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/rculist_bl.h>
28 #include <linux/atomic.h>
29 #include <linux/shrinker.h>
30 #include <linux/migrate_mode.h>
31 #include <linux/uidgid.h>
32 #include <linux/lockdep.h>
33 #include <linux/percpu-rwsem.h>
34 #include <linux/workqueue.h>
35 #include <linux/delayed_call.h>
36 #include <linux/uuid.h>
37 #include <linux/errseq.h>
38 #include <linux/ioprio.h>
39 #include <linux/fs_types.h>
40 #include <linux/build_bug.h>
41 #include <linux/stddef.h>
42 #include <linux/mount.h>
43 #include <linux/cred.h>
44
45 #include <asm/byteorder.h>
46 #include <uapi/linux/fs.h>
47
48 struct backing_dev_info;
49 struct bdi_writeback;
50 struct bio;
51 struct io_comp_batch;
52 struct export_operations;
53 struct fiemap_extent_info;
54 struct hd_geometry;
55 struct iovec;
56 struct kiocb;
57 struct kobject;
58 struct pipe_inode_info;
59 struct poll_table_struct;
60 struct kstatfs;
61 struct vm_area_struct;
62 struct vfsmount;
63 struct cred;
64 struct swap_info_struct;
65 struct seq_file;
66 struct workqueue_struct;
67 struct iov_iter;
68 struct fscrypt_info;
69 struct fscrypt_operations;
70 struct fsverity_info;
71 struct fsverity_operations;
72 struct fs_context;
73 struct fs_parameter_spec;
74 struct fileattr;
75
76 extern void __init inode_init(void);
77 extern void __init inode_init_early(void);
78 extern void __init files_init(void);
79 extern void __init files_maxfiles_init(void);
80
81 extern struct files_stat_struct files_stat;
82 extern unsigned long get_max_files(void);
83 extern unsigned int sysctl_nr_open;
84 extern struct inodes_stat_t inodes_stat;
85 extern int leases_enable, lease_break_time;
86 extern int sysctl_protected_symlinks;
87 extern int sysctl_protected_hardlinks;
88 extern int sysctl_protected_fifos;
89 extern int sysctl_protected_regular;
90
91 typedef __kernel_rwf_t rwf_t;
92
93 struct buffer_head;
94 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
95 struct buffer_head *bh_result, int create);
96 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
97 ssize_t bytes, void *private);
98
99 #define MAY_EXEC 0x00000001
100 #define MAY_WRITE 0x00000002
101 #define MAY_READ 0x00000004
102 #define MAY_APPEND 0x00000008
103 #define MAY_ACCESS 0x00000010
104 #define MAY_OPEN 0x00000020
105 #define MAY_CHDIR 0x00000040
106 /* called from RCU mode, don't block */
107 #define MAY_NOT_BLOCK 0x00000080
108
109 /*
110 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
111 * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
112 */
113
114 /* file is open for reading */
115 #define FMODE_READ ((__force fmode_t)0x1)
116 /* file is open for writing */
117 #define FMODE_WRITE ((__force fmode_t)0x2)
118 /* file is seekable */
119 #define FMODE_LSEEK ((__force fmode_t)0x4)
120 /* file can be accessed using pread */
121 #define FMODE_PREAD ((__force fmode_t)0x8)
122 /* file can be accessed using pwrite */
123 #define FMODE_PWRITE ((__force fmode_t)0x10)
124 /* File is opened for execution with sys_execve / sys_uselib */
125 #define FMODE_EXEC ((__force fmode_t)0x20)
126 /* File is opened with O_NDELAY (only set for block devices) */
127 #define FMODE_NDELAY ((__force fmode_t)0x40)
128 /* File is opened with O_EXCL (only set for block devices) */
129 #define FMODE_EXCL ((__force fmode_t)0x80)
130 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
131 (specialy hack for floppy.c) */
132 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
133 /* 32bit hashes as llseek() offset (for directories) */
134 #define FMODE_32BITHASH ((__force fmode_t)0x200)
135 /* 64bit hashes as llseek() offset (for directories) */
136 #define FMODE_64BITHASH ((__force fmode_t)0x400)
137
138 /*
139 * Don't update ctime and mtime.
140 *
141 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
142 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
143 */
144 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
145
146 /* Expect random access pattern */
147 #define FMODE_RANDOM ((__force fmode_t)0x1000)
148
149 /* File is huge (eg. /dev/mem): treat loff_t as unsigned */
150 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
151
152 /* File is opened with O_PATH; almost nothing can be done with it */
153 #define FMODE_PATH ((__force fmode_t)0x4000)
154
155 /* File needs atomic accesses to f_pos */
156 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
157 /* Write access to underlying fs */
158 #define FMODE_WRITER ((__force fmode_t)0x10000)
159 /* Has read method(s) */
160 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
161 /* Has write method(s) */
162 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
163
164 #define FMODE_OPENED ((__force fmode_t)0x80000)
165 #define FMODE_CREATED ((__force fmode_t)0x100000)
166
167 /* File is stream-like */
168 #define FMODE_STREAM ((__force fmode_t)0x200000)
169
170 /* File was opened by fanotify and shouldn't generate fanotify events */
171 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
172
173 /* File is capable of returning -EAGAIN if I/O will block */
174 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
175
176 /* File represents mount that needs unmounting */
177 #define FMODE_NEED_UNMOUNT ((__force fmode_t)0x10000000)
178
179 /* File does not contribute to nr_files count */
180 #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000)
181
182 /* File supports async buffered reads */
183 #define FMODE_BUF_RASYNC ((__force fmode_t)0x40000000)
184
185 /*
186 * Attribute flags. These should be or-ed together to figure out what
187 * has been changed!
188 */
189 #define ATTR_MODE (1 << 0)
190 #define ATTR_UID (1 << 1)
191 #define ATTR_GID (1 << 2)
192 #define ATTR_SIZE (1 << 3)
193 #define ATTR_ATIME (1 << 4)
194 #define ATTR_MTIME (1 << 5)
195 #define ATTR_CTIME (1 << 6)
196 #define ATTR_ATIME_SET (1 << 7)
197 #define ATTR_MTIME_SET (1 << 8)
198 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
199 #define ATTR_KILL_SUID (1 << 11)
200 #define ATTR_KILL_SGID (1 << 12)
201 #define ATTR_FILE (1 << 13)
202 #define ATTR_KILL_PRIV (1 << 14)
203 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
204 #define ATTR_TIMES_SET (1 << 16)
205 #define ATTR_TOUCH (1 << 17)
206
207 /*
208 * Whiteout is represented by a char device. The following constants define the
209 * mode and device number to use.
210 */
211 #define WHITEOUT_MODE 0
212 #define WHITEOUT_DEV 0
213
214 /*
215 * This is the Inode Attributes structure, used for notify_change(). It
216 * uses the above definitions as flags, to know which values have changed.
217 * Also, in this manner, a Filesystem can look at only the values it cares
218 * about. Basically, these are the attributes that the VFS layer can
219 * request to change from the FS layer.
220 *
221 * Derek Atkins <warlord@MIT.EDU> 94-10-20
222 */
223 struct iattr {
224 unsigned int ia_valid;
225 umode_t ia_mode;
226 kuid_t ia_uid;
227 kgid_t ia_gid;
228 loff_t ia_size;
229 struct timespec64 ia_atime;
230 struct timespec64 ia_mtime;
231 struct timespec64 ia_ctime;
232
233 /*
234 * Not an attribute, but an auxiliary info for filesystems wanting to
235 * implement an ftruncate() like method. NOTE: filesystem should
236 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
237 */
238 struct file *ia_file;
239 };
240
241 /*
242 * Includes for diskquotas.
243 */
244 #include <linux/quota.h>
245
246 /*
247 * Maximum number of layers of fs stack. Needs to be limited to
248 * prevent kernel stack overflow
249 */
250 #define FILESYSTEM_MAX_STACK_DEPTH 2
251
252 /**
253 * enum positive_aop_returns - aop return codes with specific semantics
254 *
255 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
256 * completed, that the page is still locked, and
257 * should be considered active. The VM uses this hint
258 * to return the page to the active list -- it won't
259 * be a candidate for writeback again in the near
260 * future. Other callers must be careful to unlock
261 * the page if they get this return. Returned by
262 * writepage();
263 *
264 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
265 * unlocked it and the page might have been truncated.
266 * The caller should back up to acquiring a new page and
267 * trying again. The aop will be taking reasonable
268 * precautions not to livelock. If the caller held a page
269 * reference, it should drop it before retrying. Returned
270 * by readpage().
271 *
272 * address_space_operation functions return these large constants to indicate
273 * special semantics to the caller. These are much larger than the bytes in a
274 * page to allow for functions that return the number of bytes operated on in a
275 * given page.
276 */
277
278 enum positive_aop_returns {
279 AOP_WRITEPAGE_ACTIVATE = 0x80000,
280 AOP_TRUNCATED_PAGE = 0x80001,
281 };
282
283 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
284 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
285 * helper code (eg buffer layer)
286 * to clear GFP_FS from alloc */
287
288 /*
289 * oh the beauties of C type declarations.
290 */
291 struct page;
292 struct address_space;
293 struct writeback_control;
294 struct readahead_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 /* Match RWF_* bits to IOCB bits */
310 #define IOCB_HIPRI (__force int) RWF_HIPRI
311 #define IOCB_DSYNC (__force int) RWF_DSYNC
312 #define IOCB_SYNC (__force int) RWF_SYNC
313 #define IOCB_NOWAIT (__force int) RWF_NOWAIT
314 #define IOCB_APPEND (__force int) RWF_APPEND
315
316 /* non-RWF related bits - start at 16 */
317 #define IOCB_EVENTFD (1 << 16)
318 #define IOCB_DIRECT (1 << 17)
319 #define IOCB_WRITE (1 << 18)
320 /* iocb->ki_waitq is valid */
321 #define IOCB_WAITQ (1 << 19)
322 #define IOCB_NOIO (1 << 20)
323 /* can use bio alloc cache */
324 #define IOCB_ALLOC_CACHE (1 << 21)
325
326 struct kiocb {
327 struct file *ki_filp;
328
329 /* The 'ki_filp' pointer is shared in a union for aio */
330 randomized_struct_fields_start
331
332 loff_t ki_pos;
333 void (*ki_complete)(struct kiocb *iocb, long ret);
334 void *private;
335 int ki_flags;
336 u16 ki_hint;
337 u16 ki_ioprio; /* See linux/ioprio.h */
338 struct wait_page_queue *ki_waitq; /* for async buffered IO */
339 randomized_struct_fields_end
340 };
341
is_sync_kiocb(struct kiocb * kiocb)342 static inline bool is_sync_kiocb(struct kiocb *kiocb)
343 {
344 return kiocb->ki_complete == NULL;
345 }
346
347 /*
348 * "descriptor" for what we're up to with a read.
349 * This allows us to use the same read code yet
350 * have multiple different users of the data that
351 * we read from a file.
352 *
353 * The simplest case just copies the data to user
354 * mode.
355 */
356 typedef struct {
357 size_t written;
358 size_t count;
359 union {
360 char __user *buf;
361 void *data;
362 } arg;
363 int error;
364 } read_descriptor_t;
365
366 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
367 unsigned long, unsigned long);
368
369 struct address_space_operations {
370 int (*writepage)(struct page *page, struct writeback_control *wbc);
371 int (*readpage)(struct file *, struct page *);
372
373 /* Write back some dirty pages from this mapping. */
374 int (*writepages)(struct address_space *, struct writeback_control *);
375
376 /* Set a page dirty. Return true if this dirtied it */
377 int (*set_page_dirty)(struct page *page);
378
379 /*
380 * Reads in the requested pages. Unlike ->readpage(), this is
381 * PURELY used for read-ahead!.
382 */
383 int (*readpages)(struct file *filp, struct address_space *mapping,
384 struct list_head *pages, unsigned nr_pages);
385 void (*readahead)(struct readahead_control *);
386
387 int (*write_begin)(struct file *, struct address_space *mapping,
388 loff_t pos, unsigned len, unsigned flags,
389 struct page **pagep, void **fsdata);
390 int (*write_end)(struct file *, struct address_space *mapping,
391 loff_t pos, unsigned len, unsigned copied,
392 struct page *page, void *fsdata);
393
394 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
395 sector_t (*bmap)(struct address_space *, sector_t);
396 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
397 int (*releasepage) (struct page *, gfp_t);
398 void (*freepage)(struct page *);
399 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
400 /*
401 * migrate the contents of a page to the specified target. If
402 * migrate_mode is MIGRATE_ASYNC, it must not block.
403 */
404 int (*migratepage) (struct address_space *,
405 struct page *, struct page *, enum migrate_mode);
406 bool (*isolate_page)(struct page *, isolate_mode_t);
407 void (*putback_page)(struct page *);
408 int (*launder_page) (struct page *);
409 int (*is_partially_uptodate) (struct page *, unsigned long,
410 unsigned long);
411 void (*is_dirty_writeback) (struct page *, bool *, bool *);
412 int (*error_remove_page)(struct address_space *, struct page *);
413
414 /* swapfile support */
415 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
416 sector_t *span);
417 void (*swap_deactivate)(struct file *file);
418 };
419
420 extern const struct address_space_operations empty_aops;
421
422 /*
423 * pagecache_write_begin/pagecache_write_end must be used by general code
424 * to write into the pagecache.
425 */
426 int pagecache_write_begin(struct file *, struct address_space *mapping,
427 loff_t pos, unsigned len, unsigned flags,
428 struct page **pagep, void **fsdata);
429
430 int pagecache_write_end(struct file *, struct address_space *mapping,
431 loff_t pos, unsigned len, unsigned copied,
432 struct page *page, void *fsdata);
433
434 /**
435 * struct address_space - Contents of a cacheable, mappable object.
436 * @host: Owner, either the inode or the block_device.
437 * @i_pages: Cached pages.
438 * @invalidate_lock: Guards coherency between page cache contents and
439 * file offset->disk block mappings in the filesystem during invalidates.
440 * It is also used to block modification of page cache contents through
441 * memory mappings.
442 * @gfp_mask: Memory allocation flags to use for allocating pages.
443 * @i_mmap_writable: Number of VM_SHARED mappings.
444 * @nr_thps: Number of THPs in the pagecache (non-shmem only).
445 * @i_mmap: Tree of private and shared mappings.
446 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
447 * @nrpages: Number of page entries, protected by the i_pages lock.
448 * @writeback_index: Writeback starts here.
449 * @a_ops: Methods.
450 * @flags: Error bits and flags (AS_*).
451 * @wb_err: The most recent error which has occurred.
452 * @private_lock: For use by the owner of the address_space.
453 * @private_list: For use by the owner of the address_space.
454 * @private_data: For use by the owner of the address_space.
455 */
456 struct address_space {
457 struct inode *host;
458 struct xarray i_pages;
459 struct rw_semaphore invalidate_lock;
460 gfp_t gfp_mask;
461 atomic_t i_mmap_writable;
462 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
463 /* number of thp, only for non-shmem files */
464 atomic_t nr_thps;
465 #endif
466 struct rb_root_cached i_mmap;
467 struct rw_semaphore i_mmap_rwsem;
468 unsigned long nrpages;
469 pgoff_t writeback_index;
470 const struct address_space_operations *a_ops;
471 unsigned long flags;
472 errseq_t wb_err;
473 spinlock_t private_lock;
474 struct list_head private_list;
475 void *private_data;
476 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
477 /*
478 * On most architectures that alignment is already the case; but
479 * must be enforced here for CRIS, to let the least significant bit
480 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
481 */
482
483 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
484 #define PAGECACHE_TAG_DIRTY XA_MARK_0
485 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
486 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
487
488 /*
489 * Returns true if any of the pages in the mapping are marked with the tag.
490 */
mapping_tagged(struct address_space * mapping,xa_mark_t tag)491 static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag)
492 {
493 return xa_marked(&mapping->i_pages, tag);
494 }
495
i_mmap_lock_write(struct address_space * mapping)496 static inline void i_mmap_lock_write(struct address_space *mapping)
497 {
498 down_write(&mapping->i_mmap_rwsem);
499 }
500
i_mmap_trylock_write(struct address_space * mapping)501 static inline int i_mmap_trylock_write(struct address_space *mapping)
502 {
503 return down_write_trylock(&mapping->i_mmap_rwsem);
504 }
505
i_mmap_unlock_write(struct address_space * mapping)506 static inline void i_mmap_unlock_write(struct address_space *mapping)
507 {
508 up_write(&mapping->i_mmap_rwsem);
509 }
510
i_mmap_lock_read(struct address_space * mapping)511 static inline void i_mmap_lock_read(struct address_space *mapping)
512 {
513 down_read(&mapping->i_mmap_rwsem);
514 }
515
i_mmap_unlock_read(struct address_space * mapping)516 static inline void i_mmap_unlock_read(struct address_space *mapping)
517 {
518 up_read(&mapping->i_mmap_rwsem);
519 }
520
i_mmap_assert_locked(struct address_space * mapping)521 static inline void i_mmap_assert_locked(struct address_space *mapping)
522 {
523 lockdep_assert_held(&mapping->i_mmap_rwsem);
524 }
525
i_mmap_assert_write_locked(struct address_space * mapping)526 static inline void i_mmap_assert_write_locked(struct address_space *mapping)
527 {
528 lockdep_assert_held_write(&mapping->i_mmap_rwsem);
529 }
530
531 /*
532 * Might pages of this file be mapped into userspace?
