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/fiemap.h>
28 #include <linux/rculist_bl.h>
29 #include <linux/atomic.h>
30 #include <linux/shrinker.h>
31 #include <linux/migrate_mode.h>
32 #include <linux/uidgid.h>
33 #include <linux/lockdep.h>
34 #include <linux/percpu-rwsem.h>
35 #include <linux/workqueue.h>
36 #include <linux/delayed_call.h>
37 #include <linux/uuid.h>
38 #include <linux/errseq.h>
39 #include <linux/ioprio.h>
40 #include <linux/fs_types.h>
41 #include <linux/build_bug.h>
42 #include <linux/stddef.h>
44 #include <asm/byteorder.h>
45 #include <uapi/linux/fs.h>
47 struct backing_dev_info
;
50 struct export_operations
;
55 struct pipe_inode_info
;
56 struct poll_table_struct
;
58 struct vm_area_struct
;
61 struct swap_info_struct
;
63 struct workqueue_struct
;
66 struct fscrypt_operations
;
68 struct fsverity_operations
;
70 struct fs_parameter_spec
;
72 extern void __init
inode_init(void);
73 extern void __init
inode_init_early(void);
74 extern void __init
files_init(void);
75 extern void __init
files_maxfiles_init(void);
77 extern struct files_stat_struct files_stat
;
78 extern unsigned long get_max_files(void);
79 extern unsigned int sysctl_nr_open
;
80 extern struct inodes_stat_t inodes_stat
;
81 extern int leases_enable
, lease_break_time
;
82 extern int sysctl_protected_symlinks
;
83 extern int sysctl_protected_hardlinks
;
84 extern int sysctl_protected_fifos
;
85 extern int sysctl_protected_regular
;
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/kmem): 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 was opened by fanotify and shouldn't generate fanotify events */
167 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
169 /* File is capable of returning -EAGAIN if I/O will block */
170 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
172 /* File represents mount that needs unmounting */
173 #define FMODE_NEED_UNMOUNT ((__force fmode_t)0x10000000)
175 /* File does not contribute to nr_files count */
176 #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000)
179 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
180 * that indicates that they should check the contents of the iovec are
181 * valid, but not check the memory that the iovec elements
184 #define CHECK_IOVEC_ONLY -1
187 * Attribute flags. These should be or-ed together to figure out what
190 #define ATTR_MODE (1 << 0)
191 #define ATTR_UID (1 << 1)
192 #define ATTR_GID (1 << 2)
193 #define ATTR_SIZE (1 << 3)
194 #define ATTR_ATIME (1 << 4)
195 #define ATTR_MTIME (1 << 5)
196 #define ATTR_CTIME (1 << 6)
197 #define ATTR_ATIME_SET (1 << 7)
198 #define ATTR_MTIME_SET (1 << 8)
199 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
200 #define ATTR_KILL_SUID (1 << 11)
201 #define ATTR_KILL_SGID (1 << 12)
202 #define ATTR_FILE (1 << 13)
203 #define ATTR_KILL_PRIV (1 << 14)
204 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
205 #define ATTR_TIMES_SET (1 << 16)
206 #define ATTR_TOUCH (1 << 17)
209 * Whiteout is represented by a char device. The following constants define the
210 * mode and device number to use.
212 #define WHITEOUT_MODE 0
213 #define WHITEOUT_DEV 0
216 * This is the Inode Attributes structure, used for notify_change(). It
217 * uses the above definitions as flags, to know which values have changed.
218 * Also, in this manner, a Filesystem can look at only the values it cares
219 * about. Basically, these are the attributes that the VFS layer can
220 * request to change from the FS layer.
222 * Derek Atkins <warlord@MIT.EDU> 94-10-20
225 unsigned int ia_valid
;
230 struct timespec64 ia_atime
;
231 struct timespec64 ia_mtime
;
232 struct timespec64 ia_ctime
;
235 * Not an attribute, but an auxiliary info for filesystems wanting to
236 * implement an ftruncate() like method. NOTE: filesystem should
237 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
239 struct file
*ia_file
;
243 * Includes for diskquotas.
245 #include <linux/quota.h>
248 * Maximum number of layers of fs stack. Needs to be limited to
249 * prevent kernel stack overflow
251 #define FILESYSTEM_MAX_STACK_DEPTH 2
254 * enum positive_aop_returns - aop return codes with specific semantics
256 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
257 * completed, that the page is still locked, and
258 * should be considered active. The VM uses this hint
259 * to return the page to the active list -- it won't
260 * be a candidate for writeback again in the near
261 * future. Other callers must be careful to unlock
262 * the page if they get this return. Returned by
265 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
266 * unlocked it and the page might have been truncated.
267 * The caller should back up to acquiring a new page and
268 * trying again. The aop will be taking reasonable
269 * precautions not to livelock. If the caller held a page
270 * reference, it should drop it before retrying. Returned
273 * address_space_operation functions return these large constants to indicate
274 * special semantics to the caller. These are much larger than the bytes in a
275 * page to allow for functions that return the number of bytes operated on in a
279 enum positive_aop_returns
{
280 AOP_WRITEPAGE_ACTIVATE
= 0x80000,
281 AOP_TRUNCATED_PAGE
= 0x80001,
284 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
285 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
286 * helper code (eg buffer layer)
287 * to clear GFP_FS from alloc */
290 * oh the beauties of C type declarations.
293 struct address_space
;
294 struct writeback_control
;
295 struct readahead_control
;
298 * Write life time hint values.
299 * Stored in struct inode as u8.
302 WRITE_LIFE_NOT_SET
= 0,
303 WRITE_LIFE_NONE
= RWH_WRITE_LIFE_NONE
,
304 WRITE_LIFE_SHORT
= RWH_WRITE_LIFE_SHORT
,
305 WRITE_LIFE_MEDIUM
= RWH_WRITE_LIFE_MEDIUM
,
306 WRITE_LIFE_LONG
= RWH_WRITE_LIFE_LONG
,
307 WRITE_LIFE_EXTREME
= RWH_WRITE_LIFE_EXTREME
,
310 #define IOCB_EVENTFD (1 << 0)
311 #define IOCB_APPEND (1 << 1)
312 #define IOCB_DIRECT (1 << 2)
313 #define IOCB_HIPRI (1 << 3)
314 #define IOCB_DSYNC (1 << 4)
315 #define IOCB_SYNC (1 << 5)
316 #define IOCB_WRITE (1 << 6)
317 #define IOCB_NOWAIT (1 << 7)
320 struct file
*ki_filp
;
322 /* The 'ki_filp' pointer is shared in a union for aio */
323 randomized_struct_fields_start
326 void (*ki_complete
)(struct kiocb
*iocb
, long ret
, long ret2
);
330 u16 ki_ioprio
; /* See linux/ioprio.h */
331 unsigned int ki_cookie
; /* for ->iopoll */
333 randomized_struct_fields_end
336 static inline bool is_sync_kiocb(struct kiocb
*kiocb
)
338 return kiocb
->ki_complete
== NULL
;
342 * "descriptor" for what we're up to with a read.
343 * This allows us to use the same read code yet
344 * have multiple different users of the data that
345 * we read from a file.
347 * The simplest case just copies the data to user
360 typedef int (*read_actor_t
)(read_descriptor_t
*, struct page
*,
361 unsigned long, unsigned long);
363 struct address_space_operations
{
364 int (*writepage
)(struct page
*page
, struct writeback_control
*wbc
);
365 int (*readpage
)(struct file
*, struct page
*);
367 /* Write back some dirty pages from this mapping. */
368 int (*writepages
)(struct address_space
*, struct writeback_control
*);
370 /* Set a page dirty. Return true if this dirtied it */
371 int (*set_page_dirty
)(struct page
*page
);
374 * Reads in the requested pages. Unlike ->readpage(), this is
375 * PURELY used for read-ahead!.
