1 /* SPDX-License-Identifier: GPL-2.0 */
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 #include <linux/mnt_idmapping.h>
45 #include <linux/slab.h>
47 #include <asm/byteorder.h>
48 #include <uapi/linux/fs.h>
50 struct backing_dev_info
;
54 struct export_operations
;
55 struct fiemap_extent_info
;
60 struct pipe_inode_info
;
61 struct poll_table_struct
;
63 struct vm_area_struct
;
66 struct swap_info_struct
;
68 struct workqueue_struct
;
70 struct fscrypt_inode_info
;
71 struct fscrypt_operations
;
73 struct fsverity_operations
;
75 struct fs_parameter_spec
;
79 extern void __init
inode_init(void);
80 extern void __init
inode_init_early(void);
81 extern void __init
files_init(void);
82 extern void __init
files_maxfiles_init(void);
84 extern unsigned long get_max_files(void);
85 extern unsigned int sysctl_nr_open
;
87 typedef __kernel_rwf_t rwf_t
;
90 typedef int (get_block_t
)(struct inode
*inode
, sector_t iblock
,
91 struct buffer_head
*bh_result
, int create
);
92 typedef int (dio_iodone_t
)(struct kiocb
*iocb
, loff_t offset
,
93 ssize_t bytes
, void *private);
95 #define MAY_EXEC 0x00000001
96 #define MAY_WRITE 0x00000002
97 #define MAY_READ 0x00000004
98 #define MAY_APPEND 0x00000008
99 #define MAY_ACCESS 0x00000010
100 #define MAY_OPEN 0x00000020
101 #define MAY_CHDIR 0x00000040
102 /* called from RCU mode, don't block */
103 #define MAY_NOT_BLOCK 0x00000080
106 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
107 * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
110 /* file is open for reading */
111 #define FMODE_READ ((__force fmode_t)0x1)
112 /* file is open for writing */
113 #define FMODE_WRITE ((__force fmode_t)0x2)
114 /* file is seekable */
115 #define FMODE_LSEEK ((__force fmode_t)0x4)
116 /* file can be accessed using pread */
117 #define FMODE_PREAD ((__force fmode_t)0x8)
118 /* file can be accessed using pwrite */
119 #define FMODE_PWRITE ((__force fmode_t)0x10)
120 /* File is opened for execution with sys_execve / sys_uselib */
121 #define FMODE_EXEC ((__force fmode_t)0x20)
122 /* 32bit hashes as llseek() offset (for directories) */
123 #define FMODE_32BITHASH ((__force fmode_t)0x200)
124 /* 64bit hashes as llseek() offset (for directories) */
125 #define FMODE_64BITHASH ((__force fmode_t)0x400)
128 * Don't update ctime and mtime.
130 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
131 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
133 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
135 /* Expect random access pattern */
136 #define FMODE_RANDOM ((__force fmode_t)0x1000)
138 /* File is huge (eg. /dev/mem): treat loff_t as unsigned */
139 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
141 /* File is opened with O_PATH; almost nothing can be done with it */
142 #define FMODE_PATH ((__force fmode_t)0x4000)
144 /* File needs atomic accesses to f_pos */
145 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
146 /* Write access to underlying fs */
147 #define FMODE_WRITER ((__force fmode_t)0x10000)
148 /* Has read method(s) */
149 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
150 /* Has write method(s) */
151 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
153 #define FMODE_OPENED ((__force fmode_t)0x80000)
154 #define FMODE_CREATED ((__force fmode_t)0x100000)
156 /* File is stream-like */
157 #define FMODE_STREAM ((__force fmode_t)0x200000)
159 /* File supports DIRECT IO */
160 #define FMODE_CAN_ODIRECT ((__force fmode_t)0x400000)
162 #define FMODE_NOREUSE ((__force fmode_t)0x800000)
164 /* File supports non-exclusive O_DIRECT writes from multiple threads */
165 #define FMODE_DIO_PARALLEL_WRITE ((__force fmode_t)0x1000000)
167 /* File is embedded in backing_file object */
168 #define FMODE_BACKING ((__force fmode_t)0x2000000)
170 /* File was opened by fanotify and shouldn't generate fanotify events */
171 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
173 /* File is capable of returning -EAGAIN if I/O will block */
174 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
176 /* File represents mount that needs unmounting */
177 #define FMODE_NEED_UNMOUNT ((__force fmode_t)0x10000000)
179 /* File does not contribute to nr_files count */
180 #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000)
182 /* File supports async buffered reads */
183 #define FMODE_BUF_RASYNC ((__force fmode_t)0x40000000)
185 /* File supports async nowait buffered writes */
186 #define FMODE_BUF_WASYNC ((__force fmode_t)0x80000000)
189 * Attribute flags. These should be or-ed together to figure out what
192 #define ATTR_MODE (1 << 0)
193 #define ATTR_UID (1 << 1)
194 #define ATTR_GID (1 << 2)
195 #define ATTR_SIZE (1 << 3)
196 #define ATTR_ATIME (1 << 4)
197 #define ATTR_MTIME (1 << 5)
198 #define ATTR_CTIME (1 << 6)
199 #define ATTR_ATIME_SET (1 << 7)
200 #define ATTR_MTIME_SET (1 << 8)
201 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
202 #define ATTR_KILL_SUID (1 << 11)
203 #define ATTR_KILL_SGID (1 << 12)
204 #define ATTR_FILE (1 << 13)
205 #define ATTR_KILL_PRIV (1 << 14)
206 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
207 #define ATTR_TIMES_SET (1 << 16)
208 #define ATTR_TOUCH (1 << 17)
211 * Whiteout is represented by a char device. The following constants define the
212 * mode and device number to use.
214 #define WHITEOUT_MODE 0
215 #define WHITEOUT_DEV 0
218 * This is the Inode Attributes structure, used for notify_change(). It
219 * uses the above definitions as flags, to know which values have changed.
220 * Also, in this manner, a Filesystem can look at only the values it cares
221 * about. Basically, these are the attributes that the VFS layer can
222 * request to change from the FS layer.
224 * Derek Atkins <warlord@MIT.EDU> 94-10-20
227 unsigned int ia_valid
;
230 * The two anonymous unions wrap structures with the same member.
232 * Filesystems raising FS_ALLOW_IDMAP need to use ia_vfs{g,u}id which
233 * are a dedicated type requiring the filesystem to use the dedicated
234 * helpers. Other filesystem can continue to use ia_{g,u}id until they
237 * They always contain the same value. In other words FS_ALLOW_IDMAP
238 * pass down the same value on idmapped mounts as they would on regular
250 struct timespec64 ia_atime
;
251 struct timespec64 ia_mtime
;
252 struct timespec64 ia_ctime
;
255 * Not an attribute, but an auxiliary info for filesystems wanting to
256 * implement an ftruncate() like method. NOTE: filesystem should
257 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
259 struct file
*ia_file
;
263 * Includes for diskquotas.
265 #include <linux/quota.h>
268 * Maximum number of layers of fs stack. Needs to be limited to
269 * prevent kernel stack overflow
271 #define FILESYSTEM_MAX_STACK_DEPTH 2
274 * enum positive_aop_returns - aop return codes with specific semantics
276 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
277 * completed, that the page is still locked, and
278 * should be considered active. The VM uses this hint
279 * to return the page to the active list -- it won't
280 * be a candidate for writeback again in the near
281 * future. Other callers must be careful to unlock
282 * the page if they get this return. Returned by
285 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
286 * unlocked it and the page might have been truncated.
287 * The caller should back up to acquiring a new page and
288 * trying again. The aop will be taking reasonable
289 * precautions not to livelock. If the caller held a page
290 * reference, it should drop it before retrying. Returned
293 * address_space_operation functions return these large constants to indicate
294 * special semantics to the caller. These are much larger than the bytes in a
295 * page to allow for functions that return the number of bytes operated on in a
299 enum positive_aop_returns
{
300 AOP_WRITEPAGE_ACTIVATE
= 0x80000,
301 AOP_TRUNCATED_PAGE
= 0x80001,
305 * oh the beauties of C type declarations.
308 struct address_space
;
309 struct writeback_control
;
310 struct readahead_control
;
313 * Write life time hint values.
314 * Stored in struct inode as u8.
317 WRITE_LIFE_NOT_SET
= 0,
318 WRITE_LIFE_NONE
= RWH_WRITE_LIFE_NONE
,
319 WRITE_LIFE_SHORT
= RWH_WRITE_LIFE_SHORT
,
320 WRITE_LIFE_MEDIUM
= RWH_WRITE_LIFE_MEDIUM
,
321 WRITE_LIFE_LONG
= RWH_WRITE_LIFE_LONG
,
322 WRITE_LIFE_EXTREME
= RWH_WRITE_LIFE_EXTREME
,
325 /* Match RWF_* bits to IOCB bits */
326 #define IOCB_HIPRI (__force int) RWF_HIPRI
327 #define IOCB_DSYNC (__force int) RWF_DSYNC
328 #define IOCB_SYNC (__force int) RWF_SYNC
329 #define IOCB_NOWAIT (__force int) RWF_NOWAIT
330 #define IOCB_APPEND (__force int) RWF_APPEND
332 /* non-RWF related bits - start at 16 */
333 #define IOCB_EVENTFD (1 << 16)
334 #define IOCB_DIRECT (1 << 17)
335 #define IOCB_WRITE (1 << 18)
336 /* iocb->ki_waitq is valid */
337 #define IOCB_WAITQ (1 << 19)
338 #define IOCB_NOIO (1 << 20)
339 /* can use bio alloc cache */
340 #define IOCB_ALLOC_CACHE (1 << 21)
342 * IOCB_DIO_CALLER_COMP can be set by the iocb owner, to indicate that the
343 * iocb completion can be passed back to the owner for execution from a safe
344 * context rather than needing to be punted through a workqueue. If this
345 * flag is set, the bio completion handling may set iocb->dio_complete to a
346 * handler function and iocb->private to context information for that handler.
347 * The issuer should call the handler with that context information from task
348 * context to complete the processing of the iocb. Note that while this
349 * provides a task context for the dio_complete() callback, it should only be
350 * used on the completion side for non-IO generating completions. It's fine to
351 * call blocking functions from this callback, but they should not wait for
352 * unrelated IO (like cache flushing, new IO generation, etc).
354 #define IOCB_DIO_CALLER_COMP (1 << 22)
356 /* for use in trace events */
357 #define TRACE_IOCB_STRINGS \
358 { IOCB_HIPRI, "HIPRI" }, \
359 { IOCB_DSYNC, "DSYNC" }, \
360 { IOCB_SYNC, "SYNC" }, \
361 { IOCB_NOWAIT, "NOWAIT" }, \
362 { IOCB_APPEND, "APPEND" }, \
363 { IOCB_EVENTFD, "EVENTFD"}, \
364 { IOCB_DIRECT, "DIRECT" }, \
365 { IOCB_WRITE, "WRITE" }, \
366 { IOCB_WAITQ, "WAITQ" }, \
367 { IOCB_NOIO, "NOIO" }, \
368 { IOCB_ALLOC_CACHE, "ALLOC_CACHE" }, \
369 { IOCB_DIO_CALLER_COMP, "CALLER_COMP" }
372 struct file
*ki_filp
;
374 void (*ki_complete
)(struct kiocb
*iocb
, long ret
);
377 u16 ki_ioprio
; /* See linux/ioprio.h */
380 * Only used for async buffered reads, where it denotes the
381 * page waitqueue associated with completing the read. Valid
382 * IFF IOCB_WAITQ is set.
384 struct wait_page_queue
*ki_waitq
;
386 * Can be used for O_DIRECT IO, where the completion handling
387 * is punted back to the issuer of the IO. May only be set
388 * if IOCB_DIO_CALLER_COMP is set by the issuer, and the issuer
389 * must then check for presence of this handler when ki_complete
390 * is invoked. The data passed in to this handler must be
391 * assigned to ->private when dio_complete is assigned.
393 ssize_t (*dio_complete
)(void *data
);
397 static inline bool is_sync_kiocb(struct kiocb
*kiocb
)
399 return kiocb
->ki_complete
== NULL
;
402 struct address_space_operations
{
403 int (*writepage
)(struct page
*page
, struct writeback_control
*wbc
);
404 int (*read_folio
)(struct file
*, struct folio
*);
406 /* Write back some dirty pages from this mapping. */
407 int (*writepages
)(struct address_space
*, struct writeback_control
*);
409 /* Mark a folio dirty. Return true if this dirtied it */
410 bool (*dirty_folio
)(struct address_space
*, struct folio
*);
412 void (*readahead
)(struct readahead_control
*);
414 int (*write_begin
)(struct file
*, struct address_space
*mapping
,
415 loff_t pos
, unsigned len
,
416 struct page
**pagep
, void **fsdata
);
417 int (*write_end
)(struct file
*, struct address_space
*mapping
,
418 loff_t pos
, unsigned len
, unsigned copied
,
419 struct page
*page
, void *fsdata
);
421 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
422 sector_t (*bmap
)(struct address_space
*, sector_t
);
423 void (*invalidate_folio
) (struct folio
*, size_t offset
, size_t len
);
424 bool (*release_folio
)(struct folio
*, gfp_t
);
425 void (*free_folio
)(struct folio
*folio
);
426 ssize_t (*direct_IO
)(struct kiocb
*, struct iov_iter
*iter
);
428 * migrate the contents of a folio to the specified target. If
429 * migrate_mode is MIGRATE_ASYNC, it must not block.
431 int (*migrate_folio
)(struct address_space
*, struct folio
*dst
,
432 struct folio
*src
, enum migrate_mode
);
433 int (*launder_folio
)(struct folio
*);
434 bool (*is_partially_uptodate
) (struct folio
*, size_t from
,
436 void (*is_dirty_writeback
) (struct folio
*, bool *dirty
, bool *wb
);
437 int (*error_remove_page
)(struct address_space
*, struct page
*);
439 /* swapfile support */
440 int (*swap_activate
)(struct swap_info_struct
*sis
, struct file
*file
,
442 void (*swap_deactivate
)(struct file
*file
);
443 int (*swap_rw
)(struct kiocb
*iocb
, struct iov_iter
*iter
);
446 extern const struct address_space_operations empty_aops
;
449 * struct address_space - Contents of a cacheable, mappable object.
450 * @host: Owner, either the inode or the block_device.
451 * @i_pages: Cached pages.
452 * @invalidate_lock: Guards coherency between page cache contents and
453 * file offset->disk block mappings in the filesystem during invalidates.
454 * It is also used to block modification of page cache contents through
456 * @gfp_mask: Memory allocation flags to use for allocating pages.
457 * @i_mmap_writable: Number of VM_SHARED, VM_MAYWRITE mappings.
458 * @nr_thps: Number of THPs in the pagecache (non-shmem only).
459 * @i_mmap: Tree of private and shared mappings.
