999 | ||
1da177e4 | 3000 | extern void __iget(struct inode * inode); |
b46980fe | 3001 | extern void iget_failed(struct inode *); |
dbd5768f | 3002 | extern void clear_inode(struct inode *); |
2e00c97e | 3003 | extern void __destroy_inode(struct inode *); |
a209dfc7 ED |
3004 | extern struct inode *new_inode_pseudo(struct super_block *sb); |
3005 | extern struct inode *new_inode(struct super_block *sb); | |
ff0c7d15 | 3006 | extern void free_inode_nonrcu(struct inode *inode); |
9452e93e | 3007 | extern int setattr_should_drop_suidgid(struct mnt_idmap *, struct inode *); |
5fa8e0a1 | 3008 | extern int file_remove_privs(struct file *); |
4f704d9a CB |
3009 | int setattr_should_drop_sgid(struct mnt_idmap *idmap, |
3010 | const struct inode *inode); | |
1da177e4 | 3011 | |
8b9f3ac5 MS |
3012 | /* |
3013 | * This must be used for allocating filesystems specific inodes to set | |
3014 | * up the inode reclaim context correctly. | |
3015 | */ | |
3016 | static inline void * | |
3017 | alloc_inode_sb(struct super_block *sb, struct kmem_cache *cache, gfp_t gfp) | |
3018 | { | |
3019 | return kmem_cache_alloc_lru(cache, &sb->s_inode_lru, gfp); | |
3020 | } | |
3021 | ||
1da177e4 | 3022 | extern void __insert_inode_hash(struct inode *, unsigned long hashval); |
646ec461 CH |
3023 | static inline void insert_inode_hash(struct inode *inode) |
3024 | { | |
1da177e4 LT |
3025 | __insert_inode_hash(inode, inode->i_ino); |
3026 | } | |
f2ee7abf ED |
3027 | |
3028 | extern void __remove_inode_hash(struct inode *); | |
3029 | static inline void remove_inode_hash(struct inode *inode) | |
3030 | { | |
cbedaac6 | 3031 | if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash)) |
f2ee7abf ED |
3032 | __remove_inode_hash(inode); |
3033 | } | |
3034 | ||
646ec461 | 3035 | extern void inode_sb_list_add(struct inode *inode); |
51b8c1fe | 3036 | extern void inode_add_lru(struct inode *inode); |
1da177e4 | 3037 | |
1da177e4 LT |
3038 | extern int sb_set_blocksize(struct super_block *, int); |
3039 | extern int sb_min_blocksize(struct super_block *, int); | |
3040 | ||
3041 | extern int generic_file_mmap(struct file *, struct vm_area_struct *); | |
3042 | extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *); | |
3309dd04 | 3043 | extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *); |
f6f7a25a | 3044 | int generic_write_checks_count(struct kiocb *iocb, loff_t *count); |
02e83f46 DW |
3045 | extern int generic_write_check_limits(struct file *file, loff_t pos, |
3046 | loff_t *count); | |
a3171351 | 3047 | extern int generic_file_rw_checks(struct file *file_in, struct file *file_out); |
87fa0f3e CH |
3048 | ssize_t filemap_read(struct kiocb *iocb, struct iov_iter *to, |
3049 | ssize_t already_read); | |
ed978a81 | 3050 | extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *); |
8174202b | 3051 | extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *); |
8174202b | 3052 | extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *); |
1af5bb49 | 3053 | extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *); |
800ba295 | 3054 | ssize_t generic_perform_write(struct kiocb *, struct iov_iter *); |
44fff0fa CH |
3055 | ssize_t direct_write_fallback(struct kiocb *iocb, struct iov_iter *iter, |
3056 | ssize_t direct_written, ssize_t buffered_written); | |
88e6faef | 3057 | |
