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> |
1da177e4 | 45 | |
1da177e4 | 46 | #include <asm/byteorder.h> |
607ca46e | 47 | #include <uapi/linux/fs.h> |
1da177e4 | 48 | |
b83ae6d4 | 49 | struct backing_dev_info; |
52ebea74 | 50 | struct bdi_writeback; |
2f8b5444 | 51 | struct bio; |
5a72e899 | 52 | struct io_comp_batch; |
a5694255 | 53 | struct export_operations; |
10c5db28 | 54 | struct fiemap_extent_info; |
a885c8c4 | 55 | struct hd_geometry; |
1da177e4 | 56 | struct iovec; |
92198f7e | 57 | struct kiocb; |
57cc7215 | 58 | struct kobject; |
1da177e4 LT |
59 | struct pipe_inode_info; |
60 | struct poll_table_struct; | |
61 | struct kstatfs; | |
62 | struct vm_area_struct; | |
63 | struct vfsmount; | |
745ca247 | 64 | struct cred; |
a509bc1a | 65 | struct swap_info_struct; |
55985dd7 | 66 | struct seq_file; |
7b7a8665 | 67 | struct workqueue_struct; |
3b93f911 | 68 | struct iov_iter; |
0b81d077 JK |
69 | struct fscrypt_info; |
70 | struct fscrypt_operations; | |
5585f2af EB |
71 | struct fsverity_info; |
72 | struct fsverity_operations; | |
f3a09c92 | 73 | struct fs_context; |
d7167b14 | 74 | struct fs_parameter_spec; |
4c5b4799 | 75 | struct fileattr; |
1da177e4 | 76 | |
74bf17cf | 77 | extern void __init inode_init(void); |
1da177e4 | 78 | extern void __init inode_init_early(void); |
4248b0da MG |
79 | extern void __init files_init(void); |
80 | extern void __init files_maxfiles_init(void); | |
1da177e4 | 81 | |
dded4f4d | 82 | extern struct files_stat_struct files_stat; |
518de9b3 | 83 | extern unsigned long get_max_files(void); |
9b80a184 | 84 | extern unsigned int sysctl_nr_open; |
dded4f4d JE |
85 | extern struct inodes_stat_t inodes_stat; |
86 | extern int leases_enable, lease_break_time; | |
800179c9 KC |
87 | extern int sysctl_protected_symlinks; |
88 | extern int sysctl_protected_hardlinks; | |
30aba665 SM |
89 | extern int sysctl_protected_fifos; |
90 | extern int sysctl_protected_regular; | |
dded4f4d | 91 | |
ddef7ed2 CH |
92 | typedef __kernel_rwf_t rwf_t; |
93 | ||
1da177e4 LT |
94 | struct buffer_head; |
95 | typedef int (get_block_t)(struct inode *inode, sector_t iblock, | |
96 | struct buffer_head *bh_result, int create); | |
187372a3 | 97 | typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset, |
7b7a8665 | 98 | ssize_t bytes, void *private); |
1da177e4 | 99 | |
bbc1096a DH |
100 | #define MAY_EXEC 0x00000001 |
101 | #define MAY_WRITE 0x00000002 | |
102 | #define MAY_READ 0x00000004 | |
103 | #define MAY_APPEND 0x00000008 | |
104 | #define MAY_ACCESS 0x00000010 | |
105 | #define MAY_OPEN 0x00000020 | |
106 | #define MAY_CHDIR 0x00000040 | |
107 | /* called from RCU mode, don't block */ | |
108 | #define MAY_NOT_BLOCK 0x00000080 | |
109 | ||
110 | /* | |
111 | * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond | |
75abe329 | 112 | * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open() |
bbc1096a DH |
113 | */ |
114 | ||
115 | /* file is open for reading */ | |
116 | #define FMODE_READ ((__force fmode_t)0x1) | |
117 | /* file is open for writing */ | |
118 | #define FMODE_WRITE ((__force fmode_t)0x2) | |
119 | /* file is seekable */ | |
120 | #define FMODE_LSEEK ((__force fmode_t)0x4) | |
121 | /* file can be accessed using pread */ | |
122 | #define FMODE_PREAD ((__force fmode_t)0x8) | |
123 | /* file can be accessed using pwrite */ | |
124 | #define FMODE_PWRITE ((__force fmode_t)0x10) | |
125 | /* File is opened for execution with sys_execve / sys_uselib */ | |
126 | #define FMODE_EXEC ((__force fmode_t)0x20) | |
127 | /* File is opened with O_NDELAY (only set for block devices) */ | |
128 | #define FMODE_NDELAY ((__force fmode_t)0x40) | |
129 | /* File is opened with O_EXCL (only set for block devices) */ | |
130 | #define FMODE_EXCL ((__force fmode_t)0x80) | |
131 | /* File is opened using open(.., 3, ..) and is writeable only for ioctls | |
132 | (specialy hack for floppy.c) */ | |
133 | #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100) | |
134 | /* 32bit hashes as llseek() offset (for directories) */ | |
135 | #define FMODE_32BITHASH ((__force fmode_t)0x200) | |
136 | /* 64bit hashes as llseek() offset (for directories) */ | |
137 | #define FMODE_64BITHASH ((__force fmode_t)0x400) | |
138 | ||
139 | /* | |
140 | * Don't update ctime and mtime. | |
141 | * | |
142 | * Currently a special hack for the XFS open_by_handle ioctl, but we'll | |
143 | * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon. | |
144 | */ | |
145 | #define FMODE_NOCMTIME ((__force fmode_t)0x800) | |
146 | ||
147 | /* Expect random access pattern */ | |
148 | #define FMODE_RANDOM ((__force fmode_t)0x1000) | |
149 | ||
bbcd53c9 | 150 | /* File is huge (eg. /dev/mem): treat loff_t as unsigned */ |
bbc1096a DH |
151 | #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000) |
152 | ||
153 | /* File is opened with O_PATH; almost nothing can be done with it */ | |
154 | #define FMODE_PATH ((__force fmode_t)0x4000) | |
155 | ||
2be7d348 LT |
156 | /* File needs atomic accesses to f_pos */ |
157 | #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000) | |
83f936c7 AV |
158 | /* Write access to underlying fs */ |
159 | #define FMODE_WRITER ((__force fmode_t)0x10000) | |
7f7f25e8 AV |
160 | /* Has read method(s) */ |
161 | #define FMODE_CAN_READ ((__force fmode_t)0x20000) | |
162 | /* Has write method(s) */ | |
163 | #define FMODE_CAN_WRITE ((__force fmode_t)0x40000) | |
9c225f26 | 164 | |
f5d11409 | 165 | #define FMODE_OPENED ((__force fmode_t)0x80000) |
73a09dd9 | 166 | #define FMODE_CREATED ((__force fmode_t)0x100000) |
f5d11409 | 167 | |
10dce8af KS |
168 | /* File is stream-like */ |
169 | #define FMODE_STREAM ((__force fmode_t)0x200000) | |
170 | ||
bbc1096a | 171 | /* File was opened by fanotify and shouldn't generate fanotify events */ |
75069f2b | 172 | #define FMODE_NONOTIFY ((__force fmode_t)0x4000000) |
bbc1096a | 173 | |
91f9943e | 174 | /* File is capable of returning -EAGAIN if I/O will block */ |
a07b2000 AV |
175 | #define FMODE_NOWAIT ((__force fmode_t)0x8000000) |
176 | ||
177 | /* File represents mount that needs unmounting */ | |
178 | #define FMODE_NEED_UNMOUNT ((__force fmode_t)0x10000000) | |
b745fafa | 179 | |
d3b1084d | 180 | /* File does not contribute to nr_files count */ |
a07b2000 | 181 | #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000) |
d3b1084d | 182 | |
c2a25ec0 JA |
183 | /* File supports async buffered reads */ |
184 | #define FMODE_BUF_RASYNC ((__force fmode_t)0x40000000) | |
185 | ||
1da177e4 LT |
186 | /* |
187 | * Attribute flags. These should be or-ed together to figure out what | |
188 | * has been changed! | |
189 | */ | |
9767d749 MS |
190 | #define ATTR_MODE (1 << 0) |
191 | #define ATTR_UID (1 << 1) | |
192 | #define ATTR_GID (1 << 2) | |
193 | #define ATTR_SIZE (1 << 3) | |
194 | #define ATTR_ATIME (1 << 4) | |
195 | #define ATTR_MTIME (1 << 5) | |
196 | #define ATTR_CTIME (1 << 6) | |
197 | #define ATTR_ATIME_SET (1 << 7) | |
198 | #define ATTR_MTIME_SET (1 << 8) | |
199 | #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */ | |
9767d749 MS |
200 | #define ATTR_KILL_SUID (1 << 11) |
201 | #define ATTR_KILL_SGID (1 << 12) | |
202 | #define ATTR_FILE (1 << 13) | |
203 | #define ATTR_KILL_PRIV (1 << 14) | |
204 | #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */ | |
205 | #define ATTR_TIMES_SET (1 << 16) | |
f2b20f6e | 206 | #define ATTR_TOUCH (1 << 17) |
1da177e4 | 207 | |
787fb6bc MS |
208 | /* |
209 | * Whiteout is represented by a char device. The following constants define the | |
210 | * mode and device number to use. | |
211 | */ | |
212 | #define WHITEOUT_MODE 0 | |
213 | #define WHITEOUT_DEV 0 | |
214 | ||
1da177e4 LT |
215 | /* |
216 | * This is the Inode Attributes structure, used for notify_change(). It | |
217 | * uses the above definitions as flags, to know which values have changed. | |
218 | * Also, in this manner, a Filesystem can look at only the values it cares | |
219 | * about. Basically, these are the attributes that the VFS layer can | |
220 | * request to change from the FS layer. | |
221 | * | |
222 | * Derek Atkins <warlord@MIT.EDU> 94-10-20 | |
223 | */ | |
224 | struct iattr { | |
225 | unsigned int ia_valid; | |
226 | umode_t ia_mode; | |
92361636 EB |
227 | kuid_t ia_uid; |
228 | kgid_t ia_gid; | |
1da177e4 | 229 | loff_t ia_size; |
95582b00 DD |
230 | struct timespec64 ia_atime; |
231 | struct timespec64 ia_mtime; | |
232 | struct timespec64 ia_ctime; | |
cc4e69de MS |
233 | |
234 | /* | |
25985edc | 235 | * Not an attribute, but an auxiliary info for filesystems wanting to |
cc4e69de MS |
236 | * implement an ftruncate() like method. NOTE: filesystem should |
237 | * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL). | |
238 | */ | |
239 | struct file *ia_file; | |
1da177e4 LT |
240 | }; |
241 | ||
1da177e4 LT |
242 | /* |
243 | * Includes for diskquotas. | |
244 | */ | |
245 | #include <linux/quota.h> | |
246 | ||
69c433ed MS |
247 | /* |
248 | * Maximum number of layers of fs stack. Needs to be limited to | |
249 | * prevent kernel stack overflow | |
250 | */ | |
251 | #define FILESYSTEM_MAX_STACK_DEPTH 2 | |
252 | ||
994fc28c ZB |
253 | /** |
254 | * enum positive_aop_returns - aop return codes with specific semantics | |
255 | * | |
256 | * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has | |
257 | * completed, that the page is still locked, and | |
258 | * should be considered active. The VM uses this hint | |
259 | * to return the page to the active list -- it won't | |
260 | * be a candidate for writeback again in the near | |
261 | * future. Other callers must be careful to unlock | |
262 | * the page if they get this return. Returned by | |
263 | * writepage(); | |
264 | * | |
265 | * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has | |
266 | * unlocked it and the page might have been truncated. | |
267 | * The caller should back up to acquiring a new page and | |
268 | * trying again. The aop will be taking reasonable | |
269 | * precautions not to livelock. If the caller held a page | |
270 | * reference, it should drop it before retrying. Returned | |
55144768 | 271 | * by readpage(). |
994fc28c ZB |
272 | * |
273 | * address_space_operation functions return these large constants to indicate | |
274 | * special semantics to the caller. These are much larger than the bytes in a | |
275 | * page to allow for functions that return the number of bytes operated on in a | |
276 | * given page. | |
277 | */ | |
278 | ||
279 | enum positive_aop_returns { | |
280 | AOP_WRITEPAGE_ACTIVATE = 0x80000, | |
281 | AOP_TRUNCATED_PAGE = 0x80001, | |
282 | }; | |
283 | ||
c718a975 TH |
284 | #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */ |
285 | #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct | |
54566b2c NP |
286 | * helper code (eg buffer layer) |
287 | * to clear GFP_FS from alloc */ | |
afddba49 | 288 | |
1da177e4 LT |
289 | /* |
290 | * oh the beauties of C type declarations. | |
291 | */ | |
292 | struct page; | |
293 | struct address_space; | |
294 | struct writeback_control; | |
8151b4c8 | 295 | struct readahead_control; |
1da177e4 | 296 | |
c75b1d94 JA |
297 | /* |
298 | * Write life time hint values. | |
69448867 | 299 | * Stored in struct inode as u8. |
c75b1d94 JA |
300 | */ |
301 | enum rw_hint { | |
302 | WRITE_LIFE_NOT_SET = 0, | |
303 | WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE, | |
304 | WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT, | |
305 | WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM, | |
306 | WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG, | |
307 | WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME, | |
308 | }; | |
309 | ||
ce71bfea JA |
310 | /* Match RWF_* bits to IOCB bits */ |
311 | #define IOCB_HIPRI (__force int) RWF_HIPRI | |
312 | #define IOCB_DSYNC (__force int) RWF_DSYNC | |
313 | #define IOCB_SYNC (__force int) RWF_SYNC | |
314 | #define IOCB_NOWAIT (__force int) RWF_NOWAIT | |
315 | #define IOCB_APPEND (__force int) RWF_APPEND | |
316 | ||
317 | /* non-RWF related bits - start at 16 */ | |
318 | #define IOCB_EVENTFD (1 << 16) | |
319 | #define IOCB_DIRECT (1 << 17) | |
320 | #define IOCB_WRITE (1 << 18) | |
dd3e6d50 | 321 | /* iocb->ki_waitq is valid */ |
ce71bfea JA |
322 | #define IOCB_WAITQ (1 << 19) |
323 | #define IOCB_NOIO (1 << 20) | |
6c7ef543 JA |
324 | /* can use bio alloc cache */ |
325 | #define IOCB_ALLOC_CACHE (1 << 21) | |
e2e40f2c CH |
326 | |
327 | struct kiocb { | |
328 | struct file *ki_filp; | |
84c4e1f8 LT |
329 | |
330 | /* The 'ki_filp' pointer is shared in a union for aio */ | |
331 | randomized_struct_fields_start | |
332 | ||
e2e40f2c | 333 | loff_t ki_pos; |
6b19b766 | 334 | void (*ki_complete)(struct kiocb *iocb, long ret); |
e2e40f2c CH |
335 | void *private; |
336 | int ki_flags; | |
fc28724d | 337 | u16 ki_hint; |
d9a08a9e | 338 | u16 ki_ioprio; /* See linux/ioprio.h */ |
3e08773c | 339 | struct wait_page_queue *ki_waitq; /* for async buffered IO */ |
84c4e1f8 LT |
340 | randomized_struct_fields_end |
341 | }; | |
e2e40f2c CH |
342 | |
343 | static inline bool is_sync_kiocb(struct kiocb *kiocb) | |
344 | { | |
345 | return kiocb->ki_complete == NULL; | |
346 | } | |
347 | ||
8ab22b9a HH |
348 | /* |
349 | * "descriptor" for what we're up to with a read. | |
350 | * This allows us to use the same read code yet | |
351 | * have multiple different users of the data that | |
352 | * we read from a file. | |
353 | * | |
354 | * The simplest case just copies the data to user | |
355 | * mode. | |
356 | */ | |
357 | typedef struct { | |
358 | size_t written; | |
359 | size_t count; | |
360 | union { | |
361 | char __user *buf; | |
362 | void *data; | |
363 | } arg; | |
364 | int error; | |
365 | } read_descriptor_t; | |
366 | ||
367 | typedef int (*read_actor_t)(read_descriptor_t *, struct page *, | |
368 | unsigned long, unsigned long); | |
2f718ffc | 369 | |
1da177e4 LT |
370 | struct address_space_operations { |
371 | int (*writepage)(struct page *page, struct writeback_control *wbc); | |
372 | int (*readpage)(struct file *, struct page *); | |
1da177e4 LT |
373 | |
374 | /* Write back some dirty pages from this mapping. */ | |
375 | int (*writepages)(struct address_space *, struct writeback_control *); | |
376 | ||
4741c9fd | 377 | /* Set a page dirty. Return true if this dirtied it */ |
1da177e4 LT |
378 | int (*set_page_dirty)(struct page *page); |
379 | ||
74c8164e JA |
380 | /* |
381 | * Reads in the requested pages. Unlike ->readpage(), this is | |
382 | * PURELY used for read-ahead!. | |
383 | */ | |
1da177e4 LT |
384 | int (*readpages)(struct file *filp, struct address_space *mapping, |
385 | struct list_head *pages, unsigned nr_pages); | |
8151b4c8 | 386 | void (*readahead)(struct readahead_control *); |
1da177e4 | 387 | |
afddba49 NP |
388 | int (*write_begin)(struct file *, struct address_space *mapping, |
389 | loff_t pos, unsigned len, unsigned flags, | |
390 | struct page **pagep, void **fsdata); | |
391 | int (*write_end)(struct file *, struct address_space *mapping, | |
392 | loff_t pos, unsigned len, unsigned copied, | |
393 | struct page *page, void *fsdata); | |
394 | ||
1da177e4 LT |
395 | /* Unfortunately this kludge is needed for FIBMAP. Don't use it */ |
396 | sector_t (*bmap)(struct address_space *, sector_t); | |
d47992f8 | 397 | void (*invalidatepage) (struct page *, unsigned int, unsigned int); |
27496a8c | 398 | int (*releasepage) (struct page *, gfp_t); |
6072d13c | 399 | void (*freepage)(struct page *); |
c8b8e32d | 400 | ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter); |
b969c4ab | 401 | /* |
ef277c73 TC |
402 | * migrate the contents of a page to the specified target. If |
403 | * migrate_mode is MIGRATE_ASYNC, it must not block. | |
b969c4ab | 404 | */ |
2d1db3b1 | 405 | int (*migratepage) (struct address_space *, |
a6bc32b8 | 406 | struct page *, struct page *, enum migrate_mode); |
bda807d4 MK |
407 | bool (*isolate_page)(struct page *, isolate_mode_t); |
408 | void (*putback_page)(struct page *); | |
e3db7691 | 409 | int (*launder_page) (struct page *); |
c186afb4 | 410 | int (*is_partially_uptodate) (struct page *, unsigned long, |
8ab22b9a | 411 | unsigned long); |
b4597226 | 412 | void (*is_dirty_writeback) (struct page *, bool *, bool *); |
25718736 | 413 | int (*error_remove_page)(struct address_space *, struct page *); |
62c230bc MG |
414 | |
415 | /* swapfile support */ | |
a509bc1a MG |
416 | int (*swap_activate)(struct swap_info_struct *sis, struct file *file, |
417 | sector_t *span); | |
418 | void (*swap_deactivate)(struct file *file); | |
1da177e4 LT |
419 | }; |
420 | ||
7dcda1c9 JA |
421 | extern const struct address_space_operations empty_aops; |
422 | ||
afddba49 NP |
423 | /* |
424 | * pagecache_write_begin/pagecache_write_end must be used by general code | |
425 | * to write into the pagecache. | |
426 | */ | |
427 | int pagecache_write_begin(struct file *, struct address_space *mapping, | |
428 | loff_t pos, unsigned len, unsigned flags, | |
429 | struct page **pagep, void **fsdata); | |
430 | ||
431 | int pagecache_write_end(struct file *, struct address_space *mapping, | |
432 | loff_t pos, unsigned len, unsigned copied, | |
433 | struct page *page, void *fsdata); | |
434 | ||
eb797a8e MW |
435 | /** |
436 | * struct address_space - Contents of a cacheable, mappable object. | |
437 | * @host: Owner, either the inode or the block_device. | |
438 | * @i_pages: Cached pages. | |
730633f0 JK |
439 | * @invalidate_lock: Guards coherency between page cache contents and |
440 | * file offset->disk block mappings in the filesystem during invalidates. | |
441 | * It is also used to block modification of page cache contents through | |
442 | * memory mappings. | |
eb797a8e MW |
443 | * @gfp_mask: Memory allocation flags to use for allocating pages. |
444 | * @i_mmap_writable: Number of VM_SHARED mappings. | |
09d91cda | 445 | * @nr_thps: Number of THPs in the pagecache (non-shmem only). |
eb797a8e MW |
446 | * @i_mmap: Tree of private and shared mappings. |
447 | * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable. | |
448 | * @nrpages: Number of page entries, protected by the i_pages lock. | |
eb797a8e MW |
449 | * @writeback_index: Writeback starts here. |
450 | * @a_ops: Methods. | |
451 | * @flags: Error bits and flags (AS_*). | |
452 | * @wb_err: The most recent error which has occurred. | |
453 | * @private_lock: For use by the owner of the address_space. | |
454 | * @private_list: For use by the owner of the address_space. | |
455 | * @private_data: For use by the owner of the address_space. | |
456 | */ | |
1da177e4 | 457 | struct address_space { |
eb797a8e MW |
458 | struct inode *host; |
459 | struct xarray i_pages; | |
730633f0 | 460 | struct rw_semaphore invalidate_lock; |
eb797a8e MW |
461 | gfp_t gfp_mask; |
462 | atomic_t i_mmap_writable; | |
09d91cda SL |
463 | #ifdef CONFIG_READ_ONLY_THP_FOR_FS |
464 | /* number of thp, only for non-shmem files */ | |
465 | atomic_t nr_thps; | |
466 | #endif | |
eb797a8e MW |
467 | struct rb_root_cached i_mmap; |
468 | struct rw_semaphore i_mmap_rwsem; | |
469 | unsigned long nrpages; | |
eb797a8e MW |
470 | pgoff_t writeback_index; |
471 | const struct address_space_operations *a_ops; | |
472 | unsigned long flags; | |
5660e13d | 473 | errseq_t wb_err; |
eb797a8e MW |
474 | spinlock_t private_lock; |
475 | struct list_head private_list; | |
476 | void *private_data; | |
3859a271 | 477 | } __attribute__((aligned(sizeof(long)))) __randomize_layout; |
1da177e4 LT |
478 | /* |
479 | * On most architectures that alignment is already the case; but | |
25985edc | 480 | * must be enforced here for CRIS, to let the least significant bit |
1da177e4 LT |
481 | * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON. |
482 | */ | |
1da177e4 | 483 | |
ff9c745b MW |
484 | /* XArray tags, for tagging dirty and writeback pages in the pagecache. */ |
485 | #define PAGECACHE_TAG_DIRTY XA_MARK_0 | |
486 | #define PAGECACHE_TAG_WRITEBACK XA_MARK_1 | |
487 | #define PAGECACHE_TAG_TOWRITE XA_MARK_2 | |
488 | ||
1da177e4 | 489 | /* |
ff9c745b | 490 | * Returns true if any of the pages in the mapping are marked with the tag. |
1da177e4 | 491 | */ |
ff9c745b MW |
492 | static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag) |
493 | { | |
494 | return xa_marked(&mapping->i_pages, tag); | |
495 | } | |
1da177e4 | 496 | |
8b28f621 DB |
497 | static inline void i_mmap_lock_write(struct address_space *mapping) |
498 | { | |
c8c06efa | 499 | down_write(&mapping->i_mmap_rwsem); |
8b28f621 DB |
500 | } |
501 | ||
c0d0381a MK |
502 | static inline int i_mmap_trylock_write(struct address_space *mapping) |
503 | { | |
504 | return down_write_trylock(&mapping->i_mmap_rwsem); | |
505 | } | |
506 | ||
8b28f621 DB |
507 | static inline void i_mmap_unlock_write(struct address_space *mapping) |
508 | { | |
c8c06efa | 509 | up_write(&mapping->i_mmap_rwsem); |