533 */
mapping_mapped(struct address_space * mapping)534 static inline int mapping_mapped(struct address_space *mapping)
535 {
536 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
537 }
538
539 /*
540 * Might pages of this file have been modified in userspace?
541 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap
542 * marks vma as VM_SHARED if it is shared, and the file was opened for
543 * writing i.e. vma may be mprotected writable even if now readonly.
544 *
545 * If i_mmap_writable is negative, no new writable mappings are allowed. You
546 * can only deny writable mappings, if none exists right now.
547 */
mapping_writably_mapped(struct address_space * mapping)548 static inline int mapping_writably_mapped(struct address_space *mapping)
549 {
550 return atomic_read(&mapping->i_mmap_writable) > 0;
551 }
552
mapping_map_writable(struct address_space * mapping)553 static inline int mapping_map_writable(struct address_space *mapping)
554 {
555 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
556 0 : -EPERM;
557 }
558
mapping_unmap_writable(struct address_space * mapping)559 static inline void mapping_unmap_writable(struct address_space *mapping)
560 {
561 atomic_dec(&mapping->i_mmap_writable);
562 }
563
mapping_deny_writable(struct address_space * mapping)564 static inline int mapping_deny_writable(struct address_space *mapping)
565 {
566 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
567 0 : -EBUSY;
568 }
569
mapping_allow_writable(struct address_space * mapping)570 static inline void mapping_allow_writable(struct address_space *mapping)
571 {
572 atomic_inc(&mapping->i_mmap_writable);
573 }
574
575 /*
576 * Use sequence counter to get consistent i_size on 32-bit processors.
577 */
578 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
579 #include <linux/seqlock.h>
580 #define __NEED_I_SIZE_ORDERED
581 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
582 #else
583 #define i_size_ordered_init(inode) do { } while (0)
584 #endif
585
586 struct posix_acl;
587 #define ACL_NOT_CACHED ((void *)(-1))
588 /*
589 * ACL_DONT_CACHE is for stacked filesystems, that rely on underlying fs to
590 * cache the ACL. This also means that ->get_acl() can be called in RCU mode
591 * with the LOOKUP_RCU flag.
592 */
593 #define ACL_DONT_CACHE ((void *)(-3))
594
595 static inline struct posix_acl *
uncached_acl_sentinel(struct task_struct * task)596 uncached_acl_sentinel(struct task_struct *task)
597 {
598 return (void *)task + 1;
599 }
600
601 static inline bool
is_uncached_acl(struct posix_acl * acl)602 is_uncached_acl(struct posix_acl *acl)
603 {
604 return (long)acl & 1;
605 }
606
607 #define IOP_FASTPERM 0x0001
608 #define IOP_LOOKUP 0x0002
609 #define IOP_NOFOLLOW 0x0004
610 #define IOP_XATTR 0x0008
611 #define IOP_DEFAULT_READLINK 0x0010
612
613 struct fsnotify_mark_connector;
614
615 /*
616 * Keep mostly read-only and often accessed (especially for
617 * the RCU path lookup and 'stat' data) fields at the beginning
618 * of the 'struct inode'
619 */
620 struct inode {
621 umode_t i_mode;
622 unsigned short i_opflags;
623 kuid_t i_uid;
624 kgid_t i_gid;
625 unsigned int i_flags;
626
627 #ifdef CONFIG_FS_POSIX_ACL
628 struct posix_acl *i_acl;
629 struct posix_acl *i_default_acl;
630 #endif
631
632 const struct inode_operations *i_op;
633 struct super_block *i_sb;
634 struct address_space *i_mapping;
635
636 #ifdef CONFIG_SECURITY
637 void *i_security;
638 #endif
639
640 /* Stat data, not accessed from path walking */
641 unsigned long i_ino;
642 /*
643 * Filesystems may only read i_nlink directly. They shall use the
644 * following functions for modification:
645 *
646 * (set|clear|inc|drop)_nlink
647 * inode_(inc|dec)_link_count
648 */
649 union {
650 const unsigned int i_nlink;
651 unsigned int __i_nlink;
652 };
653 dev_t i_rdev;
654 loff_t i_size;
655 struct timespec64 i_atime;
656 struct timespec64 i_mtime;
657 struct timespec64 i_ctime;
658 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
659 unsigned short i_bytes;
660 u8 i_blkbits;
661 u8 i_write_hint;
662 blkcnt_t i_blocks;
663
664 #ifdef __NEED_I_SIZE_ORDERED
665 seqcount_t i_size_seqcount;
666 #endif
667
668 /* Misc */
669 unsigned long i_state;
670 struct rw_semaphore i_rwsem;
671
672 unsigned long dirtied_when; /* jiffies of first dirtying */
673 unsigned long dirtied_time_when;
674
675 struct hlist_node i_hash;
676 struct list_head i_io_list; /* backing dev IO list */
677 #ifdef CONFIG_CGROUP_WRITEBACK
678 struct bdi_writeback *i_wb; /* the associated cgroup wb */
679
680 /* foreign inode detection, see wbc_detach_inode() */
681 int i_wb_frn_winner;
682 u16 i_wb_frn_avg_time;
683 u16 i_wb_frn_history;
684 #endif
685 struct list_head i_lru; /* inode LRU list */
686 struct list_head i_sb_list;
687 struct list_head i_wb_list; /* backing dev writeback list */
688 union {
689 struct hlist_head i_dentry;
690 struct rcu_head i_rcu;
691 };
692 atomic64_t i_version;
693 atomic64_t i_sequence; /* see futex */
694 atomic_t i_count;
695 atomic_t i_dio_count;
696 atomic_t i_writecount;
697 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
698 atomic_t i_readcount; /* struct files open RO */
699 #endif
700 union {
701 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
702 void (*free_inode)(struct inode *);
703 };
704 struct file_lock_context *i_flctx;
705 struct address_space i_data;
706 struct list_head i_devices;
707 union {
708 struct pipe_inode_info *i_pipe;
709 struct cdev *i_cdev;
710 char *i_link;
711 unsigned i_dir_seq;
712 };
713
714 __u32 i_generation;
715
716 #ifdef CONFIG_FSNOTIFY
717 __u32 i_fsnotify_mask; /* all events this inode cares about */
718 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
719 #endif
720
721 #ifdef CONFIG_FS_ENCRYPTION
722 struct fscrypt_info *i_crypt_info;
723 #endif
724
725 #ifdef CONFIG_FS_VERITY
726 struct fsverity_info *i_verity_info;
727 #endif
728
729 void *i_private; /* fs or device private pointer */
730 } __randomize_layout;
731
732 struct timespec64 timestamp_truncate(struct timespec64 t, struct inode *inode);
733
i_blocksize(const struct inode * node)734 static inline unsigned int i_blocksize(const struct inode *node)
735 {
736 return (1 << node->i_blkbits);
737 }
738
inode_unhashed(struct inode * inode)739 static inline int inode_unhashed(struct inode *inode)
740 {
741 return hlist_unhashed(&inode->i_hash);
742 }
743
744 /*
745 * __mark_inode_dirty expects inodes to be hashed. Since we don't
746 * want special inodes in the fileset inode space, we make them
747 * appear hashed, but do not put on any lists. hlist_del()
748 * will work fine and require no locking.
749 */
inode_fake_hash(struct inode * inode)750 static inline void inode_fake_hash(struct inode *inode)
751 {
752 hlist_add_fake(&inode->i_hash);
753 }
754
755 /*
756 * inode->i_mutex nesting subclasses for the lock validator:
757 *
758 * 0: the object of the current VFS operation
759 * 1: parent
760 * 2: child/target
761 * 3: xattr
762 * 4: second non-directory
763 * 5: second parent (when locking independent directories in rename)
764 *
765 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
766 * non-directories at once.
767 *
768 * The locking order between these classes is
769 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
770 */
771 enum inode_i_mutex_lock_class
772 {
773 I_MUTEX_NORMAL,
774 I_MUTEX_PARENT,
775 I_MUTEX_CHILD,
776 I_MUTEX_XATTR,
777 I_MUTEX_NONDIR2,
778 I_MUTEX_PARENT2,
779 };
780
inode_lock(struct inode * inode)781 static inline void inode_lock(struct inode *inode)
782 {
783 down_write(&inode->i_rwsem);
784 }
785
inode_unlock(struct inode * inode)786 static inline void inode_unlock(struct inode *inode)
787 {
788 up_write(&inode->i_rwsem);
789 }
790
inode_lock_shared(struct inode * inode)791 static inline void inode_lock_shared(struct inode *inode)
792 {
793 down_read(&inode->i_rwsem);
794 }
795
inode_unlock_shared(struct inode * inode)796 static inline void inode_unlock_shared(struct inode *inode)
797 {
798 up_read(&inode->i_rwsem);
799 }
800
inode_trylock(struct inode * inode)801 static inline int inode_trylock(struct inode *inode)
802 {
803 return down_write_trylock(&inode->i_rwsem);
804 }
805
inode_trylock_shared(struct inode * inode)806 static inline int inode_trylock_shared(struct inode *inode)
807 {
808 return down_read_trylock(&inode->i_rwsem);
809 }
810
inode_is_locked(struct inode * inode)811 static inline int inode_is_locked(struct inode *inode)
812 {
813 return rwsem_is_locked(&inode->i_rwsem);
814 }
815
inode_lock_nested(struct inode * inode,unsigned subclass)816 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
817 {
818 down_write_nested(&inode->i_rwsem, subclass);
819 }
820
inode_lock_shared_nested(struct inode * inode,unsigned subclass)821 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
822 {
823 down_read_nested(&inode->i_rwsem, subclass);
824 }
825
filemap_invalidate_lock(struct address_space * mapping)826 static inline void filemap_invalidate_lock(struct address_space *mapping)
827 {
828 down_write(&mapping->invalidate_lock);
829 }
830
filemap_invalidate_unlock(struct address_space * mapping)831 static inline void filemap_invalidate_unlock(struct address_space *mapping)
832 {
833 up_write(&mapping->invalidate_lock);
834 }
835
filemap_invalidate_lock_shared(struct address_space * mapping)836 static inline void filemap_invalidate_lock_shared(struct address_space *mapping)
837 {
838 down_read(&mapping->invalidate_lock);
839 }
840
filemap_invalidate_trylock_shared(struct address_space * mapping)841 static inline int filemap_invalidate_trylock_shared(
842 struct address_space *mapping)
843 {
844 return down_read_trylock(&mapping->invalidate_lock);
845 }
846
filemap_invalidate_unlock_shared(struct address_space * mapping)847 static inline void filemap_invalidate_unlock_shared(
848 struct address_space *mapping)
849 {
850 up_read(&mapping->invalidate_lock);
851 }
852
853 void lock_two_nondirectories(struct inode *, struct inode*);
854 void unlock_two_nondirectories(struct inode *, struct inode*);
855
856 void filemap_invalidate_lock_two(struct address_space *mapping1,
857 struct address_space *mapping2);
858 void filemap_invalidate_unlock_two(struct address_space *mapping1,
859 struct address_space *mapping2);
860
861
862 /*
863 * NOTE: in a 32bit arch with a preemptable kernel and
864 * an UP compile the i_size_read/write must be atomic
865 * with respect to the local cpu (unlike with preempt disabled),
866 * but they don't need to be atomic with respect to other cpus like in
867 * true SMP (so they need either to either locally disable irq around
868 * the read or for example on x86 they can be still implemented as a
869 * cmpxchg8b without the need of the lock prefix). For SMP compiles
870 * and 64bit archs it makes no difference if preempt is enabled or not.
871 */
i_size_read(const struct inode * inode)872 static inline loff_t i_size_read(const struct inode *inode)
873 {
874 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
875 loff_t i_size;
876 unsigned int seq;
877
878 do {
879 seq = read_seqcount_begin(&inode->i_size_seqcount);
880 i_size = inode->i_size;
881 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
882 return i_size;
883 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
884 loff_t i_size;
885
886 preempt_disable();
887 i_size = inode->i_size;
888 preempt_enable();
889 return i_size;
890 #else
891 return inode->i_size;
892 #endif
893 }
894
895 /*
896 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
897 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
898 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
899 */
i_size_write(struct inode * inode,loff_t i_size)900 static inline void i_size_write(struct inode *inode, loff_t i_size)
901 {
902 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
903 preempt_disable();
904 write_seqcount_begin(&inode->i_size_seqcount);
905 inode->i_size = i_size;
906 write_seqcount_end(&inode->i_size_seqcount);
907 preempt_enable();
908 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
909 preempt_disable();
910 inode->i_size = i_size;
911 preempt_enable();
912 #else
913 inode->i_size = i_size;
914 #endif
915 }
916
iminor(const struct inode * inode)917 static inline unsigned iminor(const struct inode *inode)
918 {
919 return MINOR(inode->i_rdev);
920 }
921
imajor(const struct inode * inode)922 static inline unsigned imajor(const struct inode *inode)
923 {
924 return MAJOR(inode->i_rdev);
925 }
926
927 struct fown_struct {
928 rwlock_t lock; /* protects pid, uid, euid fields */
929 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
930 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
931 kuid_t uid, euid; /* uid/euid of process setting the owner */
932 int signum; /* posix.1b rt signal to be delivered on IO */
933 };
934
935 /**
936 * struct file_ra_state - Track a file's readahead state.
937 * @start: Where the most recent readahead started.
938 * @size: Number of pages read in the most recent readahead.
939 * @async_size: Start next readahead when this many pages are left.
940 * @ra_pages: Maximum size of a readahead request.
941 * @mmap_miss: How many mmap accesses missed in the page cache.
942 * @prev_pos: The last byte in the most recent read request.
943 */
944 struct file_ra_state {
945 pgoff_t start;
946 unsigned int size;
947 unsigned int async_size;
948 unsigned int ra_pages;
949 unsigned int mmap_miss;
950 loff_t prev_pos;
951 };
952
953 /*
954 * Check if @index falls in the readahead windows.
955 */
ra_has_index(struct file_ra_state * ra,pgoff_t index)956 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
957 {
958 return (index >= ra->start &&
959 index < ra->start + ra->size);
960 }
961
962 struct file {
963 union {
964 struct llist_node fu_llist;
965 struct rcu_head fu_rcuhead;
966 } f_u;
967 struct path f_path;
968 struct inode *f_inode; /* cached value */
969 const struct file_operations *f_op;
970
971 /*
972 * Protects f_ep, f_flags.
973 * Must not be taken from IRQ context.
974 */
975 spinlock_t f_lock;
976 enum rw_hint f_write_hint;
977 atomic_long_t f_count;
978 unsigned int f_flags;
979 fmode_t f_mode;
980 struct mutex f_pos_lock;
981 loff_t f_pos;
982 struct fown_struct f_owner;
983 const struct cred *f_cred;
984 struct file_ra_state f_ra;
985
986 u64 f_version;
987 #ifdef CONFIG_SECURITY
988 void *f_security;
989 #endif
990 /* needed for tty driver, and maybe others */
991 void *private_data;
992
993 #ifdef CONFIG_EPOLL
994 /* Used by fs/eventpoll.c to link all the hooks to this file */
995 struct hlist_head *f_ep;
996 #endif /* #ifdef CONFIG_EPOLL */
997 struct address_space *f_mapping;
998 errseq_t f_wb_err;
999 errseq_t f_sb_err; /* for syncfs */
1000 } __randomize_layout
1001 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
1002
1003 struct file_handle {
1004 __u32 handle_bytes;
1005 int handle_type;
1006 /* file identifier */
1007 unsigned char f_handle[];
1008 };
1009
get_file(struct file * f)1010 static inline struct file *get_file(struct file *f)
1011 {
1012 atomic_long_inc(&f->f_count);
1013 return f;
1014 }
1015 #define get_file_rcu_many(x, cnt) \
1016 atomic_long_add_unless(&(x)->f_count, (cnt), 0)
1017 #define get_file_rcu(x) get_file_rcu_many((x), 1)
1018 #define file_count(x) atomic_long_read(&(x)->f_count)
1019
1020 #define MAX_NON_LFS ((1UL<<31) - 1)
1021
1022 /* Page cache limit. The filesystems should put that into their s_maxbytes
1023 limits, otherwise bad things can happen in VM. */
1024 #if BITS_PER_LONG==32
1025 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
1026 #elif BITS_PER_LONG==64
1027 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
1028 #endif
1029
1030 #define FL_POSIX 1
1031 #define FL_FLOCK 2
1032 #define FL_DELEG 4 /* NFSv4 delegation */
1033 #define FL_ACCESS 8 /* not trying to lock, just looking */
1034 #define FL_EXISTS 16 /* when unlocking, test for existence */
1035 #define FL_LEASE 32 /* lease held on this file */
1036 #define FL_CLOSE 64 /* unlock on close */
1037 #define FL_SLEEP 128 /* A blocking lock */
1038 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
1039 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
1040 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
1041 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
1042 #define FL_RECLAIM 4096 /* reclaiming from a reboot server */
1043
1044 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
1045
1046 /*
1047 * Special return value from posix_lock_file() and vfs_lock_file() for
1048 * asynchronous locking.