18e9710e | 3058 | ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos, |
ddef7ed2 | 3059 | rwf_t flags); |
abbb6589 | 3060 | ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos, |
ddef7ed2 | 3061 | rwf_t flags); |
5dcdc43e JX |
3062 | ssize_t vfs_iocb_iter_read(struct file *file, struct kiocb *iocb, |
3063 | struct iov_iter *iter); | |
3064 | ssize_t vfs_iocb_iter_write(struct file *file, struct kiocb *iocb, | |
3065 | struct iov_iter *iter); | |
dbe4e192 | 3066 | |
88e6faef | 3067 | /* fs/splice.c */ |
07073eb0 DH |
3068 | ssize_t filemap_splice_read(struct file *in, loff_t *ppos, |
3069 | struct pipe_inode_info *pipe, | |
3070 | size_t len, unsigned int flags); | |
69df79a4 DH |
3071 | ssize_t copy_splice_read(struct file *in, loff_t *ppos, |
3072 | struct pipe_inode_info *pipe, | |
3073 | size_t len, unsigned int flags); | |
8d020765 | 3074 | extern ssize_t iter_file_splice_write(struct pipe_inode_info *, |
cbb7e577 | 3075 | struct file *, loff_t *, size_t, unsigned int); |
1c118596 | 3076 | |
88e6faef | 3077 | |
1da177e4 LT |
3078 | extern void |
3079 | file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping); | |
965c8e59 | 3080 | extern loff_t noop_llseek(struct file *file, loff_t offset, int whence); |
868941b1 | 3081 | #define no_llseek NULL |
46a1c2c7 | 3082 | extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize); |
965c8e59 | 3083 | extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence); |
5760495a | 3084 | extern loff_t generic_file_llseek_size(struct file *file, loff_t offset, |
965c8e59 | 3085 | int whence, loff_t maxsize, loff_t eof); |
1bf9d14d AV |
3086 | extern loff_t fixed_size_llseek(struct file *file, loff_t offset, |
3087 | int whence, loff_t size); | |
b25472f9 AV |
3088 | extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t); |
3089 | extern loff_t no_seek_end_llseek(struct file *, loff_t, int); | |
87112933 | 3090 | int rw_verify_area(int, struct file *, const loff_t *, size_t); |
1da177e4 LT |
3091 | extern int generic_file_open(struct inode * inode, struct file * filp); |
3092 | extern int nonseekable_open(struct inode * inode, struct file * filp); | |
10dce8af | 3093 | extern int stream_open(struct inode * inode, struct file * filp); |
1da177e4 | 3094 | |
9361401e | 3095 | #ifdef CONFIG_BLOCK |
8a4c1e42 | 3096 | typedef void (dio_submit_t)(struct bio *bio, struct inode *inode, |
facd07b0 | 3097 | loff_t file_offset); |
1da177e4 LT |
3098 | |
3099 | enum { | |
1e431f5c CH |
3100 | /* need locking between buffered and direct access */ |
3101 | DIO_LOCKING = 0x01, | |
3102 | ||
3103 | /* filesystem does not support filling holes */ | |
3104 | DIO_SKIP_HOLES = 0x02, | |
1da177e4 LT |
3105 | }; |
3106 | ||
17f8c842 OS |
3107 | ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode, |
3108 | struct block_device *bdev, struct iov_iter *iter, | |
c8b8e32d | 3109 | get_block_t get_block, |
0aaf08de | 3110 | dio_iodone_t end_io, |
17f8c842 | 3111 | int flags); |
7bb46a67 | 3112 | |
17f8c842 OS |
3113 | static inline ssize_t blockdev_direct_IO(struct kiocb *iocb, |
3114 | struct inode *inode, | |
c8b8e32d | 3115 | struct iov_iter *iter, |
17f8c842 | 3116 | get_block_t get_block) |
1da177e4 | 3117 | { |
17f8c842 | 3118 | return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter, |
0aaf08de | 3119 | get_block, NULL, DIO_LOCKING | DIO_SKIP_HOLES); |
