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
;
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 /* File is opened with O_NDELAY (only set for block devices) */
123 #define FMODE_NDELAY ((__force fmode_t)0x40)
124 /* File is opened with O_EXCL (only set for block devices) */
125 #define FMODE_EXCL ((__force fmode_t)0x80)
126 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
127 (specialy hack for floppy.c) */
128 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
129 /* 32bit hashes as llseek() offset (for directories) */
130 #define FMODE_32BITHASH ((__force fmode_t)0x200)
131 /* 64bit hashes as llseek() offset (for directories) */
132 #define FMODE_64BITHASH ((__force fmode_t)0x400)
135 * Don't update ctime and mtime.
137 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
138 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
140 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
142 /* Expect random access pattern */
143 #define FMODE_RANDOM ((__force fmode_t)0x1000)
145 /* File is huge (eg. /dev/mem): treat loff_t as unsigned */
146 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
148 /* File is opened with O_PATH; almost nothing can be done with it */
149 #define FMODE_PATH ((__force fmode_t)0x4000)
151 /* File needs atomic accesses to f_pos */
152 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
153 /* Write access to underlying fs */
154 #define FMODE_WRITER ((__force fmode_t)0x10000)
155 /* Has read method(s) */
156 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
157 /* Has write method(s) */
158 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
160 #define FMODE_OPENED ((__force fmode_t)0x80000)
161 #define FMODE_CREATED ((__force fmode_t)0x100000)
163 /* File is stream-like */
164 #define FMODE_STREAM ((__force fmode_t)0x200000)
166 /* File supports DIRECT IO */
167 #define FMODE_CAN_ODIRECT ((__force fmode_t)0x400000)
169 /* File was opened by fanotify and shouldn't generate fanotify events */
170 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
172 /* File is capable of returning -EAGAIN if I/O will block */
173 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
175 /* File represents mount that needs unmounting */
176 #define FMODE_NEED_UNMOUNT ((__force fmode_t)0x10000000)
178 /* File does not contribute to nr_files count */
179 #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000)
181 /* File supports async buffered reads */
182 #define FMODE_BUF_RASYNC ((__force fmode_t)0x40000000)
184 /* File supports async nowait buffered writes */
185 #define FMODE_BUF_WASYNC ((__force fmode_t)0x80000000)
188 * Attribute flags. These should be or-ed together to figure out what
191 #define ATTR_MODE (1 << 0)
192 #define ATTR_UID (1 << 1)
193 #define ATTR_GID (1 << 2)
194 #define ATTR_SIZE (1 << 3)
195 #define ATTR_ATIME (1 << 4)
196 #define ATTR_MTIME (1 << 5)
197 #define ATTR_CTIME (1 << 6)
198 #define ATTR_ATIME_SET (1 << 7)
199 #define ATTR_MTIME_SET (1 << 8)
200 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
201 #define ATTR_KILL_SUID (1 << 11)
202 #define ATTR_KILL_SGID (1 << 12)
203 #define ATTR_FILE (1 << 13)
204 #define ATTR_KILL_PRIV (1 << 14)
205 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
206 #define ATTR_TIMES_SET (1 << 16)
207 #define ATTR_TOUCH (1 << 17)
210 * Whiteout is represented by a char device. The following constants define the
211 * mode and device number to use.
213 #define WHITEOUT_MODE 0
214 #define WHITEOUT_DEV 0
217 * This is the Inode Attributes structure, used for notify_change(). It
218 * uses the above definitions as flags, to know which values have changed.
219 * Also, in this manner, a Filesystem can look at only the values it cares
220 * about. Basically, these are the attributes that the VFS layer can
221 * request to change from the FS layer.
223 * Derek Atkins <warlord@MIT.EDU> 94-10-20
226 unsigned int ia_valid
;
229 * The two anonymous unions wrap structures with the same member.
231 * Filesystems raising FS_ALLOW_IDMAP need to use ia_vfs{g,u}id which
232 * are a dedicated type requiring the filesystem to use the dedicated
233 * helpers. Other filesystem can continue to use ia_{g,u}id until they
236 * They always contain the same value. In other words FS_ALLOW_IDMAP
237 * pass down the same value on idmapped mounts as they would on regular
249 struct timespec64 ia_atime
;
250 struct timespec64 ia_mtime
;
251 struct timespec64 ia_ctime
;
254 * Not an attribute, but an auxiliary info for filesystems wanting to
255 * implement an ftruncate() like method. NOTE: filesystem should
256 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
258 struct file
*ia_file
;
262 * Includes for diskquotas.
264 #include <linux/quota.h>
267 * Maximum number of layers of fs stack. Needs to be limited to
268 * prevent kernel stack overflow
270 #define FILESYSTEM_MAX_STACK_DEPTH 2
273 * enum positive_aop_returns - aop return codes with specific semantics
275 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
276 * completed, that the page is still locked, and
277 * should be considered active. The VM uses this hint
278 * to return the page to the active list -- it won't
279 * be a candidate for writeback again in the near
280 * future. Other callers must be careful to unlock
281 * the page if they get this return. Returned by
284 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
285 * unlocked it and the page might have been truncated.
286 * The caller should back up to acquiring a new page and
287 * trying again. The aop will be taking reasonable
288 * precautions not to livelock. If the caller held a page
289 * reference, it should drop it before retrying. Returned
292 * address_space_operation functions return these large constants to indicate
293 * special semantics to the caller. These are much larger than the bytes in a
294 * page to allow for functions that return the number of bytes operated on in a
298 enum positive_aop_returns
{
299 AOP_WRITEPAGE_ACTIVATE
= 0x80000,
300 AOP_TRUNCATED_PAGE
= 0x80001,
304 * oh the beauties of C type declarations.
307 struct address_space
;
308 struct writeback_control
;
309 struct readahead_control
;
312 * Write life time hint values.
313 * Stored in struct inode as u8.
316 WRITE_LIFE_NOT_SET
= 0,
317 WRITE_LIFE_NONE
= RWH_WRITE_LIFE_NONE
,
318 WRITE_LIFE_SHORT
= RWH_WRITE_LIFE_SHORT
,
319 WRITE_LIFE_MEDIUM
= RWH_WRITE_LIFE_MEDIUM
,
320 WRITE_LIFE_LONG
= RWH_WRITE_LIFE_LONG
,
321 WRITE_LIFE_EXTREME
= RWH_WRITE_LIFE_EXTREME
,
324 /* Match RWF_* bits to IOCB bits */
325 #define IOCB_HIPRI (__force int) RWF_HIPRI
326 #define IOCB_DSYNC (__force int) RWF_DSYNC
327 #define IOCB_SYNC (__force int) RWF_SYNC
328 #define IOCB_NOWAIT (__force int) RWF_NOWAIT
329 #define IOCB_APPEND (__force int) RWF_APPEND
331 /* non-RWF related bits - start at 16 */
332 #define IOCB_EVENTFD (1 << 16)
333 #define IOCB_DIRECT (1 << 17)
334 #define IOCB_WRITE (1 << 18)
335 /* iocb->ki_waitq is valid */
336 #define IOCB_WAITQ (1 << 19)
337 #define IOCB_NOIO (1 << 20)
338 /* can use bio alloc cache */
339 #define IOCB_ALLOC_CACHE (1 << 21)
342 struct file
*ki_filp
;
344 void (*ki_complete
)(struct kiocb
*iocb
, long ret
);
347 u16 ki_ioprio
; /* See linux/ioprio.h */
348 struct wait_page_queue
*ki_waitq
; /* for async buffered IO */
351 static inline bool is_sync_kiocb(struct kiocb
*kiocb
)
353 return kiocb
->ki_complete
== NULL
;
356 struct address_space_operations
{
357 int (*writepage
)(struct page
*page
, struct writeback_control
*wbc
);
358 int (*read_folio
)(struct file
*, struct folio
*);
360 /* Write back some dirty pages from this mapping. */
361 int (*writepages
)(struct address_space
*, struct writeback_control
*);
363 /* Mark a folio dirty. Return true if this dirtied it */
364 bool (*dirty_folio
)(struct address_space
*, struct folio
*);
366 void (*readahead
)(struct readahead_control
*);
368 int (*write_begin
)(struct file
*, struct address_space
*mapping
,
369 loff_t pos
, unsigned len
,
370 struct page
**pagep
, void **fsdata
);
371 int (*write_end
)(struct file
*, struct address_space
*mapping
,
372 loff_t pos
, unsigned len
, unsigned copied
,
373 struct page
*page
, void *fsdata
);
375 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
376 sector_t (*bmap
)(struct address_space
*, sector_t
);
377 void (*invalidate_folio
) (struct folio
*, size_t offset
, size_t len
);
378 bool (*release_folio
)(struct folio
*, gfp_t
);
379 void (*free_folio
)(struct folio
*folio
);
380 ssize_t (*direct_IO
)(struct kiocb
*, struct iov_iter
*iter
);
382 * migrate the contents of a folio to the specified target. If
383 * migrate_mode is MIGRATE_ASYNC, it must not block.
385 int (*migrate_folio
)(struct address_space
*, struct folio
*dst
,
386 struct folio
*src
, enum migrate_mode
);
387 int (*launder_folio
)(struct folio
*);
388 bool (*is_partially_uptodate
) (struct folio
*, size_t from
,
390 void (*is_dirty_writeback
) (struct folio
*, bool *dirty
, bool *wb
);
391 int (*error_remove_page
)(struct address_space
*, struct page
*);
393 /* swapfile support */
394 int (*swap_activate
)(struct swap_info_struct
*sis
, struct file
*file
,
396 void (*swap_deactivate
)(struct file
*file
);
397 int (*swap_rw
)(struct kiocb
*iocb
, struct iov_iter
*iter
);
400 extern const struct address_space_operations empty_aops
;
403 * struct address_space - Contents of a cacheable, mappable object.
404 * @host: Owner, either the inode or the block_device.
405 * @i_pages: Cached pages.
406 * @invalidate_lock: Guards coherency between page cache contents and
407 * file offset->disk block mappings in the filesystem during invalidates.
408 * It is also used to block modification of page cache contents through
410 * @gfp_mask: Memory allocation flags to use for allocating pages.
411 * @i_mmap_writable: Number of VM_SHARED mappings.
412 * @nr_thps: Number of THPs in the pagecache (non-shmem only).
413 * @i_mmap: Tree of private and shared mappings.
414 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
415 * @nrpages: Number of page entries, protected by the i_pages lock.
416 * @writeback_index: Writeback starts here.
418 * @flags: Error bits and flags (AS_*).
419 * @wb_err: The most recent error which has occurred.
420 * @private_lock: For use by the owner of the address_space.
421 * @private_list: For use by the owner of the address_space.
422 * @private_data: For use by the owner of the address_space.
424 struct address_space
{
426 struct xarray i_pages
;
427 struct rw_semaphore invalidate_lock
;
429 atomic_t i_mmap_writable
;
430 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
431 /* number of thp, only for non-shmem files */
434 struct rb_root_cached i_mmap
;
435 struct rw_semaphore i_mmap_rwsem
;
436 unsigned long nrpages
;
437 pgoff_t writeback_index
;
438 const struct address_space_operations
*a_ops
;
441 spinlock_t private_lock
;
442 struct list_head private_list
;
444 } __attribute__((aligned(sizeof(long)))) __randomize_layout
;
446 * On most architectures that alignment is already the case; but
447 * must be enforced here for CRIS, to let the least significant bit
448 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
451 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
452 #define PAGECACHE_TAG_DIRTY XA_MARK_0
453 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
454 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
457 * Returns true if any of the pages in the mapping are marked with the tag.
459 static inline bool mapping_tagged(struct address_space
*mapping
, xa_mark_t tag
)
461 return xa_marked(&mapping
->i_pages
, tag
);
464 static inline void i_mmap_lock_write(struct address_space
*mapping
)
466 down_write(&mapping
->i_mmap_rwsem
);
469 static inline int i_mmap_trylock_write(struct address_space
*mapping
)
471 return down_write_trylock(&mapping
->i_mmap_rwsem
);
474 static inline void i_mmap_unlock_write(struct address_space
*mapping
)
476 up_write(&mapping
->i_mmap_rwsem
);
479 static inline int i_mmap_trylock_read(struct address_space
*mapping
)
481 return down_read_trylock(&mapping
->i_mmap_rwsem
);
484 static inline void i_mmap_lock_read(struct address_space
*mapping
)
486 down_read(&mapping
->i_mmap_rwsem
);
489 static inline void i_mmap_unlock_read(struct address_space
*mapping
)
491 up_read(&mapping
->i_mmap_rwsem
);
494 static inline void i_mmap_assert_locked(struct address_space
*mapping
)
496 lockdep_assert_held(&mapping
->i_mmap_rwsem
);
499 static inline void i_mmap_assert_write_locked(struct address_space
*mapping
)
501 lockdep_assert_held_write(&mapping
->i_mmap_rwsem
);
505 * Might pages of this file be mapped into userspace?
507 static inline int mapping_mapped(struct address_space
*mapping
)
509 return !RB_EMPTY_ROOT(&mapping
->i_mmap
.rb_root
);
513 * Might pages of this file have been modified in userspace?
514 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap
515 * marks vma as VM_SHARED if it is shared, and the file was opened for
516 * writing i.e. vma may be mprotected writable even if now readonly.
518 * If i_mmap_writable is negative, no new writable mappings are allowed. You
519 * can only deny writable mappings, if none exists right now.
521 static inline int mapping_writably_mapped(struct address_space
*mapping
)
523 return atomic_read(&mapping
->i_mmap_writable
) > 0;
526 static inline int mapping_map_writable(struct address_space
*mapping
)
528 return atomic_inc_unless_negative(&mapping
->i_mmap_writable
) ?
532 static inline void mapping_unmap_writable(struct address_space
*mapping
)
534 atomic_dec(&mapping
->i_mmap_writable
);
537 static inline int mapping_deny_writable(struct address_space
*mapping
)
539 return atomic_dec_unless_positive(&mapping
->i_mmap_writable
) ?
543 static inline void mapping_allow_writable(struct address_space
*mapping
)
545 atomic_inc(&mapping
->i_mmap_writable
);
549 * Use sequence counter to get consistent i_size on 32-bit processors.
551 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
552 #include <linux/seqlock.h>
553 #define __NEED_I_SIZE_ORDERED
554 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
556 #define i_size_ordered_init(inode) do { } while (0)
560 #define ACL_NOT_CACHED ((void *)(-1))
562 * ACL_DONT_CACHE is for stacked filesystems, that rely on underlying fs to
563 * cache the ACL. This also means that ->get_inode_acl() can be called in RCU
564 * mode with the LOOKUP_RCU flag.