460 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
461 * @nrpages: Number of page entries, protected by the i_pages lock.
462 * @writeback_index: Writeback starts here.
464 * @flags: Error bits and flags (AS_*).
465 * @wb_err: The most recent error which has occurred.
466 * @i_private_lock: For use by the owner of the address_space.
467 * @i_private_list: For use by the owner of the address_space.
468 * @i_private_data: For use by the owner of the address_space.
470 struct address_space
{
472 struct xarray i_pages
;
473 struct rw_semaphore invalidate_lock
;
475 atomic_t i_mmap_writable
;
476 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
477 /* number of thp, only for non-shmem files */
480 struct rb_root_cached i_mmap
;
481 unsigned long nrpages
;
482 pgoff_t writeback_index
;
483 const struct address_space_operations
*a_ops
;
485 struct rw_semaphore i_mmap_rwsem
;
487 spinlock_t i_private_lock
;
488 struct list_head i_private_list
;
489 void * i_private_data
;
490 } __attribute__((aligned(sizeof(long)))) __randomize_layout
;
492 * On most architectures that alignment is already the case; but
493 * must be enforced here for CRIS, to let the least significant bit
494 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
497 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
498 #define PAGECACHE_TAG_DIRTY XA_MARK_0
499 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
500 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
503 * Returns true if any of the pages in the mapping are marked with the tag.
505 static inline bool mapping_tagged(struct address_space
*mapping
, xa_mark_t tag
)
507 return xa_marked(&mapping
->i_pages
, tag
);
510 static inline void i_mmap_lock_write(struct address_space
*mapping
)
512 down_write(&mapping
->i_mmap_rwsem
);
515 static inline int i_mmap_trylock_write(struct address_space
*mapping
)
517 return down_write_trylock(&mapping
->i_mmap_rwsem
);
520 static inline void i_mmap_unlock_write(struct address_space
*mapping
)
522 up_write(&mapping
->i_mmap_rwsem
);
525 static inline int i_mmap_trylock_read(struct address_space
*mapping
)
527 return down_read_trylock(&mapping
->i_mmap_rwsem
);
530 static inline void i_mmap_lock_read(struct address_space
*mapping
)
532 down_read(&mapping
->i_mmap_rwsem
);
535 static inline void i_mmap_unlock_read(struct address_space
*mapping
)
537 up_read(&mapping
->i_mmap_rwsem
);
540 static inline void i_mmap_assert_locked(struct address_space
*mapping
)
542 lockdep_assert_held(&mapping
->i_mmap_rwsem
);
545 static inline void i_mmap_assert_write_locked(struct address_space
*mapping
)
547 lockdep_assert_held_write(&mapping
->i_mmap_rwsem
);
551 * Might pages of this file be mapped into userspace?
553 static inline int mapping_mapped(struct address_space
*mapping
)
555 return !RB_EMPTY_ROOT(&mapping
->i_mmap
.rb_root
);
559 * Might pages of this file have been modified in userspace?
560 * Note that i_mmap_writable counts all VM_SHARED, VM_MAYWRITE vmas: do_mmap
561 * marks vma as VM_SHARED if it is shared, and the file was opened for
562 * writing i.e. vma may be mprotected writable even if now readonly.
564 * If i_mmap_writable is negative, no new writable mappings are allowed. You
565 * can only deny writable mappings, if none exists right now.
567 static inline int mapping_writably_mapped(struct address_space
*mapping
)
569 return atomic_read(&mapping
->i_mmap_writable
) > 0;
572 static inline int mapping_map_writable(struct address_space
*mapping
)
574 return atomic_inc_unless_negative(&mapping
->i_mmap_writable
) ?
578 static inline void mapping_unmap_writable(struct address_space
*mapping
)
580 atomic_dec(&mapping
->i_mmap_writable
);
583 static inline int mapping_deny_writable(struct address_space
*mapping
)
585 return atomic_dec_unless_positive(&mapping
->i_mmap_writable
) ?
589 static inline void mapping_allow_writable(struct address_space
*mapping
)
591 atomic_inc(&mapping
->i_mmap_writable
);
595 * Use sequence counter to get consistent i_size on 32-bit processors.
597 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
598 #include <linux/seqlock.h>
599 #define __NEED_I_SIZE_ORDERED
600 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
602 #define i_size_ordered_init(inode) do { } while (0)
606 #define ACL_NOT_CACHED ((void *)(-1))
608 * ACL_DONT_CACHE is for stacked filesystems, that rely on underlying fs to
609 * cache the ACL. This also means that ->get_inode_acl() can be called in RCU
610 * mode with the LOOKUP_RCU flag.
612 #define ACL_DONT_CACHE ((void *)(-3))
614 static inline struct posix_acl
*
615 uncached_acl_sentinel(struct task_struct
*task
)
617 return (void *)task
+ 1;
621 is_uncached_acl(struct posix_acl
*acl
)
623 return (long)acl
& 1;
626 #define IOP_FASTPERM 0x0001
627 #define IOP_LOOKUP 0x0002
628 #define IOP_NOFOLLOW 0x0004
629 #define IOP_XATTR 0x0008
630 #define IOP_DEFAULT_READLINK 0x0010
632 struct fsnotify_mark_connector
;
635 * Keep mostly read-only and often accessed (especially for
636 * the RCU path lookup and 'stat' data) fields at the beginning
637 * of the 'struct inode'
641 unsigned short i_opflags
;
644 unsigned int i_flags
;
646 #ifdef CONFIG_FS_POSIX_ACL
647 struct posix_acl
*i_acl
;
648 struct posix_acl
*i_default_acl
;
651 const struct inode_operations
*i_op
;
652 struct super_block
*i_sb
;
653 struct address_space
*i_mapping
;
655 #ifdef CONFIG_SECURITY
659 /* Stat data, not accessed from path walking */
662 * Filesystems may only read i_nlink directly. They shall use the
663 * following functions for modification:
665 * (set|clear|inc|drop)_nlink
666 * inode_(inc|dec)_link_count
669 const unsigned int i_nlink
;
670 unsigned int __i_nlink
;
674 struct timespec64 __i_atime
;
675 struct timespec64 __i_mtime
;
676 struct timespec64 __i_ctime
; /* use inode_*_ctime accessors! */
677 spinlock_t i_lock
; /* i_blocks, i_bytes, maybe i_size */
678 unsigned short i_bytes
;
683 #ifdef __NEED_I_SIZE_ORDERED
684 seqcount_t i_size_seqcount
;
688 unsigned long i_state
;
689 struct rw_semaphore i_rwsem
;
691 unsigned long dirtied_when
; /* jiffies of first dirtying */
692 unsigned long dirtied_time_when
;
694 struct hlist_node i_hash
;
695 struct list_head i_io_list
; /* backing dev IO list */
696 #ifdef CONFIG_CGROUP_WRITEBACK
697 struct bdi_writeback
*i_wb
; /* the associated cgroup wb */
699 /* foreign inode detection, see wbc_detach_inode() */
701 u16 i_wb_frn_avg_time
;
702 u16 i_wb_frn_history
;
704 struct list_head i_lru
; /* inode LRU list */
705 struct list_head i_sb_list
;
706 struct list_head i_wb_list
; /* backing dev writeback list */
708 struct hlist_head i_dentry
;
709 struct rcu_head i_rcu
;
711 atomic64_t i_version
;
712 atomic64_t i_sequence
; /* see futex */
714 atomic_t i_dio_count
;
715 atomic_t i_writecount
;
716 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
717 atomic_t i_readcount
; /* struct files open RO */
720 const struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
721 void (*free_inode
)(struct inode
*);
723 struct file_lock_context
*i_flctx
;
724 struct address_space i_data
;
725 struct list_head i_devices
;
727 struct pipe_inode_info
*i_pipe
;
735 #ifdef CONFIG_FSNOTIFY
736 __u32 i_fsnotify_mask
; /* all events this inode cares about */
737 struct fsnotify_mark_connector __rcu
*i_fsnotify_marks
;
740 #ifdef CONFIG_FS_ENCRYPTION
741 struct fscrypt_inode_info
*i_crypt_info
;
744 #ifdef CONFIG_FS_VERITY
745 struct fsverity_info
*i_verity_info
;
748 void *i_private
; /* fs or device private pointer */
749 } __randomize_layout
;
751 struct timespec64
timestamp_truncate(struct timespec64 t
, struct inode
*inode
);
753 static inline unsigned int i_blocksize(const struct inode
*node
)
755 return (1 << node
->i_blkbits
);
758 static inline int inode_unhashed(struct inode
*inode
)
760 return hlist_unhashed(&inode
->i_hash
);
764 * __mark_inode_dirty expects inodes to be hashed. Since we don't
765 * want special inodes in the fileset inode space, we make them
766 * appear hashed, but do not put on any lists. hlist_del()
767 * will work fine and require no locking.
769 static inline void inode_fake_hash(struct inode
*inode
)
771 hlist_add_fake(&inode
->i_hash
);
775 * inode->i_mutex nesting subclasses for the lock validator:
777 * 0: the object of the current VFS operation
781 * 4: second non-directory
782 * 5: second parent (when locking independent directories in rename)
784 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
785 * non-directories at once.
787 * The locking order between these classes is
788 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
790 enum inode_i_mutex_lock_class
800 static inline void inode_lock(struct inode
*inode
)
802 down_write(&inode
->i_rwsem
);
805 static inline void inode_unlock(struct inode
*inode
)
807 up_write(&inode
->i_rwsem
);
810 static inline void inode_lock_shared(struct inode
*inode
)
812 down_read(&inode
->i_rwsem
);
815 static inline void inode_unlock_shared(struct inode
*inode
)
817 up_read(&inode
->i_rwsem
);
820 static inline int inode_trylock(struct inode
*inode
)
822 return down_write_trylock(&inode
->i_rwsem
);
825 static inline int inode_trylock_shared(struct inode
*inode
)
827 return down_read_trylock(&inode
->i_rwsem
);
830 static inline int inode_is_locked(struct inode
*inode
)
832 return rwsem_is_locked(&inode
->i_rwsem
);
835 static inline void inode_lock_nested(struct inode
*inode
, unsigned subclass
)
837 down_write_nested(&inode
->i_rwsem
, subclass
);
840 static inline void inode_lock_shared_nested(struct inode
*inode
, unsigned subclass
)
842 down_read_nested(&inode
->i_rwsem
, subclass
);
845 static inline void filemap_invalidate_lock(struct address_space
*mapping
)
847 down_write(&mapping
->invalidate_lock
);
850 static inline void filemap_invalidate_unlock(struct address_space
*mapping
)
852 up_write(&mapping
->invalidate_lock
);
855 static inline void filemap_invalidate_lock_shared(struct address_space
*mapping
)
857 down_read(&mapping
->invalidate_lock
);
860 static inline int filemap_invalidate_trylock_shared(
861 struct address_space
*mapping
)
863 return down_read_trylock(&mapping
->invalidate_lock
);
866 static inline void filemap_invalidate_unlock_shared(
867 struct address_space
*mapping
)
869 up_read(&mapping
->invalidate_lock
);
872 void lock_two_nondirectories(struct inode
*, struct inode
*);
873 void unlock_two_nondirectories(struct inode
*, struct inode
*);
875 void filemap_invalidate_lock_two(struct address_space
*mapping1
,
876 struct address_space
*mapping2
);
877 void filemap_invalidate_unlock_two(struct address_space
*mapping1
,
878 struct address_space
*mapping2
);
882 * NOTE: in a 32bit arch with a preemptable kernel and
883 * an UP compile the i_size_read/write must be atomic
884 * with respect to the local cpu (unlike with preempt disabled),
885 * but they don't need to be atomic with respect to other cpus like in
886 * true SMP (so they need either to either locally disable irq around
887 * the read or for example on x86 they can be still implemented as a
888 * cmpxchg8b without the need of the lock prefix). For SMP compiles
889 * and 64bit archs it makes no difference if preempt is enabled or not.
891 static inline loff_t
i_size_read(const struct inode
*inode
)
893 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
898 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
899 i_size
= inode
->i_size
;
900 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
902 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
906 i_size
= inode
->i_size
;
910 return inode
->i_size
;
915 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
916 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
917 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
919 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
921 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
923 write_seqcount_begin(&inode
->i_size_seqcount
);
924 inode
->i_size
= i_size
;
925 write_seqcount_end(&inode
->i_size_seqcount
);
927 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
929 inode
->i_size
= i_size
;
932 inode
->i_size
= i_size
;
936 static inline unsigned iminor(const struct inode
*inode
)
938 return MINOR(inode
->i_rdev
);
941 static inline unsigned imajor(const struct inode
*inode
)
943 return MAJOR(inode
->i_rdev
);
947 rwlock_t lock
; /* protects pid, uid, euid fields */
948 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
949 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
950 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
951 int signum
; /* posix.1b rt signal to be delivered on IO */
955 * struct file_ra_state - Track a file's readahead state.
956 * @start: Where the most recent readahead started.
957 * @size: Number of pages read in the most recent readahead.
958 * @async_size: Numer of pages that were/are not needed immediately
959 * and so were/are genuinely "ahead". Start next readahead when
960 * the first of these pages is accessed.
961 * @ra_pages: Maximum size of a readahead request, copied from the bdi.
962 * @mmap_miss: How many mmap accesses missed in the page cache.
963 * @prev_pos: The last byte in the most recent read request.
965 * When this structure is passed to ->readahead(), the "most recent"
966 * readahead means the current readahead.
968 struct file_ra_state
{
971 unsigned int async_size
;
972 unsigned int ra_pages
;
973 unsigned int mmap_miss
;
978 * Check if @index falls in the readahead windows.
980 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
982 return (index
>= ra
->start
&&
983 index
< ra
->start
+ ra
->size
);
987 * f_{lock,count,pos_lock} members can be highly contended and share
988 * the same cacheline. f_{lock,mode} are very frequently used together
989 * and so share the same cacheline as well. The read-mostly
990 * f_{path,inode,op} are kept on a separate cacheline.
994 /* fput() uses task work when closing and freeing file (default). */
995 struct callback_head f_task_work
;
996 /* fput() must use workqueue (most kernel threads). */
997 struct llist_node f_llist
;
998 unsigned int f_iocb_flags
;
1002 * Protects f_ep, f_flags.
1003 * Must not be taken from IRQ context.