377 int (*readpages
)(struct file
*filp
, struct address_space
*mapping
,
378 struct list_head
*pages
, unsigned nr_pages
);
379 void (*readahead
)(struct readahead_control
*);
381 int (*write_begin
)(struct file
*, struct address_space
*mapping
,
382 loff_t pos
, unsigned len
, unsigned flags
,
383 struct page
**pagep
, void **fsdata
);
384 int (*write_end
)(struct file
*, struct address_space
*mapping
,
385 loff_t pos
, unsigned len
, unsigned copied
,
386 struct page
*page
, void *fsdata
);
388 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
389 sector_t (*bmap
)(struct address_space
*, sector_t
);
390 void (*invalidatepage
) (struct page
*, unsigned int, unsigned int);
391 int (*releasepage
) (struct page
*, gfp_t
);
392 void (*freepage
)(struct page
*);
393 ssize_t (*direct_IO
)(struct kiocb
*, struct iov_iter
*iter
);
395 * migrate the contents of a page to the specified target. If
396 * migrate_mode is MIGRATE_ASYNC, it must not block.
398 int (*migratepage
) (struct address_space
*,
399 struct page
*, struct page
*, enum migrate_mode
);
400 bool (*isolate_page
)(struct page
*, isolate_mode_t
);
401 void (*putback_page
)(struct page
*);
402 int (*launder_page
) (struct page
*);
403 int (*is_partially_uptodate
) (struct page
*, unsigned long,
405 void (*is_dirty_writeback
) (struct page
*, bool *, bool *);
406 int (*error_remove_page
)(struct address_space
*, struct page
*);
408 /* swapfile support */
409 int (*swap_activate
)(struct swap_info_struct
*sis
, struct file
*file
,
411 void (*swap_deactivate
)(struct file
*file
);
414 extern const struct address_space_operations empty_aops
;
417 * pagecache_write_begin/pagecache_write_end must be used by general code
418 * to write into the pagecache.
420 int pagecache_write_begin(struct file
*, struct address_space
*mapping
,
421 loff_t pos
, unsigned len
, unsigned flags
,
422 struct page
**pagep
, void **fsdata
);
424 int pagecache_write_end(struct file
*, struct address_space
*mapping
,
425 loff_t pos
, unsigned len
, unsigned copied
,
426 struct page
*page
, void *fsdata
);
429 * struct address_space - Contents of a cacheable, mappable object.
430 * @host: Owner, either the inode or the block_device.
431 * @i_pages: Cached pages.
432 * @gfp_mask: Memory allocation flags to use for allocating pages.
433 * @i_mmap_writable: Number of VM_SHARED mappings.
434 * @nr_thps: Number of THPs in the pagecache (non-shmem only).
435 * @i_mmap: Tree of private and shared mappings.
436 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
437 * @nrpages: Number of page entries, protected by the i_pages lock.
438 * @nrexceptional: Shadow or DAX entries, protected by the i_pages lock.
439 * @writeback_index: Writeback starts here.
441 * @flags: Error bits and flags (AS_*).
442 * @wb_err: The most recent error which has occurred.
443 * @private_lock: For use by the owner of the address_space.
444 * @private_list: For use by the owner of the address_space.
445 * @private_data: For use by the owner of the address_space.
447 struct address_space
{
449 struct xarray i_pages
;
451 atomic_t i_mmap_writable
;
452 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
453 /* number of thp, only for non-shmem files */
456 struct rb_root_cached i_mmap
;
457 struct rw_semaphore i_mmap_rwsem
;
458 unsigned long nrpages
;
459 unsigned long nrexceptional
;
460 pgoff_t writeback_index
;
461 const struct address_space_operations
*a_ops
;
464 spinlock_t private_lock
;
465 struct list_head private_list
;
467 } __attribute__((aligned(sizeof(long)))) __randomize_layout
;
469 * On most architectures that alignment is already the case; but
470 * must be enforced here for CRIS, to let the least significant bit
471 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
473 struct request_queue
;
475 struct block_device
{
476 dev_t bd_dev
; /* not a kdev_t - it's a search key */
478 struct inode
* bd_inode
; /* will die */
479 struct super_block
* bd_super
;
480 struct mutex bd_mutex
; /* open/close mutex */
484 bool bd_write_holder
;
486 struct list_head bd_holder_disks
;
488 struct block_device
* bd_contains
;
489 unsigned bd_block_size
;
491 struct hd_struct
* bd_part
;
492 /* number of times partitions within this device have been opened. */
493 unsigned bd_part_count
;
495 struct gendisk
* bd_disk
;
496 struct request_queue
* bd_queue
;
497 struct backing_dev_info
*bd_bdi
;
498 struct list_head bd_list
;
500 * Private data. You must have bd_claim'ed the block_device
501 * to use this. NOTE: bd_claim allows an owner to claim
502 * the same device multiple times, the owner must take special
503 * care to not mess up bd_private for that case.
505 unsigned long bd_private
;
507 /* The counter of freeze processes */
508 int bd_fsfreeze_count
;
509 /* Mutex for freeze */
510 struct mutex bd_fsfreeze_mutex
;
511 } __randomize_layout
;
513 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
514 #define PAGECACHE_TAG_DIRTY XA_MARK_0
515 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
516 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
519 * Returns true if any of the pages in the mapping are marked with the tag.
521 static inline bool mapping_tagged(struct address_space
*mapping
, xa_mark_t tag
)
523 return xa_marked(&mapping
->i_pages
, tag
);
526 static inline void i_mmap_lock_write(struct address_space
*mapping
)
528 down_write(&mapping
->i_mmap_rwsem
);
531 static inline int i_mmap_trylock_write(struct address_space
*mapping
)
533 return down_write_trylock(&mapping
->i_mmap_rwsem
);
536 static inline void i_mmap_unlock_write(struct address_space
*mapping
)
538 up_write(&mapping
->i_mmap_rwsem
);
541 static inline void i_mmap_lock_read(struct address_space
*mapping
)
543 down_read(&mapping
->i_mmap_rwsem
);
546 static inline void i_mmap_unlock_read(struct address_space
*mapping
)
548 up_read(&mapping
->i_mmap_rwsem
);
552 * Might pages of this file be mapped into userspace?
554 static inline int mapping_mapped(struct address_space
*mapping
)
556 return !RB_EMPTY_ROOT(&mapping
->i_mmap
.rb_root
);
560 * Might pages of this file have been modified in userspace?
561 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
562 * marks vma as VM_SHARED if it is shared, and the file was opened for
563 * writing i.e. vma may be mprotected writable even if now readonly.
565 * If i_mmap_writable is negative, no new writable mappings are allowed. You
566 * can only deny writable mappings, if none exists right now.
568 static inline int mapping_writably_mapped(struct address_space
*mapping
)
570 return atomic_read(&mapping
->i_mmap_writable
) > 0;
573 static inline int mapping_map_writable(struct address_space
*mapping
)
575 return atomic_inc_unless_negative(&mapping
->i_mmap_writable
) ?
579 static inline void mapping_unmap_writable(struct address_space
*mapping
)
581 atomic_dec(&mapping
->i_mmap_writable
);
584 static inline int mapping_deny_writable(struct address_space
*mapping
)
586 return atomic_dec_unless_positive(&mapping
->i_mmap_writable
) ?
590 static inline void mapping_allow_writable(struct address_space
*mapping
)
592 atomic_inc(&mapping
->i_mmap_writable
);
596 * Use sequence counter to get consistent i_size on 32-bit processors.