566 #define ACL_DONT_CACHE ((void *)(-3))
568 static inline struct posix_acl
*
569 uncached_acl_sentinel(struct task_struct
*task
)
571 return (void *)task
+ 1;
575 is_uncached_acl(struct posix_acl
*acl
)
577 return (long)acl
& 1;
580 #define IOP_FASTPERM 0x0001
581 #define IOP_LOOKUP 0x0002
582 #define IOP_NOFOLLOW 0x0004
583 #define IOP_XATTR 0x0008
584 #define IOP_DEFAULT_READLINK 0x0010
586 struct fsnotify_mark_connector
;
589 * Keep mostly read-only and often accessed (especially for
590 * the RCU path lookup and 'stat' data) fields at the beginning
591 * of the 'struct inode'
595 unsigned short i_opflags
;
598 unsigned int i_flags
;
600 #ifdef CONFIG_FS_POSIX_ACL
601 struct posix_acl
*i_acl
;
602 struct posix_acl
*i_default_acl
;
605 const struct inode_operations
*i_op
;
606 struct super_block
*i_sb
;
607 struct address_space
*i_mapping
;
609 #ifdef CONFIG_SECURITY
613 /* Stat data, not accessed from path walking */
616 * Filesystems may only read i_nlink directly. They shall use the
617 * following functions for modification:
619 * (set|clear|inc|drop)_nlink
620 * inode_(inc|dec)_link_count
623 const unsigned int i_nlink
;
624 unsigned int __i_nlink
;
628 struct timespec64 i_atime
;
629 struct timespec64 i_mtime
;
630 struct timespec64 i_ctime
;
631 spinlock_t i_lock
; /* i_blocks, i_bytes, maybe i_size */
632 unsigned short i_bytes
;
637 #ifdef __NEED_I_SIZE_ORDERED
638 seqcount_t i_size_seqcount
;
642 unsigned long i_state
;
643 struct rw_semaphore i_rwsem
;
645 unsigned long dirtied_when
; /* jiffies of first dirtying */
646 unsigned long dirtied_time_when
;
648 struct hlist_node i_hash
;
649 struct list_head i_io_list
; /* backing dev IO list */
650 #ifdef CONFIG_CGROUP_WRITEBACK
651 struct bdi_writeback
*i_wb
; /* the associated cgroup wb */
653 /* foreign inode detection, see wbc_detach_inode() */
655 u16 i_wb_frn_avg_time
;
656 u16 i_wb_frn_history
;
658 struct list_head i_lru
; /* inode LRU list */
659 struct list_head i_sb_list
;
660 struct list_head i_wb_list
; /* backing dev writeback list */
662 struct hlist_head i_dentry
;
663 struct rcu_head i_rcu
;
665 atomic64_t i_version
;
666 atomic64_t i_sequence
; /* see futex */
668 atomic_t i_dio_count
;
669 atomic_t i_writecount
;
670 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
671 atomic_t i_readcount
; /* struct files open RO */
674 const struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
675 void (*free_inode
)(struct inode
*);
677 struct file_lock_context
*i_flctx
;
678 struct address_space i_data
;
679 struct list_head i_devices
;
681 struct pipe_inode_info
*i_pipe
;
689 #ifdef CONFIG_FSNOTIFY
690 __u32 i_fsnotify_mask
; /* all events this inode cares about */
691 struct fsnotify_mark_connector __rcu
*i_fsnotify_marks
;
694 #ifdef CONFIG_FS_ENCRYPTION
695 struct fscrypt_info
*i_crypt_info
;
698 #ifdef CONFIG_FS_VERITY
699 struct fsverity_info
*i_verity_info
;
702 void *i_private
; /* fs or device private pointer */
703 } __randomize_layout
;
705 struct timespec64
timestamp_truncate(struct timespec64 t
, struct inode
*inode
);
707 static inline unsigned int i_blocksize(const struct inode
*node
)
709 return (1 << node
->i_blkbits
);
712 static inline int inode_unhashed(struct inode
*inode
)
714 return hlist_unhashed(&inode
->i_hash
);
718 * __mark_inode_dirty expects inodes to be hashed. Since we don't
719 * want special inodes in the fileset inode space, we make them
720 * appear hashed, but do not put on any lists. hlist_del()
721 * will work fine and require no locking.
723 static inline void inode_fake_hash(struct inode
*inode
)
725 hlist_add_fake(&inode
->i_hash
);
729 * inode->i_mutex nesting subclasses for the lock validator:
731 * 0: the object of the current VFS operation
735 * 4: second non-directory
736 * 5: second parent (when locking independent directories in rename)
738 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
739 * non-directories at once.
741 * The locking order between these classes is
742 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
744 enum inode_i_mutex_lock_class
754 static inline void inode_lock(struct inode
*inode
)
756 down_write(&inode
->i_rwsem
);
759 static inline void inode_unlock(struct inode
*inode
)
761 up_write(&inode
->i_rwsem
);
764 static inline void inode_lock_shared(struct inode
*inode
)
766 down_read(&inode
->i_rwsem
);
769 static inline void inode_unlock_shared(struct inode
*inode
)
771 up_read(&inode
->i_rwsem
);
774 static inline int inode_trylock(struct inode
*inode
)
776 return down_write_trylock(&inode
->i_rwsem
);
779 static inline int inode_trylock_shared(struct inode
*inode
)
781 return down_read_trylock(&inode
->i_rwsem
);
784 static inline int inode_is_locked(struct inode
*inode
)
786 return rwsem_is_locked(&inode
->i_rwsem
);
789 static inline void inode_lock_nested(struct inode
*inode
, unsigned subclass
)
791 down_write_nested(&inode
->i_rwsem
, subclass
);
794 static inline void inode_lock_shared_nested(struct inode
*inode
, unsigned subclass
)
796 down_read_nested(&inode
->i_rwsem
, subclass
);
799 static inline void filemap_invalidate_lock(struct address_space
*mapping
)
801 down_write(&mapping
->invalidate_lock
);
804 static inline void filemap_invalidate_unlock(struct address_space
*mapping
)
806 up_write(&mapping
->invalidate_lock
);
809 static inline void filemap_invalidate_lock_shared(struct address_space
*mapping
)
811 down_read(&mapping
->invalidate_lock
);
814 static inline int filemap_invalidate_trylock_shared(
815 struct address_space
*mapping
)
817 return down_read_trylock(&mapping
->invalidate_lock
);
820 static inline void filemap_invalidate_unlock_shared(
821 struct address_space
*mapping
)
823 up_read(&mapping
->invalidate_lock
);
826 void lock_two_nondirectories(struct inode
*, struct inode
*);
827 void unlock_two_nondirectories(struct inode
*, struct inode
*);
829 void filemap_invalidate_lock_two(struct address_space
*mapping1
,
830 struct address_space
*mapping2
);
831 void filemap_invalidate_unlock_two(struct address_space
*mapping1
,
832 struct address_space
*mapping2
);
836 * NOTE: in a 32bit arch with a preemptable kernel and
837 * an UP compile the i_size_read/write must be atomic
838 * with respect to the local cpu (unlike with preempt disabled),
839 * but they don't need to be atomic with respect to other cpus like in
840 * true SMP (so they need either to either locally disable irq around
841 * the read or for example on x86 they can be still implemented as a
842 * cmpxchg8b without the need of the lock prefix). For SMP compiles
843 * and 64bit archs it makes no difference if preempt is enabled or not.
845 static inline loff_t
i_size_read(const struct inode
*inode
)
847 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
852 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
853 i_size
= inode
->i_size
;
854 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
856 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
860 i_size
= inode
->i_size
;
864 return inode
->i_size
;
869 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
870 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
871 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
873 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
875 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
877 write_seqcount_begin(&inode
->i_size_seqcount
);
878 inode
->i_size
= i_size
;
879 write_seqcount_end(&inode
->i_size_seqcount
);
881 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
883 inode
->i_size
= i_size
;
886 inode
->i_size
= i_size
;
890 static inline unsigned iminor(const struct inode
*inode
)
892 return MINOR(inode
->i_rdev
);
895 static inline unsigned imajor(const struct inode
*inode
)
897 return MAJOR(inode
->i_rdev
);
901 rwlock_t lock
; /* protects pid, uid, euid fields */
902 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
903 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
904 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
905 int signum
; /* posix.1b rt signal to be delivered on IO */
909 * struct file_ra_state - Track a file's readahead state.
910 * @start: Where the most recent readahead started.
911 * @size: Number of pages read in the most recent readahead.
912 * @async_size: Numer of pages that were/are not needed immediately
913 * and so were/are genuinely "ahead". Start next readahead when
914 * the first of these pages is accessed.
915 * @ra_pages: Maximum size of a readahead request, copied from the bdi.
916 * @mmap_miss: How many mmap accesses missed in the page cache.
917 * @prev_pos: The last byte in the most recent read request.
919 * When this structure is passed to ->readahead(), the "most recent"
920 * readahead means the current readahead.
922 struct file_ra_state
{
925 unsigned int async_size
;
926 unsigned int ra_pages
;
927 unsigned int mmap_miss
;
932 * Check if @index falls in the readahead windows.
934 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
936 return (index
>= ra
->start
&&
937 index
< ra
->start
+ ra
->size
);
942 struct llist_node f_llist
;
943 struct rcu_head f_rcuhead
;
944 unsigned int f_iocb_flags
;
947 struct inode
*f_inode
; /* cached value */
948 const struct file_operations
*f_op
;
951 * Protects f_ep, f_flags.
952 * Must not be taken from IRQ context.
955 atomic_long_t f_count
;
956 unsigned int f_flags
;
958 struct mutex f_pos_lock
;
960 struct fown_struct f_owner
;
961 const struct cred
*f_cred
;
962 struct file_ra_state f_ra
;
965 #ifdef CONFIG_SECURITY
968 /* needed for tty driver, and maybe others */
972 /* Used by fs/eventpoll.c to link all the hooks to this file */
973 struct hlist_head
*f_ep
;
974 #endif /* #ifdef CONFIG_EPOLL */
975 struct address_space
*f_mapping
;
977 errseq_t f_sb_err
; /* for syncfs */
979 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
984 /* file identifier */
985 unsigned char f_handle
[];
988 static inline struct file
*get_file(struct file
*f
)
990 atomic_long_inc(&f
->f_count
);
993 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
994 #define file_count(x) atomic_long_read(&(x)->f_count)
996 #define MAX_NON_LFS ((1UL<<31) - 1)
998 /* Page cache limit. The filesystems should put that into their s_maxbytes
999 limits, otherwise bad things can happen in VM. */
1000 #if BITS_PER_LONG==32
1001 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
1002 #elif BITS_PER_LONG==64
1003 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
1008 #define FL_DELEG 4 /* NFSv4 delegation */
1009 #define FL_ACCESS 8 /* not trying to lock, just looking */
1010 #define FL_EXISTS 16 /* when unlocking, test for existence */
1011 #define FL_LEASE 32 /* lease held on this file */
1012 #define FL_CLOSE 64 /* unlock on close */
1013 #define FL_SLEEP 128 /* A blocking lock */
1014 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
1015 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
1016 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
1017 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
1018 #define FL_RECLAIM 4096 /* reclaiming from a reboot server */
1020 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
1023 * Special return value from posix_lock_file() and vfs_lock_file() for
1024 * asynchronous locking.
1026 #define FILE_LOCK_DEFERRED 1
1028 /* legacy typedef, should eventually be removed */
1029 typedef void *fl_owner_t
;
1033 struct file_lock_operations
{
1034 void (*fl_copy_lock
)(struct file_lock
*, struct file_lock
*);
1035 void (*fl_release_private
)(struct file_lock
*);
1038 struct lock_manager_operations
{
1040 fl_owner_t (*lm_get_owner
)(fl_owner_t
);
1041 void (*lm_put_owner
)(fl_owner_t
);
1042 void (*lm_notify
)(struct file_lock
*); /* unblock callback */
1043 int (*lm_grant
)(struct file_lock
*, int);
1044 bool (*lm_break
)(struct file_lock
*);
1045 int (*lm_change
)(struct file_lock
*, int, struct list_head
*);
1046 void (*lm_setup
)(struct file_lock
*, void **);
1047 bool (*lm_breaker_owns_lease
)(struct file_lock
*);
1048 bool (*lm_lock_expirable
)(struct file_lock
*cfl
);
1049 void (*lm_expire_lock
)(void);
1052 struct lock_manager
{
1053 struct list_head list
;
1055 * NFSv4 and up also want opens blocked during the grace period;
1062 void locks_start_grace(struct net
*, struct lock_manager
*);
1063 void locks_end_grace(struct lock_manager
*);
1064 bool locks_in_grace(struct net
*);
1065 bool opens_in_grace(struct net
*);
1067 /* that will die - we need it for nfs_lock_info */
1068 #include <linux/nfs_fs_i.h>
1071 * struct file_lock represents a generic "file lock". It's used to represent
1072 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1073 * note that the same struct is used to represent both a request for a lock and
1074 * the lock itself, but the same object is never used for both.
1076 * FIXME: should we create a separate "struct lock_request" to help distinguish
1079 * The varous i_flctx lists are ordered by:
1082 * 2) lock range start
1085 * Obviously, the last two criteria only matter for POSIX locks.
