Branch data Line data Source code
1 : : /* SPDX-License-Identifier: GPL-2.0 */
2 : : #ifndef _LINUX_FS_H
3 : : #define _LINUX_FS_H
4 : :
5 : : #include <linux/linkage.h>
6 : : #include <linux/wait_bit.h>
7 : : #include <linux/kdev_t.h>
8 : : #include <linux/dcache.h>
9 : : #include <linux/path.h>
10 : : #include <linux/stat.h>
11 : : #include <linux/cache.h>
12 : : #include <linux/list.h>
13 : : #include <linux/list_lru.h>
14 : : #include <linux/llist.h>
15 : : #include <linux/radix-tree.h>
16 : : #include <linux/xarray.h>
17 : : #include <linux/rbtree.h>
18 : : #include <linux/init.h>
19 : : #include <linux/pid.h>
20 : : #include <linux/bug.h>
21 : : #include <linux/mutex.h>
22 : : #include <linux/rwsem.h>
23 : : #include <linux/mm_types.h>
24 : : #include <linux/capability.h>
25 : : #include <linux/semaphore.h>
26 : : #include <linux/fcntl.h>
27 : : #include <linux/fiemap.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 : :
44 : : #include <asm/byteorder.h>
45 : : #include <uapi/linux/fs.h>
46 : :
47 : : struct backing_dev_info;
48 : : struct bdi_writeback;
49 : : struct bio;
50 : : struct export_operations;
51 : : struct hd_geometry;
52 : : struct iovec;
53 : : struct kiocb;
54 : : struct kobject;
55 : : struct pipe_inode_info;
56 : : struct poll_table_struct;
57 : : struct kstatfs;
58 : : struct vm_area_struct;
59 : : struct vfsmount;
60 : : struct cred;
61 : : struct swap_info_struct;
62 : : struct seq_file;
63 : : struct workqueue_struct;
64 : : struct iov_iter;
65 : : struct fscrypt_info;
66 : : struct fscrypt_operations;
67 : : struct fsverity_info;
68 : : struct fsverity_operations;
69 : : struct fs_context;
70 : : struct fs_parameter_description;
71 : :
72 : : extern void __init inode_init(void);
73 : : extern void __init inode_init_early(void);
74 : : extern void __init files_init(void);
75 : : extern void __init files_maxfiles_init(void);
76 : :
77 : : extern struct files_stat_struct files_stat;
78 : : extern unsigned long get_max_files(void);
79 : : extern unsigned int sysctl_nr_open;
80 : : extern struct inodes_stat_t inodes_stat;
81 : : extern int leases_enable, lease_break_time;
82 : : extern int sysctl_protected_symlinks;
83 : : extern int sysctl_protected_hardlinks;
84 : : extern int sysctl_protected_fifos;
85 : : extern int sysctl_protected_regular;
86 : :
87 : : typedef __kernel_rwf_t rwf_t;
88 : :
89 : : struct buffer_head;
90 : : typedef int (get_block_t)(struct inode *inode, sector_t iblock,
91 : : struct buffer_head *bh_result, int create);
92 : : typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
93 : : ssize_t bytes, void *private);
94 : :
95 : : #define MAY_EXEC 0x00000001
96 : : #define MAY_WRITE 0x00000002
97 : : #define MAY_READ 0x00000004
98 : : #define MAY_APPEND 0x00000008
99 : : #define MAY_ACCESS 0x00000010
100 : : #define MAY_OPEN 0x00000020
101 : : #define MAY_CHDIR 0x00000040
102 : : /* called from RCU mode, don't block */
103 : : #define MAY_NOT_BLOCK 0x00000080
104 : :
105 : : /*
106 : : * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
107 : : * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
108 : : */
109 : :
110 : : /* file is open for reading */
111 : : #define FMODE_READ ((__force fmode_t)0x1)
112 : : /* file is open for writing */
113 : : #define FMODE_WRITE ((__force fmode_t)0x2)
114 : : /* file is seekable */
115 : : #define FMODE_LSEEK ((__force fmode_t)0x4)
116 : : /* file can be accessed using pread */
117 : : #define FMODE_PREAD ((__force fmode_t)0x8)
118 : : /* file can be accessed using pwrite */
119 : : #define FMODE_PWRITE ((__force fmode_t)0x10)
120 : : /* File is opened for execution with sys_execve / sys_uselib */
121 : : #define FMODE_EXEC ((__force fmode_t)0x20)
122 : : /* File is opened with O_NDELAY (only set for block devices) */
123 : : #define FMODE_NDELAY ((__force fmode_t)0x40)
124 : : /* File is opened with O_EXCL (only set for block devices) */
125 : : #define FMODE_EXCL ((__force fmode_t)0x80)
126 : : /* File is opened using open(.., 3, ..) and is writeable only for ioctls
127 : : (specialy hack for floppy.c) */
128 : : #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
129 : : /* 32bit hashes as llseek() offset (for directories) */
130 : : #define FMODE_32BITHASH ((__force fmode_t)0x200)
131 : : /* 64bit hashes as llseek() offset (for directories) */
132 : : #define FMODE_64BITHASH ((__force fmode_t)0x400)
133 : :
134 : : /*
135 : : * Don't update ctime and mtime.
136 : : *
137 : : * Currently a special hack for the XFS open_by_handle ioctl, but we'll
138 : : * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
139 : : */
140 : : #define FMODE_NOCMTIME ((__force fmode_t)0x800)
141 : :
142 : : /* Expect random access pattern */
143 : : #define FMODE_RANDOM ((__force fmode_t)0x1000)
144 : :
145 : : /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
146 : : #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
147 : :
148 : : /* File is opened with O_PATH; almost nothing can be done with it */
149 : : #define FMODE_PATH ((__force fmode_t)0x4000)
150 : :
151 : : /* File needs atomic accesses to f_pos */
152 : : #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
153 : : /* Write access to underlying fs */
154 : : #define FMODE_WRITER ((__force fmode_t)0x10000)
155 : : /* Has read method(s) */
156 : : #define FMODE_CAN_READ ((__force fmode_t)0x20000)
157 : : /* Has write method(s) */
158 : : #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
159 : :
160 : : #define FMODE_OPENED ((__force fmode_t)0x80000)
161 : : #define FMODE_CREATED ((__force fmode_t)0x100000)
162 : :
163 : : /* File is stream-like */
164 : : #define FMODE_STREAM ((__force fmode_t)0x200000)
165 : :
166 : : /* File was opened by fanotify and shouldn't generate fanotify events */
167 : : #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
168 : :
169 : : /* File is capable of returning -EAGAIN if I/O will block */
170 : : #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
171 : :
172 : : /* File represents mount that needs unmounting */
173 : : #define FMODE_NEED_UNMOUNT ((__force fmode_t)0x10000000)
174 : :
175 : : /* File does not contribute to nr_files count */
176 : : #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000)
177 : :
178 : : /*
179 : : * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
180 : : * that indicates that they should check the contents of the iovec are
181 : : * valid, but not check the memory that the iovec elements
182 : : * points too.
183 : : */
184 : : #define CHECK_IOVEC_ONLY -1
185 : :
186 : : /*
187 : : * Attribute flags. These should be or-ed together to figure out what
188 : : * has been changed!
189 : : */
190 : : #define ATTR_MODE (1 << 0)
191 : : #define ATTR_UID (1 << 1)
192 : : #define ATTR_GID (1 << 2)
193 : : #define ATTR_SIZE (1 << 3)
194 : : #define ATTR_ATIME (1 << 4)
195 : : #define ATTR_MTIME (1 << 5)
196 : : #define ATTR_CTIME (1 << 6)
197 : : #define ATTR_ATIME_SET (1 << 7)
198 : : #define ATTR_MTIME_SET (1 << 8)
199 : : #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
200 : : #define ATTR_KILL_SUID (1 << 11)
201 : : #define ATTR_KILL_SGID (1 << 12)
202 : : #define ATTR_FILE (1 << 13)
203 : : #define ATTR_KILL_PRIV (1 << 14)
204 : : #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
205 : : #define ATTR_TIMES_SET (1 << 16)
206 : : #define ATTR_TOUCH (1 << 17)
207 : :
208 : : /*
209 : : * Whiteout is represented by a char device. The following constants define the
210 : : * mode and device number to use.
211 : : */
212 : : #define WHITEOUT_MODE 0
213 : : #define WHITEOUT_DEV 0
214 : :
215 : : /*
216 : : * This is the Inode Attributes structure, used for notify_change(). It
217 : : * uses the above definitions as flags, to know which values have changed.
218 : : * Also, in this manner, a Filesystem can look at only the values it cares
219 : : * about. Basically, these are the attributes that the VFS layer can
220 : : * request to change from the FS layer.
221 : : *
222 : : * Derek Atkins <warlord@MIT.EDU> 94-10-20
223 : : */
224 : : struct iattr {
225 : : unsigned int ia_valid;
226 : : umode_t ia_mode;
227 : : kuid_t ia_uid;
228 : : kgid_t ia_gid;
229 : : loff_t ia_size;
230 : : struct timespec64 ia_atime;
231 : : struct timespec64 ia_mtime;
232 : : struct timespec64 ia_ctime;
233 : :
234 : : /*
235 : : * Not an attribute, but an auxiliary info for filesystems wanting to
236 : : * implement an ftruncate() like method. NOTE: filesystem should
237 : : * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
238 : : */
239 : : struct file *ia_file;
240 : : };
241 : :
242 : : /*
243 : : * Includes for diskquotas.
244 : : */
245 : : #include <linux/quota.h>
246 : :
247 : : /*
248 : : * Maximum number of layers of fs stack. Needs to be limited to
249 : : * prevent kernel stack overflow
250 : : */
251 : : #define FILESYSTEM_MAX_STACK_DEPTH 2
252 : :
253 : : /**
254 : : * enum positive_aop_returns - aop return codes with specific semantics
255 : : *
256 : : * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
257 : : * completed, that the page is still locked, and
258 : : * should be considered active. The VM uses this hint
259 : : * to return the page to the active list -- it won't
260 : : * be a candidate for writeback again in the near
261 : : * future. Other callers must be careful to unlock
262 : : * the page if they get this return. Returned by
263 : : * writepage();
264 : : *
265 : : * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
266 : : * unlocked it and the page might have been truncated.
267 : : * The caller should back up to acquiring a new page and
268 : : * trying again. The aop will be taking reasonable
269 : : * precautions not to livelock. If the caller held a page
270 : : * reference, it should drop it before retrying. Returned
271 : : * by readpage().
272 : : *
273 : : * address_space_operation functions return these large constants to indicate
274 : : * special semantics to the caller. These are much larger than the bytes in a
275 : : * page to allow for functions that return the number of bytes operated on in a
276 : : * given page.
277 : : */
278 : :
279 : : enum positive_aop_returns {
280 : : AOP_WRITEPAGE_ACTIVATE = 0x80000,
281 : : AOP_TRUNCATED_PAGE = 0x80001,
282 : : };
283 : :
284 : : #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
285 : : #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
286 : : * helper code (eg buffer layer)
287 : : * to clear GFP_FS from alloc */
288 : :
289 : : /*
290 : : * oh the beauties of C type declarations.
291 : : */
292 : : struct page;
293 : : struct address_space;
294 : : struct writeback_control;
295 : :
296 : : /*
297 : : * Write life time hint values.
298 : : * Stored in struct inode as u8.
299 : : */
300 : : enum rw_hint {
301 : : WRITE_LIFE_NOT_SET = 0,
302 : : WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
303 : : WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
304 : : WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
305 : : WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
306 : : WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
307 : : };
308 : :
309 : : #define IOCB_EVENTFD (1 << 0)
310 : : #define IOCB_APPEND (1 << 1)
311 : : #define IOCB_DIRECT (1 << 2)
312 : : #define IOCB_HIPRI (1 << 3)
313 : : #define IOCB_DSYNC (1 << 4)
314 : : #define IOCB_SYNC (1 << 5)
315 : : #define IOCB_WRITE (1 << 6)
316 : : #define IOCB_NOWAIT (1 << 7)
317 : :
318 : : struct kiocb {
319 : : struct file *ki_filp;
320 : :
321 : : /* The 'ki_filp' pointer is shared in a union for aio */
322 : : randomized_struct_fields_start
323 : :
324 : : loff_t ki_pos;
325 : : void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
326 : : void *private;
327 : : int ki_flags;
328 : : u16 ki_hint;
329 : : u16 ki_ioprio; /* See linux/ioprio.h */
330 : : unsigned int ki_cookie; /* for ->iopoll */
331 : :
332 : : randomized_struct_fields_end
333 : : };
334 : :
335 : : static inline bool is_sync_kiocb(struct kiocb *kiocb)
336 : : {
337 : 0 : return kiocb->ki_complete == NULL;
338 : : }
339 : :
340 : : /*
341 : : * "descriptor" for what we're up to with a read.
342 : : * This allows us to use the same read code yet
343 : : * have multiple different users of the data that
344 : : * we read from a file.
345 : : *
346 : : * The simplest case just copies the data to user
347 : : * mode.
348 : : */
349 : : typedef struct {
350 : : size_t written;
351 : : size_t count;
352 : : union {
353 : : char __user *buf;
354 : : void *data;
355 : : } arg;
356 : : int error;
357 : : } read_descriptor_t;
358 : :
359 : : typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
360 : : unsigned long, unsigned long);
361 : :
362 : : struct address_space_operations {
363 : : int (*writepage)(struct page *page, struct writeback_control *wbc);
364 : : int (*readpage)(struct file *, struct page *);
365 : :
366 : : /* Write back some dirty pages from this mapping. */
367 : : int (*writepages)(struct address_space *, struct writeback_control *);
368 : :
369 : : /* Set a page dirty. Return true if this dirtied it */
370 : : int (*set_page_dirty)(struct page *page);
371 : :
372 : : /*
373 : : * Reads in the requested pages. Unlike ->readpage(), this is
374 : : * PURELY used for read-ahead!.
375 : : */
376 : : int (*readpages)(struct file *filp, struct address_space *mapping,
377 : : struct list_head *pages, unsigned nr_pages);
378 : :
379 : : int (*write_begin)(struct file *, struct address_space *mapping,
380 : : loff_t pos, unsigned len, unsigned flags,
381 : : struct page **pagep, void **fsdata);
382 : : int (*write_end)(struct file *, struct address_space *mapping,
383 : : loff_t pos, unsigned len, unsigned copied,
384 : : struct page *page, void *fsdata);
385 : :
386 : : /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
387 : : sector_t (*bmap)(struct address_space *, sector_t);
388 : : void (*invalidatepage) (struct page *, unsigned int, unsigned int);
389 : : int (*releasepage) (struct page *, gfp_t);
390 : : void (*freepage)(struct page *);
391 : : ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
392 : : /*
393 : : * migrate the contents of a page to the specified target. If
394 : : * migrate_mode is MIGRATE_ASYNC, it must not block.
395 : : */
396 : : int (*migratepage) (struct address_space *,
397 : : struct page *, struct page *, enum migrate_mode);
398 : : bool (*isolate_page)(struct page *, isolate_mode_t);
399 : : void (*putback_page)(struct page *);
400 : : int (*launder_page) (struct page *);
401 : : int (*is_partially_uptodate) (struct page *, unsigned long,
402 : : unsigned long);
403 : : void (*is_dirty_writeback) (struct page *, bool *, bool *);
404 : : int (*error_remove_page)(struct address_space *, struct page *);
405 : :
406 : : /* swapfile support */
407 : : int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
408 : : sector_t *span);
409 : : void (*swap_deactivate)(struct file *file);
410 : : };
411 : :
412 : : extern const struct address_space_operations empty_aops;
413 : :
414 : : /*
415 : : * pagecache_write_begin/pagecache_write_end must be used by general code
416 : : * to write into the pagecache.
417 : : */
418 : : int pagecache_write_begin(struct file *, struct address_space *mapping,
419 : : loff_t pos, unsigned len, unsigned flags,
420 : : struct page **pagep, void **fsdata);
421 : :
422 : : int pagecache_write_end(struct file *, struct address_space *mapping,
423 : : loff_t pos, unsigned len, unsigned copied,
424 : : struct page *page, void *fsdata);
425 : :
426 : : /**
427 : : * struct address_space - Contents of a cacheable, mappable object.
428 : : * @host: Owner, either the inode or the block_device.
429 : : * @i_pages: Cached pages.
430 : : * @gfp_mask: Memory allocation flags to use for allocating pages.
431 : : * @i_mmap_writable: Number of VM_SHARED mappings.
432 : : * @nr_thps: Number of THPs in the pagecache (non-shmem only).
433 : : * @i_mmap: Tree of private and shared mappings.
434 : : * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
435 : : * @nrpages: Number of page entries, protected by the i_pages lock.
436 : : * @nrexceptional: Shadow or DAX entries, protected by the i_pages lock.
437 : : * @writeback_index: Writeback starts here.
438 : : * @a_ops: Methods.
439 : : * @flags: Error bits and flags (AS_*).
440 : : * @wb_err: The most recent error which has occurred.
441 : : * @private_lock: For use by the owner of the address_space.
442 : : * @private_list: For use by the owner of the address_space.
443 : : * @private_data: For use by the owner of the address_space.
