1 /* SPDX-License-Identifier: GPL-2.0 */
2 #ifndef _LINUX_FS_H
3 #define _LINUX_FS_H
4
5 #include <linux/vfsdebug.h>
6 #include <linux/linkage.h>
7 #include <linux/wait_bit.h>
8 #include <linux/kdev_t.h>
9 #include <linux/dcache.h>
10 #include <linux/path.h>
11 #include <linux/stat.h>
12 #include <linux/cache.h>
13 #include <linux/list.h>
14 #include <linux/list_lru.h>
15 #include <linux/llist.h>
16 #include <linux/radix-tree.h>
17 #include <linux/xarray.h>
18 #include <linux/rbtree.h>
19 #include <linux/init.h>
20 #include <linux/pid.h>
21 #include <linux/bug.h>
22 #include <linux/mutex.h>
23 #include <linux/rwsem.h>
24 #include <linux/mm_types.h>
25 #include <linux/capability.h>
26 #include <linux/semaphore.h>
27 #include <linux/fcntl.h>
28 #include <linux/rculist_bl.h>
29 #include <linux/atomic.h>
30 #include <linux/shrinker.h>
31 #include <linux/migrate_mode.h>
32 #include <linux/uidgid.h>
33 #include <linux/lockdep.h>
34 #include <linux/percpu-rwsem.h>
35 #include <linux/workqueue.h>
36 #include <linux/delayed_call.h>
37 #include <linux/uuid.h>
38 #include <linux/errseq.h>
39 #include <linux/ioprio.h>
40 #include <linux/fs_types.h>
41 #include <linux/build_bug.h>
42 #include <linux/stddef.h>
43 #include <linux/mount.h>
44 #include <linux/cred.h>
45 #include <linux/mnt_idmapping.h>
46 #include <linux/slab.h>
47 #include <linux/maple_tree.h>
48 #include <linux/rw_hint.h>
49 #include <linux/file_ref.h>
50 #include <linux/unicode.h>
51
52 #include <asm/byteorder.h>
53 #include <uapi/linux/fs.h>
54
55 struct backing_dev_info;
56 struct bdi_writeback;
57 struct bio;
58 struct io_comp_batch;
59 struct export_operations;
60 struct fiemap_extent_info;
61 struct hd_geometry;
62 struct iovec;
63 struct kiocb;
64 struct kobject;
65 struct pipe_inode_info;
66 struct poll_table_struct;
67 struct kstatfs;
68 struct vm_area_struct;
69 struct vfsmount;
70 struct cred;
71 struct swap_info_struct;
72 struct seq_file;
73 struct workqueue_struct;
74 struct iov_iter;
75 struct fscrypt_inode_info;
76 struct fscrypt_operations;
77 struct fsverity_info;
78 struct fsverity_operations;
79 struct fsnotify_mark_connector;
80 struct fsnotify_sb_info;
81 struct fs_context;
82 struct fs_parameter_spec;
83 struct fileattr;
84 struct iomap_ops;
85
86 extern void __init inode_init(void);
87 extern void __init inode_init_early(void);
88 extern void __init files_init(void);
89 extern void __init files_maxfiles_init(void);
90
91 extern unsigned long get_max_files(void);
92 extern unsigned int sysctl_nr_open;
93
94 typedef __kernel_rwf_t rwf_t;
95
96 struct buffer_head;
97 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
98 struct buffer_head *bh_result, int create);
99 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
100 ssize_t bytes, void *private);
101
102 #define MAY_EXEC 0x00000001
103 #define MAY_WRITE 0x00000002
104 #define MAY_READ 0x00000004
105 #define MAY_APPEND 0x00000008
106 #define MAY_ACCESS 0x00000010
107 #define MAY_OPEN 0x00000020
108 #define MAY_CHDIR 0x00000040
109 /* called from RCU mode, don't block */
110 #define MAY_NOT_BLOCK 0x00000080
111
112 /*
113 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
114 * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
115 */
116
117 /* file is open for reading */
118 #define FMODE_READ ((__force fmode_t)(1 << 0))
119 /* file is open for writing */
120 #define FMODE_WRITE ((__force fmode_t)(1 << 1))
121 /* file is seekable */
122 #define FMODE_LSEEK ((__force fmode_t)(1 << 2))
123 /* file can be accessed using pread */
124 #define FMODE_PREAD ((__force fmode_t)(1 << 3))
125 /* file can be accessed using pwrite */
126 #define FMODE_PWRITE ((__force fmode_t)(1 << 4))
127 /* File is opened for execution with sys_execve / sys_uselib */
128 #define FMODE_EXEC ((__force fmode_t)(1 << 5))
129 /* File writes are restricted (block device specific) */
130 #define FMODE_WRITE_RESTRICTED ((__force fmode_t)(1 << 6))
131 /* File supports atomic writes */
132 #define FMODE_CAN_ATOMIC_WRITE ((__force fmode_t)(1 << 7))
133
134 /* FMODE_* bit 8 */
135
136 /* 32bit hashes as llseek() offset (for directories) */
137 #define FMODE_32BITHASH ((__force fmode_t)(1 << 9))
138 /* 64bit hashes as llseek() offset (for directories) */
139 #define FMODE_64BITHASH ((__force fmode_t)(1 << 10))
140
141 /*
142 * Don't update ctime and mtime.
143 *
144 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
145 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
146 */
147 #define FMODE_NOCMTIME ((__force fmode_t)(1 << 11))
148
149 /* Expect random access pattern */
150 #define FMODE_RANDOM ((__force fmode_t)(1 << 12))
151
152 /* FMODE_* bit 13 */
153
154 /* File is opened with O_PATH; almost nothing can be done with it */
155 #define FMODE_PATH ((__force fmode_t)(1 << 14))
156
157 /* File needs atomic accesses to f_pos */
158 #define FMODE_ATOMIC_POS ((__force fmode_t)(1 << 15))
159 /* Write access to underlying fs */
160 #define FMODE_WRITER ((__force fmode_t)(1 << 16))
161 /* Has read method(s) */
162 #define FMODE_CAN_READ ((__force fmode_t)(1 << 17))
163 /* Has write method(s) */
164 #define FMODE_CAN_WRITE ((__force fmode_t)(1 << 18))
165
166 #define FMODE_OPENED ((__force fmode_t)(1 << 19))
167 #define FMODE_CREATED ((__force fmode_t)(1 << 20))
168
169 /* File is stream-like */
170 #define FMODE_STREAM ((__force fmode_t)(1 << 21))
171
172 /* File supports DIRECT IO */
173 #define FMODE_CAN_ODIRECT ((__force fmode_t)(1 << 22))
174
175 #define FMODE_NOREUSE ((__force fmode_t)(1 << 23))
176
177 /* File is embedded in backing_file object */
178 #define FMODE_BACKING ((__force fmode_t)(1 << 24))
179
180 /*
181 * Together with FMODE_NONOTIFY_PERM defines which fsnotify events shouldn't be
182 * generated (see below)
183 */
184 #define FMODE_NONOTIFY ((__force fmode_t)(1 << 25))
185
186 /*
187 * Together with FMODE_NONOTIFY defines which fsnotify events shouldn't be
188 * generated (see below)
189 */
190 #define FMODE_NONOTIFY_PERM ((__force fmode_t)(1 << 26))
191
192 /* File is capable of returning -EAGAIN if I/O will block */
193 #define FMODE_NOWAIT ((__force fmode_t)(1 << 27))
194
195 /* File represents mount that needs unmounting */
196 #define FMODE_NEED_UNMOUNT ((__force fmode_t)(1 << 28))
197
198 /* File does not contribute to nr_files count */
199 #define FMODE_NOACCOUNT ((__force fmode_t)(1 << 29))
200
201 /*
202 * The two FMODE_NONOTIFY* define which fsnotify events should not be generated
203 * for a file. These are the possible values of (f->f_mode &
204 * FMODE_FSNOTIFY_MASK) and their meaning:
205 *
206 * FMODE_NONOTIFY - suppress all (incl. non-permission) events.
207 * FMODE_NONOTIFY_PERM - suppress permission (incl. pre-content) events.
208 * FMODE_NONOTIFY | FMODE_NONOTIFY_PERM - suppress only pre-content events.
209 */
210 #define FMODE_FSNOTIFY_MASK \
211 (FMODE_NONOTIFY | FMODE_NONOTIFY_PERM)
212
213 #define FMODE_FSNOTIFY_NONE(mode) \
214 ((mode & FMODE_FSNOTIFY_MASK) == FMODE_NONOTIFY)
215 #ifdef CONFIG_FANOTIFY_ACCESS_PERMISSIONS
216 #define FMODE_FSNOTIFY_PERM(mode) \
217 ((mode & FMODE_FSNOTIFY_MASK) == 0 || \
218 (mode & FMODE_FSNOTIFY_MASK) == (FMODE_NONOTIFY | FMODE_NONOTIFY_PERM))
219 #define FMODE_FSNOTIFY_HSM(mode) \
220 ((mode & FMODE_FSNOTIFY_MASK) == 0)
221 #else
222 #define FMODE_FSNOTIFY_PERM(mode) 0
223 #define FMODE_FSNOTIFY_HSM(mode) 0
224 #endif
225
226 /*
227 * Attribute flags. These should be or-ed together to figure out what
228 * has been changed!
229 */
230 #define ATTR_MODE (1 << 0)
231 #define ATTR_UID (1 << 1)
232 #define ATTR_GID (1 << 2)
233 #define ATTR_SIZE (1 << 3)
234 #define ATTR_ATIME (1 << 4)
235 #define ATTR_MTIME (1 << 5)
236 #define ATTR_CTIME (1 << 6)
237 #define ATTR_ATIME_SET (1 << 7)
238 #define ATTR_MTIME_SET (1 << 8)
239 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
240 #define ATTR_KILL_SUID (1 << 11)
241 #define ATTR_KILL_SGID (1 << 12)
242 #define ATTR_FILE (1 << 13)
243 #define ATTR_KILL_PRIV (1 << 14)
244 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
245 #define ATTR_TIMES_SET (1 << 16)
246 #define ATTR_TOUCH (1 << 17)
247 #define ATTR_DELEG (1 << 18) /* Delegated attrs. Don't break write delegations */
248
249 /*
250 * Whiteout is represented by a char device. The following constants define the
251 * mode and device number to use.
252 */
253 #define WHITEOUT_MODE 0
254 #define WHITEOUT_DEV 0
255
256 /*
257 * This is the Inode Attributes structure, used for notify_change(). It
258 * uses the above definitions as flags, to know which values have changed.
259 * Also, in this manner, a Filesystem can look at only the values it cares
260 * about. Basically, these are the attributes that the VFS layer can
261 * request to change from the FS layer.
262 *
263 * Derek Atkins <warlord@MIT.EDU> 94-10-20
264 */
265 struct iattr {
266 unsigned int ia_valid;
267 umode_t ia_mode;
268 /*
269 * The two anonymous unions wrap structures with the same member.
270 *
271 * Filesystems raising FS_ALLOW_IDMAP need to use ia_vfs{g,u}id which
272 * are a dedicated type requiring the filesystem to use the dedicated
273 * helpers. Other filesystem can continue to use ia_{g,u}id until they
274 * have been ported.
275 *
276 * They always contain the same value. In other words FS_ALLOW_IDMAP
277 * pass down the same value on idmapped mounts as they would on regular
278 * mounts.
279 */
280 union {
281 kuid_t ia_uid;
282 vfsuid_t ia_vfsuid;
283 };
284 union {
285 kgid_t ia_gid;
286 vfsgid_t ia_vfsgid;
287 };
288 loff_t ia_size;
289 struct timespec64 ia_atime;
290 struct timespec64 ia_mtime;
291 struct timespec64 ia_ctime;
292
293 /*
294 * Not an attribute, but an auxiliary info for filesystems wanting to
295 * implement an ftruncate() like method. NOTE: filesystem should
296 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
297 */
298 struct file *ia_file;
299 };
300
301 /*
302 * Includes for diskquotas.
303 */
304 #include <linux/quota.h>
305
306 /*
307 * Maximum number of layers of fs stack. Needs to be limited to
308 * prevent kernel stack overflow
309 */
310 #define FILESYSTEM_MAX_STACK_DEPTH 2
311
312 /**
313 * enum positive_aop_returns - aop return codes with specific semantics
314 *
315 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
316 * completed, that the page is still locked, and
317 * should be considered active. The VM uses this hint
318 * to return the page to the active list -- it won't
319 * be a candidate for writeback again in the near
320 * future. Other callers must be careful to unlock
321 * the page if they get this return. Returned by
322 * writepage();
323 *
324 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
325 * unlocked it and the page might have been truncated.
326 * The caller should back up to acquiring a new page and
327 * trying again. The aop will be taking reasonable
328 * precautions not to livelock. If the caller held a page
329 * reference, it should drop it before retrying. Returned
330 * by read_folio().
331 *
332 * address_space_operation functions return these large constants to indicate
333 * special semantics to the caller. These are much larger than the bytes in a
334 * page to allow for functions that return the number of bytes operated on in a
335 * given page.
336 */
337
338 enum positive_aop_returns {
339 AOP_WRITEPAGE_ACTIVATE = 0x80000,
340 AOP_TRUNCATED_PAGE = 0x80001,
341 };
342
343 /*
344 * oh the beauties of C type declarations.
345 */
346 struct page;
347 struct address_space;
348 struct writeback_control;
349 struct readahead_control;
350
351 /* Match RWF_* bits to IOCB bits */
352 #define IOCB_HIPRI (__force int) RWF_HIPRI
353 #define IOCB_DSYNC (__force int) RWF_DSYNC
354 #define IOCB_SYNC (__force int) RWF_SYNC
355 #define IOCB_NOWAIT (__force int) RWF_NOWAIT
356 #define IOCB_APPEND (__force int) RWF_APPEND
357 #define IOCB_ATOMIC (__force int) RWF_ATOMIC
358 #define IOCB_DONTCACHE (__force int) RWF_DONTCACHE
359
360 /* non-RWF related bits - start at 16 */
361 #define IOCB_EVENTFD (1 << 16)
362 #define IOCB_DIRECT (1 << 17)
363 #define IOCB_WRITE (1 << 18)
364 /* iocb->ki_waitq is valid */
365 #define IOCB_WAITQ (1 << 19)
366 #define IOCB_NOIO (1 << 20)
367 /* can use bio alloc cache */
368 #define IOCB_ALLOC_CACHE (1 << 21)
369 /*
370 * IOCB_DIO_CALLER_COMP can be set by the iocb owner, to indicate that the
371 * iocb completion can be passed back to the owner for execution from a safe
372 * context rather than needing to be punted through a workqueue. If this
373 * flag is set, the bio completion handling may set iocb->dio_complete to a
374 * handler function and iocb->private to context information for that handler.
375 * The issuer should call the handler with that context information from task
376 * context to complete the processing of the iocb. Note that while this
377 * provides a task context for the dio_complete() callback, it should only be
378 * used on the completion side for non-IO generating completions. It's fine to
379 * call blocking functions from this callback, but they should not wait for
380 * unrelated IO (like cache flushing, new IO generation, etc).
381 */
382 #define IOCB_DIO_CALLER_COMP (1 << 22)
383 /* kiocb is a read or write operation submitted by fs/aio.c. */
384 #define IOCB_AIO_RW (1 << 23)
385 #define IOCB_HAS_METADATA (1 << 24)
386
387 /* for use in trace events */
388 #define TRACE_IOCB_STRINGS \
389 { IOCB_HIPRI, "HIPRI" }, \
390 { IOCB_DSYNC, "DSYNC" }, \
391 { IOCB_SYNC, "SYNC" }, \
392 { IOCB_NOWAIT, "NOWAIT" }, \
393 { IOCB_APPEND, "APPEND" }, \
394 { IOCB_ATOMIC, "ATOMIC" }, \
395 { IOCB_DONTCACHE, "DONTCACHE" }, \
396 { IOCB_EVENTFD, "EVENTFD"}, \
397 { IOCB_DIRECT, "DIRECT" }, \
398 { IOCB_WRITE, "WRITE" }, \
399 { IOCB_WAITQ, "WAITQ" }, \
400 { IOCB_NOIO, "NOIO" }, \
401 { IOCB_ALLOC_CACHE, "ALLOC_CACHE" }, \
402 { IOCB_DIO_CALLER_COMP, "CALLER_COMP" }
403
404 struct kiocb {
405 struct file *ki_filp;
406 loff_t ki_pos;
407 void (*ki_complete)(struct kiocb *iocb, long ret);
408 void *private;
409 int ki_flags;
410 u16 ki_ioprio; /* See linux/ioprio.h */
411 union {
412 /*
413 * Only used for async buffered reads, where it denotes the
414 * page waitqueue associated with completing the read. Valid
415 * IFF IOCB_WAITQ is set.
416 */
417 struct wait_page_queue *ki_waitq;
418 /*
419 * Can be used for O_DIRECT IO, where the completion handling
420 * is punted back to the issuer of the IO. May only be set
421 * if IOCB_DIO_CALLER_COMP is set by the issuer, and the issuer
422 * must then check for presence of this handler when ki_complete
423 * is invoked. The data passed in to this handler must be
424 * assigned to ->private when dio_complete is assigned.
425 */
426 ssize_t (*dio_complete)(void *data);
427 };
428 };
429
is_sync_kiocb(struct kiocb * kiocb)430 static inline bool is_sync_kiocb(struct kiocb *kiocb)
431 {
432 return kiocb->ki_complete == NULL;
433 }
434
435 struct address_space_operations {
436 int (*writepage)(struct page *page, struct writeback_control *wbc);
437 int (*read_folio)(struct file *, struct folio *);
438
439 /* Write back some dirty pages from this mapping. */
440 int (*writepages)(struct address_space *, struct writeback_control *);
441
442 /* Mark a folio dirty. Return true if this dirtied it */
443 bool (*dirty_folio)(struct address_space *, struct folio *);
444
445 void (*readahead)(struct readahead_control *);
446
447 int (*write_begin)(struct file *, struct address_space *mapping,
448 loff_t pos, unsigned len,
449 struct folio **foliop, void **fsdata);
450 int (*write_end)(struct file *, struct address_space *mapping,
451 loff_t pos, unsigned len, unsigned copied,
452 struct folio *folio, void *fsdata);
453
454 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
455 sector_t (*bmap)(struct address_space *, sector_t);
456 void (*invalidate_folio) (struct folio *, size_t offset, size_t len);
457 bool (*release_folio)(struct folio *, gfp_t);
458 void (*free_folio)(struct folio *folio);
459 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
460 /*
461 * migrate the contents of a folio to the specified target. If
462 * migrate_mode is MIGRATE_ASYNC, it must not block.
463 */
464 int (*migrate_folio)(struct address_space *, struct folio *dst,
465 struct folio *src, enum migrate_mode);
466 int (*launder_folio)(struct folio *);
467 bool (*is_partially_uptodate) (struct folio *, size_t from,
468 size_t count);
469 void (*is_dirty_writeback) (struct folio *, bool *dirty, bool *wb);
470 int (*error_remove_folio)(struct address_space *, struct folio *);
471
472 /* swapfile support */
473 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
474 sector_t *span);
475 void (*swap_deactivate)(struct file *file);
476 int (*swap_rw)(struct kiocb *iocb, struct iov_iter *iter);
477 };
478
479 extern const struct address_space_operations empty_aops;
480
481 /**
482 * struct address_space - Contents of a cacheable, mappable object.
483 * @host: Owner, either the inode or the block_device.
484 * @i_pages: Cached pages.
485 * @invalidate_lock: Guards coherency between page cache contents and
486 * file offset->disk block mappings in the filesystem during invalidates.
487 * It is also used to block modification of page cache contents through
488 * memory mappings.
489 * @gfp_mask: Memory allocation flags to use for allocating pages.