8b28f621 DB |
510 | } |
511 | ||
3dec0ba0 DB |
512 | static inline void i_mmap_lock_read(struct address_space *mapping) |
513 | { | |
514 | down_read(&mapping->i_mmap_rwsem); | |
515 | } | |
516 | ||
517 | static inline void i_mmap_unlock_read(struct address_space *mapping) | |
518 | { | |
519 | up_read(&mapping->i_mmap_rwsem); | |
520 | } | |
521 | ||
34ae204f MK |
522 | static inline void i_mmap_assert_locked(struct address_space *mapping) |
523 | { | |
524 | lockdep_assert_held(&mapping->i_mmap_rwsem); | |
525 | } | |
526 | ||
527 | static inline void i_mmap_assert_write_locked(struct address_space *mapping) | |
528 | { | |
529 | lockdep_assert_held_write(&mapping->i_mmap_rwsem); | |
530 | } | |
531 | ||
1da177e4 LT |
532 | /* |
533 | * Might pages of this file be mapped into userspace? | |
534 | */ | |
535 | static inline int mapping_mapped(struct address_space *mapping) | |
536 | { | |
f808c13f | 537 | return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root); |
1da177e4 LT |
538 | } |
539 | ||
540 | /* | |
541 | * Might pages of this file have been modified in userspace? | |
45e55300 | 542 | * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap |
1da177e4 LT |
543 | * marks vma as VM_SHARED if it is shared, and the file was opened for |
544 | * writing i.e. vma may be mprotected writable even if now readonly. | |
4bb5f5d9 DH |
545 | * |
546 | * If i_mmap_writable is negative, no new writable mappings are allowed. You | |
547 | * can only deny writable mappings, if none exists right now. | |
1da177e4 LT |
548 | */ |
549 | static inline int mapping_writably_mapped(struct address_space *mapping) | |
550 | { | |
4bb5f5d9 DH |
551 | return atomic_read(&mapping->i_mmap_writable) > 0; |
552 | } | |
553 | ||
554 | static inline int mapping_map_writable(struct address_space *mapping) | |
555 | { | |
556 | return atomic_inc_unless_negative(&mapping->i_mmap_writable) ? | |
557 | 0 : -EPERM; | |
558 | } | |
559 | ||
560 | static inline void mapping_unmap_writable(struct address_space *mapping) | |
561 | { | |
562 | atomic_dec(&mapping->i_mmap_writable); | |
563 | } | |
564 | ||
565 | static inline int mapping_deny_writable(struct address_space *mapping) | |
566 | { | |
567 | return atomic_dec_unless_positive(&mapping->i_mmap_writable) ? | |
568 | 0 : -EBUSY; | |
569 | } | |
570 | ||
571 | static inline void mapping_allow_writable(struct address_space *mapping) | |
572 | { | |
573 | atomic_inc(&mapping->i_mmap_writable); | |
1da177e4 LT |
574 | } |
575 | ||
576 | /* | |
577 | * Use sequence counter to get consistent i_size on 32-bit processors. | |
578 | */ | |
579 | #if BITS_PER_LONG==32 && defined(CONFIG_SMP) | |
580 | #include <linux/seqlock.h> | |
581 | #define __NEED_I_SIZE_ORDERED | |
582 | #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount) | |
583 | #else | |
584 | #define i_size_ordered_init(inode) do { } while (0) | |
585 | #endif | |
586 | ||
f19d4a8f AV |
587 | struct posix_acl; |
588 | #define ACL_NOT_CACHED ((void *)(-1)) | |
332f606b MS |
589 | /* |
590 | * ACL_DONT_CACHE is for stacked filesystems, that rely on underlying fs to | |
591 | * cache the ACL. This also means that ->get_acl() can be called in RCU mode | |
592 | * with the LOOKUP_RCU flag. | |
593 | */ | |
2a3a2a3f | 594 | #define ACL_DONT_CACHE ((void *)(-3)) |
f19d4a8f | 595 | |
b8a7a3a6 AG |
596 | static inline struct posix_acl * |
597 | uncached_acl_sentinel(struct task_struct *task) | |
598 | { | |
599 | return (void *)task + 1; | |
600 | } | |
601 | ||
602 | static inline bool | |
603 | is_uncached_acl(struct posix_acl *acl) | |
604 | { | |
605 | return (long)acl & 1; | |
606 | } | |
607 | ||
3ddcd056 LT |
608 | #define IOP_FASTPERM 0x0001 |
609 | #define IOP_LOOKUP 0x0002 | |
610 | #define IOP_NOFOLLOW 0x0004 | |
d0a5b995 | 611 | #define IOP_XATTR 0x0008 |
76fca90e | 612 | #define IOP_DEFAULT_READLINK 0x0010 |
3ddcd056 | 613 | |
9dd813c1 JK |
614 | struct fsnotify_mark_connector; |
615 | ||
3ddcd056 LT |
616 | /* |
617 | * Keep mostly read-only and often accessed (especially for | |
618 | * the RCU path lookup and 'stat' data) fields at the beginning | |
619 | * of the 'struct inode' | |
620 | */ | |
1da177e4 | 621 | struct inode { |
44a7d7a8 | 622 | umode_t i_mode; |
3ddcd056 | 623 | unsigned short i_opflags; |
92361636 EB |
624 | kuid_t i_uid; |
625 | kgid_t i_gid; | |
3ddcd056 LT |
626 | unsigned int i_flags; |
627 | ||
628 | #ifdef CONFIG_FS_POSIX_ACL | |
629 | struct posix_acl *i_acl; | |
630 | struct posix_acl *i_default_acl; | |
631 | #endif | |
632 | ||
44a7d7a8 NP |
633 | const struct inode_operations *i_op; |
634 | struct super_block *i_sb; | |
3ddcd056 | 635 | struct address_space *i_mapping; |
44a7d7a8 | 636 | |
13e12d14 LT |
637 | #ifdef CONFIG_SECURITY |
638 | void *i_security; | |
639 | #endif | |
44a7d7a8 | 640 | |
3ddcd056 LT |
641 | /* Stat data, not accessed from path walking */ |
642 | unsigned long i_ino; | |
a78ef704 MS |
643 | /* |
644 | * Filesystems may only read i_nlink directly. They shall use the | |
645 | * following functions for modification: | |
646 | * | |
647 | * (set|clear|inc|drop)_nlink | |
648 | * inode_(inc|dec)_link_count | |
649 | */ | |
650 | union { | |
651 | const unsigned int i_nlink; | |
652 | unsigned int __i_nlink; | |
653 | }; | |
3ddcd056 | 654 | dev_t i_rdev; |
2f9d3df8 | 655 | loff_t i_size; |
95582b00 DD |
656 | struct timespec64 i_atime; |
657 | struct timespec64 i_mtime; | |
658 | struct timespec64 i_ctime; | |
6cdbb0ef TT |
659 | spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */ |
660 | unsigned short i_bytes; | |
69448867 AG |
661 | u8 i_blkbits; |
662 | u8 i_write_hint; | |
3ddcd056 LT |
663 | blkcnt_t i_blocks; |
664 | ||
665 | #ifdef __NEED_I_SIZE_ORDERED | |
666 | seqcount_t i_size_seqcount; | |
667 | #endif | |
668 | ||
669 | /* Misc */ | |
670 | unsigned long i_state; | |
9902af79 | 671 | struct rw_semaphore i_rwsem; |
13e12d14 | 672 | |
44a7d7a8 | 673 | unsigned long dirtied_when; /* jiffies of first dirtying */ |
a2f48706 | 674 | unsigned long dirtied_time_when; |
44a7d7a8 | 675 | |
1da177e4 | 676 | struct hlist_node i_hash; |
c7f54084 | 677 | struct list_head i_io_list; /* backing dev IO list */ |
52ebea74 TH |
678 | #ifdef CONFIG_CGROUP_WRITEBACK |
679 | struct bdi_writeback *i_wb; /* the associated cgroup wb */ | |
2a814908 TH |
680 | |
681 | /* foreign inode detection, see wbc_detach_inode() */ | |
682 | int i_wb_frn_winner; | |
683 | u16 i_wb_frn_avg_time; | |
684 | u16 i_wb_frn_history; | |
52ebea74 | 685 | #endif |
7ccf19a8 | 686 | struct list_head i_lru; /* inode LRU list */ |
1da177e4 | 687 | struct list_head i_sb_list; |
6c60d2b5 | 688 | struct list_head i_wb_list; /* backing dev writeback list */ |
fa0d7e3d | 689 | union { |
b3d9b7a3 | 690 | struct hlist_head i_dentry; |
fa0d7e3d NP |
691 | struct rcu_head i_rcu; |
692 | }; | |
f02a9ad1 | 693 | atomic64_t i_version; |
8019ad13 | 694 | atomic64_t i_sequence; /* see futex */ |
2f9d3df8 | 695 | atomic_t i_count; |
bd5fe6c5 | 696 | atomic_t i_dio_count; |
6cdbb0ef | 697 | atomic_t i_writecount; |
387e3746 | 698 | #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING) |
d984ea60 MZ |
699 | atomic_t i_readcount; /* struct files open RO */ |
700 | #endif | |
fdb0da89 AV |
701 | union { |
702 | const struct file_operations *i_fop; /* former ->i_op->default_file_ops */ | |
703 | void (*free_inode)(struct inode *); | |
704 | }; | |
4a075e39 | 705 | struct file_lock_context *i_flctx; |
1da177e4 | 706 | struct address_space i_data; |
1da177e4 | 707 | struct list_head i_devices; |
4c154168 TT |
708 | union { |
709 | struct pipe_inode_info *i_pipe; | |
577c4eb0 | 710 | struct cdev *i_cdev; |
61ba64fc | 711 | char *i_link; |
84e710da | 712 | unsigned i_dir_seq; |
4c154168 | 713 | }; |
1da177e4 LT |
714 | |
715 | __u32 i_generation; | |
716 | ||
3be25f49 EP |
717 | #ifdef CONFIG_FSNOTIFY |
718 | __u32 i_fsnotify_mask; /* all events this inode cares about */ | |
08991e83 | 719 | struct fsnotify_mark_connector __rcu *i_fsnotify_marks; |
0eeca283 RL |
720 | #endif |
721 | ||
643fa961 | 722 | #ifdef CONFIG_FS_ENCRYPTION |
0b81d077 JK |
723 | struct fscrypt_info *i_crypt_info; |
724 | #endif | |
725 | ||
5585f2af EB |
726 | #ifdef CONFIG_FS_VERITY |
727 | struct fsverity_info *i_verity_info; | |
728 | #endif | |
729 | ||
8e18e294 | 730 | void *i_private; /* fs or device private pointer */ |
3859a271 | 731 | } __randomize_layout; |
1da177e4 | 732 | |
50e17c00 DD |
733 | struct timespec64 timestamp_truncate(struct timespec64 t, struct inode *inode); |
734 | ||
93407472 FF |
735 | static inline unsigned int i_blocksize(const struct inode *node) |
736 | { | |
737 | return (1 << node->i_blkbits); | |
738 | } | |
739 | ||
1d3382cb AV |
740 | static inline int inode_unhashed(struct inode *inode) |
741 | { | |
742 | return hlist_unhashed(&inode->i_hash); | |
743 | } | |
744 | ||
5bef9151 AV |
745 | /* |
746 | * __mark_inode_dirty expects inodes to be hashed. Since we don't | |
747 | * want special inodes in the fileset inode space, we make them | |
748 | * appear hashed, but do not put on any lists. hlist_del() | |
749 | * will work fine and require no locking. | |
750 | */ | |
751 | static inline void inode_fake_hash(struct inode *inode) | |
752 | { | |
753 | hlist_add_fake(&inode->i_hash); | |
754 | } | |
755 | ||
f2eace23 IM |
756 | /* |
757 | * inode->i_mutex nesting subclasses for the lock validator: | |
758 | * | |
759 | * 0: the object of the current VFS operation | |
760 | * 1: parent | |
761 | * 2: child/target | |
40bd22c9 BF |
762 | * 3: xattr |
763 | * 4: second non-directory | |
d1b72cc6 MS |
764 | * 5: second parent (when locking independent directories in rename) |
765 | * | |
766 | * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two | |
40bd22c9 | 767 | * non-directories at once. |
f2eace23 IM |
768 | * |
769 | * The locking order between these classes is | |
d1b72cc6 | 770 | * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory |
f2eace23 IM |
771 | */ |
772 | enum inode_i_mutex_lock_class | |
773 | { | |
774 | I_MUTEX_NORMAL, | |
775 | I_MUTEX_PARENT, | |
776 | I_MUTEX_CHILD, | |
4df46240 | 777 | I_MUTEX_XATTR, |
d1b72cc6 MS |
778 | I_MUTEX_NONDIR2, |
779 | I_MUTEX_PARENT2, | |
f2eace23 IM |
780 | }; |
781 | ||
5955102c AV |
782 | static inline void inode_lock(struct inode *inode) |
783 | { | |
9902af79 | 784 | down_write(&inode->i_rwsem); |
5955102c AV |
785 | } |
786 | ||
787 | static inline void inode_unlock(struct inode *inode) | |
788 | { | |
9902af79 AV |
789 | up_write(&inode->i_rwsem); |
790 | } | |
791 | ||
792 | static inline void inode_lock_shared(struct inode *inode) | |
793 | { | |
794 | down_read(&inode->i_rwsem); | |
795 | } | |
796 | ||
797 | static inline void inode_unlock_shared(struct inode *inode) | |
798 | { | |
799 | up_read(&inode->i_rwsem); | |
5955102c AV |
800 | } |
801 | ||
802 | static inline int inode_trylock(struct inode *inode) | |
803 | { | |
9902af79 AV |
804 | return down_write_trylock(&inode->i_rwsem); |
805 | } | |
806 | ||
807 | static inline int inode_trylock_shared(struct inode *inode) | |
808 | { | |
809 | return down_read_trylock(&inode->i_rwsem); | |
5955102c AV |
810 | } |
811 | ||
812 | static inline int inode_is_locked(struct inode *inode) | |
813 | { | |
9902af79 | 814 | return rwsem_is_locked(&inode->i_rwsem); |
5955102c AV |
815 | } |
816 | ||
817 | static inline void inode_lock_nested(struct inode *inode, unsigned subclass) | |
818 | { | |
9902af79 | 819 | down_write_nested(&inode->i_rwsem, subclass); |
5955102c AV |
820 | } |
821 | ||
01c2e13d DW |
822 | static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass) |
823 | { | |
824 | down_read_nested(&inode->i_rwsem, subclass); | |
825 | } | |
826 | ||
730633f0 JK |
827 | static inline void filemap_invalidate_lock(struct address_space *mapping) |
828 | { | |
829 | down_write(&mapping->invalidate_lock); | |
830 | } | |
831 | ||
832 | static inline void filemap_invalidate_unlock(struct address_space *mapping) | |
833 | { | |
834 | up_write(&mapping->invalidate_lock); | |
835 | } | |
836 | ||
837 | static inline void filemap_invalidate_lock_shared(struct address_space *mapping) | |
838 | { | |
839 | down_read(&mapping->invalidate_lock); | |
840 | } | |
841 | ||
842 | static inline int filemap_invalidate_trylock_shared( | |
843 | struct address_space *mapping) | |
844 | { | |
845 | return down_read_trylock(&mapping->invalidate_lock); | |
846 | } | |
847 | ||
848 | static inline void filemap_invalidate_unlock_shared( | |
849 | struct address_space *mapping) | |
850 | { | |
851 | up_read(&mapping->invalidate_lock); | |
852 | } | |
853 | ||
375e289e BF |
854 | void lock_two_nondirectories(struct inode *, struct inode*); |
855 | void unlock_two_nondirectories(struct inode *, struct inode*); | |
856 | ||
7506ae6a JK |
857 | void filemap_invalidate_lock_two(struct address_space *mapping1, |
858 | struct address_space *mapping2); | |
859 | void filemap_invalidate_unlock_two(struct address_space *mapping1, | |
860 | struct address_space *mapping2); | |
861 | ||
862 | ||
1da177e4 LT |
863 | /* |
864 | * NOTE: in a 32bit arch with a preemptable kernel and | |
865 | * an UP compile the i_size_read/write must be atomic | |
866 | * with respect to the local cpu (unlike with preempt disabled), | |
867 | * but they don't need to be atomic with respect to other cpus like in | |
868 | * true SMP (so they need either to either locally disable irq around | |
869 | * the read or for example on x86 they can be still implemented as a | |
870 | * cmpxchg8b without the need of the lock prefix). For SMP compiles | |
871 | * and 64bit archs it makes no difference if preempt is enabled or not. | |
872 | */ | |
48ed214d | 873 | static inline loff_t i_size_read(const struct inode *inode) |
1da177e4 LT |
874 | { |
875 | #if BITS_PER_LONG==32 && defined(CONFIG_SMP) | |
876 | loff_t i_size; | |
877 | unsigned int seq; | |
878 | ||
879 | do { | |
880 | seq = read_seqcount_begin(&inode->i_size_seqcount); | |
881 | i_size = inode->i_size; | |
882 | } while (read_seqcount_retry(&inode->i_size_seqcount, seq)); | |
883 | return i_size; | |
2496396f | 884 | #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION) |
1da177e4 LT |
885 | loff_t i_size; |
886 | ||
887 | preempt_disable(); | |
888 | i_size = inode->i_size; | |
889 | preempt_enable(); | |
890 | return i_size; | |
891 | #else | |
892 | return inode->i_size; | |
893 | #endif | |
894 | } | |
895 | ||
7762f5a0 MS |
896 | /* |
897 | * NOTE: unlike i_size_read(), i_size_write() does need locking around it | |
898 | * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount | |
899 | * can be lost, resulting in subsequent i_size_read() calls spinning forever. | |
900 | */ | |
1da177e4 LT |
901 | static inline void i_size_write(struct inode *inode, loff_t i_size) |
902 | { | |
903 | #if BITS_PER_LONG==32 && defined(CONFIG_SMP) | |
74e3d1e1 | 904 | preempt_disable(); |
1da177e4 LT |
905 | write_seqcount_begin(&inode->i_size_seqcount); |
906 | inode->i_size = i_size; | |
907 | write_seqcount_end(&inode->i_size_seqcount); | |
74e3d1e1 | 908 | preempt_enable(); |
2496396f | 909 | #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION) |
1da177e4 LT |
910 | preempt_disable(); |
911 | inode->i_size = i_size; | |
912 | preempt_enable(); | |
913 | #else | |
914 | inode->i_size = i_size; | |
915 | #endif | |
916 | } | |
917 | ||
48ed214d | 918 | static inline unsigned iminor(const struct inode *inode) |
1da177e4 LT |
919 | { |
920 | return MINOR(inode->i_rdev); | |
921 | } | |
922 | ||
48ed214d | 923 | static inline unsigned imajor(const struct inode *inode) |
1da177e4 LT |
924 | { |
925 | return MAJOR(inode->i_rdev); | |
926 | } | |
927 | ||
1da177e4 LT |
928 | struct fown_struct { |
929 | rwlock_t lock; /* protects pid, uid, euid fields */ | |
609d7fa9 EB |
930 | struct pid *pid; /* pid or -pgrp where SIGIO should be sent */ |
931 | enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */ | |
92361636 | 932 | kuid_t uid, euid; /* uid/euid of process setting the owner */ |
1da177e4 LT |
933 | int signum; /* posix.1b rt signal to be delivered on IO */ |
934 | }; | |
935 | ||
c790fbf2 MWO |
936 | /** |
937 | * struct file_ra_state - Track a file's readahead state. | |
938 | * @start: Where the most recent readahead started. | |
939 | * @size: Number of pages read in the most recent readahead. | |
940 | * @async_size: Start next readahead when this many pages are left. | |
941 | * @ra_pages: Maximum size of a readahead request. | |
942 | * @mmap_miss: How many mmap accesses missed in the page cache. | |
943 | * @prev_pos: The last byte in the most recent read request. | |
1da177e4 LT |
944 | */ |
945 | struct file_ra_state { | |
c790fbf2 MWO |
946 | pgoff_t start; |
947 | unsigned int size; | |
948 | unsigned int async_size; | |
949 | unsigned int ra_pages; | |
950 | unsigned int mmap_miss; | |
951 | loff_t prev_pos; | |
1da177e4 | 952 | }; |
1da177e4 | 953 | |
5ce1110b FW |
954 | /* |
955 | * Check if @index falls in the readahead windows. | |
956 | */ | |
957 | static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index) | |
958 | { | |
f9acc8c7 FW |
959 | return (index >= ra->start && |
960 | index < ra->start + ra->size); | |
5ce1110b FW |
961 | } |
962 | ||
1da177e4 | 963 | struct file { |
2f512016 | 964 | union { |
4f5e65a1 | 965 | struct llist_node fu_llist; |
2f512016 ED |
966 | struct rcu_head fu_rcuhead; |
967 | } f_u; | |
0f7fc9e4 | 968 | struct path f_path; |
dd37978c | 969 | struct inode *f_inode; /* cached value */ |
99ac48f5 | 970 | const struct file_operations *f_op; |
ef3d0fd2 AK |
971 | |
972 | /* | |
319c1517 | 973 | * Protects f_ep, f_flags. |
ef3d0fd2 AK |
974 | * Must not be taken from IRQ context. |
975 | */ | |
976 | spinlock_t f_lock; | |
c75b1d94 | 977 | enum rw_hint f_write_hint; |
516e0cc5 | 978 | atomic_long_t f_count; |
1da177e4 | 979 | unsigned int f_flags; |
aeb5d727 | 980 | fmode_t f_mode; |
9c225f26 | 981 | struct mutex f_pos_lock; |
1da177e4 LT |
982 | loff_t f_pos; |
983 | struct fown_struct f_owner; | |
d76b0d9b | 984 | const struct cred *f_cred; |
1da177e4 LT |
985 | struct file_ra_state f_ra; |
986 | ||
2b47c361 | 987 | u64 f_version; |
50462062 | 988 | #ifdef CONFIG_SECURITY |
1da177e4 | 989 | void *f_security; |
50462062 | 990 | #endif |
1da177e4 LT |
991 | /* needed for tty driver, and maybe others */ |
992 | void *private_data; | |
993 | ||
994 | #ifdef CONFIG_EPOLL | |
995 | /* Used by fs/eventpoll.c to link all the hooks to this file */ | |
319c1517 | 996 | struct hlist_head *f_ep; |
1da177e4 LT |
997 | #endif /* #ifdef CONFIG_EPOLL */ |
998 | struct address_space *f_mapping; | |
5660e13d | 999 | errseq_t f_wb_err; |
735e4ae5 | 1000 | errseq_t f_sb_err; /* for syncfs */ |
3859a271 KC |
1001 | } __randomize_layout |
1002 | __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */ | |
1da177e4 | 1003 | |
990d6c2d AK |
1004 | struct file_handle { |
1005 | __u32 handle_bytes; | |
1006 | int handle_type; | |
1007 | /* file identifier */ | |
9d82973e | 1008 | unsigned char f_handle[]; |
990d6c2d AK |
1009 | }; |
1010 | ||
cb0942b8 AV |
1011 | static inline struct file *get_file(struct file *f) |
1012 | { | |
1013 | atomic_long_inc(&f->f_count); | |
1014 | return f; | |
1015 | } | |
091141a4 JA |
1016 | #define get_file_rcu_many(x, cnt) \ |
1017 | atomic_long_add_unless(&(x)->f_count, (cnt), 0) | |
1018 | #define get_file_rcu(x) get_file_rcu_many((x), 1) | |
516e0cc5 | 1019 | #define file_count(x) atomic_long_read(&(x)->f_count) |
1da177e4 LT |
1020 | |
1021 | #define MAX_NON_LFS ((1UL<<31) - 1) | |
1022 | ||
1023 | /* Page cache limit. The filesystems should put that into their s_maxbytes | |
1024 | limits, otherwise bad things can happen in VM. */ | |
1025 | #if BITS_PER_LONG==32 | |
0cc3b0ec | 1026 | #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT) |
1da177e4 | 1027 | #elif BITS_PER_LONG==64 |
0cc3b0ec | 1028 | #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX) |
1da177e4 LT |
1029 | #endif |
1030 | ||
1031 | #define FL_POSIX 1 | |
1032 | #define FL_FLOCK 2 | |
617588d5 | 1033 | #define FL_DELEG 4 /* NFSv4 delegation */ |
1da177e4 | 1034 | #define FL_ACCESS 8 /* not trying to lock, just looking */ |
f475ae95 | 1035 | #define FL_EXISTS 16 /* when unlocking, test for existence */ |
1da177e4 | 1036 | #define FL_LEASE 32 /* lease held on this file */ |
75e1fcc0 | 1037 | #define FL_CLOSE 64 /* unlock on close */ |
1da177e4 | 1038 | #define FL_SLEEP 128 /* A blocking lock */ |
778fc546 BF |
1039 | #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */ |
1040 | #define FL_UNLOCK_PENDING 512 /* Lease is being broken */ | |
cff2fce5 | 1041 | #define FL_OFDLCK 1024 /* lock is "owned" by struct file */ |
11afe9f7 | 1042 | #define FL_LAYOUT 2048 /* outstanding pNFS layout */ |
bb0a55bb | 1043 | #define FL_RECLAIM 4096 /* reclaiming from a reboot server */ |
1da177e4 | 1044 | |
50f2112c BC |