444 : : */
445 : : struct address_space {
446 : : struct inode *host;
447 : : struct xarray i_pages;
448 : : gfp_t gfp_mask;
449 : : atomic_t i_mmap_writable;
450 : : #ifdef CONFIG_READ_ONLY_THP_FOR_FS
451 : : /* number of thp, only for non-shmem files */
452 : : atomic_t nr_thps;
453 : : #endif
454 : : struct rb_root_cached i_mmap;
455 : : struct rw_semaphore i_mmap_rwsem;
456 : : unsigned long nrpages;
457 : : unsigned long nrexceptional;
458 : : pgoff_t writeback_index;
459 : : const struct address_space_operations *a_ops;
460 : : unsigned long flags;
461 : : errseq_t wb_err;
462 : : spinlock_t private_lock;
463 : : struct list_head private_list;
464 : : void *private_data;
465 : : } __attribute__((aligned(sizeof(long)))) __randomize_layout;
466 : : /*
467 : : * On most architectures that alignment is already the case; but
468 : : * must be enforced here for CRIS, to let the least significant bit
469 : : * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
470 : : */
471 : : struct request_queue;
472 : :
473 : : struct block_device {
474 : : dev_t bd_dev; /* not a kdev_t - it's a search key */
475 : : int bd_openers;
476 : : struct inode * bd_inode; /* will die */
477 : : struct super_block * bd_super;
478 : : struct mutex bd_mutex; /* open/close mutex */
479 : : void * bd_claiming;
480 : : void * bd_holder;
481 : : int bd_holders;
482 : : bool bd_write_holder;
483 : : #ifdef CONFIG_SYSFS
484 : : struct list_head bd_holder_disks;
485 : : #endif
486 : : struct block_device * bd_contains;
487 : : unsigned bd_block_size;
488 : : u8 bd_partno;
489 : : struct hd_struct * bd_part;
490 : : /* number of times partitions within this device have been opened. */
491 : : unsigned bd_part_count;
492 : : int bd_invalidated;
493 : : struct gendisk * bd_disk;
494 : : struct request_queue * bd_queue;
495 : : struct backing_dev_info *bd_bdi;
496 : : struct list_head bd_list;
497 : : /*
498 : : * Private data. You must have bd_claim'ed the block_device
499 : : * to use this. NOTE: bd_claim allows an owner to claim
500 : : * the same device multiple times, the owner must take special
501 : : * care to not mess up bd_private for that case.
502 : : */
503 : : unsigned long bd_private;
504 : :
505 : : /* The counter of freeze processes */
506 : : int bd_fsfreeze_count;
507 : : /* Mutex for freeze */
508 : : struct mutex bd_fsfreeze_mutex;
509 : : } __randomize_layout;
510 : :
511 : : /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
512 : : #define PAGECACHE_TAG_DIRTY XA_MARK_0
513 : : #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
514 : : #define PAGECACHE_TAG_TOWRITE XA_MARK_2
515 : :
516 : : /*
517 : : * Returns true if any of the pages in the mapping are marked with the tag.
518 : : */
519 : : static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag)
520 : : {
521 : : return xa_marked(&mapping->i_pages, tag);
522 : : }
523 : :
524 : : static inline void i_mmap_lock_write(struct address_space *mapping)
525 : : {
526 : 3 : down_write(&mapping->i_mmap_rwsem);
527 : : }
528 : :
529 : : static inline void i_mmap_unlock_write(struct address_space *mapping)
530 : : {
531 : 3 : up_write(&mapping->i_mmap_rwsem);
532 : : }
533 : :
534 : : static inline void i_mmap_lock_read(struct address_space *mapping)
535 : : {
536 : 2 : down_read(&mapping->i_mmap_rwsem);
537 : : }
538 : :
539 : : static inline void i_mmap_unlock_read(struct address_space *mapping)
540 : : {
541 : 2 : up_read(&mapping->i_mmap_rwsem);
542 : : }
543 : :
544 : : /*
545 : : * Might pages of this file be mapped into userspace?
546 : : */
547 : : static inline int mapping_mapped(struct address_space *mapping)
548 : : {
549 : 0 : return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
550 : : }
551 : :
552 : : /*
553 : : * Might pages of this file have been modified in userspace?
554 : : * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
555 : : * marks vma as VM_SHARED if it is shared, and the file was opened for
556 : : * writing i.e. vma may be mprotected writable even if now readonly.
557 : : *
558 : : * If i_mmap_writable is negative, no new writable mappings are allowed. You
559 : : * can only deny writable mappings, if none exists right now.
560 : : */
561 : : static inline int mapping_writably_mapped(struct address_space *mapping)
562 : : {
563 : 3 : return atomic_read(&mapping->i_mmap_writable) > 0;
564 : : }
565 : :
566 : : static inline int mapping_map_writable(struct address_space *mapping)
567 : : {
568 : 3 : return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
569 : 3 : 0 : -EPERM;
570 : : }
571 : :
572 : : static inline void mapping_unmap_writable(struct address_space *mapping)
573 : : {
574 : 3 : atomic_dec(&mapping->i_mmap_writable);
575 : : }
576 : :
577 : : static inline int mapping_deny_writable(struct address_space *mapping)
578 : : {
579 : 0 : return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
580 : 0 : 0 : -EBUSY;
581 : : }
582 : :
583 : : static inline void mapping_allow_writable(struct address_space *mapping)
584 : : {
585 : 0 : atomic_inc(&mapping->i_mmap_writable);
586 : : }
587 : :
588 : : /*
589 : : * Use sequence counter to get consistent i_size on 32-bit processors.
590 : : */
591 : : #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
592 : : #include <linux/seqlock.h>
593 : : #define __NEED_I_SIZE_ORDERED
594 : : #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
595 : : #else
596 : : #define i_size_ordered_init(inode) do { } while (0)
597 : : #endif
598 : :
599 : : struct posix_acl;
600 : : #define ACL_NOT_CACHED ((void *)(-1))
601 : : #define ACL_DONT_CACHE ((void *)(-3))
602 : :
603 : : static inline struct posix_acl *
604 : : uncached_acl_sentinel(struct task_struct *task)
605 : : {
606 : 3 : return (void *)task + 1;
607 : : }
608 : :
609 : : static inline bool
610 : : is_uncached_acl(struct posix_acl *acl)
611 : : {
612 : 3 : return (long)acl & 1;
613 : : }
614 : :
615 : : #define IOP_FASTPERM 0x0001
616 : : #define IOP_LOOKUP 0x0002
617 : : #define IOP_NOFOLLOW 0x0004
618 : : #define IOP_XATTR 0x0008
619 : : #define IOP_DEFAULT_READLINK 0x0010
620 : :
621 : : struct fsnotify_mark_connector;
622 : :
623 : : /*
624 : : * Keep mostly read-only and often accessed (especially for
625 : : * the RCU path lookup and 'stat' data) fields at the beginning
626 : : * of the 'struct inode'
627 : : */
628 : : struct inode {
629 : : umode_t i_mode;
630 : : unsigned short i_opflags;
631 : : kuid_t i_uid;
632 : : kgid_t i_gid;
633 : : unsigned int i_flags;
634 : :
635 : : #ifdef CONFIG_FS_POSIX_ACL
636 : : struct posix_acl *i_acl;
637 : : struct posix_acl *i_default_acl;
638 : : #endif
639 : :
640 : : const struct inode_operations *i_op;
641 : : struct super_block *i_sb;
642 : : struct address_space *i_mapping;
643 : :
644 : : #ifdef CONFIG_SECURITY
645 : : void *i_security;
646 : : #endif
647 : :
648 : : /* Stat data, not accessed from path walking */
649 : : unsigned long i_ino;
650 : : /*
651 : : * Filesystems may only read i_nlink directly. They shall use the
652 : : * following functions for modification:
653 : : *
654 : : * (set|clear|inc|drop)_nlink
655 : : * inode_(inc|dec)_link_count
656 : : */
657 : : union {
658 : : const unsigned int i_nlink;
659 : : unsigned int __i_nlink;
660 : : };
661 : : dev_t i_rdev;
662 : : loff_t i_size;
663 : : struct timespec64 i_atime;
664 : : struct timespec64 i_mtime;
665 : : struct timespec64 i_ctime;
666 : : spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
667 : : unsigned short i_bytes;
668 : : u8 i_blkbits;
669 : : u8 i_write_hint;
670 : : blkcnt_t i_blocks;
671 : :
672 : : #ifdef __NEED_I_SIZE_ORDERED
673 : : seqcount_t i_size_seqcount;
674 : : #endif
675 : :
676 : : /* Misc */
677 : : unsigned long i_state;
678 : : struct rw_semaphore i_rwsem;
679 : :
680 : : unsigned long dirtied_when; /* jiffies of first dirtying */
681 : : unsigned long dirtied_time_when;
682 : :
683 : : struct hlist_node i_hash;
684 : : struct list_head i_io_list; /* backing dev IO list */
685 : : #ifdef CONFIG_CGROUP_WRITEBACK
686 : : struct bdi_writeback *i_wb; /* the associated cgroup wb */
687 : :
688 : : /* foreign inode detection, see wbc_detach_inode() */
689 : : int i_wb_frn_winner;
690 : : u16 i_wb_frn_avg_time;
691 : : u16 i_wb_frn_history;
692 : : #endif
693 : : struct list_head i_lru; /* inode LRU list */
694 : : struct list_head i_sb_list;
695 : : struct list_head i_wb_list; /* backing dev writeback list */
696 : : union {
697 : : struct hlist_head i_dentry;
698 : : struct rcu_head i_rcu;
699 : : };
700 : : atomic64_t i_version;
701 : : atomic64_t i_sequence; /* see futex */
702 : : atomic_t i_count;
703 : : atomic_t i_dio_count;
704 : : atomic_t i_writecount;
705 : : #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
706 : : atomic_t i_readcount; /* struct files open RO */
707 : : #endif
708 : : union {
709 : : const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
710 : : void (*free_inode)(struct inode *);
711 : : };
712 : : struct file_lock_context *i_flctx;
713 : : struct address_space i_data;
714 : : struct list_head i_devices;
715 : : union {
716 : : struct pipe_inode_info *i_pipe;
717 : : struct block_device *i_bdev;
718 : : struct cdev *i_cdev;
719 : : char *i_link;
720 : : unsigned i_dir_seq;
721 : : };
722 : :
723 : : __u32 i_generation;
724 : :
725 : : #ifdef CONFIG_FSNOTIFY
726 : : __u32 i_fsnotify_mask; /* all events this inode cares about */
727 : : struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
728 : : #endif
729 : :
730 : : #ifdef CONFIG_FS_ENCRYPTION
731 : : struct fscrypt_info *i_crypt_info;
732 : : #endif
733 : :
734 : : #ifdef CONFIG_FS_VERITY
735 : : struct fsverity_info *i_verity_info;
736 : : #endif
737 : :
738 : : void *i_private; /* fs or device private pointer */
739 : : } __randomize_layout;
740 : :
741 : : struct timespec64 timestamp_truncate(struct timespec64 t, struct inode *inode);
742 : :
743 : : static inline unsigned int i_blocksize(const struct inode *node)
744 : : {
745 : 3 : return (1 << node->i_blkbits);
746 : : }
747 : :
748 : : static inline int inode_unhashed(struct inode *inode)
749 : : {
750 : : return hlist_unhashed(&inode->i_hash);
751 : : }
752 : :
753 : : /*
754 : : * __mark_inode_dirty expects inodes to be hashed. Since we don't
755 : : * want special inodes in the fileset inode space, we make them
756 : : * appear hashed, but do not put on any lists. hlist_del()
757 : : * will work fine and require no locking.
758 : : */
759 : : static inline void inode_fake_hash(struct inode *inode)
760 : : {
761 : : hlist_add_fake(&inode->i_hash);
762 : : }
763 : :
764 : : /*
765 : : * inode->i_mutex nesting subclasses for the lock validator:
766 : : *
767 : : * 0: the object of the current VFS operation
768 : : * 1: parent
769 : : * 2: child/target
770 : : * 3: xattr
771 : : * 4: second non-directory
772 : : * 5: second parent (when locking independent directories in rename)
773 : : *
774 : : * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
775 : : * non-directories at once.
776 : : *
777 : : * The locking order between these classes is
778 : : * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
779 : : */
780 : : enum inode_i_mutex_lock_class
781 : : {
782 : : I_MUTEX_NORMAL,
783 : : I_MUTEX_PARENT,
784 : : I_MUTEX_CHILD,
785 : : I_MUTEX_XATTR,
786 : : I_MUTEX_NONDIR2,
787 : : I_MUTEX_PARENT2,
788 : : };
789 : :
790 : : static inline void inode_lock(struct inode *inode)
791 : : {
792 : 3 : down_write(&inode->i_rwsem);
793 : : }
794 : :
795 : : static inline void inode_unlock(struct inode *inode)
796 : : {
797 : 3 : up_write(&inode->i_rwsem);
798 : : }
799 : :
800 : : static inline void inode_lock_shared(struct inode *inode)
801 : : {
802 : 3 : down_read(&inode->i_rwsem);
803 : : }
804 : :
805 : : static inline void inode_unlock_shared(struct inode *inode)
806 : : {
807 : 3 : up_read(&inode->i_rwsem);
808 : : }
809 : :
810 : : static inline int inode_trylock(struct inode *inode)
811 : : {
812 : 3 : return down_write_trylock(&inode->i_rwsem);
813 : : }
814 : :
815 : : static inline int inode_trylock_shared(struct inode *inode)
816 : : {
817 : 0 : return down_read_trylock(&inode->i_rwsem);
818 : : }
819 : :
820 : : static inline int inode_is_locked(struct inode *inode)
821 : : {
822 : : return rwsem_is_locked(&inode->i_rwsem);
823 : : }
824 : :
825 : : static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
826 : : {
827 : 3 : down_write_nested(&inode->i_rwsem, subclass);
828 : : }
829 : :
830 : : static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
831 : : {
832 : : down_read_nested(&inode->i_rwsem, subclass);
833 : : }
834 : :
835 : : void lock_two_nondirectories(struct inode *, struct inode*);
836 : : void unlock_two_nondirectories(struct inode *, struct inode*);
837 : :
838 : : /*
839 : : * NOTE: in a 32bit arch with a preemptable kernel and
840 : : * an UP compile the i_size_read/write must be atomic
841 : : * with respect to the local cpu (unlike with preempt disabled),
842 : : * but they don't need to be atomic with respect to other cpus like in
843 : : * true SMP (so they need either to either locally disable irq around
844 : : * the read or for example on x86 they can be still implemented as a
845 : : * cmpxchg8b without the need of the lock prefix). For SMP compiles
846 : : * and 64bit archs it makes no difference if preempt is enabled or not.
847 : : */
848 : : static inline loff_t i_size_read(const struct inode *inode)
849 : : {
850 : : #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
851 : : loff_t i_size;
852 : : unsigned int seq;
853 : :
854 : : do {
855 : : seq = read_seqcount_begin(&inode->i_size_seqcount);
856 : 3 : i_size = inode->i_size;
857 : 3 : } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
858 : 3 : return i_size;
859 : : #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
860 : : loff_t i_size;
861 : :
862 : : preempt_disable();
863 : : i_size = inode->i_size;
864 : : preempt_enable();
865 : : return i_size;
866 : : #else
867 : : return inode->i_size;
868 : : #endif
869 : : }
870 : :
871 : : /*
872 : : * NOTE: unlike i_size_read(), i_size_write() does need locking around it
873 : : * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
874 : : * can be lost, resulting in subsequent i_size_read() calls spinning forever.
875 : : */
876 : : static inline void i_size_write(struct inode *inode, loff_t i_size)
877 : : {
878 : : #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
879 : 3 : preempt_disable();
880 : : write_seqcount_begin(&inode->i_size_seqcount);
881 : 3 : inode->i_size = i_size;
882 : : write_seqcount_end(&inode->i_size_seqcount);
883 : 3 : preempt_enable();
884 : : #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
885 : : preempt_disable();
886 : : inode->i_size = i_size;
887 : : preempt_enable();
888 : : #else
889 : : inode->i_size = i_size;
890 : : #endif
891 : : }
892 : :
893 : : static inline unsigned iminor(const struct inode *inode)
894 : : {
895 : 3 : return MINOR(inode->i_rdev);
896 : : }
897 : :
898 : : static inline unsigned imajor(const struct inode *inode)
899 : : {
900 : 3 : return MAJOR(inode->i_rdev);
901 : : }
902 : :
903 : : extern struct block_device *I_BDEV(struct inode *inode);
904 : :
905 : : struct fown_struct {
906 : : rwlock_t lock; /* protects pid, uid, euid fields */
907 : : struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
908 : : enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
909 : : kuid_t uid, euid; /* uid/euid of process setting the owner */
910 : : int signum; /* posix.1b rt signal to be delivered on IO */
911 : : };
912 : :
913 : : /*
914 : : * Track a single file's readahead state
915 : : */
916 : : struct file_ra_state {
917 : : pgoff_t start; /* where readahead started */
918 : : unsigned int size; /* # of readahead pages */
919 : : unsigned int async_size; /* do asynchronous readahead when
920 : : there are only # of pages ahead */
921 : :
922 : : unsigned int ra_pages; /* Maximum readahead window */
923 : : unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
924 : : loff_t prev_pos; /* Cache last read() position */
925 : : };
926 : :
927 : : /*
928 : : * Check if @index falls in the readahead windows.
929 : : */
930 : : static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
931 : : {
932 : 0 : return (index >= ra->start &&
933 : 0 : index < ra->start + ra->size);
934 : : }
935 : :
936 : : struct file {
937 : : union {
938 : : struct llist_node fu_llist;
939 : : struct rcu_head fu_rcuhead;
940 : : } f_u;
941 : : struct path f_path;
942 : : struct inode *f_inode; /* cached value */
943 : : const struct file_operations *f_op;
944 : :
945 : : /*
946 : : * Protects f_ep_links, f_flags.
947 : : * Must not be taken from IRQ context.