1007 atomic_long_t f_count
;
1008 struct mutex f_pos_lock
;
1010 unsigned int f_flags
;
1011 struct fown_struct f_owner
;
1012 const struct cred
*f_cred
;
1013 struct file_ra_state f_ra
;
1015 struct inode
*f_inode
; /* cached value */
1016 const struct file_operations
*f_op
;
1019 #ifdef CONFIG_SECURITY
1022 /* needed for tty driver, and maybe others */
1026 /* Used by fs/eventpoll.c to link all the hooks to this file */
1027 struct hlist_head
*f_ep
;
1028 #endif /* #ifdef CONFIG_EPOLL */
1029 struct address_space
*f_mapping
;
1031 errseq_t f_sb_err
; /* for syncfs */
1032 } __randomize_layout
1033 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
1035 struct file_handle
{
1038 /* file identifier */
1039 unsigned char f_handle
[];
1042 static inline struct file
*get_file(struct file
*f
)
1044 atomic_long_inc(&f
->f_count
);
1048 struct file
*get_file_rcu(struct file __rcu
**f
);
1049 struct file
*get_file_active(struct file
**f
);
1051 #define file_count(x) atomic_long_read(&(x)->f_count)
1053 #define MAX_NON_LFS ((1UL<<31) - 1)
1055 /* Page cache limit. The filesystems should put that into their s_maxbytes
1056 limits, otherwise bad things can happen in VM. */
1057 #if BITS_PER_LONG==32
1058 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
1059 #elif BITS_PER_LONG==64
1060 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
1063 /* legacy typedef, should eventually be removed */
1064 typedef void *fl_owner_t
;
1068 /* The following constant reflects the upper bound of the file/locking space */
1070 #define OFFSET_MAX type_max(loff_t)
1071 #define OFFT_OFFSET_MAX type_max(off_t)
1074 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
1076 static inline struct inode
*file_inode(const struct file
*f
)
1081 static inline struct dentry
*file_dentry(const struct file
*file
)
1083 return d_real(file
->f_path
.dentry
, file_inode(file
));
1086 struct fasync_struct
{
1090 struct fasync_struct
*fa_next
; /* singly linked list */
1091 struct file
*fa_file
;
1092 struct rcu_head fa_rcu
;
1095 #define FASYNC_MAGIC 0x4601
1097 /* SMP safe fasync helpers: */
1098 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1099 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1100 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1101 extern struct fasync_struct
*fasync_alloc(void);
1102 extern void fasync_free(struct fasync_struct
*);
1104 /* can be called from interrupts */
1105 extern void kill_fasync(struct fasync_struct
**, int, int);
1107 extern void __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1108 extern int f_setown(struct file
*filp
, int who
, int force
);
1109 extern void f_delown(struct file
*filp
);
1110 extern pid_t
f_getown(struct file
*filp
);
1111 extern int send_sigurg(struct fown_struct
*fown
);
1114 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1115 * represented in both.
1117 #define SB_RDONLY BIT(0) /* Mount read-only */
1118 #define SB_NOSUID BIT(1) /* Ignore suid and sgid bits */
1119 #define SB_NODEV BIT(2) /* Disallow access to device special files */
1120 #define SB_NOEXEC BIT(3) /* Disallow program execution */
1121 #define SB_SYNCHRONOUS BIT(4) /* Writes are synced at once */
1122 #define SB_MANDLOCK BIT(6) /* Allow mandatory locks on an FS */
1123 #define SB_DIRSYNC BIT(7) /* Directory modifications are synchronous */
1124 #define SB_NOATIME BIT(10) /* Do not update access times. */
1125 #define SB_NODIRATIME BIT(11) /* Do not update directory access times */
1126 #define SB_SILENT BIT(15)
1127 #define SB_POSIXACL BIT(16) /* Supports POSIX ACLs */
1128 #define SB_INLINECRYPT BIT(17) /* Use blk-crypto for encrypted files */
1129 #define SB_KERNMOUNT BIT(22) /* this is a kern_mount call */
1130 #define SB_I_VERSION BIT(23) /* Update inode I_version field */
1131 #define SB_LAZYTIME BIT(25) /* Update the on-disk [acm]times lazily */
1133 /* These sb flags are internal to the kernel */
1134 #define SB_DEAD BIT(21)
1135 #define SB_DYING BIT(24)
1136 #define SB_SUBMOUNT BIT(26)
1137 #define SB_FORCE BIT(27)
1138 #define SB_NOSEC BIT(28)
1139 #define SB_BORN BIT(29)
1140 #define SB_ACTIVE BIT(30)
1141 #define SB_NOUSER BIT(31)
1143 /* These flags relate to encoding and casefolding */
1144 #define SB_ENC_STRICT_MODE_FL (1 << 0)
1146 #define sb_has_strict_encoding(sb) \
1147 (sb->s_encoding_flags & SB_ENC_STRICT_MODE_FL)
1153 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1154 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1155 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1156 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1157 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1160 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1161 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1162 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1163 #define SB_I_STABLE_WRITES 0x00000008 /* don't modify blks until WB is done */
1165 /* sb->s_iflags to limit user namespace mounts */
1166 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1167 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1168 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1170 #define SB_I_SKIP_SYNC 0x00000100 /* Skip superblock at global sync */
1171 #define SB_I_PERSB_BDI 0x00000200 /* has a per-sb bdi */
1172 #define SB_I_TS_EXPIRY_WARNED 0x00000400 /* warned about timestamp range expiry */
1173 #define SB_I_RETIRED 0x00000800 /* superblock shouldn't be reused */
1174 #define SB_I_NOUMASK 0x00001000 /* VFS does not apply umask */
1176 /* Possible states of 'frozen' field */
1178 SB_UNFROZEN
= 0, /* FS is unfrozen */
1179 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1180 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1181 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1182 * internal threads if needed) */
1183 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1186 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1189 unsigned short frozen
; /* Is sb frozen? */
1190 unsigned short freeze_holders
; /* Who froze fs? */
1191 struct percpu_rw_semaphore rw_sem
[SB_FREEZE_LEVELS
];
1194 struct super_block
{
1195 struct list_head s_list
; /* Keep this first */
1196 dev_t s_dev
; /* search index; _not_ kdev_t */
1197 unsigned char s_blocksize_bits
;
1198 unsigned long s_blocksize
;
1199 loff_t s_maxbytes
; /* Max file size */
1200 struct file_system_type
*s_type
;
1201 const struct super_operations
*s_op
;
1202 const struct dquot_operations
*dq_op
;
1203 const struct quotactl_ops
*s_qcop
;
1204 const struct export_operations
*s_export_op
;
1205 unsigned long s_flags
;
1206 unsigned long s_iflags
; /* internal SB_I_* flags */
1207 unsigned long s_magic
;
1208 struct dentry
*s_root
;
1209 struct rw_semaphore s_umount
;
1212 #ifdef CONFIG_SECURITY
1215 const struct xattr_handler
* const *s_xattr
;
1216 #ifdef CONFIG_FS_ENCRYPTION
1217 const struct fscrypt_operations
*s_cop
;
1218 struct fscrypt_keyring
*s_master_keys
; /* master crypto keys in use */
1220 #ifdef CONFIG_FS_VERITY
1221 const struct fsverity_operations
*s_vop
;
1223 #if IS_ENABLED(CONFIG_UNICODE)
1224 struct unicode_map
*s_encoding
;
1225 __u16 s_encoding_flags
;
1227 struct hlist_bl_head s_roots
; /* alternate root dentries for NFS */
1228 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1229 struct block_device
*s_bdev
;
1230 struct bdev_handle
*s_bdev_handle
;
1231 struct backing_dev_info
*s_bdi
;
1232 struct mtd_info
*s_mtd
;
1233 struct hlist_node s_instances
;
1234 unsigned int s_quota_types
; /* Bitmask of supported quota types */
1235 struct quota_info s_dquot
; /* Diskquota specific options */
1237 struct sb_writers s_writers
;
1240 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1241 * s_fsnotify_marks together for cache efficiency. They are frequently
1242 * accessed and rarely modified.
1244 void *s_fs_info
; /* Filesystem private info */
1246 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1248 /* Time limits for c/m/atime in seconds */
1249 time64_t s_time_min
;
1250 time64_t s_time_max
;
1251 #ifdef CONFIG_FSNOTIFY
1252 __u32 s_fsnotify_mask
;
1253 struct fsnotify_mark_connector __rcu
*s_fsnotify_marks
;
1256 char s_id
[32]; /* Informational name */
1257 uuid_t s_uuid
; /* UUID */
1259 unsigned int s_max_links
;
1262 * The next field is for VFS *only*. No filesystems have any business
1263 * even looking at it. You had been warned.
1265 struct mutex s_vfs_rename_mutex
; /* Kludge */
1268 * Filesystem subtype. If non-empty the filesystem type field
1269 * in /proc/mounts will be "type.subtype"
1271 const char *s_subtype
;
1273 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1275 struct shrinker
*s_shrink
; /* per-sb shrinker handle */
1277 /* Number of inodes with nlink == 0 but still referenced */
1278 atomic_long_t s_remove_count
;
1281 * Number of inode/mount/sb objects that are being watched, note that
1282 * inodes objects are currently double-accounted.
1284 atomic_long_t s_fsnotify_connectors
;
1286 /* Read-only state of the superblock is being changed */
1287 int s_readonly_remount
;
1289 /* per-sb errseq_t for reporting writeback errors via syncfs */
1292 /* AIO completions deferred from interrupt context */
1293 struct workqueue_struct
*s_dio_done_wq
;
1294 struct hlist_head s_pins
;
1297 * Owning user namespace and default context in which to
1298 * interpret filesystem uids, gids, quotas, device nodes,
1299 * xattrs and security labels.
1301 struct user_namespace
*s_user_ns
;
1304 * The list_lru structure is essentially just a pointer to a table
1305 * of per-node lru lists, each of which has its own spinlock.
1306 * There is no need to put them into separate cachelines.
1308 struct list_lru s_dentry_lru
;
1309 struct list_lru s_inode_lru
;
1310 struct rcu_head rcu
;
1311 struct work_struct destroy_work
;
1313 struct mutex s_sync_lock
; /* sync serialisation lock */
1316 * Indicates how deep in a filesystem stack this SB is
1320 /* s_inode_list_lock protects s_inodes */
1321 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp
;
1322 struct list_head s_inodes
; /* all inodes */
1324 spinlock_t s_inode_wblist_lock
;
1325 struct list_head s_inodes_wb
; /* writeback inodes */
1326 } __randomize_layout
;
1328 static inline struct user_namespace
*i_user_ns(const struct inode
*inode
)
1330 return inode
->i_sb
->s_user_ns
;
1333 /* Helper functions so that in most cases filesystems will
1334 * not need to deal directly with kuid_t and kgid_t and can
1335 * instead deal with the raw numeric values that are stored
1336 * in the filesystem.
1338 static inline uid_t
i_uid_read(const struct inode
*inode
)
1340 return from_kuid(i_user_ns(inode
), inode
->i_uid
);
1343 static inline gid_t
i_gid_read(const struct inode
*inode
)
1345 return from_kgid(i_user_ns(inode
), inode
->i_gid
);
1348 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
1350 inode
->i_uid
= make_kuid(i_user_ns(inode
), uid
);
1353 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
1355 inode
->i_gid
= make_kgid(i_user_ns(inode
), gid
);
1359 * i_uid_into_vfsuid - map an inode's i_uid down according to an idmapping
1360 * @idmap: idmap of the mount the inode was found from
1361 * @inode: inode to map
1363 * Return: whe inode's i_uid mapped down according to @idmap.
1364 * If the inode's i_uid has no mapping INVALID_VFSUID is returned.
1366 static inline vfsuid_t
i_uid_into_vfsuid(struct mnt_idmap
*idmap
,
1367 const struct inode
*inode
)
1369 return make_vfsuid(idmap
, i_user_ns(inode
), inode
->i_uid
);
1373 * i_uid_needs_update - check whether inode's i_uid needs to be updated
1374 * @idmap: idmap of the mount the inode was found from
1375 * @attr: the new attributes of @inode
1376 * @inode: the inode to update
1378 * Check whether the $inode's i_uid field needs to be updated taking idmapped
1379 * mounts into account if the filesystem supports it.
1381 * Return: true if @inode's i_uid field needs to be updated, false if not.
1383 static inline bool i_uid_needs_update(struct mnt_idmap
*idmap
,
1384 const struct iattr
*attr
,
1385 const struct inode
*inode
)
1387 return ((attr
->ia_valid
& ATTR_UID
) &&
1388 !vfsuid_eq(attr
->ia_vfsuid
,
1389 i_uid_into_vfsuid(idmap
, inode
)));
1393 * i_uid_update - update @inode's i_uid field
1394 * @idmap: idmap of the mount the inode was found from
1395 * @attr: the new attributes of @inode
1396 * @inode: the inode to update
1398 * Safely update @inode's i_uid field translating the vfsuid of any idmapped
1399 * mount into the filesystem kuid.
1401 static inline void i_uid_update(struct mnt_idmap
*idmap
,
1402 const struct iattr
*attr
,
1403 struct inode
*inode
)
1405 if (attr
->ia_valid
& ATTR_UID
)
1406 inode
->i_uid
= from_vfsuid(idmap
, i_user_ns(inode
),
1411 * i_gid_into_vfsgid - map an inode's i_gid down according to an idmapping
1412 * @idmap: idmap of the mount the inode was found from
1413 * @inode: inode to map
1415 * Return: the inode's i_gid mapped down according to @idmap.
1416 * If the inode's i_gid has no mapping INVALID_VFSGID is returned.
1418 static inline vfsgid_t
i_gid_into_vfsgid(struct mnt_idmap
*idmap
,
1419 const struct inode
*inode
)
1421 return make_vfsgid(idmap
, i_user_ns(inode
), inode
->i_gid
);
1425 * i_gid_needs_update - check whether inode's i_gid needs to be updated
1426 * @idmap: idmap of the mount the inode was found from
1427 * @attr: the new attributes of @inode
1428 * @inode: the inode to update
1430 * Check whether the $inode's i_gid field needs to be updated taking idmapped
1431 * mounts into account if the filesystem supports it.
1433 * Return: true if @inode's i_gid field needs to be updated, false if not.
1435 static inline bool i_gid_needs_update(struct mnt_idmap
*idmap
,
1436 const struct iattr
*attr
,
1437 const struct inode
*inode
)
1439 return ((attr
->ia_valid
& ATTR_GID
) &&
1440 !vfsgid_eq(attr
->ia_vfsgid
,
1441 i_gid_into_vfsgid(idmap
, inode
)));
1445 * i_gid_update - update @inode's i_gid field
1446 * @idmap: idmap of the mount the inode was found from
1447 * @attr: the new attributes of @inode
1448 * @inode: the inode to update
1450 * Safely update @inode's i_gid field translating the vfsgid of any idmapped
1451 * mount into the filesystem kgid.
1453 static inline void i_gid_update(struct mnt_idmap
*idmap
,
1454 const struct iattr
*attr
,
1455 struct inode
*inode
)
1457 if (attr
->ia_valid
& ATTR_GID
)
1458 inode
->i_gid
= from_vfsgid(idmap
, i_user_ns(inode
),
1463 * inode_fsuid_set - initialize inode's i_uid field with callers fsuid
1464 * @inode: inode to initialize
1465 * @idmap: idmap of the mount the inode was found from
1467 * Initialize the i_uid field of @inode. If the inode was found/created via
1468 * an idmapped mount map the caller's fsuid according to @idmap.