1049 */
1050 #define FILE_LOCK_DEFERRED 1
1051
1052 /* legacy typedef, should eventually be removed */
1053 typedef void *fl_owner_t;
1054
1055 struct file_lock;
1056
1057 struct file_lock_operations {
1058 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1059 void (*fl_release_private)(struct file_lock *);
1060 };
1061
1062 struct lock_manager_operations {
1063 fl_owner_t (*lm_get_owner)(fl_owner_t);
1064 void (*lm_put_owner)(fl_owner_t);
1065 void (*lm_notify)(struct file_lock *); /* unblock callback */
1066 int (*lm_grant)(struct file_lock *, int);
1067 bool (*lm_break)(struct file_lock *);
1068 int (*lm_change)(struct file_lock *, int, struct list_head *);
1069 void (*lm_setup)(struct file_lock *, void **);
1070 bool (*lm_breaker_owns_lease)(struct file_lock *);
1071 };
1072
1073 struct lock_manager {
1074 struct list_head list;
1075 /*
1076 * NFSv4 and up also want opens blocked during the grace period;
1077 * NLM doesn't care:
1078 */
1079 bool block_opens;
1080 };
1081
1082 struct net;
1083 void locks_start_grace(struct net *, struct lock_manager *);
1084 void locks_end_grace(struct lock_manager *);
1085 bool locks_in_grace(struct net *);
1086 bool opens_in_grace(struct net *);
1087
1088 /* that will die - we need it for nfs_lock_info */
1089 #include <linux/nfs_fs_i.h>
1090
1091 /*
1092 * struct file_lock represents a generic "file lock". It's used to represent
1093 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1094 * note that the same struct is used to represent both a request for a lock and
1095 * the lock itself, but the same object is never used for both.
1096 *
1097 * FIXME: should we create a separate "struct lock_request" to help distinguish
1098 * these two uses?
1099 *
1100 * The varous i_flctx lists are ordered by:
1101 *
1102 * 1) lock owner
1103 * 2) lock range start
1104 * 3) lock range end
1105 *
1106 * Obviously, the last two criteria only matter for POSIX locks.
1107 */
1108 struct file_lock {
1109 struct file_lock *fl_blocker; /* The lock, that is blocking us */
1110 struct list_head fl_list; /* link into file_lock_context */
1111 struct hlist_node fl_link; /* node in global lists */
1112 struct list_head fl_blocked_requests; /* list of requests with
1113 * ->fl_blocker pointing here
1114 */
1115 struct list_head fl_blocked_member; /* node in
1116 * ->fl_blocker->fl_blocked_requests
1117 */
1118 fl_owner_t fl_owner;
1119 unsigned int fl_flags;
1120 unsigned char fl_type;
1121 unsigned int fl_pid;
1122 int fl_link_cpu; /* what cpu's list is this on? */
1123 wait_queue_head_t fl_wait;
1124 struct file *fl_file;
1125 loff_t fl_start;
1126 loff_t fl_end;
1127
1128 struct fasync_struct * fl_fasync; /* for lease break notifications */
1129 /* for lease breaks: */
1130 unsigned long fl_break_time;
1131 unsigned long fl_downgrade_time;
1132
1133 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1134 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1135 union {
1136 struct nfs_lock_info nfs_fl;
1137 struct nfs4_lock_info nfs4_fl;
1138 struct {
1139 struct list_head link; /* link in AFS vnode's pending_locks list */
1140 int state; /* state of grant or error if -ve */
1141 unsigned int debug_id;
1142 } afs;
1143 } fl_u;
1144 } __randomize_layout;
1145
1146 struct file_lock_context {
1147 spinlock_t flc_lock;
1148 struct list_head flc_flock;
1149 struct list_head flc_posix;
1150 struct list_head flc_lease;
1151 };
1152
1153 /* The following constant reflects the upper bound of the file/locking space */
1154 #ifndef OFFSET_MAX
1155 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1156 #define OFFSET_MAX INT_LIMIT(loff_t)
1157 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1158 #endif
1159
1160 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1161
1162 #define locks_inode(f) file_inode(f)
1163
1164 #ifdef CONFIG_FILE_LOCKING
1165 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1166 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1167 struct flock *);
1168
1169 #if BITS_PER_LONG == 32
1170 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1171 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1172 struct flock64 *);
1173 #endif
1174
1175 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1176 extern int fcntl_getlease(struct file *filp);
1177
1178 /* fs/locks.c */
1179 void locks_free_lock_context(struct inode *inode);
1180 void locks_free_lock(struct file_lock *fl);
1181 extern void locks_init_lock(struct file_lock *);
1182 extern struct file_lock * locks_alloc_lock(void);
1183 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1184 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1185 extern void locks_remove_posix(struct file *, fl_owner_t);
1186 extern void locks_remove_file(struct file *);
1187 extern void locks_release_private(struct file_lock *);
1188 extern void posix_test_lock(struct file *, struct file_lock *);
1189 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1190 extern int locks_delete_block(struct file_lock *);
1191 extern int vfs_test_lock(struct file *, struct file_lock *);
1192 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1193 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1194 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1195 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1196 extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1197 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1198 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1199 extern int lease_modify(struct file_lock *, int, struct list_head *);
1200
1201 struct notifier_block;
1202 extern int lease_register_notifier(struct notifier_block *);
1203 extern void lease_unregister_notifier(struct notifier_block *);
1204
1205 struct files_struct;
1206 extern void show_fd_locks(struct seq_file *f,
1207 struct file *filp, struct files_struct *files);
1208 #else /* !CONFIG_FILE_LOCKING */
fcntl_getlk(struct file * file,unsigned int cmd,struct flock __user * user)1209 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1210 struct flock __user *user)
1211 {
1212 return -EINVAL;
1213 }
1214
fcntl_setlk(unsigned int fd,struct file * file,unsigned int cmd,struct flock __user * user)1215 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1216 unsigned int cmd, struct flock __user *user)
1217 {
1218 return -EACCES;
1219 }
1220
1221 #if BITS_PER_LONG == 32
fcntl_getlk64(struct file * file,unsigned int cmd,struct flock64 __user * user)1222 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1223 struct flock64 __user *user)
1224 {
1225 return -EINVAL;
1226 }
1227
fcntl_setlk64(unsigned int fd,struct file * file,unsigned int cmd,struct flock64 __user * user)1228 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1229 unsigned int cmd, struct flock64 __user *user)
1230 {
1231 return -EACCES;
1232 }
1233 #endif
fcntl_setlease(unsigned int fd,struct file * filp,long arg)1234 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1235 {
1236 return -EINVAL;
1237 }
1238
fcntl_getlease(struct file * filp)1239 static inline int fcntl_getlease(struct file *filp)
1240 {
1241 return F_UNLCK;
1242 }
1243
1244 static inline void
locks_free_lock_context(struct inode * inode)1245 locks_free_lock_context(struct inode *inode)
1246 {
1247 }
1248
locks_init_lock(struct file_lock * fl)1249 static inline void locks_init_lock(struct file_lock *fl)
1250 {
1251 return;
1252 }
1253
locks_copy_conflock(struct file_lock * new,struct file_lock * fl)1254 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1255 {
1256 return;
1257 }
1258
locks_copy_lock(struct file_lock * new,struct file_lock * fl)1259 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1260 {
1261 return;
1262 }
1263
locks_remove_posix(struct file * filp,fl_owner_t owner)1264 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1265 {
1266 return;
1267 }
1268
locks_remove_file(struct file * filp)1269 static inline void locks_remove_file(struct file *filp)
1270 {
1271 return;
1272 }
1273
posix_test_lock(struct file * filp,struct file_lock * fl)1274 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1275 {
1276 return;
1277 }
1278
posix_lock_file(struct file * filp,struct file_lock * fl,struct file_lock * conflock)1279 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1280 struct file_lock *conflock)
1281 {
1282 return -ENOLCK;
1283 }
1284
locks_delete_block(struct file_lock * waiter)1285 static inline int locks_delete_block(struct file_lock *waiter)
1286 {
1287 return -ENOENT;
1288 }
1289
vfs_test_lock(struct file * filp,struct file_lock * fl)1290 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1291 {
1292 return 0;
1293 }
1294
vfs_lock_file(struct file * filp,unsigned int cmd,struct file_lock * fl,struct file_lock * conf)1295 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1296 struct file_lock *fl, struct file_lock *conf)
1297 {
1298 return -ENOLCK;
1299 }
1300
vfs_cancel_lock(struct file * filp,struct file_lock * fl)1301 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1302 {
1303 return 0;
1304 }
1305
locks_lock_inode_wait(struct inode * inode,struct file_lock * fl)1306 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1307 {
1308 return -ENOLCK;
1309 }
1310
__break_lease(struct inode * inode,unsigned int mode,unsigned int type)1311 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1312 {
1313 return 0;
1314 }
1315
lease_get_mtime(struct inode * inode,struct timespec64 * time)1316 static inline void lease_get_mtime(struct inode *inode,
1317 struct timespec64 *time)
1318 {
1319 return;
1320 }
1321
generic_setlease(struct file * filp,long arg,struct file_lock ** flp,void ** priv)1322 static inline int generic_setlease(struct file *filp, long arg,
1323 struct file_lock **flp, void **priv)
1324 {
1325 return -EINVAL;
1326 }
1327
vfs_setlease(struct file * filp,long arg,struct file_lock ** lease,void ** priv)1328 static inline int vfs_setlease(struct file *filp, long arg,
1329 struct file_lock **lease, void **priv)
1330 {
1331 return -EINVAL;
1332 }
1333
lease_modify(struct file_lock * fl,int arg,struct list_head * dispose)1334 static inline int lease_modify(struct file_lock *fl, int arg,
1335 struct list_head *dispose)
1336 {
1337 return -EINVAL;
1338 }
1339
1340 struct files_struct;
show_fd_locks(struct seq_file * f,struct file * filp,struct files_struct * files)1341 static inline void show_fd_locks(struct seq_file *f,
1342 struct file *filp, struct files_struct *files) {}
1343 #endif /* !CONFIG_FILE_LOCKING */
1344
file_inode(const struct file * f)1345 static inline struct inode *file_inode(const struct file *f)
1346 {
1347 return f->f_inode;
1348 }
1349
file_dentry(const struct file * file)1350 static inline struct dentry *file_dentry(const struct file *file)
1351 {
1352 return d_real(file->f_path.dentry, file_inode(file));
1353 }
1354
locks_lock_file_wait(struct file * filp,struct file_lock * fl)1355 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1356 {
1357 return locks_lock_inode_wait(locks_inode(filp), fl);
1358 }
1359
1360 struct fasync_struct {
1361 rwlock_t fa_lock;
1362 int magic;
1363 int fa_fd;
1364 struct fasync_struct *fa_next; /* singly linked list */
1365 struct file *fa_file;
1366 struct rcu_head fa_rcu;
1367 };
1368
1369 #define FASYNC_MAGIC 0x4601
1370
1371 /* SMP safe fasync helpers: */
1372 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1373 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1374 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1375 extern struct fasync_struct *fasync_alloc(void);
1376 extern void fasync_free(struct fasync_struct *);
1377
1378 /* can be called from interrupts */
1379 extern void kill_fasync(struct fasync_struct **, int, int);
1380
1381 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1382 extern int f_setown(struct file *filp, unsigned long arg, int force);
1383 extern void f_delown(struct file *filp);
1384 extern pid_t f_getown(struct file *filp);
1385 extern int send_sigurg(struct fown_struct *fown);
1386
1387 /*
1388 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1389 * represented in both.
1390 */
1391 #define SB_RDONLY 1 /* Mount read-only */
1392 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1393 #define SB_NODEV 4 /* Disallow access to device special files */
1394 #define SB_NOEXEC 8 /* Disallow program execution */
1395 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1396 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1397 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1398 #define SB_NOATIME 1024 /* Do not update access times. */
1399 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1400 #define SB_SILENT 32768
1401 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1402 #define SB_INLINECRYPT (1<<17) /* Use blk-crypto for encrypted files */
1403 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1404 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1405 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1406
1407 /* These sb flags are internal to the kernel */
1408 #define SB_SUBMOUNT (1<<26)
1409 #define SB_FORCE (1<<27)
1410 #define SB_NOSEC (1<<28)
1411 #define SB_BORN (1<<29)
1412 #define SB_ACTIVE (1<<30)
1413 #define SB_NOUSER (1<<31)
1414
1415 /* These flags relate to encoding and casefolding */
1416 #define SB_ENC_STRICT_MODE_FL (1 << 0)
1417
1418 #define sb_has_strict_encoding(sb) \
1419 (sb->s_encoding_flags & SB_ENC_STRICT_MODE_FL)
1420
1421 /*
1422 * Umount options
1423 */
1424
1425 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1426 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1427 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1428 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1429 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1430
1431 /* sb->s_iflags */
1432 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1433 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1434 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1435 #define SB_I_STABLE_WRITES 0x00000008 /* don't modify blks until WB is done */
1436
1437 /* sb->s_iflags to limit user namespace mounts */
1438 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1439 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1440 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1441
1442 #define SB_I_SKIP_SYNC 0x00000100 /* Skip superblock at global sync */
1443 #define SB_I_PERSB_BDI 0x00000200 /* has a per-sb bdi */
1444
1445 /* Possible states of 'frozen' field */
1446 enum {
1447 SB_UNFROZEN = 0, /* FS is unfrozen */
1448 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1449 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1450 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1451 * internal threads if needed) */
1452 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1453 };
1454
1455 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1456
1457 struct sb_writers {
1458 int frozen; /* Is sb frozen? */
1459 wait_queue_head_t wait_unfrozen; /* wait for thaw */
1460 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1461 };
1462
1463 struct super_block {
1464 struct list_head s_list; /* Keep this first */
1465 dev_t s_dev; /* search index; _not_ kdev_t */
1466 unsigned char s_blocksize_bits;
1467 unsigned long s_blocksize;
1468 loff_t s_maxbytes; /* Max file size */
1469 struct file_system_type *s_type;
1470 const struct super_operations *s_op;
1471 const struct dquot_operations *dq_op;
1472 const struct quotactl_ops *s_qcop;
1473 const struct export_operations *s_export_op;
1474 unsigned long s_flags;
1475 unsigned long s_iflags; /* internal SB_I_* flags */
1476 unsigned long s_magic;
1477 struct dentry *s_root;
1478 struct rw_semaphore s_umount;
1479 int s_count;
1480 atomic_t s_active;
1481 #ifdef CONFIG_SECURITY
1482 void *s_security;
1483 #endif
1484 const struct xattr_handler **s_xattr;
1485 #ifdef CONFIG_FS_ENCRYPTION
1486 const struct fscrypt_operations *s_cop;
1487 struct key *s_master_keys; /* master crypto keys in use */
1488 #endif
1489 #ifdef CONFIG_FS_VERITY
1490 const struct fsverity_operations *s_vop;
1491 #endif
1492 #ifdef CONFIG_UNICODE
1493 struct unicode_map *s_encoding;
1494 __u16 s_encoding_flags;
1495 #endif
1496 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1497 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1498 struct block_device *s_bdev;
1499 struct backing_dev_info *s_bdi;
1500 struct mtd_info *s_mtd;
1501 struct hlist_node s_instances;
1502 unsigned int s_quota_types; /* Bitmask of supported quota types */
1503 struct quota_info s_dquot; /* Diskquota specific options */
1504
1505 struct sb_writers s_writers;
1506
1507 /*
1508 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1509 * s_fsnotify_marks together for cache efficiency. They are frequently
1510 * accessed and rarely modified.
1511 */
1512 void *s_fs_info; /* Filesystem private info */
1513
1514 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1515 u32 s_time_gran;
1516 /* Time limits for c/m/atime in seconds */
1517 time64_t s_time_min;
1518 time64_t s_time_max;
1519 #ifdef CONFIG_FSNOTIFY
1520 __u32 s_fsnotify_mask;
1521 struct fsnotify_mark_connector __rcu *s_fsnotify_marks;
1522 #endif
1523
1524 char s_id[32]; /* Informational name */
1525 uuid_t s_uuid; /* UUID */
1526
1527 unsigned int s_max_links;
1528 fmode_t s_mode;
1529
1530 /*
1531 * The next field is for VFS *only*. No filesystems have any business
1532 * even looking at it. You had been warned.