598 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
599 #include <linux/seqlock.h>
600 #define __NEED_I_SIZE_ORDERED
601 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
603 #define i_size_ordered_init(inode) do { } while (0)
607 #define ACL_NOT_CACHED ((void *)(-1))
608 #define ACL_DONT_CACHE ((void *)(-3))
610 static inline struct posix_acl
*
611 uncached_acl_sentinel(struct task_struct
*task
)
613 return (void *)task
+ 1;
617 is_uncached_acl(struct posix_acl
*acl
)
619 return (long)acl
& 1;
622 #define IOP_FASTPERM 0x0001
623 #define IOP_LOOKUP 0x0002
624 #define IOP_NOFOLLOW 0x0004
625 #define IOP_XATTR 0x0008
626 #define IOP_DEFAULT_READLINK 0x0010
628 struct fsnotify_mark_connector
;
631 * Keep mostly read-only and often accessed (especially for
632 * the RCU path lookup and 'stat' data) fields at the beginning
633 * of the 'struct inode'
637 unsigned short i_opflags
;
640 unsigned int i_flags
;
642 #ifdef CONFIG_FS_POSIX_ACL
643 struct posix_acl
*i_acl
;
644 struct posix_acl
*i_default_acl
;
647 const struct inode_operations
*i_op
;
648 struct super_block
*i_sb
;
649 struct address_space
*i_mapping
;
651 #ifdef CONFIG_SECURITY
655 /* Stat data, not accessed from path walking */
658 * Filesystems may only read i_nlink directly. They shall use the
659 * following functions for modification:
661 * (set|clear|inc|drop)_nlink
662 * inode_(inc|dec)_link_count
665 const unsigned int i_nlink
;
666 unsigned int __i_nlink
;
670 struct timespec64 i_atime
;
671 struct timespec64 i_mtime
;
672 struct timespec64 i_ctime
;
673 spinlock_t i_lock
; /* i_blocks, i_bytes, maybe i_size */
674 unsigned short i_bytes
;
679 #ifdef __NEED_I_SIZE_ORDERED
680 seqcount_t i_size_seqcount
;
684 unsigned long i_state
;
685 struct rw_semaphore i_rwsem
;
687 unsigned long dirtied_when
; /* jiffies of first dirtying */
688 unsigned long dirtied_time_when
;
690 struct hlist_node i_hash
;
691 struct list_head i_io_list
; /* backing dev IO list */
692 #ifdef CONFIG_CGROUP_WRITEBACK
693 struct bdi_writeback
*i_wb
; /* the associated cgroup wb */
695 /* foreign inode detection, see wbc_detach_inode() */
697 u16 i_wb_frn_avg_time
;
698 u16 i_wb_frn_history
;
700 struct list_head i_lru
; /* inode LRU list */
701 struct list_head i_sb_list
;
702 struct list_head i_wb_list
; /* backing dev writeback list */
704 struct hlist_head i_dentry
;
705 struct rcu_head i_rcu
;
707 atomic64_t i_version
;
708 atomic64_t i_sequence
; /* see futex */
710 atomic_t i_dio_count
;
711 atomic_t i_writecount
;
712 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
713 atomic_t i_readcount
; /* struct files open RO */
716 const struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
717 void (*free_inode
)(struct inode
*);
719 struct file_lock_context
*i_flctx
;
720 struct address_space i_data
;
721 struct list_head i_devices
;
723 struct pipe_inode_info
*i_pipe
;
724 struct block_device
*i_bdev
;
732 #ifdef CONFIG_FSNOTIFY
733 __u32 i_fsnotify_mask
; /* all events this inode cares about */
734 struct fsnotify_mark_connector __rcu
*i_fsnotify_marks
;
737 #ifdef CONFIG_FS_ENCRYPTION
738 struct fscrypt_info
*i_crypt_info
;
741 #ifdef CONFIG_FS_VERITY
742 struct fsverity_info
*i_verity_info
;
745 void *i_private
; /* fs or device private pointer */
746 } __randomize_layout
;
748 struct timespec64
timestamp_truncate(struct timespec64 t
, struct inode
*inode
);
750 static inline unsigned int i_blocksize(const struct inode
*node
)
752 return (1 << node
->i_blkbits
);
755 static inline int inode_unhashed(struct inode
*inode
)
757 return hlist_unhashed(&inode
->i_hash
);
761 * __mark_inode_dirty expects inodes to be hashed. Since we don't
762 * want special inodes in the fileset inode space, we make them
763 * appear hashed, but do not put on any lists. hlist_del()
764 * will work fine and require no locking.
766 static inline void inode_fake_hash(struct inode
*inode
)
768 hlist_add_fake(&inode
->i_hash
);
772 * inode->i_mutex nesting subclasses for the lock validator:
774 * 0: the object of the current VFS operation
778 * 4: second non-directory
779 * 5: second parent (when locking independent directories in rename)
781 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
782 * non-directories at once.
784 * The locking order between these classes is
785 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
787 enum inode_i_mutex_lock_class
797 static inline void inode_lock(struct inode
*inode
)
799 down_write(&inode
->i_rwsem
);
802 static inline void inode_unlock(struct inode
*inode
)
804 up_write(&inode
->i_rwsem
);
807 static inline void inode_lock_shared(struct inode
*inode
)
809 down_read(&inode
->i_rwsem
);
812 static inline void inode_unlock_shared(struct inode
*inode
)
814 up_read(&inode
->i_rwsem
);
817 static inline int inode_trylock(struct inode
*inode
)
819 return down_write_trylock(&inode
->i_rwsem
);
822 static inline int inode_trylock_shared(struct inode
*inode
)
824 return down_read_trylock(&inode
->i_rwsem
);
827 static inline int inode_is_locked(struct inode
*inode
)
829 return rwsem_is_locked(&inode
->i_rwsem
);
832 static inline void inode_lock_nested(struct inode
*inode
, unsigned subclass
)
834 down_write_nested(&inode
->i_rwsem
, subclass
);
837 static inline void inode_lock_shared_nested(struct inode
*inode
, unsigned subclass
)
839 down_read_nested(&inode
->i_rwsem
, subclass
);
842 void lock_two_nondirectories(struct inode
*, struct inode
*);
843 void unlock_two_nondirectories(struct inode
*, struct inode
*);
846 * NOTE: in a 32bit arch with a preemptable kernel and
847 * an UP compile the i_size_read/write must be atomic
848 * with respect to the local cpu (unlike with preempt disabled),
849 * but they don't need to be atomic with respect to other cpus like in
850 * true SMP (so they need either to either locally disable irq around
851 * the read or for example on x86 they can be still implemented as a
852 * cmpxchg8b without the need of the lock prefix). For SMP compiles
853 * and 64bit archs it makes no difference if preempt is enabled or not.
855 static inline loff_t
i_size_read(const struct inode
*inode
)
857 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
862 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
863 i_size
= inode
->i_size
;
864 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
866 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
870 i_size
= inode
->i_size
;
874 return inode
->i_size
;
879 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
880 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
881 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
883 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
885 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
887 write_seqcount_begin(&inode
->i_size_seqcount
);
888 inode
->i_size
= i_size
;
889 write_seqcount_end(&inode
->i_size_seqcount
);
891 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
893 inode
->i_size
= i_size
;
896 inode
->i_size
= i_size
;
900 static inline unsigned iminor(const struct inode
*inode
)
902 return MINOR(inode
->i_rdev
);
905 static inline unsigned imajor(const struct inode
*inode
)
907 return MAJOR(inode
->i_rdev
);
910 extern struct block_device
*I_BDEV(struct inode
*inode
);
913 rwlock_t lock
; /* protects pid, uid, euid fields */
914 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
915 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
916 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
917 int signum
; /* posix.1b rt signal to be delivered on IO */
921 * Track a single file's readahead state
923 struct file_ra_state
{
924 pgoff_t start
; /* where readahead started */
925 unsigned int size
; /* # of readahead pages */
926 unsigned int async_size
; /* do asynchronous readahead when
927 there are only # of pages ahead */
929 unsigned int ra_pages
; /* Maximum readahead window */
930 unsigned int mmap_miss
; /* Cache miss stat for mmap accesses */
931 loff_t prev_pos
; /* Cache last read() position */
935 * Check if @index falls in the readahead windows.
937 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
939 return (index
>= ra
->start
&&
940 index
< ra
->start
+ ra
->size
);
945 struct llist_node fu_llist
;
946 struct rcu_head fu_rcuhead
;
949 struct inode
*f_inode
; /* cached value */
950 const struct file_operations
*f_op
;
953 * Protects f_ep_links, f_flags.
954 * Must not be taken from IRQ context.