1470 static inline void inode_fsuid_set(struct inode
*inode
,
1471 struct mnt_idmap
*idmap
)
1473 inode
->i_uid
= mapped_fsuid(idmap
, i_user_ns(inode
));
1477 * inode_fsgid_set - initialize inode's i_gid field with callers fsgid
1478 * @inode: inode to initialize
1479 * @idmap: idmap of the mount the inode was found from
1481 * Initialize the i_gid field of @inode. If the inode was found/created via
1482 * an idmapped mount map the caller's fsgid according to @idmap.
1484 static inline void inode_fsgid_set(struct inode
*inode
,
1485 struct mnt_idmap
*idmap
)
1487 inode
->i_gid
= mapped_fsgid(idmap
, i_user_ns(inode
));
1491 * fsuidgid_has_mapping() - check whether caller's fsuid/fsgid is mapped
1492 * @sb: the superblock we want a mapping in
1493 * @idmap: idmap of the relevant mount
1495 * Check whether the caller's fsuid and fsgid have a valid mapping in the
1496 * s_user_ns of the superblock @sb. If the caller is on an idmapped mount map
1497 * the caller's fsuid and fsgid according to the @idmap first.
1499 * Return: true if fsuid and fsgid is mapped, false if not.
1501 static inline bool fsuidgid_has_mapping(struct super_block
*sb
,
1502 struct mnt_idmap
*idmap
)
1504 struct user_namespace
*fs_userns
= sb
->s_user_ns
;
1508 kuid
= mapped_fsuid(idmap
, fs_userns
);
1509 if (!uid_valid(kuid
))
1511 kgid
= mapped_fsgid(idmap
, fs_userns
);
1512 if (!gid_valid(kgid
))
1514 return kuid_has_mapping(fs_userns
, kuid
) &&
1515 kgid_has_mapping(fs_userns
, kgid
);
1518 struct timespec64
current_time(struct inode
*inode
);
1519 struct timespec64
inode_set_ctime_current(struct inode
*inode
);
1521 static inline time64_t
inode_get_atime_sec(const struct inode
*inode
)
1523 return inode
->__i_atime
.tv_sec
;
1526 static inline long inode_get_atime_nsec(const struct inode
*inode
)
1528 return inode
->__i_atime
.tv_nsec
;
1531 static inline struct timespec64
inode_get_atime(const struct inode
*inode
)
1533 return inode
->__i_atime
;
1536 static inline struct timespec64
inode_set_atime_to_ts(struct inode
*inode
,
1537 struct timespec64 ts
)
1539 inode
->__i_atime
= ts
;
1543 static inline struct timespec64
inode_set_atime(struct inode
*inode
,
1544 time64_t sec
, long nsec
)
1546 struct timespec64 ts
= { .tv_sec
= sec
,
1548 return inode_set_atime_to_ts(inode
, ts
);
1551 static inline time64_t
inode_get_mtime_sec(const struct inode
*inode
)
1553 return inode
->__i_mtime
.tv_sec
;
1556 static inline long inode_get_mtime_nsec(const struct inode
*inode
)
1558 return inode
->__i_mtime
.tv_nsec
;
1561 static inline struct timespec64
inode_get_mtime(const struct inode
*inode
)
1563 return inode
->__i_mtime
;
1566 static inline struct timespec64
inode_set_mtime_to_ts(struct inode
*inode
,
1567 struct timespec64 ts
)
1569 inode
->__i_mtime
= ts
;
1573 static inline struct timespec64
inode_set_mtime(struct inode
*inode
,
1574 time64_t sec
, long nsec
)
1576 struct timespec64 ts
= { .tv_sec
= sec
,
1578 return inode_set_mtime_to_ts(inode
, ts
);
1581 static inline time64_t
inode_get_ctime_sec(const struct inode
*inode
)
1583 return inode
->__i_ctime
.tv_sec
;
1586 static inline long inode_get_ctime_nsec(const struct inode
*inode
)
1588 return inode
->__i_ctime
.tv_nsec
;
1591 static inline struct timespec64
inode_get_ctime(const struct inode
*inode
)
1593 return inode
->__i_ctime
;
1596 static inline struct timespec64
inode_set_ctime_to_ts(struct inode
*inode
,
1597 struct timespec64 ts
)
1599 inode
->__i_ctime
= ts
;
1604 * inode_set_ctime - set the ctime in the inode
1605 * @inode: inode in which to set the ctime
1606 * @sec: tv_sec value to set
1607 * @nsec: tv_nsec value to set
1609 * Set the ctime in @inode to { @sec, @nsec }
1611 static inline struct timespec64
inode_set_ctime(struct inode
*inode
,
1612 time64_t sec
, long nsec
)
1614 struct timespec64 ts
= { .tv_sec
= sec
,
1617 return inode_set_ctime_to_ts(inode
, ts
);
1620 struct timespec64
simple_inode_init_ts(struct inode
*inode
);
1623 * Snapshotting support.
1627 * These are internal functions, please use sb_start_{write,pagefault,intwrite}
1630 static inline void __sb_end_write(struct super_block
*sb
, int level
)
1632 percpu_up_read(sb
->s_writers
.rw_sem
+ level
-1);
1635 static inline void __sb_start_write(struct super_block
*sb
, int level
)
1637 percpu_down_read(sb
->s_writers
.rw_sem
+ level
- 1);
1640 static inline bool __sb_start_write_trylock(struct super_block
*sb
, int level
)
1642 return percpu_down_read_trylock(sb
->s_writers
.rw_sem
+ level
- 1);
1645 #define __sb_writers_acquired(sb, lev) \
1646 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1647 #define __sb_writers_release(sb, lev) \
1648 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1650 static inline bool sb_write_started(const struct super_block
*sb
)
1652 return lockdep_is_held_type(sb
->s_writers
.rw_sem
+ SB_FREEZE_WRITE
- 1, 1);
1656 * sb_end_write - drop write access to a superblock
1657 * @sb: the super we wrote to
1659 * Decrement number of writers to the filesystem. Wake up possible waiters
1660 * wanting to freeze the filesystem.
1662 static inline void sb_end_write(struct super_block
*sb
)
1664 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1668 * sb_end_pagefault - drop write access to a superblock from a page fault
1669 * @sb: the super we wrote to
1671 * Decrement number of processes handling write page fault to the filesystem.
1672 * Wake up possible waiters wanting to freeze the filesystem.
1674 static inline void sb_end_pagefault(struct super_block
*sb
)
1676 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1680 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1681 * @sb: the super we wrote to
1683 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1684 * waiters wanting to freeze the filesystem.
1686 static inline void sb_end_intwrite(struct super_block
*sb
)
1688 __sb_end_write(sb
, SB_FREEZE_FS
);
1692 * sb_start_write - get write access to a superblock
1693 * @sb: the super we write to
1695 * When a process wants to write data or metadata to a file system (i.e. dirty
1696 * a page or an inode), it should embed the operation in a sb_start_write() -
1697 * sb_end_write() pair to get exclusion against file system freezing. This
1698 * function increments number of writers preventing freezing. If the file
1699 * system is already frozen, the function waits until the file system is
1702 * Since freeze protection behaves as a lock, users have to preserve
1703 * ordering of freeze protection and other filesystem locks. Generally,
1704 * freeze protection should be the outermost lock. In particular, we have:
1707 * -> i_mutex (write path, truncate, directory ops, ...)
1708 * -> s_umount (freeze_super, thaw_super)
1710 static inline void sb_start_write(struct super_block
*sb
)
1712 __sb_start_write(sb
, SB_FREEZE_WRITE
);
1715 static inline bool sb_start_write_trylock(struct super_block
*sb
)
1717 return __sb_start_write_trylock(sb
, SB_FREEZE_WRITE
);
1721 * sb_start_pagefault - get write access to a superblock from a page fault
1722 * @sb: the super we write to
1724 * When a process starts handling write page fault, it should embed the
1725 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1726 * exclusion against file system freezing. This is needed since the page fault
1727 * is going to dirty a page. This function increments number of running page
1728 * faults preventing freezing. If the file system is already frozen, the
1729 * function waits until the file system is thawed.
1731 * Since page fault freeze protection behaves as a lock, users have to preserve
1732 * ordering of freeze protection and other filesystem locks. It is advised to
1733 * put sb_start_pagefault() close to mmap_lock in lock ordering. Page fault
1734 * handling code implies lock dependency:
1737 * -> sb_start_pagefault
1739 static inline void sb_start_pagefault(struct super_block
*sb
)
1741 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
);
1745 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1746 * @sb: the super we write to
1748 * This is the third level of protection against filesystem freezing. It is
1749 * free for use by a filesystem. The only requirement is that it must rank
1750 * below sb_start_pagefault.
1752 * For example filesystem can call sb_start_intwrite() when starting a
1753 * transaction which somewhat eases handling of freezing for internal sources
1754 * of filesystem changes (internal fs threads, discarding preallocation on file
1757 static inline void sb_start_intwrite(struct super_block
*sb
)
1759 __sb_start_write(sb
, SB_FREEZE_FS
);
1762 static inline bool sb_start_intwrite_trylock(struct super_block
*sb
)
1764 return __sb_start_write_trylock(sb
, SB_FREEZE_FS
);
1767 bool inode_owner_or_capable(struct mnt_idmap
*idmap
,
1768 const struct inode
*inode
);
1771 * VFS helper functions..
1773 int vfs_create(struct mnt_idmap
*, struct inode
*,
1774 struct dentry
*, umode_t
, bool);
1775 int vfs_mkdir(struct mnt_idmap
*, struct inode
*,
1776 struct dentry
*, umode_t
);
1777 int vfs_mknod(struct mnt_idmap
*, struct inode
*, struct dentry
*,
1779 int vfs_symlink(struct mnt_idmap
*, struct inode
*,
1780 struct dentry
*, const char *);
1781 int vfs_link(struct dentry
*, struct mnt_idmap
*, struct inode
*,
1782 struct dentry
*, struct inode
**);
1783 int vfs_rmdir(struct mnt_idmap
*, struct inode
*, struct dentry
*);
1784 int vfs_unlink(struct mnt_idmap
*, struct inode
*, struct dentry
*,
1788 * struct renamedata - contains all information required for renaming
1789 * @old_mnt_idmap: idmap of the old mount the inode was found from
1790 * @old_dir: parent of source
1791 * @old_dentry: source
1792 * @new_mnt_idmap: idmap of the new mount the inode was found from
1793 * @new_dir: parent of destination
1794 * @new_dentry: destination
1795 * @delegated_inode: returns an inode needing a delegation break
1796 * @flags: rename flags
1799 struct mnt_idmap
*old_mnt_idmap
;
1800 struct inode
*old_dir
;
1801 struct dentry
*old_dentry
;
1802 struct mnt_idmap
*new_mnt_idmap
;
1803 struct inode
*new_dir
;
1804 struct dentry
*new_dentry
;
1805 struct inode
**delegated_inode
;
1807 } __randomize_layout
;
1809 int vfs_rename(struct renamedata
*);
1811 static inline int vfs_whiteout(struct mnt_idmap
*idmap
,
1812 struct inode
*dir
, struct dentry
*dentry
)
1814 return vfs_mknod(idmap
, dir
, dentry
, S_IFCHR
| WHITEOUT_MODE
,
1818 struct file
*kernel_tmpfile_open(struct mnt_idmap
*idmap
,
1819 const struct path
*parentpath
,
1820 umode_t mode
, int open_flag
,
1821 const struct cred
*cred
);
1822 struct file
*kernel_file_open(const struct path
*path
, int flags
,
1823 struct inode
*inode
, const struct cred
*cred
);
1825 int vfs_mkobj(struct dentry
*, umode_t
,
1826 int (*f
)(struct dentry
*, umode_t
, void *),
1829 int vfs_fchown(struct file
*file
, uid_t user
, gid_t group
);
1830 int vfs_fchmod(struct file
*file
, umode_t mode
);
1831 int vfs_utimes(const struct path
*path
, struct timespec64
*times
);
1833 extern long vfs_ioctl(struct file
*file
, unsigned int cmd
, unsigned long arg
);
1835 #ifdef CONFIG_COMPAT
1836 extern long compat_ptr_ioctl(struct file
*file
, unsigned int cmd
,
1839 #define compat_ptr_ioctl NULL
1843 * VFS file helper functions.
1845 void inode_init_owner(struct mnt_idmap
*idmap
, struct inode
*inode
,
1846 const struct inode
*dir
, umode_t mode
);
1847 extern bool may_open_dev(const struct path
*path
);
1848 umode_t
mode_strip_sgid(struct mnt_idmap
*idmap
,
1849 const struct inode
*dir
, umode_t mode
);
1852 * This is the "filldir" function type, used by readdir() to let
1853 * the kernel specify what kind of dirent layout it wants to have.
1854 * This allows the kernel to read directories into kernel space or
1855 * to have different dirent layouts depending on the binary type.
1856 * Return 'true' to keep going and 'false' if there are no more entries.
1859 typedef bool (*filldir_t
)(struct dir_context
*, const char *, int, loff_t
, u64
,
1862 struct dir_context
{
1868 * These flags let !MMU mmap() govern direct device mapping vs immediate
1869 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1871 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1872 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1873 * NOMMU_MAP_READ: Can be mapped for reading
1874 * NOMMU_MAP_WRITE: Can be mapped for writing
1875 * NOMMU_MAP_EXEC: Can be mapped for execution
1877 #define NOMMU_MAP_COPY 0x00000001
1878 #define NOMMU_MAP_DIRECT 0x00000008
1879 #define NOMMU_MAP_READ VM_MAYREAD
1880 #define NOMMU_MAP_WRITE VM_MAYWRITE
1881 #define NOMMU_MAP_EXEC VM_MAYEXEC
1883 #define NOMMU_VMFLAGS \
1884 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1887 * These flags control the behavior of the remap_file_range function pointer.
1888 * If it is called with len == 0 that means "remap to end of source file".
1889 * See Documentation/filesystems/vfs.rst for more details about this call.
1891 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
1892 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
1894 #define REMAP_FILE_DEDUP (1 << 0)
1895 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
1898 * These flags signal that the caller is ok with altering various aspects of
1899 * the behavior of the remap operation. The changes must be made by the
1900 * implementation; the vfs remap helper functions can take advantage of them.
1901 * Flags in this category exist to preserve the quirky behavior of the hoisted
1902 * btrfs clone/dedupe ioctls.
1904 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
1907 * These flags control the behavior of vfs_copy_file_range().
1908 * They are not available to the user via syscall.