1533 */
1534 struct mutex s_vfs_rename_mutex; /* Kludge */
1535
1536 /*
1537 * Filesystem subtype. If non-empty the filesystem type field
1538 * in /proc/mounts will be "type.subtype"
1539 */
1540 const char *s_subtype;
1541
1542 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1543
1544 /*
1545 * Saved pool identifier for cleancache (-1 means none)
1546 */
1547 int cleancache_poolid;
1548
1549 struct shrinker s_shrink; /* per-sb shrinker handle */
1550
1551 /* Number of inodes with nlink == 0 but still referenced */
1552 atomic_long_t s_remove_count;
1553
1554 /*
1555 * Number of inode/mount/sb objects that are being watched, note that
1556 * inodes objects are currently double-accounted.
1557 */
1558 atomic_long_t s_fsnotify_connectors;
1559
1560 /* Being remounted read-only */
1561 int s_readonly_remount;
1562
1563 /* per-sb errseq_t for reporting writeback errors via syncfs */
1564 errseq_t s_wb_err;
1565
1566 /* AIO completions deferred from interrupt context */
1567 struct workqueue_struct *s_dio_done_wq;
1568 struct hlist_head s_pins;
1569
1570 /*
1571 * Owning user namespace and default context in which to
1572 * interpret filesystem uids, gids, quotas, device nodes,
1573 * xattrs and security labels.
1574 */
1575 struct user_namespace *s_user_ns;
1576
1577 /*
1578 * The list_lru structure is essentially just a pointer to a table
1579 * of per-node lru lists, each of which has its own spinlock.
1580 * There is no need to put them into separate cachelines.
1581 */
1582 struct list_lru s_dentry_lru;
1583 struct list_lru s_inode_lru;
1584 struct rcu_head rcu;
1585 struct work_struct destroy_work;
1586
1587 struct mutex s_sync_lock; /* sync serialisation lock */
1588
1589 /*
1590 * Indicates how deep in a filesystem stack this SB is
1591 */
1592 int s_stack_depth;
1593
1594 /* s_inode_list_lock protects s_inodes */
1595 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1596 struct list_head s_inodes; /* all inodes */
1597
1598 spinlock_t s_inode_wblist_lock;
1599 struct list_head s_inodes_wb; /* writeback inodes */
1600 } __randomize_layout;
1601
1602 /* Helper functions so that in most cases filesystems will
1603 * not need to deal directly with kuid_t and kgid_t and can
1604 * instead deal with the raw numeric values that are stored
1605 * in the filesystem.
1606 */
i_uid_read(const struct inode * inode)1607 static inline uid_t i_uid_read(const struct inode *inode)
1608 {
1609 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1610 }
1611
i_gid_read(const struct inode * inode)1612 static inline gid_t i_gid_read(const struct inode *inode)
1613 {
1614 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1615 }
1616
i_uid_write(struct inode * inode,uid_t uid)1617 static inline void i_uid_write(struct inode *inode, uid_t uid)
1618 {
1619 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1620 }
1621
i_gid_write(struct inode * inode,gid_t gid)1622 static inline void i_gid_write(struct inode *inode, gid_t gid)
1623 {
1624 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1625 }
1626
1627 /**
1628 * kuid_into_mnt - map a kuid down into a mnt_userns
1629 * @mnt_userns: user namespace of the relevant mount
1630 * @kuid: kuid to be mapped
1631 *
1632 * Return: @kuid mapped according to @mnt_userns.
1633 * If @kuid has no mapping INVALID_UID is returned.
1634 */
kuid_into_mnt(struct user_namespace * mnt_userns,kuid_t kuid)1635 static inline kuid_t kuid_into_mnt(struct user_namespace *mnt_userns,
1636 kuid_t kuid)
1637 {
1638 return make_kuid(mnt_userns, __kuid_val(kuid));
1639 }
1640
1641 /**
1642 * kgid_into_mnt - map a kgid down into a mnt_userns
1643 * @mnt_userns: user namespace of the relevant mount
1644 * @kgid: kgid to be mapped
1645 *
1646 * Return: @kgid mapped according to @mnt_userns.
1647 * If @kgid has no mapping INVALID_GID is returned.
1648 */
kgid_into_mnt(struct user_namespace * mnt_userns,kgid_t kgid)1649 static inline kgid_t kgid_into_mnt(struct user_namespace *mnt_userns,
1650 kgid_t kgid)
1651 {
1652 return make_kgid(mnt_userns, __kgid_val(kgid));
1653 }
1654
1655 /**
1656 * i_uid_into_mnt - map an inode's i_uid down into a mnt_userns
1657 * @mnt_userns: user namespace of the mount the inode was found from
1658 * @inode: inode to map
1659 *
1660 * Return: the inode's i_uid mapped down according to @mnt_userns.
1661 * If the inode's i_uid has no mapping INVALID_UID is returned.
1662 */
i_uid_into_mnt(struct user_namespace * mnt_userns,const struct inode * inode)1663 static inline kuid_t i_uid_into_mnt(struct user_namespace *mnt_userns,
1664 const struct inode *inode)
1665 {
1666 return kuid_into_mnt(mnt_userns, inode->i_uid);
1667 }
1668
1669 /**
1670 * i_gid_into_mnt - map an inode's i_gid down into a mnt_userns
1671 * @mnt_userns: user namespace of the mount the inode was found from
1672 * @inode: inode to map
1673 *
1674 * Return: the inode's i_gid mapped down according to @mnt_userns.
1675 * If the inode's i_gid has no mapping INVALID_GID is returned.
1676 */
i_gid_into_mnt(struct user_namespace * mnt_userns,const struct inode * inode)1677 static inline kgid_t i_gid_into_mnt(struct user_namespace *mnt_userns,
1678 const struct inode *inode)
1679 {
1680 return kgid_into_mnt(mnt_userns, inode->i_gid);
1681 }
1682
1683 /**
1684 * kuid_from_mnt - map a kuid up into a mnt_userns
1685 * @mnt_userns: user namespace of the relevant mount
1686 * @kuid: kuid to be mapped
1687 *
1688 * Return: @kuid mapped up according to @mnt_userns.
1689 * If @kuid has no mapping INVALID_UID is returned.
1690 */
kuid_from_mnt(struct user_namespace * mnt_userns,kuid_t kuid)1691 static inline kuid_t kuid_from_mnt(struct user_namespace *mnt_userns,
1692 kuid_t kuid)
1693 {
1694 return KUIDT_INIT(from_kuid(mnt_userns, kuid));
1695 }
1696
1697 /**
1698 * kgid_from_mnt - map a kgid up into a mnt_userns
1699 * @mnt_userns: user namespace of the relevant mount
1700 * @kgid: kgid to be mapped
1701 *
1702 * Return: @kgid mapped up according to @mnt_userns.
1703 * If @kgid has no mapping INVALID_GID is returned.
1704 */
kgid_from_mnt(struct user_namespace * mnt_userns,kgid_t kgid)1705 static inline kgid_t kgid_from_mnt(struct user_namespace *mnt_userns,
1706 kgid_t kgid)
1707 {
1708 return KGIDT_INIT(from_kgid(mnt_userns, kgid));
1709 }
1710
1711 /**
1712 * mapped_fsuid - return caller's fsuid mapped up into a mnt_userns
1713 * @mnt_userns: user namespace of the relevant mount
1714 *
1715 * Use this helper to initialize a new vfs or filesystem object based on
1716 * the caller's fsuid. A common example is initializing the i_uid field of
1717 * a newly allocated inode triggered by a creation event such as mkdir or
1718 * O_CREAT. Other examples include the allocation of quotas for a specific
1719 * user.
1720 *
1721 * Return: the caller's current fsuid mapped up according to @mnt_userns.
1722 */
mapped_fsuid(struct user_namespace * mnt_userns)1723 static inline kuid_t mapped_fsuid(struct user_namespace *mnt_userns)
1724 {
1725 return kuid_from_mnt(mnt_userns, current_fsuid());
1726 }
1727
1728 /**
1729 * mapped_fsgid - return caller's fsgid mapped up into a mnt_userns
1730 * @mnt_userns: user namespace of the relevant mount
1731 *
1732 * Use this helper to initialize a new vfs or filesystem object based on
1733 * the caller's fsgid. A common example is initializing the i_gid field of
1734 * a newly allocated inode triggered by a creation event such as mkdir or
1735 * O_CREAT. Other examples include the allocation of quotas for a specific
1736 * user.
1737 *
1738 * Return: the caller's current fsgid mapped up according to @mnt_userns.
1739 */
mapped_fsgid(struct user_namespace * mnt_userns)1740 static inline kgid_t mapped_fsgid(struct user_namespace *mnt_userns)
1741 {
1742 return kgid_from_mnt(mnt_userns, current_fsgid());
1743 }
1744
1745 /**
1746 * inode_fsuid_set - initialize inode's i_uid field with callers fsuid
1747 * @inode: inode to initialize
1748 * @mnt_userns: user namespace of the mount the inode was found from
1749 *
1750 * Initialize the i_uid field of @inode. If the inode was found/created via
1751 * an idmapped mount map the caller's fsuid according to @mnt_users.
1752 */
inode_fsuid_set(struct inode * inode,struct user_namespace * mnt_userns)1753 static inline void inode_fsuid_set(struct inode *inode,
1754 struct user_namespace *mnt_userns)
1755 {
1756 inode->i_uid = mapped_fsuid(mnt_userns);
1757 }
1758
1759 /**
1760 * inode_fsgid_set - initialize inode's i_gid field with callers fsgid
1761 * @inode: inode to initialize
1762 * @mnt_userns: user namespace of the mount the inode was found from
1763 *
1764 * Initialize the i_gid field of @inode. If the inode was found/created via
1765 * an idmapped mount map the caller's fsgid according to @mnt_users.
1766 */
inode_fsgid_set(struct inode * inode,struct user_namespace * mnt_userns)1767 static inline void inode_fsgid_set(struct inode *inode,
1768 struct user_namespace *mnt_userns)
1769 {
1770 inode->i_gid = mapped_fsgid(mnt_userns);
1771 }
1772
1773 /**
1774 * fsuidgid_has_mapping() - check whether caller's fsuid/fsgid is mapped
1775 * @sb: the superblock we want a mapping in
1776 * @mnt_userns: user namespace of the relevant mount
1777 *
1778 * Check whether the caller's fsuid and fsgid have a valid mapping in the
1779 * s_user_ns of the superblock @sb. If the caller is on an idmapped mount map
1780 * the caller's fsuid and fsgid according to the @mnt_userns first.
1781 *
1782 * Return: true if fsuid and fsgid is mapped, false if not.
1783 */
fsuidgid_has_mapping(struct super_block * sb,struct user_namespace * mnt_userns)1784 static inline bool fsuidgid_has_mapping(struct super_block *sb,
1785 struct user_namespace *mnt_userns)
1786 {
1787 struct user_namespace *s_user_ns = sb->s_user_ns;
1788
1789 return kuid_has_mapping(s_user_ns, mapped_fsuid(mnt_userns)) &&
1790 kgid_has_mapping(s_user_ns, mapped_fsgid(mnt_userns));
1791 }
1792
1793 extern struct timespec64 current_time(struct inode *inode);
1794
1795 /*
1796 * Snapshotting support.
1797 */
1798
1799 /*
1800 * These are internal functions, please use sb_start_{write,pagefault,intwrite}
1801 * instead.
1802 */
__sb_end_write(struct super_block * sb,int level)1803 static inline void __sb_end_write(struct super_block *sb, int level)
1804 {
1805 percpu_up_read(sb->s_writers.rw_sem + level-1);
1806 }
1807
__sb_start_write(struct super_block * sb,int level)1808 static inline void __sb_start_write(struct super_block *sb, int level)
1809 {
1810 percpu_down_read(sb->s_writers.rw_sem + level - 1);
1811 }
1812
__sb_start_write_trylock(struct super_block * sb,int level)1813 static inline bool __sb_start_write_trylock(struct super_block *sb, int level)
1814 {
1815 return percpu_down_read_trylock(sb->s_writers.rw_sem + level - 1);
1816 }
1817
1818 #define __sb_writers_acquired(sb, lev) \
1819 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1820 #define __sb_writers_release(sb, lev) \
1821 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1822
1823 /**
1824 * sb_end_write - drop write access to a superblock
1825 * @sb: the super we wrote to
1826 *
1827 * Decrement number of writers to the filesystem. Wake up possible waiters
1828 * wanting to freeze the filesystem.
1829 */
sb_end_write(struct super_block * sb)1830 static inline void sb_end_write(struct super_block *sb)
1831 {
1832 __sb_end_write(sb, SB_FREEZE_WRITE);
1833 }
1834
1835 /**
1836 * sb_end_pagefault - drop write access to a superblock from a page fault
1837 * @sb: the super we wrote to
1838 *
1839 * Decrement number of processes handling write page fault to the filesystem.
1840 * Wake up possible waiters wanting to freeze the filesystem.
1841 */
sb_end_pagefault(struct super_block * sb)1842 static inline void sb_end_pagefault(struct super_block *sb)
1843 {
1844 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1845 }
1846
1847 /**
1848 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1849 * @sb: the super we wrote to
1850 *
1851 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1852 * waiters wanting to freeze the filesystem.
1853 */
sb_end_intwrite(struct super_block * sb)1854 static inline void sb_end_intwrite(struct super_block *sb)
1855 {
1856 __sb_end_write(sb, SB_FREEZE_FS);
1857 }
1858
1859 /**
1860 * sb_start_write - get write access to a superblock
1861 * @sb: the super we write to
1862 *
1863 * When a process wants to write data or metadata to a file system (i.e. dirty
1864 * a page or an inode), it should embed the operation in a sb_start_write() -
1865 * sb_end_write() pair to get exclusion against file system freezing. This
1866 * function increments number of writers preventing freezing. If the file
1867 * system is already frozen, the function waits until the file system is
1868 * thawed.
1869 *
1870 * Since freeze protection behaves as a lock, users have to preserve
1871 * ordering of freeze protection and other filesystem locks. Generally,
1872 * freeze protection should be the outermost lock. In particular, we have:
1873 *
1874 * sb_start_write
1875 * -> i_mutex (write path, truncate, directory ops, ...)
1876 * -> s_umount (freeze_super, thaw_super)
1877 */
sb_start_write(struct super_block * sb)1878 static inline void sb_start_write(struct super_block *sb)
1879 {
1880 __sb_start_write(sb, SB_FREEZE_WRITE);
1881 }
1882
sb_start_write_trylock(struct super_block * sb)1883 static inline bool sb_start_write_trylock(struct super_block *sb)
1884 {
1885 return __sb_start_write_trylock(sb, SB_FREEZE_WRITE);
1886 }
1887
1888 /**
1889 * sb_start_pagefault - get write access to a superblock from a page fault
1890 * @sb: the super we write to
1891 *
1892 * When a process starts handling write page fault, it should embed the
1893 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1894 * exclusion against file system freezing. This is needed since the page fault
1895 * is going to dirty a page. This function increments number of running page
1896 * faults preventing freezing. If the file system is already frozen, the
1897 * function waits until the file system is thawed.
1898 *
1899 * Since page fault freeze protection behaves as a lock, users have to preserve
1900 * ordering of freeze protection and other filesystem locks. It is advised to
1901 * put sb_start_pagefault() close to mmap_lock in lock ordering. Page fault
1902 * handling code implies lock dependency:
1903 *
1904 * mmap_lock
1905 * -> sb_start_pagefault
1906 */
sb_start_pagefault(struct super_block * sb)1907 static inline void sb_start_pagefault(struct super_block *sb)
1908 {
1909 __sb_start_write(sb, SB_FREEZE_PAGEFAULT);
1910 }
1911
1912 /**
1913 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1914 * @sb: the super we write to
1915 *
1916 * This is the third level of protection against filesystem freezing. It is
1917 * free for use by a filesystem. The only requirement is that it must rank
1918 * below sb_start_pagefault.
1919 *
1920 * For example filesystem can call sb_start_intwrite() when starting a
1921 * transaction which somewhat eases handling of freezing for internal sources
1922 * of filesystem changes (internal fs threads, discarding preallocation on file
1923 * close, etc.).
1924 */
sb_start_intwrite(struct super_block * sb)1925 static inline void sb_start_intwrite(struct super_block *sb)
1926 {
1927 __sb_start_write(sb, SB_FREEZE_FS);
1928 }
1929
sb_start_intwrite_trylock(struct super_block * sb)1930 static inline bool sb_start_intwrite_trylock(struct super_block *sb)
1931 {
1932 return __sb_start_write_trylock(sb, SB_FREEZE_FS);
1933 }
1934
1935 bool inode_owner_or_capable(struct user_namespace *mnt_userns,
1936 const struct inode *inode);
1937
1938 /*
1939 * VFS helper functions..