1da177e4 | 3120 | } |
9361401e | 3121 | #endif |
1da177e4 | 3122 | |
1d59d61f | 3123 | void inode_dio_wait(struct inode *inode); |
fe0f07d0 | 3124 | |
39015399 | 3125 | /** |
fe0f07d0 JA |
3126 | * inode_dio_begin - signal start of a direct I/O requests |
3127 | * @inode: inode the direct I/O happens on | |
3128 | * | |
3129 | * This is called once we've finished processing a direct I/O request, | |
3130 | * and is used to wake up callers waiting for direct I/O to be quiesced. | |
3131 | */ | |
3132 | static inline void inode_dio_begin(struct inode *inode) | |
3133 | { | |
3134 | atomic_inc(&inode->i_dio_count); | |
3135 | } | |
3136 | ||
39015399 | 3137 | /** |
fe0f07d0 JA |
3138 | * inode_dio_end - signal finish of a direct I/O requests |
3139 | * @inode: inode the direct I/O happens on | |
3140 | * | |
3141 | * This is called once we've finished processing a direct I/O request, | |
3142 | * and is used to wake up callers waiting for direct I/O to be quiesced. | |
3143 | */ | |
3144 | static inline void inode_dio_end(struct inode *inode) | |
3145 | { | |
3146 | if (atomic_dec_and_test(&inode->i_dio_count)) | |
3147 | wake_up_bit(&inode->i_state, __I_DIO_WAKEUP); | |
3148 | } | |
1d59d61f | 3149 | |
5f16f322 TT |
3150 | extern void inode_set_flags(struct inode *inode, unsigned int flags, |
3151 | unsigned int mask); | |
3152 | ||
4b6f5d20 | 3153 | extern const struct file_operations generic_ro_fops; |
1da177e4 LT |
3154 | |
3155 | #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m)) | |
3156 | ||
5d826c84 | 3157 | extern int readlink_copy(char __user *, int, const char *); |
1da177e4 | 3158 | extern int page_readlink(struct dentry *, char __user *, int); |
fceef393 AV |
3159 | extern const char *page_get_link(struct dentry *, struct inode *, |
3160 | struct delayed_call *); | |
3161 | extern void page_put_link(void *); | |
1da177e4 | 3162 | extern int page_symlink(struct inode *inode, const char *symname, int len); |
c5ef1c42 | 3163 | extern const struct inode_operations page_symlink_inode_operations; |
fceef393 | 3164 | extern void kfree_link(void *); |
0d72b928 | 3165 | void generic_fillattr(struct mnt_idmap *, u32, struct inode *, struct kstat *); |
4f911138 | 3166 | void generic_fill_statx_attr(struct inode *inode, struct kstat *stat); |
a528d35e DH |
3167 | extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int); |
3168 | extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int); | |
b462707e | 3169 | void __inode_add_bytes(struct inode *inode, loff_t bytes); |
1da177e4 | 3170 | void inode_add_bytes(struct inode *inode, loff_t bytes); |
1c8924eb | 3171 | void __inode_sub_bytes(struct inode *inode, loff_t bytes); |
1da177e4 | 3172 | void inode_sub_bytes(struct inode *inode, loff_t bytes); |
f4a8116a JK |
3173 | static inline loff_t __inode_get_bytes(struct inode *inode) |
3174 | { | |
3175 | return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes; | |
3176 | } | |
1da177e4 LT |
3177 | loff_t inode_get_bytes(struct inode *inode); |
3178 | void inode_set_bytes(struct inode *inode, loff_t bytes); | |
fceef393 AV |
3179 | const char *simple_get_link(struct dentry *, struct inode *, |
3180 | struct delayed_call *); | |
61ba64fc | 3181 | extern const struct inode_operations simple_symlink_inode_operations; |
1da177e4 | 3182 | |