1088 struct file_lock
*fl_blocker
; /* The lock, that is blocking us */
1089 struct list_head fl_list
; /* link into file_lock_context */
1090 struct hlist_node fl_link
; /* node in global lists */
1091 struct list_head fl_blocked_requests
; /* list of requests with
1092 * ->fl_blocker pointing here
1094 struct list_head fl_blocked_member
; /* node in
1095 * ->fl_blocker->fl_blocked_requests
1097 fl_owner_t fl_owner
;
1098 unsigned int fl_flags
;
1099 unsigned char fl_type
;
1100 unsigned int fl_pid
;
1101 int fl_link_cpu
; /* what cpu's list is this on? */
1102 wait_queue_head_t fl_wait
;
1103 struct file
*fl_file
;
1107 struct fasync_struct
* fl_fasync
; /* for lease break notifications */
1108 /* for lease breaks: */
1109 unsigned long fl_break_time
;
1110 unsigned long fl_downgrade_time
;
1112 const struct file_lock_operations
*fl_ops
; /* Callbacks for filesystems */
1113 const struct lock_manager_operations
*fl_lmops
; /* Callbacks for lockmanagers */
1115 struct nfs_lock_info nfs_fl
;
1116 struct nfs4_lock_info nfs4_fl
;
1118 struct list_head link
; /* link in AFS vnode's pending_locks list */
1119 int state
; /* state of grant or error if -ve */
1120 unsigned int debug_id
;
1123 } __randomize_layout
;
1125 struct file_lock_context
{
1126 spinlock_t flc_lock
;
1127 struct list_head flc_flock
;
1128 struct list_head flc_posix
;
1129 struct list_head flc_lease
;
1132 /* The following constant reflects the upper bound of the file/locking space */
1134 #define OFFSET_MAX type_max(loff_t)
1135 #define OFFT_OFFSET_MAX type_max(off_t)
1138 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
1140 #define locks_inode(f) file_inode(f)
1142 #ifdef CONFIG_FILE_LOCKING
1143 extern int fcntl_getlk(struct file
*, unsigned int, struct flock
*);
1144 extern int fcntl_setlk(unsigned int, struct file
*, unsigned int,
1147 #if BITS_PER_LONG == 32
1148 extern int fcntl_getlk64(struct file
*, unsigned int, struct flock64
*);
1149 extern int fcntl_setlk64(unsigned int, struct file
*, unsigned int,
1153 extern int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
);
1154 extern int fcntl_getlease(struct file
*filp
);
1157 void locks_free_lock_context(struct inode
*inode
);
1158 void locks_free_lock(struct file_lock
*fl
);
1159 extern void locks_init_lock(struct file_lock
*);
1160 extern struct file_lock
* locks_alloc_lock(void);
1161 extern void locks_copy_lock(struct file_lock
*, struct file_lock
*);
1162 extern void locks_copy_conflock(struct file_lock
*, struct file_lock
*);
1163 extern void locks_remove_posix(struct file
*, fl_owner_t
);
1164 extern void locks_remove_file(struct file
*);
1165 extern void locks_release_private(struct file_lock
*);
1166 extern void posix_test_lock(struct file
*, struct file_lock
*);
1167 extern int posix_lock_file(struct file
*, struct file_lock
*, struct file_lock
*);
1168 extern int locks_delete_block(struct file_lock
*);
1169 extern int vfs_test_lock(struct file
*, struct file_lock
*);
1170 extern int vfs_lock_file(struct file
*, unsigned int, struct file_lock
*, struct file_lock
*);
1171 extern int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
);
1172 bool vfs_inode_has_locks(struct inode
*inode
);
1173 extern int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
);
1174 extern int __break_lease(struct inode
*inode
, unsigned int flags
, unsigned int type
);
1175 extern void lease_get_mtime(struct inode
*, struct timespec64
*time
);
1176 extern int generic_setlease(struct file
*, long, struct file_lock
**, void **priv
);
1177 extern int vfs_setlease(struct file
*, long, struct file_lock
**, void **);
1178 extern int lease_modify(struct file_lock
*, int, struct list_head
*);
1180 struct notifier_block
;
1181 extern int lease_register_notifier(struct notifier_block
*);
1182 extern void lease_unregister_notifier(struct notifier_block
*);
1184 struct files_struct
;
1185 extern void show_fd_locks(struct seq_file
*f
,
1186 struct file
*filp
, struct files_struct
*files
);
1187 extern bool locks_owner_has_blockers(struct file_lock_context
*flctx
,
1190 static inline struct file_lock_context
*
1191 locks_inode_context(const struct inode
*inode
)
1193 return smp_load_acquire(&inode
->i_flctx
);
1196 #else /* !CONFIG_FILE_LOCKING */
1197 static inline int fcntl_getlk(struct file
*file
, unsigned int cmd
,
1198 struct flock __user
*user
)
1203 static inline int fcntl_setlk(unsigned int fd
, struct file
*file
,
1204 unsigned int cmd
, struct flock __user
*user
)
1209 #if BITS_PER_LONG == 32
1210 static inline int fcntl_getlk64(struct file
*file
, unsigned int cmd
,
1211 struct flock64
*user
)
1216 static inline int fcntl_setlk64(unsigned int fd
, struct file
*file
,
1217 unsigned int cmd
, struct flock64
*user
)
1222 static inline int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
)
1227 static inline int fcntl_getlease(struct file
*filp
)
1233 locks_free_lock_context(struct inode
*inode
)
1237 static inline void locks_init_lock(struct file_lock
*fl
)
1242 static inline void locks_copy_conflock(struct file_lock
*new, struct file_lock
*fl
)
1247 static inline void locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
1252 static inline void locks_remove_posix(struct file
*filp
, fl_owner_t owner
)
1257 static inline void locks_remove_file(struct file
*filp
)
1262 static inline void posix_test_lock(struct file
*filp
, struct file_lock
*fl
)
1267 static inline int posix_lock_file(struct file
*filp
, struct file_lock
*fl
,
1268 struct file_lock
*conflock
)
1273 static inline int locks_delete_block(struct file_lock
*waiter
)
1278 static inline int vfs_test_lock(struct file
*filp
, struct file_lock
*fl
)
1283 static inline int vfs_lock_file(struct file
*filp
, unsigned int cmd
,
1284 struct file_lock
*fl
, struct file_lock
*conf
)
1289 static inline int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
)
1294 static inline bool vfs_inode_has_locks(struct inode
*inode
)
1299 static inline int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
)
1304 static inline int __break_lease(struct inode
*inode
, unsigned int mode
, unsigned int type
)
1309 static inline void lease_get_mtime(struct inode
*inode
,
1310 struct timespec64
*time
)
1315 static inline int generic_setlease(struct file
*filp
, long arg
,
1316 struct file_lock
**flp
, void **priv
)
1321 static inline int vfs_setlease(struct file
*filp
, long arg
,
1322 struct file_lock
**lease
, void **priv
)
1327 static inline int lease_modify(struct file_lock
*fl
, int arg
,
1328 struct list_head
*dispose
)
1333 struct files_struct
;
1334 static inline void show_fd_locks(struct seq_file
*f
,
1335 struct file
*filp
, struct files_struct
*files
) {}
1336 static inline bool locks_owner_has_blockers(struct file_lock_context
*flctx
,
1342 static inline struct file_lock_context
*
1343 locks_inode_context(const struct inode
*inode
)
1348 #endif /* !CONFIG_FILE_LOCKING */
1350 static inline struct inode
*file_inode(const struct file
*f
)
1355 static inline struct dentry
*file_dentry(const struct file
*file
)
1357 return d_real(file
->f_path
.dentry
, file_inode(file
));
1360 static inline int locks_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1362 return locks_lock_inode_wait(locks_inode(filp
), fl
);
1365 struct fasync_struct
{
1369 struct fasync_struct
*fa_next
; /* singly linked list */
1370 struct file
*fa_file
;
1371 struct rcu_head fa_rcu
;
1374 #define FASYNC_MAGIC 0x4601
1376 /* SMP safe fasync helpers: */
1377 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1378 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1379 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1380 extern struct fasync_struct
*fasync_alloc(void);
1381 extern void fasync_free(struct fasync_struct
*);
1383 /* can be called from interrupts */
1384 extern void kill_fasync(struct fasync_struct
**, int, int);
1386 extern void __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1387 extern int f_setown(struct file
*filp
, unsigned long arg
, int force
);
1388 extern void f_delown(struct file
*filp
);
1389 extern pid_t
f_getown(struct file
*filp
);
1390 extern int send_sigurg(struct fown_struct
*fown
);
1393 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1394 * represented in both.
1396 #define SB_RDONLY 1 /* Mount read-only */
1397 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1398 #define SB_NODEV 4 /* Disallow access to device special files */
1399 #define SB_NOEXEC 8 /* Disallow program execution */
1400 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1401 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1402 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1403 #define SB_NOATIME 1024 /* Do not update access times. */
1404 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1405 #define SB_SILENT 32768
1406 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1407 #define SB_INLINECRYPT (1<<17) /* Use blk-crypto for encrypted files */
1408 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1409 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1410 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1412 /* These sb flags are internal to the kernel */
1413 #define SB_SUBMOUNT (1<<26)
1414 #define SB_FORCE (1<<27)
1415 #define SB_NOSEC (1<<28)
1416 #define SB_BORN (1<<29)
1417 #define SB_ACTIVE (1<<30)
1418 #define SB_NOUSER (1<<31)
1420 /* These flags relate to encoding and casefolding */
1421 #define SB_ENC_STRICT_MODE_FL (1 << 0)
1423 #define sb_has_strict_encoding(sb) \
1424 (sb->s_encoding_flags & SB_ENC_STRICT_MODE_FL)
1430 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1431 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1432 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1433 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1434 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1437 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1438 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1439 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1440 #define SB_I_STABLE_WRITES 0x00000008 /* don't modify blks until WB is done */
1442 /* sb->s_iflags to limit user namespace mounts */
1443 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1444 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1445 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1447 #define SB_I_SKIP_SYNC 0x00000100 /* Skip superblock at global sync */
1448 #define SB_I_PERSB_BDI 0x00000200 /* has a per-sb bdi */
1449 #define SB_I_TS_EXPIRY_WARNED 0x00000400 /* warned about timestamp range expiry */
1450 #define SB_I_RETIRED 0x00000800 /* superblock shouldn't be reused */
1452 /* Possible states of 'frozen' field */
1454 SB_UNFROZEN
= 0, /* FS is unfrozen */
1455 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1456 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1457 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1458 * internal threads if needed) */
1459 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1462 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1465 int frozen
; /* Is sb frozen? */
1466 wait_queue_head_t wait_unfrozen
; /* wait for thaw */
1467 struct percpu_rw_semaphore rw_sem
[SB_FREEZE_LEVELS
];
1470 struct super_block
{
1471 struct list_head s_list
; /* Keep this first */
1472 dev_t s_dev
; /* search index; _not_ kdev_t */
1473 unsigned char s_blocksize_bits
;
1474 unsigned long s_blocksize
;
1475 loff_t s_maxbytes
; /* Max file size */
1476 struct file_system_type
*s_type
;
1477 const struct super_operations
*s_op
;
1478 const struct dquot_operations
*dq_op
;
1479 const struct quotactl_ops
*s_qcop
;
1480 const struct export_operations
*s_export_op
;
1481 unsigned long s_flags
;
1482 unsigned long s_iflags
; /* internal SB_I_* flags */
1483 unsigned long s_magic
;
1484 struct dentry
*s_root
;
1485 struct rw_semaphore s_umount
;
1488 #ifdef CONFIG_SECURITY
1491 const struct xattr_handler
**s_xattr
;
1492 #ifdef CONFIG_FS_ENCRYPTION
1493 const struct fscrypt_operations
*s_cop
;
1494 struct fscrypt_keyring
*s_master_keys
; /* master crypto keys in use */
1496 #ifdef CONFIG_FS_VERITY
1497 const struct fsverity_operations
*s_vop
;
1499 #if IS_ENABLED(CONFIG_UNICODE)
1500 struct unicode_map
*s_encoding
;
1501 __u16 s_encoding_flags
;
1503 struct hlist_bl_head s_roots
; /* alternate root dentries for NFS */
1504 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1505 struct block_device
*s_bdev
;
1506 struct backing_dev_info
*s_bdi
;
1507 struct mtd_info
*s_mtd
;
1508 struct hlist_node s_instances
;
1509 unsigned int s_quota_types
; /* Bitmask of supported quota types */
1510 struct quota_info s_dquot
; /* Diskquota specific options */
1512 struct sb_writers s_writers
;
1515 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1516 * s_fsnotify_marks together for cache efficiency. They are frequently
1517 * accessed and rarely modified.
1519 void *s_fs_info
; /* Filesystem private info */
1521 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1523 /* Time limits for c/m/atime in seconds */
1524 time64_t s_time_min
;
1525 time64_t s_time_max
;
1526 #ifdef CONFIG_FSNOTIFY
1527 __u32 s_fsnotify_mask
;
1528 struct fsnotify_mark_connector __rcu
*s_fsnotify_marks
;
1531 char s_id
[32]; /* Informational name */
1532 uuid_t s_uuid
; /* UUID */
1534 unsigned int s_max_links
;
1538 * The next field is for VFS *only*. No filesystems have any business
1539 * even looking at it. You had been warned.
1541 struct mutex s_vfs_rename_mutex
; /* Kludge */
1544 * Filesystem subtype. If non-empty the filesystem type field
1545 * in /proc/mounts will be "type.subtype"
1547 const char *s_subtype
;
1549 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1551 struct shrinker s_shrink
; /* per-sb shrinker handle */
1553 /* Number of inodes with nlink == 0 but still referenced */
1554 atomic_long_t s_remove_count
;
1557 * Number of inode/mount/sb objects that are being watched, note that
1558 * inodes objects are currently double-accounted.
1560 atomic_long_t s_fsnotify_connectors
;
1562 /* Being remounted read-only */
1563 int s_readonly_remount
;
1565 /* per-sb errseq_t for reporting writeback errors via syncfs */
1568 /* AIO completions deferred from interrupt context */
1569 struct workqueue_struct
*s_dio_done_wq
;
1570 struct hlist_head s_pins
;
1573 * Owning user namespace and default context in which to
1574 * interpret filesystem uids, gids, quotas, device nodes,
1575 * xattrs and security labels.
1577 struct user_namespace
*s_user_ns
;
1580 * The list_lru structure is essentially just a pointer to a table
1581 * of per-node lru lists, each of which has its own spinlock.
1582 * There is no need to put them into separate cachelines.
1584 struct list_lru s_dentry_lru
;
1585 struct list_lru s_inode_lru
;
1586 struct rcu_head rcu
;
1587 struct work_struct destroy_work
;
1589 struct mutex s_sync_lock
; /* sync serialisation lock */
1592 * Indicates how deep in a filesystem stack this SB is
1596 /* s_inode_list_lock protects s_inodes */
1597 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp
;
1598 struct list_head s_inodes
; /* all inodes */
1600 spinlock_t s_inode_wblist_lock
;
1601 struct list_head s_inodes_wb
; /* writeback inodes */
1602 } __randomize_layout
;
1604 static inline struct user_namespace
*i_user_ns(const struct inode
*inode
)
1606 return inode
->i_sb
->s_user_ns
;
1609 /* Helper functions so that in most cases filesystems will
1610 * not need to deal directly with kuid_t and kgid_t and can
1611 * instead deal with the raw numeric values that are stored
1612 * in the filesystem.