1940 */
1941 int vfs_create(struct user_namespace *, struct inode *,
1942 struct dentry *, umode_t, bool);
1943 int vfs_mkdir(struct user_namespace *, struct inode *,
1944 struct dentry *, umode_t);
1945 int vfs_mknod(struct user_namespace *, struct inode *, struct dentry *,
1946 umode_t, dev_t);
1947 int vfs_symlink(struct user_namespace *, struct inode *,
1948 struct dentry *, const char *);
1949 int vfs_link(struct dentry *, struct user_namespace *, struct inode *,
1950 struct dentry *, struct inode **);
1951 int vfs_rmdir(struct user_namespace *, struct inode *, struct dentry *);
1952 int vfs_unlink(struct user_namespace *, struct inode *, struct dentry *,
1953 struct inode **);
1954
1955 /**
1956 * struct renamedata - contains all information required for renaming
1957 * @old_mnt_userns: old user namespace of the mount the inode was found from
1958 * @old_dir: parent of source
1959 * @old_dentry: source
1960 * @new_mnt_userns: new user namespace of the mount the inode was found from
1961 * @new_dir: parent of destination
1962 * @new_dentry: destination
1963 * @delegated_inode: returns an inode needing a delegation break
1964 * @flags: rename flags
1965 */
1966 struct renamedata {
1967 struct user_namespace *old_mnt_userns;
1968 struct inode *old_dir;
1969 struct dentry *old_dentry;
1970 struct user_namespace *new_mnt_userns;
1971 struct inode *new_dir;
1972 struct dentry *new_dentry;
1973 struct inode **delegated_inode;
1974 unsigned int flags;
1975 } __randomize_layout;
1976
1977 int vfs_rename(struct renamedata *);
1978
vfs_whiteout(struct user_namespace * mnt_userns,struct inode * dir,struct dentry * dentry)1979 static inline int vfs_whiteout(struct user_namespace *mnt_userns,
1980 struct inode *dir, struct dentry *dentry)
1981 {
1982 return vfs_mknod(mnt_userns, dir, dentry, S_IFCHR | WHITEOUT_MODE,
1983 WHITEOUT_DEV);
1984 }
1985
1986 struct dentry *vfs_tmpfile(struct user_namespace *mnt_userns,
1987 struct dentry *dentry, umode_t mode, int open_flag);
1988
1989 int vfs_mkobj(struct dentry *, umode_t,
1990 int (*f)(struct dentry *, umode_t, void *),
1991 void *);
1992
1993 int vfs_fchown(struct file *file, uid_t user, gid_t group);
1994 int vfs_fchmod(struct file *file, umode_t mode);
1995 int vfs_utimes(const struct path *path, struct timespec64 *times);
1996
1997 extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1998
1999 #ifdef CONFIG_COMPAT
2000 extern long compat_ptr_ioctl(struct file *file, unsigned int cmd,
2001 unsigned long arg);
2002 #else
2003 #define compat_ptr_ioctl NULL
2004 #endif
2005
2006 /*
2007 * VFS file helper functions.
2008 */
2009 void inode_init_owner(struct user_namespace *mnt_userns, struct inode *inode,
2010 const struct inode *dir, umode_t mode);
2011 extern bool may_open_dev(const struct path *path);
2012
2013 /*
2014 * This is the "filldir" function type, used by readdir() to let
2015 * the kernel specify what kind of dirent layout it wants to have.
2016 * This allows the kernel to read directories into kernel space or
2017 * to have different dirent layouts depending on the binary type.
2018 */
2019 struct dir_context;
2020 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
2021 unsigned);
2022
2023 struct dir_context {
2024 filldir_t actor;
2025 loff_t pos;
2026 };
2027
2028 /*
2029 * These flags let !MMU mmap() govern direct device mapping vs immediate
2030 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
2031 *
2032 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
2033 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
2034 * NOMMU_MAP_READ: Can be mapped for reading
2035 * NOMMU_MAP_WRITE: Can be mapped for writing
2036 * NOMMU_MAP_EXEC: Can be mapped for execution
2037 */
2038 #define NOMMU_MAP_COPY 0x00000001
2039 #define NOMMU_MAP_DIRECT 0x00000008
2040 #define NOMMU_MAP_READ VM_MAYREAD
2041 #define NOMMU_MAP_WRITE VM_MAYWRITE
2042 #define NOMMU_MAP_EXEC VM_MAYEXEC
2043
2044 #define NOMMU_VMFLAGS \
2045 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
2046
2047 /*
2048 * These flags control the behavior of the remap_file_range function pointer.
2049 * If it is called with len == 0 that means "remap to end of source file".
2050 * See Documentation/filesystems/vfs.rst for more details about this call.
2051 *
2052 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
2053 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
2054 */
2055 #define REMAP_FILE_DEDUP (1 << 0)
2056 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
2057
2058 /*
2059 * These flags signal that the caller is ok with altering various aspects of
2060 * the behavior of the remap operation. The changes must be made by the
2061 * implementation; the vfs remap helper functions can take advantage of them.
2062 * Flags in this category exist to preserve the quirky behavior of the hoisted
2063 * btrfs clone/dedupe ioctls.
2064 */
2065 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
2066
2067 struct iov_iter;
2068
2069 struct file_operations {
2070 struct module *owner;
2071 loff_t (*llseek) (struct file *, loff_t, int);
2072 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
2073 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
2074 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
2075 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
2076 int (*iopoll)(struct kiocb *kiocb, struct io_comp_batch *,
2077 unsigned int flags);
2078 int (*iterate) (struct file *, struct dir_context *);
2079 int (*iterate_shared) (struct file *, struct dir_context *);
2080 __poll_t (*poll) (struct file *, struct poll_table_struct *);
2081 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
2082 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
2083 int (*mmap) (struct file *, struct vm_area_struct *);
2084 unsigned long mmap_supported_flags;
2085 int (*open) (struct inode *, struct file *);
2086 int (*flush) (struct file *, fl_owner_t id);
2087 int (*release) (struct inode *, struct file *);
2088 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
2089 int (*fasync) (int, struct file *, int);
2090 int (*lock) (struct file *, int, struct file_lock *);
2091 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
2092 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
2093 int (*check_flags)(int);
2094 int (*flock) (struct file *, int, struct file_lock *);
2095 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
2096 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
2097 int (*setlease)(struct file *, long, struct file_lock **, void **);
2098 long (*fallocate)(struct file *file, int mode, loff_t offset,
2099 loff_t len);
2100 void (*show_fdinfo)(struct seq_file *m, struct file *f);
2101 #ifndef CONFIG_MMU
2102 unsigned (*mmap_capabilities)(struct file *);
2103 #endif
2104 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
2105 loff_t, size_t, unsigned int);
2106 loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
2107 struct file *file_out, loff_t pos_out,
2108 loff_t len, unsigned int remap_flags);
2109 int (*fadvise)(struct file *, loff_t, loff_t, int);
2110 } __randomize_layout;
2111
2112 struct inode_operations {
2113 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
2114 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
2115 int (*permission) (struct user_namespace *, struct inode *, int);
2116 struct posix_acl * (*get_acl)(struct inode *, int, bool);
2117
2118 int (*readlink) (struct dentry *, char __user *,int);
2119
2120 int (*create) (struct user_namespace *, struct inode *,struct dentry *,
2121 umode_t, bool);
2122 int (*link) (struct dentry *,struct inode *,struct dentry *);
2123 int (*unlink) (struct inode *,struct dentry *);
2124 int (*symlink) (struct user_namespace *, struct inode *,struct dentry *,
2125 const char *);
2126 int (*mkdir) (struct user_namespace *, struct inode *,struct dentry *,
2127 umode_t);
2128 int (*rmdir) (struct inode *,struct dentry *);
2129 int (*mknod) (struct user_namespace *, struct inode *,struct dentry *,
2130 umode_t,dev_t);
2131 int (*rename) (struct user_namespace *, struct inode *, struct dentry *,
2132 struct inode *, struct dentry *, unsigned int);
2133 int (*setattr) (struct user_namespace *, struct dentry *,
2134 struct iattr *);
2135 int (*getattr) (struct user_namespace *, const struct path *,
2136 struct kstat *, u32, unsigned int);
2137 ssize_t (*listxattr) (struct dentry *, char *, size_t);
2138 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
2139 u64 len);
2140 int (*update_time)(struct inode *, struct timespec64 *, int);
2141 int (*atomic_open)(struct inode *, struct dentry *,
2142 struct file *, unsigned open_flag,
2143 umode_t create_mode);
2144 int (*tmpfile) (struct user_namespace *, struct inode *,
2145 struct dentry *, umode_t);
2146 int (*set_acl)(struct user_namespace *, struct inode *,
2147 struct posix_acl *, int);
2148 int (*fileattr_set)(struct user_namespace *mnt_userns,
2149 struct dentry *dentry, struct fileattr *fa);
2150 int (*fileattr_get)(struct dentry *dentry, struct fileattr *fa);
2151 } ____cacheline_aligned;
2152
call_read_iter(struct file * file,struct kiocb * kio,struct iov_iter * iter)2153 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
2154 struct iov_iter *iter)
2155 {
2156 return file->f_op->read_iter(kio, iter);
2157 }
2158
call_write_iter(struct file * file,struct kiocb * kio,struct iov_iter * iter)2159 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
2160 struct iov_iter *iter)
2161 {
2162 return file->f_op->write_iter(kio, iter);
2163 }
2164
call_mmap(struct file * file,struct vm_area_struct * vma)2165 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
2166 {
2167 return file->f_op->mmap(file, vma);
2168 }
2169
2170 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
2171 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
2172 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
2173 loff_t, size_t, unsigned int);
2174 extern ssize_t generic_copy_file_range(struct file *file_in, loff_t pos_in,
2175 struct file *file_out, loff_t pos_out,
2176 size_t len, unsigned int flags);
2177 extern int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
2178 struct file *file_out, loff_t pos_out,
2179 loff_t *count,
2180 unsigned int remap_flags);
2181 extern loff_t do_clone_file_range(struct file *file_in, loff_t pos_in,
2182 struct file *file_out, loff_t pos_out,
2183 loff_t len, unsigned int remap_flags);
2184 extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in,
2185 struct file *file_out, loff_t pos_out,
2186 loff_t len, unsigned int remap_flags);
2187 extern int vfs_dedupe_file_range(struct file *file,
2188 struct file_dedupe_range *same);
2189 extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
2190 struct file *dst_file, loff_t dst_pos,
2191 loff_t len, unsigned int remap_flags);
2192
2193
2194 struct super_operations {
2195 struct inode *(*alloc_inode)(struct super_block *sb);
2196 void (*destroy_inode)(struct inode *);
2197 void (*free_inode)(struct inode *);
2198
2199 void (*dirty_inode) (struct inode *, int flags);
2200 int (*write_inode) (struct inode *, struct writeback_control *wbc);
2201 int (*drop_inode) (struct inode *);
2202 void (*evict_inode) (struct inode *);
2203 void (*put_super) (struct super_block *);
2204 int (*sync_fs)(struct super_block *sb, int wait);
2205 int (*freeze_super) (struct super_block *);
2206 int (*freeze_fs) (struct super_block *);
2207 int (*thaw_super) (struct super_block *);
2208 int (*unfreeze_fs) (struct super_block *);
2209 int (*statfs) (struct dentry *, struct kstatfs *);
2210 int (*remount_fs) (struct super_block *, int *, char *);
2211 void (*umount_begin) (struct super_block *);
2212
2213 int (*show_options)(struct seq_file *, struct dentry *);
2214 int (*show_devname)(struct seq_file *, struct dentry *);
2215 int (*show_path)(struct seq_file *, struct dentry *);
2216 int (*show_stats)(struct seq_file *, struct dentry *);
2217 #ifdef CONFIG_QUOTA
2218 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
2219 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
2220 struct dquot **(*get_dquots)(struct inode *);
2221 #endif
2222 long (*nr_cached_objects)(struct super_block *,
2223 struct shrink_control *);
2224 long (*free_cached_objects)(struct super_block *,
2225 struct shrink_control *);
2226 };
2227
2228 /*
2229 * Inode flags - they have no relation to superblock flags now
2230 */
2231 #define S_SYNC (1 << 0) /* Writes are synced at once */
2232 #define S_NOATIME (1 << 1) /* Do not update access times */
2233 #define S_APPEND (1 << 2) /* Append-only file */
2234 #define S_IMMUTABLE (1 << 3) /* Immutable file */
2235 #define S_DEAD (1 << 4) /* removed, but still open directory */
2236 #define S_NOQUOTA (1 << 5) /* Inode is not counted to quota */
2237 #define S_DIRSYNC (1 << 6) /* Directory modifications are synchronous */
2238 #define S_NOCMTIME (1 << 7) /* Do not update file c/mtime */
2239 #define S_SWAPFILE (1 << 8) /* Do not truncate: swapon got its bmaps */
2240 #define S_PRIVATE (1 << 9) /* Inode is fs-internal */
2241 #define S_IMA (1 << 10) /* Inode has an associated IMA struct */
2242 #define S_AUTOMOUNT (1 << 11) /* Automount/referral quasi-directory */
2243 #define S_NOSEC (1 << 12) /* no suid or xattr security attributes */
2244 #ifdef CONFIG_FS_DAX
2245 #define S_DAX (1 << 13) /* Direct Access, avoiding the page cache */
2246 #else
2247 #define S_DAX 0 /* Make all the DAX code disappear */
2248 #endif
2249 #define S_ENCRYPTED (1 << 14) /* Encrypted file (using fs/crypto/) */
2250 #define S_CASEFOLD (1 << 15) /* Casefolded file */
2251 #define S_VERITY (1 << 16) /* Verity file (using fs/verity/) */
2252
2253 /*
2254 * Note that nosuid etc flags are inode-specific: setting some file-system
2255 * flags just means all the inodes inherit those flags by default. It might be
2256 * possible to override it selectively if you really wanted to with some
2257 * ioctl() that is not currently implemented.
2258 *
2259 * Exception: SB_RDONLY is always applied to the entire file system.
2260 *
2261 * Unfortunately, it is possible to change a filesystems flags with it mounted
2262 * with files in use. This means that all of the inodes will not have their
2263 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
2264 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
2265 */
2266 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
2267
sb_rdonly(const struct super_block * sb)2268 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
2269 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
2270 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
2271 ((inode)->i_flags & S_SYNC))
2272 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
2273 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
2274 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
2275 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
2276 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
2277
2278 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
2279 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
2280 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
2281 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
2282
2283 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
2284 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
2285 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
2286 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
2287 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
2288 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2289 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2290 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2291 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2292 #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD)
2293 #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY)
2294
2295 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2296 (inode)->i_rdev == WHITEOUT_DEV)
2297
HAS_UNMAPPED_ID(struct user_namespace * mnt_userns,struct inode * inode)2298 static inline bool HAS_UNMAPPED_ID(struct user_namespace *mnt_userns,
2299 struct inode *inode)
2300 {
2301 return !uid_valid(i_uid_into_mnt(mnt_userns, inode)) ||
2302 !gid_valid(i_gid_into_mnt(mnt_userns, inode));
2303 }
2304
file_write_hint(struct file * file)2305 static inline enum rw_hint file_write_hint(struct file *file)
2306 {
2307 if (file->f_write_hint != WRITE_LIFE_NOT_SET)
2308 return file->f_write_hint;
2309
2310 return file_inode(file)->i_write_hint;
2311 }
2312
2313 static inline int iocb_flags(struct file *file);
2314
ki_hint_validate(enum rw_hint hint)2315 static inline u16 ki_hint_validate(enum rw_hint hint)
2316 {
2317 typeof(((struct kiocb *)0)->ki_hint) max_hint = -1;
2318
2319 if (hint <= max_hint)
2320 return hint;
2321 return 0;
2322 }
2323
init_sync_kiocb(struct kiocb * kiocb,struct file * filp)2324 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
2325 {
2326 *kiocb = (struct kiocb) {
2327 .ki_filp = filp,
2328 .ki_flags = iocb_flags(filp),
2329 .ki_hint = ki_hint_validate(file_write_hint(filp)),
2330 .ki_ioprio = get_current_ioprio(),
2331 };
2332 }
2333
kiocb_clone(struct kiocb * kiocb,struct kiocb * kiocb_src,struct file * filp)2334 static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src,
2335 struct file *filp)
2336 {
2337 *kiocb = (struct kiocb) {
2338 .ki_filp = filp,
2339 .ki_flags = kiocb_src->ki_flags,
2340 .ki_hint = kiocb_src->ki_hint,
2341 .ki_ioprio = kiocb_src->ki_ioprio,
2342 .ki_pos = kiocb_src->ki_pos,
2343 };
2344 }
2345
2346 /*
2347 * Inode state bits. Protected by inode->i_lock
2348 *
2349 * Four bits determine the dirty state of the inode: I_DIRTY_SYNC,
2350 * I_DIRTY_DATASYNC, I_DIRTY_PAGES, and I_DIRTY_TIME.
2351 *
2352 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2353 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2354 * various stages of removing an inode.
2355 *
2356 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2357 *
2358 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2359 * fdatasync() (unless I_DIRTY_DATASYNC is also set).
2360 * Timestamp updates are the usual cause.
2361 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2362 * these changes separately from I_DIRTY_SYNC so that we
2363 * don't have to write inode on fdatasync() when only
2364 * e.g. the timestamps have changed.