1045 | #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE) |
1046 | ||
bde74e4b MS |
1047 | /* |
1048 | * Special return value from posix_lock_file() and vfs_lock_file() for | |
1049 | * asynchronous locking. | |
1050 | */ | |
1051 | #define FILE_LOCK_DEFERRED 1 | |
1052 | ||
7ca76311 | 1053 | /* legacy typedef, should eventually be removed */ |
17fa388d | 1054 | typedef void *fl_owner_t; |
1da177e4 | 1055 | |
a7231a97 JL |
1056 | struct file_lock; |
1057 | ||
1da177e4 | 1058 | struct file_lock_operations { |
1da177e4 LT |
1059 | void (*fl_copy_lock)(struct file_lock *, struct file_lock *); |
1060 | void (*fl_release_private)(struct file_lock *); | |
1061 | }; | |
1062 | ||
1063 | struct lock_manager_operations { | |
cae80b30 JL |
1064 | fl_owner_t (*lm_get_owner)(fl_owner_t); |
1065 | void (*lm_put_owner)(fl_owner_t); | |
8fb47a4f | 1066 | void (*lm_notify)(struct file_lock *); /* unblock callback */ |
d0449b90 | 1067 | int (*lm_grant)(struct file_lock *, int); |
4d01b7f5 | 1068 | bool (*lm_break)(struct file_lock *); |
7448cc37 | 1069 | int (*lm_change)(struct file_lock *, int, struct list_head *); |
1c7dd2ff | 1070 | void (*lm_setup)(struct file_lock *, void **); |
28df3d15 | 1071 | bool (*lm_breaker_owns_lease)(struct file_lock *); |
1da177e4 LT |
1072 | }; |
1073 | ||
af558e33 BF |
1074 | struct lock_manager { |
1075 | struct list_head list; | |
c87fb4a3 BF |
1076 | /* |
1077 | * NFSv4 and up also want opens blocked during the grace period; | |
1078 | * NLM doesn't care: | |
1079 | */ | |
1080 | bool block_opens; | |
af558e33 BF |
1081 | }; |
1082 | ||
5ccb0066 SK |
1083 | struct net; |
1084 | void locks_start_grace(struct net *, struct lock_manager *); | |
af558e33 | 1085 | void locks_end_grace(struct lock_manager *); |
003278e4 CL |
1086 | bool locks_in_grace(struct net *); |
1087 | bool opens_in_grace(struct net *); | |
af558e33 | 1088 | |
1da177e4 LT |
1089 | /* that will die - we need it for nfs_lock_info */ |
1090 | #include <linux/nfs_fs_i.h> | |
1091 | ||
1cb36012 JL |
1092 | /* |
1093 | * struct file_lock represents a generic "file lock". It's used to represent | |
1094 | * POSIX byte range locks, BSD (flock) locks, and leases. It's important to | |
1095 | * note that the same struct is used to represent both a request for a lock and | |
1096 | * the lock itself, but the same object is never used for both. | |
1097 | * | |
1098 | * FIXME: should we create a separate "struct lock_request" to help distinguish | |
1099 | * these two uses? | |
1100 | * | |
8116bf4c | 1101 | * The varous i_flctx lists are ordered by: |
1cb36012 | 1102 | * |
8116bf4c JL |
1103 | * 1) lock owner |
1104 | * 2) lock range start | |
1105 | * 3) lock range end | |
1cb36012 JL |
1106 | * |
1107 | * Obviously, the last two criteria only matter for POSIX locks. | |
1108 | */ | |
1da177e4 | 1109 | struct file_lock { |
ada5c1da | 1110 | struct file_lock *fl_blocker; /* The lock, that is blocking us */ |
6dee60f6 | 1111 | struct list_head fl_list; /* link into file_lock_context */ |
139ca04e | 1112 | struct hlist_node fl_link; /* node in global lists */ |
ada5c1da N |
1113 | struct list_head fl_blocked_requests; /* list of requests with |
1114 | * ->fl_blocker pointing here | |
1115 | */ | |
1116 | struct list_head fl_blocked_member; /* node in | |
1117 | * ->fl_blocker->fl_blocked_requests | |
1118 | */ | |
1da177e4 | 1119 | fl_owner_t fl_owner; |
710b7216 | 1120 | unsigned int fl_flags; |
afc1246f | 1121 | unsigned char fl_type; |
1da177e4 | 1122 | unsigned int fl_pid; |
7012b02a | 1123 | int fl_link_cpu; /* what cpu's list is this on? */ |
1da177e4 LT |
1124 | wait_queue_head_t fl_wait; |
1125 | struct file *fl_file; | |
1da177e4 LT |
1126 | loff_t fl_start; |
1127 | loff_t fl_end; | |
1128 | ||
1129 | struct fasync_struct * fl_fasync; /* for lease break notifications */ | |
778fc546 BF |
1130 | /* for lease breaks: */ |
1131 | unsigned long fl_break_time; | |
1132 | unsigned long fl_downgrade_time; | |
1da177e4 | 1133 | |
6aed6285 | 1134 | const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */ |
7b021967 | 1135 | const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */ |
1da177e4 LT |
1136 | union { |
1137 | struct nfs_lock_info nfs_fl; | |
8d0a8a9d | 1138 | struct nfs4_lock_info nfs4_fl; |
e8d6c554 DH |
1139 | struct { |
1140 | struct list_head link; /* link in AFS vnode's pending_locks list */ | |
1141 | int state; /* state of grant or error if -ve */ | |
d4696601 | 1142 | unsigned int debug_id; |
e8d6c554 | 1143 | } afs; |
1da177e4 | 1144 | } fl_u; |
3859a271 | 1145 | } __randomize_layout; |
1da177e4 | 1146 | |
4a075e39 | 1147 | struct file_lock_context { |
6109c850 | 1148 | spinlock_t flc_lock; |
4a075e39 | 1149 | struct list_head flc_flock; |
bd61e0a9 | 1150 | struct list_head flc_posix; |
8634b51f | 1151 | struct list_head flc_lease; |
4a075e39 JL |
1152 | }; |
1153 | ||
1da177e4 LT |
1154 | /* The following constant reflects the upper bound of the file/locking space */ |
1155 | #ifndef OFFSET_MAX | |
1156 | #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1))) | |
1157 | #define OFFSET_MAX INT_LIMIT(loff_t) | |
1158 | #define OFFT_OFFSET_MAX INT_LIMIT(off_t) | |
1159 | #endif | |
1160 | ||
bfcd17a6 TP |
1161 | extern void send_sigio(struct fown_struct *fown, int fd, int band); |
1162 | ||
de2a4a50 | 1163 | #define locks_inode(f) file_inode(f) |
c568d683 | 1164 | |
bfcd17a6 | 1165 | #ifdef CONFIG_FILE_LOCKING |
a75d30c7 | 1166 | extern int fcntl_getlk(struct file *, unsigned int, struct flock *); |
c293621b | 1167 | extern int fcntl_setlk(unsigned int, struct file *, unsigned int, |
a75d30c7 | 1168 | struct flock *); |
1da177e4 LT |
1169 | |
1170 | #if BITS_PER_LONG == 32 | |
a75d30c7 | 1171 | extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *); |
c293621b | 1172 | extern int fcntl_setlk64(unsigned int, struct file *, unsigned int, |
a75d30c7 | 1173 | struct flock64 *); |
1da177e4 LT |
1174 | #endif |
1175 | ||
1da177e4 LT |
1176 | extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg); |
1177 | extern int fcntl_getlease(struct file *filp); | |
1178 | ||
1179 | /* fs/locks.c */ | |
f27a0fe0 | 1180 | void locks_free_lock_context(struct inode *inode); |
05fa3135 | 1181 | void locks_free_lock(struct file_lock *fl); |
1da177e4 | 1182 | extern void locks_init_lock(struct file_lock *); |
c5b1f0d9 | 1183 | extern struct file_lock * locks_alloc_lock(void); |
1da177e4 | 1184 | extern void locks_copy_lock(struct file_lock *, struct file_lock *); |
3fe0fff1 | 1185 | extern void locks_copy_conflock(struct file_lock *, struct file_lock *); |
1da177e4 | 1186 | extern void locks_remove_posix(struct file *, fl_owner_t); |
78ed8a13 | 1187 | extern void locks_remove_file(struct file *); |
a9e61e25 | 1188 | extern void locks_release_private(struct file_lock *); |
6d34ac19 | 1189 | extern void posix_test_lock(struct file *, struct file_lock *); |
150b3934 | 1190 | extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *); |
cb03f94f | 1191 | extern int locks_delete_block(struct file_lock *); |
3ee17abd | 1192 | extern int vfs_test_lock(struct file *, struct file_lock *); |
150b3934 | 1193 | extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *); |
9b9d2ab4 | 1194 | extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl); |
e55c34a6 | 1195 | extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl); |
df4e8d2c | 1196 | extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type); |
95582b00 | 1197 | extern void lease_get_mtime(struct inode *, struct timespec64 *time); |
e6f5c789 JL |
1198 | extern int generic_setlease(struct file *, long, struct file_lock **, void **priv); |
1199 | extern int vfs_setlease(struct file *, long, struct file_lock **, void **); | |
7448cc37 | 1200 | extern int lease_modify(struct file_lock *, int, struct list_head *); |
18f6622e JL |
1201 | |
1202 | struct notifier_block; | |
1203 | extern int lease_register_notifier(struct notifier_block *); | |
1204 | extern void lease_unregister_notifier(struct notifier_block *); | |
1205 | ||
6c8c9031 AV |
1206 | struct files_struct; |
1207 | extern void show_fd_locks(struct seq_file *f, | |
1208 | struct file *filp, struct files_struct *files); | |
bfcd17a6 | 1209 | #else /* !CONFIG_FILE_LOCKING */ |
c1e62b8f JL |
1210 | static inline int fcntl_getlk(struct file *file, unsigned int cmd, |
1211 | struct flock __user *user) | |
c2aca5e5 SW |
1212 | { |
1213 | return -EINVAL; | |
1214 | } | |
1215 | ||
1216 | static inline int fcntl_setlk(unsigned int fd, struct file *file, | |
1217 | unsigned int cmd, struct flock __user *user) | |
1218 | { | |
1219 | return -EACCES; | |
1220 | } | |
1221 | ||
bfcd17a6 | 1222 | #if BITS_PER_LONG == 32 |
c1e62b8f | 1223 | static inline int fcntl_getlk64(struct file *file, unsigned int cmd, |
0ea9fc15 | 1224 | struct flock64 *user) |
c2aca5e5 SW |
1225 | { |
1226 | return -EINVAL; | |
1227 | } | |
1228 | ||
1229 | static inline int fcntl_setlk64(unsigned int fd, struct file *file, | |
0ea9fc15 | 1230 | unsigned int cmd, struct flock64 *user) |
c2aca5e5 SW |
1231 | { |
1232 | return -EACCES; | |
1233 | } | |
bfcd17a6 | 1234 | #endif |
c2aca5e5 SW |
1235 | static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg) |
1236 | { | |
1b2b32dc | 1237 | return -EINVAL; |
c2aca5e5 SW |
1238 | } |
1239 | ||
1240 | static inline int fcntl_getlease(struct file *filp) | |
1241 | { | |
1b2b32dc | 1242 | return F_UNLCK; |
c2aca5e5 SW |
1243 | } |
1244 | ||
4a075e39 | 1245 | static inline void |
f27a0fe0 | 1246 | locks_free_lock_context(struct inode *inode) |
4a075e39 JL |
1247 | { |
1248 | } | |
1249 | ||
c2aca5e5 SW |
1250 | static inline void locks_init_lock(struct file_lock *fl) |
1251 | { | |
1252 | return; | |
1253 | } | |
1254 | ||
3fe0fff1 | 1255 | static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl) |
c2aca5e5 SW |
1256 | { |
1257 | return; | |
1258 | } | |
1259 | ||
1260 | static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl) | |
1261 | { | |
1262 | return; | |
1263 | } | |
1264 | ||
1265 | static inline void locks_remove_posix(struct file *filp, fl_owner_t owner) | |
1266 | { | |
1267 | return; | |
1268 | } | |
1269 | ||
78ed8a13 | 1270 | static inline void locks_remove_file(struct file *filp) |
c2aca5e5 SW |
1271 | { |
1272 | return; | |
1273 | } | |
1274 | ||
1275 | static inline void posix_test_lock(struct file *filp, struct file_lock *fl) | |
1276 | { | |
1277 | return; | |
1278 | } | |
1279 | ||
1280 | static inline int posix_lock_file(struct file *filp, struct file_lock *fl, | |
1281 | struct file_lock *conflock) | |
1282 | { | |
1283 | return -ENOLCK; | |
1284 | } | |
1285 | ||
cb03f94f | 1286 | static inline int locks_delete_block(struct file_lock *waiter) |
c2aca5e5 SW |
1287 | { |
1288 | return -ENOENT; | |
1289 | } | |
1290 | ||
1291 | static inline int vfs_test_lock(struct file *filp, struct file_lock *fl) | |
1292 | { | |
1293 | return 0; | |
1294 | } | |
1295 | ||
1296 | static inline int vfs_lock_file(struct file *filp, unsigned int cmd, | |
1297 | struct file_lock *fl, struct file_lock *conf) | |
1298 | { | |
1299 | return -ENOLCK; | |
1300 | } | |
1301 | ||
1302 | static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl) | |
1303 | { | |
1304 | return 0; | |
1305 | } | |
1306 | ||
e55c34a6 BC |
1307 | static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl) |
1308 | { | |
1309 | return -ENOLCK; | |
1310 | } | |
1311 | ||
df4e8d2c | 1312 | static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type) |
c2aca5e5 SW |
1313 | { |
1314 | return 0; | |
1315 | } | |
1316 | ||
95582b00 DD |
1317 | static inline void lease_get_mtime(struct inode *inode, |
1318 | struct timespec64 *time) | |
c2aca5e5 SW |
1319 | { |
1320 | return; | |
1321 | } | |
1322 | ||
1323 | static inline int generic_setlease(struct file *filp, long arg, | |
e6f5c789 | 1324 | struct file_lock **flp, void **priv) |
c2aca5e5 SW |
1325 | { |
1326 | return -EINVAL; | |
1327 | } | |
1328 | ||
1329 | static inline int vfs_setlease(struct file *filp, long arg, | |
e6f5c789 | 1330 | struct file_lock **lease, void **priv) |
c2aca5e5 SW |
1331 | { |
1332 | return -EINVAL; | |
1333 | } | |
1334 | ||
7448cc37 | 1335 | static inline int lease_modify(struct file_lock *fl, int arg, |
c45198ed | 1336 | struct list_head *dispose) |
c2aca5e5 SW |
1337 | { |
1338 | return -EINVAL; | |
1339 | } | |
6c8c9031 AV |
1340 | |
1341 | struct files_struct; | |
1342 | static inline void show_fd_locks(struct seq_file *f, | |
1343 | struct file *filp, struct files_struct *files) {} | |
bfcd17a6 TP |
1344 | #endif /* !CONFIG_FILE_LOCKING */ |
1345 | ||
ee296d7c JL |
1346 | static inline struct inode *file_inode(const struct file *f) |
1347 | { | |
1348 | return f->f_inode; | |
1349 | } | |
1350 | ||
d101a125 MS |
1351 | static inline struct dentry *file_dentry(const struct file *file) |
1352 | { | |
fb16043b | 1353 | return d_real(file->f_path.dentry, file_inode(file)); |
d101a125 MS |
1354 | } |
1355 | ||
e55c34a6 BC |
1356 | static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl) |
1357 | { | |
c568d683 | 1358 | return locks_lock_inode_wait(locks_inode(filp), fl); |
e55c34a6 BC |
1359 | } |
1360 | ||
1da177e4 | 1361 | struct fasync_struct { |
7a107c0f | 1362 | rwlock_t fa_lock; |
989a2979 ED |
1363 | int magic; |
1364 | int fa_fd; | |
1365 | struct fasync_struct *fa_next; /* singly linked list */ | |
1366 | struct file *fa_file; | |
1367 | struct rcu_head fa_rcu; | |
1da177e4 LT |
1368 | }; |
1369 | ||
1370 | #define FASYNC_MAGIC 0x4601 | |
1371 | ||
1372 | /* SMP safe fasync helpers: */ | |
1373 | extern int fasync_helper(int, struct file *, int, struct fasync_struct **); | |
f7347ce4 LT |
1374 | extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *); |
1375 | extern int fasync_remove_entry(struct file *, struct fasync_struct **); | |
1376 | extern struct fasync_struct *fasync_alloc(void); | |
1377 | extern void fasync_free(struct fasync_struct *); | |
1378 | ||
1da177e4 LT |
1379 | /* can be called from interrupts */ |
1380 | extern void kill_fasync(struct fasync_struct **, int, int); | |
1da177e4 | 1381 | |
e0b93edd | 1382 | extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force); |
393cc3f5 | 1383 | extern int f_setown(struct file *filp, unsigned long arg, int force); |
1da177e4 | 1384 | extern void f_delown(struct file *filp); |
609d7fa9 | 1385 | extern pid_t f_getown(struct file *filp); |
1da177e4 LT |
1386 | extern int send_sigurg(struct fown_struct *fown); |
1387 | ||
e462ec50 DH |
1388 | /* |
1389 | * sb->s_flags. Note that these mirror the equivalent MS_* flags where | |
1390 | * represented in both. | |
1391 | */ | |
1392 | #define SB_RDONLY 1 /* Mount read-only */ | |
1393 | #define SB_NOSUID 2 /* Ignore suid and sgid bits */ | |
1394 | #define SB_NODEV 4 /* Disallow access to device special files */ | |
1395 | #define SB_NOEXEC 8 /* Disallow program execution */ | |
1396 | #define SB_SYNCHRONOUS 16 /* Writes are synced at once */ | |
1397 | #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */ | |
1398 | #define SB_DIRSYNC 128 /* Directory modifications are synchronous */ | |
1399 | #define SB_NOATIME 1024 /* Do not update access times. */ | |
1400 | #define SB_NODIRATIME 2048 /* Do not update directory access times */ | |
1401 | #define SB_SILENT 32768 | |
1402 | #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */ | |
457e7a13 | 1403 | #define SB_INLINECRYPT (1<<17) /* Use blk-crypto for encrypted files */ |
e462ec50 DH |
1404 | #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */ |
1405 | #define SB_I_VERSION (1<<23) /* Update inode I_version field */ | |
1406 | #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */ | |
1407 | ||
1408 | /* These sb flags are internal to the kernel */ | |
1409 | #define SB_SUBMOUNT (1<<26) | |
8d0347f6 | 1410 | #define SB_FORCE (1<<27) |
e462ec50 DH |
1411 | #define SB_NOSEC (1<<28) |
1412 | #define SB_BORN (1<<29) | |
1413 | #define SB_ACTIVE (1<<30) | |
1414 | #define SB_NOUSER (1<<31) | |
1415 | ||
c843843e DR |
1416 | /* These flags relate to encoding and casefolding */ |
1417 | #define SB_ENC_STRICT_MODE_FL (1 << 0) | |
1418 | ||
1419 | #define sb_has_strict_encoding(sb) \ | |
1420 | (sb->s_encoding_flags & SB_ENC_STRICT_MODE_FL) | |
1421 | ||
1da177e4 LT |
1422 | /* |
1423 | * Umount options | |
1424 | */ | |
1425 | ||
1426 | #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */ | |
1427 | #define MNT_DETACH 0x00000002 /* Just detach from the tree */ | |
1428 | #define MNT_EXPIRE 0x00000004 /* Mark for expiry */ | |
db1f05bb MS |
1429 | #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */ |
1430 | #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */ | |
1da177e4 | 1431 | |
46b15caa TH |
1432 | /* sb->s_iflags */ |
1433 | #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */ | |
90f8572b | 1434 | #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */ |
a2982cc9 | 1435 | #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */ |
1cb039f3 | 1436 | #define SB_I_STABLE_WRITES 0x00000008 /* don't modify blks until WB is done */ |
1da177e4 | 1437 | |
8654df4e EB |
1438 | /* sb->s_iflags to limit user namespace mounts */ |
1439 | #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */ | |
57b56ac6 MZ |
1440 | #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020 |
1441 | #define SB_I_UNTRUSTED_MOUNTER 0x00000040 | |
1da177e4 | 1442 | |
32b1924b | 1443 | #define SB_I_SKIP_SYNC 0x00000100 /* Skip superblock at global sync */ |
0b3ea092 | 1444 | #define SB_I_PERSB_BDI 0x00000200 /* has a per-sb bdi */ |
32b1924b | 1445 | |
5accdf82 JK |
1446 | /* Possible states of 'frozen' field */ |
1447 | enum { | |
1448 | SB_UNFROZEN = 0, /* FS is unfrozen */ | |
1449 | SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */ | |
5accdf82 JK |
1450 | SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */ |
1451 | SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop | |
1452 | * internal threads if needed) */ | |
1453 | SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */ | |
1454 | }; | |
1455 | ||
1456 | #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1) | |
1457 | ||
1458 | struct sb_writers { | |
8129ed29 | 1459 | int frozen; /* Is sb frozen? */ |
60b49885 | 1460 | wait_queue_head_t wait_unfrozen; /* wait for thaw */ |
8129ed29 | 1461 | struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS]; |
5accdf82 JK |
1462 | }; |
1463 | ||
1da177e4 LT |
1464 | struct super_block { |
1465 | struct list_head s_list; /* Keep this first */ | |
1466 | dev_t s_dev; /* search index; _not_ kdev_t */ | |
270ba5f7 RK |
1467 | unsigned char s_blocksize_bits; |
1468 | unsigned long s_blocksize; | |
42cb56ae | 1469 | loff_t s_maxbytes; /* Max file size */ |
1da177e4 | 1470 | struct file_system_type *s_type; |
ee9b6d61 | 1471 | const struct super_operations *s_op; |
61e225dc | 1472 | const struct dquot_operations *dq_op; |
0d54b217 | 1473 | const struct quotactl_ops *s_qcop; |
39655164 | 1474 | const struct export_operations *s_export_op; |
1da177e4 | 1475 | unsigned long s_flags; |
46b15caa | 1476 | unsigned long s_iflags; /* internal SB_I_* flags */ |
1da177e4 LT |
1477 | unsigned long s_magic; |
1478 | struct dentry *s_root; | |
1479 | struct rw_semaphore s_umount; | |
1da177e4 | 1480 | int s_count; |
1da177e4 | 1481 | atomic_t s_active; |
50462062 | 1482 | #ifdef CONFIG_SECURITY |
1da177e4 | 1483 | void *s_security; |
50462062 | 1484 | #endif |
bb435453 | 1485 | const struct xattr_handler **s_xattr; |
643fa961 | 1486 | #ifdef CONFIG_FS_ENCRYPTION |
0b81d077 | 1487 | const struct fscrypt_operations *s_cop; |
22d94f49 | 1488 | struct key *s_master_keys; /* master crypto keys in use */ |
5585f2af EB |
1489 | #endif |
1490 | #ifdef CONFIG_FS_VERITY | |
1491 | const struct fsverity_operations *s_vop; | |
c843843e | 1492 | #endif |
5298d4bf | 1493 | #if IS_ENABLED(CONFIG_UNICODE) |
c843843e DR |
1494 | struct unicode_map *s_encoding; |
1495 | __u16 s_encoding_flags; | |
bbbc3fb6 | 1496 | #endif |
f1ee6162 | 1497 | struct hlist_bl_head s_roots; /* alternate root dentries for NFS */ |
39f7c4db | 1498 | struct list_head s_mounts; /* list of mounts; _not_ for fs use */ |
1da177e4 | 1499 | struct block_device *s_bdev; |
32a88aa1 | 1500 | struct backing_dev_info *s_bdi; |
acaebfd8 | 1501 | struct mtd_info *s_mtd; |
a5166169 | 1502 | struct hlist_node s_instances; |
2c5f648a | 1503 | unsigned int s_quota_types; /* Bitmask of supported quota types */ |
1da177e4 LT |
1504 | struct quota_info s_dquot; /* Diskquota specific options */ |
1505 | ||
5accdf82 | 1506 | struct sb_writers s_writers; |
1da177e4 | 1507 | |
99c228a9 AG |
1508 | /* |
1509 | * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and | |
1510 | * s_fsnotify_marks together for cache efficiency. They are frequently | |
1511 | * accessed and rarely modified. | |
1512 | */ | |
1513 | void *s_fs_info; /* Filesystem private info */ | |
1514 | ||
1515 | /* Granularity of c/m/atime in ns (cannot be worse than a second) */ | |