490 * @i_mmap_writable: Number of VM_SHARED, VM_MAYWRITE mappings.
491 * @nr_thps: Number of THPs in the pagecache (non-shmem only).
492 * @i_mmap: Tree of private and shared mappings.
493 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
494 * @nrpages: Number of page entries, protected by the i_pages lock.
495 * @writeback_index: Writeback starts here.
496 * @a_ops: Methods.
497 * @flags: Error bits and flags (AS_*).
498 * @wb_err: The most recent error which has occurred.
499 * @i_private_lock: For use by the owner of the address_space.
500 * @i_private_list: For use by the owner of the address_space.
501 * @i_private_data: For use by the owner of the address_space.
502 */
503 struct address_space {
504 struct inode *host;
505 struct xarray i_pages;
506 struct rw_semaphore invalidate_lock;
507 gfp_t gfp_mask;
508 atomic_t i_mmap_writable;
509 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
510 /* number of thp, only for non-shmem files */
511 atomic_t nr_thps;
512 #endif
513 struct rb_root_cached i_mmap;
514 unsigned long nrpages;
515 pgoff_t writeback_index;
516 const struct address_space_operations *a_ops;
517 unsigned long flags;
518 errseq_t wb_err;
519 spinlock_t i_private_lock;
520 struct list_head i_private_list;
521 struct rw_semaphore i_mmap_rwsem;
522 void * i_private_data;
523 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
524 /*
525 * On most architectures that alignment is already the case; but
526 * must be enforced here for CRIS, to let the least significant bit
527 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
528 */
529
530 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
531 #define PAGECACHE_TAG_DIRTY XA_MARK_0
532 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
533 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
534
535 /*
536 * Returns true if any of the pages in the mapping are marked with the tag.
537 */
mapping_tagged(struct address_space * mapping,xa_mark_t tag)538 static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag)
539 {
540 return xa_marked(&mapping->i_pages, tag);
541 }
542
i_mmap_lock_write(struct address_space * mapping)543 static inline void i_mmap_lock_write(struct address_space *mapping)
544 {
545 down_write(&mapping->i_mmap_rwsem);
546 }
547
i_mmap_trylock_write(struct address_space * mapping)548 static inline int i_mmap_trylock_write(struct address_space *mapping)
549 {
550 return down_write_trylock(&mapping->i_mmap_rwsem);
551 }
552
i_mmap_unlock_write(struct address_space * mapping)553 static inline void i_mmap_unlock_write(struct address_space *mapping)
554 {
555 up_write(&mapping->i_mmap_rwsem);
556 }
557
i_mmap_trylock_read(struct address_space * mapping)558 static inline int i_mmap_trylock_read(struct address_space *mapping)
559 {
560 return down_read_trylock(&mapping->i_mmap_rwsem);
561 }
562
i_mmap_lock_read(struct address_space * mapping)563 static inline void i_mmap_lock_read(struct address_space *mapping)
564 {
565 down_read(&mapping->i_mmap_rwsem);
566 }
567
i_mmap_unlock_read(struct address_space * mapping)568 static inline void i_mmap_unlock_read(struct address_space *mapping)
569 {
570 up_read(&mapping->i_mmap_rwsem);
571 }
572
i_mmap_assert_locked(struct address_space * mapping)573 static inline void i_mmap_assert_locked(struct address_space *mapping)
574 {
575 lockdep_assert_held(&mapping->i_mmap_rwsem);
576 }
577
i_mmap_assert_write_locked(struct address_space * mapping)578 static inline void i_mmap_assert_write_locked(struct address_space *mapping)
579 {
580 lockdep_assert_held_write(&mapping->i_mmap_rwsem);
581 }
582
583 /*
584 * Might pages of this file be mapped into userspace?
585 */
mapping_mapped(struct address_space * mapping)586 static inline int mapping_mapped(struct address_space *mapping)
587 {
588 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
589 }
590
591 /*
592 * Might pages of this file have been modified in userspace?
593 * Note that i_mmap_writable counts all VM_SHARED, VM_MAYWRITE vmas: do_mmap
594 * marks vma as VM_SHARED if it is shared, and the file was opened for
595 * writing i.e. vma may be mprotected writable even if now readonly.
596 *
597 * If i_mmap_writable is negative, no new writable mappings are allowed. You
598 * can only deny writable mappings, if none exists right now.
599 */
mapping_writably_mapped(struct address_space * mapping)600 static inline int mapping_writably_mapped(struct address_space *mapping)
601 {
602 return atomic_read(&mapping->i_mmap_writable) > 0;
603 }
604
mapping_map_writable(struct address_space * mapping)605 static inline int mapping_map_writable(struct address_space *mapping)
606 {
607 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
608 0 : -EPERM;
609 }
610
mapping_unmap_writable(struct address_space * mapping)611 static inline void mapping_unmap_writable(struct address_space *mapping)
612 {
613 atomic_dec(&mapping->i_mmap_writable);
614 }
615
mapping_deny_writable(struct address_space * mapping)616 static inline int mapping_deny_writable(struct address_space *mapping)
617 {
618 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
619 0 : -EBUSY;
620 }
621
mapping_allow_writable(struct address_space * mapping)622 static inline void mapping_allow_writable(struct address_space *mapping)
623 {
624 atomic_inc(&mapping->i_mmap_writable);
625 }
626
627 /*
628 * Use sequence counter to get consistent i_size on 32-bit processors.
629 */
630 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
631 #include <linux/seqlock.h>
632 #define __NEED_I_SIZE_ORDERED
633 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
634 #else
635 #define i_size_ordered_init(inode) do { } while (0)
636 #endif
637
638 struct posix_acl;
639 #define ACL_NOT_CACHED ((void *)(-1))
640 /*
641 * ACL_DONT_CACHE is for stacked filesystems, that rely on underlying fs to
642 * cache the ACL. This also means that ->get_inode_acl() can be called in RCU
643 * mode with the LOOKUP_RCU flag.
644 */
645 #define ACL_DONT_CACHE ((void *)(-3))
646
647 static inline struct posix_acl *
uncached_acl_sentinel(struct task_struct * task)648 uncached_acl_sentinel(struct task_struct *task)
649 {
650 return (void *)task + 1;
651 }
652
653 static inline bool
is_uncached_acl(struct posix_acl * acl)654 is_uncached_acl(struct posix_acl *acl)
655 {
656 return (long)acl & 1;
657 }
658
659 #define IOP_FASTPERM 0x0001
660 #define IOP_LOOKUP 0x0002
661 #define IOP_NOFOLLOW 0x0004
662 #define IOP_XATTR 0x0008
663 #define IOP_DEFAULT_READLINK 0x0010
664 #define IOP_MGTIME 0x0020
665 #define IOP_CACHED_LINK 0x0040
666
667 /*
668 * Keep mostly read-only and often accessed (especially for
669 * the RCU path lookup and 'stat' data) fields at the beginning
670 * of the 'struct inode'
671 */
672 struct inode {
673 umode_t i_mode;
674 unsigned short i_opflags;
675 kuid_t i_uid;
676 kgid_t i_gid;
677 unsigned int i_flags;
678
679 #ifdef CONFIG_FS_POSIX_ACL
680 struct posix_acl *i_acl;
681 struct posix_acl *i_default_acl;
682 #endif
683
684 const struct inode_operations *i_op;
685 struct super_block *i_sb;
686 struct address_space *i_mapping;
687
688 #ifdef CONFIG_SECURITY
689 void *i_security;
690 #endif
691
692 /* Stat data, not accessed from path walking */
693 unsigned long i_ino;
694 /*
695 * Filesystems may only read i_nlink directly. They shall use the
696 * following functions for modification:
697 *
698 * (set|clear|inc|drop)_nlink
699 * inode_(inc|dec)_link_count
700 */
701 union {
702 const unsigned int i_nlink;
703 unsigned int __i_nlink;
704 };
705 dev_t i_rdev;
706 loff_t i_size;
707 time64_t i_atime_sec;
708 time64_t i_mtime_sec;
709 time64_t i_ctime_sec;
710 u32 i_atime_nsec;
711 u32 i_mtime_nsec;
712 u32 i_ctime_nsec;
713 u32 i_generation;
714 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
715 unsigned short i_bytes;
716 u8 i_blkbits;
717 enum rw_hint i_write_hint;
718 blkcnt_t i_blocks;
719
720 #ifdef __NEED_I_SIZE_ORDERED
721 seqcount_t i_size_seqcount;
722 #endif
723
724 /* Misc */
725 u32 i_state;
726 /* 32-bit hole */
727 struct rw_semaphore i_rwsem;
728
729 unsigned long dirtied_when; /* jiffies of first dirtying */
730 unsigned long dirtied_time_when;
731
732 struct hlist_node i_hash;
733 struct list_head i_io_list; /* backing dev IO list */
734 #ifdef CONFIG_CGROUP_WRITEBACK
735 struct bdi_writeback *i_wb; /* the associated cgroup wb */
736
737 /* foreign inode detection, see wbc_detach_inode() */
738 int i_wb_frn_winner;
739 u16 i_wb_frn_avg_time;
740 u16 i_wb_frn_history;
741 #endif
742 struct list_head i_lru; /* inode LRU list */
743 struct list_head i_sb_list;
744 struct list_head i_wb_list; /* backing dev writeback list */
745 union {
746 struct hlist_head i_dentry;
747 struct rcu_head i_rcu;
748 };
749 atomic64_t i_version;
750 atomic64_t i_sequence; /* see futex */
751 atomic_t i_count;
752 atomic_t i_dio_count;
753 atomic_t i_writecount;
754 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
755 atomic_t i_readcount; /* struct files open RO */
756 #endif
757 union {
758 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
759 void (*free_inode)(struct inode *);
760 };
761 struct file_lock_context *i_flctx;
762 struct address_space i_data;
763 union {
764 struct list_head i_devices;
765 int i_linklen;
766 };
767 union {
768 struct pipe_inode_info *i_pipe;
769 struct cdev *i_cdev;
770 char *i_link;
771 unsigned i_dir_seq;
772 };
773
774
775 #ifdef CONFIG_FSNOTIFY
776 __u32 i_fsnotify_mask; /* all events this inode cares about */
777 /* 32-bit hole reserved for expanding i_fsnotify_mask */
778 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
779 #endif
780
781 #ifdef CONFIG_FS_ENCRYPTION
782 struct fscrypt_inode_info *i_crypt_info;
783 #endif
784
785 #ifdef CONFIG_FS_VERITY
786 struct fsverity_info *i_verity_info;
787 #endif
788
789 void *i_private; /* fs or device private pointer */
790 } __randomize_layout;
791
inode_set_cached_link(struct inode * inode,char * link,int linklen)792 static inline void inode_set_cached_link(struct inode *inode, char *link, int linklen)
793 {
794 VFS_WARN_ON_INODE(strlen(link) != linklen, inode);
795 VFS_WARN_ON_INODE(inode->i_opflags & IOP_CACHED_LINK, inode);
796 inode->i_link = link;
797 inode->i_linklen = linklen;
798 inode->i_opflags |= IOP_CACHED_LINK;
799 }
800
801 /*
802 * Get bit address from inode->i_state to use with wait_var_event()
803 * infrastructre.
804 */
805 #define inode_state_wait_address(inode, bit) ((char *)&(inode)->i_state + (bit))
806
807 struct wait_queue_head *inode_bit_waitqueue(struct wait_bit_queue_entry *wqe,
808 struct inode *inode, u32 bit);
809
inode_wake_up_bit(struct inode * inode,u32 bit)810 static inline void inode_wake_up_bit(struct inode *inode, u32 bit)
811 {
812 /* Caller is responsible for correct memory barriers. */
813 wake_up_var(inode_state_wait_address(inode, bit));
814 }
815
816 struct timespec64 timestamp_truncate(struct timespec64 t, struct inode *inode);
817
i_blocksize(const struct inode * node)818 static inline unsigned int i_blocksize(const struct inode *node)
819 {
820 return (1 << node->i_blkbits);
821 }
822
inode_unhashed(struct inode * inode)823 static inline int inode_unhashed(struct inode *inode)
824 {
825 return hlist_unhashed(&inode->i_hash);
826 }
827
828 /*
829 * __mark_inode_dirty expects inodes to be hashed. Since we don't
830 * want special inodes in the fileset inode space, we make them
831 * appear hashed, but do not put on any lists. hlist_del()
832 * will work fine and require no locking.
833 */
inode_fake_hash(struct inode * inode)834 static inline void inode_fake_hash(struct inode *inode)
835 {
836 hlist_add_fake(&inode->i_hash);
837 }
838
839 /*
840 * inode->i_mutex nesting subclasses for the lock validator:
841 *
842 * 0: the object of the current VFS operation
843 * 1: parent
844 * 2: child/target
845 * 3: xattr
846 * 4: second non-directory
847 * 5: second parent (when locking independent directories in rename)
848 *
849 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
850 * non-directories at once.
851 *
852 * The locking order between these classes is
853 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
854 */
855 enum inode_i_mutex_lock_class
856 {
857 I_MUTEX_NORMAL,
858 I_MUTEX_PARENT,
859 I_MUTEX_CHILD,
860 I_MUTEX_XATTR,
861 I_MUTEX_NONDIR2,
862 I_MUTEX_PARENT2,
863 };
864
inode_lock(struct inode * inode)865 static inline void inode_lock(struct inode *inode)
866 {
867 down_write(&inode->i_rwsem);
868 }
869
inode_unlock(struct inode * inode)870 static inline void inode_unlock(struct inode *inode)
871 {
872 up_write(&inode->i_rwsem);
873 }
874
inode_lock_shared(struct inode * inode)875 static inline void inode_lock_shared(struct inode *inode)
876 {
877 down_read(&inode->i_rwsem);
878 }
879
inode_unlock_shared(struct inode * inode)880 static inline void inode_unlock_shared(struct inode *inode)
881 {
882 up_read(&inode->i_rwsem);
883 }
884
inode_trylock(struct inode * inode)885 static inline int inode_trylock(struct inode *inode)
886 {
887 return down_write_trylock(&inode->i_rwsem);
888 }
889
inode_trylock_shared(struct inode * inode)890 static inline int inode_trylock_shared(struct inode *inode)
891 {
892 return down_read_trylock(&inode->i_rwsem);
893 }
894
inode_is_locked(struct inode * inode)895 static inline int inode_is_locked(struct inode *inode)
896 {
897 return rwsem_is_locked(&inode->i_rwsem);
898 }
899
inode_lock_nested(struct inode * inode,unsigned subclass)900 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
901 {
902 down_write_nested(&inode->i_rwsem, subclass);
903 }
904
inode_lock_shared_nested(struct inode * inode,unsigned subclass)905 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
906 {
907 down_read_nested(&inode->i_rwsem, subclass);
908 }
909
filemap_invalidate_lock(struct address_space * mapping)910 static inline void filemap_invalidate_lock(struct address_space *mapping)
911 {
912 down_write(&mapping->invalidate_lock);
913 }
914
filemap_invalidate_unlock(struct address_space * mapping)915 static inline void filemap_invalidate_unlock(struct address_space *mapping)
916 {
917 up_write(&mapping->invalidate_lock);
918 }
919
filemap_invalidate_lock_shared(struct address_space * mapping)920 static inline void filemap_invalidate_lock_shared(struct address_space *mapping)
921 {
922 down_read(&mapping->invalidate_lock);
923 }
924
filemap_invalidate_trylock_shared(struct address_space * mapping)925 static inline int filemap_invalidate_trylock_shared(
926 struct address_space *mapping)
927 {
928 return down_read_trylock(&mapping->invalidate_lock);
929 }
930
filemap_invalidate_unlock_shared(struct address_space * mapping)931 static inline void filemap_invalidate_unlock_shared(
932 struct address_space *mapping)
933 {
934 up_read(&mapping->invalidate_lock);
935 }
936
937 void lock_two_nondirectories(struct inode *, struct inode*);
938 void unlock_two_nondirectories(struct inode *, struct inode*);
939
940 void filemap_invalidate_lock_two(struct address_space *mapping1,
941 struct address_space *mapping2);
942 void filemap_invalidate_unlock_two(struct address_space *mapping1,
943 struct address_space *mapping2);
944
945
946 /*
947 * NOTE: in a 32bit arch with a preemptable kernel and
948 * an UP compile the i_size_read/write must be atomic
949 * with respect to the local cpu (unlike with preempt disabled),
950 * but they don't need to be atomic with respect to other cpus like in
951 * true SMP (so they need either to either locally disable irq around
952 * the read or for example on x86 they can be still implemented as a
953 * cmpxchg8b without the need of the lock prefix). For SMP compiles
954 * and 64bit archs it makes no difference if preempt is enabled or not.
955 */
i_size_read(const struct inode * inode)956 static inline loff_t i_size_read(const struct inode *inode)
957 {
958 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
959 loff_t i_size;
960 unsigned int seq;
961
962 do {
963 seq = read_seqcount_begin(&inode->i_size_seqcount);
964 i_size = inode->i_size;
965 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
966 return i_size;
967 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
968 loff_t i_size;
969
970 preempt_disable();
971 i_size = inode->i_size;
972 preempt_enable();
973 return i_size;
974 #else
975 /* Pairs with smp_store_release() in i_size_write() */
976 return smp_load_acquire(&inode->i_size);
977 #endif
978 }
979
980 /*
981 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
982 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
983 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
984 */
i_size_write(struct inode * inode,loff_t i_size)985 static inline void i_size_write(struct inode *inode, loff_t i_size)
986 {
987 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
988 preempt_disable();
989 write_seqcount_begin(&inode->i_size_seqcount);
990 inode->i_size = i_size;
991 write_seqcount_end(&inode->i_size_seqcount);
992 preempt_enable();
993 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
994 preempt_disable();
995 inode->i_size = i_size;
996 preempt_enable();
997 #else
998 /*
999 * Pairs with smp_load_acquire() in i_size_read() to ensure
1000 * changes related to inode size (such as page contents) are
1001 * visible before we see the changed inode size.
1002 */
1003 smp_store_release(&inode->i_size, i_size);
1004 #endif
1005 }
1006
iminor(const struct inode * inode)1007 static inline unsigned iminor(const struct inode *inode)
1008 {
1009 return MINOR(inode->i_rdev);
1010 }
1011
imajor(const struct inode * inode)1012 static inline unsigned imajor(const struct inode *inode)
1013 {
1014 return MAJOR(inode->i_rdev);
1015 }
1016
1017 struct fown_struct {
1018 struct file *file; /* backpointer for security modules */
1019 rwlock_t lock; /* protects pid, uid, euid fields */
1020 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
1021 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
1022 kuid_t uid, euid; /* uid/euid of process setting the owner */
1023 int signum; /* posix.1b rt signal to be delivered on IO */
1024 };
1025
1026 /**
1027 * struct file_ra_state - Track a file's readahead state.
1028 * @start: Where the most recent readahead started.
1029 * @size: Number of pages read in the most recent readahead.
1030 * @async_size: Numer of pages that were/are not needed immediately
1031 * and so were/are genuinely "ahead". Start next readahead when
1032 * the first of these pages is accessed.
1033 * @ra_pages: Maximum size of a readahead request, copied from the bdi.
1034 * @mmap_miss: How many mmap accesses missed in the page cache.
1035 * @prev_pos: The last byte in the most recent read request.
1036 *
1037 * When this structure is passed to ->readahead(), the "most recent"
1038 * readahead means the current readahead.
1039 */
1040 struct file_ra_state {
1041 pgoff_t start;
1042 unsigned int size;
1043 unsigned int async_size;
1044 unsigned int ra_pages;
1045 unsigned int mmap_miss;
1046 loff_t prev_pos;
1047 };
1048
1049 /*
1050 * Check if @index falls in the readahead windows.
1051 */
ra_has_index(struct file_ra_state * ra,pgoff_t index)1052 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
1053 {
1054 return (index >= ra->start &&
1055 index < ra->start + ra->size);
1056 }
1057
1058 /**
1059 * struct file - Represents a file
1060 * @f_lock: Protects f_ep, f_flags. Must not be taken from IRQ context.