948 : : */
949 : : spinlock_t f_lock;
950 : : enum rw_hint f_write_hint;
951 : : atomic_long_t f_count;
952 : : unsigned int f_flags;
953 : : fmode_t f_mode;
954 : : struct mutex f_pos_lock;
955 : : loff_t f_pos;
956 : : struct fown_struct f_owner;
957 : : const struct cred *f_cred;
958 : : struct file_ra_state f_ra;
959 : :
960 : : u64 f_version;
961 : : #ifdef CONFIG_SECURITY
962 : : void *f_security;
963 : : #endif
964 : : /* needed for tty driver, and maybe others */
965 : : void *private_data;
966 : :
967 : : #ifdef CONFIG_EPOLL
968 : : /* Used by fs/eventpoll.c to link all the hooks to this file */
969 : : struct list_head f_ep_links;
970 : : struct list_head f_tfile_llink;
971 : : #endif /* #ifdef CONFIG_EPOLL */
972 : : struct address_space *f_mapping;
973 : : errseq_t f_wb_err;
974 : : } __randomize_layout
975 : : __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
976 : :
977 : : struct file_handle {
978 : : __u32 handle_bytes;
979 : : int handle_type;
980 : : /* file identifier */
981 : : unsigned char f_handle[];
982 : : };
983 : :
984 : : static inline struct file *get_file(struct file *f)
985 : : {
986 : 3 : atomic_long_inc(&f->f_count);
987 : : return f;
988 : : }
989 : : #define get_file_rcu_many(x, cnt) \
990 : : atomic_long_add_unless(&(x)->f_count, (cnt), 0)
991 : : #define get_file_rcu(x) get_file_rcu_many((x), 1)
992 : : #define file_count(x) atomic_long_read(&(x)->f_count)
993 : :
994 : : #define MAX_NON_LFS ((1UL<<31) - 1)
995 : :
996 : : /* Page cache limit. The filesystems should put that into their s_maxbytes
997 : : limits, otherwise bad things can happen in VM. */
998 : : #if BITS_PER_LONG==32
999 : : #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
1000 : : #elif BITS_PER_LONG==64
1001 : : #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
1002 : : #endif
1003 : :
1004 : : #define FL_POSIX 1
1005 : : #define FL_FLOCK 2
1006 : : #define FL_DELEG 4 /* NFSv4 delegation */
1007 : : #define FL_ACCESS 8 /* not trying to lock, just looking */
1008 : : #define FL_EXISTS 16 /* when unlocking, test for existence */
1009 : : #define FL_LEASE 32 /* lease held on this file */
1010 : : #define FL_CLOSE 64 /* unlock on close */
1011 : : #define FL_SLEEP 128 /* A blocking lock */
1012 : : #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
1013 : : #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
1014 : : #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
1015 : : #define FL_LAYOUT 2048 /* outstanding pNFS layout */
1016 : :
1017 : : #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
1018 : :
1019 : : /*
1020 : : * Special return value from posix_lock_file() and vfs_lock_file() for
1021 : : * asynchronous locking.
1022 : : */
1023 : : #define FILE_LOCK_DEFERRED 1
1024 : :
1025 : : /* legacy typedef, should eventually be removed */
1026 : : typedef void *fl_owner_t;
1027 : :
1028 : : struct file_lock;
1029 : :
1030 : : struct file_lock_operations {
1031 : : void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1032 : : void (*fl_release_private)(struct file_lock *);
1033 : : };
1034 : :
1035 : : struct lock_manager_operations {
1036 : : fl_owner_t (*lm_get_owner)(fl_owner_t);
1037 : : void (*lm_put_owner)(fl_owner_t);
1038 : : void (*lm_notify)(struct file_lock *); /* unblock callback */
1039 : : int (*lm_grant)(struct file_lock *, int);
1040 : : bool (*lm_break)(struct file_lock *);
1041 : : int (*lm_change)(struct file_lock *, int, struct list_head *);
1042 : : void (*lm_setup)(struct file_lock *, void **);
1043 : : };
1044 : :
1045 : : struct lock_manager {
1046 : : struct list_head list;
1047 : : /*
1048 : : * NFSv4 and up also want opens blocked during the grace period;
1049 : : * NLM doesn't care:
1050 : : */
1051 : : bool block_opens;
1052 : : };
1053 : :
1054 : : struct net;
1055 : : void locks_start_grace(struct net *, struct lock_manager *);
1056 : : void locks_end_grace(struct lock_manager *);
1057 : : bool locks_in_grace(struct net *);
1058 : : bool opens_in_grace(struct net *);
1059 : :
1060 : : /* that will die - we need it for nfs_lock_info */
1061 : : #include <linux/nfs_fs_i.h>
1062 : :
1063 : : /*
1064 : : * struct file_lock represents a generic "file lock". It's used to represent
1065 : : * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1066 : : * note that the same struct is used to represent both a request for a lock and
1067 : : * the lock itself, but the same object is never used for both.
1068 : : *
1069 : : * FIXME: should we create a separate "struct lock_request" to help distinguish
1070 : : * these two uses?
1071 : : *
1072 : : * The varous i_flctx lists are ordered by:
1073 : : *
1074 : : * 1) lock owner
1075 : : * 2) lock range start
1076 : : * 3) lock range end
1077 : : *
1078 : : * Obviously, the last two criteria only matter for POSIX locks.
1079 : : */
1080 : : struct file_lock {
1081 : : struct file_lock *fl_blocker; /* The lock, that is blocking us */
1082 : : struct list_head fl_list; /* link into file_lock_context */
1083 : : struct hlist_node fl_link; /* node in global lists */
1084 : : struct list_head fl_blocked_requests; /* list of requests with
1085 : : * ->fl_blocker pointing here
1086 : : */
1087 : : struct list_head fl_blocked_member; /* node in
1088 : : * ->fl_blocker->fl_blocked_requests
1089 : : */
1090 : : fl_owner_t fl_owner;
1091 : : unsigned int fl_flags;
1092 : : unsigned char fl_type;
1093 : : unsigned int fl_pid;
1094 : : int fl_link_cpu; /* what cpu's list is this on? */
1095 : : wait_queue_head_t fl_wait;
1096 : : struct file *fl_file;
1097 : : loff_t fl_start;
1098 : : loff_t fl_end;
1099 : :
1100 : : struct fasync_struct * fl_fasync; /* for lease break notifications */
1101 : : /* for lease breaks: */
1102 : : unsigned long fl_break_time;
1103 : : unsigned long fl_downgrade_time;
1104 : :
1105 : : const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1106 : : const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1107 : : union {
1108 : : struct nfs_lock_info nfs_fl;
1109 : : struct nfs4_lock_info nfs4_fl;
1110 : : struct {
1111 : : struct list_head link; /* link in AFS vnode's pending_locks list */
1112 : : int state; /* state of grant or error if -ve */
1113 : : unsigned int debug_id;
1114 : : } afs;
1115 : : } fl_u;
1116 : : } __randomize_layout;
1117 : :
1118 : : struct file_lock_context {
1119 : : spinlock_t flc_lock;
1120 : : struct list_head flc_flock;
1121 : : struct list_head flc_posix;
1122 : : struct list_head flc_lease;
1123 : : };
1124 : :
1125 : : /* The following constant reflects the upper bound of the file/locking space */
1126 : : #ifndef OFFSET_MAX
1127 : : #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1128 : : #define OFFSET_MAX INT_LIMIT(loff_t)
1129 : : #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1130 : : #endif
1131 : :
1132 : : extern void send_sigio(struct fown_struct *fown, int fd, int band);
1133 : :
1134 : : #define locks_inode(f) file_inode(f)
1135 : :
1136 : : #ifdef CONFIG_FILE_LOCKING
1137 : : extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1138 : : extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1139 : : struct flock *);
1140 : :
1141 : : #if BITS_PER_LONG == 32
1142 : : extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1143 : : extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1144 : : struct flock64 *);
1145 : : #endif
1146 : :
1147 : : extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1148 : : extern int fcntl_getlease(struct file *filp);
1149 : :
1150 : : /* fs/locks.c */
1151 : : void locks_free_lock_context(struct inode *inode);
1152 : : void locks_free_lock(struct file_lock *fl);
1153 : : extern void locks_init_lock(struct file_lock *);
1154 : : extern struct file_lock * locks_alloc_lock(void);
1155 : : extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1156 : : extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1157 : : extern void locks_remove_posix(struct file *, fl_owner_t);
1158 : : extern void locks_remove_file(struct file *);
1159 : : extern void locks_release_private(struct file_lock *);
1160 : : extern void posix_test_lock(struct file *, struct file_lock *);
1161 : : extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1162 : : extern int locks_delete_block(struct file_lock *);
1163 : : extern int vfs_test_lock(struct file *, struct file_lock *);
1164 : : extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1165 : : extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1166 : : extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1167 : : extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1168 : : extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1169 : : extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1170 : : extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1171 : : extern int lease_modify(struct file_lock *, int, struct list_head *);
1172 : :
1173 : : struct notifier_block;
1174 : : extern int lease_register_notifier(struct notifier_block *);
1175 : : extern void lease_unregister_notifier(struct notifier_block *);
1176 : :
1177 : : struct files_struct;
1178 : : extern void show_fd_locks(struct seq_file *f,
1179 : : struct file *filp, struct files_struct *files);
1180 : : #else /* !CONFIG_FILE_LOCKING */
1181 : : static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1182 : : struct flock __user *user)
1183 : : {
1184 : : return -EINVAL;
1185 : : }
1186 : :
1187 : : static inline int fcntl_setlk(unsigned int fd, struct file *file,
1188 : : unsigned int cmd, struct flock __user *user)
1189 : : {
1190 : : return -EACCES;
1191 : : }
1192 : :
1193 : : #if BITS_PER_LONG == 32
1194 : : static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1195 : : struct flock64 __user *user)
1196 : : {
1197 : : return -EINVAL;
1198 : : }
1199 : :
1200 : : static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1201 : : unsigned int cmd, struct flock64 __user *user)
1202 : : {
1203 : : return -EACCES;
1204 : : }
1205 : : #endif
1206 : : static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1207 : : {
1208 : : return -EINVAL;
1209 : : }
1210 : :
1211 : : static inline int fcntl_getlease(struct file *filp)
1212 : : {
1213 : : return F_UNLCK;
1214 : : }
1215 : :
1216 : : static inline void
1217 : : locks_free_lock_context(struct inode *inode)
1218 : : {
1219 : : }
1220 : :
1221 : : static inline void locks_init_lock(struct file_lock *fl)
1222 : : {
1223 : : return;
1224 : : }
1225 : :
1226 : : static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1227 : : {
1228 : : return;
1229 : : }
1230 : :
1231 : : static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1232 : : {
1233 : : return;
1234 : : }
1235 : :
1236 : : static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1237 : : {
1238 : : return;
1239 : : }
1240 : :
1241 : : static inline void locks_remove_file(struct file *filp)
1242 : : {
1243 : : return;
1244 : : }
1245 : :
1246 : : static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1247 : : {
1248 : : return;
1249 : : }
1250 : :
1251 : : static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1252 : : struct file_lock *conflock)
1253 : : {
1254 : : return -ENOLCK;
1255 : : }
1256 : :
1257 : : static inline int locks_delete_block(struct file_lock *waiter)
1258 : : {
1259 : : return -ENOENT;
1260 : : }
1261 : :
1262 : : static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1263 : : {
1264 : : return 0;
1265 : : }
1266 : :
1267 : : static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1268 : : struct file_lock *fl, struct file_lock *conf)
1269 : : {
1270 : : return -ENOLCK;
1271 : : }
1272 : :
1273 : : static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1274 : : {
1275 : : return 0;
1276 : : }
1277 : :
1278 : : static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1279 : : {
1280 : : return -ENOLCK;
1281 : : }
1282 : :
1283 : : static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1284 : : {
1285 : : return 0;
1286 : : }
1287 : :
1288 : : static inline void lease_get_mtime(struct inode *inode,
1289 : : struct timespec64 *time)
1290 : : {
1291 : : return;
1292 : : }
1293 : :
1294 : : static inline int generic_setlease(struct file *filp, long arg,
1295 : : struct file_lock **flp, void **priv)
1296 : : {
1297 : : return -EINVAL;
1298 : : }
1299 : :
1300 : : static inline int vfs_setlease(struct file *filp, long arg,
1301 : : struct file_lock **lease, void **priv)
1302 : : {
1303 : : return -EINVAL;
1304 : : }
1305 : :
1306 : : static inline int lease_modify(struct file_lock *fl, int arg,
1307 : : struct list_head *dispose)
1308 : : {
1309 : : return -EINVAL;
1310 : : }
1311 : :
1312 : : struct files_struct;
1313 : : static inline void show_fd_locks(struct seq_file *f,
1314 : : struct file *filp, struct files_struct *files) {}
1315 : : #endif /* !CONFIG_FILE_LOCKING */
1316 : :
1317 : : static inline struct inode *file_inode(const struct file *f)
1318 : : {
1319 : 3 : return f->f_inode;
1320 : : }
1321 : :
1322 : : static inline struct dentry *file_dentry(const struct file *file)
1323 : : {
1324 : 3 : return d_real(file->f_path.dentry, file_inode(file));
1325 : : }
1326 : :
1327 : : static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1328 : : {
1329 : 3 : return locks_lock_inode_wait(locks_inode(filp), fl);
1330 : : }
1331 : :
1332 : : struct fasync_struct {
1333 : : rwlock_t fa_lock;
1334 : : int magic;
1335 : : int fa_fd;
1336 : : struct fasync_struct *fa_next; /* singly linked list */
1337 : : struct file *fa_file;
1338 : : struct rcu_head fa_rcu;
1339 : : };
1340 : :
1341 : : #define FASYNC_MAGIC 0x4601
1342 : :
1343 : : /* SMP safe fasync helpers: */
1344 : : extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1345 : : extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1346 : : extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1347 : : extern struct fasync_struct *fasync_alloc(void);
1348 : : extern void fasync_free(struct fasync_struct *);
1349 : :
1350 : : /* can be called from interrupts */
1351 : : extern void kill_fasync(struct fasync_struct **, int, int);
1352 : :
1353 : : extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1354 : : extern int f_setown(struct file *filp, unsigned long arg, int force);
1355 : : extern void f_delown(struct file *filp);
1356 : : extern pid_t f_getown(struct file *filp);
1357 : : extern int send_sigurg(struct fown_struct *fown);
1358 : :
1359 : : /*
1360 : : * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1361 : : * represented in both.
1362 : : */
1363 : : #define SB_RDONLY 1 /* Mount read-only */
1364 : : #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1365 : : #define SB_NODEV 4 /* Disallow access to device special files */
1366 : : #define SB_NOEXEC 8 /* Disallow program execution */
1367 : : #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1368 : : #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1369 : : #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1370 : : #define SB_NOATIME 1024 /* Do not update access times. */
1371 : : #define SB_NODIRATIME 2048 /* Do not update directory access times */
1372 : : #define SB_SILENT 32768
1373 : : #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1374 : : #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1375 : : #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1376 : : #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1377 : :
1378 : : /* These sb flags are internal to the kernel */
1379 : : #define SB_SUBMOUNT (1<<26)
1380 : : #define SB_FORCE (1<<27)
1381 : : #define SB_NOSEC (1<<28)
1382 : : #define SB_BORN (1<<29)
1383 : : #define SB_ACTIVE (1<<30)
1384 : : #define SB_NOUSER (1<<31)
1385 : :
1386 : : /*
1387 : : * Umount options
1388 : : */
1389 : :
1390 : : #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1391 : : #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1392 : : #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1393 : : #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1394 : : #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1395 : :
1396 : : /* sb->s_iflags */
1397 : : #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1398 : : #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1399 : : #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1400 : : #define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */
1401 : :
1402 : : /* sb->s_iflags to limit user namespace mounts */
1403 : : #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1404 : : #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1405 : : #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1406 : :
1407 : : /* Possible states of 'frozen' field */
1408 : : enum {
1409 : : SB_UNFROZEN = 0, /* FS is unfrozen */
1410 : : SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1411 : : SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1412 : : SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1413 : : * internal threads if needed) */
1414 : : SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1415 : : };
1416 : :
1417 : : #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1418 : :
1419 : : struct sb_writers {
1420 : : int frozen; /* Is sb frozen? */
1421 : : wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1422 : : struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1423 : : };
1424 : :
1425 : : struct super_block {
1426 : : struct list_head s_list; /* Keep this first */
1427 : : dev_t s_dev; /* search index; _not_ kdev_t */
1428 : : unsigned char s_blocksize_bits;
1429 : : unsigned long s_blocksize;
1430 : : loff_t s_maxbytes; /* Max file size */
1431 : : struct file_system_type *s_type;
1432 : : const struct super_operations *s_op;
1433 : : const struct dquot_operations *dq_op;
1434 : : const struct quotactl_ops *s_qcop;
1435 : : const struct export_operations *s_export_op;
1436 : : unsigned long s_flags;
1437 : : unsigned long s_iflags; /* internal SB_I_* flags */
1438 : : unsigned long s_magic;
1439 : : struct dentry *s_root;
1440 : : struct rw_semaphore s_umount;
1441 : : int s_count;
1442 : : atomic_t s_active;
1443 : : #ifdef CONFIG_SECURITY
1444 : : void *s_security;
1445 : : #endif
1446 : : const struct xattr_handler **s_xattr;
1447 : : #ifdef CONFIG_FS_ENCRYPTION
1448 : : const struct fscrypt_operations *s_cop;
1449 : : struct key *s_master_keys; /* master crypto keys in use */
1450 : : #endif
1451 : : #ifdef CONFIG_FS_VERITY
1452 : : const struct fsverity_operations *s_vop;
1453 : : #endif
1454 : : struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1455 : : struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1456 : : struct block_device *s_bdev;
1457 : : struct backing_dev_info *s_bdi;
1458 : : struct mtd_info *s_mtd;
1459 : : struct hlist_node s_instances;
1460 : : unsigned int s_quota_types; /* Bitmask of supported quota types */
1461 : : struct quota_info s_dquot; /* Diskquota specific options */
1462 : :
1463 : : struct sb_writers s_writers;
1464 : :
1465 : : /*
1466 : : * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1467 : : * s_fsnotify_marks together for cache efficiency. They are frequently
1468 : : * accessed and rarely modified.