1614 static inline uid_t
i_uid_read(const struct inode
*inode
)
1616 return from_kuid(i_user_ns(inode
), inode
->i_uid
);
1619 static inline gid_t
i_gid_read(const struct inode
*inode
)
1621 return from_kgid(i_user_ns(inode
), inode
->i_gid
);
1624 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
1626 inode
->i_uid
= make_kuid(i_user_ns(inode
), uid
);
1629 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
1631 inode
->i_gid
= make_kgid(i_user_ns(inode
), gid
);
1635 * i_uid_into_vfsuid - map an inode's i_uid down according to an idmapping
1636 * @idmap: idmap of the mount the inode was found from
1637 * @inode: inode to map
1639 * Return: whe inode's i_uid mapped down according to @idmap.
1640 * If the inode's i_uid has no mapping INVALID_VFSUID is returned.
1642 static inline vfsuid_t
i_uid_into_vfsuid(struct mnt_idmap
*idmap
,
1643 const struct inode
*inode
)
1645 return make_vfsuid(idmap
, i_user_ns(inode
), inode
->i_uid
);
1649 * i_uid_needs_update - check whether inode's i_uid needs to be updated
1650 * @idmap: idmap of the mount the inode was found from
1651 * @attr: the new attributes of @inode
1652 * @inode: the inode to update
1654 * Check whether the $inode's i_uid field needs to be updated taking idmapped
1655 * mounts into account if the filesystem supports it.
1657 * Return: true if @inode's i_uid field needs to be updated, false if not.
1659 static inline bool i_uid_needs_update(struct mnt_idmap
*idmap
,
1660 const struct iattr
*attr
,
1661 const struct inode
*inode
)
1663 return ((attr
->ia_valid
& ATTR_UID
) &&
1664 !vfsuid_eq(attr
->ia_vfsuid
,
1665 i_uid_into_vfsuid(idmap
, inode
)));
1669 * i_uid_update - update @inode's i_uid field
1670 * @idmap: idmap of the mount the inode was found from
1671 * @attr: the new attributes of @inode
1672 * @inode: the inode to update
1674 * Safely update @inode's i_uid field translating the vfsuid of any idmapped
1675 * mount into the filesystem kuid.
1677 static inline void i_uid_update(struct mnt_idmap
*idmap
,
1678 const struct iattr
*attr
,
1679 struct inode
*inode
)
1681 if (attr
->ia_valid
& ATTR_UID
)
1682 inode
->i_uid
= from_vfsuid(idmap
, i_user_ns(inode
),
1687 * i_gid_into_vfsgid - map an inode's i_gid down according to an idmapping
1688 * @idmap: idmap of the mount the inode was found from
1689 * @inode: inode to map
1691 * Return: the inode's i_gid mapped down according to @idmap.
1692 * If the inode's i_gid has no mapping INVALID_VFSGID is returned.
1694 static inline vfsgid_t
i_gid_into_vfsgid(struct mnt_idmap
*idmap
,
1695 const struct inode
*inode
)
1697 return make_vfsgid(idmap
, i_user_ns(inode
), inode
->i_gid
);
1701 * i_gid_needs_update - check whether inode's i_gid needs to be updated
1702 * @idmap: idmap of the mount the inode was found from
1703 * @attr: the new attributes of @inode
1704 * @inode: the inode to update
1706 * Check whether the $inode's i_gid field needs to be updated taking idmapped
1707 * mounts into account if the filesystem supports it.
1709 * Return: true if @inode's i_gid field needs to be updated, false if not.
1711 static inline bool i_gid_needs_update(struct mnt_idmap
*idmap
,
1712 const struct iattr
*attr
,
1713 const struct inode
*inode
)
1715 return ((attr
->ia_valid
& ATTR_GID
) &&
1716 !vfsgid_eq(attr
->ia_vfsgid
,
1717 i_gid_into_vfsgid(idmap
, inode
)));
1721 * i_gid_update - update @inode's i_gid field
1722 * @idmap: idmap of the mount the inode was found from
1723 * @attr: the new attributes of @inode
1724 * @inode: the inode to update
1726 * Safely update @inode's i_gid field translating the vfsgid of any idmapped
1727 * mount into the filesystem kgid.
1729 static inline void i_gid_update(struct mnt_idmap
*idmap
,
1730 const struct iattr
*attr
,
1731 struct inode
*inode
)
1733 if (attr
->ia_valid
& ATTR_GID
)
1734 inode
->i_gid
= from_vfsgid(idmap
, i_user_ns(inode
),
1739 * inode_fsuid_set - initialize inode's i_uid field with callers fsuid
1740 * @inode: inode to initialize
1741 * @idmap: idmap of the mount the inode was found from
1743 * Initialize the i_uid field of @inode. If the inode was found/created via
1744 * an idmapped mount map the caller's fsuid according to @idmap.
1746 static inline void inode_fsuid_set(struct inode
*inode
,
1747 struct mnt_idmap
*idmap
)
1749 inode
->i_uid
= mapped_fsuid(idmap
, i_user_ns(inode
));
1753 * inode_fsgid_set - initialize inode's i_gid field with callers fsgid
1754 * @inode: inode to initialize
1755 * @idmap: idmap of the mount the inode was found from
1757 * Initialize the i_gid field of @inode. If the inode was found/created via
1758 * an idmapped mount map the caller's fsgid according to @idmap.
1760 static inline void inode_fsgid_set(struct inode
*inode
,
1761 struct mnt_idmap
*idmap
)
1763 inode
->i_gid
= mapped_fsgid(idmap
, i_user_ns(inode
));
1767 * fsuidgid_has_mapping() - check whether caller's fsuid/fsgid is mapped
1768 * @sb: the superblock we want a mapping in
1769 * @idmap: idmap of the relevant mount
1771 * Check whether the caller's fsuid and fsgid have a valid mapping in the
1772 * s_user_ns of the superblock @sb. If the caller is on an idmapped mount map
1773 * the caller's fsuid and fsgid according to the @idmap first.
1775 * Return: true if fsuid and fsgid is mapped, false if not.
1777 static inline bool fsuidgid_has_mapping(struct super_block
*sb
,
1778 struct mnt_idmap
*idmap
)
1780 struct user_namespace
*fs_userns
= sb
->s_user_ns
;
1784 kuid
= mapped_fsuid(idmap
, fs_userns
);
1785 if (!uid_valid(kuid
))
1787 kgid
= mapped_fsgid(idmap
, fs_userns
);
1788 if (!gid_valid(kgid
))
1790 return kuid_has_mapping(fs_userns
, kuid
) &&
1791 kgid_has_mapping(fs_userns
, kgid
);
1794 extern struct timespec64
current_time(struct inode
*inode
);
1797 * Snapshotting support.
1801 * These are internal functions, please use sb_start_{write,pagefault,intwrite}
1804 static inline void __sb_end_write(struct super_block
*sb
, int level
)
1806 percpu_up_read(sb
->s_writers
.rw_sem
+ level
-1);
1809 static inline void __sb_start_write(struct super_block
*sb
, int level
)
1811 percpu_down_read(sb
->s_writers
.rw_sem
+ level
- 1);
1814 static inline bool __sb_start_write_trylock(struct super_block
*sb
, int level
)
1816 return percpu_down_read_trylock(sb
->s_writers
.rw_sem
+ level
- 1);
1819 #define __sb_writers_acquired(sb, lev) \
1820 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1821 #define __sb_writers_release(sb, lev) \
1822 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1824 static inline bool sb_write_started(const struct super_block
*sb
)
1826 return lockdep_is_held_type(sb
->s_writers
.rw_sem
+ SB_FREEZE_WRITE
- 1, 1);
1830 * sb_end_write - drop write access to a superblock
1831 * @sb: the super we wrote to
1833 * Decrement number of writers to the filesystem. Wake up possible waiters
1834 * wanting to freeze the filesystem.
1836 static inline void sb_end_write(struct super_block
*sb
)
1838 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1842 * sb_end_pagefault - drop write access to a superblock from a page fault
1843 * @sb: the super we wrote to
1845 * Decrement number of processes handling write page fault to the filesystem.
1846 * Wake up possible waiters wanting to freeze the filesystem.
1848 static inline void sb_end_pagefault(struct super_block
*sb
)
1850 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1854 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1855 * @sb: the super we wrote to
1857 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1858 * waiters wanting to freeze the filesystem.
1860 static inline void sb_end_intwrite(struct super_block
*sb
)
1862 __sb_end_write(sb
, SB_FREEZE_FS
);
1866 * sb_start_write - get write access to a superblock
1867 * @sb: the super we write to
1869 * When a process wants to write data or metadata to a file system (i.e. dirty
1870 * a page or an inode), it should embed the operation in a sb_start_write() -
1871 * sb_end_write() pair to get exclusion against file system freezing. This
1872 * function increments number of writers preventing freezing. If the file
1873 * system is already frozen, the function waits until the file system is
1876 * Since freeze protection behaves as a lock, users have to preserve
1877 * ordering of freeze protection and other filesystem locks. Generally,
1878 * freeze protection should be the outermost lock. In particular, we have:
1881 * -> i_mutex (write path, truncate, directory ops, ...)
1882 * -> s_umount (freeze_super, thaw_super)
1884 static inline void sb_start_write(struct super_block
*sb
)
1886 __sb_start_write(sb
, SB_FREEZE_WRITE
);
1889 static inline bool sb_start_write_trylock(struct super_block
*sb
)
1891 return __sb_start_write_trylock(sb
, SB_FREEZE_WRITE
);
1895 * sb_start_pagefault - get write access to a superblock from a page fault
1896 * @sb: the super we write to
1898 * When a process starts handling write page fault, it should embed the
1899 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1900 * exclusion against file system freezing. This is needed since the page fault
1901 * is going to dirty a page. This function increments number of running page
1902 * faults preventing freezing. If the file system is already frozen, the
1903 * function waits until the file system is thawed.
1905 * Since page fault freeze protection behaves as a lock, users have to preserve
1906 * ordering of freeze protection and other filesystem locks. It is advised to
1907 * put sb_start_pagefault() close to mmap_lock in lock ordering. Page fault
1908 * handling code implies lock dependency:
1911 * -> sb_start_pagefault
1913 static inline void sb_start_pagefault(struct super_block
*sb
)
1915 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
);
1919 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1920 * @sb: the super we write to
1922 * This is the third level of protection against filesystem freezing. It is
1923 * free for use by a filesystem. The only requirement is that it must rank
1924 * below sb_start_pagefault.
1926 * For example filesystem can call sb_start_intwrite() when starting a
1927 * transaction which somewhat eases handling of freezing for internal sources
1928 * of filesystem changes (internal fs threads, discarding preallocation on file
1931 static inline void sb_start_intwrite(struct super_block
*sb
)
1933 __sb_start_write(sb
, SB_FREEZE_FS
);
1936 static inline bool sb_start_intwrite_trylock(struct super_block
*sb
)
1938 return __sb_start_write_trylock(sb
, SB_FREEZE_FS
);
1941 bool inode_owner_or_capable(struct mnt_idmap
*idmap
,
1942 const struct inode
*inode
);
1945 * VFS helper functions..
1947 int vfs_create(struct mnt_idmap
*, struct inode
*,
1948 struct dentry
*, umode_t
, bool);
1949 int vfs_mkdir(struct mnt_idmap
*, struct inode
*,
1950 struct dentry
*, umode_t
);
1951 int vfs_mknod(struct mnt_idmap
*, struct inode
*, struct dentry
*,
1953 int vfs_symlink(struct mnt_idmap
*, struct inode
*,
1954 struct dentry
*, const char *);
1955 int vfs_link(struct dentry
*, struct mnt_idmap
*, struct inode
*,
1956 struct dentry
*, struct inode
**);
1957 int vfs_rmdir(struct mnt_idmap
*, struct inode
*, struct dentry
*);
1958 int vfs_unlink(struct mnt_idmap
*, struct inode
*, struct dentry
*,
1962 * struct renamedata - contains all information required for renaming
1963 * @old_mnt_idmap: idmap of the old mount the inode was found from
1964 * @old_dir: parent of source
1965 * @old_dentry: source
1966 * @new_mnt_idmap: idmap of the new mount the inode was found from
1967 * @new_dir: parent of destination
1968 * @new_dentry: destination
1969 * @delegated_inode: returns an inode needing a delegation break
1970 * @flags: rename flags
1973 struct mnt_idmap
*old_mnt_idmap
;
1974 struct inode
*old_dir
;
1975 struct dentry
*old_dentry
;
1976 struct mnt_idmap
*new_mnt_idmap
;
1977 struct inode
*new_dir
;
1978 struct dentry
*new_dentry
;
1979 struct inode
**delegated_inode
;
1981 } __randomize_layout
;
1983 int vfs_rename(struct renamedata
*);
1985 static inline int vfs_whiteout(struct mnt_idmap
*idmap
,
1986 struct inode
*dir
, struct dentry
*dentry
)
1988 return vfs_mknod(idmap
, dir
, dentry
, S_IFCHR
| WHITEOUT_MODE
,
1992 struct file
*vfs_tmpfile_open(struct mnt_idmap
*idmap
,
1993 const struct path
*parentpath
,
1994 umode_t mode
, int open_flag
, const struct cred
*cred
);
1996 int vfs_mkobj(struct dentry
*, umode_t
,
1997 int (*f
)(struct dentry
*, umode_t
, void *),
2000 int vfs_fchown(struct file
*file
, uid_t user
, gid_t group
);
2001 int vfs_fchmod(struct file
*file
, umode_t mode
);
2002 int vfs_utimes(const struct path
*path
, struct timespec64
*times
);
2004 extern long vfs_ioctl(struct file
*file
, unsigned int cmd
, unsigned long arg
);
2006 #ifdef CONFIG_COMPAT
2007 extern long compat_ptr_ioctl(struct file
*file
, unsigned int cmd
,
2010 #define compat_ptr_ioctl NULL
2014 * VFS file helper functions.
2016 void inode_init_owner(struct mnt_idmap
*idmap
, struct inode
*inode
,
2017 const struct inode
*dir
, umode_t mode
);
2018 extern bool may_open_dev(const struct path
*path
);
2019 umode_t
mode_strip_sgid(struct mnt_idmap
*idmap
,
2020 const struct inode
*dir
, umode_t mode
);
2023 * This is the "filldir" function type, used by readdir() to let
2024 * the kernel specify what kind of dirent layout it wants to have.
2025 * This allows the kernel to read directories into kernel space or
2026 * to have different dirent layouts depending on the binary type.
2027 * Return 'true' to keep going and 'false' if there are no more entries.