2365 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2366 * I_DIRTY_TIME The inode itself only has dirty timestamps, and the
2367 * lazytime mount option is enabled. We keep track of this
2368 * separately from I_DIRTY_SYNC in order to implement
2369 * lazytime. This gets cleared if I_DIRTY_INODE
2370 * (I_DIRTY_SYNC and/or I_DIRTY_DATASYNC) gets set. I.e.
2371 * either I_DIRTY_TIME *or* I_DIRTY_INODE can be set in
2372 * i_state, but not both. I_DIRTY_PAGES may still be set.
2373 * I_NEW Serves as both a mutex and completion notification.
2374 * New inodes set I_NEW. If two processes both create
2375 * the same inode, one of them will release its inode and
2376 * wait for I_NEW to be released before returning.
2377 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2378 * also cause waiting on I_NEW, without I_NEW actually
2379 * being set. find_inode() uses this to prevent returning
2380 * nearly-dead inodes.
2381 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2382 * is zero. I_FREEING must be set when I_WILL_FREE is
2383 * cleared.
2384 * I_FREEING Set when inode is about to be freed but still has dirty
2385 * pages or buffers attached or the inode itself is still
2386 * dirty.
2387 * I_CLEAR Added by clear_inode(). In this state the inode is
2388 * clean and can be destroyed. Inode keeps I_FREEING.
2389 *
2390 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2391 * prohibited for many purposes. iget() must wait for
2392 * the inode to be completely released, then create it
2393 * anew. Other functions will just ignore such inodes,
2394 * if appropriate. I_NEW is used for waiting.
2395 *
2396 * I_SYNC Writeback of inode is running. The bit is set during
2397 * data writeback, and cleared with a wakeup on the bit
2398 * address once it is done. The bit is also used to pin
2399 * the inode in memory for flusher thread.
2400 *
2401 * I_REFERENCED Marks the inode as recently references on the LRU list.
2402 *
2403 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2404 *
2405 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2406 * synchronize competing switching instances and to tell
2407 * wb stat updates to grab the i_pages lock. See
2408 * inode_switch_wbs_work_fn() for details.
2409 *
2410 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2411 * and work dirs among overlayfs mounts.
2412 *
2413 * I_CREATING New object's inode in the middle of setting up.
2414 *
2415 * I_DONTCACHE Evict inode as soon as it is not used anymore.
2416 *
2417 * I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists.
2418 * Used to detect that mark_inode_dirty() should not move
2419 * inode between dirty lists.
2420 *
2421 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2422 */
2423 #define I_DIRTY_SYNC (1 << 0)
2424 #define I_DIRTY_DATASYNC (1 << 1)
2425 #define I_DIRTY_PAGES (1 << 2)
2426 #define __I_NEW 3
2427 #define I_NEW (1 << __I_NEW)
2428 #define I_WILL_FREE (1 << 4)
2429 #define I_FREEING (1 << 5)
2430 #define I_CLEAR (1 << 6)
2431 #define __I_SYNC 7
2432 #define I_SYNC (1 << __I_SYNC)
2433 #define I_REFERENCED (1 << 8)
2434 #define __I_DIO_WAKEUP 9
2435 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2436 #define I_LINKABLE (1 << 10)
2437 #define I_DIRTY_TIME (1 << 11)
2438 #define I_WB_SWITCH (1 << 13)
2439 #define I_OVL_INUSE (1 << 14)
2440 #define I_CREATING (1 << 15)
2441 #define I_DONTCACHE (1 << 16)
2442 #define I_SYNC_QUEUED (1 << 17)
2443
2444 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2445 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2446 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2447
2448 extern void __mark_inode_dirty(struct inode *, int);
mark_inode_dirty(struct inode * inode)2449 static inline void mark_inode_dirty(struct inode *inode)
2450 {
2451 __mark_inode_dirty(inode, I_DIRTY);
2452 }
2453
mark_inode_dirty_sync(struct inode * inode)2454 static inline void mark_inode_dirty_sync(struct inode *inode)
2455 {
2456 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2457 }
2458
2459 /*
2460 * Returns true if the given inode itself only has dirty timestamps (its pages
2461 * may still be dirty) and isn't currently being allocated or freed.
2462 * Filesystems should call this if when writing an inode when lazytime is
2463 * enabled, they want to opportunistically write the timestamps of other inodes
2464 * located very nearby on-disk, e.g. in the same inode block. This returns true
2465 * if the given inode is in need of such an opportunistic update. Requires
2466 * i_lock, or at least later re-checking under i_lock.
2467 */
inode_is_dirtytime_only(struct inode * inode)2468 static inline bool inode_is_dirtytime_only(struct inode *inode)
2469 {
2470 return (inode->i_state & (I_DIRTY_TIME | I_NEW |
2471 I_FREEING | I_WILL_FREE)) == I_DIRTY_TIME;
2472 }
2473
2474 extern void inc_nlink(struct inode *inode);
2475 extern void drop_nlink(struct inode *inode);
2476 extern void clear_nlink(struct inode *inode);
2477 extern void set_nlink(struct inode *inode, unsigned int nlink);
2478
inode_inc_link_count(struct inode * inode)2479 static inline void inode_inc_link_count(struct inode *inode)
2480 {
2481 inc_nlink(inode);
2482 mark_inode_dirty(inode);
2483 }
2484
inode_dec_link_count(struct inode * inode)2485 static inline void inode_dec_link_count(struct inode *inode)
2486 {
2487 drop_nlink(inode);
2488 mark_inode_dirty(inode);
2489 }
2490
2491 enum file_time_flags {
2492 S_ATIME = 1,
2493 S_MTIME = 2,
2494 S_CTIME = 4,
2495 S_VERSION = 8,
2496 };
2497
2498 extern bool atime_needs_update(const struct path *, struct inode *);
2499 extern void touch_atime(const struct path *);
2500 int inode_update_time(struct inode *inode, struct timespec64 *time, int flags);
2501
file_accessed(struct file * file)2502 static inline void file_accessed(struct file *file)
2503 {
2504 if (!(file->f_flags & O_NOATIME))
2505 touch_atime(&file->f_path);
2506 }
2507
2508 extern int file_modified(struct file *file);
2509
2510 int sync_inode_metadata(struct inode *inode, int wait);
2511
2512 struct file_system_type {
2513 const char *name;
2514 int fs_flags;
2515 #define FS_REQUIRES_DEV 1
2516 #define FS_BINARY_MOUNTDATA 2
2517 #define FS_HAS_SUBTYPE 4
2518 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2519 #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */
2520 #define FS_ALLOW_IDMAP 32 /* FS has been updated to handle vfs idmappings. */
2521 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2522 int (*init_fs_context)(struct fs_context *);
2523 const struct fs_parameter_spec *parameters;
2524 struct dentry *(*mount) (struct file_system_type *, int,
2525 const char *, void *);
2526 void (*kill_sb) (struct super_block *);
2527 struct module *owner;
2528 struct file_system_type * next;
2529 struct hlist_head fs_supers;
2530
2531 struct lock_class_key s_lock_key;
2532 struct lock_class_key s_umount_key;
2533 struct lock_class_key s_vfs_rename_key;
2534 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2535
2536 struct lock_class_key i_lock_key;
2537 struct lock_class_key i_mutex_key;
2538 struct lock_class_key invalidate_lock_key;
2539 struct lock_class_key i_mutex_dir_key;
2540 };
2541
2542 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2543
2544 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2545 int flags, const char *dev_name, void *data,
2546 int (*fill_super)(struct super_block *, void *, int));
2547 extern struct dentry *mount_single(struct file_system_type *fs_type,
2548 int flags, void *data,
2549 int (*fill_super)(struct super_block *, void *, int));
2550 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2551 int flags, void *data,
2552 int (*fill_super)(struct super_block *, void *, int));
2553 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2554 void generic_shutdown_super(struct super_block *sb);
2555 void kill_block_super(struct super_block *sb);
2556 void kill_anon_super(struct super_block *sb);
2557 void kill_litter_super(struct super_block *sb);
2558 void deactivate_super(struct super_block *sb);
2559 void deactivate_locked_super(struct super_block *sb);
2560 int set_anon_super(struct super_block *s, void *data);
2561 int set_anon_super_fc(struct super_block *s, struct fs_context *fc);
2562 int get_anon_bdev(dev_t *);
2563 void free_anon_bdev(dev_t);
2564 struct super_block *sget_fc(struct fs_context *fc,
2565 int (*test)(struct super_block *, struct fs_context *),
2566 int (*set)(struct super_block *, struct fs_context *));
2567 struct super_block *sget(struct file_system_type *type,
2568 int (*test)(struct super_block *,void *),
2569 int (*set)(struct super_block *,void *),
2570 int flags, void *data);
2571
2572 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2573 #define fops_get(fops) \
2574 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2575 #define fops_put(fops) \
2576 do { if (fops) module_put((fops)->owner); } while(0)
2577 /*
2578 * This one is to be used *ONLY* from ->open() instances.
2579 * fops must be non-NULL, pinned down *and* module dependencies
2580 * should be sufficient to pin the caller down as well.
2581 */
2582 #define replace_fops(f, fops) \
2583 do { \
2584 struct file *__file = (f); \
2585 fops_put(__file->f_op); \
2586 BUG_ON(!(__file->f_op = (fops))); \
2587 } while(0)
2588
2589 extern int register_filesystem(struct file_system_type *);
2590 extern int unregister_filesystem(struct file_system_type *);
2591 extern struct vfsmount *kern_mount(struct file_system_type *);
2592 extern void kern_unmount(struct vfsmount *mnt);
2593 extern int may_umount_tree(struct vfsmount *);
2594 extern int may_umount(struct vfsmount *);
2595 extern long do_mount(const char *, const char __user *,
2596 const char *, unsigned long, void *);
2597 extern struct vfsmount *collect_mounts(const struct path *);
2598 extern void drop_collected_mounts(struct vfsmount *);
2599 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2600 struct vfsmount *);
2601 extern int vfs_statfs(const struct path *, struct kstatfs *);
2602 extern int user_statfs(const char __user *, struct kstatfs *);
2603 extern int fd_statfs(int, struct kstatfs *);
2604 extern int freeze_super(struct super_block *super);
2605 extern int thaw_super(struct super_block *super);
2606 extern bool our_mnt(struct vfsmount *mnt);
2607 extern __printf(2, 3)
2608 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2609 extern int super_setup_bdi(struct super_block *sb);
2610
2611 extern int current_umask(void);
2612
2613 extern void ihold(struct inode * inode);
2614 extern void iput(struct inode *);
2615 extern int generic_update_time(struct inode *, struct timespec64 *, int);
2616
2617 /* /sys/fs */
2618 extern struct kobject *fs_kobj;
2619
2620 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2621
2622 #ifdef CONFIG_FILE_LOCKING
break_lease(struct inode * inode,unsigned int mode)2623 static inline int break_lease(struct inode *inode, unsigned int mode)
2624 {
2625 /*
2626 * Since this check is lockless, we must ensure that any refcounts
2627 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2628 * could end up racing with tasks trying to set a new lease on this
2629 * file.
2630 */
2631 smp_mb();
2632 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2633 return __break_lease(inode, mode, FL_LEASE);
2634 return 0;
2635 }
2636
break_deleg(struct inode * inode,unsigned int mode)2637 static inline int break_deleg(struct inode *inode, unsigned int mode)
2638 {
2639 /*
2640 * Since this check is lockless, we must ensure that any refcounts
2641 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2642 * could end up racing with tasks trying to set a new lease on this
2643 * file.
2644 */
2645 smp_mb();
2646 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2647 return __break_lease(inode, mode, FL_DELEG);
2648 return 0;
2649 }
2650
try_break_deleg(struct inode * inode,struct inode ** delegated_inode)2651 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2652 {
2653 int ret;
2654
2655 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2656 if (ret == -EWOULDBLOCK && delegated_inode) {
2657 *delegated_inode = inode;
2658 ihold(inode);
2659 }
2660 return ret;
2661 }
2662
break_deleg_wait(struct inode ** delegated_inode)2663 static inline int break_deleg_wait(struct inode **delegated_inode)
2664 {
2665 int ret;
2666
2667 ret = break_deleg(*delegated_inode, O_WRONLY);
2668 iput(*delegated_inode);
2669 *delegated_inode = NULL;
2670 return ret;
2671 }
2672
break_layout(struct inode * inode,bool wait)2673 static inline int break_layout(struct inode *inode, bool wait)
2674 {
2675 smp_mb();
2676 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2677 return __break_lease(inode,
2678 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2679 FL_LAYOUT);
2680 return 0;
2681 }
2682
2683 #else /* !CONFIG_FILE_LOCKING */
break_lease(struct inode * inode,unsigned int mode)2684 static inline int break_lease(struct inode *inode, unsigned int mode)
2685 {
2686 return 0;
2687 }
2688
break_deleg(struct inode * inode,unsigned int mode)2689 static inline int break_deleg(struct inode *inode, unsigned int mode)
2690 {
2691 return 0;
2692 }
2693
try_break_deleg(struct inode * inode,struct inode ** delegated_inode)2694 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2695 {
2696 return 0;
2697 }
2698
break_deleg_wait(struct inode ** delegated_inode)2699 static inline int break_deleg_wait(struct inode **delegated_inode)
2700 {
2701 BUG();
2702 return 0;
2703 }
2704
break_layout(struct inode * inode,bool wait)2705 static inline int break_layout(struct inode *inode, bool wait)
2706 {
2707 return 0;
2708 }
2709
2710 #endif /* CONFIG_FILE_LOCKING */
2711
2712 /* fs/open.c */
2713 struct audit_names;
2714 struct filename {
2715 const char *name; /* pointer to actual string */
2716 const __user char *uptr; /* original userland pointer */
2717 int refcnt;
2718 struct audit_names *aname;
2719 const char iname[];
2720 };
2721 static_assert(offsetof(struct filename, iname) % sizeof(long) == 0);
2722
file_mnt_user_ns(struct file * file)2723 static inline struct user_namespace *file_mnt_user_ns(struct file *file)
2724 {
2725 return mnt_user_ns(file->f_path.mnt);
2726 }
2727 extern long vfs_truncate(const struct path *, loff_t);
2728 int do_truncate(struct user_namespace *, struct dentry *, loff_t start,
2729 unsigned int time_attrs, struct file *filp);
2730 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2731 loff_t len);
2732 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2733 umode_t mode);
2734 extern struct file *file_open_name(struct filename *, int, umode_t);
2735 extern struct file *filp_open(const char *, int, umode_t);
2736 extern struct file *file_open_root(const struct path *,
2737 const char *, int, umode_t);
file_open_root_mnt(struct vfsmount * mnt,const char * name,int flags,umode_t mode)2738 static inline struct file *file_open_root_mnt(struct vfsmount *mnt,
2739 const char *name, int flags, umode_t mode)
2740 {
2741 return file_open_root(&(struct path){.mnt = mnt, .dentry = mnt->mnt_root},
2742 name, flags, mode);
2743 }
2744 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2745 extern struct file * open_with_fake_path(const struct path *, int,
2746 struct inode*, const struct cred *);
file_clone_open(struct file * file)2747 static inline struct file *file_clone_open(struct file *file)
2748 {
2749 return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2750 }
2751 extern int filp_close(struct file *, fl_owner_t id);
2752
2753 extern struct filename *getname_flags(const char __user *, int, int *);
2754 extern struct filename *getname_uflags(const char __user *, int);
2755 extern struct filename *getname(const char __user *);
2756 extern struct filename *getname_kernel(const char *);
2757 extern void putname(struct filename *name);
2758
2759 extern int finish_open(struct file *file, struct dentry *dentry,
2760 int (*open)(struct inode *, struct file *));
2761 extern int finish_no_open(struct file *file, struct dentry *dentry);
2762
2763 /* fs/dcache.c */
2764 extern void __init vfs_caches_init_early(void);
2765 extern void __init vfs_caches_init(void);
2766
2767 extern struct kmem_cache *names_cachep;
2768
2769 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2770 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2771
2772 extern struct super_block *blockdev_superblock;
sb_is_blkdev_sb(struct super_block * sb)2773 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2774 {
2775 return IS_ENABLED(CONFIG_BLOCK) && sb == blockdev_superblock;
2776 }
2777
2778 void emergency_thaw_all(void);
2779 extern int sync_filesystem(struct super_block *);
2780 extern const struct file_operations def_blk_fops;
2781 extern const struct file_operations def_chr_fops;
2782
2783 /* fs/char_dev.c */
2784 #define CHRDEV_MAJOR_MAX 512
2785 /* Marks the bottom of the first segment of free char majors */
2786 #define CHRDEV_MAJOR_DYN_END 234
2787 /* Marks the top and bottom of the second segment of free char majors */
2788 #define CHRDEV_MAJOR_DYN_EXT_START 511
2789 #define CHRDEV_MAJOR_DYN_EXT_END 384
2790
2791 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2792 extern int register_chrdev_region(dev_t, unsigned, const char *);
2793 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2794 unsigned int count, const char *name,
2795 const struct file_operations *fops);
2796 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2797 unsigned int count, const char *name);
2798 extern void unregister_chrdev_region(dev_t, unsigned);
2799 extern void chrdev_show(struct seq_file *,off_t);
2800
register_chrdev(unsigned int major,const char * name,const struct file_operations * fops)2801 static inline int register_chrdev(unsigned int major, const char *name,
2802 const struct file_operations *fops)
2803 {
2804 return __register_chrdev(major, 0, 256, name, fops);
2805 }
2806
unregister_chrdev(unsigned int major,const char * name)2807 static inline void unregister_chrdev(unsigned int major, const char *name)
2808 {
2809 __unregister_chrdev(major, 0, 256, name);
2810 }
2811
2812 extern void init_special_inode(struct inode *, umode_t, dev_t);
2813
2814 /* Invalid inode operations -- fs/bad_inode.c */
2815 extern void make_bad_inode(struct inode *);
2816 extern bool is_bad_inode(struct inode *);
2817
2818 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2819 pgoff_t start, pgoff_t end);
2820
2821 void invalidate_mapping_pagevec(struct address_space *mapping,
2822 pgoff_t start, pgoff_t end,
2823 unsigned long *nr_pagevec);
2824
invalidate_remote_inode(struct inode * inode)2825 static inline void invalidate_remote_inode(struct inode *inode)
2826 {
2827 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2828 S_ISLNK(inode->i_mode))
2829 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2830 }
2831 extern int invalidate_inode_pages2(struct address_space *mapping);
2832 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2833 pgoff_t start, pgoff_t end);
2834 extern int write_inode_now(struct inode *, int);
2835 extern int filemap_fdatawrite(struct address_space *);
2836 extern int filemap_flush(struct address_space *);
2837 extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2838 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2839 loff_t lend);
2840 extern int filemap_fdatawait_range_keep_errors(struct address_space *mapping,
2841 loff_t start_byte, loff_t end_byte);
2842
filemap_fdatawait(struct address_space * mapping)2843 static inline int filemap_fdatawait(struct address_space *mapping)
2844 {
2845 return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2846 }
2847
2848 extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2849 loff_t lend);
2850 extern bool filemap_range_needs_writeback(struct address_space *,
2851 loff_t lstart, loff_t lend);
2852 extern int filemap_write_and_wait_range(struct address_space *mapping,
2853 loff_t lstart, loff_t lend);
2854 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2855 loff_t start, loff_t end, int sync_mode);
2856 extern int filemap_fdatawrite_range(struct address_space *mapping,
2857 loff_t start, loff_t end);
2858 extern int filemap_check_errors(struct address_space *mapping);
2859 extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2860 int filemap_fdatawrite_wbc(struct address_space *mapping,
2861 struct writeback_control *wbc);
2862
filemap_write_and_wait(struct address_space * mapping)2863 static inline int filemap_write_and_wait(struct address_space *mapping)
2864 {
2865 return filemap_write_and_wait_range(mapping, 0, LLONG_MAX);
2866 }
2867
2868 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2869 loff_t lend);
2870 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2871 extern int __must_check file_write_and_wait_range(struct file *file,
2872 loff_t start, loff_t end);
2873
file_write_and_wait(struct file * file)2874 static inline int file_write_and_wait(struct file *file)
2875 {
2876 return file_write_and_wait_range(file, 0, LLONG_MAX);
2877 }
2878
2879 /**
2880 * filemap_set_wb_err - set a writeback error on an address_space
2881 * @mapping: mapping in which to set writeback error
2882 * @err: error to be set in mapping
2883 *
2884 * When writeback fails in some way, we must record that error so that
2885 * userspace can be informed when fsync and the like are called. We endeavor
2886 * to report errors on any file that was open at the time of the error. Some
2887 * internal callers also need to know when writeback errors have occurred.