1061 * @f_mode: FMODE_* flags often used in hotpaths
1062 * @f_op: file operations
1063 * @f_mapping: Contents of a cacheable, mappable object.
1064 * @private_data: filesystem or driver specific data
1065 * @f_inode: cached inode
1066 * @f_flags: file flags
1067 * @f_iocb_flags: iocb flags
1068 * @f_cred: stashed credentials of creator/opener
1069 * @f_owner: file owner
1070 * @f_path: path of the file
1071 * @f_pos_lock: lock protecting file position
1072 * @f_pipe: specific to pipes
1073 * @f_pos: file position
1074 * @f_security: LSM security context of this file
1075 * @f_wb_err: writeback error
1076 * @f_sb_err: per sb writeback errors
1077 * @f_ep: link of all epoll hooks for this file
1078 * @f_task_work: task work entry point
1079 * @f_llist: work queue entrypoint
1080 * @f_ra: file's readahead state
1081 * @f_freeptr: Pointer used by SLAB_TYPESAFE_BY_RCU file cache (don't touch.)
1082 * @f_ref: reference count
1083 */
1084 struct file {
1085 spinlock_t f_lock;
1086 fmode_t f_mode;
1087 const struct file_operations *f_op;
1088 struct address_space *f_mapping;
1089 void *private_data;
1090 struct inode *f_inode;
1091 unsigned int f_flags;
1092 unsigned int f_iocb_flags;
1093 const struct cred *f_cred;
1094 struct fown_struct *f_owner;
1095 /* --- cacheline 1 boundary (64 bytes) --- */
1096 struct path f_path;
1097 union {
1098 /* regular files (with FMODE_ATOMIC_POS) and directories */
1099 struct mutex f_pos_lock;
1100 /* pipes */
1101 u64 f_pipe;
1102 };
1103 loff_t f_pos;
1104 #ifdef CONFIG_SECURITY
1105 void *f_security;
1106 #endif
1107 /* --- cacheline 2 boundary (128 bytes) --- */
1108 errseq_t f_wb_err;
1109 errseq_t f_sb_err;
1110 #ifdef CONFIG_EPOLL
1111 struct hlist_head *f_ep;
1112 #endif
1113 union {
1114 struct callback_head f_task_work;
1115 struct llist_node f_llist;
1116 struct file_ra_state f_ra;
1117 freeptr_t f_freeptr;
1118 };
1119 file_ref_t f_ref;
1120 /* --- cacheline 3 boundary (192 bytes) --- */
1121 } __randomize_layout
1122 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
1123
1124 struct file_handle {
1125 __u32 handle_bytes;
1126 int handle_type;
1127 /* file identifier */
1128 unsigned char f_handle[] __counted_by(handle_bytes);
1129 };
1130
get_file(struct file * f)1131 static inline struct file *get_file(struct file *f)
1132 {
1133 file_ref_inc(&f->f_ref);
1134 return f;
1135 }
1136
1137 struct file *get_file_rcu(struct file __rcu **f);
1138 struct file *get_file_active(struct file **f);
1139
1140 #define file_count(f) file_ref_read(&(f)->f_ref)
1141
1142 #define MAX_NON_LFS ((1UL<<31) - 1)
1143
1144 /* Page cache limit. The filesystems should put that into their s_maxbytes
1145 limits, otherwise bad things can happen in VM. */
1146 #if BITS_PER_LONG==32
1147 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
1148 #elif BITS_PER_LONG==64
1149 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
1150 #endif
1151
1152 /* legacy typedef, should eventually be removed */
1153 typedef void *fl_owner_t;
1154
1155 struct file_lock;
1156 struct file_lease;
1157
1158 /* The following constant reflects the upper bound of the file/locking space */
1159 #ifndef OFFSET_MAX
1160 #define OFFSET_MAX type_max(loff_t)
1161 #define OFFT_OFFSET_MAX type_max(off_t)
1162 #endif
1163
1164 int file_f_owner_allocate(struct file *file);
file_f_owner(const struct file * file)1165 static inline struct fown_struct *file_f_owner(const struct file *file)
1166 {
1167 return READ_ONCE(file->f_owner);
1168 }
1169
1170 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1171
file_inode(const struct file * f)1172 static inline struct inode *file_inode(const struct file *f)
1173 {
1174 return f->f_inode;
1175 }
1176
1177 /*
1178 * file_dentry() is a relic from the days that overlayfs was using files with a
1179 * "fake" path, meaning, f_path on overlayfs and f_inode on underlying fs.
1180 * In those days, file_dentry() was needed to get the underlying fs dentry that
1181 * matches f_inode.
1182 * Files with "fake" path should not exist nowadays, so use an assertion to make
1183 * sure that file_dentry() was not papering over filesystem bugs.
1184 */
file_dentry(const struct file * file)1185 static inline struct dentry *file_dentry(const struct file *file)
1186 {
1187 struct dentry *dentry = file->f_path.dentry;
1188
1189 WARN_ON_ONCE(d_inode(dentry) != file_inode(file));
1190 return dentry;
1191 }
1192
1193 struct fasync_struct {
1194 rwlock_t fa_lock;
1195 int magic;
1196 int fa_fd;
1197 struct fasync_struct *fa_next; /* singly linked list */
1198 struct file *fa_file;
1199 struct rcu_head fa_rcu;
1200 };
1201
1202 #define FASYNC_MAGIC 0x4601
1203
1204 /* SMP safe fasync helpers: */
1205 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1206 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1207 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1208 extern struct fasync_struct *fasync_alloc(void);
1209 extern void fasync_free(struct fasync_struct *);
1210
1211 /* can be called from interrupts */
1212 extern void kill_fasync(struct fasync_struct **, int, int);
1213
1214 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1215 extern int f_setown(struct file *filp, int who, int force);
1216 extern void f_delown(struct file *filp);
1217 extern pid_t f_getown(struct file *filp);
1218 extern int send_sigurg(struct file *file);
1219
1220 /*
1221 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1222 * represented in both.
1223 */
1224 #define SB_RDONLY BIT(0) /* Mount read-only */
1225 #define SB_NOSUID BIT(1) /* Ignore suid and sgid bits */
1226 #define SB_NODEV BIT(2) /* Disallow access to device special files */
1227 #define SB_NOEXEC BIT(3) /* Disallow program execution */
1228 #define SB_SYNCHRONOUS BIT(4) /* Writes are synced at once */
1229 #define SB_MANDLOCK BIT(6) /* Allow mandatory locks on an FS */
1230 #define SB_DIRSYNC BIT(7) /* Directory modifications are synchronous */
1231 #define SB_NOATIME BIT(10) /* Do not update access times. */
1232 #define SB_NODIRATIME BIT(11) /* Do not update directory access times */
1233 #define SB_SILENT BIT(15)
1234 #define SB_POSIXACL BIT(16) /* Supports POSIX ACLs */
1235 #define SB_INLINECRYPT BIT(17) /* Use blk-crypto for encrypted files */
1236 #define SB_KERNMOUNT BIT(22) /* this is a kern_mount call */
1237 #define SB_I_VERSION BIT(23) /* Update inode I_version field */
1238 #define SB_LAZYTIME BIT(25) /* Update the on-disk [acm]times lazily */
1239
1240 /* These sb flags are internal to the kernel */
1241 #define SB_DEAD BIT(21)
1242 #define SB_DYING BIT(24)
1243 #define SB_SUBMOUNT BIT(26)
1244 #define SB_FORCE BIT(27)
1245 #define SB_NOSEC BIT(28)
1246 #define SB_BORN BIT(29)
1247 #define SB_ACTIVE BIT(30)
1248 #define SB_NOUSER BIT(31)
1249
1250 /* These flags relate to encoding and casefolding */
1251 #define SB_ENC_STRICT_MODE_FL (1 << 0)
1252 #define SB_ENC_NO_COMPAT_FALLBACK_FL (1 << 1)
1253
1254 #define sb_has_strict_encoding(sb) \
1255 (sb->s_encoding_flags & SB_ENC_STRICT_MODE_FL)
1256
1257 #if IS_ENABLED(CONFIG_UNICODE)
1258 #define sb_no_casefold_compat_fallback(sb) \
1259 (sb->s_encoding_flags & SB_ENC_NO_COMPAT_FALLBACK_FL)
1260 #else
1261 #define sb_no_casefold_compat_fallback(sb) (1)
1262 #endif
1263
1264 /*
1265 * Umount options
1266 */
1267
1268 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1269 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1270 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1271 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1272 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1273
1274 /* sb->s_iflags */
1275 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1276 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1277 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1278 #define SB_I_STABLE_WRITES 0x00000008 /* don't modify blks until WB is done */
1279
1280 /* sb->s_iflags to limit user namespace mounts */
1281 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1282 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1283 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1284 #define SB_I_EVM_HMAC_UNSUPPORTED 0x00000080
1285
1286 #define SB_I_SKIP_SYNC 0x00000100 /* Skip superblock at global sync */
1287 #define SB_I_PERSB_BDI 0x00000200 /* has a per-sb bdi */
1288 #define SB_I_TS_EXPIRY_WARNED 0x00000400 /* warned about timestamp range expiry */
1289 #define SB_I_RETIRED 0x00000800 /* superblock shouldn't be reused */
1290 #define SB_I_NOUMASK 0x00001000 /* VFS does not apply umask */
1291 #define SB_I_NOIDMAP 0x00002000 /* No idmapped mounts on this superblock */
1292 #define SB_I_ALLOW_HSM 0x00004000 /* Allow HSM events on this superblock */
1293
1294 /* Possible states of 'frozen' field */
1295 enum {
1296 SB_UNFROZEN = 0, /* FS is unfrozen */
1297 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1298 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1299 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1300 * internal threads if needed) */
1301 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1302 };
1303
1304 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1305
1306 struct sb_writers {
1307 unsigned short frozen; /* Is sb frozen? */
1308 int freeze_kcount; /* How many kernel freeze requests? */
1309 int freeze_ucount; /* How many userspace freeze requests? */
1310 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1311 };
1312
1313 struct super_block {
1314 struct list_head s_list; /* Keep this first */
1315 dev_t s_dev; /* search index; _not_ kdev_t */
1316 unsigned char s_blocksize_bits;
1317 unsigned long s_blocksize;
1318 loff_t s_maxbytes; /* Max file size */
1319 struct file_system_type *s_type;
1320 const struct super_operations *s_op;
1321 const struct dquot_operations *dq_op;
1322 const struct quotactl_ops *s_qcop;
1323 const struct export_operations *s_export_op;
1324 unsigned long s_flags;
1325 unsigned long s_iflags; /* internal SB_I_* flags */
1326 unsigned long s_magic;
1327 struct dentry *s_root;
1328 struct rw_semaphore s_umount;
1329 int s_count;
1330 atomic_t s_active;
1331 #ifdef CONFIG_SECURITY
1332 void *s_security;
1333 #endif
1334 const struct xattr_handler * const *s_xattr;
1335 #ifdef CONFIG_FS_ENCRYPTION
1336 const struct fscrypt_operations *s_cop;
1337 struct fscrypt_keyring *s_master_keys; /* master crypto keys in use */
1338 #endif
1339 #ifdef CONFIG_FS_VERITY
1340 const struct fsverity_operations *s_vop;
1341 #endif
1342 #if IS_ENABLED(CONFIG_UNICODE)
1343 struct unicode_map *s_encoding;
1344 __u16 s_encoding_flags;
1345 #endif
1346 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1347 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1348 struct block_device *s_bdev; /* can go away once we use an accessor for @s_bdev_file */
1349 struct file *s_bdev_file;
1350 struct backing_dev_info *s_bdi;
1351 struct mtd_info *s_mtd;
1352 struct hlist_node s_instances;
1353 unsigned int s_quota_types; /* Bitmask of supported quota types */
1354 struct quota_info s_dquot; /* Diskquota specific options */
1355
1356 struct sb_writers s_writers;
1357
1358 /*
1359 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1360 * s_fsnotify_info together for cache efficiency. They are frequently
1361 * accessed and rarely modified.
1362 */
1363 void *s_fs_info; /* Filesystem private info */
1364
1365 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1366 u32 s_time_gran;
1367 /* Time limits for c/m/atime in seconds */
1368 time64_t s_time_min;
1369 time64_t s_time_max;
1370 #ifdef CONFIG_FSNOTIFY
1371 u32 s_fsnotify_mask;
1372 struct fsnotify_sb_info *s_fsnotify_info;
1373 #endif
1374
1375 /*
1376 * q: why are s_id and s_sysfs_name not the same? both are human
1377 * readable strings that identify the filesystem
1378 * a: s_id is allowed to change at runtime; it's used in log messages,
1379 * and we want to when a device starts out as single device (s_id is dev
1380 * name) but then a device is hot added and we have to switch to
1381 * identifying it by UUID
1382 * but s_sysfs_name is a handle for programmatic access, and can't
1383 * change at runtime
1384 */
1385 char s_id[32]; /* Informational name */
1386 uuid_t s_uuid; /* UUID */
1387 u8 s_uuid_len; /* Default 16, possibly smaller for weird filesystems */
1388
1389 /* if set, fs shows up under sysfs at /sys/fs/$FSTYP/s_sysfs_name */
1390 char s_sysfs_name[UUID_STRING_LEN + 1];
1391
1392 unsigned int s_max_links;
1393
1394 /*
1395 * The next field is for VFS *only*. No filesystems have any business
1396 * even looking at it. You had been warned.
1397 */
1398 struct mutex s_vfs_rename_mutex; /* Kludge */
1399
1400 /*
1401 * Filesystem subtype. If non-empty the filesystem type field
1402 * in /proc/mounts will be "type.subtype"
1403 */
1404 const char *s_subtype;
1405
1406 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1407
1408 struct shrinker *s_shrink; /* per-sb shrinker handle */
1409
1410 /* Number of inodes with nlink == 0 but still referenced */
1411 atomic_long_t s_remove_count;
1412
1413 /* Read-only state of the superblock is being changed */
1414 int s_readonly_remount;
1415
1416 /* per-sb errseq_t for reporting writeback errors via syncfs */
1417 errseq_t s_wb_err;
1418
1419 /* AIO completions deferred from interrupt context */
1420 struct workqueue_struct *s_dio_done_wq;
1421 struct hlist_head s_pins;
1422
1423 /*
1424 * Owning user namespace and default context in which to
1425 * interpret filesystem uids, gids, quotas, device nodes,
1426 * xattrs and security labels.
1427 */
1428 struct user_namespace *s_user_ns;
1429
1430 /*
1431 * The list_lru structure is essentially just a pointer to a table
1432 * of per-node lru lists, each of which has its own spinlock.
1433 * There is no need to put them into separate cachelines.
1434 */
1435 struct list_lru s_dentry_lru;
1436 struct list_lru s_inode_lru;
1437 struct rcu_head rcu;
1438 struct work_struct destroy_work;
1439
1440 struct mutex s_sync_lock; /* sync serialisation lock */
1441
1442 /*
1443 * Indicates how deep in a filesystem stack this SB is
1444 */
1445 int s_stack_depth;
1446
1447 /* s_inode_list_lock protects s_inodes */
1448 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1449 struct list_head s_inodes; /* all inodes */
1450
1451 spinlock_t s_inode_wblist_lock;
1452 struct list_head s_inodes_wb; /* writeback inodes */
1453 } __randomize_layout;
1454
i_user_ns(const struct inode * inode)1455 static inline struct user_namespace *i_user_ns(const struct inode *inode)
1456 {
1457 return inode->i_sb->s_user_ns;
1458 }
1459
1460 /* Helper functions so that in most cases filesystems will
1461 * not need to deal directly with kuid_t and kgid_t and can
1462 * instead deal with the raw numeric values that are stored
1463 * in the filesystem.
1464 */
i_uid_read(const struct inode * inode)1465 static inline uid_t i_uid_read(const struct inode *inode)
1466 {
1467 return from_kuid(i_user_ns(inode), inode->i_uid);
1468 }
1469
i_gid_read(const struct inode * inode)1470 static inline gid_t i_gid_read(const struct inode *inode)
1471 {
1472 return from_kgid(i_user_ns(inode), inode->i_gid);
1473 }
1474
i_uid_write(struct inode * inode,uid_t uid)1475 static inline void i_uid_write(struct inode *inode, uid_t uid)
1476 {
1477 inode->i_uid = make_kuid(i_user_ns(inode), uid);
1478 }
1479
i_gid_write(struct inode * inode,gid_t gid)1480 static inline void i_gid_write(struct inode *inode, gid_t gid)
1481 {
1482 inode->i_gid = make_kgid(i_user_ns(inode), gid);
1483 }
1484
1485 /**
1486 * i_uid_into_vfsuid - map an inode's i_uid down according to an idmapping
1487 * @idmap: idmap of the mount the inode was found from
1488 * @inode: inode to map
1489 *
1490 * Return: whe inode's i_uid mapped down according to @idmap.
1491 * If the inode's i_uid has no mapping INVALID_VFSUID is returned.
1492 */
i_uid_into_vfsuid(struct mnt_idmap * idmap,const struct inode * inode)1493 static inline vfsuid_t i_uid_into_vfsuid(struct mnt_idmap *idmap,
1494 const struct inode *inode)
1495 {
1496 return make_vfsuid(idmap, i_user_ns(inode), inode->i_uid);
1497 }
1498
1499 /**
1500 * i_uid_needs_update - check whether inode's i_uid needs to be updated
1501 * @idmap: idmap of the mount the inode was found from
1502 * @attr: the new attributes of @inode
1503 * @inode: the inode to update
1504 *
1505 * Check whether the $inode's i_uid field needs to be updated taking idmapped
1506 * mounts into account if the filesystem supports it.
1507 *
1508 * Return: true if @inode's i_uid field needs to be updated, false if not.
1509 */
i_uid_needs_update(struct mnt_idmap * idmap,const struct iattr * attr,const struct inode * inode)1510 static inline bool i_uid_needs_update(struct mnt_idmap *idmap,
1511 const struct iattr *attr,
1512 const struct inode *inode)
1513 {
1514 return ((attr->ia_valid & ATTR_UID) &&
1515 !vfsuid_eq(attr->ia_vfsuid,
1516 i_uid_into_vfsuid(idmap, inode)));
1517 }
1518
1519 /**
1520 * i_uid_update - update @inode's i_uid field
1521 * @idmap: idmap of the mount the inode was found from
1522 * @attr: the new attributes of @inode
1523 * @inode: the inode to update
1524 *
1525 * Safely update @inode's i_uid field translating the vfsuid of any idmapped
1526 * mount into the filesystem kuid.
1527 */
i_uid_update(struct mnt_idmap * idmap,const struct iattr * attr,struct inode * inode)1528 static inline void i_uid_update(struct mnt_idmap *idmap,
1529 const struct iattr *attr,
1530 struct inode *inode)
1531 {
1532 if (attr->ia_valid & ATTR_UID)
1533 inode->i_uid = from_vfsuid(idmap, i_user_ns(inode),
1534 attr->ia_vfsuid);
1535 }
1536
1537 /**
1538 * i_gid_into_vfsgid - map an inode's i_gid down according to an idmapping
1539 * @idmap: idmap of the mount the inode was found from
1540 * @inode: inode to map
1541 *
1542 * Return: the inode's i_gid mapped down according to @idmap.
1543 * If the inode's i_gid has no mapping INVALID_VFSGID is returned.