1516 | u32 s_time_gran; | |
188d20bc DD |
1517 | /* Time limits for c/m/atime in seconds */ |
1518 | time64_t s_time_min; | |
1519 | time64_t s_time_max; | |
99c228a9 AG |
1520 | #ifdef CONFIG_FSNOTIFY |
1521 | __u32 s_fsnotify_mask; | |
1522 | struct fsnotify_mark_connector __rcu *s_fsnotify_marks; | |
1523 | #endif | |
1524 | ||
85787090 CH |
1525 | char s_id[32]; /* Informational name */ |
1526 | uuid_t s_uuid; /* UUID */ | |
1da177e4 | 1527 | |
8de52778 | 1528 | unsigned int s_max_links; |
30c40d2c | 1529 | fmode_t s_mode; |
1da177e4 LT |
1530 | |
1531 | /* | |
1532 | * The next field is for VFS *only*. No filesystems have any business | |
1533 | * even looking at it. You had been warned. | |
1534 | */ | |
a11f3a05 | 1535 | struct mutex s_vfs_rename_mutex; /* Kludge */ |
1da177e4 | 1536 | |
79c0b2df MS |
1537 | /* |
1538 | * Filesystem subtype. If non-empty the filesystem type field | |
1539 | * in /proc/mounts will be "type.subtype" | |
1540 | */ | |
a0c9a8b8 | 1541 | const char *s_subtype; |
b3b304a2 | 1542 | |
c8aebb0c | 1543 | const struct dentry_operations *s_d_op; /* default d_op for dentries */ |
9fdfdcf1 DM |
1544 | |
1545 | /* | |
1546 | * Saved pool identifier for cleancache (-1 means none) | |
1547 | */ | |
1548 | int cleancache_poolid; | |
b0d40c92 DC |
1549 | |
1550 | struct shrinker s_shrink; /* per-sb shrinker handle */ | |
4ed5e82f | 1551 | |
7ada4db8 MS |
1552 | /* Number of inodes with nlink == 0 but still referenced */ |
1553 | atomic_long_t s_remove_count; | |
1554 | ||
ec44610f AG |
1555 | /* |
1556 | * Number of inode/mount/sb objects that are being watched, note that | |
1557 | * inodes objects are currently double-accounted. | |
1558 | */ | |
1559 | atomic_long_t s_fsnotify_connectors; | |
721fb6fb | 1560 | |
4ed5e82f MS |
1561 | /* Being remounted read-only */ |
1562 | int s_readonly_remount; | |
7b7a8665 | 1563 | |
735e4ae5 JL |
1564 | /* per-sb errseq_t for reporting writeback errors via syncfs */ |
1565 | errseq_t s_wb_err; | |
1566 | ||
7b7a8665 CH |
1567 | /* AIO completions deferred from interrupt context */ |
1568 | struct workqueue_struct *s_dio_done_wq; | |
215752fc | 1569 | struct hlist_head s_pins; |
f6041567 | 1570 | |
6e4eab57 EB |
1571 | /* |
1572 | * Owning user namespace and default context in which to | |
1573 | * interpret filesystem uids, gids, quotas, device nodes, | |
1574 | * xattrs and security labels. | |
1575 | */ | |
1576 | struct user_namespace *s_user_ns; | |
1577 | ||
f6041567 | 1578 | /* |
7d10f70f WL |
1579 | * The list_lru structure is essentially just a pointer to a table |
1580 | * of per-node lru lists, each of which has its own spinlock. | |
1581 | * There is no need to put them into separate cachelines. | |
f6041567 | 1582 | */ |
7d10f70f WL |
1583 | struct list_lru s_dentry_lru; |
1584 | struct list_lru s_inode_lru; | |
e2fec7c3 | 1585 | struct rcu_head rcu; |
853b39a7 | 1586 | struct work_struct destroy_work; |
69c433ed | 1587 | |
e97fedb9 | 1588 | struct mutex s_sync_lock; /* sync serialisation lock */ |
69c433ed MS |
1589 | |
1590 | /* | |
1591 | * Indicates how deep in a filesystem stack this SB is | |
1592 | */ | |
1593 | int s_stack_depth; | |
74278da9 DC |
1594 | |
1595 | /* s_inode_list_lock protects s_inodes */ | |
1596 | spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp; | |
1597 | struct list_head s_inodes; /* all inodes */ | |
6c60d2b5 DC |
1598 | |
1599 | spinlock_t s_inode_wblist_lock; | |
1600 | struct list_head s_inodes_wb; /* writeback inodes */ | |
3859a271 | 1601 | } __randomize_layout; |
1da177e4 | 1602 | |
a1ec9040 CB |
1603 | static inline struct user_namespace *i_user_ns(const struct inode *inode) |
1604 | { | |
1605 | return inode->i_sb->s_user_ns; | |
1606 | } | |
1607 | ||
81754357 SF |
1608 | /* Helper functions so that in most cases filesystems will |
1609 | * not need to deal directly with kuid_t and kgid_t and can | |
1610 | * instead deal with the raw numeric values that are stored | |
1611 | * in the filesystem. | |
1612 | */ | |
1613 | static inline uid_t i_uid_read(const struct inode *inode) | |
1614 | { | |
a1ec9040 | 1615 | return from_kuid(i_user_ns(inode), inode->i_uid); |
81754357 SF |
1616 | } |
1617 | ||
1618 | static inline gid_t i_gid_read(const struct inode *inode) | |
1619 | { | |
a1ec9040 | 1620 | return from_kgid(i_user_ns(inode), inode->i_gid); |
81754357 SF |
1621 | } |
1622 | ||
1623 | static inline void i_uid_write(struct inode *inode, uid_t uid) | |
1624 | { | |
a1ec9040 | 1625 | inode->i_uid = make_kuid(i_user_ns(inode), uid); |
81754357 SF |
1626 | } |
1627 | ||
1628 | static inline void i_gid_write(struct inode *inode, gid_t gid) | |
1629 | { | |
a1ec9040 | 1630 | inode->i_gid = make_kgid(i_user_ns(inode), gid); |
81754357 SF |
1631 | } |
1632 | ||
1bd66c1a CB |
1633 | /** |
1634 | * i_uid_into_mnt - map an inode's i_uid down into a mnt_userns | |
1635 | * @mnt_userns: user namespace of the mount the inode was found from | |
1636 | * @inode: inode to map | |
1637 | * | |
1638 | * Return: the inode's i_uid mapped down according to @mnt_userns. | |
1639 | * If the inode's i_uid has no mapping INVALID_UID is returned. | |
1640 | */ | |
e6c9a714 CB |
1641 | static inline kuid_t i_uid_into_mnt(struct user_namespace *mnt_userns, |
1642 | const struct inode *inode) | |
1643 | { | |
bd303368 | 1644 | return mapped_kuid_fs(mnt_userns, i_user_ns(inode), inode->i_uid); |
e6c9a714 CB |
1645 | } |
1646 | ||
1bd66c1a CB |
1647 | /** |
1648 | * i_gid_into_mnt - map an inode's i_gid down into a mnt_userns | |
1649 | * @mnt_userns: user namespace of the mount the inode was found from | |
1650 | * @inode: inode to map | |
1651 | * | |
1652 | * Return: the inode's i_gid mapped down according to @mnt_userns. | |
1653 | * If the inode's i_gid has no mapping INVALID_GID is returned. | |
1654 | */ | |
e6c9a714 CB |
1655 | static inline kgid_t i_gid_into_mnt(struct user_namespace *mnt_userns, |
1656 | const struct inode *inode) | |
1657 | { | |
bd303368 | 1658 | return mapped_kgid_fs(mnt_userns, i_user_ns(inode), inode->i_gid); |
e6c9a714 CB |
1659 | } |
1660 | ||
db998553 CB |
1661 | /** |
1662 | * inode_fsuid_set - initialize inode's i_uid field with callers fsuid | |
1663 | * @inode: inode to initialize | |
1664 | * @mnt_userns: user namespace of the mount the inode was found from | |
1665 | * | |
1666 | * Initialize the i_uid field of @inode. If the inode was found/created via | |
1667 | * an idmapped mount map the caller's fsuid according to @mnt_users. | |
1668 | */ | |
1669 | static inline void inode_fsuid_set(struct inode *inode, | |
1670 | struct user_namespace *mnt_userns) | |
1671 | { | |
bd303368 | 1672 | inode->i_uid = mapped_fsuid(mnt_userns, i_user_ns(inode)); |
db998553 CB |
1673 | } |
1674 | ||
1675 | /** | |
1676 | * inode_fsgid_set - initialize inode's i_gid field with callers fsgid | |
1677 | * @inode: inode to initialize | |
1678 | * @mnt_userns: user namespace of the mount the inode was found from | |
1679 | * | |
1680 | * Initialize the i_gid field of @inode. If the inode was found/created via | |
1681 | * an idmapped mount map the caller's fsgid according to @mnt_users. | |
1682 | */ | |
1683 | static inline void inode_fsgid_set(struct inode *inode, | |
1684 | struct user_namespace *mnt_userns) | |
1685 | { | |
bd303368 | 1686 | inode->i_gid = mapped_fsgid(mnt_userns, i_user_ns(inode)); |
db998553 CB |
1687 | } |
1688 | ||
8e538913 CB |
1689 | /** |
1690 | * fsuidgid_has_mapping() - check whether caller's fsuid/fsgid is mapped | |
1691 | * @sb: the superblock we want a mapping in | |
1692 | * @mnt_userns: user namespace of the relevant mount | |
1693 | * | |
1694 | * Check whether the caller's fsuid and fsgid have a valid mapping in the | |
1695 | * s_user_ns of the superblock @sb. If the caller is on an idmapped mount map | |
1696 | * the caller's fsuid and fsgid according to the @mnt_userns first. | |
1697 | * | |
1698 | * Return: true if fsuid and fsgid is mapped, false if not. | |
1699 | */ | |
1700 | static inline bool fsuidgid_has_mapping(struct super_block *sb, | |
1701 | struct user_namespace *mnt_userns) | |
1702 | { | |
476860b3 CB |
1703 | struct user_namespace *fs_userns = sb->s_user_ns; |
1704 | kuid_t kuid; | |
1705 | kgid_t kgid; | |
1706 | ||
bd303368 | 1707 | kuid = mapped_fsuid(mnt_userns, fs_userns); |
476860b3 CB |
1708 | if (!uid_valid(kuid)) |
1709 | return false; | |
bd303368 | 1710 | kgid = mapped_fsgid(mnt_userns, fs_userns); |
476860b3 CB |
1711 | if (!gid_valid(kgid)) |
1712 | return false; | |
1713 | return kuid_has_mapping(fs_userns, kuid) && | |
1714 | kgid_has_mapping(fs_userns, kgid); | |
8e538913 CB |
1715 | } |
1716 | ||
95582b00 | 1717 | extern struct timespec64 current_time(struct inode *inode); |
1da177e4 LT |
1718 | |
1719 | /* | |
1720 | * Snapshotting support. | |
1721 | */ | |
1da177e4 | 1722 | |
9b852342 DW |
1723 | /* |
1724 | * These are internal functions, please use sb_start_{write,pagefault,intwrite} | |
1725 | * instead. | |
1726 | */ | |
1727 | static inline void __sb_end_write(struct super_block *sb, int level) | |
1728 | { | |
1729 | percpu_up_read(sb->s_writers.rw_sem + level-1); | |
1730 | } | |
1731 | ||
1732 | static inline void __sb_start_write(struct super_block *sb, int level) | |
1733 | { | |
1734 | percpu_down_read(sb->s_writers.rw_sem + level - 1); | |
1735 | } | |
1736 | ||
1737 | static inline bool __sb_start_write_trylock(struct super_block *sb, int level) | |
1738 | { | |
1739 | return percpu_down_read_trylock(sb->s_writers.rw_sem + level - 1); | |
1740 | } | |
5accdf82 | 1741 | |
bee9182d | 1742 | #define __sb_writers_acquired(sb, lev) \ |
8129ed29 | 1743 | percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_) |
bee9182d | 1744 | #define __sb_writers_release(sb, lev) \ |
8129ed29 | 1745 | percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_) |
bee9182d | 1746 | |
5accdf82 JK |
1747 | /** |
1748 | * sb_end_write - drop write access to a superblock | |
1749 | * @sb: the super we wrote to | |
1750 | * | |
1751 | * Decrement number of writers to the filesystem. Wake up possible waiters | |
1752 | * wanting to freeze the filesystem. | |
1753 | */ | |
1754 | static inline void sb_end_write(struct super_block *sb) | |
1755 | { | |
1756 | __sb_end_write(sb, SB_FREEZE_WRITE); | |
1757 | } | |
1758 | ||
1759 | /** | |
1760 | * sb_end_pagefault - drop write access to a superblock from a page fault | |
1761 | * @sb: the super we wrote to | |
1762 | * | |
1763 | * Decrement number of processes handling write page fault to the filesystem. | |
1764 | * Wake up possible waiters wanting to freeze the filesystem. | |
1765 | */ | |
1766 | static inline void sb_end_pagefault(struct super_block *sb) | |
1767 | { | |
1768 | __sb_end_write(sb, SB_FREEZE_PAGEFAULT); | |
1769 | } | |
1770 | ||
1771 | /** | |
1772 | * sb_end_intwrite - drop write access to a superblock for internal fs purposes | |
1773 | * @sb: the super we wrote to | |
1774 | * | |
1775 | * Decrement fs-internal number of writers to the filesystem. Wake up possible | |
1776 | * waiters wanting to freeze the filesystem. | |
1777 | */ | |
1778 | static inline void sb_end_intwrite(struct super_block *sb) | |
1779 | { | |
1780 | __sb_end_write(sb, SB_FREEZE_FS); | |
1781 | } | |
1782 | ||
1783 | /** | |
1784 | * sb_start_write - get write access to a superblock | |
1785 | * @sb: the super we write to | |
1786 | * | |
1787 | * When a process wants to write data or metadata to a file system (i.e. dirty | |
1788 | * a page or an inode), it should embed the operation in a sb_start_write() - | |
1789 | * sb_end_write() pair to get exclusion against file system freezing. This | |
1790 | * function increments number of writers preventing freezing. If the file | |
1791 | * system is already frozen, the function waits until the file system is | |
1792 | * thawed. | |
1793 | * | |
1794 | * Since freeze protection behaves as a lock, users have to preserve | |
1795 | * ordering of freeze protection and other filesystem locks. Generally, | |
1796 | * freeze protection should be the outermost lock. In particular, we have: | |
1797 | * | |
1798 | * sb_start_write | |
1799 | * -> i_mutex (write path, truncate, directory ops, ...) | |
1800 | * -> s_umount (freeze_super, thaw_super) | |
1801 | */ | |
1802 | static inline void sb_start_write(struct super_block *sb) | |
1803 | { | |
8a3c84b6 | 1804 | __sb_start_write(sb, SB_FREEZE_WRITE); |
5accdf82 JK |
1805 | } |
1806 | ||
8a3c84b6 | 1807 | static inline bool sb_start_write_trylock(struct super_block *sb) |
5accdf82 | 1808 | { |
8a3c84b6 | 1809 | return __sb_start_write_trylock(sb, SB_FREEZE_WRITE); |
5accdf82 JK |
1810 | } |
1811 | ||
1812 | /** | |
1813 | * sb_start_pagefault - get write access to a superblock from a page fault | |
1814 | * @sb: the super we write to | |
1815 | * | |
1816 | * When a process starts handling write page fault, it should embed the | |
1817 | * operation into sb_start_pagefault() - sb_end_pagefault() pair to get | |
1818 | * exclusion against file system freezing. This is needed since the page fault | |
1819 | * is going to dirty a page. This function increments number of running page | |
1820 | * faults preventing freezing. If the file system is already frozen, the | |
1821 | * function waits until the file system is thawed. | |
1822 | * | |
1823 | * Since page fault freeze protection behaves as a lock, users have to preserve | |
1824 | * ordering of freeze protection and other filesystem locks. It is advised to | |
c1e8d7c6 | 1825 | * put sb_start_pagefault() close to mmap_lock in lock ordering. Page fault |
5accdf82 JK |
1826 | * handling code implies lock dependency: |
1827 | * | |
c1e8d7c6 | 1828 | * mmap_lock |
5accdf82 JK |
1829 | * -> sb_start_pagefault |
1830 | */ | |
1831 | static inline void sb_start_pagefault(struct super_block *sb) | |
1832 | { | |
8a3c84b6 | 1833 | __sb_start_write(sb, SB_FREEZE_PAGEFAULT); |
5accdf82 JK |
1834 | } |
1835 | ||
39015399 | 1836 | /** |
5accdf82 JK |
1837 | * sb_start_intwrite - get write access to a superblock for internal fs purposes |
1838 | * @sb: the super we write to | |
1839 | * | |
1840 | * This is the third level of protection against filesystem freezing. It is | |
1841 | * free for use by a filesystem. The only requirement is that it must rank | |
1842 | * below sb_start_pagefault. | |
1843 | * | |
1844 | * For example filesystem can call sb_start_intwrite() when starting a | |
1845 | * transaction which somewhat eases handling of freezing for internal sources | |
1846 | * of filesystem changes (internal fs threads, discarding preallocation on file | |
1847 | * close, etc.). | |
1848 | */ | |
1849 | static inline void sb_start_intwrite(struct super_block *sb) | |
1850 | { | |
8a3c84b6 | 1851 | __sb_start_write(sb, SB_FREEZE_FS); |
5accdf82 | 1852 | } |
1da177e4 | 1853 | |
8a3c84b6 | 1854 | static inline bool sb_start_intwrite_trylock(struct super_block *sb) |
0c8e3fe3 | 1855 | { |
8a3c84b6 | 1856 | return __sb_start_write_trylock(sb, SB_FREEZE_FS); |
0c8e3fe3 AG |
1857 | } |
1858 | ||
21cb47be CB |
1859 | bool inode_owner_or_capable(struct user_namespace *mnt_userns, |
1860 | const struct inode *inode); | |
3bd858ab | 1861 | |
1da177e4 LT |
1862 | /* |
1863 | * VFS helper functions.. | |
1864 | */ | |
6521f891 CB |
1865 | int vfs_create(struct user_namespace *, struct inode *, |
1866 | struct dentry *, umode_t, bool); | |
1867 | int vfs_mkdir(struct user_namespace *, struct inode *, | |
1868 | struct dentry *, umode_t); | |
1869 | int vfs_mknod(struct user_namespace *, struct inode *, struct dentry *, | |
1870 | umode_t, dev_t); | |
1871 | int vfs_symlink(struct user_namespace *, struct inode *, | |
1872 | struct dentry *, const char *); | |
1873 | int vfs_link(struct dentry *, struct user_namespace *, struct inode *, | |
1874 | struct dentry *, struct inode **); | |
1875 | int vfs_rmdir(struct user_namespace *, struct inode *, struct dentry *); | |
1876 | int vfs_unlink(struct user_namespace *, struct inode *, struct dentry *, | |
1877 | struct inode **); | |
9fe61450 | 1878 | |
92cb01c7 CB |
1879 | /** |
1880 | * struct renamedata - contains all information required for renaming | |
1881 | * @old_mnt_userns: old user namespace of the mount the inode was found from | |
1882 | * @old_dir: parent of source | |
1883 | * @old_dentry: source | |
1884 | * @new_mnt_userns: new user namespace of the mount the inode was found from | |
1885 | * @new_dir: parent of destination | |
1886 | * @new_dentry: destination | |
1887 | * @delegated_inode: returns an inode needing a delegation break | |
1888 | * @flags: rename flags | |
1889 | */ | |
9fe61450 | 1890 | struct renamedata { |
6521f891 | 1891 | struct user_namespace *old_mnt_userns; |
9fe61450 CB |
1892 | struct inode *old_dir; |
1893 | struct dentry *old_dentry; | |
6521f891 | 1894 | struct user_namespace *new_mnt_userns; |
9fe61450 CB |
1895 | struct inode *new_dir; |
1896 | struct dentry *new_dentry; | |
1897 | struct inode **delegated_inode; | |
1898 | unsigned int flags; | |
1899 | } __randomize_layout; | |
a3c751a5 | 1900 | |
9fe61450 | 1901 | int vfs_rename(struct renamedata *); |
a3c751a5 | 1902 | |
6521f891 CB |
1903 | static inline int vfs_whiteout(struct user_namespace *mnt_userns, |
1904 | struct inode *dir, struct dentry *dentry) | |
a3c751a5 | 1905 | { |
6521f891 CB |
1906 | return vfs_mknod(mnt_userns, dir, dentry, S_IFCHR | WHITEOUT_MODE, |
1907 | WHITEOUT_DEV); | |
a3c751a5 | 1908 | } |
1da177e4 | 1909 | |
6521f891 CB |
1910 | struct dentry *vfs_tmpfile(struct user_namespace *mnt_userns, |
1911 | struct dentry *dentry, umode_t mode, int open_flag); | |
af7bd4dc | 1912 | |
8e6c848e AV |
1913 | int vfs_mkobj(struct dentry *, umode_t, |
1914 | int (*f)(struct dentry *, umode_t, void *), | |
1915 | void *); | |
1916 | ||
c04011fe | 1917 | int vfs_fchown(struct file *file, uid_t user, gid_t group); |
9e96c8c0 | 1918 | int vfs_fchmod(struct file *file, umode_t mode); |
fd5ad30c | 1919 | int vfs_utimes(const struct path *path, struct timespec64 *times); |
c04011fe | 1920 | |
9df6702a MS |
1921 | extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg); |
1922 | ||
2952db0f AB |
1923 | #ifdef CONFIG_COMPAT |
1924 | extern long compat_ptr_ioctl(struct file *file, unsigned int cmd, | |
1925 | unsigned long arg); | |
1926 | #else | |
1927 | #define compat_ptr_ioctl NULL | |
1928 | #endif | |
1929 | ||
8c744fb8 CH |
1930 | /* |
1931 | * VFS file helper functions. | |
1932 | */ | |
21cb47be CB |
1933 | void inode_init_owner(struct user_namespace *mnt_userns, struct inode *inode, |
1934 | const struct inode *dir, umode_t mode); | |
a2982cc9 | 1935 | extern bool may_open_dev(const struct path *path); |
c4b929b8 | 1936 | |
1da177e4 LT |
1937 | /* |
1938 | * This is the "filldir" function type, used by readdir() to let | |
1939 | * the kernel specify what kind of dirent layout it wants to have. | |
1940 | * This allows the kernel to read directories into kernel space or | |
1941 | * to have different dirent layouts depending on the binary type. | |
1942 | */ | |
ac7576f4 MS |
1943 | struct dir_context; |
1944 | typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64, | |
1945 | unsigned); | |
1946 | ||
5c0ba4e0 | 1947 | struct dir_context { |
a09acf4b | 1948 | filldir_t actor; |
bb6f619b | 1949 | loff_t pos; |
5c0ba4e0 | 1950 | }; |
bb6f619b | 1951 | |
b4caecd4 CH |
1952 | /* |
1953 | * These flags let !MMU mmap() govern direct device mapping vs immediate | |
1954 | * copying more easily for MAP_PRIVATE, especially for ROM filesystems. | |
1955 | * | |
1956 | * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE) | |
1957 | * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED) | |
1958 | * NOMMU_MAP_READ: Can be mapped for reading | |
1959 | * NOMMU_MAP_WRITE: Can be mapped for writing | |
1960 | * NOMMU_MAP_EXEC: Can be mapped for execution | |
1961 | */ | |
1962 | #define NOMMU_MAP_COPY 0x00000001 | |
1963 | #define NOMMU_MAP_DIRECT 0x00000008 | |
1964 | #define NOMMU_MAP_READ VM_MAYREAD | |
1965 | #define NOMMU_MAP_WRITE VM_MAYWRITE | |
1966 | #define NOMMU_MAP_EXEC VM_MAYEXEC | |
1967 | ||
1968 | #define NOMMU_VMFLAGS \ | |
1969 | (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC) | |
1970 | ||
2e5dfc99 DW |
1971 | /* |
1972 | * These flags control the behavior of the remap_file_range function pointer. | |
1973 | * If it is called with len == 0 that means "remap to end of source file". | |
5c437fa2 | 1974 | * See Documentation/filesystems/vfs.rst for more details about this call. |
2e5dfc99 DW |
1975 | * |
1976 | * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate) | |
eca3654e | 1977 | * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request |
2e5dfc99 DW |
1978 | */ |
1979 | #define REMAP_FILE_DEDUP (1 << 0) | |
eca3654e | 1980 | #define REMAP_FILE_CAN_SHORTEN (1 << 1) |
2e5dfc99 DW |
1981 | |
1982 | /* | |