1469 : : */
1470 : : void *s_fs_info; /* Filesystem private info */
1471 : :
1472 : : /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1473 : : u32 s_time_gran;
1474 : : /* Time limits for c/m/atime in seconds */
1475 : : time64_t s_time_min;
1476 : : time64_t s_time_max;
1477 : : #ifdef CONFIG_FSNOTIFY
1478 : : __u32 s_fsnotify_mask;
1479 : : struct fsnotify_mark_connector __rcu *s_fsnotify_marks;
1480 : : #endif
1481 : :
1482 : : char s_id[32]; /* Informational name */
1483 : : uuid_t s_uuid; /* UUID */
1484 : :
1485 : : unsigned int s_max_links;
1486 : : fmode_t s_mode;
1487 : :
1488 : : /*
1489 : : * The next field is for VFS *only*. No filesystems have any business
1490 : : * even looking at it. You had been warned.
1491 : : */
1492 : : struct mutex s_vfs_rename_mutex; /* Kludge */
1493 : :
1494 : : /*
1495 : : * Filesystem subtype. If non-empty the filesystem type field
1496 : : * in /proc/mounts will be "type.subtype"
1497 : : */
1498 : : const char *s_subtype;
1499 : :
1500 : : const struct dentry_operations *s_d_op; /* default d_op for dentries */
1501 : :
1502 : : /*
1503 : : * Saved pool identifier for cleancache (-1 means none)
1504 : : */
1505 : : int cleancache_poolid;
1506 : :
1507 : : struct shrinker s_shrink; /* per-sb shrinker handle */
1508 : :
1509 : : /* Number of inodes with nlink == 0 but still referenced */
1510 : : atomic_long_t s_remove_count;
1511 : :
1512 : : /* Pending fsnotify inode refs */
1513 : : atomic_long_t s_fsnotify_inode_refs;
1514 : :
1515 : : /* Being remounted read-only */
1516 : : int s_readonly_remount;
1517 : :
1518 : : /* AIO completions deferred from interrupt context */
1519 : : struct workqueue_struct *s_dio_done_wq;
1520 : : struct hlist_head s_pins;
1521 : :
1522 : : /*
1523 : : * Owning user namespace and default context in which to
1524 : : * interpret filesystem uids, gids, quotas, device nodes,
1525 : : * xattrs and security labels.
1526 : : */
1527 : : struct user_namespace *s_user_ns;
1528 : :
1529 : : /*
1530 : : * The list_lru structure is essentially just a pointer to a table
1531 : : * of per-node lru lists, each of which has its own spinlock.
1532 : : * There is no need to put them into separate cachelines.
1533 : : */
1534 : : struct list_lru s_dentry_lru;
1535 : : struct list_lru s_inode_lru;
1536 : : struct rcu_head rcu;
1537 : : struct work_struct destroy_work;
1538 : :
1539 : : struct mutex s_sync_lock; /* sync serialisation lock */
1540 : :
1541 : : /*
1542 : : * Indicates how deep in a filesystem stack this SB is
1543 : : */
1544 : : int s_stack_depth;
1545 : :
1546 : : /* s_inode_list_lock protects s_inodes */
1547 : : spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1548 : : struct list_head s_inodes; /* all inodes */
1549 : :
1550 : : spinlock_t s_inode_wblist_lock;
1551 : : struct list_head s_inodes_wb; /* writeback inodes */
1552 : : } __randomize_layout;
1553 : :
1554 : : /* Helper functions so that in most cases filesystems will
1555 : : * not need to deal directly with kuid_t and kgid_t and can
1556 : : * instead deal with the raw numeric values that are stored
1557 : : * in the filesystem.
1558 : : */
1559 : : static inline uid_t i_uid_read(const struct inode *inode)
1560 : : {
1561 : 3 : return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1562 : : }
1563 : :
1564 : : static inline gid_t i_gid_read(const struct inode *inode)
1565 : : {
1566 : 3 : return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1567 : : }
1568 : :
1569 : : static inline void i_uid_write(struct inode *inode, uid_t uid)
1570 : : {
1571 : 3 : inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1572 : : }
1573 : :
1574 : : static inline void i_gid_write(struct inode *inode, gid_t gid)
1575 : : {
1576 : 3 : inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1577 : : }
1578 : :
1579 : : extern struct timespec64 timespec64_trunc(struct timespec64 t, unsigned gran);
1580 : : extern struct timespec64 current_time(struct inode *inode);
1581 : :
1582 : : /*
1583 : : * Snapshotting support.
1584 : : */
1585 : :
1586 : : void __sb_end_write(struct super_block *sb, int level);
1587 : : int __sb_start_write(struct super_block *sb, int level, bool wait);
1588 : :
1589 : : #define __sb_writers_acquired(sb, lev) \
1590 : : percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1591 : : #define __sb_writers_release(sb, lev) \
1592 : : percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1593 : :
1594 : : /**
1595 : : * sb_end_write - drop write access to a superblock
1596 : : * @sb: the super we wrote to
1597 : : *
1598 : : * Decrement number of writers to the filesystem. Wake up possible waiters
1599 : : * wanting to freeze the filesystem.
1600 : : */
1601 : : static inline void sb_end_write(struct super_block *sb)
1602 : : {
1603 : 3 : __sb_end_write(sb, SB_FREEZE_WRITE);
1604 : : }
1605 : :
1606 : : /**
1607 : : * sb_end_pagefault - drop write access to a superblock from a page fault
1608 : : * @sb: the super we wrote to
1609 : : *
1610 : : * Decrement number of processes handling write page fault to the filesystem.
1611 : : * Wake up possible waiters wanting to freeze the filesystem.
1612 : : */
1613 : : static inline void sb_end_pagefault(struct super_block *sb)
1614 : : {
1615 : 3 : __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1616 : : }
1617 : :
1618 : : /**
1619 : : * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1620 : : * @sb: the super we wrote to
1621 : : *
1622 : : * Decrement fs-internal number of writers to the filesystem. Wake up possible
1623 : : * waiters wanting to freeze the filesystem.
1624 : : */
1625 : : static inline void sb_end_intwrite(struct super_block *sb)
1626 : : {
1627 : 3 : __sb_end_write(sb, SB_FREEZE_FS);
1628 : : }
1629 : :
1630 : : /**
1631 : : * sb_start_write - get write access to a superblock
1632 : : * @sb: the super we write to
1633 : : *
1634 : : * When a process wants to write data or metadata to a file system (i.e. dirty
1635 : : * a page or an inode), it should embed the operation in a sb_start_write() -
1636 : : * sb_end_write() pair to get exclusion against file system freezing. This
1637 : : * function increments number of writers preventing freezing. If the file
1638 : : * system is already frozen, the function waits until the file system is
1639 : : * thawed.
1640 : : *
1641 : : * Since freeze protection behaves as a lock, users have to preserve
1642 : : * ordering of freeze protection and other filesystem locks. Generally,
1643 : : * freeze protection should be the outermost lock. In particular, we have:
1644 : : *
1645 : : * sb_start_write
1646 : : * -> i_mutex (write path, truncate, directory ops, ...)
1647 : : * -> s_umount (freeze_super, thaw_super)
1648 : : */
1649 : : static inline void sb_start_write(struct super_block *sb)
1650 : : {
1651 : 3 : __sb_start_write(sb, SB_FREEZE_WRITE, true);
1652 : : }
1653 : :
1654 : : static inline int sb_start_write_trylock(struct super_block *sb)
1655 : : {
1656 : 3 : return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1657 : : }
1658 : :
1659 : : /**
1660 : : * sb_start_pagefault - get write access to a superblock from a page fault
1661 : : * @sb: the super we write to
1662 : : *
1663 : : * When a process starts handling write page fault, it should embed the
1664 : : * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1665 : : * exclusion against file system freezing. This is needed since the page fault
1666 : : * is going to dirty a page. This function increments number of running page
1667 : : * faults preventing freezing. If the file system is already frozen, the
1668 : : * function waits until the file system is thawed.
1669 : : *
1670 : : * Since page fault freeze protection behaves as a lock, users have to preserve
1671 : : * ordering of freeze protection and other filesystem locks. It is advised to
1672 : : * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1673 : : * handling code implies lock dependency:
1674 : : *
1675 : : * mmap_sem
1676 : : * -> sb_start_pagefault
1677 : : */
1678 : : static inline void sb_start_pagefault(struct super_block *sb)
1679 : : {
1680 : 3 : __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1681 : : }
1682 : :
1683 : : /*
1684 : : * sb_start_intwrite - get write access to a superblock for internal fs purposes
1685 : : * @sb: the super we write to
1686 : : *
1687 : : * This is the third level of protection against filesystem freezing. It is
1688 : : * free for use by a filesystem. The only requirement is that it must rank
1689 : : * below sb_start_pagefault.
1690 : : *
1691 : : * For example filesystem can call sb_start_intwrite() when starting a
1692 : : * transaction which somewhat eases handling of freezing for internal sources
1693 : : * of filesystem changes (internal fs threads, discarding preallocation on file
1694 : : * close, etc.).
1695 : : */
1696 : : static inline void sb_start_intwrite(struct super_block *sb)
1697 : : {
1698 : 3 : __sb_start_write(sb, SB_FREEZE_FS, true);
1699 : : }
1700 : :
1701 : : static inline int sb_start_intwrite_trylock(struct super_block *sb)
1702 : : {
1703 : 3 : return __sb_start_write(sb, SB_FREEZE_FS, false);
1704 : : }
1705 : :
1706 : :
1707 : : extern bool inode_owner_or_capable(const struct inode *inode);
1708 : :
1709 : : /*
1710 : : * VFS helper functions..
1711 : : */
1712 : : extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1713 : : extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1714 : : extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1715 : : extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1716 : : extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1717 : : extern int vfs_rmdir(struct inode *, struct dentry *);
1718 : : extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1719 : : extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1720 : : extern int vfs_whiteout(struct inode *, struct dentry *);
1721 : :
1722 : : extern struct dentry *vfs_tmpfile(struct dentry *dentry, umode_t mode,
1723 : : int open_flag);
1724 : :
1725 : : int vfs_mkobj(struct dentry *, umode_t,
1726 : : int (*f)(struct dentry *, umode_t, void *),
1727 : : void *);
1728 : :
1729 : : extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1730 : :
1731 : : #ifdef CONFIG_COMPAT
1732 : : extern long compat_ptr_ioctl(struct file *file, unsigned int cmd,
1733 : : unsigned long arg);
1734 : : #else
1735 : : #define compat_ptr_ioctl NULL
1736 : : #endif
1737 : :
1738 : : /*
1739 : : * VFS file helper functions.
1740 : : */
1741 : : extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1742 : : umode_t mode);
1743 : : extern bool may_open_dev(const struct path *path);
1744 : : /*
1745 : : * VFS FS_IOC_FIEMAP helper definitions.
1746 : : */
1747 : : struct fiemap_extent_info {
1748 : : unsigned int fi_flags; /* Flags as passed from user */
1749 : : unsigned int fi_extents_mapped; /* Number of mapped extents */
1750 : : unsigned int fi_extents_max; /* Size of fiemap_extent array */
1751 : : struct fiemap_extent __user *fi_extents_start; /* Start of
1752 : : fiemap_extent array */
1753 : : };
1754 : : int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1755 : : u64 phys, u64 len, u32 flags);
1756 : : int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1757 : :
1758 : : /*
1759 : : * This is the "filldir" function type, used by readdir() to let
1760 : : * the kernel specify what kind of dirent layout it wants to have.
1761 : : * This allows the kernel to read directories into kernel space or
1762 : : * to have different dirent layouts depending on the binary type.
1763 : : */
1764 : : struct dir_context;
1765 : : typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1766 : : unsigned);
1767 : :
1768 : : struct dir_context {
1769 : : filldir_t actor;
1770 : : loff_t pos;
1771 : : };
1772 : :
1773 : : struct block_device_operations;
1774 : :
1775 : : /* These macros are for out of kernel modules to test that
1776 : : * the kernel supports the unlocked_ioctl and compat_ioctl
1777 : : * fields in struct file_operations. */
1778 : : #define HAVE_COMPAT_IOCTL 1
1779 : : #define HAVE_UNLOCKED_IOCTL 1
1780 : :
1781 : : /*
1782 : : * These flags let !MMU mmap() govern direct device mapping vs immediate
1783 : : * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1784 : : *
1785 : : * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1786 : : * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1787 : : * NOMMU_MAP_READ: Can be mapped for reading
1788 : : * NOMMU_MAP_WRITE: Can be mapped for writing
1789 : : * NOMMU_MAP_EXEC: Can be mapped for execution
1790 : : */
1791 : : #define NOMMU_MAP_COPY 0x00000001
1792 : : #define NOMMU_MAP_DIRECT 0x00000008
1793 : : #define NOMMU_MAP_READ VM_MAYREAD
1794 : : #define NOMMU_MAP_WRITE VM_MAYWRITE
1795 : : #define NOMMU_MAP_EXEC VM_MAYEXEC
1796 : :
1797 : : #define NOMMU_VMFLAGS \
1798 : : (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1799 : :
1800 : : /*
1801 : : * These flags control the behavior of the remap_file_range function pointer.
1802 : : * If it is called with len == 0 that means "remap to end of source file".
1803 : : * See Documentation/filesystems/vfs.rst for more details about this call.
1804 : : *
1805 : : * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
1806 : : * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
1807 : : */
1808 : : #define REMAP_FILE_DEDUP (1 << 0)
1809 : : #define REMAP_FILE_CAN_SHORTEN (1 << 1)
1810 : :
1811 : : /*
1812 : : * These flags signal that the caller is ok with altering various aspects of
1813 : : * the behavior of the remap operation. The changes must be made by the
1814 : : * implementation; the vfs remap helper functions can take advantage of them.
1815 : : * Flags in this category exist to preserve the quirky behavior of the hoisted
1816 : : * btrfs clone/dedupe ioctls.