2888 *
2889 * When a writeback error occurs, most filesystems will want to call
2890 * filemap_set_wb_err to record the error in the mapping so that it will be
2891 * automatically reported whenever fsync is called on the file.
2892 */
filemap_set_wb_err(struct address_space * mapping,int err)2893 static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2894 {
2895 /* Fastpath for common case of no error */
2896 if (unlikely(err))
2897 __filemap_set_wb_err(mapping, err);
2898 }
2899
2900 /**
2901 * filemap_check_wb_err - has an error occurred since the mark was sampled?
2902 * @mapping: mapping to check for writeback errors
2903 * @since: previously-sampled errseq_t
2904 *
2905 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2906 * the given value was sampled.
2907 *
2908 * If it has then report the latest error set, otherwise return 0.
2909 */
filemap_check_wb_err(struct address_space * mapping,errseq_t since)2910 static inline int filemap_check_wb_err(struct address_space *mapping,
2911 errseq_t since)
2912 {
2913 return errseq_check(&mapping->wb_err, since);
2914 }
2915
2916 /**
2917 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2918 * @mapping: mapping to be sampled
2919 *
2920 * Writeback errors are always reported relative to a particular sample point
2921 * in the past. This function provides those sample points.
2922 */
filemap_sample_wb_err(struct address_space * mapping)2923 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2924 {
2925 return errseq_sample(&mapping->wb_err);
2926 }
2927
2928 /**
2929 * file_sample_sb_err - sample the current errseq_t to test for later errors
2930 * @file: file pointer to be sampled
2931 *
2932 * Grab the most current superblock-level errseq_t value for the given
2933 * struct file.
2934 */
file_sample_sb_err(struct file * file)2935 static inline errseq_t file_sample_sb_err(struct file *file)
2936 {
2937 return errseq_sample(&file->f_path.dentry->d_sb->s_wb_err);
2938 }
2939
2940 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2941 int datasync);
2942 extern int vfs_fsync(struct file *file, int datasync);
2943
2944 extern int sync_file_range(struct file *file, loff_t offset, loff_t nbytes,
2945 unsigned int flags);
2946
2947 /*
2948 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2949 * to already be updated for the write, and will return either the amount
2950 * of bytes passed in, or an error if syncing the file failed.
2951 */
generic_write_sync(struct kiocb * iocb,ssize_t count)2952 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2953 {
2954 if (iocb->ki_flags & IOCB_DSYNC) {
2955 int ret = vfs_fsync_range(iocb->ki_filp,
2956 iocb->ki_pos - count, iocb->ki_pos - 1,
2957 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2958 if (ret)
2959 return ret;
2960 }
2961
2962 return count;
2963 }
2964
2965 extern void emergency_sync(void);
2966 extern void emergency_remount(void);
2967
2968 #ifdef CONFIG_BLOCK
2969 extern int bmap(struct inode *inode, sector_t *block);
2970 #else
bmap(struct inode * inode,sector_t * block)2971 static inline int bmap(struct inode *inode, sector_t *block)
2972 {
2973 return -EINVAL;
2974 }
2975 #endif
2976
2977 int notify_change(struct user_namespace *, struct dentry *,
2978 struct iattr *, struct inode **);
2979 int inode_permission(struct user_namespace *, struct inode *, int);
2980 int generic_permission(struct user_namespace *, struct inode *, int);
file_permission(struct file * file,int mask)2981 static inline int file_permission(struct file *file, int mask)
2982 {
2983 return inode_permission(file_mnt_user_ns(file),
2984 file_inode(file), mask);
2985 }
path_permission(const struct path * path,int mask)2986 static inline int path_permission(const struct path *path, int mask)
2987 {
2988 return inode_permission(mnt_user_ns(path->mnt),
2989 d_inode(path->dentry), mask);
2990 }
2991 int __check_sticky(struct user_namespace *mnt_userns, struct inode *dir,
2992 struct inode *inode);
2993
execute_ok(struct inode * inode)2994 static inline bool execute_ok(struct inode *inode)
2995 {
2996 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2997 }
2998
inode_wrong_type(const struct inode * inode,umode_t mode)2999 static inline bool inode_wrong_type(const struct inode *inode, umode_t mode)
3000 {
3001 return (inode->i_mode ^ mode) & S_IFMT;
3002 }
3003
file_start_write(struct file * file)3004 static inline void file_start_write(struct file *file)
3005 {
3006 if (!S_ISREG(file_inode(file)->i_mode))
3007 return;
3008 sb_start_write(file_inode(file)->i_sb);
3009 }
3010
file_start_write_trylock(struct file * file)3011 static inline bool file_start_write_trylock(struct file *file)
3012 {
3013 if (!S_ISREG(file_inode(file)->i_mode))
3014 return true;
3015 return sb_start_write_trylock(file_inode(file)->i_sb);
3016 }
3017
file_end_write(struct file * file)3018 static inline void file_end_write(struct file *file)
3019 {
3020 if (!S_ISREG(file_inode(file)->i_mode))
3021 return;
3022 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
3023 }
3024
3025 /*
3026 * This is used for regular files where some users -- especially the
3027 * currently executed binary in a process, previously handled via
3028 * VM_DENYWRITE -- cannot handle concurrent write (and maybe mmap
3029 * read-write shared) accesses.
3030 *
3031 * get_write_access() gets write permission for a file.
3032 * put_write_access() releases this write permission.
3033 * deny_write_access() denies write access to a file.
3034 * allow_write_access() re-enables write access to a file.
3035 *
3036 * The i_writecount field of an inode can have the following values:
3037 * 0: no write access, no denied write access
3038 * < 0: (-i_writecount) users that denied write access to the file.
3039 * > 0: (i_writecount) users that have write access to the file.
3040 *
3041 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
3042 * except for the cases where we don't hold i_writecount yet. Then we need to
3043 * use {get,deny}_write_access() - these functions check the sign and refuse
3044 * to do the change if sign is wrong.
3045 */
get_write_access(struct inode * inode)3046 static inline int get_write_access(struct inode *inode)
3047 {
3048 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
3049 }
deny_write_access(struct file * file)3050 static inline int deny_write_access(struct file *file)
3051 {
3052 struct inode *inode = file_inode(file);
3053 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
3054 }
put_write_access(struct inode * inode)3055 static inline void put_write_access(struct inode * inode)
3056 {
3057 atomic_dec(&inode->i_writecount);
3058 }
allow_write_access(struct file * file)3059 static inline void allow_write_access(struct file *file)
3060 {
3061 if (file)
3062 atomic_inc(&file_inode(file)->i_writecount);
3063 }
inode_is_open_for_write(const struct inode * inode)3064 static inline bool inode_is_open_for_write(const struct inode *inode)
3065 {
3066 return atomic_read(&inode->i_writecount) > 0;
3067 }
3068
3069 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
i_readcount_dec(struct inode * inode)3070 static inline void i_readcount_dec(struct inode *inode)
3071 {
3072 BUG_ON(!atomic_read(&inode->i_readcount));
3073 atomic_dec(&inode->i_readcount);
3074 }
i_readcount_inc(struct inode * inode)3075 static inline void i_readcount_inc(struct inode *inode)
3076 {
3077 atomic_inc(&inode->i_readcount);
3078 }
3079 #else
i_readcount_dec(struct inode * inode)3080 static inline void i_readcount_dec(struct inode *inode)
3081 {
3082 return;
3083 }
i_readcount_inc(struct inode * inode)3084 static inline void i_readcount_inc(struct inode *inode)
3085 {
3086 return;
3087 }
3088 #endif
3089 extern int do_pipe_flags(int *, int);
3090
3091 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
3092 ssize_t __kernel_read(struct file *file, void *buf, size_t count, loff_t *pos);
3093 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
3094 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
3095 extern struct file * open_exec(const char *);
3096
3097 /* fs/dcache.c -- generic fs support functions */
3098 extern bool is_subdir(struct dentry *, struct dentry *);
3099 extern bool path_is_under(const struct path *, const struct path *);
3100
3101 extern char *file_path(struct file *, char *, int);
3102
3103 #include <linux/err.h>
3104
3105 /* needed for stackable file system support */
3106 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
3107
3108 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
3109
3110 extern int inode_init_always(struct super_block *, struct inode *);
3111 extern void inode_init_once(struct inode *);
3112 extern void address_space_init_once(struct address_space *mapping);
3113 extern struct inode * igrab(struct inode *);
3114 extern ino_t iunique(struct super_block *, ino_t);
3115 extern int inode_needs_sync(struct inode *inode);
3116 extern int generic_delete_inode(struct inode *inode);
generic_drop_inode(struct inode * inode)3117 static inline int generic_drop_inode(struct inode *inode)
3118 {
3119 return !inode->i_nlink || inode_unhashed(inode);
3120 }
3121 extern void d_mark_dontcache(struct inode *inode);
3122
3123 extern struct inode *ilookup5_nowait(struct super_block *sb,
3124 unsigned long hashval, int (*test)(struct inode *, void *),
3125 void *data);
3126 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
3127 int (*test)(struct inode *, void *), void *data);
3128 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
3129
3130 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
3131 int (*test)(struct inode *, void *),
3132 int (*set)(struct inode *, void *),
3133 void *data);
3134 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
3135 extern struct inode * iget_locked(struct super_block *, unsigned long);
3136 extern struct inode *find_inode_nowait(struct super_block *,
3137 unsigned long,
3138 int (*match)(struct inode *,
3139 unsigned long, void *),
3140 void *data);
3141 extern struct inode *find_inode_rcu(struct super_block *, unsigned long,
3142 int (*)(struct inode *, void *), void *);
3143 extern struct inode *find_inode_by_ino_rcu(struct super_block *, unsigned long);
3144 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
3145 extern int insert_inode_locked(struct inode *);
3146 #ifdef CONFIG_DEBUG_LOCK_ALLOC
3147 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
3148 #else
lockdep_annotate_inode_mutex_key(struct inode * inode)3149 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
3150 #endif
3151 extern void unlock_new_inode(struct inode *);
3152 extern void discard_new_inode(struct inode *);
3153 extern unsigned int get_next_ino(void);
3154 extern void evict_inodes(struct super_block *sb);
3155
3156 /*
3157 * Userspace may rely on the the inode number being non-zero. For example, glibc
3158 * simply ignores files with zero i_ino in unlink() and other places.
3159 *
3160 * As an additional complication, if userspace was compiled with
3161 * _FILE_OFFSET_BITS=32 on a 64-bit kernel we'll only end up reading out the
3162 * lower 32 bits, so we need to check that those aren't zero explicitly. With
3163 * _FILE_OFFSET_BITS=64, this may cause some harmless false-negatives, but
3164 * better safe than sorry.
3165 */
is_zero_ino(ino_t ino)3166 static inline bool is_zero_ino(ino_t ino)
3167 {
3168 return (u32)ino == 0;
3169 }
3170
3171 extern void __iget(struct inode * inode);
3172 extern void iget_failed(struct inode *);
3173 extern void clear_inode(struct inode *);
3174 extern void __destroy_inode(struct inode *);
3175 extern struct inode *new_inode_pseudo(struct super_block *sb);
3176 extern struct inode *new_inode(struct super_block *sb);
3177 extern void free_inode_nonrcu(struct inode *inode);
3178 extern int should_remove_suid(struct dentry *);
3179 extern int file_remove_privs(struct file *);
3180
3181 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
insert_inode_hash(struct inode * inode)3182 static inline void insert_inode_hash(struct inode *inode)
3183 {
3184 __insert_inode_hash(inode, inode->i_ino);
3185 }
3186
3187 extern void __remove_inode_hash(struct inode *);
remove_inode_hash(struct inode * inode)3188 static inline void remove_inode_hash(struct inode *inode)
3189 {
3190 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
3191 __remove_inode_hash(inode);
3192 }
3193
3194 extern void inode_sb_list_add(struct inode *inode);
3195 extern void inode_add_lru(struct inode *inode);
3196
3197 extern int sb_set_blocksize(struct super_block *, int);
3198 extern int sb_min_blocksize(struct super_block *, int);
3199
3200 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
3201 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
3202 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
3203 extern int generic_write_check_limits(struct file *file, loff_t pos,
3204 loff_t *count);
3205 extern int generic_file_rw_checks(struct file *file_in, struct file *file_out);
3206 ssize_t filemap_read(struct kiocb *iocb, struct iov_iter *to,
3207 ssize_t already_read);
3208 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
3209 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
3210 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
3211 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
3212 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
3213
3214 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
3215 rwf_t flags);
3216 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
3217 rwf_t flags);
3218 ssize_t vfs_iocb_iter_read(struct file *file, struct kiocb *iocb,
3219 struct iov_iter *iter);
3220 ssize_t vfs_iocb_iter_write(struct file *file, struct kiocb *iocb,
3221 struct iov_iter *iter);
3222
3223 /* fs/splice.c */
3224 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
3225 struct pipe_inode_info *, size_t, unsigned int);
3226 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
3227 struct file *, loff_t *, size_t, unsigned int);
3228 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
3229 struct file *out, loff_t *, size_t len, unsigned int flags);
3230 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
3231 loff_t *opos, size_t len, unsigned int flags);
3232
3233
3234 extern void
3235 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3236 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3237 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
3238 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3239 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3240 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3241 int whence, loff_t maxsize, loff_t eof);
3242 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3243 int whence, loff_t size);
3244 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3245 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3246 extern int generic_file_open(struct inode * inode, struct file * filp);
3247 extern int nonseekable_open(struct inode * inode, struct file * filp);
3248 extern int stream_open(struct inode * inode, struct file * filp);
3249
3250 #ifdef CONFIG_BLOCK
3251 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3252 loff_t file_offset);
3253
3254 enum {
3255 /* need locking between buffered and direct access */
3256 DIO_LOCKING = 0x01,
3257
3258 /* filesystem does not support filling holes */
3259 DIO_SKIP_HOLES = 0x02,
3260 };
3261
3262 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3263 struct block_device *bdev, struct iov_iter *iter,
3264 get_block_t get_block,
3265 dio_iodone_t end_io, dio_submit_t submit_io,
3266 int flags);
3267
blockdev_direct_IO(struct kiocb * iocb,struct inode * inode,struct iov_iter * iter,get_block_t get_block)3268 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3269 struct inode *inode,
3270 struct iov_iter *iter,
3271 get_block_t get_block)
3272 {
3273 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3274 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3275 }
3276 #endif
3277
3278 void inode_dio_wait(struct inode *inode);
3279
3280 /**
3281 * inode_dio_begin - signal start of a direct I/O requests
3282 * @inode: inode the direct I/O happens on
3283 *
3284 * This is called once we've finished processing a direct I/O request,
3285 * and is used to wake up callers waiting for direct I/O to be quiesced.