1910 * COPY_FILE_SPLICE: call splice direct instead of fs clone/copy ops
1912 #define COPY_FILE_SPLICE (1 << 0)
1915 struct io_uring_cmd
;
1918 struct file_operations
{
1919 struct module
*owner
;
1920 loff_t (*llseek
) (struct file
*, loff_t
, int);
1921 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
1922 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
1923 ssize_t (*read_iter
) (struct kiocb
*, struct iov_iter
*);
1924 ssize_t (*write_iter
) (struct kiocb
*, struct iov_iter
*);
1925 int (*iopoll
)(struct kiocb
*kiocb
, struct io_comp_batch
*,
1926 unsigned int flags
);
1927 int (*iterate_shared
) (struct file
*, struct dir_context
*);
1928 __poll_t (*poll
) (struct file
*, struct poll_table_struct
*);
1929 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
1930 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
1931 int (*mmap
) (struct file
*, struct vm_area_struct
*);
1932 unsigned long mmap_supported_flags
;
1933 int (*open
) (struct inode
*, struct file
*);
1934 int (*flush
) (struct file
*, fl_owner_t id
);
1935 int (*release
) (struct inode
*, struct file
*);
1936 int (*fsync
) (struct file
*, loff_t
, loff_t
, int datasync
);
1937 int (*fasync
) (int, struct file
*, int);
1938 int (*lock
) (struct file
*, int, struct file_lock
*);
1939 unsigned long (*get_unmapped_area
)(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1940 int (*check_flags
)(int);
1941 int (*flock
) (struct file
*, int, struct file_lock
*);
1942 ssize_t (*splice_write
)(struct pipe_inode_info
*, struct file
*, loff_t
*, size_t, unsigned int);
1943 ssize_t (*splice_read
)(struct file
*, loff_t
*, struct pipe_inode_info
*, size_t, unsigned int);
1944 void (*splice_eof
)(struct file
*file
);
1945 int (*setlease
)(struct file
*, int, struct file_lock
**, void **);
1946 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
1948 void (*show_fdinfo
)(struct seq_file
*m
, struct file
*f
);
1950 unsigned (*mmap_capabilities
)(struct file
*);
1952 ssize_t (*copy_file_range
)(struct file
*, loff_t
, struct file
*,
1953 loff_t
, size_t, unsigned int);
1954 loff_t (*remap_file_range
)(struct file
*file_in
, loff_t pos_in
,
1955 struct file
*file_out
, loff_t pos_out
,
1956 loff_t len
, unsigned int remap_flags
);
1957 int (*fadvise
)(struct file
*, loff_t
, loff_t
, int);
1958 int (*uring_cmd
)(struct io_uring_cmd
*ioucmd
, unsigned int issue_flags
);
1959 int (*uring_cmd_iopoll
)(struct io_uring_cmd
*, struct io_comp_batch
*,
1960 unsigned int poll_flags
);
1961 } __randomize_layout
;
1963 /* Wrap a directory iterator that needs exclusive inode access */
1964 int wrap_directory_iterator(struct file
*, struct dir_context
*,
1965 int (*) (struct file
*, struct dir_context
*));
1966 #define WRAP_DIR_ITER(x) \
1967 static int shared_##x(struct file *file , struct dir_context *ctx) \
1968 { return wrap_directory_iterator(file, ctx, x); }
1970 struct inode_operations
{
1971 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*, unsigned int);
1972 const char * (*get_link
) (struct dentry
*, struct inode
*, struct delayed_call
*);
1973 int (*permission
) (struct mnt_idmap
*, struct inode
*, int);
1974 struct posix_acl
* (*get_inode_acl
)(struct inode
*, int, bool);
1976 int (*readlink
) (struct dentry
*, char __user
*,int);
1978 int (*create
) (struct mnt_idmap
*, struct inode
*,struct dentry
*,
1980 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
1981 int (*unlink
) (struct inode
*,struct dentry
*);
1982 int (*symlink
) (struct mnt_idmap
*, struct inode
*,struct dentry
*,
1984 int (*mkdir
) (struct mnt_idmap
*, struct inode
*,struct dentry
*,
1986 int (*rmdir
) (struct inode
*,struct dentry
*);
1987 int (*mknod
) (struct mnt_idmap
*, struct inode
*,struct dentry
*,
1989 int (*rename
) (struct mnt_idmap
*, struct inode
*, struct dentry
*,
1990 struct inode
*, struct dentry
*, unsigned int);
1991 int (*setattr
) (struct mnt_idmap
*, struct dentry
*, struct iattr
*);
1992 int (*getattr
) (struct mnt_idmap
*, const struct path
*,
1993 struct kstat
*, u32
, unsigned int);
1994 ssize_t (*listxattr
) (struct dentry
*, char *, size_t);
1995 int (*fiemap
)(struct inode
*, struct fiemap_extent_info
*, u64 start
,
1997 int (*update_time
)(struct inode
*, int);
1998 int (*atomic_open
)(struct inode
*, struct dentry
*,
1999 struct file
*, unsigned open_flag
,
2000 umode_t create_mode
);
2001 int (*tmpfile
) (struct mnt_idmap
*, struct inode
*,
2002 struct file
*, umode_t
);
2003 struct posix_acl
*(*get_acl
)(struct mnt_idmap
*, struct dentry
*,
2005 int (*set_acl
)(struct mnt_idmap
*, struct dentry
*,
2006 struct posix_acl
*, int);
2007 int (*fileattr_set
)(struct mnt_idmap
*idmap
,
2008 struct dentry
*dentry
, struct fileattr
*fa
);
2009 int (*fileattr_get
)(struct dentry
*dentry
, struct fileattr
*fa
);
2010 struct offset_ctx
*(*get_offset_ctx
)(struct inode
*inode
);
2011 } ____cacheline_aligned
;
2013 static inline ssize_t
call_read_iter(struct file
*file
, struct kiocb
*kio
,
2014 struct iov_iter
*iter
)
2016 return file
->f_op
->read_iter(kio
, iter
);
2019 static inline ssize_t
call_write_iter(struct file
*file
, struct kiocb
*kio
,
2020 struct iov_iter
*iter
)
2022 return file
->f_op
->write_iter(kio
, iter
);
2025 static inline int call_mmap(struct file
*file
, struct vm_area_struct
*vma
)
2027 return file
->f_op
->mmap(file
, vma
);
2030 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
2031 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
2032 extern ssize_t
vfs_copy_file_range(struct file
*, loff_t
, struct file
*,
2033 loff_t
, size_t, unsigned int);
2034 extern ssize_t
generic_copy_file_range(struct file
*file_in
, loff_t pos_in
,
2035 struct file
*file_out
, loff_t pos_out
,
2036 size_t len
, unsigned int flags
);
2037 int __generic_remap_file_range_prep(struct file
*file_in
, loff_t pos_in
,
2038 struct file
*file_out
, loff_t pos_out
,
2039 loff_t
*len
, unsigned int remap_flags
,
2040 const struct iomap_ops
*dax_read_ops
);
2041 int generic_remap_file_range_prep(struct file
*file_in
, loff_t pos_in
,
2042 struct file
*file_out
, loff_t pos_out
,
2043 loff_t
*count
, unsigned int remap_flags
);
2044 extern loff_t
do_clone_file_range(struct file
*file_in
, loff_t pos_in
,
2045 struct file
*file_out
, loff_t pos_out
,
2046 loff_t len
, unsigned int remap_flags
);
2047 extern loff_t
vfs_clone_file_range(struct file
*file_in
, loff_t pos_in
,
2048 struct file
*file_out
, loff_t pos_out
,
2049 loff_t len
, unsigned int remap_flags
);
2050 extern int vfs_dedupe_file_range(struct file
*file
,
2051 struct file_dedupe_range
*same
);
2052 extern loff_t
vfs_dedupe_file_range_one(struct file
*src_file
, loff_t src_pos
,
2053 struct file
*dst_file
, loff_t dst_pos
,
2054 loff_t len
, unsigned int remap_flags
);
2056 enum freeze_holder
{
2057 FREEZE_HOLDER_KERNEL
= (1U << 0),
2058 FREEZE_HOLDER_USERSPACE
= (1U << 1),
2061 struct super_operations
{
2062 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
2063 void (*destroy_inode
)(struct inode
*);
2064 void (*free_inode
)(struct inode
*);
2066 void (*dirty_inode
) (struct inode
*, int flags
);
2067 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
2068 int (*drop_inode
) (struct inode
*);
2069 void (*evict_inode
) (struct inode
*);
2070 void (*put_super
) (struct super_block
*);
2071 int (*sync_fs
)(struct super_block
*sb
, int wait
);
2072 int (*freeze_super
) (struct super_block
*, enum freeze_holder who
);
2073 int (*freeze_fs
) (struct super_block
*);
2074 int (*thaw_super
) (struct super_block
*, enum freeze_holder who
);
2075 int (*unfreeze_fs
) (struct super_block
*);
2076 int (*statfs
) (struct dentry
*, struct kstatfs
*);
2077 int (*remount_fs
) (struct super_block
*, int *, char *);
2078 void (*umount_begin
) (struct super_block
*);
2080 int (*show_options
)(struct seq_file
*, struct dentry
*);
2081 int (*show_devname
)(struct seq_file
*, struct dentry
*);
2082 int (*show_path
)(struct seq_file
*, struct dentry
*);
2083 int (*show_stats
)(struct seq_file
*, struct dentry
*);
2085 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
2086 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
2087 struct dquot
**(*get_dquots
)(struct inode
*);
2089 long (*nr_cached_objects
)(struct super_block
*,
2090 struct shrink_control
*);
2091 long (*free_cached_objects
)(struct super_block
*,
2092 struct shrink_control
*);
2093 void (*shutdown
)(struct super_block
*sb
);
2097 * Inode flags - they have no relation to superblock flags now
2099 #define S_SYNC (1 << 0) /* Writes are synced at once */
2100 #define S_NOATIME (1 << 1) /* Do not update access times */
2101 #define S_APPEND (1 << 2) /* Append-only file */
2102 #define S_IMMUTABLE (1 << 3) /* Immutable file */
2103 #define S_DEAD (1 << 4) /* removed, but still open directory */
2104 #define S_NOQUOTA (1 << 5) /* Inode is not counted to quota */
2105 #define S_DIRSYNC (1 << 6) /* Directory modifications are synchronous */
2106 #define S_NOCMTIME (1 << 7) /* Do not update file c/mtime */
2107 #define S_SWAPFILE (1 << 8) /* Do not truncate: swapon got its bmaps */
2108 #define S_PRIVATE (1 << 9) /* Inode is fs-internal */
2109 #define S_IMA (1 << 10) /* Inode has an associated IMA struct */
2110 #define S_AUTOMOUNT (1 << 11) /* Automount/referral quasi-directory */
2111 #define S_NOSEC (1 << 12) /* no suid or xattr security attributes */
2112 #ifdef CONFIG_FS_DAX
2113 #define S_DAX (1 << 13) /* Direct Access, avoiding the page cache */
2115 #define S_DAX 0 /* Make all the DAX code disappear */
2117 #define S_ENCRYPTED (1 << 14) /* Encrypted file (using fs/crypto/) */
2118 #define S_CASEFOLD (1 << 15) /* Casefolded file */
2119 #define S_VERITY (1 << 16) /* Verity file (using fs/verity/) */
2120 #define S_KERNEL_FILE (1 << 17) /* File is in use by the kernel (eg. fs/cachefiles) */
2123 * Note that nosuid etc flags are inode-specific: setting some file-system
2124 * flags just means all the inodes inherit those flags by default. It might be
2125 * possible to override it selectively if you really wanted to with some
2126 * ioctl() that is not currently implemented.
2128 * Exception: SB_RDONLY is always applied to the entire file system.
2130 * Unfortunately, it is possible to change a filesystems flags with it mounted
2131 * with files in use. This means that all of the inodes will not have their
2132 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
2133 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
2135 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
2137 static inline bool sb_rdonly(const struct super_block
*sb
) { return sb
->s_flags
& SB_RDONLY
; }
2138 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
2139 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
2140 ((inode)->i_flags & S_SYNC))
2141 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
2142 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
2143 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
2144 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
2145 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
2147 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
2148 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
2149 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
2151 #ifdef CONFIG_FS_POSIX_ACL
2152 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
2154 #define IS_POSIXACL(inode) 0
2157 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
2158 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
2159 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
2160 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
2161 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
2162 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2163 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2164 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2165 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2166 #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD)
2167 #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY)
2169 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2170 (inode)->i_rdev == WHITEOUT_DEV)
2172 static inline bool HAS_UNMAPPED_ID(struct mnt_idmap
*idmap
,
2173 struct inode
*inode
)
2175 return !vfsuid_valid(i_uid_into_vfsuid(idmap
, inode
)) ||
2176 !vfsgid_valid(i_gid_into_vfsgid(idmap
, inode
));
2179 static inline void init_sync_kiocb(struct kiocb
*kiocb
, struct file
*filp
)
2181 *kiocb
= (struct kiocb
) {
2183 .ki_flags
= filp
->f_iocb_flags
,
2184 .ki_ioprio
= get_current_ioprio(),
2188 static inline void kiocb_clone(struct kiocb
*kiocb
, struct kiocb
*kiocb_src
,
2191 *kiocb
= (struct kiocb
) {
2193 .ki_flags
= kiocb_src
->ki_flags
,
2194 .ki_ioprio
= kiocb_src
->ki_ioprio
,
2195 .ki_pos
= kiocb_src
->ki_pos
,
2200 * Inode state bits. Protected by inode->i_lock
2202 * Four bits determine the dirty state of the inode: I_DIRTY_SYNC,
2203 * I_DIRTY_DATASYNC, I_DIRTY_PAGES, and I_DIRTY_TIME.
2205 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2206 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2207 * various stages of removing an inode.
2209 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2211 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2212 * fdatasync() (unless I_DIRTY_DATASYNC is also set).
2213 * Timestamp updates are the usual cause.
2214 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2215 * these changes separately from I_DIRTY_SYNC so that we
2216 * don't have to write inode on fdatasync() when only
2217 * e.g. the timestamps have changed.
2218 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2219 * I_DIRTY_TIME The inode itself has dirty timestamps, and the
2220 * lazytime mount option is enabled. We keep track of this
2221 * separately from I_DIRTY_SYNC in order to implement
2222 * lazytime. This gets cleared if I_DIRTY_INODE
2223 * (I_DIRTY_SYNC and/or I_DIRTY_DATASYNC) gets set. But
2224 * I_DIRTY_TIME can still be set if I_DIRTY_SYNC is already
2225 * in place because writeback might already be in progress
2226 * and we don't want to lose the time update
2227 * I_NEW Serves as both a mutex and completion notification.
2228 * New inodes set I_NEW. If two processes both create
2229 * the same inode, one of them will release its inode and
2230 * wait for I_NEW to be released before returning.
2231 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2232 * also cause waiting on I_NEW, without I_NEW actually
2233 * being set. find_inode() uses this to prevent returning
2234 * nearly-dead inodes.