1817 : : */
1818 : : #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
1819 : :
1820 : : struct iov_iter;
1821 : :
1822 : : struct file_operations {
1823 : : struct module *owner;
1824 : : loff_t (*llseek) (struct file *, loff_t, int);
1825 : : ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1826 : : ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1827 : : ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1828 : : ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1829 : : int (*iopoll)(struct kiocb *kiocb, bool spin);
1830 : : int (*iterate) (struct file *, struct dir_context *);
1831 : : int (*iterate_shared) (struct file *, struct dir_context *);
1832 : : __poll_t (*poll) (struct file *, struct poll_table_struct *);
1833 : : long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1834 : : long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1835 : : int (*mmap) (struct file *, struct vm_area_struct *);
1836 : : unsigned long mmap_supported_flags;
1837 : : int (*open) (struct inode *, struct file *);
1838 : : int (*flush) (struct file *, fl_owner_t id);
1839 : : int (*release) (struct inode *, struct file *);
1840 : : int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1841 : : int (*fasync) (int, struct file *, int);
1842 : : int (*lock) (struct file *, int, struct file_lock *);
1843 : : ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1844 : : unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1845 : : int (*check_flags)(int);
1846 : : int (*flock) (struct file *, int, struct file_lock *);
1847 : : ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1848 : : ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1849 : : int (*setlease)(struct file *, long, struct file_lock **, void **);
1850 : : long (*fallocate)(struct file *file, int mode, loff_t offset,
1851 : : loff_t len);
1852 : : void (*show_fdinfo)(struct seq_file *m, struct file *f);
1853 : : #ifndef CONFIG_MMU
1854 : : unsigned (*mmap_capabilities)(struct file *);
1855 : : #endif
1856 : : ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1857 : : loff_t, size_t, unsigned int);
1858 : : loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
1859 : : struct file *file_out, loff_t pos_out,
1860 : : loff_t len, unsigned int remap_flags);
1861 : : int (*fadvise)(struct file *, loff_t, loff_t, int);
1862 : : } __randomize_layout;
1863 : :
1864 : : struct inode_operations {
1865 : : struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1866 : : const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1867 : : int (*permission) (struct inode *, int);
1868 : : struct posix_acl * (*get_acl)(struct inode *, int);
1869 : :
1870 : : int (*readlink) (struct dentry *, char __user *,int);
1871 : :
1872 : : int (*create) (struct inode *,struct dentry *, umode_t, bool);
1873 : : int (*link) (struct dentry *,struct inode *,struct dentry *);
1874 : : int (*unlink) (struct inode *,struct dentry *);
1875 : : int (*symlink) (struct inode *,struct dentry *,const char *);
1876 : : int (*mkdir) (struct inode *,struct dentry *,umode_t);
1877 : : int (*rmdir) (struct inode *,struct dentry *);
1878 : : int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1879 : : int (*rename) (struct inode *, struct dentry *,
1880 : : struct inode *, struct dentry *, unsigned int);
1881 : : int (*setattr) (struct dentry *, struct iattr *);
1882 : : int (*getattr) (const struct path *, struct kstat *, u32, unsigned int);
1883 : : ssize_t (*listxattr) (struct dentry *, char *, size_t);
1884 : : int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1885 : : u64 len);
1886 : : int (*update_time)(struct inode *, struct timespec64 *, int);
1887 : : int (*atomic_open)(struct inode *, struct dentry *,
1888 : : struct file *, unsigned open_flag,
1889 : : umode_t create_mode);
1890 : : int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1891 : : int (*set_acl)(struct inode *, struct posix_acl *, int);
1892 : : } ____cacheline_aligned;
1893 : :
1894 : : static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
1895 : : struct iov_iter *iter)
1896 : : {
1897 : 3 : return file->f_op->read_iter(kio, iter);
1898 : : }
1899 : :
1900 : : static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
1901 : : struct iov_iter *iter)
1902 : : {
1903 : 3 : return file->f_op->write_iter(kio, iter);
1904 : : }
1905 : :
1906 : : static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
1907 : : {
1908 : 3 : return file->f_op->mmap(file, vma);
1909 : : }
1910 : :
1911 : : ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1912 : : unsigned long nr_segs, unsigned long fast_segs,
1913 : : struct iovec *fast_pointer,
1914 : : struct iovec **ret_pointer);
1915 : :
1916 : : extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1917 : : extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1918 : : extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1919 : : extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1920 : : unsigned long, loff_t *, rwf_t);
1921 : : extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1922 : : loff_t, size_t, unsigned int);
1923 : : extern ssize_t generic_copy_file_range(struct file *file_in, loff_t pos_in,
1924 : : struct file *file_out, loff_t pos_out,
1925 : : size_t len, unsigned int flags);
1926 : : extern int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
1927 : : struct file *file_out, loff_t pos_out,
1928 : : loff_t *count,
1929 : : unsigned int remap_flags);
1930 : : extern loff_t do_clone_file_range(struct file *file_in, loff_t pos_in,
1931 : : struct file *file_out, loff_t pos_out,
1932 : : loff_t len, unsigned int remap_flags);
1933 : : extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1934 : : struct file *file_out, loff_t pos_out,
1935 : : loff_t len, unsigned int remap_flags);
1936 : : extern int vfs_dedupe_file_range(struct file *file,
1937 : : struct file_dedupe_range *same);
1938 : : extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
1939 : : struct file *dst_file, loff_t dst_pos,
1940 : : loff_t len, unsigned int remap_flags);
1941 : :
1942 : :
1943 : : struct super_operations {
1944 : : struct inode *(*alloc_inode)(struct super_block *sb);
1945 : : void (*destroy_inode)(struct inode *);
1946 : : void (*free_inode)(struct inode *);
1947 : :
1948 : : void (*dirty_inode) (struct inode *, int flags);
1949 : : int (*write_inode) (struct inode *, struct writeback_control *wbc);
1950 : : int (*drop_inode) (struct inode *);
1951 : : void (*evict_inode) (struct inode *);
1952 : : void (*put_super) (struct super_block *);
1953 : : int (*sync_fs)(struct super_block *sb, int wait);
1954 : : int (*freeze_super) (struct super_block *);
1955 : : int (*freeze_fs) (struct super_block *);
1956 : : int (*thaw_super) (struct super_block *);
1957 : : int (*unfreeze_fs) (struct super_block *);
1958 : : int (*statfs) (struct dentry *, struct kstatfs *);
1959 : : int (*remount_fs) (struct super_block *, int *, char *);
1960 : : void (*umount_begin) (struct super_block *);
1961 : :
1962 : : int (*show_options)(struct seq_file *, struct dentry *);
1963 : : int (*show_devname)(struct seq_file *, struct dentry *);
1964 : : int (*show_path)(struct seq_file *, struct dentry *);
1965 : : int (*show_stats)(struct seq_file *, struct dentry *);
1966 : : #ifdef CONFIG_QUOTA
1967 : : ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1968 : : ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1969 : : struct dquot **(*get_dquots)(struct inode *);
1970 : : #endif
1971 : : int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1972 : : long (*nr_cached_objects)(struct super_block *,
1973 : : struct shrink_control *);
1974 : : long (*free_cached_objects)(struct super_block *,
1975 : : struct shrink_control *);
1976 : : };
1977 : :
1978 : : /*
1979 : : * Inode flags - they have no relation to superblock flags now
1980 : : */
1981 : : #define S_SYNC 1 /* Writes are synced at once */
1982 : : #define S_NOATIME 2 /* Do not update access times */
1983 : : #define S_APPEND 4 /* Append-only file */
1984 : : #define S_IMMUTABLE 8 /* Immutable file */
1985 : : #define S_DEAD 16 /* removed, but still open directory */
1986 : : #define S_NOQUOTA 32 /* Inode is not counted to quota */
1987 : : #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1988 : : #define S_NOCMTIME 128 /* Do not update file c/mtime */
1989 : : #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1990 : : #define S_PRIVATE 512 /* Inode is fs-internal */
1991 : : #define S_IMA 1024 /* Inode has an associated IMA struct */
1992 : : #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1993 : : #define S_NOSEC 4096 /* no suid or xattr security attributes */
1994 : : #ifdef CONFIG_FS_DAX
1995 : : #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1996 : : #else
1997 : : #define S_DAX 0 /* Make all the DAX code disappear */
1998 : : #endif
1999 : : #define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */
2000 : : #define S_CASEFOLD 32768 /* Casefolded file */
2001 : : #define S_VERITY 65536 /* Verity file (using fs/verity/) */
2002 : :
2003 : : /*
2004 : : * Note that nosuid etc flags are inode-specific: setting some file-system
2005 : : * flags just means all the inodes inherit those flags by default. It might be
2006 : : * possible to override it selectively if you really wanted to with some
2007 : : * ioctl() that is not currently implemented.
2008 : : *
2009 : : * Exception: SB_RDONLY is always applied to the entire file system.
2010 : : *
2011 : : * Unfortunately, it is possible to change a filesystems flags with it mounted
2012 : : * with files in use. This means that all of the inodes will not have their
2013 : : * i_flags updated. Hence, i_flags no longer inherit the superblock mount
2014 : : * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
2015 : : */
2016 : : #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
2017 : :
2018 : 3 : static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
2019 : : #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
2020 : : #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
2021 : : ((inode)->i_flags & S_SYNC))
2022 : : #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
2023 : : ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
2024 : : #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
2025 : : #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
2026 : : #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
2027 : :
2028 : : #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
2029 : : #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
2030 : : #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
2031 : : #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
2032 : :
2033 : : #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
2034 : : #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
2035 : : #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
2036 : : #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
2037 : : #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
2038 : : #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2039 : : #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2040 : : #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2041 : : #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2042 : : #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD)
2043 : : #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY)
2044 : :
2045 : : #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2046 : : (inode)->i_rdev == WHITEOUT_DEV)
2047 : :
2048 : : static inline bool HAS_UNMAPPED_ID(struct inode *inode)
2049 : : {
2050 : 3 : return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
2051 : : }
2052 : :
2053 : : static inline enum rw_hint file_write_hint(struct file *file)
2054 : : {
2055 : 3 : if (file->f_write_hint != WRITE_LIFE_NOT_SET)
2056 : : return file->f_write_hint;
2057 : :
2058 : 3 : return file_inode(file)->i_write_hint;
2059 : : }
2060 : :
2061 : : static inline int iocb_flags(struct file *file);
2062 : :
2063 : : static inline u16 ki_hint_validate(enum rw_hint hint)
2064 : : {
2065 : : typeof(((struct kiocb *)0)->ki_hint) max_hint = -1;
2066 : :
2067 : 3 : if (hint <= max_hint)
2068 : 3 : return hint;
2069 : : return 0;
2070 : : }
2071 : :
2072 : 3 : static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
2073 : : {
2074 : 3 : *kiocb = (struct kiocb) {
2075 : : .ki_filp = filp,
2076 : 3 : .ki_flags = iocb_flags(filp),
2077 : : .ki_hint = ki_hint_validate(file_write_hint(filp)),
2078 : : .ki_ioprio = get_current_ioprio(),
2079 : : };
2080 : 3 : }
2081 : :
2082 : : /*
2083 : : * Inode state bits. Protected by inode->i_lock
2084 : : *
2085 : : * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
2086 : : * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
2087 : : *
2088 : : * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2089 : : * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2090 : : * various stages of removing an inode.
2091 : : *
2092 : : * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2093 : : *
2094 : : * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2095 : : * fdatasync(). i_atime is the usual cause.
2096 : : * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2097 : : * these changes separately from I_DIRTY_SYNC so that we
2098 : : * don't have to write inode on fdatasync() when only
2099 : : * mtime has changed in it.
2100 : : * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2101 : : * I_NEW Serves as both a mutex and completion notification.
2102 : : * New inodes set I_NEW. If two processes both create
2103 : : * the same inode, one of them will release its inode and
2104 : : * wait for I_NEW to be released before returning.
2105 : : * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2106 : : * also cause waiting on I_NEW, without I_NEW actually
2107 : : * being set. find_inode() uses this to prevent returning
2108 : : * nearly-dead inodes.
2109 : : * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2110 : : * is zero. I_FREEING must be set when I_WILL_FREE is
2111 : : * cleared.
2112 : : * I_FREEING Set when inode is about to be freed but still has dirty
2113 : : * pages or buffers attached or the inode itself is still
2114 : : * dirty.
2115 : : * I_CLEAR Added by clear_inode(). In this state the inode is
2116 : : * clean and can be destroyed. Inode keeps I_FREEING.
2117 : : *
2118 : : * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2119 : : * prohibited for many purposes. iget() must wait for
2120 : : * the inode to be completely released, then create it
2121 : : * anew. Other functions will just ignore such inodes,
2122 : : * if appropriate. I_NEW is used for waiting.
2123 : : *
2124 : : * I_SYNC Writeback of inode is running. The bit is set during
2125 : : * data writeback, and cleared with a wakeup on the bit
2126 : : * address once it is done. The bit is also used to pin
2127 : : * the inode in memory for flusher thread.
2128 : : *
2129 : : * I_REFERENCED Marks the inode as recently references on the LRU list.
2130 : : *
2131 : : * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2132 : : *
2133 : : * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2134 : : * synchronize competing switching instances and to tell
2135 : : * wb stat updates to grab the i_pages lock. See
2136 : : * inode_switch_wbs_work_fn() for details.
2137 : : *
2138 : : * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2139 : : * and work dirs among overlayfs mounts.
2140 : : *
2141 : : * I_CREATING New object's inode in the middle of setting up.
2142 : : *
2143 : : * Q: What is the difference between I_WILL_FREE and I_FREEING?
2144 : : */
2145 : : #define I_DIRTY_SYNC (1 << 0)
2146 : : #define I_DIRTY_DATASYNC (1 << 1)
2147 : : #define I_DIRTY_PAGES (1 << 2)
2148 : : #define __I_NEW 3
2149 : : #define I_NEW (1 << __I_NEW)
2150 : : #define I_WILL_FREE (1 << 4)
2151 : : #define I_FREEING (1 << 5)
2152 : : #define I_CLEAR (1 << 6)
2153 : : #define __I_SYNC 7
2154 : : #define I_SYNC (1 << __I_SYNC)
2155 : : #define I_REFERENCED (1 << 8)
2156 : : #define __I_DIO_WAKEUP 9
2157 : : #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2158 : : #define I_LINKABLE (1 << 10)
2159 : : #define I_DIRTY_TIME (1 << 11)
2160 : : #define __I_DIRTY_TIME_EXPIRED 12
2161 : : #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
2162 : : #define I_WB_SWITCH (1 << 13)
2163 : : #define I_OVL_INUSE (1 << 14)
2164 : : #define I_CREATING (1 << 15)
2165 : :
2166 : : #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2167 : : #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2168 : : #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2169 : :
2170 : : extern void __mark_inode_dirty(struct inode *, int);
2171 : : static inline void mark_inode_dirty(struct inode *inode)
2172 : : {
2173 : 3 : __mark_inode_dirty(inode, I_DIRTY);
2174 : : }
2175 : :
2176 : : static inline void mark_inode_dirty_sync(struct inode *inode)
2177 : : {
2178 : 3 : __mark_inode_dirty(inode, I_DIRTY_SYNC);
2179 : : }
2180 : :
2181 : : extern void inc_nlink(struct inode *inode);
2182 : : extern void drop_nlink(struct inode *inode);
2183 : : extern void clear_nlink(struct inode *inode);
2184 : : extern void set_nlink(struct inode *inode, unsigned int nlink);
2185 : :
2186 : : static inline void inode_inc_link_count(struct inode *inode)
2187 : : {
2188 : : inc_nlink(inode);
2189 : : mark_inode_dirty(inode);
2190 : : }
2191 : :
2192 : : static inline void inode_dec_link_count(struct inode *inode)
2193 : : {
2194 : 3 : drop_nlink(inode);
2195 : : mark_inode_dirty(inode);
2196 : : }
2197 : :
2198 : : enum file_time_flags {
2199 : : S_ATIME = 1,
2200 : : S_MTIME = 2,
2201 : : S_CTIME = 4,
2202 : : S_VERSION = 8,
2203 : : };
2204 : :
2205 : : extern bool atime_needs_update(const struct path *, struct inode *);
2206 : : extern void touch_atime(const struct path *);
2207 : : static inline void file_accessed(struct file *file)
2208 : : {
2209 : 3 : if (!(file->f_flags & O_NOATIME))
2210 : 3 : touch_atime(&file->f_path);
2211 : : }
2212 : :
2213 : : extern int file_modified(struct file *file);
2214 : :
2215 : : int sync_inode(struct inode *inode, struct writeback_control *wbc);
2216 : : int sync_inode_metadata(struct inode *inode, int wait);
2217 : :
2218 : : struct file_system_type {
2219 : : const char *name;
2220 : : int fs_flags;
2221 : : #define FS_REQUIRES_DEV 1
2222 : : #define FS_BINARY_MOUNTDATA 2
2223 : : #define FS_HAS_SUBTYPE 4
2224 : : #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2225 : : #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */
2226 : : #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2227 : : int (*init_fs_context)(struct fs_context *);
2228 : : const struct fs_parameter_description *parameters;
2229 : : struct dentry *(*mount) (struct file_system_type *, int,
2230 : : const char *, void *);
2231 : : void (*kill_sb) (struct super_block *);
2232 : : struct module *owner;
2233 : : struct file_system_type * next;
2234 : : struct hlist_head fs_supers;
2235 : :
2236 : : struct lock_class_key s_lock_key;
2237 : : struct lock_class_key s_umount_key;
2238 : : struct lock_class_key s_vfs_rename_key;
2239 : : struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2240 : :
2241 : : struct lock_class_key i_lock_key;
2242 : : struct lock_class_key i_mutex_key;
2243 : : struct lock_class_key i_mutex_dir_key;
2244 : : };
2245 : :
2246 : : #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2247 : :
2248 : : #ifdef CONFIG_BLOCK
2249 : : extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2250 : : int flags, const char *dev_name, void *data,
2251 : : int (*fill_super)(struct super_block *, void *, int));
2252 : : #else
2253 : : static inline struct dentry *mount_bdev(struct file_system_type *fs_type,
2254 : : int flags, const char *dev_name, void *data,
2255 : : int (*fill_super)(struct super_block *, void *, int))
2256 : : {
2257 : : return ERR_PTR(-ENODEV);
2258 : : }
2259 : : #endif
2260 : : extern struct dentry *mount_single(struct file_system_type *fs_type,
2261 : : int flags, void *data,
2262 : : int (*fill_super)(struct super_block *, void *, int));
2263 : : extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2264 : : int flags, void *data,
2265 : : int (*fill_super)(struct super_block *, void *, int));
2266 : : extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2267 : : void generic_shutdown_super(struct super_block *sb);
2268 : : #ifdef CONFIG_BLOCK
2269 : : void kill_block_super(struct super_block *sb);
2270 : : #else
2271 : : static inline void kill_block_super(struct super_block *sb)
2272 : : {
2273 : : BUG();
2274 : : }
2275 : : #endif
2276 : : void kill_anon_super(struct super_block *sb);
2277 : : void kill_litter_super(struct super_block *sb);
2278 : : void deactivate_super(struct super_block *sb);
2279 : : void deactivate_locked_super(struct super_block *sb);
2280 : : int set_anon_super(struct super_block *s, void *data);
2281 : : int set_anon_super_fc(struct super_block *s, struct fs_context *fc);
2282 : : int get_anon_bdev(dev_t *);
2283 : : void free_anon_bdev(dev_t);
2284 : : struct super_block *sget_fc(struct fs_context *fc,
2285 : : int (*test)(struct super_block *, struct fs_context *),
2286 : : int (*set)(struct super_block *, struct fs_context *));
2287 : : struct super_block *sget(struct file_system_type *type,
2288 : : int (*test)(struct super_block *,void *),
2289 : : int (*set)(struct super_block *,void *),
2290 : : int flags, void *data);
2291 : :
2292 : : /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2293 : : #define fops_get(fops) \
2294 : : (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2295 : : #define fops_put(fops) \
2296 : : do { if (fops) module_put((fops)->owner); } while(0)
2297 : : /*
2298 : : * This one is to be used *ONLY* from ->open() instances.
2299 : : * fops must be non-NULL, pinned down *and* module dependencies
2300 : : * should be sufficient to pin the caller down as well.