5c0ba4e0 | 3183 | extern int iterate_dir(struct file *, struct dir_context *); |
1da177e4 | 3184 | |
09f1bde4 CH |
3185 | int vfs_fstatat(int dfd, const char __user *filename, struct kstat *stat, |
3186 | int flags); | |
da9aa5d9 | 3187 | int vfs_fstat(int fd, struct kstat *stat); |
a528d35e DH |
3188 | |
3189 | static inline int vfs_stat(const char __user *filename, struct kstat *stat) | |
3190 | { | |
0b2c6693 | 3191 | return vfs_fstatat(AT_FDCWD, filename, stat, 0); |
a528d35e DH |
3192 | } |
3193 | static inline int vfs_lstat(const char __user *name, struct kstat *stat) | |
3194 | { | |
0b2c6693 | 3195 | return vfs_fstatat(AT_FDCWD, name, stat, AT_SYMLINK_NOFOLLOW); |
a528d35e | 3196 | } |
a528d35e | 3197 | |
d60874cd | 3198 | extern const char *vfs_get_link(struct dentry *, struct delayed_call *); |
fd4a0edf | 3199 | extern int vfs_readlink(struct dentry *, char __user *, int); |
1da177e4 | 3200 | |
ee416bcd | 3201 | extern struct file_system_type *get_filesystem(struct file_system_type *fs); |
c18479fe | 3202 | extern void put_filesystem(struct file_system_type *fs); |
1da177e4 | 3203 | extern struct file_system_type *get_fs_type(const char *name); |
1da177e4 | 3204 | extern void drop_super(struct super_block *sb); |
ba6379f7 | 3205 | extern void drop_super_exclusive(struct super_block *sb); |
01a05b33 | 3206 | extern void iterate_supers(void (*)(struct super_block *, void *), void *); |
43e15cdb AV |
3207 | extern void iterate_supers_type(struct file_system_type *, |
3208 | void (*)(struct super_block *, void *), void *); | |
1da177e4 LT |
3209 | |
3210 | extern int dcache_dir_open(struct inode *, struct file *); | |
3211 | extern int dcache_dir_close(struct inode *, struct file *); | |
3212 | extern loff_t dcache_dir_lseek(struct file *, loff_t, int); | |
5f99f4e7 | 3213 | extern int dcache_readdir(struct file *, struct dir_context *); |
c1632a0f | 3214 | extern int simple_setattr(struct mnt_idmap *, struct dentry *, |
549c7297 | 3215 | struct iattr *); |
b74d24f7 | 3216 | extern int simple_getattr(struct mnt_idmap *, const struct path *, |
549c7297 | 3217 | struct kstat *, u32, unsigned int); |
726c3342 | 3218 | extern int simple_statfs(struct dentry *, struct kstatfs *); |
20955e89 | 3219 | extern int simple_open(struct inode *inode, struct file *file); |
1da177e4 LT |
3220 | extern int simple_link(struct dentry *, struct inode *, struct dentry *); |
3221 | extern int simple_unlink(struct inode *, struct dentry *); | |
3222 | extern int simple_rmdir(struct inode *, struct dentry *); | |
0c476792 JL |
3223 | void simple_rename_timestamp(struct inode *old_dir, struct dentry *old_dentry, |
3224 | struct inode *new_dir, struct dentry *new_dentry); | |
6429e463 LB |
3225 | extern int simple_rename_exchange(struct inode *old_dir, struct dentry *old_dentry, |
3226 | struct inode *new_dir, struct dentry *new_dentry); | |
e18275ae | 3227 | extern int simple_rename(struct mnt_idmap *, struct inode *, |
549c7297 CB |
3228 | struct dentry *, struct inode *, struct dentry *, |
3229 | unsigned int); | |
a3d1e7eb AV |
3230 | extern void simple_recursive_removal(struct dentry *, |
3231 | void (*callback)(struct dentry *)); | |
02c24a82 | 3232 | extern int noop_fsync(struct file *, loff_t, loff_t, int); |
f44c7763 | 3233 | extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter); |
1da177e4 | 3234 | extern int simple_empty(struct dentry *); |