957 enum rw_hint f_write_hint
;
958 atomic_long_t f_count
;
959 unsigned int f_flags
;
961 struct mutex f_pos_lock
;
963 struct fown_struct f_owner
;
964 const struct cred
*f_cred
;
965 struct file_ra_state f_ra
;
968 #ifdef CONFIG_SECURITY
971 /* needed for tty driver, and maybe others */
975 /* Used by fs/eventpoll.c to link all the hooks to this file */
976 struct list_head f_ep_links
;
977 struct list_head f_tfile_llink
;
978 #endif /* #ifdef CONFIG_EPOLL */
979 struct address_space
*f_mapping
;
981 errseq_t f_sb_err
; /* for syncfs */
983 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
988 /* file identifier */
989 unsigned char f_handle
[];
992 static inline struct file
*get_file(struct file
*f
)
994 atomic_long_inc(&f
->f_count
);
997 #define get_file_rcu_many(x, cnt) \
998 atomic_long_add_unless(&(x)->f_count, (cnt), 0)
999 #define get_file_rcu(x) get_file_rcu_many((x), 1)
1000 #define file_count(x) atomic_long_read(&(x)->f_count)
1002 #define MAX_NON_LFS ((1UL<<31) - 1)
1004 /* Page cache limit. The filesystems should put that into their s_maxbytes
1005 limits, otherwise bad things can happen in VM. */
1006 #if BITS_PER_LONG==32
1007 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
1008 #elif BITS_PER_LONG==64
1009 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
1014 #define FL_DELEG 4 /* NFSv4 delegation */
1015 #define FL_ACCESS 8 /* not trying to lock, just looking */
1016 #define FL_EXISTS 16 /* when unlocking, test for existence */
1017 #define FL_LEASE 32 /* lease held on this file */
1018 #define FL_CLOSE 64 /* unlock on close */
1019 #define FL_SLEEP 128 /* A blocking lock */
1020 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
1021 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
1022 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
1023 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
1025 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
1028 * Special return value from posix_lock_file() and vfs_lock_file() for
1029 * asynchronous locking.
1031 #define FILE_LOCK_DEFERRED 1
1033 /* legacy typedef, should eventually be removed */
1034 typedef void *fl_owner_t
;
1038 struct file_lock_operations
{
1039 void (*fl_copy_lock
)(struct file_lock
*, struct file_lock
*);
1040 void (*fl_release_private
)(struct file_lock
*);
1043 struct lock_manager_operations
{
1044 fl_owner_t (*lm_get_owner
)(fl_owner_t
);
1045 void (*lm_put_owner
)(fl_owner_t
);
1046 void (*lm_notify
)(struct file_lock
*); /* unblock callback */
1047 int (*lm_grant
)(struct file_lock
*, int);
1048 bool (*lm_break
)(struct file_lock
*);
1049 int (*lm_change
)(struct file_lock
*, int, struct list_head
*);
1050 void (*lm_setup
)(struct file_lock
*, void **);
1053 struct lock_manager
{
1054 struct list_head list
;
1056 * NFSv4 and up also want opens blocked during the grace period;
1063 void locks_start_grace(struct net
*, struct lock_manager
*);
1064 void locks_end_grace(struct lock_manager
*);
1065 bool locks_in_grace(struct net
*);
1066 bool opens_in_grace(struct net
*);
1068 /* that will die - we need it for nfs_lock_info */
1069 #include <linux/nfs_fs_i.h>
1072 * struct file_lock represents a generic "file lock". It's used to represent
1073 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1074 * note that the same struct is used to represent both a request for a lock and
1075 * the lock itself, but the same object is never used for both.
1077 * FIXME: should we create a separate "struct lock_request" to help distinguish
1080 * The varous i_flctx lists are ordered by:
1083 * 2) lock range start
1086 * Obviously, the last two criteria only matter for POSIX locks.
1089 struct file_lock
*fl_blocker
; /* The lock, that is blocking us */
1090 struct list_head fl_list
; /* link into file_lock_context */
1091 struct hlist_node fl_link
; /* node in global lists */
1092 struct list_head fl_blocked_requests
; /* list of requests with
1093 * ->fl_blocker pointing here
1095 struct list_head fl_blocked_member
; /* node in
1096 * ->fl_blocker->fl_blocked_requests
1098 fl_owner_t fl_owner
;
1099 unsigned int fl_flags
;
1100 unsigned char fl_type
;
1101 unsigned int fl_pid
;
1102 int fl_link_cpu
; /* what cpu's list is this on? */
1103 wait_queue_head_t fl_wait
;
1104 struct file
*fl_file
;
1108 struct fasync_struct
* fl_fasync
; /* for lease break notifications */
1109 /* for lease breaks: */
1110 unsigned long fl_break_time
;
1111 unsigned long fl_downgrade_time
;
1113 const struct file_lock_operations
*fl_ops
; /* Callbacks for filesystems */
1114 const struct lock_manager_operations
*fl_lmops
; /* Callbacks for lockmanagers */
1116 struct nfs_lock_info nfs_fl
;
1117 struct nfs4_lock_info nfs4_fl
;
1119 struct list_head link
; /* link in AFS vnode's pending_locks list */
1120 int state
; /* state of grant or error if -ve */
1121 unsigned int debug_id
;
1124 } __randomize_layout
;
1126 struct file_lock_context
{
1127 spinlock_t flc_lock
;
1128 struct list_head flc_flock
;
1129 struct list_head flc_posix
;
1130 struct list_head flc_lease
;
1133 /* The following constant reflects the upper bound of the file/locking space */
1135 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1136 #define OFFSET_MAX INT_LIMIT(loff_t)
1137 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1140 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
1142 #define locks_inode(f) file_inode(f)
1144 #ifdef CONFIG_FILE_LOCKING
1145 extern int fcntl_getlk(struct file
*, unsigned int, struct flock
*);
1146 extern int fcntl_setlk(unsigned int, struct file
*, unsigned int,
1149 #if BITS_PER_LONG == 32
1150 extern int fcntl_getlk64(struct file
*, unsigned int, struct flock64
*);
1151 extern int fcntl_setlk64(unsigned int, struct file
*, unsigned int,
1155 extern int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
);
1156 extern int fcntl_getlease(struct file
*filp
);
1159 void locks_free_lock_context(struct inode
*inode
);
1160 void locks_free_lock(struct file_lock
*fl
);
1161 extern void locks_init_lock(struct file_lock
*);
1162 extern struct file_lock
* locks_alloc_lock(void);
1163 extern void locks_copy_lock(struct file_lock
*, struct file_lock
*);
1164 extern void locks_copy_conflock(struct file_lock
*, struct file_lock
*);
1165 extern void locks_remove_posix(struct file
*, fl_owner_t
);
1166 extern void locks_remove_file(struct file
*);
1167 extern void locks_release_private(struct file_lock
*);
1168 extern void posix_test_lock(struct file
*, struct file_lock
*);
1169 extern int posix_lock_file(struct file
*, struct file_lock
*, struct file_lock
*);
1170 extern int locks_delete_block(struct file_lock
*);
1171 extern int vfs_test_lock(struct file
*, struct file_lock
*);
1172 extern int vfs_lock_file(struct file
*, unsigned int, struct file_lock
*, struct file_lock
*);
1173 extern int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
);
1174 extern int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
);
1175 extern int __break_lease(struct inode
*inode
, unsigned int flags
, unsigned int type
);
1176 extern void lease_get_mtime(struct inode
*, struct timespec64
*time
);
1177 extern int generic_setlease(struct file
*, long, struct file_lock
**, void **priv
);
1178 extern int vfs_setlease(struct file
*, long, struct file_lock
**, void **);
1179 extern int lease_modify(struct file_lock
*, int, struct list_head
*);
1181 struct notifier_block
;
1182 extern int lease_register_notifier(struct notifier_block
*);
1183 extern void lease_unregister_notifier(struct notifier_block
*);
1185 struct files_struct
;
1186 extern void show_fd_locks(struct seq_file
*f
,
1187 struct file
*filp
, struct files_struct
*files
);
1188 #else /* !CONFIG_FILE_LOCKING */
1189 static inline int fcntl_getlk(struct file
*file
, unsigned int cmd
,
1190 struct flock __user
*user
)
1195 static inline int fcntl_setlk(unsigned int fd
, struct file
*file
,
1196 unsigned int cmd
, struct flock __user
*user
)
1201 #if BITS_PER_LONG == 32
1202 static inline int fcntl_getlk64(struct file
*file
, unsigned int cmd
,
1203 struct flock64 __user
*user
)
1208 static inline int fcntl_setlk64(unsigned int fd
, struct file
*file
,
1209 unsigned int cmd
, struct flock64 __user
*user
)
1214 static inline int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
)
1219 static inline int fcntl_getlease(struct file
*filp
)
1225 locks_free_lock_context(struct inode
*inode
)
1229 static inline void locks_init_lock(struct file_lock
*fl
)