2235 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2236 * is zero. I_FREEING must be set when I_WILL_FREE is
2238 * I_FREEING Set when inode is about to be freed but still has dirty
2239 * pages or buffers attached or the inode itself is still
2241 * I_CLEAR Added by clear_inode(). In this state the inode is
2242 * clean and can be destroyed. Inode keeps I_FREEING.
2244 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2245 * prohibited for many purposes. iget() must wait for
2246 * the inode to be completely released, then create it
2247 * anew. Other functions will just ignore such inodes,
2248 * if appropriate. I_NEW is used for waiting.
2250 * I_SYNC Writeback of inode is running. The bit is set during
2251 * data writeback, and cleared with a wakeup on the bit
2252 * address once it is done. The bit is also used to pin
2253 * the inode in memory for flusher thread.
2255 * I_REFERENCED Marks the inode as recently references on the LRU list.
2257 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2259 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2260 * synchronize competing switching instances and to tell
2261 * wb stat updates to grab the i_pages lock. See
2262 * inode_switch_wbs_work_fn() for details.
2264 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2265 * and work dirs among overlayfs mounts.
2267 * I_CREATING New object's inode in the middle of setting up.
2269 * I_DONTCACHE Evict inode as soon as it is not used anymore.
2271 * I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists.
2272 * Used to detect that mark_inode_dirty() should not move
2273 * inode between dirty lists.
2275 * I_PINNING_FSCACHE_WB Inode is pinning an fscache object for writeback.
2277 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2279 #define I_DIRTY_SYNC (1 << 0)
2280 #define I_DIRTY_DATASYNC (1 << 1)
2281 #define I_DIRTY_PAGES (1 << 2)
2283 #define I_NEW (1 << __I_NEW)
2284 #define I_WILL_FREE (1 << 4)
2285 #define I_FREEING (1 << 5)
2286 #define I_CLEAR (1 << 6)
2288 #define I_SYNC (1 << __I_SYNC)
2289 #define I_REFERENCED (1 << 8)
2290 #define __I_DIO_WAKEUP 9
2291 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2292 #define I_LINKABLE (1 << 10)
2293 #define I_DIRTY_TIME (1 << 11)
2294 #define I_WB_SWITCH (1 << 13)
2295 #define I_OVL_INUSE (1 << 14)
2296 #define I_CREATING (1 << 15)
2297 #define I_DONTCACHE (1 << 16)
2298 #define I_SYNC_QUEUED (1 << 17)
2299 #define I_PINNING_FSCACHE_WB (1 << 18)
2301 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2302 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2303 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2305 extern void __mark_inode_dirty(struct inode
*, int);
2306 static inline void mark_inode_dirty(struct inode
*inode
)
2308 __mark_inode_dirty(inode
, I_DIRTY
);
2311 static inline void mark_inode_dirty_sync(struct inode
*inode
)
2313 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
2317 * Returns true if the given inode itself only has dirty timestamps (its pages
2318 * may still be dirty) and isn't currently being allocated or freed.
2319 * Filesystems should call this if when writing an inode when lazytime is
2320 * enabled, they want to opportunistically write the timestamps of other inodes
2321 * located very nearby on-disk, e.g. in the same inode block. This returns true
2322 * if the given inode is in need of such an opportunistic update. Requires
2323 * i_lock, or at least later re-checking under i_lock.
2325 static inline bool inode_is_dirtytime_only(struct inode
*inode
)
2327 return (inode
->i_state
& (I_DIRTY_TIME
| I_NEW
|
2328 I_FREEING
| I_WILL_FREE
)) == I_DIRTY_TIME
;
2331 extern void inc_nlink(struct inode
*inode
);
2332 extern void drop_nlink(struct inode
*inode
);
2333 extern void clear_nlink(struct inode
*inode
);
2334 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
2336 static inline void inode_inc_link_count(struct inode
*inode
)
2339 mark_inode_dirty(inode
);
2342 static inline void inode_dec_link_count(struct inode
*inode
)
2345 mark_inode_dirty(inode
);
2348 enum file_time_flags
{
2355 extern bool atime_needs_update(const struct path
*, struct inode
*);
2356 extern void touch_atime(const struct path
*);
2357 int inode_update_time(struct inode
*inode
, int flags
);
2359 static inline void file_accessed(struct file
*file
)
2361 if (!(file
->f_flags
& O_NOATIME
))
2362 touch_atime(&file
->f_path
);
2365 extern int file_modified(struct file
*file
);
2366 int kiocb_modified(struct kiocb
*iocb
);
2368 int sync_inode_metadata(struct inode
*inode
, int wait
);
2370 struct file_system_type
{
2373 #define FS_REQUIRES_DEV 1
2374 #define FS_BINARY_MOUNTDATA 2
2375 #define FS_HAS_SUBTYPE 4
2376 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2377 #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */
2378 #define FS_ALLOW_IDMAP 32 /* FS has been updated to handle vfs idmappings. */
2379 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2380 int (*init_fs_context
)(struct fs_context
*);
2381 const struct fs_parameter_spec
*parameters
;
2382 struct dentry
*(*mount
) (struct file_system_type
*, int,
2383 const char *, void *);
2384 void (*kill_sb
) (struct super_block
*);
2385 struct module
*owner
;
2386 struct file_system_type
* next
;
2387 struct hlist_head fs_supers
;
2389 struct lock_class_key s_lock_key
;
2390 struct lock_class_key s_umount_key
;
2391 struct lock_class_key s_vfs_rename_key
;
2392 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
2394 struct lock_class_key i_lock_key
;
2395 struct lock_class_key i_mutex_key
;
2396 struct lock_class_key invalidate_lock_key
;
2397 struct lock_class_key i_mutex_dir_key
;
2400 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2402 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
2403 int flags
, const char *dev_name
, void *data
,
2404 int (*fill_super
)(struct super_block
*, void *, int));
2405 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
2406 int flags
, void *data
,
2407 int (*fill_super
)(struct super_block
*, void *, int));
2408 extern struct dentry
*mount_nodev(struct file_system_type
*fs_type
,
2409 int flags
, void *data
,
2410 int (*fill_super
)(struct super_block
*, void *, int));
2411 extern struct dentry
*mount_subtree(struct vfsmount
*mnt
, const char *path
);
2412 void retire_super(struct super_block
*sb
);
2413 void generic_shutdown_super(struct super_block
*sb
);
2414 void kill_block_super(struct super_block
*sb
);
2415 void kill_anon_super(struct super_block
*sb
);
2416 void kill_litter_super(struct super_block
*sb
);
2417 void deactivate_super(struct super_block
*sb
);
2418 void deactivate_locked_super(struct super_block
*sb
);
2419 int set_anon_super(struct super_block
*s
, void *data
);
2420 int set_anon_super_fc(struct super_block
*s
, struct fs_context
*fc
);
2421 int get_anon_bdev(dev_t
*);
2422 void free_anon_bdev(dev_t
);
2423 struct super_block
*sget_fc(struct fs_context
*fc
,
2424 int (*test
)(struct super_block
*, struct fs_context
*),
2425 int (*set
)(struct super_block
*, struct fs_context
*));
2426 struct super_block
*sget(struct file_system_type
*type
,
2427 int (*test
)(struct super_block
*,void *),
2428 int (*set
)(struct super_block
*,void *),
2429 int flags
, void *data
);
2430 struct super_block
*sget_dev(struct fs_context
*fc
, dev_t dev
);
2432 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2433 #define fops_get(fops) \
2434 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2435 #define fops_put(fops) \
2436 do { if (fops) module_put((fops)->owner); } while(0)
2438 * This one is to be used *ONLY* from ->open() instances.
2439 * fops must be non-NULL, pinned down *and* module dependencies
2440 * should be sufficient to pin the caller down as well.
2442 #define replace_fops(f, fops) \
2444 struct file *__file = (f); \
2445 fops_put(__file->f_op); \
2446 BUG_ON(!(__file->f_op = (fops))); \
2449 extern int register_filesystem(struct file_system_type
*);
2450 extern int unregister_filesystem(struct file_system_type
*);
2451 extern int vfs_statfs(const struct path
*, struct kstatfs
*);
2452 extern int user_statfs(const char __user
*, struct kstatfs
*);
2453 extern int fd_statfs(int, struct kstatfs
*);
2454 int freeze_super(struct super_block
*super
, enum freeze_holder who
);
2455 int thaw_super(struct super_block
*super
, enum freeze_holder who
);
2456 extern __printf(2, 3)
2457 int super_setup_bdi_name(struct super_block
*sb
, char *fmt
, ...);
2458 extern int super_setup_bdi(struct super_block
*sb
);
2460 extern int current_umask(void);
2462 extern void ihold(struct inode
* inode
);
2463 extern void iput(struct inode
*);
2464 int inode_update_timestamps(struct inode
*inode
, int flags
);
2465 int generic_update_time(struct inode
*, int);
2468 extern struct kobject
*fs_kobj
;
2470 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2475 const char *name
; /* pointer to actual string */
2476 const __user
char *uptr
; /* original userland pointer */
2478 struct audit_names
*aname
;
2481 static_assert(offsetof(struct filename
, iname
) % sizeof(long) == 0);
2483 static inline struct mnt_idmap
*file_mnt_idmap(const struct file
*file
)
2485 return mnt_idmap(file
->f_path
.mnt
);
2489 * is_idmapped_mnt - check whether a mount is mapped
2490 * @mnt: the mount to check
2492 * If @mnt has an non @nop_mnt_idmap attached to it then @mnt is mapped.
2494 * Return: true if mount is mapped, false if not.
2496 static inline bool is_idmapped_mnt(const struct vfsmount
*mnt
)
2498 return mnt_idmap(mnt
) != &nop_mnt_idmap
;
2501 extern long vfs_truncate(const struct path
*, loff_t
);
2502 int do_truncate(struct mnt_idmap
*, struct dentry
*, loff_t start
,
2503 unsigned int time_attrs
, struct file
*filp
);
2504 extern int vfs_fallocate(struct file
*file
, int mode
, loff_t offset
,
2506 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
2508 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
2509 extern struct file
*filp_open(const char *, int, umode_t
);
2510 extern struct file
*file_open_root(const struct path
*,
2511 const char *, int, umode_t
);
2512 static inline struct file
*file_open_root_mnt(struct vfsmount
*mnt
,
2513 const char *name
, int flags
, umode_t mode
)
2515 return file_open_root(&(struct path
){.mnt
= mnt
, .dentry
= mnt
->mnt_root
},
2518 struct file
*dentry_open(const struct path
*path
, int flags
,
2519 const struct cred
*creds
);
2520 struct file
*dentry_create(const struct path
*path
, int flags
, umode_t mode
,
2521 const struct cred
*cred
);
2522 struct file
*backing_file_open(const struct path
*user_path
, int flags
,
2523 const struct path
*real_path
,
2524 const struct cred
*cred
);
2525 struct path
*backing_file_user_path(struct file
*f
);
2528 * When mmapping a file on a stackable filesystem (e.g., overlayfs), the file
2529 * stored in ->vm_file is a backing file whose f_inode is on the underlying
2530 * filesystem. When the mapped file path and inode number are displayed to
2531 * user (e.g. via /proc/<pid>/maps), these helpers should be used to get the
2532 * path and inode number to display to the user, which is the path of the fd
2533 * that user has requested to map and the inode number that would be returned
2534 * by fstat() on that same fd.
2536 /* Get the path to display in /proc/<pid>/maps */
2537 static inline const struct path
*file_user_path(struct file
*f
)
2539 if (unlikely(f
->f_mode
& FMODE_BACKING
))
2540 return backing_file_user_path(f
);
2543 /* Get the inode whose inode number to display in /proc/<pid>/maps */
2544 static inline const struct inode
*file_user_inode(struct file
*f
)
2546 if (unlikely(f
->f_mode
& FMODE_BACKING
))
2547 return d_inode(backing_file_user_path(f
)->dentry
);
2548 return file_inode(f
);
2551 static inline struct file
*file_clone_open(struct file
*file
)
2553 return dentry_open(&file
->f_path
, file
->f_flags
, file
->f_cred
);
2555 extern int filp_close(struct file
*, fl_owner_t id
);
2557 extern struct filename
*getname_flags(const char __user
*, int, int *);
2558 extern struct filename
*getname_uflags(const char __user
*, int);
2559 extern struct filename
*getname(const char __user
*);
2560 extern struct filename
*getname_kernel(const char *);
2561 extern void putname(struct filename
*name
);
2563 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
2564 int (*open
)(struct inode
*, struct file
*));
2565 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
2567 /* Helper for the simple case when original dentry is used */
2568 static inline int finish_open_simple(struct file
*file
, int error
)
2573 return finish_open(file
, file
->f_path
.dentry
, NULL
);
2577 extern void __init
vfs_caches_init_early(void);
2578 extern void __init
vfs_caches_init(void);
2580 extern struct kmem_cache
*names_cachep
;
2582 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2583 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2585 extern struct super_block
*blockdev_superblock
;
2586 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2588 return IS_ENABLED(CONFIG_BLOCK
) && sb
== blockdev_superblock
;
2591 void emergency_thaw_all(void);
2592 extern int sync_filesystem(struct super_block
*);
2593 extern const struct file_operations def_blk_fops
;
2594 extern const struct file_operations def_chr_fops
;
2597 #define CHRDEV_MAJOR_MAX 512
2598 /* Marks the bottom of the first segment of free char majors */
2599 #define CHRDEV_MAJOR_DYN_END 234
2600 /* Marks the top and bottom of the second segment of free char majors */
2601 #define CHRDEV_MAJOR_DYN_EXT_START 511
2602 #define CHRDEV_MAJOR_DYN_EXT_END 384
2604 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
2605 extern int register_chrdev_region(dev_t
, unsigned, const char *);
2606 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
2607 unsigned int count
, const char *name
,
2608 const struct file_operations
*fops
);
2609 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
2610 unsigned int count
, const char *name
);
2611 extern void unregister_chrdev_region(dev_t
, unsigned);
2612 extern void chrdev_show(struct seq_file
*,off_t
);
2614 static inline int register_chrdev(unsigned int major
, const char *name
,
2615 const struct file_operations
*fops
)
2617 return __register_chrdev(major
, 0, 256, name
, fops
);
2620 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2622 __unregister_chrdev(major
, 0, 256, name
);
2625 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2627 /* Invalid inode operations -- fs/bad_inode.c */
2628 extern void make_bad_inode(struct inode
*);
2629 extern bool is_bad_inode(struct inode
*);
2631 extern int __must_check
file_fdatawait_range(struct file
*file
, loff_t lstart
,
2633 extern int __must_check
file_check_and_advance_wb_err(struct file
*file
);
2634 extern int __must_check
file_write_and_wait_range(struct file
*file
,
2635 loff_t start
, loff_t end
);
2637 static inline int file_write_and_wait(struct file
*file
)
2639 return file_write_and_wait_range(file
, 0, LLONG_MAX
);
2642 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
2644 extern int vfs_fsync(struct file
*file
, int datasync
);
2646 extern int sync_file_range(struct file
*file
, loff_t offset
, loff_t nbytes
,
2647 unsigned int flags
);
2649 static inline bool iocb_is_dsync(const struct kiocb
*iocb
)
2651 return (iocb
->ki_flags
& IOCB_DSYNC
) ||
2652 IS_SYNC(iocb
->ki_filp
->f_mapping
->host
);
2656 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2657 * to already be updated for the write, and will return either the amount
2658 * of bytes passed in, or an error if syncing the file failed.