afddba49 | 3235 | extern int simple_write_begin(struct file *file, struct address_space *mapping, |
9d6b0cd7 | 3236 | loff_t pos, unsigned len, |
afddba49 | 3237 | struct page **pagep, void **fsdata); |
c1e3dbe9 | 3238 | extern const struct address_space_operations ram_aops; |
b26d4cd3 | 3239 | extern int always_delete_dentry(const struct dentry *); |
6987843f | 3240 | extern struct inode *alloc_anon_inode(struct super_block *); |
ed5f17f6 | 3241 | extern int simple_nosetlease(struct file *, int, struct file_lock **, void **); |
b26d4cd3 | 3242 | extern const struct dentry_operations simple_dentry_operations; |
1da177e4 | 3243 | |
00cd8dd3 | 3244 | extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags); |
1da177e4 | 3245 | extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *); |
4b6f5d20 | 3246 | extern const struct file_operations simple_dir_operations; |
c5ef1c42 | 3247 | extern const struct inode_operations simple_dir_inode_operations; |
fbabfd0f EB |
3248 | extern void make_empty_dir_inode(struct inode *inode); |
3249 | extern bool is_empty_dir_inode(struct inode *inode); | |
cda37124 | 3250 | struct tree_descr { const char *name; const struct file_operations *ops; int mode; }; |
1da177e4 | 3251 | struct dentry *d_alloc_name(struct dentry *, const char *); |
cda37124 EB |
3252 | extern int simple_fill_super(struct super_block *, unsigned long, |
3253 | const struct tree_descr *); | |
1f5ce9e9 | 3254 | extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count); |
1da177e4 LT |
3255 | extern void simple_release_fs(struct vfsmount **mount, int *count); |
3256 | ||
93b07113 AM |
3257 | extern ssize_t simple_read_from_buffer(void __user *to, size_t count, |
3258 | loff_t *ppos, const void *from, size_t available); | |
6a727b43 JS |
3259 | extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos, |
3260 | const void __user *from, size_t count); | |
1da177e4 | 3261 | |
6faddda6 CL |
3262 | struct offset_ctx { |
3263 | struct xarray xa; | |
3264 | u32 next_offset; | |
3265 | }; | |
3266 | ||
3267 | void simple_offset_init(struct offset_ctx *octx); | |
3268 | int simple_offset_add(struct offset_ctx *octx, struct dentry *dentry); | |
3269 | void simple_offset_remove(struct offset_ctx *octx, struct dentry *dentry); | |
3270 | int simple_offset_rename_exchange(struct inode *old_dir, | |
3271 | struct dentry *old_dentry, | |
3272 | struct inode *new_dir, | |
3273 | struct dentry *new_dentry); | |
3274 | void simple_offset_destroy(struct offset_ctx *octx); | |
3275 | ||
3276 | extern const struct file_operations simple_offset_dir_operations; | |
3277 | ||
ac13a829 | 3278 | extern int __generic_file_fsync(struct file *, loff_t, loff_t, int); |
02c24a82 | 3279 | extern int generic_file_fsync(struct file *, loff_t, loff_t, int); |
d5aacad5 | 3280 | |
30ca22c7 PL |
3281 | extern int generic_check_addressable(unsigned, u64); |
3282 | ||
608af703 | 3283 | extern void generic_set_encrypted_ci_d_ops(struct dentry *dentry); |
c843843e | 3284 | |
4609e1f1 | 3285 | int may_setattr(struct mnt_idmap *idmap, struct inode *inode, |
7bb698f0 | 3286 | unsigned int ia_valid); |
c1632a0f | 3287 | int setattr_prepare(struct mnt_idmap *, struct dentry *, struct iattr *); |
25d9e2d1 | 3288 | extern int inode_newsize_ok(const struct inode *, loff_t offset); |