1983 | * These flags signal that the caller is ok with altering various aspects of | |
1984 | * the behavior of the remap operation. The changes must be made by the | |
1985 | * implementation; the vfs remap helper functions can take advantage of them. | |
1986 | * Flags in this category exist to preserve the quirky behavior of the hoisted | |
1987 | * btrfs clone/dedupe ioctls. | |
2e5dfc99 | 1988 | */ |
eca3654e | 1989 | #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN) |
b4caecd4 | 1990 | |
293bc982 AV |
1991 | struct iov_iter; |
1992 | ||
1da177e4 LT |
1993 | struct file_operations { |
1994 | struct module *owner; | |
1995 | loff_t (*llseek) (struct file *, loff_t, int); | |
1996 | ssize_t (*read) (struct file *, char __user *, size_t, loff_t *); | |
1da177e4 | 1997 | ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *); |
293bc982 AV |
1998 | ssize_t (*read_iter) (struct kiocb *, struct iov_iter *); |
1999 | ssize_t (*write_iter) (struct kiocb *, struct iov_iter *); | |
5a72e899 JA |
2000 | int (*iopoll)(struct kiocb *kiocb, struct io_comp_batch *, |
2001 | unsigned int flags); | |
bb6f619b | 2002 | int (*iterate) (struct file *, struct dir_context *); |
61922694 | 2003 | int (*iterate_shared) (struct file *, struct dir_context *); |
a3f8683b | 2004 | __poll_t (*poll) (struct file *, struct poll_table_struct *); |
1da177e4 LT |
2005 | long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long); |
2006 | long (*compat_ioctl) (struct file *, unsigned int, unsigned long); | |
2007 | int (*mmap) (struct file *, struct vm_area_struct *); | |
1c972597 | 2008 | unsigned long mmap_supported_flags; |
1da177e4 | 2009 | int (*open) (struct inode *, struct file *); |
75e1fcc0 | 2010 | int (*flush) (struct file *, fl_owner_t id); |
1da177e4 | 2011 | int (*release) (struct inode *, struct file *); |
02c24a82 | 2012 | int (*fsync) (struct file *, loff_t, loff_t, int datasync); |
1da177e4 LT |
2013 | int (*fasync) (int, struct file *, int); |
2014 | int (*lock) (struct file *, int, struct file_lock *); | |
1da177e4 LT |
2015 | ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int); |
2016 | unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); | |
2017 | int (*check_flags)(int); | |
1da177e4 | 2018 | int (*flock) (struct file *, int, struct file_lock *); |
cbb7e577 JA |
2019 | ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int); |
2020 | ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int); | |
e6f5c789 | 2021 | int (*setlease)(struct file *, long, struct file_lock **, void **); |
2fe17c10 CH |
2022 | long (*fallocate)(struct file *file, int mode, loff_t offset, |
2023 | loff_t len); | |
a3816ab0 | 2024 | void (*show_fdinfo)(struct seq_file *m, struct file *f); |
b4caecd4 CH |
2025 | #ifndef CONFIG_MMU |
2026 | unsigned (*mmap_capabilities)(struct file *); | |
2027 | #endif | |
04b38d60 CH |
2028 | ssize_t (*copy_file_range)(struct file *, loff_t, struct file *, |
2029 | loff_t, size_t, unsigned int); | |
42ec3d4c DW |
2030 | loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in, |
2031 | struct file *file_out, loff_t pos_out, | |
2032 | loff_t len, unsigned int remap_flags); | |
45cd0faa | 2033 | int (*fadvise)(struct file *, loff_t, loff_t, int); |
3859a271 | 2034 | } __randomize_layout; |
1da177e4 LT |
2035 | |
2036 | struct inode_operations { | |
00cd8dd3 | 2037 | struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int); |
fceef393 | 2038 | const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *); |
549c7297 | 2039 | int (*permission) (struct user_namespace *, struct inode *, int); |
0cad6246 | 2040 | struct posix_acl * (*get_acl)(struct inode *, int, bool); |
44a7d7a8 NP |
2041 | |
2042 | int (*readlink) (struct dentry *, char __user *,int); | |
44a7d7a8 | 2043 | |
549c7297 CB |
2044 | int (*create) (struct user_namespace *, struct inode *,struct dentry *, |
2045 | umode_t, bool); | |
1da177e4 LT |
2046 | int (*link) (struct dentry *,struct inode *,struct dentry *); |
2047 | int (*unlink) (struct inode *,struct dentry *); | |
549c7297 CB |
2048 | int (*symlink) (struct user_namespace *, struct inode *,struct dentry *, |
2049 | const char *); | |
2050 | int (*mkdir) (struct user_namespace *, struct inode *,struct dentry *, | |
2051 | umode_t); | |
1da177e4 | 2052 | int (*rmdir) (struct inode *,struct dentry *); |
549c7297 CB |
2053 | int (*mknod) (struct user_namespace *, struct inode *,struct dentry *, |
2054 | umode_t,dev_t); | |
2055 | int (*rename) (struct user_namespace *, struct inode *, struct dentry *, | |
520c8b16 | 2056 | struct inode *, struct dentry *, unsigned int); |
549c7297 CB |
2057 | int (*setattr) (struct user_namespace *, struct dentry *, |
2058 | struct iattr *); | |
2059 | int (*getattr) (struct user_namespace *, const struct path *, | |
2060 | struct kstat *, u32, unsigned int); | |
1da177e4 | 2061 | ssize_t (*listxattr) (struct dentry *, char *, size_t); |
c4b929b8 MF |
2062 | int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start, |
2063 | u64 len); | |
95582b00 | 2064 | int (*update_time)(struct inode *, struct timespec64 *, int); |
d9585277 | 2065 | int (*atomic_open)(struct inode *, struct dentry *, |
30d90494 | 2066 | struct file *, unsigned open_flag, |
44907d79 | 2067 | umode_t create_mode); |
549c7297 CB |
2068 | int (*tmpfile) (struct user_namespace *, struct inode *, |
2069 | struct dentry *, umode_t); | |
2070 | int (*set_acl)(struct user_namespace *, struct inode *, | |
2071 | struct posix_acl *, int); | |
4c5b4799 MS |
2072 | int (*fileattr_set)(struct user_namespace *mnt_userns, |
2073 | struct dentry *dentry, struct fileattr *fa); | |
2074 | int (*fileattr_get)(struct dentry *dentry, struct fileattr *fa); | |
44a7d7a8 | 2075 | } ____cacheline_aligned; |
1da177e4 | 2076 | |
bb7462b6 MS |
2077 | static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio, |
2078 | struct iov_iter *iter) | |
2079 | { | |
2080 | return file->f_op->read_iter(kio, iter); | |
2081 | } | |
2082 | ||
2083 | static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio, | |
2084 | struct iov_iter *iter) | |
2085 | { | |
2086 | return file->f_op->write_iter(kio, iter); | |
2087 | } | |
2088 | ||
f74ac015 MS |
2089 | static inline int call_mmap(struct file *file, struct vm_area_struct *vma) |
2090 | { | |
2091 | return file->f_op->mmap(file, vma); | |
2092 | } | |
2093 | ||
1da177e4 LT |
2094 | extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *); |
2095 | extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *); | |
29732938 ZB |
2096 | extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *, |
2097 | loff_t, size_t, unsigned int); | |
f16acc9d DC |
2098 | extern ssize_t generic_copy_file_range(struct file *file_in, loff_t pos_in, |
2099 | struct file *file_out, loff_t pos_out, | |
2100 | size_t len, unsigned int flags); | |
a83ab01a DW |
2101 | extern int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in, |
2102 | struct file *file_out, loff_t pos_out, | |
42ec3d4c DW |
2103 | loff_t *count, |
2104 | unsigned int remap_flags); | |
2105 | extern loff_t do_clone_file_range(struct file *file_in, loff_t pos_in, | |
2106 | struct file *file_out, loff_t pos_out, | |
452ce659 | 2107 | loff_t len, unsigned int remap_flags); |
42ec3d4c DW |
2108 | extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in, |
2109 | struct file *file_out, loff_t pos_out, | |
452ce659 | 2110 | loff_t len, unsigned int remap_flags); |
54dbc151 DW |
2111 | extern int vfs_dedupe_file_range(struct file *file, |
2112 | struct file_dedupe_range *same); | |
42ec3d4c DW |
2113 | extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos, |
2114 | struct file *dst_file, loff_t dst_pos, | |
df365836 | 2115 | loff_t len, unsigned int remap_flags); |
f1825366 | 2116 | |
1da177e4 | 2117 | |
1da177e4 LT |
2118 | struct super_operations { |
2119 | struct inode *(*alloc_inode)(struct super_block *sb); | |
2120 | void (*destroy_inode)(struct inode *); | |
fdb0da89 | 2121 | void (*free_inode)(struct inode *); |
1da177e4 | 2122 | |
aa385729 | 2123 | void (*dirty_inode) (struct inode *, int flags); |
a9185b41 | 2124 | int (*write_inode) (struct inode *, struct writeback_control *wbc); |
45321ac5 | 2125 | int (*drop_inode) (struct inode *); |
be7ce416 | 2126 | void (*evict_inode) (struct inode *); |
1da177e4 | 2127 | void (*put_super) (struct super_block *); |
1da177e4 | 2128 | int (*sync_fs)(struct super_block *sb, int wait); |
48b6bca6 | 2129 | int (*freeze_super) (struct super_block *); |
c4be0c1d | 2130 | int (*freeze_fs) (struct super_block *); |
48b6bca6 | 2131 | int (*thaw_super) (struct super_block *); |
c4be0c1d | 2132 | int (*unfreeze_fs) (struct super_block *); |
726c3342 | 2133 | int (*statfs) (struct dentry *, struct kstatfs *); |
1da177e4 | 2134 | int (*remount_fs) (struct super_block *, int *, char *); |
42faad99 | 2135 | void (*umount_begin) (struct super_block *); |
1da177e4 | 2136 | |
34c80b1d | 2137 | int (*show_options)(struct seq_file *, struct dentry *); |
d861c630 | 2138 | int (*show_devname)(struct seq_file *, struct dentry *); |
a6322de6 | 2139 | int (*show_path)(struct seq_file *, struct dentry *); |
64132379 | 2140 | int (*show_stats)(struct seq_file *, struct dentry *); |
0e51a720 | 2141 | #ifdef CONFIG_QUOTA |
1da177e4 LT |
2142 | ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t); |
2143 | ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t); | |
2d0fa467 | 2144 | struct dquot **(*get_dquots)(struct inode *); |
0e51a720 | 2145 | #endif |
4101b624 VD |
2146 | long (*nr_cached_objects)(struct super_block *, |
2147 | struct shrink_control *); | |
2148 | long (*free_cached_objects)(struct super_block *, | |
2149 | struct shrink_control *); | |
1da177e4 LT |
2150 | }; |
2151 | ||
bbc1096a DH |
2152 | /* |
2153 | * Inode flags - they have no relation to superblock flags now | |
2154 | */ | |
6414e9b0 EB |
2155 | #define S_SYNC (1 << 0) /* Writes are synced at once */ |
2156 | #define S_NOATIME (1 << 1) /* Do not update access times */ | |
2157 | #define S_APPEND (1 << 2) /* Append-only file */ | |
2158 | #define S_IMMUTABLE (1 << 3) /* Immutable file */ | |
2159 | #define S_DEAD (1 << 4) /* removed, but still open directory */ | |
2160 | #define S_NOQUOTA (1 << 5) /* Inode is not counted to quota */ | |
2161 | #define S_DIRSYNC (1 << 6) /* Directory modifications are synchronous */ | |
2162 | #define S_NOCMTIME (1 << 7) /* Do not update file c/mtime */ | |
2163 | #define S_SWAPFILE (1 << 8) /* Do not truncate: swapon got its bmaps */ | |
2164 | #define S_PRIVATE (1 << 9) /* Inode is fs-internal */ | |
2165 | #define S_IMA (1 << 10) /* Inode has an associated IMA struct */ | |
2166 | #define S_AUTOMOUNT (1 << 11) /* Automount/referral quasi-directory */ | |
2167 | #define S_NOSEC (1 << 12) /* no suid or xattr security attributes */ | |
6cd176a5 | 2168 | #ifdef CONFIG_FS_DAX |
6414e9b0 | 2169 | #define S_DAX (1 << 13) /* Direct Access, avoiding the page cache */ |
fbbbad4b | 2170 | #else |
6414e9b0 | 2171 | #define S_DAX 0 /* Make all the DAX code disappear */ |
fbbbad4b | 2172 | #endif |
6414e9b0 EB |
2173 | #define S_ENCRYPTED (1 << 14) /* Encrypted file (using fs/crypto/) */ |
2174 | #define S_CASEFOLD (1 << 15) /* Casefolded file */ | |
2175 | #define S_VERITY (1 << 16) /* Verity file (using fs/verity/) */ | |
1bd9c4e4 | 2176 | #define S_KERNEL_FILE (1 << 17) /* File is in use by the kernel (eg. fs/cachefiles) */ |
bbc1096a DH |
2177 | |
2178 | /* | |
2179 | * Note that nosuid etc flags are inode-specific: setting some file-system | |
2180 | * flags just means all the inodes inherit those flags by default. It might be | |
2181 | * possible to override it selectively if you really wanted to with some | |
2182 | * ioctl() that is not currently implemented. | |
2183 | * | |
e462ec50 | 2184 | * Exception: SB_RDONLY is always applied to the entire file system. |
bbc1096a DH |
2185 | * |
2186 | * Unfortunately, it is possible to change a filesystems flags with it mounted | |
2187 | * with files in use. This means that all of the inodes will not have their | |
2188 | * i_flags updated. Hence, i_flags no longer inherit the superblock mount | |
2189 | * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org | |
2190 | */ | |
2191 | #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg)) | |
2192 | ||
1751e8a6 | 2193 | static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; } |
94e92e7a | 2194 | #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb) |
e462ec50 | 2195 | #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \ |
bbc1096a | 2196 | ((inode)->i_flags & S_SYNC)) |
e462ec50 | 2197 | #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \ |
bbc1096a | 2198 | ((inode)->i_flags & (S_SYNC|S_DIRSYNC))) |
e462ec50 DH |
2199 | #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK) |
2200 | #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME) | |
2201 | #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION) | |
bbc1096a DH |
2202 | |
2203 | #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA) | |
2204 | #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND) | |
2205 | #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE) | |
e462ec50 | 2206 | #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL) |
bbc1096a DH |
2207 | |
2208 | #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD) | |
2209 | #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME) | |
2210 | #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE) | |
2211 | #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE) | |
2212 | #define IS_IMA(inode) ((inode)->i_flags & S_IMA) | |
2213 | #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT) | |
2214 | #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC) | |
fbbbad4b | 2215 | #define IS_DAX(inode) ((inode)->i_flags & S_DAX) |
2ee6a576 | 2216 | #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED) |
b886ee3e | 2217 | #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD) |
5585f2af | 2218 | #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY) |
bbc1096a | 2219 | |
787fb6bc MS |
2220 | #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \ |
2221 | (inode)->i_rdev == WHITEOUT_DEV) | |
2222 | ||
ba73d987 CB |
2223 | static inline bool HAS_UNMAPPED_ID(struct user_namespace *mnt_userns, |
2224 | struct inode *inode) | |
0bd23d09 | 2225 | { |
ba73d987 CB |
2226 | return !uid_valid(i_uid_into_mnt(mnt_userns, inode)) || |
2227 | !gid_valid(i_gid_into_mnt(mnt_userns, inode)); | |
0bd23d09 EB |
2228 | } |
2229 | ||
c75b1d94 JA |
2230 | static inline enum rw_hint file_write_hint(struct file *file) |
2231 | { | |
2232 | if (file->f_write_hint != WRITE_LIFE_NOT_SET) | |
2233 | return file->f_write_hint; | |
2234 | ||
2235 | return file_inode(file)->i_write_hint; | |
2236 | } | |
2237 | ||
2238 | static inline int iocb_flags(struct file *file); | |
2239 | ||
fc28724d AM |
2240 | static inline u16 ki_hint_validate(enum rw_hint hint) |
2241 | { | |
2242 | typeof(((struct kiocb *)0)->ki_hint) max_hint = -1; | |
2243 | ||
2244 | if (hint <= max_hint) | |
2245 | return hint; | |
2246 | return 0; | |
2247 | } | |
2248 | ||
c75b1d94 JA |
2249 | static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp) |
2250 | { | |
2251 | *kiocb = (struct kiocb) { | |
2252 | .ki_filp = filp, | |
2253 | .ki_flags = iocb_flags(filp), | |
fc28724d | 2254 | .ki_hint = ki_hint_validate(file_write_hint(filp)), |
20578bdf | 2255 | .ki_ioprio = get_current_ioprio(), |
c75b1d94 JA |
2256 | }; |
2257 | } | |
2258 | ||
5dcdc43e JX |
2259 | static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src, |
2260 | struct file *filp) | |
2261 | { | |
2262 | *kiocb = (struct kiocb) { | |
2263 | .ki_filp = filp, | |
2264 | .ki_flags = kiocb_src->ki_flags, | |
2265 | .ki_hint = kiocb_src->ki_hint, | |
2266 | .ki_ioprio = kiocb_src->ki_ioprio, | |
2267 | .ki_pos = kiocb_src->ki_pos, | |
2268 | }; | |
2269 | } | |
2270 | ||
1c0eeaf5 | 2271 | /* |
250df6ed | 2272 | * Inode state bits. Protected by inode->i_lock |
1c0eeaf5 | 2273 | * |
1e9d6333 EB |
2274 | * Four bits determine the dirty state of the inode: I_DIRTY_SYNC, |
2275 | * I_DIRTY_DATASYNC, I_DIRTY_PAGES, and I_DIRTY_TIME. | |
1c0eeaf5 JE |
2276 | * |
2277 | * Four bits define the lifetime of an inode. Initially, inodes are I_NEW, | |
2278 | * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at | |
2279 | * various stages of removing an inode. | |
2280 | * | |
eaff8079 | 2281 | * Two bits are used for locking and completion notification, I_NEW and I_SYNC. |
1c0eeaf5 | 2282 | * |
e7ca2d41 | 2283 | * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on |
1e9d6333 EB |
2284 | * fdatasync() (unless I_DIRTY_DATASYNC is also set). |
2285 | * Timestamp updates are the usual cause. | |
2286 | * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of | |
000cb48e JK |
2287 | * these changes separately from I_DIRTY_SYNC so that we |
2288 | * don't have to write inode on fdatasync() when only | |
1e9d6333 | 2289 | * e.g. the timestamps have changed. |
1c0eeaf5 | 2290 | * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean. |
1e9d6333 EB |
2291 | * I_DIRTY_TIME The inode itself only has dirty timestamps, and the |
2292 | * lazytime mount option is enabled. We keep track of this | |
2293 | * separately from I_DIRTY_SYNC in order to implement | |
2294 | * lazytime. This gets cleared if I_DIRTY_INODE | |
2295 | * (I_DIRTY_SYNC and/or I_DIRTY_DATASYNC) gets set. I.e. | |
2296 | * either I_DIRTY_TIME *or* I_DIRTY_INODE can be set in | |
2297 | * i_state, but not both. I_DIRTY_PAGES may still be set. | |
eaff8079 CH |
2298 | * I_NEW Serves as both a mutex and completion notification. |
2299 | * New inodes set I_NEW. If two processes both create | |
2300 | * the same inode, one of them will release its inode and | |
2301 | * wait for I_NEW to be released before returning. | |
2302 | * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can | |
2303 | * also cause waiting on I_NEW, without I_NEW actually | |
2304 | * being set. find_inode() uses this to prevent returning | |
2305 | * nearly-dead inodes. | |
1c0eeaf5 JE |
2306 | * I_WILL_FREE Must be set when calling write_inode_now() if i_count |
2307 | * is zero. I_FREEING must be set when I_WILL_FREE is | |
2308 | * cleared. | |
2309 | * I_FREEING Set when inode is about to be freed but still has dirty | |
2310 | * pages or buffers attached or the inode itself is still | |
2311 | * dirty. | |
dbd5768f JK |
2312 | * I_CLEAR Added by clear_inode(). In this state the inode is |
2313 | * clean and can be destroyed. Inode keeps I_FREEING. | |
1c0eeaf5 JE |
2314 | * |
2315 | * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are | |
2316 | * prohibited for many purposes. iget() must wait for | |
2317 | * the inode to be completely released, then create it | |
2318 | * anew. Other functions will just ignore such inodes, | |
eaff8079 | 2319 | * if appropriate. I_NEW is used for waiting. |
1c0eeaf5 | 2320 | * |
169ebd90 JK |
2321 | * I_SYNC Writeback of inode is running. The bit is set during |
2322 | * data writeback, and cleared with a wakeup on the bit | |
2323 | * address once it is done. The bit is also used to pin | |
2324 | * the inode in memory for flusher thread. | |
1c0eeaf5 | 2325 | * |
bd5fe6c5 CH |
2326 | * I_REFERENCED Marks the inode as recently references on the LRU list. |
2327 | * | |
2328 | * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit(). | |
2329 | * | |
682aa8e1 TH |
2330 | * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to |
2331 | * synchronize competing switching instances and to tell | |
b93b0163 | 2332 | * wb stat updates to grab the i_pages lock. See |
a9519def | 2333 | * inode_switch_wbs_work_fn() for details. |
682aa8e1 | 2334 | * |
ad0af710 AG |
2335 | * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper |
2336 | * and work dirs among overlayfs mounts. | |
2337 | * | |
c2b6d621 AV |
2338 | * I_CREATING New object's inode in the middle of setting up. |
2339 | * | |
dae2f8ed IW |
2340 | * I_DONTCACHE Evict inode as soon as it is not used anymore. |
2341 | * | |
5afced3b JK |
2342 | * I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists. |
2343 | * Used to detect that mark_inode_dirty() should not move | |
2344 | * inode between dirty lists. | |
2345 | * | |
08276bda DH |
2346 | * I_PINNING_FSCACHE_WB Inode is pinning an fscache object for writeback. |
2347 | * | |
1c0eeaf5 | 2348 | * Q: What is the difference between I_WILL_FREE and I_FREEING? |
1c0eeaf5 | 2349 | */ |
9e38d86f NP |
2350 | #define I_DIRTY_SYNC (1 << 0) |
2351 | #define I_DIRTY_DATASYNC (1 << 1) | |
2352 | #define I_DIRTY_PAGES (1 << 2) | |
eaff8079 CH |
2353 | #define __I_NEW 3 |
2354 | #define I_NEW (1 << __I_NEW) | |
9e38d86f NP |
2355 | #define I_WILL_FREE (1 << 4) |
2356 | #define I_FREEING (1 << 5) | |
2357 | #define I_CLEAR (1 << 6) | |
eaff8079 | 2358 | #define __I_SYNC 7 |
1c0eeaf5 | 2359 | #define I_SYNC (1 << __I_SYNC) |