2301 : : */
2302 : : #define replace_fops(f, fops) \
2303 : : do { \
2304 : : struct file *__file = (f); \
2305 : : fops_put(__file->f_op); \
2306 : : BUG_ON(!(__file->f_op = (fops))); \
2307 : : } while(0)
2308 : :
2309 : : extern int register_filesystem(struct file_system_type *);
2310 : : extern int unregister_filesystem(struct file_system_type *);
2311 : : extern struct vfsmount *kern_mount(struct file_system_type *);
2312 : : extern void kern_unmount(struct vfsmount *mnt);
2313 : : extern int may_umount_tree(struct vfsmount *);
2314 : : extern int may_umount(struct vfsmount *);
2315 : : extern long do_mount(const char *, const char __user *,
2316 : : const char *, unsigned long, void *);
2317 : : extern struct vfsmount *collect_mounts(const struct path *);
2318 : : extern void drop_collected_mounts(struct vfsmount *);
2319 : : extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2320 : : struct vfsmount *);
2321 : : extern int vfs_statfs(const struct path *, struct kstatfs *);
2322 : : extern int user_statfs(const char __user *, struct kstatfs *);
2323 : : extern int fd_statfs(int, struct kstatfs *);
2324 : : extern int freeze_super(struct super_block *super);
2325 : : extern int thaw_super(struct super_block *super);
2326 : : extern bool our_mnt(struct vfsmount *mnt);
2327 : : extern __printf(2, 3)
2328 : : int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2329 : : extern int super_setup_bdi(struct super_block *sb);
2330 : :
2331 : : extern int current_umask(void);
2332 : :
2333 : : extern void ihold(struct inode * inode);
2334 : : extern void iput(struct inode *);
2335 : : extern int generic_update_time(struct inode *, struct timespec64 *, int);
2336 : :
2337 : : /* /sys/fs */
2338 : : extern struct kobject *fs_kobj;
2339 : :
2340 : : #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2341 : :
2342 : : #ifdef CONFIG_MANDATORY_FILE_LOCKING
2343 : : extern int locks_mandatory_locked(struct file *);
2344 : : extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2345 : :
2346 : : /*
2347 : : * Candidates for mandatory locking have the setgid bit set
2348 : : * but no group execute bit - an otherwise meaningless combination.
2349 : : */
2350 : :
2351 : : static inline int __mandatory_lock(struct inode *ino)
2352 : : {
2353 : 0 : return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2354 : : }
2355 : :
2356 : : /*
2357 : : * ... and these candidates should be on SB_MANDLOCK mounted fs,
2358 : : * otherwise these will be advisory locks
2359 : : */
2360 : :
2361 : : static inline int mandatory_lock(struct inode *ino)
2362 : : {
2363 : 3 : return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2364 : : }
2365 : :
2366 : 3 : static inline int locks_verify_locked(struct file *file)
2367 : : {
2368 : 3 : if (mandatory_lock(locks_inode(file)))
2369 : 0 : return locks_mandatory_locked(file);
2370 : : return 0;
2371 : : }
2372 : :
2373 : 3 : static inline int locks_verify_truncate(struct inode *inode,
2374 : : struct file *f,
2375 : : loff_t size)
2376 : : {
2377 : 3 : if (!inode->i_flctx || !mandatory_lock(inode))
2378 : : return 0;
2379 : :
2380 : 0 : if (size < inode->i_size) {
2381 : 0 : return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2382 : : F_WRLCK);
2383 : : } else {
2384 : 0 : return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2385 : : F_WRLCK);
2386 : : }
2387 : : }
2388 : :
2389 : : #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2390 : :
2391 : : static inline int locks_mandatory_locked(struct file *file)
2392 : : {
2393 : : return 0;
2394 : : }
2395 : :
2396 : : static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2397 : : loff_t start, loff_t end, unsigned char type)
2398 : : {
2399 : : return 0;
2400 : : }
2401 : :
2402 : : static inline int __mandatory_lock(struct inode *inode)
2403 : : {
2404 : : return 0;
2405 : : }
2406 : :
2407 : : static inline int mandatory_lock(struct inode *inode)
2408 : : {
2409 : : return 0;
2410 : : }
2411 : :
2412 : : static inline int locks_verify_locked(struct file *file)
2413 : : {
2414 : : return 0;
2415 : : }
2416 : :
2417 : : static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2418 : : size_t size)
2419 : : {
2420 : : return 0;
2421 : : }
2422 : :
2423 : : #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2424 : :
2425 : :
2426 : : #ifdef CONFIG_FILE_LOCKING
2427 : 3 : static inline int break_lease(struct inode *inode, unsigned int mode)
2428 : : {
2429 : : /*
2430 : : * Since this check is lockless, we must ensure that any refcounts
2431 : : * taken are done before checking i_flctx->flc_lease. Otherwise, we
2432 : : * could end up racing with tasks trying to set a new lease on this
2433 : : * file.
2434 : : */
2435 : 3 : smp_mb();
2436 : 3 : if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2437 : 0 : return __break_lease(inode, mode, FL_LEASE);
2438 : : return 0;
2439 : : }
2440 : :
2441 : 3 : static inline int break_deleg(struct inode *inode, unsigned int mode)
2442 : : {
2443 : : /*
2444 : : * Since this check is lockless, we must ensure that any refcounts
2445 : : * taken are done before checking i_flctx->flc_lease. Otherwise, we
2446 : : * could end up racing with tasks trying to set a new lease on this
2447 : : * file.
2448 : : */
2449 : 3 : smp_mb();
2450 : 3 : if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2451 : 0 : return __break_lease(inode, mode, FL_DELEG);
2452 : : return 0;
2453 : : }
2454 : :
2455 : 3 : static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2456 : : {
2457 : : int ret;
2458 : :
2459 : 3 : ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2460 : 3 : if (ret == -EWOULDBLOCK && delegated_inode) {
2461 : 0 : *delegated_inode = inode;
2462 : 0 : ihold(inode);
2463 : : }
2464 : 3 : return ret;
2465 : : }
2466 : :
2467 : 0 : static inline int break_deleg_wait(struct inode **delegated_inode)
2468 : : {
2469 : : int ret;
2470 : :
2471 : 0 : ret = break_deleg(*delegated_inode, O_WRONLY);
2472 : 0 : iput(*delegated_inode);
2473 : 0 : *delegated_inode = NULL;
2474 : 0 : return ret;
2475 : : }
2476 : :
2477 : : static inline int break_layout(struct inode *inode, bool wait)
2478 : : {
2479 : : smp_mb();
2480 : : if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2481 : : return __break_lease(inode,
2482 : : wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2483 : : FL_LAYOUT);
2484 : : return 0;
2485 : : }
2486 : :
2487 : : #else /* !CONFIG_FILE_LOCKING */
2488 : : static inline int break_lease(struct inode *inode, unsigned int mode)
2489 : : {
2490 : : return 0;
2491 : : }
2492 : :
2493 : : static inline int break_deleg(struct inode *inode, unsigned int mode)
2494 : : {
2495 : : return 0;
2496 : : }
2497 : :
2498 : : static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2499 : : {
2500 : : return 0;
2501 : : }
2502 : :
2503 : : static inline int break_deleg_wait(struct inode **delegated_inode)
2504 : : {
2505 : : BUG();
2506 : : return 0;
2507 : : }
2508 : :
2509 : : static inline int break_layout(struct inode *inode, bool wait)
2510 : : {
2511 : : return 0;
2512 : : }
2513 : :
2514 : : #endif /* CONFIG_FILE_LOCKING */
2515 : :
2516 : : /* fs/open.c */
2517 : : struct audit_names;
2518 : : struct filename {
2519 : : const char *name; /* pointer to actual string */
2520 : : const __user char *uptr; /* original userland pointer */
2521 : : int refcnt;
2522 : : struct audit_names *aname;
2523 : : const char iname[];
2524 : : };
2525 : : static_assert(offsetof(struct filename, iname) % sizeof(long) == 0);
2526 : :
2527 : : extern long vfs_truncate(const struct path *, loff_t);
2528 : : extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2529 : : struct file *filp);
2530 : : extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2531 : : loff_t len);
2532 : : extern long do_sys_open(int dfd, const char __user *filename, int flags,
2533 : : umode_t mode);
2534 : : extern struct file *file_open_name(struct filename *, int, umode_t);
2535 : : extern struct file *filp_open(const char *, int, umode_t);
2536 : : extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2537 : : const char *, int, umode_t);
2538 : : extern struct file * dentry_open(const struct path *, int, const struct cred *);
2539 : : extern struct file * open_with_fake_path(const struct path *, int,
2540 : : struct inode*, const struct cred *);
2541 : : static inline struct file *file_clone_open(struct file *file)
2542 : : {
2543 : : return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2544 : : }
2545 : : extern int filp_close(struct file *, fl_owner_t id);
2546 : :
2547 : : extern struct filename *getname_flags(const char __user *, int, int *);
2548 : : extern struct filename *getname(const char __user *);
2549 : : extern struct filename *getname_kernel(const char *);
2550 : : extern void putname(struct filename *name);
2551 : :
2552 : : extern int finish_open(struct file *file, struct dentry *dentry,
2553 : : int (*open)(struct inode *, struct file *));
2554 : : extern int finish_no_open(struct file *file, struct dentry *dentry);
2555 : :
2556 : : /* fs/ioctl.c */
2557 : :
2558 : : extern int ioctl_preallocate(struct file *filp, void __user *argp);
2559 : :
2560 : : /* fs/dcache.c */
2561 : : extern void __init vfs_caches_init_early(void);
2562 : : extern void __init vfs_caches_init(void);
2563 : :
2564 : : extern struct kmem_cache *names_cachep;
2565 : :
2566 : : #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2567 : : #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2568 : :
2569 : : #ifdef CONFIG_BLOCK
2570 : : extern int register_blkdev(unsigned int, const char *);
2571 : : extern void unregister_blkdev(unsigned int, const char *);
2572 : : extern void bdev_unhash_inode(dev_t dev);
2573 : : extern struct block_device *bdget(dev_t);
2574 : : extern struct block_device *bdgrab(struct block_device *bdev);
2575 : : extern void bd_set_size(struct block_device *, loff_t size);
2576 : : extern void bd_forget(struct inode *inode);
2577 : : extern void bdput(struct block_device *);
2578 : : extern void invalidate_bdev(struct block_device *);
2579 : : extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2580 : : extern int sync_blockdev(struct block_device *bdev);
2581 : : extern void kill_bdev(struct block_device *);
2582 : : extern struct super_block *freeze_bdev(struct block_device *);
2583 : : extern void emergency_thaw_all(void);
2584 : : extern void emergency_thaw_bdev(struct super_block *sb);
2585 : : extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2586 : : extern int fsync_bdev(struct block_device *);
2587 : :
2588 : : extern struct super_block *blockdev_superblock;
2589 : :
2590 : : static inline bool sb_is_blkdev_sb(struct super_block *sb)
2591 : : {
2592 : 3 : return sb == blockdev_superblock;
2593 : : }
2594 : : #else
2595 : : static inline void bd_forget(struct inode *inode) {}
2596 : : static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2597 : : static inline void kill_bdev(struct block_device *bdev) {}
2598 : : static inline void invalidate_bdev(struct block_device *bdev) {}
2599 : :
2600 : : static inline struct super_block *freeze_bdev(struct block_device *sb)
2601 : : {
2602 : : return NULL;
2603 : : }
2604 : :
2605 : : static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2606 : : {
2607 : : return 0;
2608 : : }
2609 : :
2610 : : static inline int emergency_thaw_bdev(struct super_block *sb)
2611 : : {
2612 : : return 0;
2613 : : }
2614 : :
2615 : : static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2616 : : {
2617 : : }
2618 : :
2619 : : static inline bool sb_is_blkdev_sb(struct super_block *sb)
2620 : : {
2621 : : return false;
2622 : : }
2623 : : #endif
2624 : : extern int sync_filesystem(struct super_block *);
2625 : : extern const struct file_operations def_blk_fops;
2626 : : extern const struct file_operations def_chr_fops;
2627 : : #ifdef CONFIG_BLOCK
2628 : : extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2629 : : extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2630 : : extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2631 : : extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2632 : : extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2633 : : void *holder);
2634 : : extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2635 : : void *holder);
2636 : : extern struct block_device *bd_start_claiming(struct block_device *bdev,
2637 : : void *holder);
2638 : : extern void bd_finish_claiming(struct block_device *bdev,
2639 : : struct block_device *whole, void *holder);
2640 : : extern void bd_abort_claiming(struct block_device *bdev,
2641 : : struct block_device *whole, void *holder);
2642 : : extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2643 : : extern int __blkdev_reread_part(struct block_device *bdev);
2644 : : extern int blkdev_reread_part(struct block_device *bdev);
2645 : :
2646 : : #ifdef CONFIG_SYSFS
2647 : : extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2648 : : extern void bd_unlink_disk_holder(struct block_device *bdev,
2649 : : struct gendisk *disk);
2650 : : #else
2651 : : static inline int bd_link_disk_holder(struct block_device *bdev,
2652 : : struct gendisk *disk)
2653 : : {
2654 : : return 0;
2655 : : }
2656 : : static inline void bd_unlink_disk_holder(struct block_device *bdev,
2657 : : struct gendisk *disk)
2658 : : {
2659 : : }
2660 : : #endif
2661 : : #endif
2662 : :
2663 : : /* fs/char_dev.c */
2664 : : #define CHRDEV_MAJOR_MAX 512
2665 : : /* Marks the bottom of the first segment of free char majors */
2666 : : #define CHRDEV_MAJOR_DYN_END 234
2667 : : /* Marks the top and bottom of the second segment of free char majors */
2668 : : #define CHRDEV_MAJOR_DYN_EXT_START 511
2669 : : #define CHRDEV_MAJOR_DYN_EXT_END 384
2670 : :
2671 : : extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2672 : : extern int register_chrdev_region(dev_t, unsigned, const char *);
2673 : : extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2674 : : unsigned int count, const char *name,
2675 : : const struct file_operations *fops);
2676 : : extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2677 : : unsigned int count, const char *name);
2678 : : extern void unregister_chrdev_region(dev_t, unsigned);
2679 : : extern void chrdev_show(struct seq_file *,off_t);
2680 : :
2681 : : static inline int register_chrdev(unsigned int major, const char *name,
2682 : : const struct file_operations *fops)
2683 : : {
2684 : 3 : return __register_chrdev(major, 0, 256, name, fops);
2685 : : }
2686 : :
2687 : : static inline void unregister_chrdev(unsigned int major, const char *name)
2688 : : {
2689 : 0 : __unregister_chrdev(major, 0, 256, name);
2690 : : }
2691 : :
2692 : : /* fs/block_dev.c */
2693 : : #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2694 : : #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2695 : :
2696 : : #ifdef CONFIG_BLOCK
2697 : : #define BLKDEV_MAJOR_MAX 512
2698 : : extern const char *__bdevname(dev_t, char *buffer);
2699 : : extern const char *bdevname(struct block_device *bdev, char *buffer);
2700 : : extern struct block_device *lookup_bdev(const char *);
2701 : : extern void blkdev_show(struct seq_file *,off_t);
2702 : :
2703 : : #else
2704 : : #define BLKDEV_MAJOR_MAX 0
2705 : : #endif
2706 : :
2707 : : extern void init_special_inode(struct inode *, umode_t, dev_t);
2708 : :
2709 : : /* Invalid inode operations -- fs/bad_inode.c */
2710 : : extern void make_bad_inode(struct inode *);
2711 : : extern bool is_bad_inode(struct inode *);
2712 : :
2713 : : #ifdef CONFIG_BLOCK
2714 : : extern void check_disk_size_change(struct gendisk *disk,
2715 : : struct block_device *bdev, bool verbose);
2716 : : extern int revalidate_disk(struct gendisk *);
2717 : : extern int check_disk_change(struct block_device *);
2718 : : extern int __invalidate_device(struct block_device *, bool);
2719 : : extern int invalidate_partition(struct gendisk *, int);
2720 : : #endif
2721 : : unsigned long invalidate_mapping_pages(struct address_space *mapping,
2722 : : pgoff_t start, pgoff_t end);
2723 : :
2724 : : static inline void invalidate_remote_inode(struct inode *inode)
2725 : : {
2726 : : if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2727 : : S_ISLNK(inode->i_mode))
2728 : : invalidate_mapping_pages(inode->i_mapping, 0, -1);
2729 : : }
2730 : : extern int invalidate_inode_pages2(struct address_space *mapping);
2731 : : extern int invalidate_inode_pages2_range(struct address_space *mapping,
2732 : : pgoff_t start, pgoff_t end);
2733 : : extern int write_inode_now(struct inode *, int);
2734 : : extern int filemap_fdatawrite(struct address_space *);
2735 : : extern int filemap_flush(struct address_space *);
2736 : : extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2737 : : extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2738 : : loff_t lend);
2739 : : extern int filemap_fdatawait_range_keep_errors(struct address_space *mapping,
2740 : : loff_t start_byte, loff_t end_byte);
2741 : :
2742 : 3 : static inline int filemap_fdatawait(struct address_space *mapping)
2743 : : {
2744 : 3 : return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2745 : : }
2746 : :
2747 : : extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2748 : : loff_t lend);
2749 : : extern int filemap_write_and_wait(struct address_space *mapping);
2750 : : extern int filemap_write_and_wait_range(struct address_space *mapping,
2751 : : loff_t lstart, loff_t lend);
2752 : : extern int __filemap_fdatawrite_range(struct address_space *mapping,
2753 : : loff_t start, loff_t end, int sync_mode);
2754 : : extern int filemap_fdatawrite_range(struct address_space *mapping,
2755 : : loff_t start, loff_t end);
2756 : : extern int filemap_check_errors(struct address_space *mapping);
2757 : : extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2758 : :
2759 : : extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2760 : : loff_t lend);
2761 : : extern int __must_check file_check_and_advance_wb_err(struct file *file);
2762 : : extern int __must_check file_write_and_wait_range(struct file *file,
2763 : : loff_t start, loff_t end);
2764 : :
2765 : : static inline int file_write_and_wait(struct file *file)
2766 : : {
2767 : : return file_write_and_wait_range(file, 0, LLONG_MAX);
2768 : : }
2769 : :
2770 : : /**
2771 : : * filemap_set_wb_err - set a writeback error on an address_space
2772 : : * @mapping: mapping in which to set writeback error
2773 : : * @err: error to be set in mapping
2774 : : *
2775 : : * When writeback fails in some way, we must record that error so that
2776 : : * userspace can be informed when fsync and the like are called. We endeavor
2777 : : * to report errors on any file that was open at the time of the error. Some
2778 : : * internal callers also need to know when writeback errors have occurred.