1544 */
i_gid_into_vfsgid(struct mnt_idmap * idmap,const struct inode * inode)1545 static inline vfsgid_t i_gid_into_vfsgid(struct mnt_idmap *idmap,
1546 const struct inode *inode)
1547 {
1548 return make_vfsgid(idmap, i_user_ns(inode), inode->i_gid);
1549 }
1550
1551 /**
1552 * i_gid_needs_update - check whether inode's i_gid needs to be updated
1553 * @idmap: idmap of the mount the inode was found from
1554 * @attr: the new attributes of @inode
1555 * @inode: the inode to update
1556 *
1557 * Check whether the $inode's i_gid field needs to be updated taking idmapped
1558 * mounts into account if the filesystem supports it.
1559 *
1560 * Return: true if @inode's i_gid field needs to be updated, false if not.
1561 */
i_gid_needs_update(struct mnt_idmap * idmap,const struct iattr * attr,const struct inode * inode)1562 static inline bool i_gid_needs_update(struct mnt_idmap *idmap,
1563 const struct iattr *attr,
1564 const struct inode *inode)
1565 {
1566 return ((attr->ia_valid & ATTR_GID) &&
1567 !vfsgid_eq(attr->ia_vfsgid,
1568 i_gid_into_vfsgid(idmap, inode)));
1569 }
1570
1571 /**
1572 * i_gid_update - update @inode's i_gid field
1573 * @idmap: idmap of the mount the inode was found from
1574 * @attr: the new attributes of @inode
1575 * @inode: the inode to update
1576 *
1577 * Safely update @inode's i_gid field translating the vfsgid of any idmapped
1578 * mount into the filesystem kgid.
1579 */
i_gid_update(struct mnt_idmap * idmap,const struct iattr * attr,struct inode * inode)1580 static inline void i_gid_update(struct mnt_idmap *idmap,
1581 const struct iattr *attr,
1582 struct inode *inode)
1583 {
1584 if (attr->ia_valid & ATTR_GID)
1585 inode->i_gid = from_vfsgid(idmap, i_user_ns(inode),
1586 attr->ia_vfsgid);
1587 }
1588
1589 /**
1590 * inode_fsuid_set - initialize inode's i_uid field with callers fsuid
1591 * @inode: inode to initialize
1592 * @idmap: idmap of the mount the inode was found from
1593 *
1594 * Initialize the i_uid field of @inode. If the inode was found/created via
1595 * an idmapped mount map the caller's fsuid according to @idmap.
1596 */
inode_fsuid_set(struct inode * inode,struct mnt_idmap * idmap)1597 static inline void inode_fsuid_set(struct inode *inode,
1598 struct mnt_idmap *idmap)
1599 {
1600 inode->i_uid = mapped_fsuid(idmap, i_user_ns(inode));
1601 }
1602
1603 /**
1604 * inode_fsgid_set - initialize inode's i_gid field with callers fsgid
1605 * @inode: inode to initialize
1606 * @idmap: idmap of the mount the inode was found from
1607 *
1608 * Initialize the i_gid field of @inode. If the inode was found/created via
1609 * an idmapped mount map the caller's fsgid according to @idmap.
1610 */
inode_fsgid_set(struct inode * inode,struct mnt_idmap * idmap)1611 static inline void inode_fsgid_set(struct inode *inode,
1612 struct mnt_idmap *idmap)
1613 {
1614 inode->i_gid = mapped_fsgid(idmap, i_user_ns(inode));
1615 }
1616
1617 /**
1618 * fsuidgid_has_mapping() - check whether caller's fsuid/fsgid is mapped
1619 * @sb: the superblock we want a mapping in
1620 * @idmap: idmap of the relevant mount
1621 *
1622 * Check whether the caller's fsuid and fsgid have a valid mapping in the
1623 * s_user_ns of the superblock @sb. If the caller is on an idmapped mount map
1624 * the caller's fsuid and fsgid according to the @idmap first.
1625 *
1626 * Return: true if fsuid and fsgid is mapped, false if not.
1627 */
fsuidgid_has_mapping(struct super_block * sb,struct mnt_idmap * idmap)1628 static inline bool fsuidgid_has_mapping(struct super_block *sb,
1629 struct mnt_idmap *idmap)
1630 {
1631 struct user_namespace *fs_userns = sb->s_user_ns;
1632 kuid_t kuid;
1633 kgid_t kgid;
1634
1635 kuid = mapped_fsuid(idmap, fs_userns);
1636 if (!uid_valid(kuid))
1637 return false;
1638 kgid = mapped_fsgid(idmap, fs_userns);
1639 if (!gid_valid(kgid))
1640 return false;
1641 return kuid_has_mapping(fs_userns, kuid) &&
1642 kgid_has_mapping(fs_userns, kgid);
1643 }
1644
1645 struct timespec64 current_time(struct inode *inode);
1646 struct timespec64 inode_set_ctime_current(struct inode *inode);
1647 struct timespec64 inode_set_ctime_deleg(struct inode *inode,
1648 struct timespec64 update);
1649
inode_get_atime_sec(const struct inode * inode)1650 static inline time64_t inode_get_atime_sec(const struct inode *inode)
1651 {
1652 return inode->i_atime_sec;
1653 }
1654
inode_get_atime_nsec(const struct inode * inode)1655 static inline long inode_get_atime_nsec(const struct inode *inode)
1656 {
1657 return inode->i_atime_nsec;
1658 }
1659
inode_get_atime(const struct inode * inode)1660 static inline struct timespec64 inode_get_atime(const struct inode *inode)
1661 {
1662 struct timespec64 ts = { .tv_sec = inode_get_atime_sec(inode),
1663 .tv_nsec = inode_get_atime_nsec(inode) };
1664
1665 return ts;
1666 }
1667
inode_set_atime_to_ts(struct inode * inode,struct timespec64 ts)1668 static inline struct timespec64 inode_set_atime_to_ts(struct inode *inode,
1669 struct timespec64 ts)
1670 {
1671 inode->i_atime_sec = ts.tv_sec;
1672 inode->i_atime_nsec = ts.tv_nsec;
1673 return ts;
1674 }
1675
inode_set_atime(struct inode * inode,time64_t sec,long nsec)1676 static inline struct timespec64 inode_set_atime(struct inode *inode,
1677 time64_t sec, long nsec)
1678 {
1679 struct timespec64 ts = { .tv_sec = sec,
1680 .tv_nsec = nsec };
1681
1682 return inode_set_atime_to_ts(inode, ts);
1683 }
1684
inode_get_mtime_sec(const struct inode * inode)1685 static inline time64_t inode_get_mtime_sec(const struct inode *inode)
1686 {
1687 return inode->i_mtime_sec;
1688 }
1689
inode_get_mtime_nsec(const struct inode * inode)1690 static inline long inode_get_mtime_nsec(const struct inode *inode)
1691 {
1692 return inode->i_mtime_nsec;
1693 }
1694
inode_get_mtime(const struct inode * inode)1695 static inline struct timespec64 inode_get_mtime(const struct inode *inode)
1696 {
1697 struct timespec64 ts = { .tv_sec = inode_get_mtime_sec(inode),
1698 .tv_nsec = inode_get_mtime_nsec(inode) };
1699 return ts;
1700 }
1701
inode_set_mtime_to_ts(struct inode * inode,struct timespec64 ts)1702 static inline struct timespec64 inode_set_mtime_to_ts(struct inode *inode,
1703 struct timespec64 ts)
1704 {
1705 inode->i_mtime_sec = ts.tv_sec;
1706 inode->i_mtime_nsec = ts.tv_nsec;
1707 return ts;
1708 }
1709
inode_set_mtime(struct inode * inode,time64_t sec,long nsec)1710 static inline struct timespec64 inode_set_mtime(struct inode *inode,
1711 time64_t sec, long nsec)
1712 {
1713 struct timespec64 ts = { .tv_sec = sec,
1714 .tv_nsec = nsec };
1715 return inode_set_mtime_to_ts(inode, ts);
1716 }
1717
1718 /*
1719 * Multigrain timestamps
1720 *
1721 * Conditionally use fine-grained ctime and mtime timestamps when there
1722 * are users actively observing them via getattr. The primary use-case
1723 * for this is NFS clients that use the ctime to distinguish between
1724 * different states of the file, and that are often fooled by multiple
1725 * operations that occur in the same coarse-grained timer tick.
1726 */
1727 #define I_CTIME_QUERIED ((u32)BIT(31))
1728
inode_get_ctime_sec(const struct inode * inode)1729 static inline time64_t inode_get_ctime_sec(const struct inode *inode)
1730 {
1731 return inode->i_ctime_sec;
1732 }
1733
inode_get_ctime_nsec(const struct inode * inode)1734 static inline long inode_get_ctime_nsec(const struct inode *inode)
1735 {
1736 return inode->i_ctime_nsec & ~I_CTIME_QUERIED;
1737 }
1738
inode_get_ctime(const struct inode * inode)1739 static inline struct timespec64 inode_get_ctime(const struct inode *inode)
1740 {
1741 struct timespec64 ts = { .tv_sec = inode_get_ctime_sec(inode),
1742 .tv_nsec = inode_get_ctime_nsec(inode) };
1743
1744 return ts;
1745 }
1746
1747 struct timespec64 inode_set_ctime_to_ts(struct inode *inode, struct timespec64 ts);
1748
1749 /**
1750 * inode_set_ctime - set the ctime in the inode
1751 * @inode: inode in which to set the ctime
1752 * @sec: tv_sec value to set
1753 * @nsec: tv_nsec value to set
1754 *
1755 * Set the ctime in @inode to { @sec, @nsec }
1756 */
inode_set_ctime(struct inode * inode,time64_t sec,long nsec)1757 static inline struct timespec64 inode_set_ctime(struct inode *inode,
1758 time64_t sec, long nsec)
1759 {
1760 struct timespec64 ts = { .tv_sec = sec,
1761 .tv_nsec = nsec };
1762
1763 return inode_set_ctime_to_ts(inode, ts);
1764 }
1765
1766 struct timespec64 simple_inode_init_ts(struct inode *inode);
1767
1768 /*
1769 * Snapshotting support.
1770 */
1771
1772 /*
1773 * These are internal functions, please use sb_start_{write,pagefault,intwrite}
1774 * instead.
1775 */
__sb_end_write(struct super_block * sb,int level)1776 static inline void __sb_end_write(struct super_block *sb, int level)
1777 {
1778 percpu_up_read(sb->s_writers.rw_sem + level-1);
1779 }
1780
__sb_start_write(struct super_block * sb,int level)1781 static inline void __sb_start_write(struct super_block *sb, int level)
1782 {
1783 percpu_down_read(sb->s_writers.rw_sem + level - 1);
1784 }
1785
__sb_start_write_trylock(struct super_block * sb,int level)1786 static inline bool __sb_start_write_trylock(struct super_block *sb, int level)
1787 {
1788 return percpu_down_read_trylock(sb->s_writers.rw_sem + level - 1);
1789 }
1790
1791 #define __sb_writers_acquired(sb, lev) \
1792 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1793 #define __sb_writers_release(sb, lev) \
1794 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], _THIS_IP_)
1795
1796 /**
1797 * __sb_write_started - check if sb freeze level is held
1798 * @sb: the super we write to
1799 * @level: the freeze level
1800 *
1801 * * > 0 - sb freeze level is held
1802 * * 0 - sb freeze level is not held
1803 * * < 0 - !CONFIG_LOCKDEP/LOCK_STATE_UNKNOWN
1804 */
__sb_write_started(const struct super_block * sb,int level)1805 static inline int __sb_write_started(const struct super_block *sb, int level)
1806 {
1807 return lockdep_is_held_type(sb->s_writers.rw_sem + level - 1, 1);
1808 }
1809
1810 /**
1811 * sb_write_started - check if SB_FREEZE_WRITE is held
1812 * @sb: the super we write to
1813 *
1814 * May be false positive with !CONFIG_LOCKDEP/LOCK_STATE_UNKNOWN.
1815 */
sb_write_started(const struct super_block * sb)1816 static inline bool sb_write_started(const struct super_block *sb)
1817 {
1818 return __sb_write_started(sb, SB_FREEZE_WRITE);
1819 }
1820
1821 /**
1822 * sb_write_not_started - check if SB_FREEZE_WRITE is not held
1823 * @sb: the super we write to
1824 *
1825 * May be false positive with !CONFIG_LOCKDEP/LOCK_STATE_UNKNOWN.
1826 */
sb_write_not_started(const struct super_block * sb)1827 static inline bool sb_write_not_started(const struct super_block *sb)
1828 {
1829 return __sb_write_started(sb, SB_FREEZE_WRITE) <= 0;
1830 }
1831
1832 /**
1833 * file_write_started - check if SB_FREEZE_WRITE is held
1834 * @file: the file we write to
1835 *
1836 * May be false positive with !CONFIG_LOCKDEP/LOCK_STATE_UNKNOWN.
1837 * May be false positive with !S_ISREG, because file_start_write() has
1838 * no effect on !S_ISREG.
1839 */
file_write_started(const struct file * file)1840 static inline bool file_write_started(const struct file *file)
1841 {
1842 if (!S_ISREG(file_inode(file)->i_mode))
1843 return true;
1844 return sb_write_started(file_inode(file)->i_sb);
1845 }
1846
1847 /**
1848 * file_write_not_started - check if SB_FREEZE_WRITE is not held
1849 * @file: the file we write to
1850 *
1851 * May be false positive with !CONFIG_LOCKDEP/LOCK_STATE_UNKNOWN.
1852 * May be false positive with !S_ISREG, because file_start_write() has
1853 * no effect on !S_ISREG.
1854 */
file_write_not_started(const struct file * file)1855 static inline bool file_write_not_started(const struct file *file)
1856 {
1857 if (!S_ISREG(file_inode(file)->i_mode))
1858 return true;
1859 return sb_write_not_started(file_inode(file)->i_sb);
1860 }
1861
1862 /**
1863 * sb_end_write - drop write access to a superblock
1864 * @sb: the super we wrote to
1865 *
1866 * Decrement number of writers to the filesystem. Wake up possible waiters
1867 * wanting to freeze the filesystem.
1868 */
sb_end_write(struct super_block * sb)1869 static inline void sb_end_write(struct super_block *sb)
1870 {
1871 __sb_end_write(sb, SB_FREEZE_WRITE);
1872 }
1873
1874 /**
1875 * sb_end_pagefault - drop write access to a superblock from a page fault
1876 * @sb: the super we wrote to
1877 *
1878 * Decrement number of processes handling write page fault to the filesystem.
1879 * Wake up possible waiters wanting to freeze the filesystem.
1880 */
sb_end_pagefault(struct super_block * sb)1881 static inline void sb_end_pagefault(struct super_block *sb)
1882 {
1883 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1884 }
1885
1886 /**
1887 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1888 * @sb: the super we wrote to
1889 *
1890 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1891 * waiters wanting to freeze the filesystem.
1892 */
sb_end_intwrite(struct super_block * sb)1893 static inline void sb_end_intwrite(struct super_block *sb)
1894 {
1895 __sb_end_write(sb, SB_FREEZE_FS);
1896 }
1897
1898 /**
1899 * sb_start_write - get write access to a superblock
1900 * @sb: the super we write to
1901 *
1902 * When a process wants to write data or metadata to a file system (i.e. dirty
1903 * a page or an inode), it should embed the operation in a sb_start_write() -
1904 * sb_end_write() pair to get exclusion against file system freezing. This
1905 * function increments number of writers preventing freezing. If the file
1906 * system is already frozen, the function waits until the file system is
1907 * thawed.
1908 *
1909 * Since freeze protection behaves as a lock, users have to preserve
1910 * ordering of freeze protection and other filesystem locks. Generally,
1911 * freeze protection should be the outermost lock. In particular, we have:
1912 *
1913 * sb_start_write
1914 * -> i_mutex (write path, truncate, directory ops, ...)
1915 * -> s_umount (freeze_super, thaw_super)
1916 */
sb_start_write(struct super_block * sb)1917 static inline void sb_start_write(struct super_block *sb)
1918 {
1919 __sb_start_write(sb, SB_FREEZE_WRITE);
1920 }
1921
sb_start_write_trylock(struct super_block * sb)1922 static inline bool sb_start_write_trylock(struct super_block *sb)
1923 {
1924 return __sb_start_write_trylock(sb, SB_FREEZE_WRITE);
1925 }
1926
1927 /**
1928 * sb_start_pagefault - get write access to a superblock from a page fault
1929 * @sb: the super we write to
1930 *
1931 * When a process starts handling write page fault, it should embed the
1932 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1933 * exclusion against file system freezing. This is needed since the page fault
1934 * is going to dirty a page. This function increments number of running page
1935 * faults preventing freezing. If the file system is already frozen, the
1936 * function waits until the file system is thawed.
1937 *
1938 * Since page fault freeze protection behaves as a lock, users have to preserve
1939 * ordering of freeze protection and other filesystem locks. It is advised to
1940 * put sb_start_pagefault() close to mmap_lock in lock ordering. Page fault
1941 * handling code implies lock dependency:
1942 *
1943 * mmap_lock
1944 * -> sb_start_pagefault
1945 */
sb_start_pagefault(struct super_block * sb)1946 static inline void sb_start_pagefault(struct super_block *sb)
1947 {
1948 __sb_start_write(sb, SB_FREEZE_PAGEFAULT);
1949 }
1950
1951 /**
1952 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1953 * @sb: the super we write to
1954 *
1955 * This is the third level of protection against filesystem freezing. It is
1956 * free for use by a filesystem. The only requirement is that it must rank
1957 * below sb_start_pagefault.
1958 *
1959 * For example filesystem can call sb_start_intwrite() when starting a
1960 * transaction which somewhat eases handling of freezing for internal sources
1961 * of filesystem changes (internal fs threads, discarding preallocation on file
1962 * close, etc.).
1963 */
sb_start_intwrite(struct super_block * sb)1964 static inline void sb_start_intwrite(struct super_block *sb)
1965 {
1966 __sb_start_write(sb, SB_FREEZE_FS);
1967 }
1968
sb_start_intwrite_trylock(struct super_block * sb)1969 static inline bool sb_start_intwrite_trylock(struct super_block *sb)
1970 {
1971 return __sb_start_write_trylock(sb, SB_FREEZE_FS);
1972 }
1973
1974 bool inode_owner_or_capable(struct mnt_idmap *idmap,
1975 const struct inode *inode);
1976
1977 /*
1978 * VFS helper functions..
1979 */
1980 int vfs_create(struct mnt_idmap *, struct inode *,
1981 struct dentry *, umode_t, bool);
1982 struct dentry *vfs_mkdir(struct mnt_idmap *, struct inode *,
1983 struct dentry *, umode_t);
1984 int vfs_mknod(struct mnt_idmap *, struct inode *, struct dentry *,
1985 umode_t, dev_t);
1986 int vfs_symlink(struct mnt_idmap *, struct inode *,
1987 struct dentry *, const char *);
1988 int vfs_link(struct dentry *, struct mnt_idmap *, struct inode *,
1989 struct dentry *, struct inode **);
1990 int vfs_rmdir(struct mnt_idmap *, struct inode *, struct dentry *);
1991 int vfs_unlink(struct mnt_idmap *, struct inode *, struct dentry *,
1992 struct inode **);
1993
1994 /**
1995 * struct renamedata - contains all information required for renaming
1996 * @old_mnt_idmap: idmap of the old mount the inode was found from
1997 * @old_dir: parent of source
1998 * @old_dentry: source
1999 * @new_mnt_idmap: idmap of the new mount the inode was found from
2000 * @new_dir: parent of destination
2001 * @new_dentry: destination
2002 * @delegated_inode: returns an inode needing a delegation break
2003 * @flags: rename flags
2004 */
2005 struct renamedata {
2006 struct mnt_idmap *old_mnt_idmap;
2007 struct inode *old_dir;
2008 struct dentry *old_dentry;
2009 struct mnt_idmap *new_mnt_idmap;
2010 struct inode *new_dir;
2011 struct dentry *new_dentry;
2012 struct inode **delegated_inode;
2013 unsigned int flags;
2014 } __randomize_layout;
2015
2016 int vfs_rename(struct renamedata *);
2017
vfs_whiteout(struct mnt_idmap * idmap,struct inode * dir,struct dentry * dentry)2018 static inline int vfs_whiteout(struct mnt_idmap *idmap,
2019 struct inode *dir, struct dentry *dentry)
2020 {
2021 return vfs_mknod(idmap, dir, dentry, S_IFCHR | WHITEOUT_MODE,
2022 WHITEOUT_DEV);
2023 }
2024
2025 struct file *kernel_tmpfile_open(struct mnt_idmap *idmap,
2026 const struct path *parentpath,
2027 umode_t mode, int open_flag,
2028 const struct cred *cred);
2029 struct file *kernel_file_open(const struct path *path, int flags,
2030 const struct cred *cred);
2031
2032 int vfs_mkobj(struct dentry *, umode_t,
2033 int (*f)(struct dentry *, umode_t, void *),
2034 void *);
2035
2036 int vfs_fchown(struct file *file, uid_t user, gid_t group);
2037 int vfs_fchmod(struct file *file, umode_t mode);
2038 int vfs_utimes(const struct path *path, struct timespec64 *times);
2039
2040 int vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
2041
2042 #ifdef CONFIG_COMPAT
2043 extern long compat_ptr_ioctl(struct file *file, unsigned int cmd,
2044 unsigned long arg);
2045 #else
2046 #define compat_ptr_ioctl NULL
2047 #endif
2048
2049 /*
2050 * VFS file helper functions.