3286 */
inode_dio_begin(struct inode * inode)3287 static inline void inode_dio_begin(struct inode *inode)
3288 {
3289 atomic_inc(&inode->i_dio_count);
3290 }
3291
3292 /**
3293 * inode_dio_end - signal finish of a direct I/O requests
3294 * @inode: inode the direct I/O happens on
3295 *
3296 * This is called once we've finished processing a direct I/O request,
3297 * and is used to wake up callers waiting for direct I/O to be quiesced.
3298 */
inode_dio_end(struct inode * inode)3299 static inline void inode_dio_end(struct inode *inode)
3300 {
3301 if (atomic_dec_and_test(&inode->i_dio_count))
3302 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3303 }
3304
3305 /*
3306 * Warn about a page cache invalidation failure diring a direct I/O write.
3307 */
3308 void dio_warn_stale_pagecache(struct file *filp);
3309
3310 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3311 unsigned int mask);
3312
3313 extern const struct file_operations generic_ro_fops;
3314
3315 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3316
3317 extern int readlink_copy(char __user *, int, const char *);
3318 extern int page_readlink(struct dentry *, char __user *, int);
3319 extern const char *page_get_link(struct dentry *, struct inode *,
3320 struct delayed_call *);
3321 extern void page_put_link(void *);
3322 extern int __page_symlink(struct inode *inode, const char *symname, int len,
3323 int nofs);
3324 extern int page_symlink(struct inode *inode, const char *symname, int len);
3325 extern const struct inode_operations page_symlink_inode_operations;
3326 extern void kfree_link(void *);
3327 void generic_fillattr(struct user_namespace *, struct inode *, struct kstat *);
3328 void generic_fill_statx_attr(struct inode *inode, struct kstat *stat);
3329 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3330 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3331 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3332 void inode_add_bytes(struct inode *inode, loff_t bytes);
3333 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3334 void inode_sub_bytes(struct inode *inode, loff_t bytes);
__inode_get_bytes(struct inode * inode)3335 static inline loff_t __inode_get_bytes(struct inode *inode)
3336 {
3337 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3338 }
3339 loff_t inode_get_bytes(struct inode *inode);
3340 void inode_set_bytes(struct inode *inode, loff_t bytes);
3341 const char *simple_get_link(struct dentry *, struct inode *,
3342 struct delayed_call *);
3343 extern const struct inode_operations simple_symlink_inode_operations;
3344
3345 extern int iterate_dir(struct file *, struct dir_context *);
3346
3347 int vfs_fstatat(int dfd, const char __user *filename, struct kstat *stat,
3348 int flags);
3349 int vfs_fstat(int fd, struct kstat *stat);
3350
vfs_stat(const char __user * filename,struct kstat * stat)3351 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3352 {
3353 return vfs_fstatat(AT_FDCWD, filename, stat, 0);
3354 }
vfs_lstat(const char __user * name,struct kstat * stat)3355 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3356 {
3357 return vfs_fstatat(AT_FDCWD, name, stat, AT_SYMLINK_NOFOLLOW);
3358 }
3359
3360 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3361 extern int vfs_readlink(struct dentry *, char __user *, int);
3362
3363 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3364 extern void put_filesystem(struct file_system_type *fs);
3365 extern struct file_system_type *get_fs_type(const char *name);
3366 extern struct super_block *get_super(struct block_device *);
3367 extern struct super_block *get_active_super(struct block_device *bdev);
3368 extern void drop_super(struct super_block *sb);
3369 extern void drop_super_exclusive(struct super_block *sb);
3370 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3371 extern void iterate_supers_type(struct file_system_type *,
3372 void (*)(struct super_block *, void *), void *);
3373
3374 extern int dcache_dir_open(struct inode *, struct file *);
3375 extern int dcache_dir_close(struct inode *, struct file *);
3376 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3377 extern int dcache_readdir(struct file *, struct dir_context *);
3378 extern int simple_setattr(struct user_namespace *, struct dentry *,
3379 struct iattr *);
3380 extern int simple_getattr(struct user_namespace *, const struct path *,
3381 struct kstat *, u32, unsigned int);
3382 extern int simple_statfs(struct dentry *, struct kstatfs *);
3383 extern int simple_open(struct inode *inode, struct file *file);
3384 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3385 extern int simple_unlink(struct inode *, struct dentry *);
3386 extern int simple_rmdir(struct inode *, struct dentry *);
3387 extern int simple_rename_exchange(struct inode *old_dir, struct dentry *old_dentry,
3388 struct inode *new_dir, struct dentry *new_dentry);
3389 extern int simple_rename(struct user_namespace *, struct inode *,
3390 struct dentry *, struct inode *, struct dentry *,
3391 unsigned int);
3392 extern void simple_recursive_removal(struct dentry *,
3393 void (*callback)(struct dentry *));
3394 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3395 extern void noop_invalidatepage(struct page *page, unsigned int offset,
3396 unsigned int length);
3397 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3398 extern int simple_empty(struct dentry *);
3399 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3400 loff_t pos, unsigned len, unsigned flags,
3401 struct page **pagep, void **fsdata);
3402 extern const struct address_space_operations ram_aops;
3403 extern int always_delete_dentry(const struct dentry *);
3404 extern struct inode *alloc_anon_inode(struct super_block *);
3405 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3406 extern const struct dentry_operations simple_dentry_operations;
3407
3408 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3409 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3410 extern const struct file_operations simple_dir_operations;
3411 extern const struct inode_operations simple_dir_inode_operations;
3412 extern void make_empty_dir_inode(struct inode *inode);
3413 extern bool is_empty_dir_inode(struct inode *inode);
3414 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3415 struct dentry *d_alloc_name(struct dentry *, const char *);
3416 extern int simple_fill_super(struct super_block *, unsigned long,
3417 const struct tree_descr *);
3418 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3419 extern void simple_release_fs(struct vfsmount **mount, int *count);
3420
3421 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3422 loff_t *ppos, const void *from, size_t available);
3423 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3424 const void __user *from, size_t count);
3425
3426 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3427 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3428
3429 extern int generic_check_addressable(unsigned, u64);
3430
3431 extern void generic_set_encrypted_ci_d_ops(struct dentry *dentry);
3432
3433 #ifdef CONFIG_MIGRATION
3434 extern int buffer_migrate_page(struct address_space *,
3435 struct page *, struct page *,
3436 enum migrate_mode);
3437 extern int buffer_migrate_page_norefs(struct address_space *,
3438 struct page *, struct page *,
3439 enum migrate_mode);
3440 #else
3441 #define buffer_migrate_page NULL
3442 #define buffer_migrate_page_norefs NULL
3443 #endif
3444
3445 int may_setattr(struct user_namespace *mnt_userns, struct inode *inode,
3446 unsigned int ia_valid);
3447 int setattr_prepare(struct user_namespace *, struct dentry *, struct iattr *);
3448 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3449 void setattr_copy(struct user_namespace *, struct inode *inode,
3450 const struct iattr *attr);
3451
3452 extern int file_update_time(struct file *file);
3453
vma_is_dax(const struct vm_area_struct * vma)3454 static inline bool vma_is_dax(const struct vm_area_struct *vma)
3455 {
3456 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3457 }
3458
vma_is_fsdax(struct vm_area_struct * vma)3459 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3460 {
3461 struct inode *inode;
3462
3463 if (!IS_ENABLED(CONFIG_FS_DAX) || !vma->vm_file)
3464 return false;
3465 if (!vma_is_dax(vma))
3466 return false;
3467 inode = file_inode(vma->vm_file);
3468 if (S_ISCHR(inode->i_mode))
3469 return false; /* device-dax */
3470 return true;
3471 }
3472
iocb_flags(struct file * file)3473 static inline int iocb_flags(struct file *file)
3474 {
3475 int res = 0;
3476 if (file->f_flags & O_APPEND)
3477 res |= IOCB_APPEND;
3478 if (file->f_flags & O_DIRECT)
3479 res |= IOCB_DIRECT;
3480 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3481 res |= IOCB_DSYNC;
3482 if (file->f_flags & __O_SYNC)
3483 res |= IOCB_SYNC;
3484 return res;
3485 }
3486
kiocb_set_rw_flags(struct kiocb * ki,rwf_t flags)3487 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3488 {
3489 int kiocb_flags = 0;
3490
3491 /* make sure there's no overlap between RWF and private IOCB flags */
3492 BUILD_BUG_ON((__force int) RWF_SUPPORTED & IOCB_EVENTFD);
3493
3494 if (!flags)
3495 return 0;
3496 if (unlikely(flags & ~RWF_SUPPORTED))
3497 return -EOPNOTSUPP;
3498
3499 if (flags & RWF_NOWAIT) {
3500 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3501 return -EOPNOTSUPP;
3502 kiocb_flags |= IOCB_NOIO;
3503 }
3504 kiocb_flags |= (__force int) (flags & RWF_SUPPORTED);
3505 if (flags & RWF_SYNC)
3506 kiocb_flags |= IOCB_DSYNC;
3507
3508 ki->ki_flags |= kiocb_flags;
3509 return 0;
3510 }
3511
parent_ino(struct dentry * dentry)3512 static inline ino_t parent_ino(struct dentry *dentry)
3513 {
3514 ino_t res;
3515
3516 /*
3517 * Don't strictly need d_lock here? If the parent ino could change
3518 * then surely we'd have a deeper race in the caller?
3519 */
3520 spin_lock(&dentry->d_lock);
3521 res = dentry->d_parent->d_inode->i_ino;
3522 spin_unlock(&dentry->d_lock);
3523 return res;
3524 }
3525
3526 /* Transaction based IO helpers */
3527
3528 /*
3529 * An argresp is stored in an allocated page and holds the
3530 * size of the argument or response, along with its content
3531 */
3532 struct simple_transaction_argresp {
3533 ssize_t size;
3534 char data[];
3535 };
3536
3537 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3538
3539 char *simple_transaction_get(struct file *file, const char __user *buf,
3540 size_t size);
3541 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3542 size_t size, loff_t *pos);
3543 int simple_transaction_release(struct inode *inode, struct file *file);
3544
3545 void simple_transaction_set(struct file *file, size_t n);
3546
3547 /*
3548 * simple attribute files
3549 *
3550 * These attributes behave similar to those in sysfs:
3551 *
3552 * Writing to an attribute immediately sets a value, an open file can be
3553 * written to multiple times.
3554 *
3555 * Reading from an attribute creates a buffer from the value that might get
3556 * read with multiple read calls. When the attribute has been read
3557 * completely, no further read calls are possible until the file is opened
3558 * again.
3559 *
3560 * All attributes contain a text representation of a numeric value
3561 * that are accessed with the get() and set() functions.
3562 */
3563 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3564 static int __fops ## _open(struct inode *inode, struct file *file) \
3565 { \
3566 __simple_attr_check_format(__fmt, 0ull); \
3567 return simple_attr_open(inode, file, __get, __set, __fmt); \
3568 } \
3569 static const struct file_operations __fops = { \
3570 .owner = THIS_MODULE, \
3571 .open = __fops ## _open, \
3572 .release = simple_attr_release, \
3573 .read = simple_attr_read, \
3574 .write = simple_attr_write, \
3575 .llseek = generic_file_llseek, \
3576 }
3577
3578 static inline __printf(1, 2)
__simple_attr_check_format(const char * fmt,...)3579 void __simple_attr_check_format(const char *fmt, ...)
3580 {
3581 /* don't do anything, just let the compiler check the arguments; */
3582 }
3583
3584 int simple_attr_open(struct inode *inode, struct file *file,
3585 int (*get)(void *, u64 *), int (*set)(void *, u64),
3586 const char *fmt);
3587 int simple_attr_release(struct inode *inode, struct file *file);
3588 ssize_t simple_attr_read(struct file *file, char __user *buf,
3589 size_t len, loff_t *ppos);
3590 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3591 size_t len, loff_t *ppos);
3592
3593 struct ctl_table;
3594 int proc_nr_files(struct ctl_table *table, int write,
3595 void *buffer, size_t *lenp, loff_t *ppos);
3596 int proc_nr_dentry(struct ctl_table *table, int write,
3597 void *buffer, size_t *lenp, loff_t *ppos);
3598 int proc_nr_inodes(struct ctl_table *table, int write,
3599 void *buffer, size_t *lenp, loff_t *ppos);
3600 int __init list_bdev_fs_names(char *buf, size_t size);
3601
3602 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3603 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3604
3605 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3606 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3607 (flag & __FMODE_NONOTIFY)))
3608
is_sxid(umode_t mode)3609 static inline bool is_sxid(umode_t mode)
3610 {
3611 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3612 }
3613
check_sticky(struct user_namespace * mnt_userns,struct inode * dir,struct inode * inode)3614 static inline int check_sticky(struct user_namespace *mnt_userns,
3615 struct inode *dir, struct inode *inode)
3616 {
3617 if (!(dir->i_mode & S_ISVTX))
3618 return 0;
3619
3620 return __check_sticky(mnt_userns, dir, inode);
3621 }
3622
inode_has_no_xattr(struct inode * inode)3623 static inline void inode_has_no_xattr(struct inode *inode)
3624 {
3625 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3626 inode->i_flags |= S_NOSEC;
3627 }
3628
is_root_inode(struct inode * inode)3629 static inline bool is_root_inode(struct inode *inode)
3630 {
3631 return inode == inode->i_sb->s_root->d_inode;
3632 }
3633
dir_emit(struct dir_context * ctx,const char * name,int namelen,u64 ino,unsigned type)3634 static inline bool dir_emit(struct dir_context *ctx,
3635 const char *name, int namelen,
3636 u64 ino, unsigned type)
3637 {
3638 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3639 }
dir_emit_dot(struct file * file,struct dir_context * ctx)3640 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3641 {
3642 return ctx->actor(ctx, ".", 1, ctx->pos,
3643 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3644 }
dir_emit_dotdot(struct file * file,struct dir_context * ctx)3645 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3646 {
3647 return ctx->actor(ctx, "..", 2, ctx->pos,
3648 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3649 }
dir_emit_dots(struct file * file,struct dir_context * ctx)3650 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3651 {
3652 if (ctx->pos == 0) {
3653 if (!dir_emit_dot(file, ctx))
3654 return false;
3655 ctx->pos = 1;
3656 }
3657 if (ctx->pos == 1) {
3658 if (!dir_emit_dotdot(file, ctx))
3659 return false;
3660 ctx->pos = 2;
3661 }
3662 return true;
3663 }
dir_relax(struct inode * inode)3664 static inline bool dir_relax(struct inode *inode)
3665 {
3666 inode_unlock(inode);
3667 inode_lock(inode);
3668 return !IS_DEADDIR(inode);
3669 }
3670
dir_relax_shared(struct inode * inode)3671 static inline bool dir_relax_shared(struct inode *inode)
3672 {
3673 inode_unlock_shared(inode);
3674 inode_lock_shared(inode);
3675 return !IS_DEADDIR(inode);
3676 }
3677
3678 extern bool path_noexec(const struct path *path);
3679 extern void inode_nohighmem(struct inode *inode);
3680
3681 /* mm/fadvise.c */
3682 extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len,
3683 int advice);
3684 extern int generic_fadvise(struct file *file, loff_t offset, loff_t len,
3685 int advice);
3686
3687 /*
3688 * Flush file data before changing attributes. Caller must hold any locks
3689 * required to prevent further writes to this file until we're done setting
3690 * flags.
3691 */
inode_drain_writes(struct inode * inode)3692 static inline int inode_drain_writes(struct inode *inode)
3693 {
3694 inode_dio_wait(inode);
3695 return filemap_write_and_wait(inode->i_mapping);
3696 }
3697
3698 #endif /* _LINUX_FS_H */
3699