2030 typedef bool (*filldir_t
)(struct dir_context
*, const char *, int, loff_t
, u64
,
2033 struct dir_context
{
2039 * These flags let !MMU mmap() govern direct device mapping vs immediate
2040 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
2042 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
2043 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
2044 * NOMMU_MAP_READ: Can be mapped for reading
2045 * NOMMU_MAP_WRITE: Can be mapped for writing
2046 * NOMMU_MAP_EXEC: Can be mapped for execution
2048 #define NOMMU_MAP_COPY 0x00000001
2049 #define NOMMU_MAP_DIRECT 0x00000008
2050 #define NOMMU_MAP_READ VM_MAYREAD
2051 #define NOMMU_MAP_WRITE VM_MAYWRITE
2052 #define NOMMU_MAP_EXEC VM_MAYEXEC
2054 #define NOMMU_VMFLAGS \
2055 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
2058 * These flags control the behavior of the remap_file_range function pointer.
2059 * If it is called with len == 0 that means "remap to end of source file".
2060 * See Documentation/filesystems/vfs.rst for more details about this call.
2062 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
2063 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
2065 #define REMAP_FILE_DEDUP (1 << 0)
2066 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
2069 * These flags signal that the caller is ok with altering various aspects of
2070 * the behavior of the remap operation. The changes must be made by the
2071 * implementation; the vfs remap helper functions can take advantage of them.
2072 * Flags in this category exist to preserve the quirky behavior of the hoisted
2073 * btrfs clone/dedupe ioctls.
2075 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
2078 * These flags control the behavior of vfs_copy_file_range().
2079 * They are not available to the user via syscall.
2081 * COPY_FILE_SPLICE: call splice direct instead of fs clone/copy ops
2083 #define COPY_FILE_SPLICE (1 << 0)
2086 struct io_uring_cmd
;
2088 struct file_operations
{
2089 struct module
*owner
;
2090 loff_t (*llseek
) (struct file
*, loff_t
, int);
2091 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
2092 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
2093 ssize_t (*read_iter
) (struct kiocb
*, struct iov_iter
*);
2094 ssize_t (*write_iter
) (struct kiocb
*, struct iov_iter
*);
2095 int (*iopoll
)(struct kiocb
*kiocb
, struct io_comp_batch
*,
2096 unsigned int flags
);
2097 int (*iterate
) (struct file
*, struct dir_context
*);
2098 int (*iterate_shared
) (struct file
*, struct dir_context
*);
2099 __poll_t (*poll
) (struct file
*, struct poll_table_struct
*);
2100 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
2101 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
2102 int (*mmap
) (struct file
*, struct vm_area_struct
*);
2103 unsigned long mmap_supported_flags
;
2104 int (*open
) (struct inode
*, struct file
*);
2105 int (*flush
) (struct file
*, fl_owner_t id
);
2106 int (*release
) (struct inode
*, struct file
*);
2107 int (*fsync
) (struct file
*, loff_t
, loff_t
, int datasync
);
2108 int (*fasync
) (int, struct file
*, int);
2109 int (*lock
) (struct file
*, int, struct file_lock
*);
2110 ssize_t (*sendpage
) (struct file
*, struct page
*, int, size_t, loff_t
*, int);
2111 unsigned long (*get_unmapped_area
)(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
2112 int (*check_flags
)(int);
2113 int (*flock
) (struct file
*, int, struct file_lock
*);
2114 ssize_t (*splice_write
)(struct pipe_inode_info
*, struct file
*, loff_t
*, size_t, unsigned int);
2115 ssize_t (*splice_read
)(struct file
*, loff_t
*, struct pipe_inode_info
*, size_t, unsigned int);
2116 int (*setlease
)(struct file
*, long, struct file_lock
**, void **);
2117 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
2119 void (*show_fdinfo
)(struct seq_file
*m
, struct file
*f
);
2121 unsigned (*mmap_capabilities
)(struct file
*);
2123 ssize_t (*copy_file_range
)(struct file
*, loff_t
, struct file
*,
2124 loff_t
, size_t, unsigned int);
2125 loff_t (*remap_file_range
)(struct file
*file_in
, loff_t pos_in
,
2126 struct file
*file_out
, loff_t pos_out
,
2127 loff_t len
, unsigned int remap_flags
);
2128 int (*fadvise
)(struct file
*, loff_t
, loff_t
, int);
2129 int (*uring_cmd
)(struct io_uring_cmd
*ioucmd
, unsigned int issue_flags
);
2130 int (*uring_cmd_iopoll
)(struct io_uring_cmd
*, struct io_comp_batch
*,
2131 unsigned int poll_flags
);
2132 } __randomize_layout
;
2134 struct inode_operations
{
2135 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*, unsigned int);
2136 const char * (*get_link
) (struct dentry
*, struct inode
*, struct delayed_call
*);
2137 int (*permission
) (struct mnt_idmap
*, struct inode
*, int);
2138 struct posix_acl
* (*get_inode_acl
)(struct inode
*, int, bool);
2140 int (*readlink
) (struct dentry
*, char __user
*,int);
2142 int (*create
) (struct mnt_idmap
*, struct inode
*,struct dentry
*,
2144 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
2145 int (*unlink
) (struct inode
*,struct dentry
*);
2146 int (*symlink
) (struct mnt_idmap
*, struct inode
*,struct dentry
*,
2148 int (*mkdir
) (struct mnt_idmap
*, struct inode
*,struct dentry
*,
2150 int (*rmdir
) (struct inode
*,struct dentry
*);
2151 int (*mknod
) (struct mnt_idmap
*, struct inode
*,struct dentry
*,
2153 int (*rename
) (struct mnt_idmap
*, struct inode
*, struct dentry
*,
2154 struct inode
*, struct dentry
*, unsigned int);
2155 int (*setattr
) (struct mnt_idmap
*, struct dentry
*, struct iattr
*);
2156 int (*getattr
) (struct mnt_idmap
*, const struct path
*,
2157 struct kstat
*, u32
, unsigned int);
2158 ssize_t (*listxattr
) (struct dentry
*, char *, size_t);
2159 int (*fiemap
)(struct inode
*, struct fiemap_extent_info
*, u64 start
,
2161 int (*update_time
)(struct inode
*, struct timespec64
*, int);
2162 int (*atomic_open
)(struct inode
*, struct dentry
*,
2163 struct file
*, unsigned open_flag
,
2164 umode_t create_mode
);
2165 int (*tmpfile
) (struct mnt_idmap
*, struct inode
*,
2166 struct file
*, umode_t
);
2167 struct posix_acl
*(*get_acl
)(struct mnt_idmap
*, struct dentry
*,
2169 int (*set_acl
)(struct mnt_idmap
*, struct dentry
*,
2170 struct posix_acl
*, int);
2171 int (*fileattr_set
)(struct mnt_idmap
*idmap
,
2172 struct dentry
*dentry
, struct fileattr
*fa
);
2173 int (*fileattr_get
)(struct dentry
*dentry
, struct fileattr
*fa
);
2174 } ____cacheline_aligned
;
2176 static inline ssize_t
call_read_iter(struct file
*file
, struct kiocb
*kio
,
2177 struct iov_iter
*iter
)
2179 return file
->f_op
->read_iter(kio
, iter
);
2182 static inline ssize_t
call_write_iter(struct file
*file
, struct kiocb
*kio
,
2183 struct iov_iter
*iter
)
2185 return file
->f_op
->write_iter(kio
, iter
);
2188 static inline int call_mmap(struct file
*file
, struct vm_area_struct
*vma
)
2190 return file
->f_op
->mmap(file
, vma
);
2193 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
2194 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
2195 extern ssize_t
vfs_copy_file_range(struct file
*, loff_t
, struct file
*,
2196 loff_t
, size_t, unsigned int);
2197 extern ssize_t
generic_copy_file_range(struct file
*file_in
, loff_t pos_in
,
2198 struct file
*file_out
, loff_t pos_out
,
2199 size_t len
, unsigned int flags
);
2200 int __generic_remap_file_range_prep(struct file
*file_in
, loff_t pos_in
,
2201 struct file
*file_out
, loff_t pos_out
,
2202 loff_t
*len
, unsigned int remap_flags
,
2203 const struct iomap_ops
*dax_read_ops
);
2204 int generic_remap_file_range_prep(struct file
*file_in
, loff_t pos_in
,
2205 struct file
*file_out
, loff_t pos_out
,
2206 loff_t
*count
, unsigned int remap_flags
);
2207 extern loff_t
do_clone_file_range(struct file
*file_in
, loff_t pos_in
,
2208 struct file
*file_out
, loff_t pos_out
,
2209 loff_t len
, unsigned int remap_flags
);
2210 extern loff_t
vfs_clone_file_range(struct file
*file_in
, loff_t pos_in
,
2211 struct file
*file_out
, loff_t pos_out
,
2212 loff_t len
, unsigned int remap_flags
);
2213 extern int vfs_dedupe_file_range(struct file
*file
,
2214 struct file_dedupe_range
*same
);
2215 extern loff_t
vfs_dedupe_file_range_one(struct file
*src_file
, loff_t src_pos
,
2216 struct file
*dst_file
, loff_t dst_pos
,
2217 loff_t len
, unsigned int remap_flags
);
2220 struct super_operations
{
2221 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
2222 void (*destroy_inode
)(struct inode
*);
2223 void (*free_inode
)(struct inode
*);
2225 void (*dirty_inode
) (struct inode
*, int flags
);
2226 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
2227 int (*drop_inode
) (struct inode
*);
2228 void (*evict_inode
) (struct inode
*);
2229 void (*put_super
) (struct super_block
*);
2230 int (*sync_fs
)(struct super_block
*sb
, int wait
);
2231 int (*freeze_super
) (struct super_block
*);
2232 int (*freeze_fs
) (struct super_block
*);
2233 int (*thaw_super
) (struct super_block
*);
2234 int (*unfreeze_fs
) (struct super_block
*);
2235 int (*statfs
) (struct dentry
*, struct kstatfs
*);
2236 int (*remount_fs
) (struct super_block
*, int *, char *);
2237 void (*umount_begin
) (struct super_block
*);
2239 int (*show_options
)(struct seq_file
*, struct dentry
*);
2240 int (*show_devname
)(struct seq_file
*, struct dentry
*);
2241 int (*show_path
)(struct seq_file
*, struct dentry
*);
2242 int (*show_stats
)(struct seq_file
*, struct dentry
*);
2244 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
2245 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
2246 struct dquot
**(*get_dquots
)(struct inode
*);
2248 long (*nr_cached_objects
)(struct super_block
*,
2249 struct shrink_control
*);
2250 long (*free_cached_objects
)(struct super_block
*,
2251 struct shrink_control
*);
2255 * Inode flags - they have no relation to superblock flags now
2257 #define S_SYNC (1 << 0) /* Writes are synced at once */
2258 #define S_NOATIME (1 << 1) /* Do not update access times */
2259 #define S_APPEND (1 << 2) /* Append-only file */
2260 #define S_IMMUTABLE (1 << 3) /* Immutable file */
2261 #define S_DEAD (1 << 4) /* removed, but still open directory */
2262 #define S_NOQUOTA (1 << 5) /* Inode is not counted to quota */
2263 #define S_DIRSYNC (1 << 6) /* Directory modifications are synchronous */
2264 #define S_NOCMTIME (1 << 7) /* Do not update file c/mtime */
2265 #define S_SWAPFILE (1 << 8) /* Do not truncate: swapon got its bmaps */
2266 #define S_PRIVATE (1 << 9) /* Inode is fs-internal */
2267 #define S_IMA (1 << 10) /* Inode has an associated IMA struct */
2268 #define S_AUTOMOUNT (1 << 11) /* Automount/referral quasi-directory */
2269 #define S_NOSEC (1 << 12) /* no suid or xattr security attributes */
2270 #ifdef CONFIG_FS_DAX
2271 #define S_DAX (1 << 13) /* Direct Access, avoiding the page cache */
2273 #define S_DAX 0 /* Make all the DAX code disappear */
2275 #define S_ENCRYPTED (1 << 14) /* Encrypted file (using fs/crypto/) */
2276 #define S_CASEFOLD (1 << 15) /* Casefolded file */
2277 #define S_VERITY (1 << 16) /* Verity file (using fs/verity/) */
2278 #define S_KERNEL_FILE (1 << 17) /* File is in use by the kernel (eg. fs/cachefiles) */
2281 * Note that nosuid etc flags are inode-specific: setting some file-system
2282 * flags just means all the inodes inherit those flags by default. It might be
2283 * possible to override it selectively if you really wanted to with some
2284 * ioctl() that is not currently implemented.
2286 * Exception: SB_RDONLY is always applied to the entire file system.
2288 * Unfortunately, it is possible to change a filesystems flags with it mounted
2289 * with files in use. This means that all of the inodes will not have their
2290 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
2291 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
2293 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
2295 static inline bool sb_rdonly(const struct super_block
*sb
) { return sb
->s_flags
& SB_RDONLY
; }
2296 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
2297 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
2298 ((inode)->i_flags & S_SYNC))
2299 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
2300 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
2301 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
2302 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
2303 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
2305 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
2306 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
2307 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
2308 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
2310 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
2311 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
2312 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
2313 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
2314 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
2315 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2316 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2317 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2318 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2319 #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD)
2320 #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY)
2322 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2323 (inode)->i_rdev == WHITEOUT_DEV)
2325 static inline bool HAS_UNMAPPED_ID(struct mnt_idmap
*idmap
,
2326 struct inode
*inode
)
2328 return !vfsuid_valid(i_uid_into_vfsuid(idmap
, inode
)) ||
2329 !vfsgid_valid(i_gid_into_vfsgid(idmap
, inode
));
2332 static inline void init_sync_kiocb(struct kiocb
*kiocb
, struct file
*filp
)
2334 *kiocb
= (struct kiocb
) {
2336 .ki_flags
= filp
->f_iocb_flags
,
2337 .ki_ioprio
= get_current_ioprio(),
2341 static inline void kiocb_clone(struct kiocb
*kiocb
, struct kiocb
*kiocb_src
,
2344 *kiocb
= (struct kiocb
) {
2346 .ki_flags
= kiocb_src
->ki_flags
,
2347 .ki_ioprio
= kiocb_src
->ki_ioprio
,
2348 .ki_pos
= kiocb_src
->ki_pos
,
2353 * Inode state bits. Protected by inode->i_lock
2355 * Four bits determine the dirty state of the inode: I_DIRTY_SYNC,
2356 * I_DIRTY_DATASYNC, I_DIRTY_PAGES, and I_DIRTY_TIME.