2051 */
2052 void inode_init_owner(struct mnt_idmap *idmap, struct inode *inode,
2053 const struct inode *dir, umode_t mode);
2054 extern bool may_open_dev(const struct path *path);
2055 umode_t mode_strip_sgid(struct mnt_idmap *idmap,
2056 const struct inode *dir, umode_t mode);
2057 bool in_group_or_capable(struct mnt_idmap *idmap,
2058 const struct inode *inode, vfsgid_t vfsgid);
2059
2060 /*
2061 * This is the "filldir" function type, used by readdir() to let
2062 * the kernel specify what kind of dirent layout it wants to have.
2063 * This allows the kernel to read directories into kernel space or
2064 * to have different dirent layouts depending on the binary type.
2065 * Return 'true' to keep going and 'false' if there are no more entries.
2066 */
2067 struct dir_context;
2068 typedef bool (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
2069 unsigned);
2070
2071 struct dir_context {
2072 filldir_t actor;
2073 loff_t pos;
2074 };
2075
2076 /*
2077 * These flags let !MMU mmap() govern direct device mapping vs immediate
2078 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
2079 *
2080 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
2081 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
2082 * NOMMU_MAP_READ: Can be mapped for reading
2083 * NOMMU_MAP_WRITE: Can be mapped for writing
2084 * NOMMU_MAP_EXEC: Can be mapped for execution
2085 */
2086 #define NOMMU_MAP_COPY 0x00000001
2087 #define NOMMU_MAP_DIRECT 0x00000008
2088 #define NOMMU_MAP_READ VM_MAYREAD
2089 #define NOMMU_MAP_WRITE VM_MAYWRITE
2090 #define NOMMU_MAP_EXEC VM_MAYEXEC
2091
2092 #define NOMMU_VMFLAGS \
2093 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
2094
2095 /*
2096 * These flags control the behavior of the remap_file_range function pointer.
2097 * If it is called with len == 0 that means "remap to end of source file".
2098 * See Documentation/filesystems/vfs.rst for more details about this call.
2099 *
2100 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
2101 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
2102 */
2103 #define REMAP_FILE_DEDUP (1 << 0)
2104 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
2105
2106 /*
2107 * These flags signal that the caller is ok with altering various aspects of
2108 * the behavior of the remap operation. The changes must be made by the
2109 * implementation; the vfs remap helper functions can take advantage of them.
2110 * Flags in this category exist to preserve the quirky behavior of the hoisted
2111 * btrfs clone/dedupe ioctls.
2112 */
2113 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
2114
2115 /*
2116 * These flags control the behavior of vfs_copy_file_range().
2117 * They are not available to the user via syscall.
2118 *
2119 * COPY_FILE_SPLICE: call splice direct instead of fs clone/copy ops
2120 */
2121 #define COPY_FILE_SPLICE (1 << 0)
2122
2123 struct iov_iter;
2124 struct io_uring_cmd;
2125 struct offset_ctx;
2126
2127 typedef unsigned int __bitwise fop_flags_t;
2128
2129 struct file_operations {
2130 struct module *owner;
2131 fop_flags_t fop_flags;
2132 loff_t (*llseek) (struct file *, loff_t, int);
2133 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
2134 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
2135 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
2136 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
2137 int (*iopoll)(struct kiocb *kiocb, struct io_comp_batch *,
2138 unsigned int flags);
2139 int (*iterate_shared) (struct file *, struct dir_context *);
2140 __poll_t (*poll) (struct file *, struct poll_table_struct *);
2141 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
2142 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
2143 int (*mmap) (struct file *, struct vm_area_struct *);
2144 int (*open) (struct inode *, struct file *);
2145 int (*flush) (struct file *, fl_owner_t id);
2146 int (*release) (struct inode *, struct file *);
2147 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
2148 int (*fasync) (int, struct file *, int);
2149 int (*lock) (struct file *, int, struct file_lock *);
2150 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
2151 int (*check_flags)(int);
2152 int (*flock) (struct file *, int, struct file_lock *);
2153 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
2154 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
2155 void (*splice_eof)(struct file *file);
2156 int (*setlease)(struct file *, int, struct file_lease **, void **);
2157 long (*fallocate)(struct file *file, int mode, loff_t offset,
2158 loff_t len);
2159 void (*show_fdinfo)(struct seq_file *m, struct file *f);
2160 #ifndef CONFIG_MMU
2161 unsigned (*mmap_capabilities)(struct file *);
2162 #endif
2163 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
2164 loff_t, size_t, unsigned int);
2165 loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
2166 struct file *file_out, loff_t pos_out,
2167 loff_t len, unsigned int remap_flags);
2168 int (*fadvise)(struct file *, loff_t, loff_t, int);
2169 int (*uring_cmd)(struct io_uring_cmd *ioucmd, unsigned int issue_flags);
2170 int (*uring_cmd_iopoll)(struct io_uring_cmd *, struct io_comp_batch *,
2171 unsigned int poll_flags);
2172 } __randomize_layout;
2173
2174 /* Supports async buffered reads */
2175 #define FOP_BUFFER_RASYNC ((__force fop_flags_t)(1 << 0))
2176 /* Supports async buffered writes */
2177 #define FOP_BUFFER_WASYNC ((__force fop_flags_t)(1 << 1))
2178 /* Supports synchronous page faults for mappings */
2179 #define FOP_MMAP_SYNC ((__force fop_flags_t)(1 << 2))
2180 /* Supports non-exclusive O_DIRECT writes from multiple threads */
2181 #define FOP_DIO_PARALLEL_WRITE ((__force fop_flags_t)(1 << 3))
2182 /* Contains huge pages */
2183 #define FOP_HUGE_PAGES ((__force fop_flags_t)(1 << 4))
2184 /* Treat loff_t as unsigned (e.g., /dev/mem) */
2185 #define FOP_UNSIGNED_OFFSET ((__force fop_flags_t)(1 << 5))
2186 /* Supports asynchronous lock callbacks */
2187 #define FOP_ASYNC_LOCK ((__force fop_flags_t)(1 << 6))
2188 /* File system supports uncached read/write buffered IO */
2189 #define FOP_DONTCACHE ((__force fop_flags_t)(1 << 7))
2190
2191 /* Wrap a directory iterator that needs exclusive inode access */
2192 int wrap_directory_iterator(struct file *, struct dir_context *,
2193 int (*) (struct file *, struct dir_context *));
2194 #define WRAP_DIR_ITER(x) \
2195 static int shared_##x(struct file *file , struct dir_context *ctx) \
2196 { return wrap_directory_iterator(file, ctx, x); }
2197
2198 struct inode_operations {
2199 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
2200 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
2201 int (*permission) (struct mnt_idmap *, struct inode *, int);
2202 struct posix_acl * (*get_inode_acl)(struct inode *, int, bool);
2203
2204 int (*readlink) (struct dentry *, char __user *,int);
2205
2206 int (*create) (struct mnt_idmap *, struct inode *,struct dentry *,
2207 umode_t, bool);
2208 int (*link) (struct dentry *,struct inode *,struct dentry *);
2209 int (*unlink) (struct inode *,struct dentry *);
2210 int (*symlink) (struct mnt_idmap *, struct inode *,struct dentry *,
2211 const char *);
2212 struct dentry *(*mkdir) (struct mnt_idmap *, struct inode *,
2213 struct dentry *, umode_t);
2214 int (*rmdir) (struct inode *,struct dentry *);
2215 int (*mknod) (struct mnt_idmap *, struct inode *,struct dentry *,
2216 umode_t,dev_t);
2217 int (*rename) (struct mnt_idmap *, struct inode *, struct dentry *,
2218 struct inode *, struct dentry *, unsigned int);
2219 int (*setattr) (struct mnt_idmap *, struct dentry *, struct iattr *);
2220 int (*getattr) (struct mnt_idmap *, const struct path *,
2221 struct kstat *, u32, unsigned int);
2222 ssize_t (*listxattr) (struct dentry *, char *, size_t);
2223 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
2224 u64 len);
2225 int (*update_time)(struct inode *, int);
2226 int (*atomic_open)(struct inode *, struct dentry *,
2227 struct file *, unsigned open_flag,
2228 umode_t create_mode);
2229 int (*tmpfile) (struct mnt_idmap *, struct inode *,
2230 struct file *, umode_t);
2231 struct posix_acl *(*get_acl)(struct mnt_idmap *, struct dentry *,
2232 int);
2233 int (*set_acl)(struct mnt_idmap *, struct dentry *,
2234 struct posix_acl *, int);
2235 int (*fileattr_set)(struct mnt_idmap *idmap,
2236 struct dentry *dentry, struct fileattr *fa);
2237 int (*fileattr_get)(struct dentry *dentry, struct fileattr *fa);
2238 struct offset_ctx *(*get_offset_ctx)(struct inode *inode);
2239 } ____cacheline_aligned;
2240
call_mmap(struct file * file,struct vm_area_struct * vma)2241 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
2242 {
2243 return file->f_op->mmap(file, vma);
2244 }
2245
2246 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
2247 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
2248 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
2249 loff_t, size_t, unsigned int);
2250 int remap_verify_area(struct file *file, loff_t pos, loff_t len, bool write);
2251 int __generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
2252 struct file *file_out, loff_t pos_out,
2253 loff_t *len, unsigned int remap_flags,
2254 const struct iomap_ops *dax_read_ops);
2255 int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
2256 struct file *file_out, loff_t pos_out,
2257 loff_t *count, unsigned int remap_flags);
2258 extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in,
2259 struct file *file_out, loff_t pos_out,
2260 loff_t len, unsigned int remap_flags);
2261 extern int vfs_dedupe_file_range(struct file *file,
2262 struct file_dedupe_range *same);
2263 extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
2264 struct file *dst_file, loff_t dst_pos,
2265 loff_t len, unsigned int remap_flags);
2266
2267 /**
2268 * enum freeze_holder - holder of the freeze
2269 * @FREEZE_HOLDER_KERNEL: kernel wants to freeze or thaw filesystem
2270 * @FREEZE_HOLDER_USERSPACE: userspace wants to freeze or thaw filesystem
2271 * @FREEZE_MAY_NEST: whether nesting freeze and thaw requests is allowed
2272 *
2273 * Indicate who the owner of the freeze or thaw request is and whether
2274 * the freeze needs to be exclusive or can nest.
2275 * Without @FREEZE_MAY_NEST, multiple freeze and thaw requests from the
2276 * same holder aren't allowed. It is however allowed to hold a single
2277 * @FREEZE_HOLDER_USERSPACE and a single @FREEZE_HOLDER_KERNEL freeze at
2278 * the same time. This is relied upon by some filesystems during online
2279 * repair or similar.
2280 */
2281 enum freeze_holder {
2282 FREEZE_HOLDER_KERNEL = (1U << 0),
2283 FREEZE_HOLDER_USERSPACE = (1U << 1),
2284 FREEZE_MAY_NEST = (1U << 2),
2285 };
2286
2287 struct super_operations {
2288 struct inode *(*alloc_inode)(struct super_block *sb);
2289 void (*destroy_inode)(struct inode *);
2290 void (*free_inode)(struct inode *);
2291
2292 void (*dirty_inode) (struct inode *, int flags);
2293 int (*write_inode) (struct inode *, struct writeback_control *wbc);
2294 int (*drop_inode) (struct inode *);
2295 void (*evict_inode) (struct inode *);
2296 void (*put_super) (struct super_block *);
2297 int (*sync_fs)(struct super_block *sb, int wait);
2298 int (*freeze_super) (struct super_block *, enum freeze_holder who);
2299 int (*freeze_fs) (struct super_block *);
2300 int (*thaw_super) (struct super_block *, enum freeze_holder who);
2301 int (*unfreeze_fs) (struct super_block *);
2302 int (*statfs) (struct dentry *, struct kstatfs *);
2303 int (*remount_fs) (struct super_block *, int *, char *);
2304 void (*umount_begin) (struct super_block *);
2305
2306 int (*show_options)(struct seq_file *, struct dentry *);
2307 int (*show_devname)(struct seq_file *, struct dentry *);
2308 int (*show_path)(struct seq_file *, struct dentry *);
2309 int (*show_stats)(struct seq_file *, struct dentry *);
2310 #ifdef CONFIG_QUOTA
2311 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
2312 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
2313 struct dquot __rcu **(*get_dquots)(struct inode *);
2314 #endif
2315 long (*nr_cached_objects)(struct super_block *,
2316 struct shrink_control *);
2317 long (*free_cached_objects)(struct super_block *,
2318 struct shrink_control *);
2319 void (*shutdown)(struct super_block *sb);
2320 };
2321
2322 /*
2323 * Inode flags - they have no relation to superblock flags now
2324 */
2325 #define S_SYNC (1 << 0) /* Writes are synced at once */
2326 #define S_NOATIME (1 << 1) /* Do not update access times */
2327 #define S_APPEND (1 << 2) /* Append-only file */
2328 #define S_IMMUTABLE (1 << 3) /* Immutable file */
2329 #define S_DEAD (1 << 4) /* removed, but still open directory */
2330 #define S_NOQUOTA (1 << 5) /* Inode is not counted to quota */
2331 #define S_DIRSYNC (1 << 6) /* Directory modifications are synchronous */
2332 #define S_NOCMTIME (1 << 7) /* Do not update file c/mtime */
2333 #define S_SWAPFILE (1 << 8) /* Do not truncate: swapon got its bmaps */
2334 #define S_PRIVATE (1 << 9) /* Inode is fs-internal */
2335 #define S_IMA (1 << 10) /* Inode has an associated IMA struct */
2336 #define S_AUTOMOUNT (1 << 11) /* Automount/referral quasi-directory */
2337 #define S_NOSEC (1 << 12) /* no suid or xattr security attributes */
2338 #ifdef CONFIG_FS_DAX
2339 #define S_DAX (1 << 13) /* Direct Access, avoiding the page cache */
2340 #else
2341 #define S_DAX 0 /* Make all the DAX code disappear */
2342 #endif
2343 #define S_ENCRYPTED (1 << 14) /* Encrypted file (using fs/crypto/) */
2344 #define S_CASEFOLD (1 << 15) /* Casefolded file */
2345 #define S_VERITY (1 << 16) /* Verity file (using fs/verity/) */
2346 #define S_KERNEL_FILE (1 << 17) /* File is in use by the kernel (eg. fs/cachefiles) */
2347
2348 /*
2349 * Note that nosuid etc flags are inode-specific: setting some file-system
2350 * flags just means all the inodes inherit those flags by default. It might be
2351 * possible to override it selectively if you really wanted to with some
2352 * ioctl() that is not currently implemented.
2353 *
2354 * Exception: SB_RDONLY is always applied to the entire file system.
2355 *
2356 * Unfortunately, it is possible to change a filesystems flags with it mounted
2357 * with files in use. This means that all of the inodes will not have their
2358 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
2359 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
2360 */
2361 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
2362
sb_rdonly(const struct super_block * sb)2363 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
2364 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
2365 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
2366 ((inode)->i_flags & S_SYNC))
2367 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
2368 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
2369 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
2370 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
2371 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
2372
2373 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
2374 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
2375 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
2376
2377 #ifdef CONFIG_FS_POSIX_ACL
2378 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
2379 #else
2380 #define IS_POSIXACL(inode) 0
2381 #endif
2382
2383 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
2384 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
2385
2386 #ifdef CONFIG_SWAP
2387 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
2388 #else
2389 #define IS_SWAPFILE(inode) ((void)(inode), 0U)
2390 #endif
2391
2392 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
2393 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
2394 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2395 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2396 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2397 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2398 #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD)
2399 #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY)
2400
2401 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2402 (inode)->i_rdev == WHITEOUT_DEV)
2403
HAS_UNMAPPED_ID(struct mnt_idmap * idmap,struct inode * inode)2404 static inline bool HAS_UNMAPPED_ID(struct mnt_idmap *idmap,
2405 struct inode *inode)
2406 {
2407 return !vfsuid_valid(i_uid_into_vfsuid(idmap, inode)) ||
2408 !vfsgid_valid(i_gid_into_vfsgid(idmap, inode));
2409 }
2410
init_sync_kiocb(struct kiocb * kiocb,struct file * filp)2411 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
2412 {
2413 *kiocb = (struct kiocb) {
2414 .ki_filp = filp,
2415 .ki_flags = filp->f_iocb_flags,
2416 .ki_ioprio = get_current_ioprio(),
2417 };
2418 }
2419
kiocb_clone(struct kiocb * kiocb,struct kiocb * kiocb_src,struct file * filp)2420 static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src,
2421 struct file *filp)
2422 {
2423 *kiocb = (struct kiocb) {
2424 .ki_filp = filp,
2425 .ki_flags = kiocb_src->ki_flags,
2426 .ki_ioprio = kiocb_src->ki_ioprio,
2427 .ki_pos = kiocb_src->ki_pos,
2428 };
2429 }
2430
2431 /*
2432 * Inode state bits. Protected by inode->i_lock
2433 *
2434 * Four bits determine the dirty state of the inode: I_DIRTY_SYNC,
2435 * I_DIRTY_DATASYNC, I_DIRTY_PAGES, and I_DIRTY_TIME.
2436 *
2437 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2438 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2439 * various stages of removing an inode.
2440 *
2441 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2442 *
2443 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2444 * fdatasync() (unless I_DIRTY_DATASYNC is also set).
2445 * Timestamp updates are the usual cause.
2446 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2447 * these changes separately from I_DIRTY_SYNC so that we
2448 * don't have to write inode on fdatasync() when only
2449 * e.g. the timestamps have changed.
2450 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2451 * I_DIRTY_TIME The inode itself has dirty timestamps, and the
2452 * lazytime mount option is enabled. We keep track of this
2453 * separately from I_DIRTY_SYNC in order to implement
2454 * lazytime. This gets cleared if I_DIRTY_INODE
2455 * (I_DIRTY_SYNC and/or I_DIRTY_DATASYNC) gets set. But
2456 * I_DIRTY_TIME can still be set if I_DIRTY_SYNC is already
2457 * in place because writeback might already be in progress
2458 * and we don't want to lose the time update
2459 * I_NEW Serves as both a mutex and completion notification.
2460 * New inodes set I_NEW. If two processes both create
2461 * the same inode, one of them will release its inode and
2462 * wait for I_NEW to be released before returning.
2463 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2464 * also cause waiting on I_NEW, without I_NEW actually
2465 * being set. find_inode() uses this to prevent returning
2466 * nearly-dead inodes.