9e38d86f | 2360 | #define I_REFERENCED (1 << 8) |
bd5fe6c5 | 2361 | #define __I_DIO_WAKEUP 9 |
ac74d8d6 | 2362 | #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP) |
f4e0c30c | 2363 | #define I_LINKABLE (1 << 10) |
0ae45f63 | 2364 | #define I_DIRTY_TIME (1 << 11) |
682aa8e1 | 2365 | #define I_WB_SWITCH (1 << 13) |
c2b6d621 AV |
2366 | #define I_OVL_INUSE (1 << 14) |
2367 | #define I_CREATING (1 << 15) | |
dae2f8ed | 2368 | #define I_DONTCACHE (1 << 16) |
5afced3b | 2369 | #define I_SYNC_QUEUED (1 << 17) |
08276bda | 2370 | #define I_PINNING_FSCACHE_WB (1 << 18) |
1da177e4 | 2371 | |
0e11f644 CH |
2372 | #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC) |
2373 | #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES) | |
0ae45f63 | 2374 | #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME) |
1da177e4 LT |
2375 | |
2376 | extern void __mark_inode_dirty(struct inode *, int); | |
2377 | static inline void mark_inode_dirty(struct inode *inode) | |
2378 | { | |
2379 | __mark_inode_dirty(inode, I_DIRTY); | |
2380 | } | |
2381 | ||
2382 | static inline void mark_inode_dirty_sync(struct inode *inode) | |
2383 | { | |
2384 | __mark_inode_dirty(inode, I_DIRTY_SYNC); | |
2385 | } | |
2386 | ||
ed296c6c EB |
2387 | /* |
2388 | * Returns true if the given inode itself only has dirty timestamps (its pages | |
2389 | * may still be dirty) and isn't currently being allocated or freed. | |
2390 | * Filesystems should call this if when writing an inode when lazytime is | |
2391 | * enabled, they want to opportunistically write the timestamps of other inodes | |
2392 | * located very nearby on-disk, e.g. in the same inode block. This returns true | |
2393 | * if the given inode is in need of such an opportunistic update. Requires | |
2394 | * i_lock, or at least later re-checking under i_lock. | |
2395 | */ | |
2396 | static inline bool inode_is_dirtytime_only(struct inode *inode) | |
2397 | { | |
2398 | return (inode->i_state & (I_DIRTY_TIME | I_NEW | | |
2399 | I_FREEING | I_WILL_FREE)) == I_DIRTY_TIME; | |
2400 | } | |
2401 | ||
7ada4db8 MS |
2402 | extern void inc_nlink(struct inode *inode); |
2403 | extern void drop_nlink(struct inode *inode); | |
2404 | extern void clear_nlink(struct inode *inode); | |
2405 | extern void set_nlink(struct inode *inode, unsigned int nlink); | |
d8c76e6f DH |
2406 | |
2407 | static inline void inode_inc_link_count(struct inode *inode) | |
2408 | { | |
2409 | inc_nlink(inode); | |
66d21730 AD |
2410 | mark_inode_dirty(inode); |
2411 | } | |
2412 | ||
9a53c3a7 DH |
2413 | static inline void inode_dec_link_count(struct inode *inode) |
2414 | { | |
2415 | drop_nlink(inode); | |
66d21730 AD |
2416 | mark_inode_dirty(inode); |
2417 | } | |
2418 | ||
c3b2da31 JB |
2419 | enum file_time_flags { |
2420 | S_ATIME = 1, | |
2421 | S_MTIME = 2, | |
2422 | S_CTIME = 4, | |
2423 | S_VERSION = 8, | |
2424 | }; | |
2425 | ||
c6718543 | 2426 | extern bool atime_needs_update(const struct path *, struct inode *); |
badcf2b7 | 2427 | extern void touch_atime(const struct path *); |
e60feb44 JB |
2428 | int inode_update_time(struct inode *inode, struct timespec64 *time, int flags); |
2429 | ||
1da177e4 LT |
2430 | static inline void file_accessed(struct file *file) |
2431 | { | |
2432 | if (!(file->f_flags & O_NOATIME)) | |
68ac1234 | 2433 | touch_atime(&file->f_path); |
1da177e4 LT |
2434 | } |
2435 | ||
e38f7f53 AG |
2436 | extern int file_modified(struct file *file); |
2437 | ||
c3765016 | 2438 | int sync_inode_metadata(struct inode *inode, int wait); |
1da177e4 | 2439 | |
1da177e4 LT |
2440 | struct file_system_type { |
2441 | const char *name; | |
2442 | int fs_flags; | |
bbc1096a DH |
2443 | #define FS_REQUIRES_DEV 1 |
2444 | #define FS_BINARY_MOUNTDATA 2 | |
2445 | #define FS_HAS_SUBTYPE 4 | |
0c55cfc4 | 2446 | #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */ |
0b3b094a | 2447 | #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */ |
a6435940 | 2448 | #define FS_ALLOW_IDMAP 32 /* FS has been updated to handle vfs idmappings. */ |
bbc1096a | 2449 | #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */ |
f3a09c92 | 2450 | int (*init_fs_context)(struct fs_context *); |
d7167b14 | 2451 | const struct fs_parameter_spec *parameters; |
c96e41e9 AV |
2452 | struct dentry *(*mount) (struct file_system_type *, int, |
2453 | const char *, void *); | |
1da177e4 LT |
2454 | void (*kill_sb) (struct super_block *); |
2455 | struct module *owner; | |
2456 | struct file_system_type * next; | |
a5166169 | 2457 | struct hlist_head fs_supers; |
d475fd42 | 2458 | |
cf516249 | 2459 | struct lock_class_key s_lock_key; |
897c6ff9 | 2460 | struct lock_class_key s_umount_key; |
51ee049e | 2461 | struct lock_class_key s_vfs_rename_key; |
5accdf82 | 2462 | struct lock_class_key s_writers_key[SB_FREEZE_LEVELS]; |
d475fd42 PZ |
2463 | |
2464 | struct lock_class_key i_lock_key; | |
2465 | struct lock_class_key i_mutex_key; | |
730633f0 | 2466 | struct lock_class_key invalidate_lock_key; |
14358e6d | 2467 | struct lock_class_key i_mutex_dir_key; |
1da177e4 LT |
2468 | }; |
2469 | ||
7f78e035 EB |
2470 | #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME) |
2471 | ||
152a0836 AV |
2472 | extern struct dentry *mount_bdev(struct file_system_type *fs_type, |
2473 | int flags, const char *dev_name, void *data, | |
2474 | int (*fill_super)(struct super_block *, void *, int)); | |
fc14f2fe AV |
2475 | extern struct dentry *mount_single(struct file_system_type *fs_type, |
2476 | int flags, void *data, | |
2477 | int (*fill_super)(struct super_block *, void *, int)); | |
3c26ff6e AV |
2478 | extern struct dentry *mount_nodev(struct file_system_type *fs_type, |
2479 | int flags, void *data, | |
2480 | int (*fill_super)(struct super_block *, void *, int)); | |
ea441d11 | 2481 | extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path); |
1da177e4 LT |
2482 | void generic_shutdown_super(struct super_block *sb); |
2483 | void kill_block_super(struct super_block *sb); | |
2484 | void kill_anon_super(struct super_block *sb); | |
2485 | void kill_litter_super(struct super_block *sb); | |
2486 | void deactivate_super(struct super_block *sb); | |
74dbbdd7 | 2487 | void deactivate_locked_super(struct super_block *sb); |
1da177e4 | 2488 | int set_anon_super(struct super_block *s, void *data); |
cb50b348 | 2489 | int set_anon_super_fc(struct super_block *s, struct fs_context *fc); |
0ee5dc67 AV |
2490 | int get_anon_bdev(dev_t *); |
2491 | void free_anon_bdev(dev_t); | |
cb50b348 AV |
2492 | struct super_block *sget_fc(struct fs_context *fc, |
2493 | int (*test)(struct super_block *, struct fs_context *), | |
2494 | int (*set)(struct super_block *, struct fs_context *)); | |
1da177e4 LT |
2495 | struct super_block *sget(struct file_system_type *type, |
2496 | int (*test)(struct super_block *,void *), | |
2497 | int (*set)(struct super_block *,void *), | |
9249e17f | 2498 | int flags, void *data); |
1da177e4 LT |
2499 | |
2500 | /* Alas, no aliases. Too much hassle with bringing module.h everywhere */ | |
2501 | #define fops_get(fops) \ | |
2502 | (((fops) && try_module_get((fops)->owner) ? (fops) : NULL)) | |
2503 | #define fops_put(fops) \ | |
2504 | do { if (fops) module_put((fops)->owner); } while(0) | |
e84f9e57 AV |
2505 | /* |
2506 | * This one is to be used *ONLY* from ->open() instances. | |
2507 | * fops must be non-NULL, pinned down *and* module dependencies | |
2508 | * should be sufficient to pin the caller down as well. | |
2509 | */ | |
2510 | #define replace_fops(f, fops) \ | |
2511 | do { \ | |
2512 | struct file *__file = (f); \ | |
2513 | fops_put(__file->f_op); \ | |
2514 | BUG_ON(!(__file->f_op = (fops))); \ | |
2515 | } while(0) | |
1da177e4 LT |
2516 | |
2517 | extern int register_filesystem(struct file_system_type *); | |
2518 | extern int unregister_filesystem(struct file_system_type *); | |
d911b458 | 2519 | extern struct vfsmount *kern_mount(struct file_system_type *); |
423e0ab0 | 2520 | extern void kern_unmount(struct vfsmount *mnt); |
1da177e4 LT |
2521 | extern int may_umount_tree(struct vfsmount *); |
2522 | extern int may_umount(struct vfsmount *); | |
5e6123f3 SL |
2523 | extern long do_mount(const char *, const char __user *, |
2524 | const char *, unsigned long, void *); | |
ca71cf71 | 2525 | extern struct vfsmount *collect_mounts(const struct path *); |
8aec0809 | 2526 | extern void drop_collected_mounts(struct vfsmount *); |
1f707137 AV |
2527 | extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *, |
2528 | struct vfsmount *); | |
f0bb5aaf | 2529 | extern int vfs_statfs(const struct path *, struct kstatfs *); |
c8b91acc AV |
2530 | extern int user_statfs(const char __user *, struct kstatfs *); |
2531 | extern int fd_statfs(int, struct kstatfs *); | |
18e9e510 JB |
2532 | extern int freeze_super(struct super_block *super); |
2533 | extern int thaw_super(struct super_block *super); | |
02125a82 | 2534 | extern bool our_mnt(struct vfsmount *mnt); |
fca39346 JK |
2535 | extern __printf(2, 3) |
2536 | int super_setup_bdi_name(struct super_block *sb, char *fmt, ...); | |
2537 | extern int super_setup_bdi(struct super_block *sb); | |
1da177e4 | 2538 | |
ce3b0f8d AV |
2539 | extern int current_umask(void); |
2540 | ||
5a14696c BF |
2541 | extern void ihold(struct inode * inode); |
2542 | extern void iput(struct inode *); | |
95582b00 | 2543 | extern int generic_update_time(struct inode *, struct timespec64 *, int); |
5a14696c | 2544 | |
f87fd4c2 | 2545 | /* /sys/fs */ |
00d26666 | 2546 | extern struct kobject *fs_kobj; |
f87fd4c2 | 2547 | |
09cbfeaf | 2548 | #define MAX_RW_COUNT (INT_MAX & PAGE_MASK) |
bfcd17a6 | 2549 | |
9e8925b6 | 2550 | #ifdef CONFIG_FILE_LOCKING |
1da177e4 LT |
2551 | static inline int break_lease(struct inode *inode, unsigned int mode) |
2552 | { | |
24cbe784 JL |
2553 | /* |
2554 | * Since this check is lockless, we must ensure that any refcounts | |
8116bf4c JL |
2555 | * taken are done before checking i_flctx->flc_lease. Otherwise, we |
2556 | * could end up racing with tasks trying to set a new lease on this | |
2557 | * file. | |
24cbe784 JL |
2558 | */ |
2559 | smp_mb(); | |
8634b51f | 2560 | if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease)) |
df4e8d2c | 2561 | return __break_lease(inode, mode, FL_LEASE); |
1da177e4 LT |
2562 | return 0; |
2563 | } | |
df4e8d2c BF |
2564 | |
2565 | static inline int break_deleg(struct inode *inode, unsigned int mode) | |
2566 | { | |
962bd40b JL |
2567 | /* |
2568 | * Since this check is lockless, we must ensure that any refcounts | |
8116bf4c JL |
2569 | * taken are done before checking i_flctx->flc_lease. Otherwise, we |
2570 | * could end up racing with tasks trying to set a new lease on this | |
2571 | * file. | |
962bd40b JL |
2572 | */ |
2573 | smp_mb(); | |
8634b51f | 2574 | if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease)) |
df4e8d2c | 2575 | return __break_lease(inode, mode, FL_DELEG); |
1da177e4 LT |
2576 | return 0; |
2577 | } | |
df4e8d2c | 2578 | |
5a14696c BF |
2579 | static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode) |
2580 | { | |
2581 | int ret; | |
2582 | ||
2583 | ret = break_deleg(inode, O_WRONLY|O_NONBLOCK); | |
2584 | if (ret == -EWOULDBLOCK && delegated_inode) { | |
2585 | *delegated_inode = inode; | |
2586 | ihold(inode); | |
2587 | } | |
2588 | return ret; | |
2589 | } | |
2590 | ||
2591 | static inline int break_deleg_wait(struct inode **delegated_inode) | |
2592 | { | |
2593 | int ret; | |
2594 | ||
2595 | ret = break_deleg(*delegated_inode, O_WRONLY); | |
2596 | iput(*delegated_inode); | |
2597 | *delegated_inode = NULL; | |
2598 | return ret; | |
2599 | } | |
2600 | ||
11afe9f7 CH |
2601 | static inline int break_layout(struct inode *inode, bool wait) |
2602 | { | |
2603 | smp_mb(); | |
2604 | if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease)) | |
2605 | return __break_lease(inode, | |
2606 | wait ? O_WRONLY : O_WRONLY | O_NONBLOCK, | |
2607 | FL_LAYOUT); | |
2608 | return 0; | |
2609 | } | |
2610 | ||
bfcd17a6 | 2611 | #else /* !CONFIG_FILE_LOCKING */ |
af5df566 SW |
2612 | static inline int break_lease(struct inode *inode, unsigned int mode) |
2613 | { | |
2614 | return 0; | |
2615 | } | |
2616 | ||
df4e8d2c BF |
2617 | static inline int break_deleg(struct inode *inode, unsigned int mode) |
2618 | { | |
2619 | return 0; | |
2620 | } | |
5a14696c BF |
2621 | |
2622 | static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode) | |
2623 | { | |
2624 | return 0; | |
2625 | } | |
2626 | ||
2627 | static inline int break_deleg_wait(struct inode **delegated_inode) | |
2628 | { | |
2629 | BUG(); | |
2630 | return 0; | |
2631 | } | |
2632 | ||
11afe9f7 CH |
2633 | static inline int break_layout(struct inode *inode, bool wait) |
2634 | { | |
2635 | return 0; | |
2636 | } | |
2637 | ||
bfcd17a6 | 2638 | #endif /* CONFIG_FILE_LOCKING */ |
1da177e4 LT |
2639 | |
2640 | /* fs/open.c */ | |
adb5c247 | 2641 | struct audit_names; |
91a27b2a | 2642 | struct filename { |
adb5c247 JL |
2643 | const char *name; /* pointer to actual string */ |
2644 | const __user char *uptr; /* original userland pointer */ | |
55422d0b | 2645 | int refcnt; |
1c949843 | 2646 | struct audit_names *aname; |
fd2f7cb5 | 2647 | const char iname[]; |
91a27b2a | 2648 | }; |
f1fffbd4 | 2649 | static_assert(offsetof(struct filename, iname) % sizeof(long) == 0); |
1da177e4 | 2650 | |
a6435940 CB |
2651 | static inline struct user_namespace *file_mnt_user_ns(struct file *file) |
2652 | { | |
2653 | return mnt_user_ns(file->f_path.mnt); | |
2654 | } | |
bb49e9e7 CB |
2655 | |
2656 | /** | |
2657 | * is_idmapped_mnt - check whether a mount is mapped | |
2658 | * @mnt: the mount to check | |
2659 | * | |
bd303368 CB |
2660 | * If @mnt has an idmapping attached different from the |
2661 | * filesystem's idmapping then @mnt is mapped. | |
bb49e9e7 CB |
2662 | * |
2663 | * Return: true if mount is mapped, false if not. | |
2664 | */ | |
2665 | static inline bool is_idmapped_mnt(const struct vfsmount *mnt) | |
2666 | { | |
bd303368 | 2667 | return mnt_user_ns(mnt) != mnt->mnt_sb->s_user_ns; |
bb49e9e7 CB |
2668 | } |
2669 | ||
7df818b2 | 2670 | extern long vfs_truncate(const struct path *, loff_t); |
643fe55a CB |
2671 | int do_truncate(struct user_namespace *, struct dentry *, loff_t start, |
2672 | unsigned int time_attrs, struct file *filp); | |
72c72bdf | 2673 | extern int vfs_fallocate(struct file *file, int mode, loff_t offset, |
3e63cbb1 | 2674 | loff_t len); |
8e8a1407 | 2675 | extern long do_sys_open(int dfd, const char __user *filename, int flags, |
a218d0fd | 2676 | umode_t mode); |
669abf4e | 2677 | extern struct file *file_open_name(struct filename *, int, umode_t); |
a218d0fd | 2678 | extern struct file *filp_open(const char *, int, umode_t); |
ffb37ca3 | 2679 | extern struct file *file_open_root(const struct path *, |
378c6520 | 2680 | const char *, int, umode_t); |
ffb37ca3 AV |
2681 | static inline struct file *file_open_root_mnt(struct vfsmount *mnt, |
2682 | const char *name, int flags, umode_t mode) | |
2683 | { | |
2684 | return file_open_root(&(struct path){.mnt = mnt, .dentry = mnt->mnt_root}, | |
2685 | name, flags, mode); | |
2686 | } | |
765927b2 | 2687 | extern struct file * dentry_open(const struct path *, int, const struct cred *); |
2abc77af AV |
2688 | extern struct file * open_with_fake_path(const struct path *, int, |
2689 | struct inode*, const struct cred *); | |
19f391eb AV |
2690 | static inline struct file *file_clone_open(struct file *file) |
2691 | { | |
2692 | return dentry_open(&file->f_path, file->f_flags, file->f_cred); | |
2693 | } | |
1da177e4 | 2694 | extern int filp_close(struct file *, fl_owner_t id); |
91a27b2a | 2695 | |
51f39a1f | 2696 | extern struct filename *getname_flags(const char __user *, int, int *); |
8228e2c3 | 2697 | extern struct filename *getname_uflags(const char __user *, int); |
91a27b2a | 2698 | extern struct filename *getname(const char __user *); |
c4ad8f98 | 2699 | extern struct filename *getname_kernel(const char *); |
55422d0b | 2700 | extern void putname(struct filename *name); |
91a27b2a | 2701 | |
30d90494 | 2702 | extern int finish_open(struct file *file, struct dentry *dentry, |
be12af3e | 2703 | int (*open)(struct inode *, struct file *)); |
e45198a6 | 2704 | extern int finish_no_open(struct file *file, struct dentry *dentry); |
1da177e4 LT |
2705 | |
2706 | /* fs/dcache.c */ | |
2707 | extern void __init vfs_caches_init_early(void); | |
4248b0da | 2708 | extern void __init vfs_caches_init(void); |
1da177e4 | 2709 | |
b86c089b CL |
2710 | extern struct kmem_cache *names_cachep; |
2711 | ||
a608ca21 | 2712 | #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL) |
3446a8aa | 2713 | #define __putname(name) kmem_cache_free(names_cachep, (void *)(name)) |
1da177e4 | 2714 | |
a212b105 | 2715 | extern struct super_block *blockdev_superblock; |
a212b105 TH |
2716 | static inline bool sb_is_blkdev_sb(struct super_block *sb) |
2717 | { | |
dd0dca22 | 2718 | return IS_ENABLED(CONFIG_BLOCK) && sb == blockdev_superblock; |
47e4491b | 2719 | } |
5c0d6b60 | 2720 | |
b818f09e | 2721 | void emergency_thaw_all(void); |
60b0680f | 2722 | extern int sync_filesystem(struct super_block *); |
9361401e | 2723 | extern const struct file_operations def_blk_fops; |
4b6f5d20 | 2724 | extern const struct file_operations def_chr_fops; |
1da177e4 LT |
2725 | |
2726 | /* fs/char_dev.c */ | |
8a932f73 | 2727 | #define CHRDEV_MAJOR_MAX 512 |
49db08c3 LW |
2728 | /* Marks the bottom of the first segment of free char majors */ |
2729 | #define CHRDEV_MAJOR_DYN_END 234 | |
a5d31a3f LG |
2730 | /* Marks the top and bottom of the second segment of free char majors */ |
2731 | #define CHRDEV_MAJOR_DYN_EXT_START 511 | |
2732 | #define CHRDEV_MAJOR_DYN_EXT_END 384 | |
2733 | ||
1da177e4 LT |
2734 | extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *); |
2735 | extern int register_chrdev_region(dev_t, unsigned, const char *); | |
1905b1bf TH |
2736 | extern int __register_chrdev(unsigned int major, unsigned int baseminor, |
2737 | unsigned int count, const char *name, | |
2738 | const struct file_operations *fops); | |
2739 | extern void __unregister_chrdev(unsigned int major, unsigned int baseminor, | |
2740 | unsigned int count, const char *name); | |
1da177e4 | 2741 | extern void unregister_chrdev_region(dev_t, unsigned); |
68eef3b4 | 2742 | extern void chrdev_show(struct seq_file *,off_t); |
1da177e4 | 2743 | |
1905b1bf TH |
2744 | static inline int register_chrdev(unsigned int major, const char *name, |
2745 | const struct file_operations *fops) | |
2746 | { | |
2747 | return __register_chrdev(major, 0, 256, name, fops); | |
2748 | } | |
2749 | ||
2750 | static inline void unregister_chrdev(unsigned int major, const char *name) | |
2751 | { | |
2752 | __unregister_chrdev(major, 0, 256, name); | |
2753 | } | |
2754 | ||
1da177e4 LT |
2755 | extern void init_special_inode(struct inode *, umode_t, dev_t); |
2756 | ||
2757 | /* Invalid inode operations -- fs/bad_inode.c */ | |
2758 | extern void make_bad_inode(struct inode *); | |
0e3ef1fe | 2759 | extern bool is_bad_inode(struct inode *); |
1da177e4 | 2760 | |
1da177e4 LT |
2761 | unsigned long invalidate_mapping_pages(struct address_space *mapping, |
2762 | pgoff_t start, pgoff_t end); | |
54bc4855 | 2763 | |
eb1d7a65 YS |
2764 | void invalidate_mapping_pagevec(struct address_space *mapping, |
2765 | pgoff_t start, pgoff_t end, | |
2766 | unsigned long *nr_pagevec); | |
2767 | ||
1da177e4 LT |
2768 | static inline void invalidate_remote_inode(struct inode *inode) |
2769 | { | |
2770 | if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) || | |
2771 | S_ISLNK(inode->i_mode)) | |
fc0ecff6 | 2772 | invalidate_mapping_pages(inode->i_mapping, 0, -1); |
1da177e4 LT |
2773 | } |
2774 | extern int invalidate_inode_pages2(struct address_space *mapping); | |
2775 | extern int invalidate_inode_pages2_range(struct address_space *mapping, | |
2776 | pgoff_t start, pgoff_t end); | |
2777 | extern int write_inode_now(struct inode *, int); | |
2778 | extern int filemap_fdatawrite(struct address_space *); | |
2779 | extern int filemap_flush(struct address_space *); | |
76341cab | 2780 | extern int filemap_fdatawait_keep_errors(struct address_space *mapping); |
d3bccb6f JK |
2781 | extern int filemap_fdatawait_range(struct address_space *, loff_t lstart, |
2782 | loff_t lend); | |
aa0bfcd9 RZ |
2783 | extern int filemap_fdatawait_range_keep_errors(struct address_space *mapping, |
2784 | loff_t start_byte, loff_t end_byte); | |
ffb959bb JL |
2785 | |
2786 | static inline int filemap_fdatawait(struct address_space *mapping) | |
2787 | { | |
2788 | return filemap_fdatawait_range(mapping, 0, LLONG_MAX); | |
2789 | } | |
2790 | ||
7fc9e472 GR |
2791 | extern bool filemap_range_has_page(struct address_space *, loff_t lstart, |
2792 | loff_t lend); | |
1da177e4 LT |
2793 | extern int filemap_write_and_wait_range(struct address_space *mapping, |