c1632a0f | 3289 | void setattr_copy(struct mnt_idmap *, struct inode *inode, |
2f221d6f | 3290 | const struct iattr *attr); |
1da177e4 | 3291 | |
c3b2da31 | 3292 | extern int file_update_time(struct file *file); |
1da177e4 | 3293 | |
f05a3849 | 3294 | static inline bool vma_is_dax(const struct vm_area_struct *vma) |
baabda26 DW |
3295 | { |
3296 | return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host); | |
3297 | } | |
3298 | ||
2bb6d283 DW |
3299 | static inline bool vma_is_fsdax(struct vm_area_struct *vma) |
3300 | { | |
3301 | struct inode *inode; | |
3302 | ||
52650c8b | 3303 | if (!IS_ENABLED(CONFIG_FS_DAX) || !vma->vm_file) |
2bb6d283 DW |
3304 | return false; |
3305 | if (!vma_is_dax(vma)) | |
3306 | return false; | |
3307 | inode = file_inode(vma->vm_file); | |
230f5a89 | 3308 | if (S_ISCHR(inode->i_mode)) |
2bb6d283 DW |
3309 | return false; /* device-dax */ |
3310 | return true; | |
3311 | } | |
3312 | ||
2ba48ce5 AV |
3313 | static inline int iocb_flags(struct file *file) |
3314 | { | |
3315 | int res = 0; | |
3316 | if (file->f_flags & O_APPEND) | |
3317 | res |= IOCB_APPEND; | |
efbe3c24 | 3318 | if (file->f_flags & O_DIRECT) |
2ba48ce5 | 3319 | res |= IOCB_DIRECT; |
91b94c5d | 3320 | if (file->f_flags & O_DSYNC) |
dde0c2e7 CH |
3321 | res |= IOCB_DSYNC; |
3322 | if (file->f_flags & __O_SYNC) | |
3323 | res |= IOCB_SYNC; | |
2ba48ce5 AV |
3324 | return res; |
3325 | } | |
3326 | ||
ddef7ed2 | 3327 | static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags) |
fdd2f5b7 | 3328 | { |
1752f0ad PB |
3329 | int kiocb_flags = 0; |
3330 | ||
ce71bfea JA |
3331 | /* make sure there's no overlap between RWF and private IOCB flags */ |
3332 | BUILD_BUG_ON((__force int) RWF_SUPPORTED & IOCB_EVENTFD); | |
3333 | ||
1752f0ad PB |
3334 | if (!flags) |
3335 | return 0; | |
fdd2f5b7 GR |
3336 | if (unlikely(flags & ~RWF_SUPPORTED)) |
3337 | return -EOPNOTSUPP; | |
3338 | ||
b745fafa | 3339 | if (flags & RWF_NOWAIT) { |
91f9943e | 3340 | if (!(ki->ki_filp->f_mode & FMODE_NOWAIT)) |
b745fafa | 3341 | return -EOPNOTSUPP; |
ce71bfea | 3342 | kiocb_flags |= IOCB_NOIO; |
b745fafa | 3343 | } |
ce71bfea | 3344 | kiocb_flags |= (__force int) (flags & RWF_SUPPORTED); |
fdd2f5b7 | 3345 | if (flags & RWF_SYNC) |
ce71bfea | 3346 | kiocb_flags |= IOCB_DSYNC; |
1752f0ad PB |
3347 | |
3348 | ki->ki_flags |= kiocb_flags; | |
fdd2f5b7 GR |
3349 | return 0; |
3350 | } | |
3351 | ||
1da177e4 LT |
3352 | static inline ino_t parent_ino(struct dentry *dentry) |
3353 | { | |
3354 | ino_t res; | |
3355 | ||
b5c84bf6 NP |
3356 | /* |
3357 | * Don't strictly need d_lock here? If the parent ino could change | |
3358 | * then surely we'd have a deeper race in the caller? | |
3359 | */ | |
1da177e4 LT |
3360 | spin_lock(&dentry->d_lock); |
3361 | res = dentry->d_parent->d_inode->i_ino; | |
3362 | spin_unlock(&dentry->d_lock); | |
3363 | return res; | |
3364 | } | |
3365 | ||
1da177e4 LT |
3366 | /* Transaction based IO helpers */ |
3367 | ||
3368 | /* | |
3369 | * An argresp is stored in an allocated page and holds the | |
3370 | * size of the argument or response, along with its content | |
3371 | */ | |
3372 | struct simple_transaction_argresp { | |
3373 | ssize_t size; | |
5e01fdff | 3374 | char data[]; |
1da177e4 LT |
3375 | }; |
3376 | ||
3377 | #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp)) | |
3378 | ||