2467 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2468 * is zero. I_FREEING must be set when I_WILL_FREE is
2469 * cleared.
2470 * I_FREEING Set when inode is about to be freed but still has dirty
2471 * pages or buffers attached or the inode itself is still
2472 * dirty.
2473 * I_CLEAR Added by clear_inode(). In this state the inode is
2474 * clean and can be destroyed. Inode keeps I_FREEING.
2475 *
2476 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2477 * prohibited for many purposes. iget() must wait for
2478 * the inode to be completely released, then create it
2479 * anew. Other functions will just ignore such inodes,
2480 * if appropriate. I_NEW is used for waiting.
2481 *
2482 * I_SYNC Writeback of inode is running. The bit is set during
2483 * data writeback, and cleared with a wakeup on the bit
2484 * address once it is done. The bit is also used to pin
2485 * the inode in memory for flusher thread.
2486 *
2487 * I_REFERENCED Marks the inode as recently references on the LRU list.
2488 *
2489 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2490 * synchronize competing switching instances and to tell
2491 * wb stat updates to grab the i_pages lock. See
2492 * inode_switch_wbs_work_fn() for details.
2493 *
2494 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2495 * and work dirs among overlayfs mounts.
2496 *
2497 * I_CREATING New object's inode in the middle of setting up.
2498 *
2499 * I_DONTCACHE Evict inode as soon as it is not used anymore.
2500 *
2501 * I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists.
2502 * Used to detect that mark_inode_dirty() should not move
2503 * inode between dirty lists.
2504 *
2505 * I_PINNING_FSCACHE_WB Inode is pinning an fscache object for writeback.
2506 *
2507 * I_LRU_ISOLATING Inode is pinned being isolated from LRU without holding
2508 * i_count.
2509 *
2510 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2511 *
2512 * __I_{SYNC,NEW,LRU_ISOLATING} are used to derive unique addresses to wait
2513 * upon. There's one free address left.
2514 */
2515 #define __I_NEW 0
2516 #define I_NEW (1 << __I_NEW)
2517 #define __I_SYNC 1
2518 #define I_SYNC (1 << __I_SYNC)
2519 #define __I_LRU_ISOLATING 2
2520 #define I_LRU_ISOLATING (1 << __I_LRU_ISOLATING)
2521
2522 #define I_DIRTY_SYNC (1 << 3)
2523 #define I_DIRTY_DATASYNC (1 << 4)
2524 #define I_DIRTY_PAGES (1 << 5)
2525 #define I_WILL_FREE (1 << 6)
2526 #define I_FREEING (1 << 7)
2527 #define I_CLEAR (1 << 8)
2528 #define I_REFERENCED (1 << 9)
2529 #define I_LINKABLE (1 << 10)
2530 #define I_DIRTY_TIME (1 << 11)
2531 #define I_WB_SWITCH (1 << 12)
2532 #define I_OVL_INUSE (1 << 13)
2533 #define I_CREATING (1 << 14)
2534 #define I_DONTCACHE (1 << 15)
2535 #define I_SYNC_QUEUED (1 << 16)
2536 #define I_PINNING_NETFS_WB (1 << 17)
2537
2538 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2539 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2540 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2541
2542 extern void __mark_inode_dirty(struct inode *, int);
mark_inode_dirty(struct inode * inode)2543 static inline void mark_inode_dirty(struct inode *inode)
2544 {
2545 __mark_inode_dirty(inode, I_DIRTY);
2546 }
2547
mark_inode_dirty_sync(struct inode * inode)2548 static inline void mark_inode_dirty_sync(struct inode *inode)
2549 {
2550 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2551 }
2552
2553 /*
2554 * Returns true if the given inode itself only has dirty timestamps (its pages
2555 * may still be dirty) and isn't currently being allocated or freed.
2556 * Filesystems should call this if when writing an inode when lazytime is
2557 * enabled, they want to opportunistically write the timestamps of other inodes
2558 * located very nearby on-disk, e.g. in the same inode block. This returns true
2559 * if the given inode is in need of such an opportunistic update. Requires
2560 * i_lock, or at least later re-checking under i_lock.
2561 */
inode_is_dirtytime_only(struct inode * inode)2562 static inline bool inode_is_dirtytime_only(struct inode *inode)
2563 {
2564 return (inode->i_state & (I_DIRTY_TIME | I_NEW |
2565 I_FREEING | I_WILL_FREE)) == I_DIRTY_TIME;
2566 }
2567
2568 extern void inc_nlink(struct inode *inode);
2569 extern void drop_nlink(struct inode *inode);
2570 extern void clear_nlink(struct inode *inode);
2571 extern void set_nlink(struct inode *inode, unsigned int nlink);
2572
inode_inc_link_count(struct inode * inode)2573 static inline void inode_inc_link_count(struct inode *inode)
2574 {
2575 inc_nlink(inode);
2576 mark_inode_dirty(inode);
2577 }
2578
inode_dec_link_count(struct inode * inode)2579 static inline void inode_dec_link_count(struct inode *inode)
2580 {
2581 drop_nlink(inode);
2582 mark_inode_dirty(inode);
2583 }
2584
2585 enum file_time_flags {
2586 S_ATIME = 1,
2587 S_MTIME = 2,
2588 S_CTIME = 4,
2589 S_VERSION = 8,
2590 };
2591
2592 extern bool atime_needs_update(const struct path *, struct inode *);
2593 extern void touch_atime(const struct path *);
2594 int inode_update_time(struct inode *inode, int flags);
2595
file_accessed(struct file * file)2596 static inline void file_accessed(struct file *file)
2597 {
2598 if (!(file->f_flags & O_NOATIME))
2599 touch_atime(&file->f_path);
2600 }
2601
2602 extern int file_modified(struct file *file);
2603 int kiocb_modified(struct kiocb *iocb);
2604
2605 int sync_inode_metadata(struct inode *inode, int wait);
2606
2607 struct file_system_type {
2608 const char *name;
2609 int fs_flags;
2610 #define FS_REQUIRES_DEV 1
2611 #define FS_BINARY_MOUNTDATA 2
2612 #define FS_HAS_SUBTYPE 4
2613 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2614 #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */
2615 #define FS_ALLOW_IDMAP 32 /* FS has been updated to handle vfs idmappings. */
2616 #define FS_MGTIME 64 /* FS uses multigrain timestamps */
2617 #define FS_LBS 128 /* FS supports LBS */
2618 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2619 int (*init_fs_context)(struct fs_context *);
2620 const struct fs_parameter_spec *parameters;
2621 struct dentry *(*mount) (struct file_system_type *, int,
2622 const char *, void *);
2623 void (*kill_sb) (struct super_block *);
2624 struct module *owner;
2625 struct file_system_type * next;
2626 struct hlist_head fs_supers;
2627
2628 struct lock_class_key s_lock_key;
2629 struct lock_class_key s_umount_key;
2630 struct lock_class_key s_vfs_rename_key;
2631 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2632
2633 struct lock_class_key i_lock_key;
2634 struct lock_class_key i_mutex_key;
2635 struct lock_class_key invalidate_lock_key;
2636 struct lock_class_key i_mutex_dir_key;
2637 };
2638
2639 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2640
2641 /**
2642 * is_mgtime: is this inode using multigrain timestamps
2643 * @inode: inode to test for multigrain timestamps
2644 *
2645 * Return true if the inode uses multigrain timestamps, false otherwise.
2646 */
is_mgtime(const struct inode * inode)2647 static inline bool is_mgtime(const struct inode *inode)
2648 {
2649 return inode->i_opflags & IOP_MGTIME;
2650 }
2651
2652 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2653 int flags, const char *dev_name, void *data,
2654 int (*fill_super)(struct super_block *, void *, int));
2655 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2656 int flags, void *data,
2657 int (*fill_super)(struct super_block *, void *, int));
2658 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2659 void retire_super(struct super_block *sb);
2660 void generic_shutdown_super(struct super_block *sb);
2661 void kill_block_super(struct super_block *sb);
2662 void kill_anon_super(struct super_block *sb);
2663 void kill_litter_super(struct super_block *sb);
2664 void deactivate_super(struct super_block *sb);
2665 void deactivate_locked_super(struct super_block *sb);
2666 int set_anon_super(struct super_block *s, void *data);
2667 int set_anon_super_fc(struct super_block *s, struct fs_context *fc);
2668 int get_anon_bdev(dev_t *);
2669 void free_anon_bdev(dev_t);
2670 struct super_block *sget_fc(struct fs_context *fc,
2671 int (*test)(struct super_block *, struct fs_context *),
2672 int (*set)(struct super_block *, struct fs_context *));
2673 struct super_block *sget(struct file_system_type *type,
2674 int (*test)(struct super_block *,void *),
2675 int (*set)(struct super_block *,void *),
2676 int flags, void *data);
2677 struct super_block *sget_dev(struct fs_context *fc, dev_t dev);
2678
2679 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2680 #define fops_get(fops) ({ \
2681 const struct file_operations *_fops = (fops); \
2682 (((_fops) && try_module_get((_fops)->owner) ? (_fops) : NULL)); \
2683 })
2684
2685 #define fops_put(fops) ({ \
2686 const struct file_operations *_fops = (fops); \
2687 if (_fops) \
2688 module_put((_fops)->owner); \
2689 })
2690
2691 /*
2692 * This one is to be used *ONLY* from ->open() instances.
2693 * fops must be non-NULL, pinned down *and* module dependencies
2694 * should be sufficient to pin the caller down as well.
2695 */
2696 #define replace_fops(f, fops) \
2697 do { \
2698 struct file *__file = (f); \
2699 fops_put(__file->f_op); \
2700 BUG_ON(!(__file->f_op = (fops))); \
2701 } while(0)
2702
2703 extern int register_filesystem(struct file_system_type *);
2704 extern int unregister_filesystem(struct file_system_type *);
2705 extern int vfs_statfs(const struct path *, struct kstatfs *);
2706 extern int user_statfs(const char __user *, struct kstatfs *);
2707 extern int fd_statfs(int, struct kstatfs *);
2708 int freeze_super(struct super_block *super, enum freeze_holder who);
2709 int thaw_super(struct super_block *super, enum freeze_holder who);
2710 extern __printf(2, 3)
2711 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2712 extern int super_setup_bdi(struct super_block *sb);
2713
super_set_uuid(struct super_block * sb,const u8 * uuid,unsigned len)2714 static inline void super_set_uuid(struct super_block *sb, const u8 *uuid, unsigned len)
2715 {
2716 if (WARN_ON(len > sizeof(sb->s_uuid)))
2717 len = sizeof(sb->s_uuid);
2718 sb->s_uuid_len = len;
2719 memcpy(&sb->s_uuid, uuid, len);
2720 }
2721
2722 /* set sb sysfs name based on sb->s_bdev */
super_set_sysfs_name_bdev(struct super_block * sb)2723 static inline void super_set_sysfs_name_bdev(struct super_block *sb)
2724 {
2725 snprintf(sb->s_sysfs_name, sizeof(sb->s_sysfs_name), "%pg", sb->s_bdev);
2726 }
2727
2728 /* set sb sysfs name based on sb->s_uuid */
super_set_sysfs_name_uuid(struct super_block * sb)2729 static inline void super_set_sysfs_name_uuid(struct super_block *sb)
2730 {
2731 WARN_ON(sb->s_uuid_len != sizeof(sb->s_uuid));
2732 snprintf(sb->s_sysfs_name, sizeof(sb->s_sysfs_name), "%pU", sb->s_uuid.b);
2733 }
2734
2735 /* set sb sysfs name based on sb->s_id */
super_set_sysfs_name_id(struct super_block * sb)2736 static inline void super_set_sysfs_name_id(struct super_block *sb)
2737 {
2738 strscpy(sb->s_sysfs_name, sb->s_id, sizeof(sb->s_sysfs_name));
2739 }
2740
2741 /* try to use something standard before you use this */
2742 __printf(2, 3)
super_set_sysfs_name_generic(struct super_block * sb,const char * fmt,...)2743 static inline void super_set_sysfs_name_generic(struct super_block *sb, const char *fmt, ...)
2744 {
2745 va_list args;
2746
2747 va_start(args, fmt);
2748 vsnprintf(sb->s_sysfs_name, sizeof(sb->s_sysfs_name), fmt, args);
2749 va_end(args);
2750 }
2751
2752 extern int current_umask(void);
2753
2754 extern void ihold(struct inode * inode);
2755 extern void iput(struct inode *);
2756 int inode_update_timestamps(struct inode *inode, int flags);
2757 int generic_update_time(struct inode *, int);
2758
2759 /* /sys/fs */
2760 extern struct kobject *fs_kobj;
2761
2762 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2763
2764 /* fs/open.c */
2765 struct audit_names;
2766 struct filename {
2767 const char *name; /* pointer to actual string */
2768 const __user char *uptr; /* original userland pointer */
2769 atomic_t refcnt;
2770 struct audit_names *aname;
2771 const char iname[];
2772 };
2773 static_assert(offsetof(struct filename, iname) % sizeof(long) == 0);
2774
file_mnt_idmap(const struct file * file)2775 static inline struct mnt_idmap *file_mnt_idmap(const struct file *file)
2776 {
2777 return mnt_idmap(file->f_path.mnt);
2778 }
2779
2780 /**
2781 * is_idmapped_mnt - check whether a mount is mapped
2782 * @mnt: the mount to check
2783 *
2784 * If @mnt has an non @nop_mnt_idmap attached to it then @mnt is mapped.
2785 *
2786 * Return: true if mount is mapped, false if not.
2787 */
is_idmapped_mnt(const struct vfsmount * mnt)2788 static inline bool is_idmapped_mnt(const struct vfsmount *mnt)
2789 {
2790 return mnt_idmap(mnt) != &nop_mnt_idmap;
2791 }
2792
2793 int vfs_truncate(const struct path *, loff_t);
2794 int do_truncate(struct mnt_idmap *, struct dentry *, loff_t start,
2795 unsigned int time_attrs, struct file *filp);
2796 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2797 loff_t len);
2798 int do_sys_open(int dfd, const char __user *filename, int flags,
2799 umode_t mode);
2800 extern struct file *file_open_name(struct filename *, int, umode_t);
2801 extern struct file *filp_open(const char *, int, umode_t);
2802 extern struct file *file_open_root(const struct path *,
2803 const char *, int, umode_t);
file_open_root_mnt(struct vfsmount * mnt,const char * name,int flags,umode_t mode)2804 static inline struct file *file_open_root_mnt(struct vfsmount *mnt,
2805 const char *name, int flags, umode_t mode)
2806 {
2807 return file_open_root(&(struct path){.mnt = mnt, .dentry = mnt->mnt_root},
2808 name, flags, mode);
2809 }
2810 struct file *dentry_open(const struct path *path, int flags,
2811 const struct cred *creds);
2812 struct file *dentry_open_nonotify(const struct path *path, int flags,
2813 const struct cred *cred);
2814 struct file *dentry_create(const struct path *path, int flags, umode_t mode,
2815 const struct cred *cred);
2816 struct path *backing_file_user_path(struct file *f);
2817
2818 /*
2819 * When mmapping a file on a stackable filesystem (e.g., overlayfs), the file
2820 * stored in ->vm_file is a backing file whose f_inode is on the underlying
2821 * filesystem. When the mapped file path and inode number are displayed to
2822 * user (e.g. via /proc/<pid>/maps), these helpers should be used to get the
2823 * path and inode number to display to the user, which is the path of the fd
2824 * that user has requested to map and the inode number that would be returned
2825 * by fstat() on that same fd.
2826 */
2827 /* Get the path to display in /proc/<pid>/maps */
file_user_path(struct file * f)2828 static inline const struct path *file_user_path(struct file *f)
2829 {
2830 if (unlikely(f->f_mode & FMODE_BACKING))
2831 return backing_file_user_path(f);
2832 return &f->f_path;
2833 }
2834 /* Get the inode whose inode number to display in /proc/<pid>/maps */
file_user_inode(struct file * f)2835 static inline const struct inode *file_user_inode(struct file *f)
2836 {
2837 if (unlikely(f->f_mode & FMODE_BACKING))
2838 return d_inode(backing_file_user_path(f)->dentry);
2839 return file_inode(f);
2840 }
2841
file_clone_open(struct file * file)2842 static inline struct file *file_clone_open(struct file *file)
2843 {
2844 return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2845 }
2846 extern int filp_close(struct file *, fl_owner_t id);
2847
2848 extern struct filename *getname_flags(const char __user *, int);
2849 extern struct filename *getname_uflags(const char __user *, int);
getname(const char __user * name)2850 static inline struct filename *getname(const char __user *name)
2851 {
2852 return getname_flags(name, 0);
2853 }
2854 extern struct filename *getname_kernel(const char *);
2855 extern struct filename *__getname_maybe_null(const char __user *);
getname_maybe_null(const char __user * name,int flags)2856 static inline struct filename *getname_maybe_null(const char __user *name, int flags)
2857 {
2858 if (!(flags & AT_EMPTY_PATH))
2859 return getname(name);
2860
2861 if (!name)
2862 return NULL;
2863 return __getname_maybe_null(name);
2864 }
2865 extern void putname(struct filename *name);
2866 DEFINE_FREE(putname, struct filename *, if (!IS_ERR_OR_NULL(_T)) putname(_T))
2867
refname(struct filename * name)2868 static inline struct filename *refname(struct filename *name)
2869 {
2870 atomic_inc(&name->refcnt);
2871 return name;
2872 }
2873
2874 extern int finish_open(struct file *file, struct dentry *dentry,
2875 int (*open)(struct inode *, struct file *));
2876 extern int finish_no_open(struct file *file, struct dentry *dentry);
2877
2878 /* Helper for the simple case when original dentry is used */
finish_open_simple(struct file * file,int error)2879 static inline int finish_open_simple(struct file *file, int error)
2880 {
2881 if (error)
2882 return error;
2883
2884 return finish_open(file, file->f_path.dentry, NULL);
2885 }
2886
2887 /* fs/dcache.c */
2888 extern void __init vfs_caches_init_early(void);
2889 extern void __init vfs_caches_init(void);
2890
2891 extern struct kmem_cache *names_cachep;
2892
2893 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2894 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2895
2896 extern struct super_block *blockdev_superblock;
sb_is_blkdev_sb(struct super_block * sb)2897 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2898 {
2899 return IS_ENABLED(CONFIG_BLOCK) && sb == blockdev_superblock;
2900 }
2901
2902 void emergency_thaw_all(void);
2903 extern int sync_filesystem(struct super_block *);
2904 extern const struct file_operations def_blk_fops;
2905 extern const struct file_operations def_chr_fops;
2906
2907 /* fs/char_dev.c */
2908 #define CHRDEV_MAJOR_MAX 512
2909 /* Marks the bottom of the first segment of free char majors */
2910 #define CHRDEV_MAJOR_DYN_END 234
2911 /* Marks the top and bottom of the second segment of free char majors */
2912 #define CHRDEV_MAJOR_DYN_EXT_START 511
2913 #define CHRDEV_MAJOR_DYN_EXT_END 384
2914
2915 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2916 extern int register_chrdev_region(dev_t, unsigned, const char *);
2917 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2918 unsigned int count, const char *name,
2919 const struct file_operations *fops);
2920 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2921 unsigned int count, const char *name);
2922 extern void unregister_chrdev_region(dev_t, unsigned);
2923 extern void chrdev_show(struct seq_file *,off_t);
2924
register_chrdev(unsigned int major,const char * name,const struct file_operations * fops)2925 static inline int register_chrdev(unsigned int major, const char *name,
2926 const struct file_operations *fops)
2927 {
2928 return __register_chrdev(major, 0, 256, name, fops);
2929 }
2930
unregister_chrdev(unsigned int major,const char * name)2931 static inline void unregister_chrdev(unsigned int major, const char *name)
2932 {
2933 __unregister_chrdev(major, 0, 256, name);
2934 }
2935
2936 extern void init_special_inode(struct inode *, umode_t, dev_t);
2937
2938 /* Invalid inode operations -- fs/bad_inode.c */
2939 extern void make_bad_inode(struct inode *);
2940 extern bool is_bad_inode(struct inode *);
2941
2942 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2943 loff_t lend);
2944 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2945 extern int __must_check file_write_and_wait_range(struct file *file,
2946 loff_t start, loff_t end);
2947 int filemap_fdatawrite_range_kick(struct address_space *mapping, loff_t start,
2948 loff_t end);
2949
file_write_and_wait(struct file * file)2950 static inline int file_write_and_wait(struct file *file)
2951 {
2952 return file_write_and_wait_range(file, 0, LLONG_MAX);
2953 }
2954
2955 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2956 int datasync);
2957 extern int vfs_fsync(struct file *file, int datasync);
2958
2959 extern int sync_file_range(struct file *file, loff_t offset, loff_t nbytes,
2960 unsigned int flags);
2961
iocb_is_dsync(const struct kiocb * iocb)2962 static inline bool iocb_is_dsync(const struct kiocb *iocb)
2963 {
2964 return (iocb->ki_flags & IOCB_DSYNC) ||
2965 IS_SYNC(iocb->ki_filp->f_mapping->host);
2966 }
2967
2968 /*
2969 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2970 * to already be updated for the write, and will return either the amount
2971 * of bytes passed in, or an error if syncing the file failed.