1234 static inline void locks_copy_conflock(struct file_lock
*new, struct file_lock
*fl
)
1239 static inline void locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
1244 static inline void locks_remove_posix(struct file
*filp
, fl_owner_t owner
)
1249 static inline void locks_remove_file(struct file
*filp
)
1254 static inline void posix_test_lock(struct file
*filp
, struct file_lock
*fl
)
1259 static inline int posix_lock_file(struct file
*filp
, struct file_lock
*fl
,
1260 struct file_lock
*conflock
)
1265 static inline int locks_delete_block(struct file_lock
*waiter
)
1270 static inline int vfs_test_lock(struct file
*filp
, struct file_lock
*fl
)
1275 static inline int vfs_lock_file(struct file
*filp
, unsigned int cmd
,
1276 struct file_lock
*fl
, struct file_lock
*conf
)
1281 static inline int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
)
1286 static inline int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
)
1291 static inline int __break_lease(struct inode
*inode
, unsigned int mode
, unsigned int type
)
1296 static inline void lease_get_mtime(struct inode
*inode
,
1297 struct timespec64
*time
)
1302 static inline int generic_setlease(struct file
*filp
, long arg
,
1303 struct file_lock
**flp
, void **priv
)
1308 static inline int vfs_setlease(struct file
*filp
, long arg
,
1309 struct file_lock
**lease
, void **priv
)
1314 static inline int lease_modify(struct file_lock
*fl
, int arg
,
1315 struct list_head
*dispose
)
1320 struct files_struct
;
1321 static inline void show_fd_locks(struct seq_file
*f
,
1322 struct file
*filp
, struct files_struct
*files
) {}
1323 #endif /* !CONFIG_FILE_LOCKING */
1325 static inline struct inode
*file_inode(const struct file
*f
)
1330 static inline struct dentry
*file_dentry(const struct file
*file
)
1332 return d_real(file
->f_path
.dentry
, file_inode(file
));
1335 static inline int locks_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1337 return locks_lock_inode_wait(locks_inode(filp
), fl
);
1340 struct fasync_struct
{
1344 struct fasync_struct
*fa_next
; /* singly linked list */
1345 struct file
*fa_file
;
1346 struct rcu_head fa_rcu
;
1349 #define FASYNC_MAGIC 0x4601
1351 /* SMP safe fasync helpers: */
1352 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1353 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1354 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1355 extern struct fasync_struct
*fasync_alloc(void);
1356 extern void fasync_free(struct fasync_struct
*);
1358 /* can be called from interrupts */
1359 extern void kill_fasync(struct fasync_struct
**, int, int);
1361 extern void __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1362 extern int f_setown(struct file
*filp
, unsigned long arg
, int force
);
1363 extern void f_delown(struct file
*filp
);
1364 extern pid_t
f_getown(struct file
*filp
);
1365 extern int send_sigurg(struct fown_struct
*fown
);
1368 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1369 * represented in both.
1371 #define SB_RDONLY 1 /* Mount read-only */
1372 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1373 #define SB_NODEV 4 /* Disallow access to device special files */
1374 #define SB_NOEXEC 8 /* Disallow program execution */
1375 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1376 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1377 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1378 #define SB_NOATIME 1024 /* Do not update access times. */
1379 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1380 #define SB_SILENT 32768
1381 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1382 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1383 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1384 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1386 /* These sb flags are internal to the kernel */
1387 #define SB_SUBMOUNT (1<<26)
1388 #define SB_FORCE (1<<27)
1389 #define SB_NOSEC (1<<28)
1390 #define SB_BORN (1<<29)
1391 #define SB_ACTIVE (1<<30)
1392 #define SB_NOUSER (1<<31)
1398 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1399 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1400 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1401 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1402 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1405 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1406 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1407 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1408 #define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */
1410 /* sb->s_iflags to limit user namespace mounts */
1411 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1412 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1413 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1415 /* Possible states of 'frozen' field */
1417 SB_UNFROZEN
= 0, /* FS is unfrozen */
1418 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1419 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1420 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1421 * internal threads if needed) */
1422 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1425 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1428 int frozen
; /* Is sb frozen? */
1429 wait_queue_head_t wait_unfrozen
; /* for get_super_thawed() */
1430 struct percpu_rw_semaphore rw_sem
[SB_FREEZE_LEVELS
];
1433 struct super_block
{
1434 struct list_head s_list
; /* Keep this first */
1435 dev_t s_dev
; /* search index; _not_ kdev_t */
1436 unsigned char s_blocksize_bits
;
1437 unsigned long s_blocksize
;
1438 loff_t s_maxbytes
; /* Max file size */
1439 struct file_system_type
*s_type
;
1440 const struct super_operations
*s_op
;
1441 const struct dquot_operations
*dq_op
;
1442 const struct quotactl_ops
*s_qcop
;
1443 const struct export_operations
*s_export_op
;
1444 unsigned long s_flags
;
1445 unsigned long s_iflags
; /* internal SB_I_* flags */
1446 unsigned long s_magic
;
1447 struct dentry
*s_root
;
1448 struct rw_semaphore s_umount
;
1451 #ifdef CONFIG_SECURITY
1454 const struct xattr_handler
**s_xattr
;
1455 #ifdef CONFIG_FS_ENCRYPTION
1456 const struct fscrypt_operations
*s_cop
;
1457 struct key
*s_master_keys
; /* master crypto keys in use */
1459 #ifdef CONFIG_FS_VERITY
1460 const struct fsverity_operations
*s_vop
;
1462 struct hlist_bl_head s_roots
; /* alternate root dentries for NFS */
1463 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1464 struct block_device
*s_bdev
;
1465 struct backing_dev_info
*s_bdi
;
1466 struct mtd_info
*s_mtd
;
1467 struct hlist_node s_instances
;
1468 unsigned int s_quota_types
; /* Bitmask of supported quota types */
1469 struct quota_info s_dquot
; /* Diskquota specific options */
1471 struct sb_writers s_writers
;
1474 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1475 * s_fsnotify_marks together for cache efficiency. They are frequently
1476 * accessed and rarely modified.
1478 void *s_fs_info
; /* Filesystem private info */
1480 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1482 /* Time limits for c/m/atime in seconds */
1483 time64_t s_time_min
;
1484 time64_t s_time_max
;
1485 #ifdef CONFIG_FSNOTIFY
1486 __u32 s_fsnotify_mask
;
1487 struct fsnotify_mark_connector __rcu
*s_fsnotify_marks
;
1490 char s_id
[32]; /* Informational name */
1491 uuid_t s_uuid
; /* UUID */
1493 unsigned int s_max_links
;
1497 * The next field is for VFS *only*. No filesystems have any business
1498 * even looking at it. You had been warned.
1500 struct mutex s_vfs_rename_mutex
; /* Kludge */
1503 * Filesystem subtype. If non-empty the filesystem type field
1504 * in /proc/mounts will be "type.subtype"
1506 const char *s_subtype
;
1508 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1511 * Saved pool identifier for cleancache (-1 means none)
1513 int cleancache_poolid
;
1515 struct shrinker s_shrink
; /* per-sb shrinker handle */
1517 /* Number of inodes with nlink == 0 but still referenced */
1518 atomic_long_t s_remove_count
;
1520 /* Pending fsnotify inode refs */
1521 atomic_long_t s_fsnotify_inode_refs
;
1523 /* Being remounted read-only */
1524 int s_readonly_remount
;
1526 /* per-sb errseq_t for reporting writeback errors via syncfs */
1529 /* AIO completions deferred from interrupt context */
1530 struct workqueue_struct
*s_dio_done_wq
;
1531 struct hlist_head s_pins
;
1534 * Owning user namespace and default context in which to
1535 * interpret filesystem uids, gids, quotas, device nodes,
1536 * xattrs and security labels.