2358 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2359 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2360 * various stages of removing an inode.
2362 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2364 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2365 * fdatasync() (unless I_DIRTY_DATASYNC is also set).
2366 * Timestamp updates are the usual cause.
2367 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2368 * these changes separately from I_DIRTY_SYNC so that we
2369 * don't have to write inode on fdatasync() when only
2370 * e.g. the timestamps have changed.
2371 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2372 * I_DIRTY_TIME The inode itself has dirty timestamps, and the
2373 * lazytime mount option is enabled. We keep track of this
2374 * separately from I_DIRTY_SYNC in order to implement
2375 * lazytime. This gets cleared if I_DIRTY_INODE
2376 * (I_DIRTY_SYNC and/or I_DIRTY_DATASYNC) gets set. But
2377 * I_DIRTY_TIME can still be set if I_DIRTY_SYNC is already
2378 * in place because writeback might already be in progress
2379 * and we don't want to lose the time update
2380 * I_NEW Serves as both a mutex and completion notification.
2381 * New inodes set I_NEW. If two processes both create
2382 * the same inode, one of them will release its inode and
2383 * wait for I_NEW to be released before returning.
2384 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2385 * also cause waiting on I_NEW, without I_NEW actually
2386 * being set. find_inode() uses this to prevent returning
2387 * nearly-dead inodes.
2388 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2389 * is zero. I_FREEING must be set when I_WILL_FREE is
2391 * I_FREEING Set when inode is about to be freed but still has dirty
2392 * pages or buffers attached or the inode itself is still
2394 * I_CLEAR Added by clear_inode(). In this state the inode is
2395 * clean and can be destroyed. Inode keeps I_FREEING.
2397 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2398 * prohibited for many purposes. iget() must wait for
2399 * the inode to be completely released, then create it
2400 * anew. Other functions will just ignore such inodes,
2401 * if appropriate. I_NEW is used for waiting.
2403 * I_SYNC Writeback of inode is running. The bit is set during
2404 * data writeback, and cleared with a wakeup on the bit
2405 * address once it is done. The bit is also used to pin
2406 * the inode in memory for flusher thread.
2408 * I_REFERENCED Marks the inode as recently references on the LRU list.
2410 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2412 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2413 * synchronize competing switching instances and to tell
2414 * wb stat updates to grab the i_pages lock. See
2415 * inode_switch_wbs_work_fn() for details.
2417 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2418 * and work dirs among overlayfs mounts.
2420 * I_CREATING New object's inode in the middle of setting up.
2422 * I_DONTCACHE Evict inode as soon as it is not used anymore.
2424 * I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists.
2425 * Used to detect that mark_inode_dirty() should not move
2426 * inode between dirty lists.
2428 * I_PINNING_FSCACHE_WB Inode is pinning an fscache object for writeback.
2430 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2432 #define I_DIRTY_SYNC (1 << 0)
2433 #define I_DIRTY_DATASYNC (1 << 1)
2434 #define I_DIRTY_PAGES (1 << 2)
2436 #define I_NEW (1 << __I_NEW)
2437 #define I_WILL_FREE (1 << 4)
2438 #define I_FREEING (1 << 5)
2439 #define I_CLEAR (1 << 6)
2441 #define I_SYNC (1 << __I_SYNC)
2442 #define I_REFERENCED (1 << 8)
2443 #define __I_DIO_WAKEUP 9
2444 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2445 #define I_LINKABLE (1 << 10)
2446 #define I_DIRTY_TIME (1 << 11)
2447 #define I_WB_SWITCH (1 << 13)
2448 #define I_OVL_INUSE (1 << 14)
2449 #define I_CREATING (1 << 15)
2450 #define I_DONTCACHE (1 << 16)
2451 #define I_SYNC_QUEUED (1 << 17)
2452 #define I_PINNING_FSCACHE_WB (1 << 18)
2454 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2455 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2456 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2458 extern void __mark_inode_dirty(struct inode
*, int);
2459 static inline void mark_inode_dirty(struct inode
*inode
)
2461 __mark_inode_dirty(inode
, I_DIRTY
);
2464 static inline void mark_inode_dirty_sync(struct inode
*inode
)
2466 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
2470 * Returns true if the given inode itself only has dirty timestamps (its pages
2471 * may still be dirty) and isn't currently being allocated or freed.
2472 * Filesystems should call this if when writing an inode when lazytime is
2473 * enabled, they want to opportunistically write the timestamps of other inodes
2474 * located very nearby on-disk, e.g. in the same inode block. This returns true
2475 * if the given inode is in need of such an opportunistic update. Requires
2476 * i_lock, or at least later re-checking under i_lock.
2478 static inline bool inode_is_dirtytime_only(struct inode
*inode
)
2480 return (inode
->i_state
& (I_DIRTY_TIME
| I_NEW
|
2481 I_FREEING
| I_WILL_FREE
)) == I_DIRTY_TIME
;
2484 extern void inc_nlink(struct inode
*inode
);
2485 extern void drop_nlink(struct inode
*inode
);
2486 extern void clear_nlink(struct inode
*inode
);
2487 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
2489 static inline void inode_inc_link_count(struct inode
*inode
)
2492 mark_inode_dirty(inode
);
2495 static inline void inode_dec_link_count(struct inode
*inode
)
2498 mark_inode_dirty(inode
);
2501 enum file_time_flags
{
2508 extern bool atime_needs_update(const struct path
*, struct inode
*);
2509 extern void touch_atime(const struct path
*);
2510 int inode_update_time(struct inode
*inode
, struct timespec64
*time
, int flags
);
2512 static inline void file_accessed(struct file
*file
)
2514 if (!(file
->f_flags
& O_NOATIME
))
2515 touch_atime(&file
->f_path
);
2518 extern int file_modified(struct file
*file
);
2519 int kiocb_modified(struct kiocb
*iocb
);
2521 int sync_inode_metadata(struct inode
*inode
, int wait
);
2523 struct file_system_type
{
2526 #define FS_REQUIRES_DEV 1
2527 #define FS_BINARY_MOUNTDATA 2
2528 #define FS_HAS_SUBTYPE 4
2529 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2530 #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */
2531 #define FS_ALLOW_IDMAP 32 /* FS has been updated to handle vfs idmappings. */
2532 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2533 int (*init_fs_context
)(struct fs_context
*);
2534 const struct fs_parameter_spec
*parameters
;
2535 struct dentry
*(*mount
) (struct file_system_type
*, int,
2536 const char *, void *);
2537 void (*kill_sb
) (struct super_block
*);
2538 struct module
*owner
;
2539 struct file_system_type
* next
;
2540 struct hlist_head fs_supers
;
2542 struct lock_class_key s_lock_key
;
2543 struct lock_class_key s_umount_key
;
2544 struct lock_class_key s_vfs_rename_key
;
2545 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
2547 struct lock_class_key i_lock_key
;
2548 struct lock_class_key i_mutex_key
;
2549 struct lock_class_key invalidate_lock_key
;
2550 struct lock_class_key i_mutex_dir_key
;
2553 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2555 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
2556 int flags
, const char *dev_name
, void *data
,
2557 int (*fill_super
)(struct super_block
*, void *, int));
2558 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
2559 int flags
, void *data
,
2560 int (*fill_super
)(struct super_block
*, void *, int));
2561 extern struct dentry
*mount_nodev(struct file_system_type
*fs_type
,
2562 int flags
, void *data
,
2563 int (*fill_super
)(struct super_block
*, void *, int));
2564 extern struct dentry
*mount_subtree(struct vfsmount
*mnt
, const char *path
);
2565 void retire_super(struct super_block
*sb
);
2566 void generic_shutdown_super(struct super_block
*sb
);
2567 void kill_block_super(struct super_block
*sb
);
2568 void kill_anon_super(struct super_block
*sb
);
2569 void kill_litter_super(struct super_block
*sb
);
2570 void deactivate_super(struct super_block
*sb
);
2571 void deactivate_locked_super(struct super_block
*sb
);
2572 int set_anon_super(struct super_block
*s
, void *data
);
2573 int set_anon_super_fc(struct super_block
*s
, struct fs_context
*fc
);
2574 int get_anon_bdev(dev_t
*);
2575 void free_anon_bdev(dev_t
);
2576 struct super_block
*sget_fc(struct fs_context
*fc
,
2577 int (*test
)(struct super_block
*, struct fs_context
*),
2578 int (*set
)(struct super_block
*, struct fs_context
*));
2579 struct super_block
*sget(struct file_system_type
*type
,
2580 int (*test
)(struct super_block
*,void *),
2581 int (*set
)(struct super_block
*,void *),
2582 int flags
, void *data
);
2584 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2585 #define fops_get(fops) \
2586 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2587 #define fops_put(fops) \
2588 do { if (fops) module_put((fops)->owner); } while(0)
2590 * This one is to be used *ONLY* from ->open() instances.
2591 * fops must be non-NULL, pinned down *and* module dependencies
2592 * should be sufficient to pin the caller down as well.
2594 #define replace_fops(f, fops) \
2596 struct file *__file = (f); \
2597 fops_put(__file->f_op); \
2598 BUG_ON(!(__file->f_op = (fops))); \
2601 extern int register_filesystem(struct file_system_type
*);
2602 extern int unregister_filesystem(struct file_system_type
*);
2603 extern int vfs_statfs(const struct path
*, struct kstatfs
*);
2604 extern int user_statfs(const char __user
*, struct kstatfs
*);
2605 extern int fd_statfs(int, struct kstatfs
*);
2606 extern int freeze_super(struct super_block
*super
);
2607 extern int thaw_super(struct super_block
*super
);
2608 extern __printf(2, 3)
2609 int super_setup_bdi_name(struct super_block
*sb
, char *fmt
, ...);
2610 extern int super_setup_bdi(struct super_block
*sb
);
2612 extern int current_umask(void);
2614 extern void ihold(struct inode
* inode
);
2615 extern void iput(struct inode
*);
2616 extern int generic_update_time(struct inode
*, struct timespec64
*, int);
2619 extern struct kobject
*fs_kobj
;
2621 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2623 #ifdef CONFIG_FILE_LOCKING
2624 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2627 * Since this check is lockless, we must ensure that any refcounts
2628 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2629 * could end up racing with tasks trying to set a new lease on this
2633 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2634 return __break_lease(inode
, mode
, FL_LEASE
);
2638 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2641 * Since this check is lockless, we must ensure that any refcounts
2642 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2643 * could end up racing with tasks trying to set a new lease on this
2647 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2648 return __break_lease(inode
, mode
, FL_DELEG
);
2652 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2656 ret
= break_deleg(inode
, O_WRONLY
|O_NONBLOCK
);
2657 if (ret
== -EWOULDBLOCK
&& delegated_inode
) {
2658 *delegated_inode
= inode
;
2664 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2668 ret
= break_deleg(*delegated_inode
, O_WRONLY
);
2669 iput(*delegated_inode
);
2670 *delegated_inode
= NULL
;
2674 static inline int break_layout(struct inode
*inode
, bool wait
)
2677 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2678 return __break_lease(inode
,
2679 wait
? O_WRONLY
: O_WRONLY
| O_NONBLOCK
,
2684 #else /* !CONFIG_FILE_LOCKING */
2685 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2690 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2695 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2700 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2706 static inline int break_layout(struct inode
*inode
, bool wait
)
2711 #endif /* CONFIG_FILE_LOCKING */
2716 const char *name
; /* pointer to actual string */
2717 const __user
char *uptr
; /* original userland pointer */
2719 struct audit_names
*aname
;
2722 static_assert(offsetof(struct filename
, iname
) % sizeof(long) == 0);
2724 static inline struct mnt_idmap
*file_mnt_idmap(struct file
*file
)
2726 return mnt_idmap(file
->f_path
.mnt
);
2730 * is_idmapped_mnt - check whether a mount is mapped
2731 * @mnt: the mount to check
2733 * If @mnt has an non @nop_mnt_idmap attached to it then @mnt is mapped.
2735 * Return: true if mount is mapped, false if not.
2737 static inline bool is_idmapped_mnt(const struct vfsmount
*mnt
)
2739 return mnt_idmap(mnt
) != &nop_mnt_idmap
;
2742 extern long vfs_truncate(const struct path
*, loff_t
);
2743 int do_truncate(struct mnt_idmap
*, struct dentry
*, loff_t start
,
2744 unsigned int time_attrs
, struct file
*filp
);
2745 extern int vfs_fallocate(struct file
*file
, int mode
, loff_t offset
,
2747 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
2749 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
2750 extern struct file
*filp_open(const char *, int, umode_t
);
2751 extern struct file
*file_open_root(const struct path
*,
2752 const char *, int, umode_t
);
2753 static inline struct file
*file_open_root_mnt(struct vfsmount
*mnt
,
2754 const char *name
, int flags
, umode_t mode
)
2756 return file_open_root(&(struct path
){.mnt
= mnt
, .dentry
= mnt
->mnt_root
},
2759 extern struct file
* dentry_open(const struct path
*, int, const struct cred
*);
2760 extern struct file
*dentry_create(const struct path
*path
, int flags
,
2761 umode_t mode
, const struct cred
*cred
);
2762 extern struct file
* open_with_fake_path(const struct path
*, int,
2763 struct inode
*, const struct cred
*);
2764 static inline struct file
*file_clone_open(struct file
*file
)
2766 return dentry_open(&file
->f_path
, file
->f_flags
, file
->f_cred
);
2768 extern int filp_close(struct file
*, fl_owner_t id
);
2770 extern struct filename
*getname_flags(const char __user
*, int, int *);
2771 extern struct filename
*getname_uflags(const char __user
*, int);
2772 extern struct filename
*getname(const char __user
*);
2773 extern struct filename
*getname_kernel(const char *);
2774 extern void putname(struct filename
*name
);
2776 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
2777 int (*open
)(struct inode
*, struct file
*));
2778 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
2780 /* Helper for the simple case when original dentry is used */
2781 static inline int finish_open_simple(struct file
*file
, int error
)
2786 return finish_open(file
, file
->f_path
.dentry
, NULL
);
2790 extern void __init
vfs_caches_init_early(void);
2791 extern void __init
vfs_caches_init(void);
2793 extern struct kmem_cache
*names_cachep
;
2795 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2796 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2798 extern struct super_block
*blockdev_superblock
;
2799 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2801 return IS_ENABLED(CONFIG_BLOCK
) && sb
== blockdev_superblock
;
2804 void emergency_thaw_all(void);
2805 extern int sync_filesystem(struct super_block
*);
2806 extern const struct file_operations def_blk_fops
;
2807 extern const struct file_operations def_chr_fops
;
2810 #define CHRDEV_MAJOR_MAX 512
2811 /* Marks the bottom of the first segment of free char majors */
2812 #define CHRDEV_MAJOR_DYN_END 234
2813 /* Marks the top and bottom of the second segment of free char majors */
2814 #define CHRDEV_MAJOR_DYN_EXT_START 511
2815 #define CHRDEV_MAJOR_DYN_EXT_END 384
2817 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
2818 extern int register_chrdev_region(dev_t
, unsigned, const char *);
2819 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
2820 unsigned int count
, const char *name
,
2821 const struct file_operations
*fops
);
2822 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
2823 unsigned int count
, const char *name
);
2824 extern void unregister_chrdev_region(dev_t
, unsigned);
2825 extern void chrdev_show(struct seq_file
*,off_t
);
2827 static inline int register_chrdev(unsigned int major
, const char *name
,
2828 const struct file_operations
*fops
)
2830 return __register_chrdev(major
, 0, 256, name
, fops
);
2833 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2835 __unregister_chrdev(major
, 0, 256, name
);
2838 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2840 /* Invalid inode operations -- fs/bad_inode.c */
2841 extern void make_bad_inode(struct inode
*);
2842 extern bool is_bad_inode(struct inode
*);
2844 extern int __must_check
file_fdatawait_range(struct file
*file
, loff_t lstart
,
2846 extern int __must_check
file_check_and_advance_wb_err(struct file
*file
);
2847 extern int __must_check
file_write_and_wait_range(struct file
*file
,
2848 loff_t start
, loff_t end
);
2850 static inline int file_write_and_wait(struct file
*file
)
2852 return file_write_and_wait_range(file
, 0, LLONG_MAX
);
2855 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
2857 extern int vfs_fsync(struct file
*file
, int datasync
);
2859 extern int sync_file_range(struct file
*file
, loff_t offset
, loff_t nbytes
,
2860 unsigned int flags
);
2862 static inline bool iocb_is_dsync(const struct kiocb
*iocb
)
2864 return (iocb
->ki_flags
& IOCB_DSYNC
) ||
2865 IS_SYNC(iocb
->ki_filp
->f_mapping
->host
);
2869 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2870 * to already be updated for the write, and will return either the amount
2871 * of bytes passed in, or an error if syncing the file failed.