2972 */
generic_write_sync(struct kiocb * iocb,ssize_t count)2973 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2974 {
2975 if (iocb_is_dsync(iocb)) {
2976 int ret = vfs_fsync_range(iocb->ki_filp,
2977 iocb->ki_pos - count, iocb->ki_pos - 1,
2978 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2979 if (ret)
2980 return ret;
2981 } else if (iocb->ki_flags & IOCB_DONTCACHE) {
2982 struct address_space *mapping = iocb->ki_filp->f_mapping;
2983
2984 filemap_fdatawrite_range_kick(mapping, iocb->ki_pos - count,
2985 iocb->ki_pos - 1);
2986 }
2987
2988 return count;
2989 }
2990
2991 extern void emergency_sync(void);
2992 extern void emergency_remount(void);
2993
2994 #ifdef CONFIG_BLOCK
2995 extern int bmap(struct inode *inode, sector_t *block);
2996 #else
bmap(struct inode * inode,sector_t * block)2997 static inline int bmap(struct inode *inode, sector_t *block)
2998 {
2999 return -EINVAL;
3000 }
3001 #endif
3002
3003 int notify_change(struct mnt_idmap *, struct dentry *,
3004 struct iattr *, struct inode **);
3005 int inode_permission(struct mnt_idmap *, struct inode *, int);
3006 int generic_permission(struct mnt_idmap *, struct inode *, int);
file_permission(struct file * file,int mask)3007 static inline int file_permission(struct file *file, int mask)
3008 {
3009 return inode_permission(file_mnt_idmap(file),
3010 file_inode(file), mask);
3011 }
path_permission(const struct path * path,int mask)3012 static inline int path_permission(const struct path *path, int mask)
3013 {
3014 return inode_permission(mnt_idmap(path->mnt),
3015 d_inode(path->dentry), mask);
3016 }
3017 int __check_sticky(struct mnt_idmap *idmap, struct inode *dir,
3018 struct inode *inode);
3019
execute_ok(struct inode * inode)3020 static inline bool execute_ok(struct inode *inode)
3021 {
3022 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
3023 }
3024
inode_wrong_type(const struct inode * inode,umode_t mode)3025 static inline bool inode_wrong_type(const struct inode *inode, umode_t mode)
3026 {
3027 return (inode->i_mode ^ mode) & S_IFMT;
3028 }
3029
3030 /**
3031 * file_start_write - get write access to a superblock for regular file io
3032 * @file: the file we want to write to
3033 *
3034 * This is a variant of sb_start_write() which is a noop on non-regualr file.
3035 * Should be matched with a call to file_end_write().
3036 */
file_start_write(struct file * file)3037 static inline void file_start_write(struct file *file)
3038 {
3039 if (!S_ISREG(file_inode(file)->i_mode))
3040 return;
3041 sb_start_write(file_inode(file)->i_sb);
3042 }
3043
file_start_write_trylock(struct file * file)3044 static inline bool file_start_write_trylock(struct file *file)
3045 {
3046 if (!S_ISREG(file_inode(file)->i_mode))
3047 return true;
3048 return sb_start_write_trylock(file_inode(file)->i_sb);
3049 }
3050
3051 /**
3052 * file_end_write - drop write access to a superblock of a regular file
3053 * @file: the file we wrote to
3054 *
3055 * Should be matched with a call to file_start_write().
3056 */
file_end_write(struct file * file)3057 static inline void file_end_write(struct file *file)
3058 {
3059 if (!S_ISREG(file_inode(file)->i_mode))
3060 return;
3061 sb_end_write(file_inode(file)->i_sb);
3062 }
3063
3064 /**
3065 * kiocb_start_write - get write access to a superblock for async file io
3066 * @iocb: the io context we want to submit the write with
3067 *
3068 * This is a variant of sb_start_write() for async io submission.
3069 * Should be matched with a call to kiocb_end_write().
3070 */
kiocb_start_write(struct kiocb * iocb)3071 static inline void kiocb_start_write(struct kiocb *iocb)
3072 {
3073 struct inode *inode = file_inode(iocb->ki_filp);
3074
3075 sb_start_write(inode->i_sb);
3076 /*
3077 * Fool lockdep by telling it the lock got released so that it
3078 * doesn't complain about the held lock when we return to userspace.
3079 */
3080 __sb_writers_release(inode->i_sb, SB_FREEZE_WRITE);
3081 }
3082
3083 /**
3084 * kiocb_end_write - drop write access to a superblock after async file io
3085 * @iocb: the io context we sumbitted the write with
3086 *
3087 * Should be matched with a call to kiocb_start_write().
3088 */
kiocb_end_write(struct kiocb * iocb)3089 static inline void kiocb_end_write(struct kiocb *iocb)
3090 {
3091 struct inode *inode = file_inode(iocb->ki_filp);
3092
3093 /*
3094 * Tell lockdep we inherited freeze protection from submission thread.
3095 */
3096 __sb_writers_acquired(inode->i_sb, SB_FREEZE_WRITE);
3097 sb_end_write(inode->i_sb);
3098 }
3099
3100 /*
3101 * This is used for regular files where some users -- especially the
3102 * currently executed binary in a process, previously handled via
3103 * VM_DENYWRITE -- cannot handle concurrent write (and maybe mmap
3104 * read-write shared) accesses.
3105 *
3106 * get_write_access() gets write permission for a file.
3107 * put_write_access() releases this write permission.
3108 * deny_write_access() denies write access to a file.
3109 * allow_write_access() re-enables write access to a file.
3110 *
3111 * The i_writecount field of an inode can have the following values:
3112 * 0: no write access, no denied write access
3113 * < 0: (-i_writecount) users that denied write access to the file.
3114 * > 0: (i_writecount) users that have write access to the file.
3115 *
3116 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
3117 * except for the cases where we don't hold i_writecount yet. Then we need to
3118 * use {get,deny}_write_access() - these functions check the sign and refuse
3119 * to do the change if sign is wrong.
3120 */
get_write_access(struct inode * inode)3121 static inline int get_write_access(struct inode *inode)
3122 {
3123 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
3124 }
deny_write_access(struct file * file)3125 static inline int deny_write_access(struct file *file)
3126 {
3127 struct inode *inode = file_inode(file);
3128 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
3129 }
put_write_access(struct inode * inode)3130 static inline void put_write_access(struct inode * inode)
3131 {
3132 atomic_dec(&inode->i_writecount);
3133 }
allow_write_access(struct file * file)3134 static inline void allow_write_access(struct file *file)
3135 {
3136 if (file)
3137 atomic_inc(&file_inode(file)->i_writecount);
3138 }
3139
3140 /*
3141 * Do not prevent write to executable file when watched by pre-content events.
3142 *
3143 * Note that FMODE_FSNOTIFY_HSM mode is set depending on pre-content watches at
3144 * the time of file open and remains constant for entire lifetime of the file,
3145 * so if pre-content watches are added post execution or removed before the end
3146 * of the execution, it will not cause i_writecount reference leak.
3147 */
exe_file_deny_write_access(struct file * exe_file)3148 static inline int exe_file_deny_write_access(struct file *exe_file)
3149 {
3150 if (unlikely(FMODE_FSNOTIFY_HSM(exe_file->f_mode)))
3151 return 0;
3152 return deny_write_access(exe_file);
3153 }
exe_file_allow_write_access(struct file * exe_file)3154 static inline void exe_file_allow_write_access(struct file *exe_file)
3155 {
3156 if (unlikely(!exe_file || FMODE_FSNOTIFY_HSM(exe_file->f_mode)))
3157 return;
3158 allow_write_access(exe_file);
3159 }
3160
file_set_fsnotify_mode(struct file * file,fmode_t mode)3161 static inline void file_set_fsnotify_mode(struct file *file, fmode_t mode)
3162 {
3163 file->f_mode &= ~FMODE_FSNOTIFY_MASK;
3164 file->f_mode |= mode;
3165 }
3166
inode_is_open_for_write(const struct inode * inode)3167 static inline bool inode_is_open_for_write(const struct inode *inode)
3168 {
3169 return atomic_read(&inode->i_writecount) > 0;
3170 }
3171
3172 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
i_readcount_dec(struct inode * inode)3173 static inline void i_readcount_dec(struct inode *inode)
3174 {
3175 BUG_ON(atomic_dec_return(&inode->i_readcount) < 0);
3176 }
i_readcount_inc(struct inode * inode)3177 static inline void i_readcount_inc(struct inode *inode)
3178 {
3179 atomic_inc(&inode->i_readcount);
3180 }
3181 #else
i_readcount_dec(struct inode * inode)3182 static inline void i_readcount_dec(struct inode *inode)
3183 {
3184 return;
3185 }
i_readcount_inc(struct inode * inode)3186 static inline void i_readcount_inc(struct inode *inode)
3187 {
3188 return;
3189 }
3190 #endif
3191 extern int do_pipe_flags(int *, int);
3192
3193 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
3194 ssize_t __kernel_read(struct file *file, void *buf, size_t count, loff_t *pos);
3195 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
3196 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
3197 extern struct file * open_exec(const char *);
3198
3199 /* fs/dcache.c -- generic fs support functions */
3200 extern bool is_subdir(struct dentry *, struct dentry *);
3201 extern bool path_is_under(const struct path *, const struct path *);
3202
3203 extern char *file_path(struct file *, char *, int);
3204
3205 /**
3206 * is_dot_dotdot - returns true only if @name is "." or ".."
3207 * @name: file name to check
3208 * @len: length of file name, in bytes
3209 */
is_dot_dotdot(const char * name,size_t len)3210 static inline bool is_dot_dotdot(const char *name, size_t len)
3211 {
3212 return len && unlikely(name[0] == '.') &&
3213 (len == 1 || (len == 2 && name[1] == '.'));
3214 }
3215
3216 #include <linux/err.h>
3217
3218 /* needed for stackable file system support */
3219 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
3220
3221 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
3222
3223 extern int inode_init_always_gfp(struct super_block *, struct inode *, gfp_t);
inode_init_always(struct super_block * sb,struct inode * inode)3224 static inline int inode_init_always(struct super_block *sb, struct inode *inode)
3225 {
3226 return inode_init_always_gfp(sb, inode, GFP_NOFS);
3227 }
3228
3229 extern void inode_init_once(struct inode *);
3230 extern void address_space_init_once(struct address_space *mapping);
3231 extern struct inode * igrab(struct inode *);
3232 extern ino_t iunique(struct super_block *, ino_t);
3233 extern int inode_needs_sync(struct inode *inode);
3234 extern int generic_delete_inode(struct inode *inode);
generic_drop_inode(struct inode * inode)3235 static inline int generic_drop_inode(struct inode *inode)
3236 {
3237 return !inode->i_nlink || inode_unhashed(inode);
3238 }
3239 extern void d_mark_dontcache(struct inode *inode);
3240
3241 extern struct inode *ilookup5_nowait(struct super_block *sb,
3242 unsigned long hashval, int (*test)(struct inode *, void *),
3243 void *data);
3244 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
3245 int (*test)(struct inode *, void *), void *data);
3246 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
3247
3248 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
3249 int (*test)(struct inode *, void *),
3250 int (*set)(struct inode *, void *),
3251 void *data);
3252 struct inode *iget5_locked(struct super_block *, unsigned long,
3253 int (*test)(struct inode *, void *),
3254 int (*set)(struct inode *, void *), void *);
3255 struct inode *iget5_locked_rcu(struct super_block *, unsigned long,
3256 int (*test)(struct inode *, void *),
3257 int (*set)(struct inode *, void *), void *);
3258 extern struct inode * iget_locked(struct super_block *, unsigned long);
3259 extern struct inode *find_inode_nowait(struct super_block *,
3260 unsigned long,
3261 int (*match)(struct inode *,
3262 unsigned long, void *),
3263 void *data);
3264 extern struct inode *find_inode_rcu(struct super_block *, unsigned long,
3265 int (*)(struct inode *, void *), void *);
3266 extern struct inode *find_inode_by_ino_rcu(struct super_block *, unsigned long);
3267 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
3268 extern int insert_inode_locked(struct inode *);
3269 #ifdef CONFIG_DEBUG_LOCK_ALLOC
3270 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
3271 #else
lockdep_annotate_inode_mutex_key(struct inode * inode)3272 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
3273 #endif
3274 extern void unlock_new_inode(struct inode *);
3275 extern void discard_new_inode(struct inode *);
3276 extern unsigned int get_next_ino(void);
3277 extern void evict_inodes(struct super_block *sb);
3278 void dump_mapping(const struct address_space *);
3279
3280 /*
3281 * Userspace may rely on the inode number being non-zero. For example, glibc
3282 * simply ignores files with zero i_ino in unlink() and other places.
3283 *
3284 * As an additional complication, if userspace was compiled with
3285 * _FILE_OFFSET_BITS=32 on a 64-bit kernel we'll only end up reading out the
3286 * lower 32 bits, so we need to check that those aren't zero explicitly. With
3287 * _FILE_OFFSET_BITS=64, this may cause some harmless false-negatives, but
3288 * better safe than sorry.
3289 */
is_zero_ino(ino_t ino)3290 static inline bool is_zero_ino(ino_t ino)
3291 {
3292 return (u32)ino == 0;
3293 }
3294
3295 /*
3296 * inode->i_lock must be held
3297 */
__iget(struct inode * inode)3298 static inline void __iget(struct inode *inode)
3299 {
3300 atomic_inc(&inode->i_count);
3301 }
3302
3303 extern void iget_failed(struct inode *);
3304 extern void clear_inode(struct inode *);
3305 extern void __destroy_inode(struct inode *);
3306 struct inode *alloc_inode(struct super_block *sb);
new_inode_pseudo(struct super_block * sb)3307 static inline struct inode *new_inode_pseudo(struct super_block *sb)
3308 {
3309 return alloc_inode(sb);
3310 }
3311 extern struct inode *new_inode(struct super_block *sb);
3312 extern void free_inode_nonrcu(struct inode *inode);
3313 extern int setattr_should_drop_suidgid(struct mnt_idmap *, struct inode *);
3314 extern int file_remove_privs_flags(struct file *file, unsigned int flags);
3315 extern int file_remove_privs(struct file *);
3316 int setattr_should_drop_sgid(struct mnt_idmap *idmap,
3317 const struct inode *inode);
3318
3319 /*
3320 * This must be used for allocating filesystems specific inodes to set
3321 * up the inode reclaim context correctly.
3322 */
3323 #define alloc_inode_sb(_sb, _cache, _gfp) kmem_cache_alloc_lru(_cache, &_sb->s_inode_lru, _gfp)
3324
3325 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
insert_inode_hash(struct inode * inode)3326 static inline void insert_inode_hash(struct inode *inode)
3327 {
3328 __insert_inode_hash(inode, inode->i_ino);
3329 }
3330
3331 extern void __remove_inode_hash(struct inode *);
remove_inode_hash(struct inode * inode)3332 static inline void remove_inode_hash(struct inode *inode)
3333 {
3334 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
3335 __remove_inode_hash(inode);
3336 }
3337
3338 extern void inode_sb_list_add(struct inode *inode);
3339 extern void inode_add_lru(struct inode *inode);
3340
3341 extern int sb_set_blocksize(struct super_block *, int);
3342 extern int sb_min_blocksize(struct super_block *, int);
3343
3344 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
3345 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
3346 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
3347 int generic_write_checks_count(struct kiocb *iocb, loff_t *count);
3348 extern int generic_write_check_limits(struct file *file, loff_t pos,
3349 loff_t *count);
3350 extern int generic_file_rw_checks(struct file *file_in, struct file *file_out);
3351 ssize_t filemap_read(struct kiocb *iocb, struct iov_iter *to,
3352 ssize_t already_read);
3353 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
3354 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
3355 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
3356 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
3357 ssize_t generic_perform_write(struct kiocb *, struct iov_iter *);
3358 ssize_t direct_write_fallback(struct kiocb *iocb, struct iov_iter *iter,
3359 ssize_t direct_written, ssize_t buffered_written);
3360
3361 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
3362 rwf_t flags);
3363 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
3364 rwf_t flags);
3365 ssize_t vfs_iocb_iter_read(struct file *file, struct kiocb *iocb,
3366 struct iov_iter *iter);
3367 ssize_t vfs_iocb_iter_write(struct file *file, struct kiocb *iocb,
3368 struct iov_iter *iter);
3369
3370 /* fs/splice.c */
3371 ssize_t filemap_splice_read(struct file *in, loff_t *ppos,
3372 struct pipe_inode_info *pipe,
3373 size_t len, unsigned int flags);
3374 ssize_t copy_splice_read(struct file *in, loff_t *ppos,
3375 struct pipe_inode_info *pipe,
3376 size_t len, unsigned int flags);
3377 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
3378 struct file *, loff_t *, size_t, unsigned int);
3379
3380
3381 extern void
3382 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3383 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3384 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3385 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3386 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3387 int whence, loff_t maxsize, loff_t eof);
3388 loff_t generic_llseek_cookie(struct file *file, loff_t offset, int whence,
3389 u64 *cookie);
3390 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3391 int whence, loff_t size);
3392 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3393 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3394 int rw_verify_area(int, struct file *, const loff_t *, size_t);
3395 extern int generic_file_open(struct inode * inode, struct file * filp);
3396 extern int nonseekable_open(struct inode * inode, struct file * filp);
3397 extern int stream_open(struct inode * inode, struct file * filp);
3398
3399 #ifdef CONFIG_BLOCK
3400 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3401 loff_t file_offset);
3402
3403 enum {
3404 /* need locking between buffered and direct access */
3405 DIO_LOCKING = 0x01,
3406
3407 /* filesystem does not support filling holes */
3408 DIO_SKIP_HOLES = 0x02,
3409 };
3410
3411 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3412 struct block_device *bdev, struct iov_iter *iter,
3413 get_block_t get_block,
3414 dio_iodone_t end_io,
3415 int flags);
3416
blockdev_direct_IO(struct kiocb * iocb,struct inode * inode,struct iov_iter * iter,get_block_t get_block)3417 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3418 struct inode *inode,
3419 struct iov_iter *iter,
3420 get_block_t get_block)
3421 {
3422 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3423 get_block, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3424 }
3425 #endif
3426
3427 bool inode_dio_finished(const struct inode *inode);
3428 void inode_dio_wait(struct inode *inode);
3429 void inode_dio_wait_interruptible(struct inode *inode);
3430
3431 /**
3432 * inode_dio_begin - signal start of a direct I/O requests
3433 * @inode: inode the direct I/O happens on
3434 *
3435 * This is called once we've finished processing a direct I/O request,
3436 * and is used to wake up callers waiting for direct I/O to be quiesced.