1538 struct user_namespace
*s_user_ns
;
1541 * The list_lru structure is essentially just a pointer to a table
1542 * of per-node lru lists, each of which has its own spinlock.
1543 * There is no need to put them into separate cachelines.
1545 struct list_lru s_dentry_lru
;
1546 struct list_lru s_inode_lru
;
1547 struct rcu_head rcu
;
1548 struct work_struct destroy_work
;
1550 struct mutex s_sync_lock
; /* sync serialisation lock */
1553 * Indicates how deep in a filesystem stack this SB is
1557 /* s_inode_list_lock protects s_inodes */
1558 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp
;
1559 struct list_head s_inodes
; /* all inodes */
1561 spinlock_t s_inode_wblist_lock
;
1562 struct list_head s_inodes_wb
; /* writeback inodes */
1563 } __randomize_layout
;
1565 /* Helper functions so that in most cases filesystems will
1566 * not need to deal directly with kuid_t and kgid_t and can
1567 * instead deal with the raw numeric values that are stored
1568 * in the filesystem.
1570 static inline uid_t
i_uid_read(const struct inode
*inode
)
1572 return from_kuid(inode
->i_sb
->s_user_ns
, inode
->i_uid
);
1575 static inline gid_t
i_gid_read(const struct inode
*inode
)
1577 return from_kgid(inode
->i_sb
->s_user_ns
, inode
->i_gid
);
1580 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
1582 inode
->i_uid
= make_kuid(inode
->i_sb
->s_user_ns
, uid
);
1585 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
1587 inode
->i_gid
= make_kgid(inode
->i_sb
->s_user_ns
, gid
);
1590 extern struct timespec64
current_time(struct inode
*inode
);
1593 * Snapshotting support.
1596 void __sb_end_write(struct super_block
*sb
, int level
);
1597 int __sb_start_write(struct super_block
*sb
, int level
, bool wait
);
1599 #define __sb_writers_acquired(sb, lev) \
1600 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1601 #define __sb_writers_release(sb, lev) \
1602 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1605 * sb_end_write - drop write access to a superblock
1606 * @sb: the super we wrote to
1608 * Decrement number of writers to the filesystem. Wake up possible waiters
1609 * wanting to freeze the filesystem.
1611 static inline void sb_end_write(struct super_block
*sb
)
1613 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1617 * sb_end_pagefault - drop write access to a superblock from a page fault
1618 * @sb: the super we wrote to
1620 * Decrement number of processes handling write page fault to the filesystem.
1621 * Wake up possible waiters wanting to freeze the filesystem.
1623 static inline void sb_end_pagefault(struct super_block
*sb
)
1625 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1629 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1630 * @sb: the super we wrote to
1632 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1633 * waiters wanting to freeze the filesystem.
1635 static inline void sb_end_intwrite(struct super_block
*sb
)
1637 __sb_end_write(sb
, SB_FREEZE_FS
);
1641 * sb_start_write - get write access to a superblock
1642 * @sb: the super we write to
1644 * When a process wants to write data or metadata to a file system (i.e. dirty
1645 * a page or an inode), it should embed the operation in a sb_start_write() -
1646 * sb_end_write() pair to get exclusion against file system freezing. This
1647 * function increments number of writers preventing freezing. If the file
1648 * system is already frozen, the function waits until the file system is
1651 * Since freeze protection behaves as a lock, users have to preserve
1652 * ordering of freeze protection and other filesystem locks. Generally,
1653 * freeze protection should be the outermost lock. In particular, we have:
1656 * -> i_mutex (write path, truncate, directory ops, ...)
1657 * -> s_umount (freeze_super, thaw_super)
1659 static inline void sb_start_write(struct super_block
*sb
)
1661 __sb_start_write(sb
, SB_FREEZE_WRITE
, true);
1664 static inline int sb_start_write_trylock(struct super_block
*sb
)
1666 return __sb_start_write(sb
, SB_FREEZE_WRITE
, false);
1670 * sb_start_pagefault - get write access to a superblock from a page fault
1671 * @sb: the super we write to
1673 * When a process starts handling write page fault, it should embed the
1674 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1675 * exclusion against file system freezing. This is needed since the page fault
1676 * is going to dirty a page. This function increments number of running page
1677 * faults preventing freezing. If the file system is already frozen, the
1678 * function waits until the file system is thawed.
1680 * Since page fault freeze protection behaves as a lock, users have to preserve
1681 * ordering of freeze protection and other filesystem locks. It is advised to
1682 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1683 * handling code implies lock dependency:
1686 * -> sb_start_pagefault
1688 static inline void sb_start_pagefault(struct super_block
*sb
)
1690 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
, true);
1694 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1695 * @sb: the super we write to
1697 * This is the third level of protection against filesystem freezing. It is
1698 * free for use by a filesystem. The only requirement is that it must rank
1699 * below sb_start_pagefault.
1701 * For example filesystem can call sb_start_intwrite() when starting a
1702 * transaction which somewhat eases handling of freezing for internal sources
1703 * of filesystem changes (internal fs threads, discarding preallocation on file
1706 static inline void sb_start_intwrite(struct super_block
*sb
)
1708 __sb_start_write(sb
, SB_FREEZE_FS
, true);
1711 static inline int sb_start_intwrite_trylock(struct super_block
*sb
)
1713 return __sb_start_write(sb
, SB_FREEZE_FS
, false);
1717 extern bool inode_owner_or_capable(const struct inode
*inode
);
1720 * VFS helper functions..
1722 extern int vfs_create(struct inode
*, struct dentry
*, umode_t
, bool);
1723 extern int vfs_mkdir(struct inode
*, struct dentry
*, umode_t
);
1724 extern int vfs_mknod(struct inode
*, struct dentry
*, umode_t
, dev_t
);
1725 extern int vfs_symlink(struct inode
*, struct dentry
*, const char *);
1726 extern int vfs_link(struct dentry
*, struct inode
*, struct dentry
*, struct inode
**);
1727 extern int vfs_rmdir(struct inode
*, struct dentry
*);
1728 extern int vfs_unlink(struct inode
*, struct dentry
*, struct inode
**);
1729 extern int vfs_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*, struct inode
**, unsigned int);
1731 static inline int vfs_whiteout(struct inode
*dir
, struct dentry
*dentry
)
1733 return vfs_mknod(dir
, dentry
, S_IFCHR
| WHITEOUT_MODE
, WHITEOUT_DEV
);
1736 extern struct dentry
*vfs_tmpfile(struct dentry
*dentry
, umode_t mode
,
1739 int vfs_mkobj(struct dentry
*, umode_t
,
1740 int (*f
)(struct dentry
*, umode_t
, void *),
1743 extern long vfs_ioctl(struct file
*file
, unsigned int cmd
, unsigned long arg
);
1745 #ifdef CONFIG_COMPAT
1746 extern long compat_ptr_ioctl(struct file
*file
, unsigned int cmd
,
1749 #define compat_ptr_ioctl NULL
1753 * VFS file helper functions.
1755 extern void inode_init_owner(struct inode
*inode
, const struct inode
*dir
,
1757 extern bool may_open_dev(const struct path
*path
);
1759 * VFS FS_IOC_FIEMAP helper definitions.
1761 struct fiemap_extent_info
{
1762 unsigned int fi_flags
; /* Flags as passed from user */
1763 unsigned int fi_extents_mapped
; /* Number of mapped extents */
1764 unsigned int fi_extents_max
; /* Size of fiemap_extent array */
1765 struct fiemap_extent __user
*fi_extents_start
; /* Start of
1766 fiemap_extent array */
1768 int fiemap_fill_next_extent(struct fiemap_extent_info
*info
, u64 logical
,
1769 u64 phys
, u64 len
, u32 flags
);
1770 int fiemap_check_flags(struct fiemap_extent_info
*fieinfo
, u32 fs_flags
);
1773 * This is the "filldir" function type, used by readdir() to let
1774 * the kernel specify what kind of dirent layout it wants to have.