2660 static inline ssize_t
generic_write_sync(struct kiocb
*iocb
, ssize_t count
)
2662 if (iocb_is_dsync(iocb
)) {
2663 int ret
= vfs_fsync_range(iocb
->ki_filp
,
2664 iocb
->ki_pos
- count
, iocb
->ki_pos
- 1,
2665 (iocb
->ki_flags
& IOCB_SYNC
) ? 0 : 1);
2673 extern void emergency_sync(void);
2674 extern void emergency_remount(void);
2677 extern int bmap(struct inode
*inode
, sector_t
*block
);
2679 static inline int bmap(struct inode
*inode
, sector_t
*block
)
2685 int notify_change(struct mnt_idmap
*, struct dentry
*,
2686 struct iattr
*, struct inode
**);
2687 int inode_permission(struct mnt_idmap
*, struct inode
*, int);
2688 int generic_permission(struct mnt_idmap
*, struct inode
*, int);
2689 static inline int file_permission(struct file
*file
, int mask
)
2691 return inode_permission(file_mnt_idmap(file
),
2692 file_inode(file
), mask
);
2694 static inline int path_permission(const struct path
*path
, int mask
)
2696 return inode_permission(mnt_idmap(path
->mnt
),
2697 d_inode(path
->dentry
), mask
);
2699 int __check_sticky(struct mnt_idmap
*idmap
, struct inode
*dir
,
2700 struct inode
*inode
);
2702 static inline bool execute_ok(struct inode
*inode
)
2704 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
2707 static inline bool inode_wrong_type(const struct inode
*inode
, umode_t mode
)
2709 return (inode
->i_mode
^ mode
) & S_IFMT
;
2713 * file_start_write - get write access to a superblock for regular file io
2714 * @file: the file we want to write to
2716 * This is a variant of sb_start_write() which is a noop on non-regualr file.
2717 * Should be matched with a call to file_end_write().
2719 static inline void file_start_write(struct file
*file
)
2721 if (!S_ISREG(file_inode(file
)->i_mode
))
2723 sb_start_write(file_inode(file
)->i_sb
);
2726 static inline bool file_start_write_trylock(struct file
*file
)
2728 if (!S_ISREG(file_inode(file
)->i_mode
))
2730 return sb_start_write_trylock(file_inode(file
)->i_sb
);
2734 * file_end_write - drop write access to a superblock of a regular file
2735 * @file: the file we wrote to
2737 * Should be matched with a call to file_start_write().
2739 static inline void file_end_write(struct file
*file
)
2741 if (!S_ISREG(file_inode(file
)->i_mode
))
2743 sb_end_write(file_inode(file
)->i_sb
);
2747 * kiocb_start_write - get write access to a superblock for async file io
2748 * @iocb: the io context we want to submit the write with
2750 * This is a variant of sb_start_write() for async io submission.
2751 * Should be matched with a call to kiocb_end_write().
2753 static inline void kiocb_start_write(struct kiocb
*iocb
)
2755 struct inode
*inode
= file_inode(iocb
->ki_filp
);
2757 sb_start_write(inode
->i_sb
);
2759 * Fool lockdep by telling it the lock got released so that it
2760 * doesn't complain about the held lock when we return to userspace.
2762 __sb_writers_release(inode
->i_sb
, SB_FREEZE_WRITE
);
2766 * kiocb_end_write - drop write access to a superblock after async file io
2767 * @iocb: the io context we sumbitted the write with
2769 * Should be matched with a call to kiocb_start_write().
2771 static inline void kiocb_end_write(struct kiocb
*iocb
)
2773 struct inode
*inode
= file_inode(iocb
->ki_filp
);
2776 * Tell lockdep we inherited freeze protection from submission thread.
2778 __sb_writers_acquired(inode
->i_sb
, SB_FREEZE_WRITE
);
2779 sb_end_write(inode
->i_sb
);
2783 * This is used for regular files where some users -- especially the
2784 * currently executed binary in a process, previously handled via
2785 * VM_DENYWRITE -- cannot handle concurrent write (and maybe mmap
2786 * read-write shared) accesses.
2788 * get_write_access() gets write permission for a file.
2789 * put_write_access() releases this write permission.
2790 * deny_write_access() denies write access to a file.
2791 * allow_write_access() re-enables write access to a file.
2793 * The i_writecount field of an inode can have the following values:
2794 * 0: no write access, no denied write access
2795 * < 0: (-i_writecount) users that denied write access to the file.
2796 * > 0: (i_writecount) users that have write access to the file.
2798 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2799 * except for the cases where we don't hold i_writecount yet. Then we need to
2800 * use {get,deny}_write_access() - these functions check the sign and refuse
2801 * to do the change if sign is wrong.
2803 static inline int get_write_access(struct inode
*inode
)
2805 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2807 static inline int deny_write_access(struct file
*file
)
2809 struct inode
*inode
= file_inode(file
);
2810 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2812 static inline void put_write_access(struct inode
* inode
)
2814 atomic_dec(&inode
->i_writecount
);
2816 static inline void allow_write_access(struct file
*file
)
2819 atomic_inc(&file_inode(file
)->i_writecount
);
2821 static inline bool inode_is_open_for_write(const struct inode
*inode
)
2823 return atomic_read(&inode
->i_writecount
) > 0;
2826 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
2827 static inline void i_readcount_dec(struct inode
*inode
)
2829 BUG_ON(atomic_dec_return(&inode
->i_readcount
) < 0);
2831 static inline void i_readcount_inc(struct inode
*inode
)
2833 atomic_inc(&inode
->i_readcount
);
2836 static inline void i_readcount_dec(struct inode
*inode
)
2840 static inline void i_readcount_inc(struct inode
*inode
)
2845 extern int do_pipe_flags(int *, int);
2847 extern ssize_t
kernel_read(struct file
*, void *, size_t, loff_t
*);
2848 ssize_t
__kernel_read(struct file
*file
, void *buf
, size_t count
, loff_t
*pos
);
2849 extern ssize_t
kernel_write(struct file
*, const void *, size_t, loff_t
*);
2850 extern ssize_t
__kernel_write(struct file
*, const void *, size_t, loff_t
*);
2851 extern struct file
* open_exec(const char *);
2853 /* fs/dcache.c -- generic fs support functions */
2854 extern bool is_subdir(struct dentry
*, struct dentry
*);
2855 extern bool path_is_under(const struct path
*, const struct path
*);
2857 extern char *file_path(struct file
*, char *, int);
2859 #include <linux/err.h>
2861 /* needed for stackable file system support */
2862 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int whence
);
2864 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int whence
);
2866 extern int inode_init_always(struct super_block
*, struct inode
*);
2867 extern void inode_init_once(struct inode
*);
2868 extern void address_space_init_once(struct address_space
*mapping
);
2869 extern struct inode
* igrab(struct inode
*);
2870 extern ino_t
iunique(struct super_block
*, ino_t
);
2871 extern int inode_needs_sync(struct inode
*inode
);
2872 extern int generic_delete_inode(struct inode
*inode
);
2873 static inline int generic_drop_inode(struct inode
*inode
)
2875 return !inode
->i_nlink
|| inode_unhashed(inode
);
2877 extern void d_mark_dontcache(struct inode
*inode
);
2879 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
2880 unsigned long hashval
, int (*test
)(struct inode
*, void *),
2882 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
2883 int (*test
)(struct inode
*, void *), void *data
);
2884 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
2886 extern struct inode
*inode_insert5(struct inode
*inode
, unsigned long hashval
,
2887 int (*test
)(struct inode
*, void *),
2888 int (*set
)(struct inode
*, void *),
2890 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
2891 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
2892 extern struct inode
*find_inode_nowait(struct super_block
*,
2894 int (*match
)(struct inode
*,
2895 unsigned long, void *),
2897 extern struct inode
*find_inode_rcu(struct super_block
*, unsigned long,
2898 int (*)(struct inode
*, void *), void *);
2899 extern struct inode
*find_inode_by_ino_rcu(struct super_block
*, unsigned long);
2900 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
2901 extern int insert_inode_locked(struct inode
*);
2902 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2903 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
2905 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
2907 extern void unlock_new_inode(struct inode
*);
2908 extern void discard_new_inode(struct inode
*);
2909 extern unsigned int get_next_ino(void);
2910 extern void evict_inodes(struct super_block
*sb
);
2911 void dump_mapping(const struct address_space
*);
2914 * Userspace may rely on the inode number being non-zero. For example, glibc
2915 * simply ignores files with zero i_ino in unlink() and other places.
2917 * As an additional complication, if userspace was compiled with
2918 * _FILE_OFFSET_BITS=32 on a 64-bit kernel we'll only end up reading out the
2919 * lower 32 bits, so we need to check that those aren't zero explicitly. With
2920 * _FILE_OFFSET_BITS=64, this may cause some harmless false-negatives, but
2921 * better safe than sorry.
2923 static inline bool is_zero_ino(ino_t ino
)
2925 return (u32
)ino
== 0;
2928 extern void __iget(struct inode
* inode
);
2929 extern void iget_failed(struct inode
*);
2930 extern void clear_inode(struct inode
*);
2931 extern void __destroy_inode(struct inode
*);
2932 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
2933 extern struct inode
*new_inode(struct super_block
*sb
);
2934 extern void free_inode_nonrcu(struct inode
*inode
);
2935 extern int setattr_should_drop_suidgid(struct mnt_idmap
*, struct inode
*);
2936 extern int file_remove_privs(struct file
*);
2937 int setattr_should_drop_sgid(struct mnt_idmap
*idmap
,
2938 const struct inode
*inode
);
2941 * This must be used for allocating filesystems specific inodes to set
2942 * up the inode reclaim context correctly.
2944 static inline void *
2945 alloc_inode_sb(struct super_block
*sb
, struct kmem_cache
*cache
, gfp_t gfp
)
2947 return kmem_cache_alloc_lru(cache
, &sb
->s_inode_lru
, gfp
);
2950 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
2951 static inline void insert_inode_hash(struct inode
*inode
)
2953 __insert_inode_hash(inode
, inode
->i_ino
);
2956 extern void __remove_inode_hash(struct inode
*);
2957 static inline void remove_inode_hash(struct inode
*inode
)
2959 if (!inode_unhashed(inode
) && !hlist_fake(&inode
->i_hash
))
2960 __remove_inode_hash(inode
);
2963 extern void inode_sb_list_add(struct inode
*inode
);
2964 extern void inode_add_lru(struct inode
*inode
);
2966 extern int sb_set_blocksize(struct super_block
*, int);
2967 extern int sb_min_blocksize(struct super_block
*, int);
2969 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
2970 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
2971 extern ssize_t
generic_write_checks(struct kiocb
*, struct iov_iter
*);
2972 int generic_write_checks_count(struct kiocb
*iocb
, loff_t
*count
);
2973 extern int generic_write_check_limits(struct file
*file
, loff_t pos
,
2975 extern int generic_file_rw_checks(struct file
*file_in
, struct file
*file_out
);
2976 ssize_t
filemap_read(struct kiocb
*iocb
, struct iov_iter
*to
,
2977 ssize_t already_read
);
2978 extern ssize_t
generic_file_read_iter(struct kiocb
*, struct iov_iter
*);
2979 extern ssize_t
__generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2980 extern ssize_t
generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2981 extern ssize_t
generic_file_direct_write(struct kiocb
*, struct iov_iter
*);
2982 ssize_t
generic_perform_write(struct kiocb
*, struct iov_iter
*);
2983 ssize_t
direct_write_fallback(struct kiocb
*iocb
, struct iov_iter
*iter
,
2984 ssize_t direct_written
, ssize_t buffered_written
);
2986 ssize_t
vfs_iter_read(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
2988 ssize_t
vfs_iter_write(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
2990 ssize_t
vfs_iocb_iter_read(struct file
*file
, struct kiocb
*iocb
,
2991 struct iov_iter
*iter
);
2992 ssize_t
vfs_iocb_iter_write(struct file
*file
, struct kiocb
*iocb
,
2993 struct iov_iter
*iter
);
2996 ssize_t
filemap_splice_read(struct file
*in
, loff_t
*ppos
,
2997 struct pipe_inode_info
*pipe
,
2998 size_t len
, unsigned int flags
);
2999 ssize_t
copy_splice_read(struct file
*in
, loff_t
*ppos
,
3000 struct pipe_inode_info
*pipe
,
3001 size_t len
, unsigned int flags
);
3002 extern ssize_t
iter_file_splice_write(struct pipe_inode_info
*,
3003 struct file
*, loff_t
*, size_t, unsigned int);
3004 extern long do_splice_direct(struct file
*in
, loff_t
*ppos
, struct file
*out
,
3005 loff_t
*opos
, size_t len
, unsigned int flags
);
3009 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
3010 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int whence
);
3011 #define no_llseek NULL
3012 extern loff_t
vfs_setpos(struct file
*file
, loff_t offset
, loff_t maxsize
);
3013 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int whence
);
3014 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
3015 int whence
, loff_t maxsize
, loff_t eof
);
3016 extern loff_t
fixed_size_llseek(struct file
*file
, loff_t offset
,
3017 int whence
, loff_t size
);
3018 extern loff_t
no_seek_end_llseek_size(struct file
*, loff_t
, int, loff_t
);
3019 extern loff_t
no_seek_end_llseek(struct file
*, loff_t
, int);
3020 int rw_verify_area(int, struct file
*, const loff_t
*, size_t);
3021 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
3022 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
3023 extern int stream_open(struct inode
* inode
, struct file
* filp
);
3026 typedef void (dio_submit_t
)(struct bio
*bio
, struct inode
*inode
,
3027 loff_t file_offset
);
3030 /* need locking between buffered and direct access */
3033 /* filesystem does not support filling holes */
3034 DIO_SKIP_HOLES
= 0x02,
3037 ssize_t
__blockdev_direct_IO(struct kiocb
*iocb
, struct inode
*inode
,
3038 struct block_device
*bdev
, struct iov_iter
*iter
,
3039 get_block_t get_block
,
3040 dio_iodone_t end_io
,
3043 static inline ssize_t
blockdev_direct_IO(struct kiocb
*iocb
,
3044 struct inode
*inode
,
3045 struct iov_iter
*iter
,
3046 get_block_t get_block
)
3048 return __blockdev_direct_IO(iocb
, inode
, inode
->i_sb
->s_bdev
, iter
,
3049 get_block
, NULL
, DIO_LOCKING
| DIO_SKIP_HOLES
);
3053 void inode_dio_wait(struct inode
*inode
);
3056 * inode_dio_begin - signal start of a direct I/O requests
3057 * @inode: inode the direct I/O happens on
3059 * This is called once we've finished processing a direct I/O request,
3060 * and is used to wake up callers waiting for direct I/O to be quiesced.