3437 */
inode_dio_begin(struct inode * inode)3438 static inline void inode_dio_begin(struct inode *inode)
3439 {
3440 atomic_inc(&inode->i_dio_count);
3441 }
3442
3443 /**
3444 * inode_dio_end - signal finish of a direct I/O requests
3445 * @inode: inode the direct I/O happens on
3446 *
3447 * This is called once we've finished processing a direct I/O request,
3448 * and is used to wake up callers waiting for direct I/O to be quiesced.
3449 */
inode_dio_end(struct inode * inode)3450 static inline void inode_dio_end(struct inode *inode)
3451 {
3452 if (atomic_dec_and_test(&inode->i_dio_count))
3453 wake_up_var(&inode->i_dio_count);
3454 }
3455
3456 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3457 unsigned int mask);
3458
3459 extern const struct file_operations generic_ro_fops;
3460
3461 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3462
3463 extern int readlink_copy(char __user *, int, const char *, int);
3464 extern int page_readlink(struct dentry *, char __user *, int);
3465 extern const char *page_get_link_raw(struct dentry *, struct inode *,
3466 struct delayed_call *);
3467 extern const char *page_get_link(struct dentry *, struct inode *,
3468 struct delayed_call *);
3469 extern void page_put_link(void *);
3470 extern int page_symlink(struct inode *inode, const char *symname, int len);
3471 extern const struct inode_operations page_symlink_inode_operations;
3472 extern void kfree_link(void *);
3473 void fill_mg_cmtime(struct kstat *stat, u32 request_mask, struct inode *inode);
3474 void generic_fillattr(struct mnt_idmap *, u32, struct inode *, struct kstat *);
3475 void generic_fill_statx_attr(struct inode *inode, struct kstat *stat);
3476 void generic_fill_statx_atomic_writes(struct kstat *stat,
3477 unsigned int unit_min,
3478 unsigned int unit_max);
3479 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3480 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3481 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3482 void inode_add_bytes(struct inode *inode, loff_t bytes);
3483 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3484 void inode_sub_bytes(struct inode *inode, loff_t bytes);
__inode_get_bytes(struct inode * inode)3485 static inline loff_t __inode_get_bytes(struct inode *inode)
3486 {
3487 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3488 }
3489 loff_t inode_get_bytes(struct inode *inode);
3490 void inode_set_bytes(struct inode *inode, loff_t bytes);
3491 const char *simple_get_link(struct dentry *, struct inode *,
3492 struct delayed_call *);
3493 extern const struct inode_operations simple_symlink_inode_operations;
3494
3495 extern int iterate_dir(struct file *, struct dir_context *);
3496
3497 int vfs_fstatat(int dfd, const char __user *filename, struct kstat *stat,
3498 int flags);
3499 int vfs_fstat(int fd, struct kstat *stat);
3500
vfs_stat(const char __user * filename,struct kstat * stat)3501 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3502 {
3503 return vfs_fstatat(AT_FDCWD, filename, stat, 0);
3504 }
vfs_lstat(const char __user * name,struct kstat * stat)3505 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3506 {
3507 return vfs_fstatat(AT_FDCWD, name, stat, AT_SYMLINK_NOFOLLOW);
3508 }
3509
3510 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3511 extern int vfs_readlink(struct dentry *, char __user *, int);
3512
3513 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3514 extern void put_filesystem(struct file_system_type *fs);
3515 extern struct file_system_type *get_fs_type(const char *name);
3516 extern void drop_super(struct super_block *sb);
3517 extern void drop_super_exclusive(struct super_block *sb);
3518 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3519 extern void iterate_supers_type(struct file_system_type *,
3520 void (*)(struct super_block *, void *), void *);
3521
3522 extern int dcache_dir_open(struct inode *, struct file *);
3523 extern int dcache_dir_close(struct inode *, struct file *);
3524 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3525 extern int dcache_readdir(struct file *, struct dir_context *);
3526 extern int simple_setattr(struct mnt_idmap *, struct dentry *,
3527 struct iattr *);
3528 extern int simple_getattr(struct mnt_idmap *, const struct path *,
3529 struct kstat *, u32, unsigned int);
3530 extern int simple_statfs(struct dentry *, struct kstatfs *);
3531 extern int simple_open(struct inode *inode, struct file *file);
3532 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3533 extern int simple_unlink(struct inode *, struct dentry *);
3534 extern int simple_rmdir(struct inode *, struct dentry *);
3535 void simple_rename_timestamp(struct inode *old_dir, struct dentry *old_dentry,
3536 struct inode *new_dir, struct dentry *new_dentry);
3537 extern int simple_rename_exchange(struct inode *old_dir, struct dentry *old_dentry,
3538 struct inode *new_dir, struct dentry *new_dentry);
3539 extern int simple_rename(struct mnt_idmap *, struct inode *,
3540 struct dentry *, struct inode *, struct dentry *,
3541 unsigned int);
3542 extern void simple_recursive_removal(struct dentry *,
3543 void (*callback)(struct dentry *));
3544 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3545 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3546 extern int simple_empty(struct dentry *);
3547 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3548 loff_t pos, unsigned len,
3549 struct folio **foliop, void **fsdata);
3550 extern const struct address_space_operations ram_aops;
3551 extern int always_delete_dentry(const struct dentry *);
3552 extern struct inode *alloc_anon_inode(struct super_block *);
3553 extern int simple_nosetlease(struct file *, int, struct file_lease **, void **);
3554 extern const struct dentry_operations simple_dentry_operations;
3555
3556 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3557 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3558 extern const struct file_operations simple_dir_operations;
3559 extern const struct inode_operations simple_dir_inode_operations;
3560 extern void make_empty_dir_inode(struct inode *inode);
3561 extern bool is_empty_dir_inode(struct inode *inode);
3562 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3563 struct dentry *d_alloc_name(struct dentry *, const char *);
3564 extern int simple_fill_super(struct super_block *, unsigned long,
3565 const struct tree_descr *);
3566 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3567 extern void simple_release_fs(struct vfsmount **mount, int *count);
3568
3569 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3570 loff_t *ppos, const void *from, size_t available);
3571 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3572 const void __user *from, size_t count);
3573
3574 struct offset_ctx {
3575 struct maple_tree mt;
3576 unsigned long next_offset;
3577 };
3578
3579 void simple_offset_init(struct offset_ctx *octx);
3580 int simple_offset_add(struct offset_ctx *octx, struct dentry *dentry);
3581 void simple_offset_remove(struct offset_ctx *octx, struct dentry *dentry);
3582 int simple_offset_rename(struct inode *old_dir, struct dentry *old_dentry,
3583 struct inode *new_dir, struct dentry *new_dentry);
3584 int simple_offset_rename_exchange(struct inode *old_dir,
3585 struct dentry *old_dentry,
3586 struct inode *new_dir,
3587 struct dentry *new_dentry);
3588 void simple_offset_destroy(struct offset_ctx *octx);
3589
3590 extern const struct file_operations simple_offset_dir_operations;
3591
3592 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3593 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3594
3595 extern int generic_check_addressable(unsigned, u64);
3596
3597 extern void generic_set_sb_d_ops(struct super_block *sb);
3598 extern int generic_ci_match(const struct inode *parent,
3599 const struct qstr *name,
3600 const struct qstr *folded_name,
3601 const u8 *de_name, u32 de_name_len);
3602
3603 #if IS_ENABLED(CONFIG_UNICODE)
3604 int generic_ci_d_hash(const struct dentry *dentry, struct qstr *str);
3605 int generic_ci_d_compare(const struct dentry *dentry, unsigned int len,
3606 const char *str, const struct qstr *name);
3607
3608 /**
3609 * generic_ci_validate_strict_name - Check if a given name is suitable
3610 * for a directory
3611 *
3612 * This functions checks if the proposed filename is valid for the
3613 * parent directory. That means that only valid UTF-8 filenames will be
3614 * accepted for casefold directories from filesystems created with the
3615 * strict encoding flag. That also means that any name will be
3616 * accepted for directories that doesn't have casefold enabled, or
3617 * aren't being strict with the encoding.
3618 *
3619 * @dir: inode of the directory where the new file will be created
3620 * @name: name of the new file
3621 *
3622 * Return:
3623 * * True: if the filename is suitable for this directory. It can be
3624 * true if a given name is not suitable for a strict encoding
3625 * directory, but the directory being used isn't strict
3626 * * False if the filename isn't suitable for this directory. This only
3627 * happens when a directory is casefolded and the filesystem is strict
3628 * about its encoding.
3629 */
generic_ci_validate_strict_name(struct inode * dir,struct qstr * name)3630 static inline bool generic_ci_validate_strict_name(struct inode *dir, struct qstr *name)
3631 {
3632 if (!IS_CASEFOLDED(dir) || !sb_has_strict_encoding(dir->i_sb))
3633 return true;
3634
3635 /*
3636 * A casefold dir must have a encoding set, unless the filesystem
3637 * is corrupted
3638 */
3639 if (WARN_ON_ONCE(!dir->i_sb->s_encoding))
3640 return true;
3641
3642 return !utf8_validate(dir->i_sb->s_encoding, name);
3643 }
3644 #else
generic_ci_validate_strict_name(struct inode * dir,struct qstr * name)3645 static inline bool generic_ci_validate_strict_name(struct inode *dir, struct qstr *name)
3646 {
3647 return true;
3648 }
3649 #endif
3650
sb_has_encoding(const struct super_block * sb)3651 static inline bool sb_has_encoding(const struct super_block *sb)
3652 {
3653 #if IS_ENABLED(CONFIG_UNICODE)
3654 return !!sb->s_encoding;
3655 #else
3656 return false;
3657 #endif
3658 }
3659
3660 int may_setattr(struct mnt_idmap *idmap, struct inode *inode,
3661 unsigned int ia_valid);
3662 int setattr_prepare(struct mnt_idmap *, struct dentry *, struct iattr *);
3663 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3664 void setattr_copy(struct mnt_idmap *, struct inode *inode,
3665 const struct iattr *attr);
3666
3667 extern int file_update_time(struct file *file);
3668
vma_is_dax(const struct vm_area_struct * vma)3669 static inline bool vma_is_dax(const struct vm_area_struct *vma)
3670 {
3671 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3672 }
3673
vma_is_fsdax(struct vm_area_struct * vma)3674 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3675 {
3676 struct inode *inode;
3677
3678 if (!IS_ENABLED(CONFIG_FS_DAX) || !vma->vm_file)
3679 return false;
3680 if (!vma_is_dax(vma))
3681 return false;
3682 inode = file_inode(vma->vm_file);
3683 if (S_ISCHR(inode->i_mode))
3684 return false; /* device-dax */
3685 return true;
3686 }
3687
iocb_flags(struct file * file)3688 static inline int iocb_flags(struct file *file)
3689 {
3690 int res = 0;
3691 if (file->f_flags & O_APPEND)
3692 res |= IOCB_APPEND;
3693 if (file->f_flags & O_DIRECT)
3694 res |= IOCB_DIRECT;
3695 if (file->f_flags & O_DSYNC)
3696 res |= IOCB_DSYNC;
3697 if (file->f_flags & __O_SYNC)
3698 res |= IOCB_SYNC;
3699 return res;
3700 }
3701
kiocb_set_rw_flags(struct kiocb * ki,rwf_t flags,int rw_type)3702 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags,
3703 int rw_type)
3704 {
3705 int kiocb_flags = 0;
3706
3707 /* make sure there's no overlap between RWF and private IOCB flags */
3708 BUILD_BUG_ON((__force int) RWF_SUPPORTED & IOCB_EVENTFD);
3709
3710 if (!flags)
3711 return 0;
3712 if (unlikely(flags & ~RWF_SUPPORTED))
3713 return -EOPNOTSUPP;
3714 if (unlikely((flags & RWF_APPEND) && (flags & RWF_NOAPPEND)))
3715 return -EINVAL;
3716
3717 if (flags & RWF_NOWAIT) {
3718 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3719 return -EOPNOTSUPP;
3720 }
3721 if (flags & RWF_ATOMIC) {
3722 if (rw_type != WRITE)
3723 return -EOPNOTSUPP;
3724 if (!(ki->ki_filp->f_mode & FMODE_CAN_ATOMIC_WRITE))
3725 return -EOPNOTSUPP;
3726 }
3727 if (flags & RWF_DONTCACHE) {
3728 /* file system must support it */
3729 if (!(ki->ki_filp->f_op->fop_flags & FOP_DONTCACHE))
3730 return -EOPNOTSUPP;
3731 /* DAX mappings not supported */
3732 if (IS_DAX(ki->ki_filp->f_mapping->host))
3733 return -EOPNOTSUPP;
3734 }
3735 kiocb_flags |= (__force int) (flags & RWF_SUPPORTED);
3736 if (flags & RWF_SYNC)
3737 kiocb_flags |= IOCB_DSYNC;
3738
3739 if ((flags & RWF_NOAPPEND) && (ki->ki_flags & IOCB_APPEND)) {
3740 if (IS_APPEND(file_inode(ki->ki_filp)))
3741 return -EPERM;
3742 ki->ki_flags &= ~IOCB_APPEND;
3743 }
3744
3745 ki->ki_flags |= kiocb_flags;
3746 return 0;
3747 }
3748
3749 /* Transaction based IO helpers */
3750
3751 /*
3752 * An argresp is stored in an allocated page and holds the
3753 * size of the argument or response, along with its content
3754 */
3755 struct simple_transaction_argresp {
3756 ssize_t size;
3757 char data[];
3758 };
3759
3760 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3761
3762 char *simple_transaction_get(struct file *file, const char __user *buf,
3763 size_t size);
3764 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3765 size_t size, loff_t *pos);
3766 int simple_transaction_release(struct inode *inode, struct file *file);
3767
3768 void simple_transaction_set(struct file *file, size_t n);
3769
3770 /*
3771 * simple attribute files
3772 *
3773 * These attributes behave similar to those in sysfs:
3774 *
3775 * Writing to an attribute immediately sets a value, an open file can be
3776 * written to multiple times.
3777 *
3778 * Reading from an attribute creates a buffer from the value that might get
3779 * read with multiple read calls. When the attribute has been read
3780 * completely, no further read calls are possible until the file is opened
3781 * again.
3782 *
3783 * All attributes contain a text representation of a numeric value
3784 * that are accessed with the get() and set() functions.
3785 */
3786 #define DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, __is_signed) \
3787 static int __fops ## _open(struct inode *inode, struct file *file) \
3788 { \
3789 __simple_attr_check_format(__fmt, 0ull); \
3790 return simple_attr_open(inode, file, __get, __set, __fmt); \
3791 } \
3792 static const struct file_operations __fops = { \
3793 .owner = THIS_MODULE, \
3794 .open = __fops ## _open, \
3795 .release = simple_attr_release, \
3796 .read = simple_attr_read, \
3797 .write = (__is_signed) ? simple_attr_write_signed : simple_attr_write, \
3798 .llseek = generic_file_llseek, \
3799 }
3800
3801 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3802 DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, false)
3803
3804 #define DEFINE_SIMPLE_ATTRIBUTE_SIGNED(__fops, __get, __set, __fmt) \
3805 DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, true)
3806
3807 static inline __printf(1, 2)
__simple_attr_check_format(const char * fmt,...)3808 void __simple_attr_check_format(const char *fmt, ...)
3809 {
3810 /* don't do anything, just let the compiler check the arguments; */
3811 }
3812
3813 int simple_attr_open(struct inode *inode, struct file *file,
3814 int (*get)(void *, u64 *), int (*set)(void *, u64),
3815 const char *fmt);
3816 int simple_attr_release(struct inode *inode, struct file *file);
3817 ssize_t simple_attr_read(struct file *file, char __user *buf,
3818 size_t len, loff_t *ppos);
3819 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3820 size_t len, loff_t *ppos);
3821 ssize_t simple_attr_write_signed(struct file *file, const char __user *buf,
3822 size_t len, loff_t *ppos);
3823
3824 struct ctl_table;
3825 int __init list_bdev_fs_names(char *buf, size_t size);
3826
3827 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3828
3829 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3830 #define OPEN_FMODE(flag) ((__force fmode_t)((flag + 1) & O_ACCMODE))
3831
is_sxid(umode_t mode)3832 static inline bool is_sxid(umode_t mode)
3833 {
3834 return mode & (S_ISUID | S_ISGID);
3835 }
3836
check_sticky(struct mnt_idmap * idmap,struct inode * dir,struct inode * inode)3837 static inline int check_sticky(struct mnt_idmap *idmap,
3838 struct inode *dir, struct inode *inode)
3839 {
3840 if (!(dir->i_mode & S_ISVTX))
3841 return 0;
3842
3843 return __check_sticky(idmap, dir, inode);
3844 }
3845
inode_has_no_xattr(struct inode * inode)3846 static inline void inode_has_no_xattr(struct inode *inode)
3847 {
3848 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3849 inode->i_flags |= S_NOSEC;
3850 }
3851
is_root_inode(struct inode * inode)3852 static inline bool is_root_inode(struct inode *inode)
3853 {
3854 return inode == inode->i_sb->s_root->d_inode;
3855 }
3856
dir_emit(struct dir_context * ctx,const char * name,int namelen,u64 ino,unsigned type)3857 static inline bool dir_emit(struct dir_context *ctx,
3858 const char *name, int namelen,
3859 u64 ino, unsigned type)
3860 {
3861 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type);
3862 }
dir_emit_dot(struct file * file,struct dir_context * ctx)3863 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3864 {
3865 return ctx->actor(ctx, ".", 1, ctx->pos,
3866 file->f_path.dentry->d_inode->i_ino, DT_DIR);
3867 }
dir_emit_dotdot(struct file * file,struct dir_context * ctx)3868 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3869 {
3870 return ctx->actor(ctx, "..", 2, ctx->pos,
3871 d_parent_ino(file->f_path.dentry), DT_DIR);
3872 }
dir_emit_dots(struct file * file,struct dir_context * ctx)3873 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3874 {
3875 if (ctx->pos == 0) {
3876 if (!dir_emit_dot(file, ctx))
3877 return false;
3878 ctx->pos = 1;
3879 }
3880 if (ctx->pos == 1) {
3881 if (!dir_emit_dotdot(file, ctx))
3882 return false;
3883 ctx->pos = 2;
3884 }
3885 return true;
3886 }
dir_relax(struct inode * inode)3887 static inline bool dir_relax(struct inode *inode)
3888 {
3889 inode_unlock(inode);
3890 inode_lock(inode);
3891 return !IS_DEADDIR(inode);
3892 }
3893
dir_relax_shared(struct inode * inode)3894 static inline bool dir_relax_shared(struct inode *inode)
3895 {
3896 inode_unlock_shared(inode);
3897 inode_lock_shared(inode);
3898 return !IS_DEADDIR(inode);
3899 }
3900
3901 extern bool path_noexec(const struct path *path);
3902 extern void inode_nohighmem(struct inode *inode);
3903
3904 /* mm/fadvise.c */
3905 extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len,
3906 int advice);
3907 extern int generic_fadvise(struct file *file, loff_t offset, loff_t len,
3908 int advice);
3909
vfs_empty_path(int dfd,const char __user * path)3910 static inline bool vfs_empty_path(int dfd, const char __user *path)
3911 {
3912 char c;
3913
3914 if (dfd < 0)
3915 return false;
3916
3917 /* We now allow NULL to be used for empty path. */
3918 if (!path)
3919 return true;
3920
3921 if (unlikely(get_user(c, path)))
3922 return false;
3923
3924 return !c;
3925 }
3926
3927 int generic_atomic_write_valid(struct kiocb *iocb, struct iov_iter *iter);
3928
3929 #endif /* _LINUX_FS_H */
3930