2794 | loff_t lstart, loff_t lend); | |
ebcf28e1 AM |
2795 | extern int __filemap_fdatawrite_range(struct address_space *mapping, |
2796 | loff_t start, loff_t end, int sync_mode); | |
f4c0a0fd JK |
2797 | extern int filemap_fdatawrite_range(struct address_space *mapping, |
2798 | loff_t start, loff_t end); | |
d72d9e2a | 2799 | extern int filemap_check_errors(struct address_space *mapping); |
5660e13d | 2800 | extern void __filemap_set_wb_err(struct address_space *mapping, int err); |
5a798493 JB |
2801 | int filemap_fdatawrite_wbc(struct address_space *mapping, |
2802 | struct writeback_control *wbc); | |
a823e458 | 2803 | |
ddf8f376 IW |
2804 | static inline int filemap_write_and_wait(struct address_space *mapping) |
2805 | { | |
2806 | return filemap_write_and_wait_range(mapping, 0, LLONG_MAX); | |
2807 | } | |
2808 | ||
a823e458 JL |
2809 | extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart, |
2810 | loff_t lend); | |
5660e13d JL |
2811 | extern int __must_check file_check_and_advance_wb_err(struct file *file); |
2812 | extern int __must_check file_write_and_wait_range(struct file *file, | |
2813 | loff_t start, loff_t end); | |
2814 | ||
a823e458 JL |
2815 | static inline int file_write_and_wait(struct file *file) |
2816 | { | |
2817 | return file_write_and_wait_range(file, 0, LLONG_MAX); | |
2818 | } | |
2819 | ||
5660e13d JL |
2820 | /** |
2821 | * filemap_set_wb_err - set a writeback error on an address_space | |
2822 | * @mapping: mapping in which to set writeback error | |
2823 | * @err: error to be set in mapping | |
2824 | * | |
2825 | * When writeback fails in some way, we must record that error so that | |
2826 | * userspace can be informed when fsync and the like are called. We endeavor | |
2827 | * to report errors on any file that was open at the time of the error. Some | |
2828 | * internal callers also need to know when writeback errors have occurred. | |
2829 | * | |
2830 | * When a writeback error occurs, most filesystems will want to call | |
2831 | * filemap_set_wb_err to record the error in the mapping so that it will be | |
2832 | * automatically reported whenever fsync is called on the file. | |
5660e13d JL |
2833 | */ |
2834 | static inline void filemap_set_wb_err(struct address_space *mapping, int err) | |
2835 | { | |
2836 | /* Fastpath for common case of no error */ | |
2837 | if (unlikely(err)) | |
2838 | __filemap_set_wb_err(mapping, err); | |
2839 | } | |
2840 | ||
2841 | /** | |
c9dff084 | 2842 | * filemap_check_wb_err - has an error occurred since the mark was sampled? |
5660e13d JL |
2843 | * @mapping: mapping to check for writeback errors |
2844 | * @since: previously-sampled errseq_t | |
2845 | * | |
2846 | * Grab the errseq_t value from the mapping, and see if it has changed "since" | |
2847 | * the given value was sampled. | |
2848 | * | |
2849 | * If it has then report the latest error set, otherwise return 0. | |
2850 | */ | |
2851 | static inline int filemap_check_wb_err(struct address_space *mapping, | |
2852 | errseq_t since) | |
2853 | { | |
2854 | return errseq_check(&mapping->wb_err, since); | |
2855 | } | |
2856 | ||
2857 | /** | |
2858 | * filemap_sample_wb_err - sample the current errseq_t to test for later errors | |
2859 | * @mapping: mapping to be sampled | |
2860 | * | |
2861 | * Writeback errors are always reported relative to a particular sample point | |
2862 | * in the past. This function provides those sample points. | |
2863 | */ | |
2864 | static inline errseq_t filemap_sample_wb_err(struct address_space *mapping) | |
2865 | { | |
2866 | return errseq_sample(&mapping->wb_err); | |
2867 | } | |
2868 | ||
735e4ae5 JL |
2869 | /** |
2870 | * file_sample_sb_err - sample the current errseq_t to test for later errors | |
21b9cb34 | 2871 | * @file: file pointer to be sampled |
735e4ae5 JL |
2872 | * |
2873 | * Grab the most current superblock-level errseq_t value for the given | |
2874 | * struct file. | |
2875 | */ | |
2876 | static inline errseq_t file_sample_sb_err(struct file *file) | |
2877 | { | |
2878 | return errseq_sample(&file->f_path.dentry->d_sb->s_wb_err); | |
2879 | } | |
2880 | ||
8018ab05 CH |
2881 | extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end, |
2882 | int datasync); | |
2883 | extern int vfs_fsync(struct file *file, int datasync); | |
e2592217 | 2884 | |
22f96b38 JA |
2885 | extern int sync_file_range(struct file *file, loff_t offset, loff_t nbytes, |
2886 | unsigned int flags); | |
2887 | ||
e2592217 CH |
2888 | /* |
2889 | * Sync the bytes written if this was a synchronous write. Expect ki_pos | |
2890 | * to already be updated for the write, and will return either the amount | |
2891 | * of bytes passed in, or an error if syncing the file failed. | |
2892 | */ | |
2893 | static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count) | |
2894 | { | |
2895 | if (iocb->ki_flags & IOCB_DSYNC) { | |
2896 | int ret = vfs_fsync_range(iocb->ki_filp, | |
2897 | iocb->ki_pos - count, iocb->ki_pos - 1, | |
2898 | (iocb->ki_flags & IOCB_SYNC) ? 0 : 1); | |
2899 | if (ret) | |
2900 | return ret; | |
2901 | } | |
2902 | ||
2903 | return count; | |
d311d79d | 2904 | } |
e2592217 | 2905 | |
1da177e4 LT |
2906 | extern void emergency_sync(void); |
2907 | extern void emergency_remount(void); | |
30460e1e | 2908 | |
9361401e | 2909 | #ifdef CONFIG_BLOCK |
30460e1e CM |
2910 | extern int bmap(struct inode *inode, sector_t *block); |
2911 | #else | |
2912 | static inline int bmap(struct inode *inode, sector_t *block) | |
2913 | { | |
2914 | return -EINVAL; | |
2915 | } | |
9361401e | 2916 | #endif |
30460e1e | 2917 | |
2f221d6f CB |
2918 | int notify_change(struct user_namespace *, struct dentry *, |
2919 | struct iattr *, struct inode **); | |
47291baa CB |
2920 | int inode_permission(struct user_namespace *, struct inode *, int); |
2921 | int generic_permission(struct user_namespace *, struct inode *, int); | |
02f92b38 CB |
2922 | static inline int file_permission(struct file *file, int mask) |
2923 | { | |
47291baa CB |
2924 | return inode_permission(file_mnt_user_ns(file), |
2925 | file_inode(file), mask); | |
02f92b38 CB |
2926 | } |
2927 | static inline int path_permission(const struct path *path, int mask) | |
2928 | { | |
47291baa CB |
2929 | return inode_permission(mnt_user_ns(path->mnt), |
2930 | d_inode(path->dentry), mask); | |
02f92b38 | 2931 | } |
ba73d987 CB |
2932 | int __check_sticky(struct user_namespace *mnt_userns, struct inode *dir, |
2933 | struct inode *inode); | |
1da177e4 | 2934 | |
f696a365 MS |
2935 | static inline bool execute_ok(struct inode *inode) |
2936 | { | |
2937 | return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode); | |
2938 | } | |
2939 | ||
6e3e2c43 AV |
2940 | static inline bool inode_wrong_type(const struct inode *inode, umode_t mode) |
2941 | { | |
2942 | return (inode->i_mode ^ mode) & S_IFMT; | |
2943 | } | |
2944 | ||
8d71db4f AV |
2945 | static inline void file_start_write(struct file *file) |
2946 | { | |
2947 | if (!S_ISREG(file_inode(file)->i_mode)) | |
2948 | return; | |
8a3c84b6 | 2949 | sb_start_write(file_inode(file)->i_sb); |
8d71db4f AV |
2950 | } |
2951 | ||
5ae98f15 JK |
2952 | static inline bool file_start_write_trylock(struct file *file) |
2953 | { | |
2954 | if (!S_ISREG(file_inode(file)->i_mode)) | |
2955 | return true; | |
8a3c84b6 | 2956 | return sb_start_write_trylock(file_inode(file)->i_sb); |
5ae98f15 JK |
2957 | } |
2958 | ||
8d71db4f AV |
2959 | static inline void file_end_write(struct file *file) |
2960 | { | |
2961 | if (!S_ISREG(file_inode(file)->i_mode)) | |
2962 | return; | |
2963 | __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE); | |
2964 | } | |
bfe219d3 | 2965 | |
07b8ce1e | 2966 | /* |
592ca09b DH |
2967 | * This is used for regular files where some users -- especially the |
2968 | * currently executed binary in a process, previously handled via | |
2969 | * VM_DENYWRITE -- cannot handle concurrent write (and maybe mmap | |
2970 | * read-write shared) accesses. | |
2971 | * | |
07b8ce1e AV |
2972 | * get_write_access() gets write permission for a file. |
2973 | * put_write_access() releases this write permission. | |
592ca09b DH |
2974 | * deny_write_access() denies write access to a file. |
2975 | * allow_write_access() re-enables write access to a file. | |
2976 | * | |
2977 | * The i_writecount field of an inode can have the following values: | |
2978 | * 0: no write access, no denied write access | |
2979 | * < 0: (-i_writecount) users that denied write access to the file. | |
2980 | * > 0: (i_writecount) users that have write access to the file. | |
07b8ce1e AV |
2981 | * |
2982 | * Normally we operate on that counter with atomic_{inc,dec} and it's safe | |
2983 | * except for the cases where we don't hold i_writecount yet. Then we need to | |
2984 | * use {get,deny}_write_access() - these functions check the sign and refuse | |
2985 | * to do the change if sign is wrong. | |
2986 | */ | |
2987 | static inline int get_write_access(struct inode *inode) | |
2988 | { | |
2989 | return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY; | |
2990 | } | |
2991 | static inline int deny_write_access(struct file *file) | |
2992 | { | |
496ad9aa | 2993 | struct inode *inode = file_inode(file); |
07b8ce1e AV |
2994 | return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY; |
2995 | } | |
1da177e4 LT |
2996 | static inline void put_write_access(struct inode * inode) |
2997 | { | |
2998 | atomic_dec(&inode->i_writecount); | |
2999 | } | |
3000 | static inline void allow_write_access(struct file *file) | |
3001 | { | |
3002 | if (file) | |
496ad9aa | 3003 | atomic_inc(&file_inode(file)->i_writecount); |
1da177e4 | 3004 | } |
f1fe29b4 DH |
3005 | static inline bool inode_is_open_for_write(const struct inode *inode) |
3006 | { | |
3007 | return atomic_read(&inode->i_writecount) > 0; | |
3008 | } | |
3009 | ||
387e3746 | 3010 | #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING) |
a5c96ebf MZ |
3011 | static inline void i_readcount_dec(struct inode *inode) |
3012 | { | |
3013 | BUG_ON(!atomic_read(&inode->i_readcount)); | |
3014 | atomic_dec(&inode->i_readcount); | |
3015 | } | |
3016 | static inline void i_readcount_inc(struct inode *inode) | |
3017 | { | |
3018 | atomic_inc(&inode->i_readcount); | |
3019 | } | |
3020 | #else | |
3021 | static inline void i_readcount_dec(struct inode *inode) | |
3022 | { | |
3023 | return; | |
3024 | } | |
3025 | static inline void i_readcount_inc(struct inode *inode) | |
3026 | { | |
3027 | return; | |
3028 | } | |
3029 | #endif | |
ed8cae8b | 3030 | extern int do_pipe_flags(int *, int); |
1da177e4 | 3031 | |
bdd1d2d3 | 3032 | extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *); |
61a707c5 | 3033 | ssize_t __kernel_read(struct file *file, void *buf, size_t count, loff_t *pos); |
e13ec939 | 3034 | extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *); |
73e18f7c | 3035 | extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *); |
1da177e4 LT |
3036 | extern struct file * open_exec(const char *); |
3037 | ||
3038 | /* fs/dcache.c -- generic fs support functions */ | |
a6e5787f | 3039 | extern bool is_subdir(struct dentry *, struct dentry *); |
640eb7e7 | 3040 | extern bool path_is_under(const struct path *, const struct path *); |
1da177e4 | 3041 | |
9bf39ab2 MS |
3042 | extern char *file_path(struct file *, char *, int); |
3043 | ||
1da177e4 LT |
3044 | #include <linux/err.h> |
3045 | ||
3046 | /* needed for stackable file system support */ | |
965c8e59 | 3047 | extern loff_t default_llseek(struct file *file, loff_t offset, int whence); |
1da177e4 | 3048 | |
965c8e59 | 3049 | extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence); |
1da177e4 | 3050 | |
54e34621 | 3051 | extern int inode_init_always(struct super_block *, struct inode *); |
1da177e4 | 3052 | extern void inode_init_once(struct inode *); |
2aa15890 | 3053 | extern void address_space_init_once(struct address_space *mapping); |
1da177e4 LT |
3054 | extern struct inode * igrab(struct inode *); |
3055 | extern ino_t iunique(struct super_block *, ino_t); | |
3056 | extern int inode_needs_sync(struct inode *inode); | |
45321ac5 | 3057 | extern int generic_delete_inode(struct inode *inode); |
9bcb4b73 AV |
3058 | static inline int generic_drop_inode(struct inode *inode) |
3059 | { | |
88149082 | 3060 | return !inode->i_nlink || inode_unhashed(inode); |
9bcb4b73 | 3061 | } |
2c567af4 | 3062 | extern void d_mark_dontcache(struct inode *inode); |
1da177e4 | 3063 | |
88bd5121 AA |
3064 | extern struct inode *ilookup5_nowait(struct super_block *sb, |
3065 | unsigned long hashval, int (*test)(struct inode *, void *), | |
3066 | void *data); | |
1da177e4 LT |
3067 | extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval, |
3068 | int (*test)(struct inode *, void *), void *data); | |
3069 | extern struct inode *ilookup(struct super_block *sb, unsigned long ino); | |
3070 | ||
80ea09a0 MS |
3071 | extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval, |
3072 | int (*test)(struct inode *, void *), | |
3073 | int (*set)(struct inode *, void *), | |
3074 | void *data); | |
1da177e4 LT |
3075 | extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *); |
3076 | extern struct inode * iget_locked(struct super_block *, unsigned long); | |
fe032c42 TT |
3077 | extern struct inode *find_inode_nowait(struct super_block *, |
3078 | unsigned long, | |
3079 | int (*match)(struct inode *, | |
3080 | unsigned long, void *), | |
3081 | void *data); | |
3f19b2ab DH |
3082 | extern struct inode *find_inode_rcu(struct super_block *, unsigned long, |
3083 | int (*)(struct inode *, void *), void *); | |
3084 | extern struct inode *find_inode_by_ino_rcu(struct super_block *, unsigned long); | |
261bca86 AV |
3085 | extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *); |
3086 | extern int insert_inode_locked(struct inode *); | |
e096d0c7 JB |
3087 | #ifdef CONFIG_DEBUG_LOCK_ALLOC |
3088 | extern void lockdep_annotate_inode_mutex_key(struct inode *inode); | |
3089 | #else | |
3090 | static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { }; | |
3091 | #endif | |
1da177e4 | 3092 | extern void unlock_new_inode(struct inode *); |
c2b6d621 | 3093 | extern void discard_new_inode(struct inode *); |
85fe4025 | 3094 | extern unsigned int get_next_ino(void); |
799ea9e9 | 3095 | extern void evict_inodes(struct super_block *sb); |
3e9d80a8 | 3096 | void dump_mapping(const struct address_space *); |
1da177e4 | 3097 | |
e809d5f0 CD |
3098 | /* |
3099 | * Userspace may rely on the the inode number being non-zero. For example, glibc | |
3100 | * simply ignores files with zero i_ino in unlink() and other places. | |
3101 | * | |
3102 | * As an additional complication, if userspace was compiled with | |
3103 | * _FILE_OFFSET_BITS=32 on a 64-bit kernel we'll only end up reading out the | |
3104 | * lower 32 bits, so we need to check that those aren't zero explicitly. With | |
3105 | * _FILE_OFFSET_BITS=64, this may cause some harmless false-negatives, but | |
3106 | * better safe than sorry. | |
3107 | */ | |
3108 | static inline bool is_zero_ino(ino_t ino) | |
3109 | { | |
3110 | return (u32)ino == 0; | |
3111 | } | |
3112 | ||
1da177e4 | 3113 | extern void __iget(struct inode * inode); |
b46980fe | 3114 | extern void iget_failed(struct inode *); |
dbd5768f | 3115 | extern void clear_inode(struct inode *); |
2e00c97e | 3116 | extern void __destroy_inode(struct inode *); |
a209dfc7 ED |
3117 | extern struct inode *new_inode_pseudo(struct super_block *sb); |
3118 | extern struct inode *new_inode(struct super_block *sb); | |
ff0c7d15 | 3119 | extern void free_inode_nonrcu(struct inode *inode); |
01de85e0 | 3120 | extern int should_remove_suid(struct dentry *); |
5fa8e0a1 | 3121 | extern int file_remove_privs(struct file *); |
1da177e4 LT |
3122 | |
3123 | extern void __insert_inode_hash(struct inode *, unsigned long hashval); | |
646ec461 CH |
3124 | static inline void insert_inode_hash(struct inode *inode) |
3125 | { | |
1da177e4 LT |
3126 | __insert_inode_hash(inode, inode->i_ino); |
3127 | } | |
f2ee7abf ED |
3128 | |
3129 | extern void __remove_inode_hash(struct inode *); | |
3130 | static inline void remove_inode_hash(struct inode *inode) | |
3131 | { | |
cbedaac6 | 3132 | if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash)) |
f2ee7abf ED |
3133 | __remove_inode_hash(inode); |
3134 | } | |
3135 | ||
646ec461 | 3136 | extern void inode_sb_list_add(struct inode *inode); |
51b8c1fe | 3137 | extern void inode_add_lru(struct inode *inode); |
1da177e4 | 3138 | |
1da177e4 LT |
3139 | extern int sb_set_blocksize(struct super_block *, int); |
3140 | extern int sb_min_blocksize(struct super_block *, int); | |
3141 | ||
3142 | extern int generic_file_mmap(struct file *, struct vm_area_struct *); | |
3143 | extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *); | |
3309dd04 | 3144 | extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *); |
02e83f46 DW |
3145 | extern int generic_write_check_limits(struct file *file, loff_t pos, |
3146 | loff_t *count); | |
a3171351 | 3147 | extern int generic_file_rw_checks(struct file *file_in, struct file *file_out); |
87fa0f3e CH |
3148 | ssize_t filemap_read(struct kiocb *iocb, struct iov_iter *to, |
3149 | ssize_t already_read); | |
ed978a81 | 3150 | extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *); |
8174202b | 3151 | extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *); |
8174202b | 3152 | extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *); |
1af5bb49 | 3153 | extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *); |
3b93f911 | 3154 | extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t); |
88e6faef | 3155 | |
18e9710e | 3156 | ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos, |
ddef7ed2 | 3157 | rwf_t flags); |
abbb6589 | 3158 | ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos, |
ddef7ed2 | 3159 | rwf_t flags); |
5dcdc43e JX |
3160 | ssize_t vfs_iocb_iter_read(struct file *file, struct kiocb *iocb, |
3161 | struct iov_iter *iter); | |
3162 | ssize_t vfs_iocb_iter_write(struct file *file, struct kiocb *iocb, | |
3163 | struct iov_iter *iter); | |
dbe4e192 | 3164 | |
88e6faef | 3165 | /* fs/splice.c */ |
cbb7e577 | 3166 | extern ssize_t generic_file_splice_read(struct file *, loff_t *, |
88e6faef | 3167 | struct pipe_inode_info *, size_t, unsigned int); |
8d020765 | 3168 | extern ssize_t iter_file_splice_write(struct pipe_inode_info *, |
cbb7e577 | 3169 | struct file *, loff_t *, size_t, unsigned int); |
88e6faef | 3170 | extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe, |
cbb7e577 | 3171 | struct file *out, loff_t *, size_t len, unsigned int flags); |
1c118596 MS |
3172 | extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out, |
3173 | loff_t *opos, size_t len, unsigned int flags); | |
3174 | ||
88e6faef | 3175 | |
1da177e4 LT |
3176 | extern void |
3177 | file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping); | |
965c8e59 AM |
3178 | extern loff_t noop_llseek(struct file *file, loff_t offset, int whence); |
3179 | extern loff_t no_llseek(struct file *file, loff_t offset, int whence); | |
46a1c2c7 | 3180 | extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize); |
965c8e59 | 3181 | extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence); |
5760495a | 3182 | extern loff_t generic_file_llseek_size(struct file *file, loff_t offset, |
965c8e59 | 3183 | int whence, loff_t maxsize, loff_t eof); |
1bf9d14d AV |
3184 | extern loff_t fixed_size_llseek(struct file *file, loff_t offset, |
3185 | int whence, loff_t size); | |
b25472f9 AV |
3186 | extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t); |
3187 | extern loff_t no_seek_end_llseek(struct file *, loff_t, int); | |
1da177e4 LT |
3188 | extern int generic_file_open(struct inode * inode, struct file * filp); |
3189 | extern int nonseekable_open(struct inode * inode, struct file * filp); | |
10dce8af | 3190 | extern int stream_open(struct inode * inode, struct file * filp); |
1da177e4 | 3191 | |
9361401e | 3192 | #ifdef CONFIG_BLOCK |
8a4c1e42 | 3193 | typedef void (dio_submit_t)(struct bio *bio, struct inode *inode, |
facd07b0 | 3194 | loff_t file_offset); |
1da177e4 LT |
3195 | |
3196 | enum { | |
1e431f5c CH |
3197 | /* need locking between buffered and direct access */ |
3198 | DIO_LOCKING = 0x01, | |
3199 | ||
3200 | /* filesystem does not support filling holes */ | |
3201 | DIO_SKIP_HOLES = 0x02, | |
1da177e4 LT |
3202 | }; |
3203 | ||
17f8c842 OS |
3204 | ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode, |
3205 | struct block_device *bdev, struct iov_iter *iter, | |
c8b8e32d | 3206 | get_block_t get_block, |
17f8c842 OS |
3207 | dio_iodone_t end_io, dio_submit_t submit_io, |
3208 | int flags); | |
7bb46a67 | 3209 | |
17f8c842 OS |
3210 | static inline ssize_t blockdev_direct_IO(struct kiocb *iocb, |
3211 | struct inode *inode, | |