3062 static inline void inode_dio_begin(struct inode
*inode
)
3064 atomic_inc(&inode
->i_dio_count
);
3068 * inode_dio_end - signal finish of a direct I/O requests
3069 * @inode: inode the direct I/O happens on
3071 * This is called once we've finished processing a direct I/O request,
3072 * and is used to wake up callers waiting for direct I/O to be quiesced.
3074 static inline void inode_dio_end(struct inode
*inode
)
3076 if (atomic_dec_and_test(&inode
->i_dio_count
))
3077 wake_up_bit(&inode
->i_state
, __I_DIO_WAKEUP
);
3080 extern void inode_set_flags(struct inode
*inode
, unsigned int flags
,
3083 extern const struct file_operations generic_ro_fops
;
3085 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3087 extern int readlink_copy(char __user
*, int, const char *);
3088 extern int page_readlink(struct dentry
*, char __user
*, int);
3089 extern const char *page_get_link(struct dentry
*, struct inode
*,
3090 struct delayed_call
*);
3091 extern void page_put_link(void *);
3092 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
3093 extern const struct inode_operations page_symlink_inode_operations
;
3094 extern void kfree_link(void *);
3095 void generic_fillattr(struct mnt_idmap
*, u32
, struct inode
*, struct kstat
*);
3096 void generic_fill_statx_attr(struct inode
*inode
, struct kstat
*stat
);
3097 extern int vfs_getattr_nosec(const struct path
*, struct kstat
*, u32
, unsigned int);
3098 extern int vfs_getattr(const struct path
*, struct kstat
*, u32
, unsigned int);
3099 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
3100 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
3101 void __inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
3102 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
3103 static inline loff_t
__inode_get_bytes(struct inode
*inode
)
3105 return (((loff_t
)inode
->i_blocks
) << 9) + inode
->i_bytes
;
3107 loff_t
inode_get_bytes(struct inode
*inode
);
3108 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
3109 const char *simple_get_link(struct dentry
*, struct inode
*,
3110 struct delayed_call
*);
3111 extern const struct inode_operations simple_symlink_inode_operations
;
3113 extern int iterate_dir(struct file
*, struct dir_context
*);
3115 int vfs_fstatat(int dfd
, const char __user
*filename
, struct kstat
*stat
,
3117 int vfs_fstat(int fd
, struct kstat
*stat
);
3119 static inline int vfs_stat(const char __user
*filename
, struct kstat
*stat
)
3121 return vfs_fstatat(AT_FDCWD
, filename
, stat
, 0);
3123 static inline int vfs_lstat(const char __user
*name
, struct kstat
*stat
)
3125 return vfs_fstatat(AT_FDCWD
, name
, stat
, AT_SYMLINK_NOFOLLOW
);
3128 extern const char *vfs_get_link(struct dentry
*, struct delayed_call
*);
3129 extern int vfs_readlink(struct dentry
*, char __user
*, int);
3131 extern struct file_system_type
*get_filesystem(struct file_system_type
*fs
);
3132 extern void put_filesystem(struct file_system_type
*fs
);
3133 extern struct file_system_type
*get_fs_type(const char *name
);
3134 extern struct super_block
*get_active_super(struct block_device
*bdev
);
3135 extern void drop_super(struct super_block
*sb
);
3136 extern void drop_super_exclusive(struct super_block
*sb
);
3137 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
3138 extern void iterate_supers_type(struct file_system_type
*,
3139 void (*)(struct super_block
*, void *), void *);
3141 extern int dcache_dir_open(struct inode
*, struct file
*);
3142 extern int dcache_dir_close(struct inode
*, struct file
*);
3143 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
3144 extern int dcache_readdir(struct file
*, struct dir_context
*);
3145 extern int simple_setattr(struct mnt_idmap
*, struct dentry
*,
3147 extern int simple_getattr(struct mnt_idmap
*, const struct path
*,
3148 struct kstat
*, u32
, unsigned int);
3149 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
3150 extern int simple_open(struct inode
*inode
, struct file
*file
);
3151 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
3152 extern int simple_unlink(struct inode
*, struct dentry
*);
3153 extern int simple_rmdir(struct inode
*, struct dentry
*);
3154 void simple_rename_timestamp(struct inode
*old_dir
, struct dentry
*old_dentry
,
3155 struct inode
*new_dir
, struct dentry
*new_dentry
);
3156 extern int simple_rename_exchange(struct inode
*old_dir
, struct dentry
*old_dentry
,
3157 struct inode
*new_dir
, struct dentry
*new_dentry
);
3158 extern int simple_rename(struct mnt_idmap
*, struct inode
*,
3159 struct dentry
*, struct inode
*, struct dentry
*,
3161 extern void simple_recursive_removal(struct dentry
*,
3162 void (*callback
)(struct dentry
*));
3163 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
3164 extern ssize_t
noop_direct_IO(struct kiocb
*iocb
, struct iov_iter
*iter
);
3165 extern int simple_empty(struct dentry
*);
3166 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
3167 loff_t pos
, unsigned len
,
3168 struct page
**pagep
, void **fsdata
);
3169 extern const struct address_space_operations ram_aops
;
3170 extern int always_delete_dentry(const struct dentry
*);
3171 extern struct inode
*alloc_anon_inode(struct super_block
*);
3172 extern int simple_nosetlease(struct file
*, int, struct file_lock
**, void **);
3173 extern const struct dentry_operations simple_dentry_operations
;
3175 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
3176 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
3177 extern const struct file_operations simple_dir_operations
;
3178 extern const struct inode_operations simple_dir_inode_operations
;
3179 extern void make_empty_dir_inode(struct inode
*inode
);
3180 extern bool is_empty_dir_inode(struct inode
*inode
);
3181 struct tree_descr
{ const char *name
; const struct file_operations
*ops
; int mode
; };
3182 struct dentry
*d_alloc_name(struct dentry
*, const char *);
3183 extern int simple_fill_super(struct super_block
*, unsigned long,
3184 const struct tree_descr
*);
3185 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
3186 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
3188 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
3189 loff_t
*ppos
, const void *from
, size_t available
);
3190 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
3191 const void __user
*from
, size_t count
);
3198 void simple_offset_init(struct offset_ctx
*octx
);
3199 int simple_offset_add(struct offset_ctx
*octx
, struct dentry
*dentry
);
3200 void simple_offset_remove(struct offset_ctx
*octx
, struct dentry
*dentry
);
3201 int simple_offset_rename_exchange(struct inode
*old_dir
,
3202 struct dentry
*old_dentry
,
3203 struct inode
*new_dir
,
3204 struct dentry
*new_dentry
);
3205 void simple_offset_destroy(struct offset_ctx
*octx
);
3207 extern const struct file_operations simple_offset_dir_operations
;
3209 extern int __generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3210 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3212 extern int generic_check_addressable(unsigned, u64
);
3214 extern void generic_set_encrypted_ci_d_ops(struct dentry
*dentry
);
3216 int may_setattr(struct mnt_idmap
*idmap
, struct inode
*inode
,
3217 unsigned int ia_valid
);
3218 int setattr_prepare(struct mnt_idmap
*, struct dentry
*, struct iattr
*);
3219 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
3220 void setattr_copy(struct mnt_idmap
*, struct inode
*inode
,
3221 const struct iattr
*attr
);
3223 extern int file_update_time(struct file
*file
);
3225 static inline bool vma_is_dax(const struct vm_area_struct
*vma
)
3227 return vma
->vm_file
&& IS_DAX(vma
->vm_file
->f_mapping
->host
);
3230 static inline bool vma_is_fsdax(struct vm_area_struct
*vma
)
3232 struct inode
*inode
;
3234 if (!IS_ENABLED(CONFIG_FS_DAX
) || !vma
->vm_file
)
3236 if (!vma_is_dax(vma
))
3238 inode
= file_inode(vma
->vm_file
);
3239 if (S_ISCHR(inode
->i_mode
))
3240 return false; /* device-dax */
3244 static inline int iocb_flags(struct file
*file
)
3247 if (file
->f_flags
& O_APPEND
)
3249 if (file
->f_flags
& O_DIRECT
)
3251 if (file
->f_flags
& O_DSYNC
)
3253 if (file
->f_flags
& __O_SYNC
)
3258 static inline int kiocb_set_rw_flags(struct kiocb
*ki
, rwf_t flags
)
3260 int kiocb_flags
= 0;
3262 /* make sure there's no overlap between RWF and private IOCB flags */
3263 BUILD_BUG_ON((__force
int) RWF_SUPPORTED
& IOCB_EVENTFD
);
3267 if (unlikely(flags
& ~RWF_SUPPORTED
))
3270 if (flags
& RWF_NOWAIT
) {
3271 if (!(ki
->ki_filp
->f_mode
& FMODE_NOWAIT
))
3273 kiocb_flags
|= IOCB_NOIO
;
3275 kiocb_flags
|= (__force
int) (flags
& RWF_SUPPORTED
);
3276 if (flags
& RWF_SYNC
)
3277 kiocb_flags
|= IOCB_DSYNC
;
3279 ki
->ki_flags
|= kiocb_flags
;
3283 static inline ino_t
parent_ino(struct dentry
*dentry
)
3288 * Don't strictly need d_lock here? If the parent ino could change
3289 * then surely we'd have a deeper race in the caller?
3291 spin_lock(&dentry
->d_lock
);
3292 res
= dentry
->d_parent
->d_inode
->i_ino
;
3293 spin_unlock(&dentry
->d_lock
);
3297 /* Transaction based IO helpers */
3300 * An argresp is stored in an allocated page and holds the
3301 * size of the argument or response, along with its content
3303 struct simple_transaction_argresp
{
3308 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3310 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
3312 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
3313 size_t size
, loff_t
*pos
);
3314 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
3316 void simple_transaction_set(struct file
*file
, size_t n
);
3319 * simple attribute files
3321 * These attributes behave similar to those in sysfs:
3323 * Writing to an attribute immediately sets a value, an open file can be
3324 * written to multiple times.
3326 * Reading from an attribute creates a buffer from the value that might get
3327 * read with multiple read calls. When the attribute has been read
3328 * completely, no further read calls are possible until the file is opened
3331 * All attributes contain a text representation of a numeric value
3332 * that are accessed with the get() and set() functions.
3334 #define DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, __is_signed) \
3335 static int __fops ## _open(struct inode *inode, struct file *file) \
3337 __simple_attr_check_format(__fmt, 0ull); \
3338 return simple_attr_open(inode, file, __get, __set, __fmt); \
3340 static const struct file_operations __fops = { \
3341 .owner = THIS_MODULE, \
3342 .open = __fops ## _open, \
3343 .release = simple_attr_release, \
3344 .read = simple_attr_read, \
3345 .write = (__is_signed) ? simple_attr_write_signed : simple_attr_write, \
3346 .llseek = generic_file_llseek, \
3349 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3350 DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, false)
3352 #define DEFINE_SIMPLE_ATTRIBUTE_SIGNED(__fops, __get, __set, __fmt) \
3353 DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, true)
3355 static inline __printf(1, 2)
3356 void __simple_attr_check_format(const char *fmt
, ...)
3358 /* don't do anything, just let the compiler check the arguments; */
3361 int simple_attr_open(struct inode
*inode
, struct file
*file
,
3362 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
3364 int simple_attr_release(struct inode
*inode
, struct file
*file
);
3365 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
3366 size_t len
, loff_t
*ppos
);
3367 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
3368 size_t len
, loff_t
*ppos
);
3369 ssize_t
simple_attr_write_signed(struct file
*file
, const char __user
*buf
,
3370 size_t len
, loff_t
*ppos
);
3373 int __init
list_bdev_fs_names(char *buf
, size_t size
);
3375 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3376 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3378 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3379 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3380 (flag & __FMODE_NONOTIFY)))
3382 static inline bool is_sxid(umode_t mode
)
3384 return mode
& (S_ISUID
| S_ISGID
);
3387 static inline int check_sticky(struct mnt_idmap
*idmap
,
3388 struct inode
*dir
, struct inode
*inode
)
3390 if (!(dir
->i_mode
& S_ISVTX
))
3393 return __check_sticky(idmap
, dir
, inode
);
3396 static inline void inode_has_no_xattr(struct inode
*inode
)
3398 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& SB_NOSEC
))
3399 inode
->i_flags
|= S_NOSEC
;
3402 static inline bool is_root_inode(struct inode
*inode
)
3404 return inode
== inode
->i_sb
->s_root
->d_inode
;
3407 static inline bool dir_emit(struct dir_context
*ctx
,
3408 const char *name
, int namelen
,
3409 u64 ino
, unsigned type
)
3411 return ctx
->actor(ctx
, name
, namelen
, ctx
->pos
, ino
, type
);
3413 static inline bool dir_emit_dot(struct file
*file
, struct dir_context
*ctx
)
3415 return ctx
->actor(ctx
, ".", 1, ctx
->pos
,
3416 file
->f_path
.dentry
->d_inode
->i_ino
, DT_DIR
);
3418 static inline bool dir_emit_dotdot(struct file
*file
, struct dir_context
*ctx
)
3420 return ctx
->actor(ctx
, "..", 2, ctx
->pos
,
3421 parent_ino(file
->f_path
.dentry
), DT_DIR
);
3423 static inline bool dir_emit_dots(struct file
*file
, struct dir_context
*ctx
)
3425 if (ctx
->pos
== 0) {
3426 if (!dir_emit_dot(file
, ctx
))
3430 if (ctx
->pos
== 1) {
3431 if (!dir_emit_dotdot(file
, ctx
))
3437 static inline bool dir_relax(struct inode
*inode
)
3439 inode_unlock(inode
);
3441 return !IS_DEADDIR(inode
);
3444 static inline bool dir_relax_shared(struct inode
*inode
)
3446 inode_unlock_shared(inode
);
3447 inode_lock_shared(inode
);
3448 return !IS_DEADDIR(inode
);
3451 extern bool path_noexec(const struct path
*path
);
3452 extern void inode_nohighmem(struct inode
*inode
);
3455 extern int vfs_fadvise(struct file
*file
, loff_t offset
, loff_t len
,
3457 extern int generic_fadvise(struct file
*file
, loff_t offset
, loff_t len
,
3460 #endif /* _LINUX_FS_H */