3379 | char *simple_transaction_get(struct file *file, const char __user *buf, | |
3380 | size_t size); | |
3381 | ssize_t simple_transaction_read(struct file *file, char __user *buf, | |
3382 | size_t size, loff_t *pos); | |
3383 | int simple_transaction_release(struct inode *inode, struct file *file); | |
3384 | ||
76791ab2 | 3385 | void simple_transaction_set(struct file *file, size_t n); |
1da177e4 | 3386 | |
acaefc25 AB |
3387 | /* |
3388 | * simple attribute files | |
3389 | * | |
3390 | * These attributes behave similar to those in sysfs: | |
3391 | * | |
3392 | * Writing to an attribute immediately sets a value, an open file can be | |
3393 | * written to multiple times. | |
3394 | * | |
3395 | * Reading from an attribute creates a buffer from the value that might get | |
3396 | * read with multiple read calls. When the attribute has been read | |
3397 | * completely, no further read calls are possible until the file is opened | |
3398 | * again. | |
3399 | * | |
3400 | * All attributes contain a text representation of a numeric value | |
3401 | * that are accessed with the get() and set() functions. | |
3402 | */ | |
2e41f274 | 3403 | #define DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, __is_signed) \ |
acaefc25 AB |
3404 | static int __fops ## _open(struct inode *inode, struct file *file) \ |
3405 | { \ | |
3406 | __simple_attr_check_format(__fmt, 0ull); \ | |
3407 | return simple_attr_open(inode, file, __get, __set, __fmt); \ | |
3408 | } \ | |
828c0950 | 3409 | static const struct file_operations __fops = { \ |
acaefc25 AB |
3410 | .owner = THIS_MODULE, \ |
3411 | .open = __fops ## _open, \ | |
74bedc4d | 3412 | .release = simple_attr_release, \ |
acaefc25 | 3413 | .read = simple_attr_read, \ |
2e41f274 | 3414 | .write = (__is_signed) ? simple_attr_write_signed : simple_attr_write, \ |
1ec5584e | 3415 | .llseek = generic_file_llseek, \ |
68be3029 | 3416 | } |
acaefc25 | 3417 | |
2e41f274 AM |
3418 | #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \ |
3419 | DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, false) | |
3420 | ||
3421 | #define DEFINE_SIMPLE_ATTRIBUTE_SIGNED(__fops, __get, __set, __fmt) \ | |
3422 | DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, true) | |
3423 | ||
b9075fa9 JP |
3424 | static inline __printf(1, 2) |
3425 | void __simple_attr_check_format(const char *fmt, ...) | |
acaefc25 AB |
3426 | { |
3427 | /* don't do anything, just let the compiler check the arguments; */ | |
3428 | } | |
3429 | ||
3430 | int simple_attr_open(struct inode *inode, struct file *file, | |
8b88b099 | 3431 | int (*get)(void *, u64 *), int (*set)(void *, u64), |
acaefc25 | 3432 | const char *fmt); |
74bedc4d | 3433 | int simple_attr_release(struct inode *inode, struct file *file); |
acaefc25 AB |
3434 | ssize_t simple_attr_read(struct file *file, char __user *buf, |
3435 | size_t len, loff_t *ppos); | |
3436 | ssize_t simple_attr_write(struct file *file, const char __user *buf, | |
3437 | size_t len, loff_t *ppos); | |
2e41f274 AM |
3438 | ssize_t simple_attr_write_signed(struct file *file, const char __user *buf, |
3439 | size_t len, loff_t *ppos); | |
acaefc25 | 3440 | |
4be28540 | 3441 | struct ctl_table; |
6e7c1770 | 3442 | int __init list_bdev_fs_names(char *buf, size_t size); |
62239ac2 | 3443 | |
3cd90ea4 | 3444 | #define __FMODE_EXEC ((__force int) FMODE_EXEC) |