1775 * This allows the kernel to read directories into kernel space or
1776 * to have different dirent layouts depending on the binary type.
1779 typedef int (*filldir_t
)(struct dir_context
*, const char *, int, loff_t
, u64
,
1782 struct dir_context
{
1787 struct block_device_operations
;
1789 /* These macros are for out of kernel modules to test that
1790 * the kernel supports the unlocked_ioctl and compat_ioctl
1791 * fields in struct file_operations. */
1792 #define HAVE_COMPAT_IOCTL 1
1793 #define HAVE_UNLOCKED_IOCTL 1
1796 * These flags let !MMU mmap() govern direct device mapping vs immediate
1797 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1799 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1800 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1801 * NOMMU_MAP_READ: Can be mapped for reading
1802 * NOMMU_MAP_WRITE: Can be mapped for writing
1803 * NOMMU_MAP_EXEC: Can be mapped for execution
1805 #define NOMMU_MAP_COPY 0x00000001
1806 #define NOMMU_MAP_DIRECT 0x00000008
1807 #define NOMMU_MAP_READ VM_MAYREAD
1808 #define NOMMU_MAP_WRITE VM_MAYWRITE
1809 #define NOMMU_MAP_EXEC VM_MAYEXEC
1811 #define NOMMU_VMFLAGS \
1812 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1815 * These flags control the behavior of the remap_file_range function pointer.
1816 * If it is called with len == 0 that means "remap to end of source file".
1817 * See Documentation/filesystems/vfs.rst for more details about this call.
1819 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
1820 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
1822 #define REMAP_FILE_DEDUP (1 << 0)
1823 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
1826 * These flags signal that the caller is ok with altering various aspects of
1827 * the behavior of the remap operation. The changes must be made by the
1828 * implementation; the vfs remap helper functions can take advantage of them.
1829 * Flags in this category exist to preserve the quirky behavior of the hoisted
1830 * btrfs clone/dedupe ioctls.
1832 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
1836 struct file_operations
{
1837 struct module
*owner
;
1838 loff_t (*llseek
) (struct file
*, loff_t
, int);
1839 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
1840 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
1841 ssize_t (*read_iter
) (struct kiocb
*, struct iov_iter
*);
1842 ssize_t (*write_iter
) (struct kiocb
*, struct iov_iter
*);
1843 int (*iopoll
)(struct kiocb
*kiocb
, bool spin
);
1844 int (*iterate
) (struct file
*, struct dir_context
*);
1845 int (*iterate_shared
) (struct file
*, struct dir_context
*);
1846 __poll_t (*poll
) (struct file
*, struct poll_table_struct
*);
1847 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
1848 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
1849 int (*mmap
) (struct file
*, struct vm_area_struct
*);
1850 unsigned long mmap_supported_flags
;
1851 int (*open
) (struct inode
*, struct file
*);
1852 int (*flush
) (struct file
*, fl_owner_t id
);
1853 int (*release
) (struct inode
*, struct file
*);
1854 int (*fsync
) (struct file
*, loff_t
, loff_t
, int datasync
);
1855 int (*fasync
) (int, struct file
*, int);
1856 int (*lock
) (struct file
*, int, struct file_lock
*);
1857 ssize_t (*sendpage
) (struct file
*, struct page
*, int, size_t, loff_t
*, int);
1858 unsigned long (*get_unmapped_area
)(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1859 int (*check_flags
)(int);
1860 int (*flock
) (struct file
*, int, struct file_lock
*);
1861 ssize_t (*splice_write
)(struct pipe_inode_info
*, struct file
*, loff_t
*, size_t, unsigned int);
1862 ssize_t (*splice_read
)(struct file
*, loff_t
*, struct pipe_inode_info
*, size_t, unsigned int);
1863 int (*setlease
)(struct file
*, long, struct file_lock
**, void **);
1864 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
1866 void (*show_fdinfo
)(struct seq_file
*m
, struct file
*f
);
1868 unsigned (*mmap_capabilities
)(struct file
*);
1870 ssize_t (*copy_file_range
)(struct file
*, loff_t
, struct file
*,
1871 loff_t
, size_t, unsigned int);
1872 loff_t (*remap_file_range
)(struct file
*file_in
, loff_t pos_in
,
1873 struct file
*file_out
, loff_t pos_out
,
1874 loff_t len
, unsigned int remap_flags
);
1875 int (*fadvise
)(struct file
*, loff_t
, loff_t
, int);
1876 } __randomize_layout
;
1878 struct inode_operations
{
1879 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*, unsigned int);
1880 const char * (*get_link
) (struct dentry
*, struct inode
*, struct delayed_call
*);
1881 int (*permission
) (struct inode
*, int);
1882 struct posix_acl
* (*get_acl
)(struct inode
*, int);
1884 int (*readlink
) (struct dentry
*, char __user
*,int);
1886 int (*create
) (struct inode
*,struct dentry
*, umode_t
, bool);
1887 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
1888 int (*unlink
) (struct inode
*,struct dentry
*);
1889 int (*symlink
) (struct inode
*,struct dentry
*,const char *);
1890 int (*mkdir
) (struct inode
*,struct dentry
*,umode_t
);
1891 int (*rmdir
) (struct inode
*,struct dentry
*);
1892 int (*mknod
) (struct inode
*,struct dentry
*,umode_t
,dev_t
);
1893 int (*rename
) (struct inode
*, struct dentry
*,
1894 struct inode
*, struct dentry
*, unsigned int);
1895 int (*setattr
) (struct dentry
*, struct iattr
*);
1896 int (*getattr
) (const struct path
*, struct kstat
*, u32
, unsigned int);
1897 ssize_t (*listxattr
) (struct dentry
*, char *, size_t);
1898 int (*fiemap
)(struct inode
*, struct fiemap_extent_info
*, u64 start
,
1900 int (*update_time
)(struct inode
*, struct timespec64
*, int);
1901 int (*atomic_open
)(struct inode
*, struct dentry
*,
1902 struct file
*, unsigned open_flag
,
1903 umode_t create_mode
);
1904 int (*tmpfile
) (struct inode
*, struct dentry
*, umode_t
);
1905 int (*set_acl
)(struct inode
*, struct posix_acl
*, int);
1906 } ____cacheline_aligned
;
1908 static inline ssize_t
call_read_iter(struct file
*file
, struct kiocb
*kio
,
1909 struct iov_iter
*iter
)
1911 return file
->f_op
->read_iter(kio
, iter
);
1914 static inline ssize_t
call_write_iter(struct file
*file
, struct kiocb
*kio
,
1915 struct iov_iter
*iter
)
1917 return file
->f_op
->write_iter(kio
, iter
);
1920 static inline int call_mmap(struct file
*file
, struct vm_area_struct
*vma
)
1922 return file
->f_op
->mmap(file
, vma
);
1925 ssize_t
rw_copy_check_uvector(int type
, const struct iovec __user
* uvector
,
1926 unsigned long nr_segs
, unsigned long fast_segs
,
1927 struct iovec
*fast_pointer
,
1928 struct iovec
**ret_pointer
);
1930 extern ssize_t
__vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1931 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1932 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1933 extern ssize_t
vfs_readv(struct file
*, const struct iovec __user
*,
1934 unsigned long, loff_t
*, rwf_t
);
1935 extern ssize_t
vfs_copy_file_range(struct file
*, loff_t
, struct file
*,
1936 loff_t
, size_t, unsigned int);
1937 extern ssize_t
generic_copy_file_range(struct file
*file_in
, loff_t pos_in
,
1938 struct file
*file_out
, loff_t pos_out
,
1939 size_t len
, unsigned int flags
);
1940 extern int generic_remap_file_range_prep(struct file
*file_in
, loff_t pos_in
,
1941 struct file
*file_out
, loff_t pos_out
,
1943 unsigned int remap_flags
);
1944 extern loff_t
do_clone_file_range(struct file
*file_in
, loff_t pos_in
,
1945 struct file
*file_out
, loff_t pos_out
,
1946 loff_t len
, unsigned int remap_flags
);
1947 extern loff_t
vfs_clone_file_range(struct file
*file_in
, loff_t pos_in
,
1948 struct file
*file_out
, loff_t pos_out
,
1949 loff_t len
, unsigned int remap_flags
);
1950 extern int vfs_dedupe_file_range(struct file
*file
,
1951 struct file_dedupe_range
*same
);
1952 extern loff_t
vfs_dedupe_file_range_one(struct file
*src_file
, loff_t src_pos
,
1953 struct file
*dst_file
, loff_t dst_pos
,
1954 loff_t len
, unsigned int remap_flags
);
1957 struct super_operations
{
1958 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
1959 void (*destroy_inode
)(struct inode
*);
1960 void (*free_inode
)(struct inode
*);
1962 void (*dirty_inode
) (struct inode
*, int flags
);
1963 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
1964 int (*drop_inode
) (struct inode