c8b8e32d | 3212 | struct iov_iter *iter, |
17f8c842 | 3213 | get_block_t get_block) |
1da177e4 | 3214 | { |
17f8c842 | 3215 | return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter, |
c8b8e32d | 3216 | get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES); |
1da177e4 | 3217 | } |
9361401e | 3218 | #endif |
1da177e4 | 3219 | |
1d59d61f | 3220 | void inode_dio_wait(struct inode *inode); |
fe0f07d0 | 3221 | |
39015399 | 3222 | /** |
fe0f07d0 JA |
3223 | * inode_dio_begin - signal start of a direct I/O requests |
3224 | * @inode: inode the direct I/O happens on | |
3225 | * | |
3226 | * This is called once we've finished processing a direct I/O request, | |
3227 | * and is used to wake up callers waiting for direct I/O to be quiesced. | |
3228 | */ | |
3229 | static inline void inode_dio_begin(struct inode *inode) | |
3230 | { | |
3231 | atomic_inc(&inode->i_dio_count); | |
3232 | } | |
3233 | ||
39015399 | 3234 | /** |
fe0f07d0 JA |
3235 | * inode_dio_end - signal finish of a direct I/O requests |
3236 | * @inode: inode the direct I/O happens on | |
3237 | * | |
3238 | * This is called once we've finished processing a direct I/O request, | |
3239 | * and is used to wake up callers waiting for direct I/O to be quiesced. | |
3240 | */ | |
3241 | static inline void inode_dio_end(struct inode *inode) | |
3242 | { | |
3243 | if (atomic_dec_and_test(&inode->i_dio_count)) | |
3244 | wake_up_bit(&inode->i_state, __I_DIO_WAKEUP); | |
3245 | } | |
1d59d61f | 3246 | |
a92853b6 KK |
3247 | /* |
3248 | * Warn about a page cache invalidation failure diring a direct I/O write. | |
3249 | */ | |
3250 | void dio_warn_stale_pagecache(struct file *filp); | |
3251 | ||
5f16f322 TT |
3252 | extern void inode_set_flags(struct inode *inode, unsigned int flags, |
3253 | unsigned int mask); | |
3254 | ||
4b6f5d20 | 3255 | extern const struct file_operations generic_ro_fops; |
1da177e4 LT |
3256 | |
3257 | #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m)) | |
3258 | ||
5d826c84 | 3259 | extern int readlink_copy(char __user *, int, const char *); |
1da177e4 | 3260 | extern int page_readlink(struct dentry *, char __user *, int); |
fceef393 AV |
3261 | extern const char *page_get_link(struct dentry *, struct inode *, |
3262 | struct delayed_call *); | |
3263 | extern void page_put_link(void *); | |
0adb25d2 | 3264 | extern int __page_symlink(struct inode *inode, const char *symname, int len, |
54566b2c | 3265 | int nofs); |
1da177e4 | 3266 | extern int page_symlink(struct inode *inode, const char *symname, int len); |
c5ef1c42 | 3267 | extern const struct inode_operations page_symlink_inode_operations; |
fceef393 | 3268 | extern void kfree_link(void *); |
0d56a451 | 3269 | void generic_fillattr(struct user_namespace *, struct inode *, struct kstat *); |
4f911138 | 3270 | void generic_fill_statx_attr(struct inode *inode, struct kstat *stat); |
a528d35e DH |
3271 | extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int); |
3272 | extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int); | |
b462707e | 3273 | void __inode_add_bytes(struct inode *inode, loff_t bytes); |
1da177e4 | 3274 | void inode_add_bytes(struct inode *inode, loff_t bytes); |
1c8924eb | 3275 | void __inode_sub_bytes(struct inode *inode, loff_t bytes); |
1da177e4 | 3276 | void inode_sub_bytes(struct inode *inode, loff_t bytes); |
f4a8116a JK |
3277 | static inline loff_t __inode_get_bytes(struct inode *inode) |
3278 | { | |
3279 | return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes; | |
3280 | } | |
1da177e4 LT |
3281 | loff_t inode_get_bytes(struct inode *inode); |
3282 | void inode_set_bytes(struct inode *inode, loff_t bytes); | |
fceef393 AV |
3283 | const char *simple_get_link(struct dentry *, struct inode *, |
3284 | struct delayed_call *); | |
61ba64fc | 3285 | extern const struct inode_operations simple_symlink_inode_operations; |
1da177e4 | 3286 | |
5c0ba4e0 | 3287 | extern int iterate_dir(struct file *, struct dir_context *); |
1da177e4 | 3288 | |
09f1bde4 CH |
3289 | int vfs_fstatat(int dfd, const char __user *filename, struct kstat *stat, |
3290 | int flags); | |
da9aa5d9 | 3291 | int vfs_fstat(int fd, struct kstat *stat); |
a528d35e DH |
3292 | |
3293 | static inline int vfs_stat(const char __user *filename, struct kstat *stat) | |
3294 | { | |
0b2c6693 | 3295 | return vfs_fstatat(AT_FDCWD, filename, stat, 0); |
a528d35e DH |
3296 | } |
3297 | static inline int vfs_lstat(const char __user *name, struct kstat *stat) | |
3298 | { | |
0b2c6693 | 3299 | return vfs_fstatat(AT_FDCWD, name, stat, AT_SYMLINK_NOFOLLOW); |
a528d35e | 3300 | } |
a528d35e | 3301 | |
d60874cd | 3302 | extern const char *vfs_get_link(struct dentry *, struct delayed_call *); |
fd4a0edf | 3303 | extern int vfs_readlink(struct dentry *, char __user *, int); |
1da177e4 | 3304 | |
ee416bcd | 3305 | extern struct file_system_type *get_filesystem(struct file_system_type *fs); |
c18479fe | 3306 | extern void put_filesystem(struct file_system_type *fs); |
1da177e4 LT |
3307 | extern struct file_system_type *get_fs_type(const char *name); |
3308 | extern struct super_block *get_super(struct block_device *); | |
4504230a | 3309 | extern struct super_block *get_active_super(struct block_device *bdev); |
1da177e4 | 3310 | extern void drop_super(struct super_block *sb); |
ba6379f7 | 3311 | extern void drop_super_exclusive(struct super_block *sb); |
01a05b33 | 3312 | extern void iterate_supers(void (*)(struct super_block *, void *), void *); |
43e15cdb AV |
3313 | extern void iterate_supers_type(struct file_system_type *, |
3314 | void (*)(struct super_block *, void *), void *); | |
1da177e4 LT |
3315 | |
3316 | extern int dcache_dir_open(struct inode *, struct file *); | |
3317 | extern int dcache_dir_close(struct inode *, struct file *); | |
3318 | extern loff_t dcache_dir_lseek(struct file *, loff_t, int); | |
5f99f4e7 | 3319 | extern int dcache_readdir(struct file *, struct dir_context *); |
549c7297 CB |
3320 | extern int simple_setattr(struct user_namespace *, struct dentry *, |
3321 | struct iattr *); | |
3322 | extern int simple_getattr(struct user_namespace *, const struct path *, | |
3323 | struct kstat *, u32, unsigned int); | |
726c3342 | 3324 | extern int simple_statfs(struct dentry *, struct kstatfs *); |
20955e89 | 3325 | extern int simple_open(struct inode *inode, struct file *file); |
1da177e4 LT |
3326 | extern int simple_link(struct dentry *, struct inode *, struct dentry *); |
3327 | extern int simple_unlink(struct inode *, struct dentry *); | |
3328 | extern int simple_rmdir(struct inode *, struct dentry *); | |
6429e463 LB |
3329 | extern int simple_rename_exchange(struct inode *old_dir, struct dentry *old_dentry, |
3330 | struct inode *new_dir, struct dentry *new_dentry); | |
549c7297 CB |
3331 | extern int simple_rename(struct user_namespace *, struct inode *, |
3332 | struct dentry *, struct inode *, struct dentry *, | |
3333 | unsigned int); | |
a3d1e7eb AV |
3334 | extern void simple_recursive_removal(struct dentry *, |
3335 | void (*callback)(struct dentry *)); | |
02c24a82 | 3336 | extern int noop_fsync(struct file *, loff_t, loff_t, int); |
f44c7763 DW |
3337 | extern void noop_invalidatepage(struct page *page, unsigned int offset, |
3338 | unsigned int length); | |
3339 | extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter); | |
1da177e4 | 3340 | extern int simple_empty(struct dentry *); |
afddba49 NP |
3341 | extern int simple_write_begin(struct file *file, struct address_space *mapping, |
3342 | loff_t pos, unsigned len, unsigned flags, | |
3343 | struct page **pagep, void **fsdata); | |
c1e3dbe9 | 3344 | extern const struct address_space_operations ram_aops; |
b26d4cd3 | 3345 | extern int always_delete_dentry(const struct dentry *); |
6987843f | 3346 | extern struct inode *alloc_anon_inode(struct super_block *); |
e6f5c789 | 3347 | extern int simple_nosetlease(struct file *, long, struct file_lock **, void **); |
b26d4cd3 | 3348 | extern const struct dentry_operations simple_dentry_operations; |
1da177e4 | 3349 | |
00cd8dd3 | 3350 | extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags); |
1da177e4 | 3351 | extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *); |
4b6f5d20 | 3352 | extern const struct file_operations simple_dir_operations; |
c5ef1c42 | 3353 | extern const struct inode_operations simple_dir_inode_operations; |
fbabfd0f EB |
3354 | extern void make_empty_dir_inode(struct inode *inode); |
3355 | extern bool is_empty_dir_inode(struct inode *inode); | |
cda37124 | 3356 | struct tree_descr { const char *name; const struct file_operations *ops; int mode; }; |
1da177e4 | 3357 | struct dentry *d_alloc_name(struct dentry *, const char *); |
cda37124 EB |
3358 | extern int simple_fill_super(struct super_block *, unsigned long, |
3359 | const struct tree_descr *); | |
1f5ce9e9 | 3360 | extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count); |
1da177e4 LT |
3361 | extern void simple_release_fs(struct vfsmount **mount, int *count); |
3362 | ||
93b07113 AM |
3363 | extern ssize_t simple_read_from_buffer(void __user *to, size_t count, |
3364 | loff_t *ppos, const void *from, size_t available); | |
6a727b43 JS |
3365 | extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos, |
3366 | const void __user *from, size_t count); | |
1da177e4 | 3367 | |
ac13a829 | 3368 | extern int __generic_file_fsync(struct file *, loff_t, loff_t, int); |
02c24a82 | 3369 | extern int generic_file_fsync(struct file *, loff_t, loff_t, int); |
d5aacad5 | 3370 | |
30ca22c7 PL |
3371 | extern int generic_check_addressable(unsigned, u64); |
3372 | ||
608af703 | 3373 | extern void generic_set_encrypted_ci_d_ops(struct dentry *dentry); |
c843843e | 3374 | |
e965f963 | 3375 | #ifdef CONFIG_MIGRATION |
2d1db3b1 | 3376 | extern int buffer_migrate_page(struct address_space *, |
a6bc32b8 MG |
3377 | struct page *, struct page *, |
3378 | enum migrate_mode); | |
89cb0888 JK |
3379 | extern int buffer_migrate_page_norefs(struct address_space *, |
3380 | struct page *, struct page *, | |
3381 | enum migrate_mode); | |
e965f963 CL |
3382 | #else |
3383 | #define buffer_migrate_page NULL | |
89cb0888 | 3384 | #define buffer_migrate_page_norefs NULL |
e965f963 CL |
3385 | #endif |
3386 | ||
7bb698f0 AG |
3387 | int may_setattr(struct user_namespace *mnt_userns, struct inode *inode, |
3388 | unsigned int ia_valid); | |
2f221d6f | 3389 | int setattr_prepare(struct user_namespace *, struct dentry *, struct iattr *); |
25d9e2d1 | 3390 | extern int inode_newsize_ok(const struct inode *, loff_t offset); |
2f221d6f CB |
3391 | void setattr_copy(struct user_namespace *, struct inode *inode, |
3392 | const struct iattr *attr); | |
1da177e4 | 3393 | |
c3b2da31 | 3394 | extern int file_update_time(struct file *file); |
1da177e4 | 3395 | |
f05a3849 | 3396 | static inline bool vma_is_dax(const struct vm_area_struct *vma) |
baabda26 DW |
3397 | { |
3398 | return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host); | |
3399 | } | |
3400 | ||
2bb6d283 DW |
3401 | static inline bool vma_is_fsdax(struct vm_area_struct *vma) |
3402 | { | |
3403 | struct inode *inode; | |
3404 | ||
52650c8b | 3405 | if (!IS_ENABLED(CONFIG_FS_DAX) || !vma->vm_file) |
2bb6d283 DW |
3406 | return false; |
3407 | if (!vma_is_dax(vma)) | |
3408 | return false; | |
3409 | inode = file_inode(vma->vm_file); | |
230f5a89 | 3410 | if (S_ISCHR(inode->i_mode)) |
2bb6d283 DW |
3411 | return false; /* device-dax */ |
3412 | return true; | |
3413 | } | |
3414 | ||
2ba48ce5 AV |
3415 | static inline int iocb_flags(struct file *file) |
3416 | { | |
3417 | int res = 0; | |
3418 | if (file->f_flags & O_APPEND) | |
3419 | res |= IOCB_APPEND; | |
efbe3c24 | 3420 | if (file->f_flags & O_DIRECT) |
2ba48ce5 | 3421 | res |= IOCB_DIRECT; |
dde0c2e7 CH |
3422 | if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host)) |
3423 | res |= IOCB_DSYNC; | |
3424 | if (file->f_flags & __O_SYNC) | |
3425 | res |= IOCB_SYNC; | |
2ba48ce5 AV |
3426 | return res; |
3427 | } | |
3428 | ||
ddef7ed2 | 3429 | static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags) |
fdd2f5b7 | 3430 | { |
1752f0ad PB |
3431 | int kiocb_flags = 0; |
3432 | ||
ce71bfea JA |
3433 | /* make sure there's no overlap between RWF and private IOCB flags */ |
3434 | BUILD_BUG_ON((__force int) RWF_SUPPORTED & IOCB_EVENTFD); | |
3435 | ||
1752f0ad PB |
3436 | if (!flags) |
3437 | return 0; | |
fdd2f5b7 GR |
3438 | if (unlikely(flags & ~RWF_SUPPORTED)) |
3439 | return -EOPNOTSUPP; | |
3440 | ||
b745fafa | 3441 | if (flags & RWF_NOWAIT) { |
91f9943e | 3442 | if (!(ki->ki_filp->f_mode & FMODE_NOWAIT)) |
b745fafa | 3443 | return -EOPNOTSUPP; |
ce71bfea | 3444 | kiocb_flags |= IOCB_NOIO; |
b745fafa | 3445 | } |
ce71bfea | 3446 | kiocb_flags |= (__force int) (flags & RWF_SUPPORTED); |
fdd2f5b7 | 3447 | if (flags & RWF_SYNC) |
ce71bfea | 3448 | kiocb_flags |= IOCB_DSYNC; |
1752f0ad PB |
3449 | |
3450 | ki->ki_flags |= kiocb_flags; | |
fdd2f5b7 GR |
3451 | return 0; |
3452 | } | |
3453 | ||
1da177e4 LT |
3454 | static inline ino_t parent_ino(struct dentry *dentry) |
3455 | { | |
3456 | ino_t res; | |
3457 | ||
b5c84bf6 NP |
3458 | /* |
3459 | * Don't strictly need d_lock here? If the parent ino could change | |
3460 | * then surely we'd have a deeper race in the caller? | |
3461 | */ | |
1da177e4 LT |
3462 | spin_lock(&dentry->d_lock); |
3463 | res = dentry->d_parent->d_inode->i_ino; | |
3464 | spin_unlock(&dentry->d_lock); | |
3465 | return res; | |
3466 | } | |
3467 | ||
1da177e4 LT |
3468 | /* Transaction based IO helpers */ |
3469 | ||
3470 | /* | |
3471 | * An argresp is stored in an allocated page and holds the | |
3472 | * size of the argument or response, along with its content | |
3473 | */ | |
3474 | struct simple_transaction_argresp { | |
3475 | ssize_t size; | |
5e01fdff | 3476 | char data[]; |
1da177e4 LT |
3477 | }; |
3478 | ||
3479 | #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp)) | |
3480 | ||
3481 | char *simple_transaction_get(struct file *file, const char __user *buf, | |
3482 | size_t size); | |
3483 | ssize_t simple_transaction_read(struct file *file, char __user *buf, | |
3484 | size_t size, loff_t *pos); | |
3485 | int simple_transaction_release(struct inode *inode, struct file *file); | |
3486 | ||
76791ab2 | 3487 | void simple_transaction_set(struct file *file, size_t n); |
1da177e4 | 3488 | |
acaefc25 AB |
3489 | /* |
3490 | * simple attribute files | |
3491 | * | |
3492 | * These attributes behave similar to those in sysfs: | |
3493 | * | |
3494 | * Writing to an attribute immediately sets a value, an open file can be | |
3495 | * written to multiple times. | |
3496 | * | |
3497 | * Reading from an attribute creates a buffer from the value that might get | |
3498 | * read with multiple read calls. When the attribute has been read | |
3499 | * completely, no further read calls are possible until the file is opened | |
3500 | * again. | |
3501 | * | |
3502 | * All attributes contain a text representation of a numeric value | |
3503 | * that are accessed with the get() and set() functions. | |
3504 | */ | |
3505 | #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \ | |
3506 | static int __fops ## _open(struct inode *inode, struct file *file) \ | |
3507 | { \ | |
3508 | __simple_attr_check_format(__fmt, 0ull); \ | |
3509 | return simple_attr_open(inode, file, __get, __set, __fmt); \ | |
3510 | } \ | |
828c0950 | 3511 | static const struct file_operations __fops = { \ |
acaefc25 AB |
3512 | .owner = THIS_MODULE, \ |
3513 | .open = __fops ## _open, \ | |
74bedc4d | 3514 | .release = simple_attr_release, \ |
acaefc25 AB |
3515 | .read = simple_attr_read, \ |
3516 | .write = simple_attr_write, \ | |
1ec5584e | 3517 | .llseek = generic_file_llseek, \ |
68be3029 | 3518 | } |
acaefc25 | 3519 | |
b9075fa9 JP |
3520 | static inline __printf(1, 2) |
3521 | void __simple_attr_check_format(const char *fmt, ...) | |
acaefc25 AB |
3522 | { |
3523 | /* don't do anything, just let the compiler check the arguments; */ | |
3524 | } | |
3525 | ||
3526 | int simple_attr_open(struct inode *inode, struct file *file, | |
8b88b099 | 3527 | int (*get)(void *, u64 *), int (*set)(void *, u64), |
acaefc25 | 3528 | const char *fmt); |
74bedc4d | 3529 | int simple_attr_release(struct inode *inode, struct file *file); |
acaefc25 AB |
3530 | ssize_t simple_attr_read(struct file *file, char __user *buf, |
3531 | size_t len, loff_t *ppos); | |
3532 | ssize_t simple_attr_write(struct file *file, const char __user *buf, | |
3533 | size_t len, loff_t *ppos); | |
3534 | ||
4be28540 | 3535 | struct ctl_table; |
8d65af78 | 3536 | int proc_nr_files(struct ctl_table *table, int write, |
32927393 | 3537 | void *buffer, size_t *lenp, loff_t *ppos); |
312d3ca8 | 3538 | int proc_nr_dentry(struct ctl_table *table, int write, |
32927393 | 3539 | void *buffer, size_t *lenp, loff_t *ppos); |
cffbc8aa | 3540 | int proc_nr_inodes(struct ctl_table *table, int write, |
32927393 | 3541 | void *buffer, size_t *lenp, loff_t *ppos); |
6e7c1770 | 3542 | int __init list_bdev_fs_names(char *buf, size_t size); |
62239ac2 | 3543 | |
3cd90ea4 | 3544 | #define __FMODE_EXEC ((__force int) FMODE_EXEC) |
1a44bc8c NK |
3545 | #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY) |
3546 | ||
6d125529 | 3547 | #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE]) |
ecf081d1 | 3548 | #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \ |
1a44bc8c | 3549 | (flag & __FMODE_NONOTIFY))) |
5300990c | 3550 | |
d37177ba | 3551 | static inline bool is_sxid(umode_t mode) |
69b45732 AK |
3552 | { |
3553 | return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP)); | |
3554 | } | |
3555 | ||
ba73d987 CB |
3556 | static inline int check_sticky(struct user_namespace *mnt_userns, |
3557 | struct inode *dir, struct inode *inode) | |
cbdf35bc MS |
3558 | { |
3559 | if (!(dir->i_mode & S_ISVTX)) | |
3560 | return 0; | |
3561 | ||
ba73d987 | 3562 | return __check_sticky(mnt_userns, dir, inode); |
cbdf35bc MS |
3563 | } |
3564 | ||
69b45732 AK |
3565 | static inline void inode_has_no_xattr(struct inode *inode) |
3566 | { | |
e462ec50 | 3567 | if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC)) |
69b45732 AK |
3568 | inode->i_flags |= S_NOSEC; |
3569 | } | |
3570 | ||
a7400222 AV |
3571 | static inline bool is_root_inode(struct inode *inode) |
3572 | { | |
3573 | return inode == inode->i_sb->s_root->d_inode; | |
3574 | } | |
3575 | ||
5f99f4e7 AV |
3576 | static inline bool dir_emit(struct dir_context *ctx, |
3577 | const char *name, int namelen, | |
3578 | u64 ino, unsigned type) | |
3579 | { | |
3580 | return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0; | |
3581 | } | |
3582 | static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx) | |
3583 | { | |
3584 | return ctx->actor(ctx, ".", 1, ctx->pos, | |
3585 | file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0; | |
3586 | } | |
3587 | static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx) | |
3588 | { | |
3589 | return ctx->actor(ctx, "..", 2, ctx->pos, | |
3590 | parent_ino(file->f_path.dentry), DT_DIR) == 0; | |
3591 | } | |
3592 | static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx) | |
3593 | { | |
3594 | if (ctx->pos == 0) { | |
3595 | if (!dir_emit_dot(file, ctx)) | |
3596 | return false; | |
3597 | ctx->pos = 1; | |
3598 | } | |
3599 | if (ctx->pos == 1) { | |
3600 | if (!dir_emit_dotdot(file, ctx)) | |
3601 | return false; | |
3602 | ctx->pos = 2; | |
3603 | } | |
3604 | return true; | |
3605 | } | |
5ded75ec AV |
3606 | static inline bool dir_relax(struct inode *inode) |
3607 | { | |
5955102c AV |
3608 | inode_unlock(inode); |
3609 | inode_lock(inode); | |
5ded75ec AV |
3610 | return !IS_DEADDIR(inode); |
3611 | } | |
5f99f4e7 | 3612 | |
ae05327a AV |
3613 | static inline bool dir_relax_shared(struct inode *inode) |
3614 | { | |
3615 | inode_unlock_shared(inode); | |
3616 | inode_lock_shared(inode); | |
3617 | return !IS_DEADDIR(inode); | |
3618 | } | |
3619 | ||
90f8572b | 3620 | extern bool path_noexec(const struct path *path); |
21fc61c7 | 3621 | extern void inode_nohighmem(struct inode *inode); |
90f8572b | 3622 | |
45cd0faa AG |
3623 | /* mm/fadvise.c */ |
3624 | extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len, | |
3625 | int advice); | |
cf1ea059 JK |
3626 | extern int generic_fadvise(struct file *file, loff_t offset, loff_t len, |
3627 | int advice); | |
45cd0faa | 3628 | |
dc617f29 DW |
3629 | /* |
3630 | * Flush file data before changing attributes. Caller must hold any locks | |
3631 | * required to prevent further writes to this file until we're done setting | |
3632 | * flags. | |
3633 | */ | |
3634 | static inline int inode_drain_writes(struct inode *inode) | |
3635 | { | |
3636 | inode_dio_wait(inode); | |
3637 | return filemap_write_and_wait(inode->i_mapping); | |
3638 | } | |
3639 | ||
1da177e4 | 3640 | #endif /* _LINUX_FS_H */ |