1a44bc8c NK |
3445 | #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY) |
3446 | ||
6d125529 | 3447 | #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE]) |
ecf081d1 | 3448 | #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \ |
1a44bc8c | 3449 | (flag & __FMODE_NONOTIFY))) |
5300990c | 3450 | |
d37177ba | 3451 | static inline bool is_sxid(umode_t mode) |
69b45732 | 3452 | { |
8d84e39d | 3453 | return mode & (S_ISUID | S_ISGID); |
69b45732 AK |
3454 | } |
3455 | ||
9452e93e | 3456 | static inline int check_sticky(struct mnt_idmap *idmap, |
ba73d987 | 3457 | struct inode *dir, struct inode *inode) |
cbdf35bc MS |
3458 | { |
3459 | if (!(dir->i_mode & S_ISVTX)) | |
3460 | return 0; | |
3461 | ||
9452e93e | 3462 | return __check_sticky(idmap, dir, inode); |
cbdf35bc MS |
3463 | } |
3464 | ||
69b45732 AK |
3465 | static inline void inode_has_no_xattr(struct inode *inode) |
3466 | { | |
e462ec50 | 3467 | if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC)) |
69b45732 AK |
3468 | inode->i_flags |= S_NOSEC; |
3469 | } | |
3470 | ||
a7400222 AV |
3471 | static inline bool is_root_inode(struct inode *inode) |
3472 | { | |
3473 | return inode == inode->i_sb->s_root->d_inode; | |
3474 | } | |
3475 | ||
5f99f4e7 AV |
3476 | static inline bool dir_emit(struct dir_context *ctx, |
3477 | const char *name, int namelen, | |
3478 | u64 ino, unsigned type) | |
3479 | { | |
25885a35 | 3480 | return ctx->actor(ctx, name, namelen, ctx->pos, ino, type); |
5f99f4e7 AV |
3481 | } |
3482 | static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx) | |
3483 | { | |
3484 | return ctx->actor(ctx, ".", 1, ctx->pos, | |
25885a35 | 3485 | file->f_path.dentry->d_inode->i_ino, DT_DIR); |
5f99f4e7 AV |
3486 | } |
3487 | static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx) | |
3488 | { | |
3489 | return ctx->actor(ctx, "..", 2, ctx->pos, | |
25885a35 | 3490 | parent_ino(file->f_path.dentry), DT_DIR); |
5f99f4e7 AV |
3491 | } |
3492 | static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx) | |
3493 | { | |
3494 | if (ctx->pos == 0) { | |
3495 | if (!dir_emit_dot(file, ctx)) | |
3496 | return false; | |
3497 | ctx->pos = 1; | |
3498 | } | |
3499 | if (ctx->pos == 1) { | |
3500 | if (!dir_emit_dotdot(file, ctx)) | |
3501 | return false; | |
3502 | ctx->pos = 2; | |
3503 | } | |
3504 | return true; | |
3505 | } | |
5ded75ec AV |
3506 | static inline bool dir_relax(struct inode *inode) |
3507 | { | |
5955102c AV |
3508 | inode_unlock(inode); |
3509 | inode_lock(inode); | |
5ded75ec AV |
3510 | return !IS_DEADDIR(inode); |
3511 | } | |
5f99f4e7 | 3512 | |
ae05327a AV |
3513 | static inline bool dir_relax_shared(struct inode *inode) |
3514 | { | |
3515 | inode_unlock_shared(inode); | |
3516 | inode_lock_shared(inode); | |
3517 | return !IS_DEADDIR(inode); | |
3518 | } | |
3519 | ||
90f8572b | 3520 | extern bool path_noexec(const struct path *path); |
21fc61c7 | 3521 | extern void inode_nohighmem(struct inode *inode); |
90f8572b | 3522 | |
45cd0faa AG |
3523 | /* mm/fadvise.c */ |
3524 | extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len, | |
3525 | int advice); | |
cf1ea059 JK |
3526 | extern int generic_fadvise(struct file *file, loff_t offset, loff_t len, |
3527 | int advice); | |
45cd0faa | 3528 | |
1da177e4 | 3529 | #endif /* _LINUX_FS_H */ |