2779 : : *
2780 : : * When a writeback error occurs, most filesystems will want to call
2781 : : * filemap_set_wb_err to record the error in the mapping so that it will be
2782 : : * automatically reported whenever fsync is called on the file.
2783 : : */
2784 : : static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2785 : : {
2786 : : /* Fastpath for common case of no error */
2787 : 0 : if (unlikely(err))
2788 : 0 : __filemap_set_wb_err(mapping, err);
2789 : : }
2790 : :
2791 : : /**
2792 : : * filemap_check_wb_error - has an error occurred since the mark was sampled?
2793 : : * @mapping: mapping to check for writeback errors
2794 : : * @since: previously-sampled errseq_t
2795 : : *
2796 : : * Grab the errseq_t value from the mapping, and see if it has changed "since"
2797 : : * the given value was sampled.
2798 : : *
2799 : : * If it has then report the latest error set, otherwise return 0.
2800 : : */
2801 : : static inline int filemap_check_wb_err(struct address_space *mapping,
2802 : : errseq_t since)
2803 : : {
2804 : 0 : return errseq_check(&mapping->wb_err, since);
2805 : : }
2806 : :
2807 : : /**
2808 : : * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2809 : : * @mapping: mapping to be sampled
2810 : : *
2811 : : * Writeback errors are always reported relative to a particular sample point
2812 : : * in the past. This function provides those sample points.
2813 : : */
2814 : : static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2815 : : {
2816 : 3 : return errseq_sample(&mapping->wb_err);
2817 : : }
2818 : :
2819 : : static inline int filemap_nr_thps(struct address_space *mapping)
2820 : : {
2821 : : #ifdef CONFIG_READ_ONLY_THP_FOR_FS
2822 : : return atomic_read(&mapping->nr_thps);
2823 : : #else
2824 : : return 0;
2825 : : #endif
2826 : : }
2827 : :
2828 : : static inline void filemap_nr_thps_inc(struct address_space *mapping)
2829 : : {
2830 : : #ifdef CONFIG_READ_ONLY_THP_FOR_FS
2831 : : atomic_inc(&mapping->nr_thps);
2832 : : #else
2833 : : WARN_ON_ONCE(1);
2834 : : #endif
2835 : : }
2836 : :
2837 : : static inline void filemap_nr_thps_dec(struct address_space *mapping)
2838 : : {
2839 : : #ifdef CONFIG_READ_ONLY_THP_FOR_FS
2840 : : atomic_dec(&mapping->nr_thps);
2841 : : #else
2842 : : WARN_ON_ONCE(1);
2843 : : #endif
2844 : : }
2845 : :
2846 : : extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2847 : : int datasync);
2848 : : extern int vfs_fsync(struct file *file, int datasync);
2849 : :
2850 : : extern int sync_file_range(struct file *file, loff_t offset, loff_t nbytes,
2851 : : unsigned int flags);
2852 : :
2853 : : /*
2854 : : * Sync the bytes written if this was a synchronous write. Expect ki_pos
2855 : : * to already be updated for the write, and will return either the amount
2856 : : * of bytes passed in, or an error if syncing the file failed.
2857 : : */
2858 : 3 : static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2859 : : {
2860 : 3 : if (iocb->ki_flags & IOCB_DSYNC) {
2861 : 0 : int ret = vfs_fsync_range(iocb->ki_filp,
2862 : 0 : iocb->ki_pos - count, iocb->ki_pos - 1,
2863 : 0 : (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2864 : 0 : if (ret)
2865 : : return ret;
2866 : : }
2867 : :
2868 : 3 : return count;
2869 : : }
2870 : :
2871 : : extern void emergency_sync(void);
2872 : : extern void emergency_remount(void);
2873 : : #ifdef CONFIG_BLOCK
2874 : : extern sector_t bmap(struct inode *, sector_t);
2875 : : #endif
2876 : : extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2877 : : extern int inode_permission(struct inode *, int);
2878 : : extern int generic_permission(struct inode *, int);
2879 : : extern int __check_sticky(struct inode *dir, struct inode *inode);
2880 : :
2881 : : static inline bool execute_ok(struct inode *inode)
2882 : : {
2883 : 0 : return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2884 : : }
2885 : :
2886 : 3 : static inline void file_start_write(struct file *file)
2887 : : {
2888 : 3 : if (!S_ISREG(file_inode(file)->i_mode))
2889 : 3 : return;
2890 : 3 : __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2891 : : }
2892 : :
2893 : 0 : static inline bool file_start_write_trylock(struct file *file)
2894 : : {
2895 : 0 : if (!S_ISREG(file_inode(file)->i_mode))
2896 : : return true;
2897 : 0 : return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2898 : : }
2899 : :
2900 : 3 : static inline void file_end_write(struct file *file)
2901 : : {
2902 : 3 : if (!S_ISREG(file_inode(file)->i_mode))
2903 : 3 : return;
2904 : 3 : __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2905 : : }
2906 : :
2907 : : /*
2908 : : * get_write_access() gets write permission for a file.
2909 : : * put_write_access() releases this write permission.
2910 : : * This is used for regular files.
2911 : : * We cannot support write (and maybe mmap read-write shared) accesses and
2912 : : * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2913 : : * can have the following values:
2914 : : * 0: no writers, no VM_DENYWRITE mappings
2915 : : * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2916 : : * > 0: (i_writecount) users are writing to the file.
2917 : : *
2918 : : * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2919 : : * except for the cases where we don't hold i_writecount yet. Then we need to
2920 : : * use {get,deny}_write_access() - these functions check the sign and refuse
2921 : : * to do the change if sign is wrong.
2922 : : */
2923 : : static inline int get_write_access(struct inode *inode)
2924 : : {
2925 : 3 : return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2926 : : }
2927 : : static inline int deny_write_access(struct file *file)
2928 : : {
2929 : : struct inode *inode = file_inode(file);
2930 : 3 : return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2931 : : }
2932 : : static inline void put_write_access(struct inode * inode)
2933 : : {
2934 : 3 : atomic_dec(&inode->i_writecount);
2935 : : }
2936 : 3 : static inline void allow_write_access(struct file *file)
2937 : : {
2938 : 3 : if (file)
2939 : 3 : atomic_inc(&file_inode(file)->i_writecount);
2940 : 3 : }
2941 : : static inline bool inode_is_open_for_write(const struct inode *inode)
2942 : : {
2943 : 3 : return atomic_read(&inode->i_writecount) > 0;
2944 : : }
2945 : :
2946 : : #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
2947 : 3 : static inline void i_readcount_dec(struct inode *inode)
2948 : : {
2949 : 3 : BUG_ON(!atomic_read(&inode->i_readcount));
2950 : 3 : atomic_dec(&inode->i_readcount);
2951 : 3 : }
2952 : : static inline void i_readcount_inc(struct inode *inode)
2953 : : {
2954 : 3 : atomic_inc(&inode->i_readcount);
2955 : : }
2956 : : #else
2957 : : static inline void i_readcount_dec(struct inode *inode)
2958 : : {
2959 : : return;
2960 : : }
2961 : : static inline void i_readcount_inc(struct inode *inode)
2962 : : {
2963 : : return;
2964 : : }
2965 : : #endif
2966 : : extern int do_pipe_flags(int *, int);
2967 : :
2968 : : #define __kernel_read_file_id(id) \
2969 : : id(UNKNOWN, unknown) \
2970 : : id(FIRMWARE, firmware) \
2971 : : id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2972 : : id(MODULE, kernel-module) \
2973 : : id(KEXEC_IMAGE, kexec-image) \
2974 : : id(KEXEC_INITRAMFS, kexec-initramfs) \
2975 : : id(POLICY, security-policy) \
2976 : : id(X509_CERTIFICATE, x509-certificate) \
2977 : : id(MAX_ID, )
2978 : :
2979 : : #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2980 : : #define __fid_stringify(dummy, str) #str,
2981 : :
2982 : : enum kernel_read_file_id {
2983 : : __kernel_read_file_id(__fid_enumify)
2984 : : };
2985 : :
2986 : : static const char * const kernel_read_file_str[] = {
2987 : : __kernel_read_file_id(__fid_stringify)
2988 : : };
2989 : :
2990 : : static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
2991 : : {
2992 : : if ((unsigned)id >= READING_MAX_ID)
2993 : : return kernel_read_file_str[READING_UNKNOWN];
2994 : :
2995 : : return kernel_read_file_str[id];
2996 : : }
2997 : :
2998 : : extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2999 : : enum kernel_read_file_id);
3000 : : extern int kernel_read_file_from_path(const char *, void **, loff_t *, loff_t,
3001 : : enum kernel_read_file_id);
3002 : : extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
3003 : : enum kernel_read_file_id);
3004 : : extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
3005 : : extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
3006 : : extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
3007 : : extern struct file * open_exec(const char *);
3008 : :
3009 : : /* fs/dcache.c -- generic fs support functions */
3010 : : extern bool is_subdir(struct dentry *, struct dentry *);
3011 : : extern bool path_is_under(const struct path *, const struct path *);
3012 : :
3013 : : extern char *file_path(struct file *, char *, int);
3014 : :
3015 : : #include <linux/err.h>
3016 : :
3017 : : /* needed for stackable file system support */
3018 : : extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
3019 : :
3020 : : extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
3021 : :
3022 : : extern int inode_init_always(struct super_block *, struct inode *);
3023 : : extern void inode_init_once(struct inode *);
3024 : : extern void address_space_init_once(struct address_space *mapping);
3025 : : extern struct inode * igrab(struct inode *);
3026 : : extern ino_t iunique(struct super_block *, ino_t);
3027 : : extern int inode_needs_sync(struct inode *inode);
3028 : : extern int generic_delete_inode(struct inode *inode);
3029 : : static inline int generic_drop_inode(struct inode *inode)
3030 : : {
3031 : 3 : return !inode->i_nlink || inode_unhashed(inode);
3032 : : }
3033 : :
3034 : : extern struct inode *ilookup5_nowait(struct super_block *sb,
3035 : : unsigned long hashval, int (*test)(struct inode *, void *),
3036 : : void *data);
3037 : : extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
3038 : : int (*test)(struct inode *, void *), void *data);
3039 : : extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
3040 : :
3041 : : extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
3042 : : int (*test)(struct inode *, void *),
3043 : : int (*set)(struct inode *, void *),
3044 : : void *data);
3045 : : extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
3046 : : extern struct inode * iget_locked(struct super_block *, unsigned long);
3047 : : extern struct inode *find_inode_nowait(struct super_block *,
3048 : : unsigned long,
3049 : : int (*match)(struct inode *,
3050 : : unsigned long, void *),
3051 : : void *data);
3052 : : extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
3053 : : extern int insert_inode_locked(struct inode *);
3054 : : #ifdef CONFIG_DEBUG_LOCK_ALLOC
3055 : : extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
3056 : : #else
3057 : : static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
3058 : : #endif
3059 : : extern void unlock_new_inode(struct inode *);
3060 : : extern void discard_new_inode(struct inode *);
3061 : : extern unsigned int get_next_ino(void);
3062 : : extern void evict_inodes(struct super_block *sb);
3063 : :
3064 : : extern void __iget(struct inode * inode);
3065 : : extern void iget_failed(struct inode *);
3066 : : extern void clear_inode(struct inode *);
3067 : : extern void __destroy_inode(struct inode *);
3068 : : extern struct inode *new_inode_pseudo(struct super_block *sb);
3069 : : extern struct inode *new_inode(struct super_block *sb);
3070 : : extern void free_inode_nonrcu(struct inode *inode);
3071 : : extern int should_remove_suid(struct dentry *);
3072 : : extern int file_remove_privs(struct file *);
3073 : :
3074 : : extern void __insert_inode_hash(struct inode *, unsigned long hashval);
3075 : : static inline void insert_inode_hash(struct inode *inode)
3076 : : {
3077 : 3 : __insert_inode_hash(inode, inode->i_ino);
3078 : : }
3079 : :
3080 : : extern void __remove_inode_hash(struct inode *);
3081 : : static inline void remove_inode_hash(struct inode *inode)
3082 : : {
3083 : 3 : if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
3084 : 3 : __remove_inode_hash(inode);
3085 : : }
3086 : :
3087 : : extern void inode_sb_list_add(struct inode *inode);
3088 : :
3089 : : #ifdef CONFIG_BLOCK
3090 : : extern int bdev_read_only(struct block_device *);
3091 : : #endif
3092 : : extern int set_blocksize(struct block_device *, int);
3093 : : extern int sb_set_blocksize(struct super_block *, int);
3094 : : extern int sb_min_blocksize(struct super_block *, int);
3095 : :
3096 : : extern int generic_file_mmap(struct file *, struct vm_area_struct *);
3097 : : extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
3098 : : extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
3099 : : extern int generic_remap_checks(struct file *file_in, loff_t pos_in,
3100 : : struct file *file_out, loff_t pos_out,
3101 : : loff_t *count, unsigned int remap_flags);
3102 : : extern int generic_file_rw_checks(struct file *file_in, struct file *file_out);
3103 : : extern int generic_copy_file_checks(struct file *file_in, loff_t pos_in,
3104 : : struct file *file_out, loff_t pos_out,
3105 : : size_t *count, unsigned int flags);
3106 : : extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
3107 : : extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
3108 : : extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
3109 : : extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
3110 : : extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
3111 : :
3112 : : ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
3113 : : rwf_t flags);
3114 : : ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
3115 : : rwf_t flags);
3116 : :
3117 : : /* fs/block_dev.c */
3118 : : extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
3119 : : extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
3120 : : extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
3121 : : int datasync);
3122 : : extern void block_sync_page(struct page *page);
3123 : :
3124 : : /* fs/splice.c */
3125 : : extern ssize_t generic_file_splice_read(struct file *, loff_t *,
3126 : : struct pipe_inode_info *, size_t, unsigned int);
3127 : : extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
3128 : : struct file *, loff_t *, size_t, unsigned int);
3129 : : extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
3130 : : struct file *out, loff_t *, size_t len, unsigned int flags);
3131 : : extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
3132 : : loff_t *opos, size_t len, unsigned int flags);
3133 : :
3134 : :
3135 : : extern void
3136 : : file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3137 : : extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3138 : : extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
3139 : : extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3140 : : extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3141 : : extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3142 : : int whence, loff_t maxsize, loff_t eof);
3143 : : extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3144 : : int whence, loff_t size);
3145 : : extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3146 : : extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3147 : : extern int generic_file_open(struct inode * inode, struct file * filp);
3148 : : extern int nonseekable_open(struct inode * inode, struct file * filp);
3149 : : extern int stream_open(struct inode * inode, struct file * filp);
3150 : :
3151 : : #ifdef CONFIG_BLOCK
3152 : : typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3153 : : loff_t file_offset);
3154 : :
3155 : : enum {
3156 : : /* need locking between buffered and direct access */
3157 : : DIO_LOCKING = 0x01,
3158 : :
3159 : : /* filesystem does not support filling holes */
3160 : : DIO_SKIP_HOLES = 0x02,
3161 : : };
3162 : :
3163 : : void dio_end_io(struct bio *bio);
3164 : : void dio_warn_stale_pagecache(struct file *filp);
3165 : :
3166 : : ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3167 : : struct block_device *bdev, struct iov_iter *iter,
3168 : : get_block_t get_block,
3169 : : dio_iodone_t end_io, dio_submit_t submit_io,
3170 : : int flags);
3171 : :
3172 : 0 : static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3173 : : struct inode *inode,
3174 : : struct iov_iter *iter,
3175 : : get_block_t get_block)
3176 : : {
3177 : 0 : return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3178 : : get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3179 : : }
3180 : : #endif
3181 : :
3182 : : void inode_dio_wait(struct inode *inode);
3183 : :
3184 : : /*
3185 : : * inode_dio_begin - signal start of a direct I/O requests
3186 : : * @inode: inode the direct I/O happens on
3187 : : *
3188 : : * This is called once we've finished processing a direct I/O request,
3189 : : * and is used to wake up callers waiting for direct I/O to be quiesced.
3190 : : */
3191 : : static inline void inode_dio_begin(struct inode *inode)
3192 : : {
3193 : 0 : atomic_inc(&inode->i_dio_count);
3194 : : }
3195 : :
3196 : : /*
3197 : : * inode_dio_end - signal finish of a direct I/O requests
3198 : : * @inode: inode the direct I/O happens on
3199 : : *
3200 : : * This is called once we've finished processing a direct I/O request,
3201 : : * and is used to wake up callers waiting for direct I/O to be quiesced.