2873 static inline ssize_t
generic_write_sync(struct kiocb
*iocb
, ssize_t count
)
2875 if (iocb_is_dsync(iocb
)) {
2876 int ret
= vfs_fsync_range(iocb
->ki_filp
,
2877 iocb
->ki_pos
- count
, iocb
->ki_pos
- 1,
2878 (iocb
->ki_flags
& IOCB_SYNC
) ? 0 : 1);
2886 extern void emergency_sync(void);
2887 extern void emergency_remount(void);
2890 extern int bmap(struct inode
*inode
, sector_t
*block
);
2892 static inline int bmap(struct inode
*inode
, sector_t
*block
)
2898 int notify_change(struct mnt_idmap
*, struct dentry
*,
2899 struct iattr
*, struct inode
**);
2900 int inode_permission(struct mnt_idmap
*, struct inode
*, int);
2901 int generic_permission(struct mnt_idmap
*, struct inode
*, int);
2902 static inline int file_permission(struct file
*file
, int mask
)
2904 return inode_permission(file_mnt_idmap(file
),
2905 file_inode(file
), mask
);
2907 static inline int path_permission(const struct path
*path
, int mask
)
2909 return inode_permission(mnt_idmap(path
->mnt
),
2910 d_inode(path
->dentry
), mask
);
2912 int __check_sticky(struct mnt_idmap
*idmap
, struct inode
*dir
,
2913 struct inode
*inode
);
2915 static inline bool execute_ok(struct inode
*inode
)
2917 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
2920 static inline bool inode_wrong_type(const struct inode
*inode
, umode_t mode
)
2922 return (inode
->i_mode
^ mode
) & S_IFMT
;
2925 static inline void file_start_write(struct file
*file
)
2927 if (!S_ISREG(file_inode(file
)->i_mode
))
2929 sb_start_write(file_inode(file
)->i_sb
);
2932 static inline bool file_start_write_trylock(struct file
*file
)
2934 if (!S_ISREG(file_inode(file
)->i_mode
))
2936 return sb_start_write_trylock(file_inode(file
)->i_sb
);
2939 static inline void file_end_write(struct file
*file
)
2941 if (!S_ISREG(file_inode(file
)->i_mode
))
2943 __sb_end_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
);
2947 * This is used for regular files where some users -- especially the
2948 * currently executed binary in a process, previously handled via
2949 * VM_DENYWRITE -- cannot handle concurrent write (and maybe mmap
2950 * read-write shared) accesses.
2952 * get_write_access() gets write permission for a file.
2953 * put_write_access() releases this write permission.
2954 * deny_write_access() denies write access to a file.
2955 * allow_write_access() re-enables write access to a file.
2957 * The i_writecount field of an inode can have the following values:
2958 * 0: no write access, no denied write access
2959 * < 0: (-i_writecount) users that denied write access to the file.
2960 * > 0: (i_writecount) users that have write access to the file.
2962 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2963 * except for the cases where we don't hold i_writecount yet. Then we need to
2964 * use {get,deny}_write_access() - these functions check the sign and refuse
2965 * to do the change if sign is wrong.
2967 static inline int get_write_access(struct inode
*inode
)
2969 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2971 static inline int deny_write_access(struct file
*file
)
2973 struct inode
*inode
= file_inode(file
);
2974 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2976 static inline void put_write_access(struct inode
* inode
)
2978 atomic_dec(&inode
->i_writecount
);
2980 static inline void allow_write_access(struct file
*file
)
2983 atomic_inc(&file_inode(file
)->i_writecount
);
2985 static inline bool inode_is_open_for_write(const struct inode
*inode
)
2987 return atomic_read(&inode
->i_writecount
) > 0;
2990 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
2991 static inline void i_readcount_dec(struct inode
*inode
)
2993 BUG_ON(!atomic_read(&inode
->i_readcount
));
2994 atomic_dec(&inode
->i_readcount
);
2996 static inline void i_readcount_inc(struct inode
*inode
)
2998 atomic_inc(&inode
->i_readcount
);
3001 static inline void i_readcount_dec(struct inode
*inode
)
3005 static inline void i_readcount_inc(struct inode
*inode
)
3010 extern int do_pipe_flags(int *, int);
3012 extern ssize_t
kernel_read(struct file
*, void *, size_t, loff_t
*);
3013 ssize_t
__kernel_read(struct file
*file
, void *buf
, size_t count
, loff_t
*pos
);
3014 extern ssize_t
kernel_write(struct file
*, const void *, size_t, loff_t
*);
3015 extern ssize_t
__kernel_write(struct file
*, const void *, size_t, loff_t
*);
3016 extern struct file
* open_exec(const char *);
3018 /* fs/dcache.c -- generic fs support functions */
3019 extern bool is_subdir(struct dentry
*, struct dentry
*);
3020 extern bool path_is_under(const struct path
*, const struct path
*);
3022 extern char *file_path(struct file
*, char *, int);
3024 #include <linux/err.h>
3026 /* needed for stackable file system support */
3027 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int whence
);
3029 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int whence
);
3031 extern int inode_init_always(struct super_block
*, struct inode
*);
3032 extern void inode_init_once(struct inode
*);
3033 extern void address_space_init_once(struct address_space
*mapping
);
3034 extern struct inode
* igrab(struct inode
*);
3035 extern ino_t
iunique(struct super_block
*, ino_t
);
3036 extern int inode_needs_sync(struct inode
*inode
);
3037 extern int generic_delete_inode(struct inode
*inode
);
3038 static inline int generic_drop_inode(struct inode
*inode
)
3040 return !inode
->i_nlink
|| inode_unhashed(inode
);
3042 extern void d_mark_dontcache(struct inode
*inode
);
3044 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
3045 unsigned long hashval
, int (*test
)(struct inode
*, void *),
3047 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
3048 int (*test
)(struct inode
*, void *), void *data
);
3049 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
3051 extern struct inode
*inode_insert5(struct inode
*inode
, unsigned long hashval
,
3052 int (*test
)(struct inode
*, void *),
3053 int (*set
)(struct inode
*, void *),
3055 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
3056 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
3057 extern struct inode
*find_inode_nowait(struct super_block
*,
3059 int (*match
)(struct inode
*,
3060 unsigned long, void *),
3062 extern struct inode
*find_inode_rcu(struct super_block
*, unsigned long,
3063 int (*)(struct inode
*, void *), void *);
3064 extern struct inode
*find_inode_by_ino_rcu(struct super_block
*, unsigned long);
3065 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
3066 extern int insert_inode_locked(struct inode
*);
3067 #ifdef CONFIG_DEBUG_LOCK_ALLOC
3068 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
3070 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
3072 extern void unlock_new_inode(struct inode
*);
3073 extern void discard_new_inode(struct inode
*);
3074 extern unsigned int get_next_ino(void);
3075 extern void evict_inodes(struct super_block
*sb
);
3076 void dump_mapping(const struct address_space
*);
3079 * Userspace may rely on the the inode number being non-zero. For example, glibc
3080 * simply ignores files with zero i_ino in unlink() and other places.
3082 * As an additional complication, if userspace was compiled with
3083 * _FILE_OFFSET_BITS=32 on a 64-bit kernel we'll only end up reading out the
3084 * lower 32 bits, so we need to check that those aren't zero explicitly. With
3085 * _FILE_OFFSET_BITS=64, this may cause some harmless false-negatives, but
3086 * better safe than sorry.
3088 static inline bool is_zero_ino(ino_t ino
)
3090 return (u32
)ino
== 0;
3093 extern void __iget(struct inode
* inode
);
3094 extern void iget_failed(struct inode
*);
3095 extern void clear_inode(struct inode
*);
3096 extern void __destroy_inode(struct inode
*);
3097 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
3098 extern struct inode
*new_inode(struct super_block
*sb
);
3099 extern void free_inode_nonrcu(struct inode
*inode
);
3100 extern int setattr_should_drop_suidgid(struct mnt_idmap
*, struct inode
*);
3101 extern int file_remove_privs(struct file
*);
3104 * This must be used for allocating filesystems specific inodes to set
3105 * up the inode reclaim context correctly.
3107 static inline void *
3108 alloc_inode_sb(struct super_block
*sb
, struct kmem_cache
*cache
, gfp_t gfp
)
3110 return kmem_cache_alloc_lru(cache
, &sb
->s_inode_lru
, gfp
);
3113 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
3114 static inline void insert_inode_hash(struct inode
*inode
)
3116 __insert_inode_hash(inode
, inode
->i_ino
);
3119 extern void __remove_inode_hash(struct inode
*);
3120 static inline void remove_inode_hash(struct inode
*inode
)
3122 if (!inode_unhashed(inode
) && !hlist_fake(&inode
->i_hash
))
3123 __remove_inode_hash(inode
);
3126 extern void inode_sb_list_add(struct inode
*inode
);
3127 extern void inode_add_lru(struct inode
*inode
);
3129 extern int sb_set_blocksize(struct super_block
*, int);
3130 extern int sb_min_blocksize(struct super_block
*, int);
3132 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
3133 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
3134 extern ssize_t
generic_write_checks(struct kiocb
*, struct iov_iter
*);
3135 int generic_write_checks_count(struct kiocb
*iocb
, loff_t
*count
);
3136 extern int generic_write_check_limits(struct file
*file
, loff_t pos
,
3138 extern int generic_file_rw_checks(struct file
*file_in
, struct file
*file_out
);
3139 ssize_t
filemap_read(struct kiocb
*iocb
, struct iov_iter
*to
,
3140 ssize_t already_read
);
3141 extern ssize_t
generic_file_read_iter(struct kiocb
*, struct iov_iter
*);
3142 extern ssize_t
__generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
3143 extern ssize_t
generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
3144 extern ssize_t
generic_file_direct_write(struct kiocb
*, struct iov_iter
*);
3145 ssize_t
generic_perform_write(struct kiocb
*, struct iov_iter
*);
3147 ssize_t
vfs_iter_read(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
3149 ssize_t
vfs_iter_write(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
3151 ssize_t
vfs_iocb_iter_read(struct file
*file
, struct kiocb
*iocb
,
3152 struct iov_iter
*iter
);
3153 ssize_t
vfs_iocb_iter_write(struct file
*file
, struct kiocb
*iocb
,
3154 struct iov_iter
*iter
);
3157 extern ssize_t
generic_file_splice_read(struct file
*, loff_t
*,
3158 struct pipe_inode_info
*, size_t, unsigned int);
3159 extern ssize_t
iter_file_splice_write(struct pipe_inode_info
*,
3160 struct file
*, loff_t
*, size_t, unsigned int);
3161 extern ssize_t
generic_splice_sendpage(struct pipe_inode_info
*pipe
,
3162 struct file
*out
, loff_t
*, size_t len
, unsigned int flags
);
3163 extern long do_splice_direct(struct file
*in
, loff_t
*ppos
, struct file
*out
,
3164 loff_t
*opos
, size_t len
, unsigned int flags
);
3168 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
3169 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int whence
);
3170 #define no_llseek NULL
3171 extern loff_t
vfs_setpos(struct file
*file
, loff_t offset
, loff_t maxsize
);
3172 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int whence
);
3173 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
3174 int whence
, loff_t maxsize
, loff_t eof
);
3175 extern loff_t
fixed_size_llseek(struct file
*file
, loff_t offset
,
3176 int whence
, loff_t size
);
3177 extern loff_t
no_seek_end_llseek_size(struct file
*, loff_t
, int, loff_t
);
3178 extern loff_t
no_seek_end_llseek(struct file
*, loff_t
, int);
3179 int rw_verify_area(int, struct file
*, const loff_t
*, size_t);
3180 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
3181 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
3182 extern int stream_open(struct inode
* inode
, struct file
* filp
);
3185 typedef void (dio_submit_t
)(struct bio
*bio
, struct inode
*inode
,
3186 loff_t file_offset
);
3189 /* need locking between buffered and direct access */
3192 /* filesystem does not support filling holes */
3193 DIO_SKIP_HOLES
= 0x02,
3196 ssize_t
__blockdev_direct_IO(struct kiocb
*iocb
, struct inode
*inode
,
3197 struct block_device
*bdev
, struct iov_iter
*iter
,
3198 get_block_t get_block
,
3199 dio_iodone_t end_io
, dio_submit_t submit_io
,
3202 static inline ssize_t
blockdev_direct_IO(struct kiocb
*iocb
,
3203 struct inode
*inode
,
3204 struct iov_iter
*iter
,
3205 get_block_t get_block
)
3207 return __blockdev_direct_IO(iocb
, inode
, inode
->i_sb
->s_bdev
, iter
,
3208 get_block
, NULL
, NULL
, DIO_LOCKING
| DIO_SKIP_HOLES
);
3212 void inode_dio_wait(struct inode
*inode
);
3215 * inode_dio_begin - signal start of a direct I/O requests
3216 * @inode: inode the direct I/O happens on
3218 * This is called once we've finished processing a direct I/O request,
3219 * and is used to wake up callers waiting for direct I/O to be quiesced.