3202 : : */
3203 : 0 : static inline void inode_dio_end(struct inode *inode)
3204 : : {
3205 : 0 : if (atomic_dec_and_test(&inode->i_dio_count))
3206 : 0 : wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3207 : 0 : }
3208 : :
3209 : : extern void inode_set_flags(struct inode *inode, unsigned int flags,
3210 : : unsigned int mask);
3211 : :
3212 : : extern const struct file_operations generic_ro_fops;
3213 : :
3214 : : #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3215 : :
3216 : : extern int readlink_copy(char __user *, int, const char *);
3217 : : extern int page_readlink(struct dentry *, char __user *, int);
3218 : : extern const char *page_get_link(struct dentry *, struct inode *,
3219 : : struct delayed_call *);
3220 : : extern void page_put_link(void *);
3221 : : extern int __page_symlink(struct inode *inode, const char *symname, int len,
3222 : : int nofs);
3223 : : extern int page_symlink(struct inode *inode, const char *symname, int len);
3224 : : extern const struct inode_operations page_symlink_inode_operations;
3225 : : extern void kfree_link(void *);
3226 : : extern void generic_fillattr(struct inode *, struct kstat *);
3227 : : extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3228 : : extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3229 : : void __inode_add_bytes(struct inode *inode, loff_t bytes);
3230 : : void inode_add_bytes(struct inode *inode, loff_t bytes);
3231 : : void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3232 : : void inode_sub_bytes(struct inode *inode, loff_t bytes);
3233 : : static inline loff_t __inode_get_bytes(struct inode *inode)
3234 : : {
3235 : 0 : return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3236 : : }
3237 : : loff_t inode_get_bytes(struct inode *inode);
3238 : : void inode_set_bytes(struct inode *inode, loff_t bytes);
3239 : : const char *simple_get_link(struct dentry *, struct inode *,
3240 : : struct delayed_call *);
3241 : : extern const struct inode_operations simple_symlink_inode_operations;
3242 : :
3243 : : extern int iterate_dir(struct file *, struct dir_context *);
3244 : :
3245 : : extern int vfs_statx(int, const char __user *, int, struct kstat *, u32);
3246 : : extern int vfs_statx_fd(unsigned int, struct kstat *, u32, unsigned int);
3247 : :
3248 : : static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3249 : : {
3250 : 3 : return vfs_statx(AT_FDCWD, filename, AT_NO_AUTOMOUNT,
3251 : : stat, STATX_BASIC_STATS);
3252 : : }
3253 : : static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3254 : : {
3255 : 3 : return vfs_statx(AT_FDCWD, name, AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT,
3256 : : stat, STATX_BASIC_STATS);
3257 : : }
3258 : : static inline int vfs_fstatat(int dfd, const char __user *filename,
3259 : : struct kstat *stat, int flags)
3260 : : {
3261 : 3 : return vfs_statx(dfd, filename, flags | AT_NO_AUTOMOUNT,
3262 : : stat, STATX_BASIC_STATS);
3263 : : }
3264 : : static inline int vfs_fstat(int fd, struct kstat *stat)
3265 : : {
3266 : 3 : return vfs_statx_fd(fd, stat, STATX_BASIC_STATS, 0);
3267 : : }
3268 : :
3269 : :
3270 : : extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3271 : : extern int vfs_readlink(struct dentry *, char __user *, int);
3272 : :
3273 : : extern int __generic_block_fiemap(struct inode *inode,
3274 : : struct fiemap_extent_info *fieinfo,
3275 : : loff_t start, loff_t len,
3276 : : get_block_t *get_block);
3277 : : extern int generic_block_fiemap(struct inode *inode,
3278 : : struct fiemap_extent_info *fieinfo, u64 start,
3279 : : u64 len, get_block_t *get_block);
3280 : :
3281 : : extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3282 : : extern void put_filesystem(struct file_system_type *fs);
3283 : : extern struct file_system_type *get_fs_type(const char *name);
3284 : : extern struct super_block *get_super(struct block_device *);
3285 : : extern struct super_block *get_super_thawed(struct block_device *);
3286 : : extern struct super_block *get_super_exclusive_thawed(struct block_device *bdev);
3287 : : extern struct super_block *get_active_super(struct block_device *bdev);
3288 : : extern void drop_super(struct super_block *sb);
3289 : : extern void drop_super_exclusive(struct super_block *sb);
3290 : : extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3291 : : extern void iterate_supers_type(struct file_system_type *,
3292 : : void (*)(struct super_block *, void *), void *);
3293 : :
3294 : : extern int dcache_dir_open(struct inode *, struct file *);
3295 : : extern int dcache_dir_close(struct inode *, struct file *);
3296 : : extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3297 : : extern int dcache_readdir(struct file *, struct dir_context *);
3298 : : extern int simple_setattr(struct dentry *, struct iattr *);
3299 : : extern int simple_getattr(const struct path *, struct kstat *, u32, unsigned int);
3300 : : extern int simple_statfs(struct dentry *, struct kstatfs *);
3301 : : extern int simple_open(struct inode *inode, struct file *file);
3302 : : extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3303 : : extern int simple_unlink(struct inode *, struct dentry *);
3304 : : extern int simple_rmdir(struct inode *, struct dentry *);
3305 : : extern int simple_rename(struct inode *, struct dentry *,
3306 : : struct inode *, struct dentry *, unsigned int);
3307 : : extern int noop_fsync(struct file *, loff_t, loff_t, int);
3308 : : extern int noop_set_page_dirty(struct page *page);
3309 : : extern void noop_invalidatepage(struct page *page, unsigned int offset,
3310 : : unsigned int length);
3311 : : extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3312 : : extern int simple_empty(struct dentry *);
3313 : : extern int simple_readpage(struct file *file, struct page *page);
3314 : : extern int simple_write_begin(struct file *file, struct address_space *mapping,
3315 : : loff_t pos, unsigned len, unsigned flags,
3316 : : struct page **pagep, void **fsdata);
3317 : : extern int simple_write_end(struct file *file, struct address_space *mapping,
3318 : : loff_t pos, unsigned len, unsigned copied,
3319 : : struct page *page, void *fsdata);
3320 : : extern int always_delete_dentry(const struct dentry *);
3321 : : extern struct inode *alloc_anon_inode(struct super_block *);
3322 : : extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3323 : : extern const struct dentry_operations simple_dentry_operations;
3324 : :
3325 : : extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3326 : : extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3327 : : extern const struct file_operations simple_dir_operations;
3328 : : extern const struct inode_operations simple_dir_inode_operations;
3329 : : extern void make_empty_dir_inode(struct inode *inode);
3330 : : extern bool is_empty_dir_inode(struct inode *inode);
3331 : : struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3332 : : struct dentry *d_alloc_name(struct dentry *, const char *);
3333 : : extern int simple_fill_super(struct super_block *, unsigned long,
3334 : : const struct tree_descr *);
3335 : : extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3336 : : extern void simple_release_fs(struct vfsmount **mount, int *count);
3337 : :
3338 : : extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3339 : : loff_t *ppos, const void *from, size_t available);
3340 : : extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3341 : : const void __user *from, size_t count);
3342 : :
3343 : : extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3344 : : extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3345 : :
3346 : : extern int generic_check_addressable(unsigned, u64);
3347 : :
3348 : : #ifdef CONFIG_MIGRATION
3349 : : extern int buffer_migrate_page(struct address_space *,
3350 : : struct page *, struct page *,
3351 : : enum migrate_mode);
3352 : : extern int buffer_migrate_page_norefs(struct address_space *,
3353 : : struct page *, struct page *,
3354 : : enum migrate_mode);
3355 : : #else
3356 : : #define buffer_migrate_page NULL
3357 : : #define buffer_migrate_page_norefs NULL
3358 : : #endif
3359 : :
3360 : : extern int setattr_prepare(struct dentry *, struct iattr *);
3361 : : extern int inode_newsize_ok(const struct inode *, loff_t offset);
3362 : : extern void setattr_copy(struct inode *inode, const struct iattr *attr);
3363 : :
3364 : : extern int file_update_time(struct file *file);
3365 : :
3366 : : static inline bool io_is_direct(struct file *filp)
3367 : : {
3368 : 3 : return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
3369 : : }
3370 : :
3371 : : static inline bool vma_is_dax(struct vm_area_struct *vma)
3372 : : {
3373 : : return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3374 : : }
3375 : :
3376 : : static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3377 : : {
3378 : : struct inode *inode;
3379 : :
3380 : : if (!vma->vm_file)
3381 : : return false;
3382 : : if (!vma_is_dax(vma))
3383 : : return false;
3384 : : inode = file_inode(vma->vm_file);
3385 : : if (S_ISCHR(inode->i_mode))
3386 : : return false; /* device-dax */
3387 : : return true;
3388 : : }
3389 : :
3390 : 3 : static inline int iocb_flags(struct file *file)
3391 : : {
3392 : : int res = 0;
3393 : 3 : if (file->f_flags & O_APPEND)
3394 : : res |= IOCB_APPEND;
3395 : 3 : if (io_is_direct(file))
3396 : 0 : res |= IOCB_DIRECT;
3397 : 3 : if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3398 : 0 : res |= IOCB_DSYNC;
3399 : 3 : if (file->f_flags & __O_SYNC)
3400 : 0 : res |= IOCB_SYNC;
3401 : 3 : return res;
3402 : : }
3403 : :
3404 : 3 : static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3405 : : {
3406 : 3 : if (unlikely(flags & ~RWF_SUPPORTED))
3407 : : return -EOPNOTSUPP;
3408 : :
3409 : 3 : if (flags & RWF_NOWAIT) {
3410 : 0 : if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3411 : : return -EOPNOTSUPP;
3412 : 0 : ki->ki_flags |= IOCB_NOWAIT;
3413 : : }
3414 : 3 : if (flags & RWF_HIPRI)
3415 : 0 : ki->ki_flags |= IOCB_HIPRI;
3416 : 3 : if (flags & RWF_DSYNC)
3417 : 0 : ki->ki_flags |= IOCB_DSYNC;
3418 : 3 : if (flags & RWF_SYNC)
3419 : 0 : ki->ki_flags |= (IOCB_DSYNC | IOCB_SYNC);
3420 : 3 : if (flags & RWF_APPEND)
3421 : 0 : ki->ki_flags |= IOCB_APPEND;
3422 : : return 0;
3423 : : }
3424 : :
3425 : 3 : static inline ino_t parent_ino(struct dentry *dentry)
3426 : : {
3427 : : ino_t res;
3428 : :
3429 : : /*
3430 : : * Don't strictly need d_lock here? If the parent ino could change
3431 : : * then surely we'd have a deeper race in the caller?
3432 : : */
3433 : : spin_lock(&dentry->d_lock);
3434 : 3 : res = dentry->d_parent->d_inode->i_ino;
3435 : : spin_unlock(&dentry->d_lock);
3436 : 3 : return res;
3437 : : }
3438 : :
3439 : : /* Transaction based IO helpers */
3440 : :
3441 : : /*
3442 : : * An argresp is stored in an allocated page and holds the
3443 : : * size of the argument or response, along with its content
3444 : : */
3445 : : struct simple_transaction_argresp {
3446 : : ssize_t size;
3447 : : char data[0];
3448 : : };
3449 : :
3450 : : #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3451 : :
3452 : : char *simple_transaction_get(struct file *file, const char __user *buf,
3453 : : size_t size);
3454 : : ssize_t simple_transaction_read(struct file *file, char __user *buf,
3455 : : size_t size, loff_t *pos);
3456 : : int simple_transaction_release(struct inode *inode, struct file *file);
3457 : :
3458 : : void simple_transaction_set(struct file *file, size_t n);
3459 : :
3460 : : /*
3461 : : * simple attribute files
3462 : : *
3463 : : * These attributes behave similar to those in sysfs:
3464 : : *
3465 : : * Writing to an attribute immediately sets a value, an open file can be
3466 : : * written to multiple times.
3467 : : *
3468 : : * Reading from an attribute creates a buffer from the value that might get
3469 : : * read with multiple read calls. When the attribute has been read
3470 : : * completely, no further read calls are possible until the file is opened
3471 : : * again.
3472 : : *
3473 : : * All attributes contain a text representation of a numeric value
3474 : : * that are accessed with the get() and set() functions.
3475 : : */
3476 : : #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3477 : : static int __fops ## _open(struct inode *inode, struct file *file) \
3478 : : { \
3479 : : __simple_attr_check_format(__fmt, 0ull); \
3480 : : return simple_attr_open(inode, file, __get, __set, __fmt); \
3481 : : } \
3482 : : static const struct file_operations __fops = { \
3483 : : .owner = THIS_MODULE, \
3484 : : .open = __fops ## _open, \
3485 : : .release = simple_attr_release, \
3486 : : .read = simple_attr_read, \
3487 : : .write = simple_attr_write, \
3488 : : .llseek = generic_file_llseek, \
3489 : : }
3490 : :
3491 : : static inline __printf(1, 2)
3492 : : void __simple_attr_check_format(const char *fmt, ...)
3493 : : {
3494 : : /* don't do anything, just let the compiler check the arguments; */
3495 : : }
3496 : :
3497 : : int simple_attr_open(struct inode *inode, struct file *file,
3498 : : int (*get)(void *, u64 *), int (*set)(void *, u64),
3499 : : const char *fmt);
3500 : : int simple_attr_release(struct inode *inode, struct file *file);
3501 : : ssize_t simple_attr_read(struct file *file, char __user *buf,
3502 : : size_t len, loff_t *ppos);
3503 : : ssize_t simple_attr_write(struct file *file, const char __user *buf,
3504 : : size_t len, loff_t *ppos);
3505 : :
3506 : : struct ctl_table;
3507 : : int proc_nr_files(struct ctl_table *table, int write,
3508 : : void __user *buffer, size_t *lenp, loff_t *ppos);
3509 : : int proc_nr_dentry(struct ctl_table *table, int write,
3510 : : void __user *buffer, size_t *lenp, loff_t *ppos);
3511 : : int proc_nr_inodes(struct ctl_table *table, int write,
3512 : : void __user *buffer, size_t *lenp, loff_t *ppos);
3513 : : int __init get_filesystem_list(char *buf);
3514 : :
3515 : : #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3516 : : #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3517 : :
3518 : : #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3519 : : #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3520 : : (flag & __FMODE_NONOTIFY)))
3521 : :
3522 : : static inline bool is_sxid(umode_t mode)
3523 : : {
3524 : 3 : return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3525 : : }
3526 : :
3527 : : static inline int check_sticky(struct inode *dir, struct inode *inode)
3528 : : {
3529 : 3 : if (!(dir->i_mode & S_ISVTX))
3530 : : return 0;
3531 : :
3532 : 3 : return __check_sticky(dir, inode);
3533 : : }
3534 : :
3535 : 3 : static inline void inode_has_no_xattr(struct inode *inode)
3536 : : {
3537 : 3 : if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3538 : 3 : inode->i_flags |= S_NOSEC;
3539 : 3 : }
3540 : :
3541 : : static inline bool is_root_inode(struct inode *inode)
3542 : : {
3543 : : return inode == inode->i_sb->s_root->d_inode;
3544 : : }
3545 : :
3546 : : static inline bool dir_emit(struct dir_context *ctx,
3547 : : const char *name, int namelen,
3548 : : u64 ino, unsigned type)
3549 : : {
3550 : 3 : return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3551 : : }
3552 : 3 : static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3553 : : {
3554 : 3 : return ctx->actor(ctx, ".", 1, ctx->pos,
3555 : 3 : file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3556 : : }
3557 : 3 : static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3558 : : {
3559 : 3 : return ctx->actor(ctx, "..", 2, ctx->pos,
3560 : 3 : parent_ino(file->f_path.dentry), DT_DIR) == 0;
3561 : : }
3562 : 3 : static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3563 : : {
3564 : 3 : if (ctx->pos == 0) {
3565 : 3 : if (!dir_emit_dot(file, ctx))
3566 : : return false;
3567 : 3 : ctx->pos = 1;
3568 : : }
3569 : 3 : if (ctx->pos == 1) {
3570 : 3 : if (!dir_emit_dotdot(file, ctx))
3571 : : return false;
3572 : 3 : ctx->pos = 2;
3573 : : }
3574 : : return true;
3575 : : }
3576 : : static inline bool dir_relax(struct inode *inode)
3577 : : {
3578 : : inode_unlock(inode);
3579 : : inode_lock(inode);
3580 : : return !IS_DEADDIR(inode);
3581 : : }
3582 : :
3583 : 0 : static inline bool dir_relax_shared(struct inode *inode)
3584 : : {
3585 : : inode_unlock_shared(inode);
3586 : : inode_lock_shared(inode);
3587 : 0 : return !IS_DEADDIR(inode);
3588 : : }
3589 : :
3590 : : extern bool path_noexec(const struct path *path);
3591 : : extern void inode_nohighmem(struct inode *inode);
3592 : :
3593 : : /* mm/fadvise.c */
3594 : : extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len,
3595 : : int advice);
3596 : : extern int generic_fadvise(struct file *file, loff_t offset, loff_t len,
3597 : : int advice);
3598 : :
3599 : : #if defined(CONFIG_IO_URING)
3600 : : extern struct sock *io_uring_get_socket(struct file *file);
3601 : : #else
3602 : : static inline struct sock *io_uring_get_socket(struct file *file)
3603 : : {
3604 : : return NULL;
3605 : : }
3606 : : #endif
3607 : :
3608 : : int vfs_ioc_setflags_prepare(struct inode *inode, unsigned int oldflags,
3609 : : unsigned int flags);
3610 : :
3611 : : int vfs_ioc_fssetxattr_check(struct inode *inode, const struct fsxattr *old_fa,
3612 : : struct fsxattr *fa);
3613 : :
3614 : : static inline void simple_fill_fsxattr(struct fsxattr *fa, __u32 xflags)
3615 : : {
3616 : 0 : memset(fa, 0, sizeof(*fa));
3617 : 0 : fa->fsx_xflags = xflags;
3618 : : }
3619 : :
3620 : : /*
3621 : : * Flush file data before changing attributes. Caller must hold any locks
3622 : : * required to prevent further writes to this file until we're done setting
3623 : : * flags.
3624 : : */
3625 : : static inline int inode_drain_writes(struct inode *inode)
3626 : : {
3627 : 3 : inode_dio_wait(inode);
3628 : 3 : return filemap_write_and_wait(inode->i_mapping);
3629 : : }
3630 : :
3631 : : #endif /* _LINUX_FS_H */
|