3221 static inline void inode_dio_begin(struct inode
*inode
)
3223 atomic_inc(&inode
->i_dio_count
);
3227 * inode_dio_end - signal finish of a direct I/O requests
3228 * @inode: inode the direct I/O happens on
3230 * This is called once we've finished processing a direct I/O request,
3231 * and is used to wake up callers waiting for direct I/O to be quiesced.
3233 static inline void inode_dio_end(struct inode
*inode
)
3235 if (atomic_dec_and_test(&inode
->i_dio_count
))
3236 wake_up_bit(&inode
->i_state
, __I_DIO_WAKEUP
);
3240 * Warn about a page cache invalidation failure diring a direct I/O write.
3242 void dio_warn_stale_pagecache(struct file
*filp
);
3244 extern void inode_set_flags(struct inode
*inode
, unsigned int flags
,
3247 extern const struct file_operations generic_ro_fops
;
3249 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3251 extern int readlink_copy(char __user
*, int, const char *);
3252 extern int page_readlink(struct dentry
*, char __user
*, int);
3253 extern const char *page_get_link(struct dentry
*, struct inode
*,
3254 struct delayed_call
*);
3255 extern void page_put_link(void *);
3256 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
3257 extern const struct inode_operations page_symlink_inode_operations
;
3258 extern void kfree_link(void *);
3259 void generic_fillattr(struct mnt_idmap
*, struct inode
*, struct kstat
*);
3260 void generic_fill_statx_attr(struct inode
*inode
, struct kstat
*stat
);
3261 extern int vfs_getattr_nosec(const struct path
*, struct kstat
*, u32
, unsigned int);
3262 extern int vfs_getattr(const struct path
*, struct kstat
*, u32
, unsigned int);
3263 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
3264 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
3265 void __inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
3266 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
3267 static inline loff_t
__inode_get_bytes(struct inode
*inode
)
3269 return (((loff_t
)inode
->i_blocks
) << 9) + inode
->i_bytes
;
3271 loff_t
inode_get_bytes(struct inode
*inode
);
3272 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
3273 const char *simple_get_link(struct dentry
*, struct inode
*,
3274 struct delayed_call
*);
3275 extern const struct inode_operations simple_symlink_inode_operations
;
3277 extern int iterate_dir(struct file
*, struct dir_context
*);
3279 int vfs_fstatat(int dfd
, const char __user
*filename
, struct kstat
*stat
,
3281 int vfs_fstat(int fd
, struct kstat
*stat
);
3283 static inline int vfs_stat(const char __user
*filename
, struct kstat
*stat
)
3285 return vfs_fstatat(AT_FDCWD
, filename
, stat
, 0);
3287 static inline int vfs_lstat(const char __user
*name
, struct kstat
*stat
)
3289 return vfs_fstatat(AT_FDCWD
, name
, stat
, AT_SYMLINK_NOFOLLOW
);
3292 extern const char *vfs_get_link(struct dentry
*, struct delayed_call
*);
3293 extern int vfs_readlink(struct dentry
*, char __user
*, int);
3295 extern struct file_system_type
*get_filesystem(struct file_system_type
*fs
);
3296 extern void put_filesystem(struct file_system_type
*fs
);
3297 extern struct file_system_type
*get_fs_type(const char *name
);
3298 extern struct super_block
*get_super(struct block_device
*);
3299 extern struct super_block
*get_active_super(struct block_device
*bdev
);
3300 extern void drop_super(struct super_block
*sb
);
3301 extern void drop_super_exclusive(struct super_block
*sb
);
3302 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
3303 extern void iterate_supers_type(struct file_system_type
*,
3304 void (*)(struct super_block
*, void *), void *);
3306 extern int dcache_dir_open(struct inode
*, struct file
*);
3307 extern int dcache_dir_close(struct inode
*, struct file
*);
3308 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
3309 extern int dcache_readdir(struct file
*, struct dir_context
*);
3310 extern int simple_setattr(struct mnt_idmap
*, struct dentry
*,
3312 extern int simple_getattr(struct mnt_idmap
*, const struct path
*,
3313 struct kstat
*, u32
, unsigned int);
3314 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
3315 extern int simple_open(struct inode
*inode
, struct file
*file
);
3316 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
3317 extern int simple_unlink(struct inode
*, struct dentry
*);
3318 extern int simple_rmdir(struct inode
*, struct dentry
*);
3319 extern int simple_rename_exchange(struct inode
*old_dir
, struct dentry
*old_dentry
,
3320 struct inode
*new_dir
, struct dentry
*new_dentry
);
3321 extern int simple_rename(struct mnt_idmap
*, struct inode
*,
3322 struct dentry
*, struct inode
*, struct dentry
*,
3324 extern void simple_recursive_removal(struct dentry
*,
3325 void (*callback
)(struct dentry
*));
3326 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
3327 extern ssize_t
noop_direct_IO(struct kiocb
*iocb
, struct iov_iter
*iter
);
3328 extern int simple_empty(struct dentry
*);
3329 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
3330 loff_t pos
, unsigned len
,
3331 struct page
**pagep
, void **fsdata
);
3332 extern const struct address_space_operations ram_aops
;
3333 extern int always_delete_dentry(const struct dentry
*);
3334 extern struct inode
*alloc_anon_inode(struct super_block
*);
3335 extern int simple_nosetlease(struct file
*, long, struct file_lock
**, void **);
3336 extern const struct dentry_operations simple_dentry_operations
;
3338 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
3339 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
3340 extern const struct file_operations simple_dir_operations
;
3341 extern const struct inode_operations simple_dir_inode_operations
;
3342 extern void make_empty_dir_inode(struct inode
*inode
);
3343 extern bool is_empty_dir_inode(struct inode
*inode
);
3344 struct tree_descr
{ const char *name
; const struct file_operations
*ops
; int mode
; };
3345 struct dentry
*d_alloc_name(struct dentry
*, const char *);
3346 extern int simple_fill_super(struct super_block
*, unsigned long,
3347 const struct tree_descr
*);
3348 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
3349 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
3351 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
3352 loff_t
*ppos
, const void *from
, size_t available
);
3353 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
3354 const void __user
*from
, size_t count
);
3356 extern int __generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3357 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3359 extern int generic_check_addressable(unsigned, u64
);
3361 extern void generic_set_encrypted_ci_d_ops(struct dentry
*dentry
);
3363 int may_setattr(struct mnt_idmap
*idmap
, struct inode
*inode
,
3364 unsigned int ia_valid
);
3365 int setattr_prepare(struct mnt_idmap
*, struct dentry
*, struct iattr
*);
3366 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
3367 void setattr_copy(struct mnt_idmap
*, struct inode
*inode
,
3368 const struct iattr
*attr
);
3370 extern int file_update_time(struct file
*file
);
3372 static inline bool vma_is_dax(const struct vm_area_struct
*vma
)
3374 return vma
->vm_file
&& IS_DAX(vma
->vm_file
->f_mapping
->host
);
3377 static inline bool vma_is_fsdax(struct vm_area_struct
*vma
)
3379 struct inode
*inode
;
3381 if (!IS_ENABLED(CONFIG_FS_DAX
) || !vma
->vm_file
)
3383 if (!vma_is_dax(vma
))
3385 inode
= file_inode(vma
->vm_file
);
3386 if (S_ISCHR(inode
->i_mode
))
3387 return false; /* device-dax */
3391 static inline int iocb_flags(struct file
*file
)
3394 if (file
->f_flags
& O_APPEND
)
3396 if (file
->f_flags
& O_DIRECT
)
3398 if (file
->f_flags
& O_DSYNC
)
3400 if (file
->f_flags
& __O_SYNC
)
3405 static inline int kiocb_set_rw_flags(struct kiocb
*ki
, rwf_t flags
)
3407 int kiocb_flags
= 0;
3409 /* make sure there's no overlap between RWF and private IOCB flags */
3410 BUILD_BUG_ON((__force
int) RWF_SUPPORTED
& IOCB_EVENTFD
);
3414 if (unlikely(flags
& ~RWF_SUPPORTED
))
3417 if (flags
& RWF_NOWAIT
) {
3418 if (!(ki
->ki_filp
->f_mode
& FMODE_NOWAIT
))
3420 kiocb_flags
|= IOCB_NOIO
;
3422 kiocb_flags
|= (__force
int) (flags
& RWF_SUPPORTED
);
3423 if (flags
& RWF_SYNC
)
3424 kiocb_flags
|= IOCB_DSYNC
;
3426 ki
->ki_flags
|= kiocb_flags
;
3430 static inline ino_t
parent_ino(struct dentry
*dentry
)
3435 * Don't strictly need d_lock here? If the parent ino could change
3436 * then surely we'd have a deeper race in the caller?
3438 spin_lock(&dentry
->d_lock
);
3439 res
= dentry
->d_parent
->d_inode
->i_ino
;
3440 spin_unlock(&dentry
->d_lock
);
3444 /* Transaction based IO helpers */
3447 * An argresp is stored in an allocated page and holds the
3448 * size of the argument or response, along with its content
3450 struct simple_transaction_argresp
{
3455 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3457 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
3459 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
3460 size_t size
, loff_t
*pos
);
3461 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
3463 void simple_transaction_set(struct file
*file
, size_t n
);
3466 * simple attribute files
3468 * These attributes behave similar to those in sysfs:
3470 * Writing to an attribute immediately sets a value, an open file can be
3471 * written to multiple times.
3473 * Reading from an attribute creates a buffer from the value that might get
3474 * read with multiple read calls. When the attribute has been read
3475 * completely, no further read calls are possible until the file is opened
3478 * All attributes contain a text representation of a numeric value
3479 * that are accessed with the get() and set() functions.
3481 #define DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, __is_signed) \
3482 static int __fops ## _open(struct inode *inode, struct file *file) \
3484 __simple_attr_check_format(__fmt, 0ull); \
3485 return simple_attr_open(inode, file, __get, __set, __fmt); \
3487 static const struct file_operations __fops = { \
3488 .owner = THIS_MODULE, \
3489 .open = __fops ## _open, \
3490 .release = simple_attr_release, \
3491 .read = simple_attr_read, \
3492 .write = (__is_signed) ? simple_attr_write_signed : simple_attr_write, \
3493 .llseek = generic_file_llseek, \
3496 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3497 DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, false)
3499 #define DEFINE_SIMPLE_ATTRIBUTE_SIGNED(__fops, __get, __set, __fmt) \
3500 DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, true)
3502 static inline __printf(1, 2)
3503 void __simple_attr_check_format(const char *fmt
, ...)
3505 /* don't do anything, just let the compiler check the arguments; */
3508 int simple_attr_open(struct inode
*inode
, struct file
*file
,
3509 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
3511 int simple_attr_release(struct inode
*inode
, struct file
*file
);
3512 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
3513 size_t len
, loff_t
*ppos
);
3514 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
3515 size_t len
, loff_t
*ppos
);
3516 ssize_t
simple_attr_write_signed(struct file
*file
, const char __user
*buf
,
3517 size_t len
, loff_t
*ppos
);
3520 int __init
list_bdev_fs_names(char *buf
, size_t size
);
3522 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3523 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3525 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3526 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3527 (flag & __FMODE_NONOTIFY)))
3529 static inline bool is_sxid(umode_t mode
)
3531 return mode
& (S_ISUID
| S_ISGID
);
3534 static inline int check_sticky(struct mnt_idmap
*idmap
,
3535 struct inode
*dir
, struct inode
*inode
)
3537 if (!(dir
->i_mode
& S_ISVTX
))
3540 return __check_sticky(idmap
, dir
, inode
);
3543 static inline void inode_has_no_xattr(struct inode
*inode
)
3545 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& SB_NOSEC
))
3546 inode
->i_flags
|= S_NOSEC
;
3549 static inline bool is_root_inode(struct inode
*inode
)
3551 return inode
== inode
->i_sb
->s_root
->d_inode
;
3554 static inline bool dir_emit(struct dir_context
*ctx
,
3555 const char *name
, int namelen
,
3556 u64 ino
, unsigned type
)
3558 return ctx
->actor(ctx
, name
, namelen
, ctx
->pos
, ino
, type
);
3560 static inline bool dir_emit_dot(struct file
*file
, struct dir_context
*ctx
)
3562 return ctx
->actor(ctx
, ".", 1, ctx
->pos
,
3563 file
->f_path
.dentry
->d_inode
->i_ino
, DT_DIR
);
3565 static inline bool dir_emit_dotdot(struct file
*file
, struct dir_context
*ctx
)
3567 return ctx
->actor(ctx
, "..", 2, ctx
->pos
,
3568 parent_ino(file
->f_path
.dentry
), DT_DIR
);
3570 static inline bool dir_emit_dots(struct file
*file
, struct dir_context
*ctx
)
3572 if (ctx
->pos
== 0) {
3573 if (!dir_emit_dot(file
, ctx
))
3577 if (ctx
->pos
== 1) {
3578 if (!dir_emit_dotdot(file
, ctx
))
3584 static inline bool dir_relax(struct inode
*inode
)
3586 inode_unlock(inode
);
3588 return !IS_DEADDIR(inode
);
3591 static inline bool dir_relax_shared(struct inode
*inode
)
3593 inode_unlock_shared(inode
);
3594 inode_lock_shared(inode
);
3595 return !IS_DEADDIR(inode
);
3598 extern bool path_noexec(const struct path
*path
);
3599 extern void inode_nohighmem(struct inode
*inode
);
3602 extern int vfs_fadvise(struct file
*file
, loff_t offset
, loff_t len
,
3604 extern int generic_fadvise(struct file
*file
, loff_t offset
, loff_t len
,
3607 #endif /* _LINUX_FS_H */