*);
1965 void (*evict_inode
) (struct inode
*);
1966 void (*put_super
) (struct super_block
*);
1967 int (*sync_fs
)(struct super_block
*sb
, int wait
);
1968 int (*freeze_super
) (struct super_block
*);
1969 int (*freeze_fs
) (struct super_block
*);
1970 int (*thaw_super
) (struct super_block
*);
1971 int (*unfreeze_fs
) (struct super_block
*);
1972 int (*statfs
) (struct dentry
*, struct kstatfs
*);
1973 int (*remount_fs
) (struct super_block
*, int *, char *);
1974 void (*umount_begin
) (struct super_block
*);
1976 int (*show_options
)(struct seq_file
*, struct dentry
*);
1977 int (*show_devname
)(struct seq_file
*, struct dentry
*);
1978 int (*show_path
)(struct seq_file
*, struct dentry
*);
1979 int (*show_stats
)(struct seq_file
*, struct dentry
*);
1981 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
1982 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
1983 struct dquot
**(*get_dquots
)(struct inode
*);
1985 int (*bdev_try_to_free_page
)(struct super_block
*, struct page
*, gfp_t
);
1986 long (*nr_cached_objects
)(struct super_block
*,
1987 struct shrink_control
*);
1988 long (*free_cached_objects
)(struct super_block
*,
1989 struct shrink_control
*);
1993 * Inode flags - they have no relation to superblock flags now
1995 #define S_SYNC 1 /* Writes are synced at once */
1996 #define S_NOATIME 2 /* Do not update access times */
1997 #define S_APPEND 4 /* Append-only file */
1998 #define S_IMMUTABLE 8 /* Immutable file */
1999 #define S_DEAD 16 /* removed, but still open directory */
2000 #define S_NOQUOTA 32 /* Inode is not counted to quota */
2001 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
2002 #define S_NOCMTIME 128 /* Do not update file c/mtime */
2003 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
2004 #define S_PRIVATE 512 /* Inode is fs-internal */
2005 #define S_IMA 1024 /* Inode has an associated IMA struct */
2006 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
2007 #define S_NOSEC 4096 /* no suid or xattr security attributes */
2008 #ifdef CONFIG_FS_DAX
2009 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
2011 #define S_DAX 0 /* Make all the DAX code disappear */
2013 #define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */
2014 #define S_CASEFOLD 32768 /* Casefolded file */
2015 #define S_VERITY 65536 /* Verity file (using fs/verity/) */
2018 * Note that nosuid etc flags are inode-specific: setting some file-system
2019 * flags just means all the inodes inherit those flags by default. It might be
2020 * possible to override it selectively if you really wanted to with some
2021 * ioctl() that is not currently implemented.
2023 * Exception: SB_RDONLY is always applied to the entire file system.
2025 * Unfortunately, it is possible to change a filesystems flags with it mounted
2026 * with files in use. This means that all of the inodes will not have their
2027 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
2028 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
2030 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
2032 static inline bool sb_rdonly(const struct super_block
*sb
) { return sb
->s_flags
& SB_RDONLY
; }
2033 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
2034 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
2035 ((inode)->i_flags & S_SYNC))
2036 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
2037 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
2038 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
2039 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
2040 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
2042 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
2043 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
2044 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
2045 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
2047 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
2048 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
2049 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
2050 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
2051 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
2052 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2053 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2054 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2055 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2056 #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD)
2057 #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY)
2059 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2060 (inode)->i_rdev == WHITEOUT_DEV)
2062 static inline bool HAS_UNMAPPED_ID(struct inode
*inode
)
2064 return !uid_valid(inode
->i_uid
) || !gid_valid(inode
->i_gid
);
2067 static inline enum rw_hint
file_write_hint(struct file
*file
)
2069 if (file
->f_write_hint
!= WRITE_LIFE_NOT_SET
)
2070 return file
->f_write_hint
;
2072 return file_inode(file
)->i_write_hint
;
2075 static inline int iocb_flags(struct file
*file
);
2077 static inline u16
ki_hint_validate(enum rw_hint hint
)
2079 typeof(((struct kiocb
*)0)->ki_hint
) max_hint
= -1;
2081 if (hint
<= max_hint
)
2086 static inline void init_sync_kiocb(struct kiocb
*kiocb
, struct file
*filp
)
2088 *kiocb
= (struct kiocb
) {
2090 .ki_flags
= iocb_flags(filp
),
2091 .ki_hint
= ki_hint_validate(file_write_hint(filp
)),
2092 .ki_ioprio
= get_current_ioprio(),
2096 static inline void kiocb_clone(struct kiocb
*kiocb
, struct kiocb
*kiocb_src
,
2099 *kiocb
= (struct kiocb
) {
2101 .ki_flags
= kiocb_src
->ki_flags
,
2102 .ki_hint
= kiocb_src
->ki_hint
,
2103 .ki_ioprio
= kiocb_src
->ki_ioprio
,
2104 .ki_pos
= kiocb_src
->ki_pos
,
2109 * Inode state bits. Protected by inode->i_lock
2111 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
2112 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
2114 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2115 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2116 * various stages of removing an inode.
2118 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2120 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2121 * fdatasync(). i_atime is the usual cause.
2122 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2123 * these changes separately from I_DIRTY_SYNC so that we
2124 * don't have to write inode on fdatasync() when only
2125 * mtime has changed in it.
2126 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2127 * I_NEW Serves as both a mutex and completion notification.
2128 * New inodes set I_NEW. If two processes both create
2129 * the same inode, one of them will release its inode and
2130 * wait for I_NEW to be released before returning.
2131 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2132 * also cause waiting on I_NEW, without I_NEW actually
2133 * being set. find_inode() uses this to prevent returning
2134 * nearly-dead inodes.
2135 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2136 * is zero. I_FREEING must be set when I_WILL_FREE is
2138 * I_FREEING Set when inode is about to be freed but still has dirty
2139 * pages or buffers attached or the inode itself is still
2141 * I_CLEAR Added by clear_inode(). In this state the inode is
2142 * clean and can be destroyed. Inode keeps I_FREEING.
2144 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2145 * prohibited for many purposes. iget() must wait for
2146 * the inode to be completely released, then create it
2147 * anew. Other functions will just ignore such inodes,
2148 * if appropriate. I_NEW is used for waiting.
2150 * I_SYNC Writeback of inode is running. The bit is set during
2151 * data writeback, and cleared with a wakeup on the bit
2152 * address once it is done. The bit is also used to pin
2153 * the inode in memory for flusher thread.
2155 * I_REFERENCED Marks the inode as recently references on the LRU list.
2157 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2159 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2160 * synchronize competing switching instances and to tell
2161 * wb stat updates to grab the i_pages lock. See
2162 * inode_switch_wbs_work_fn() for details.
2164 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2165 * and work dirs among overlayfs mounts.
2167 * I_CREATING New object's inode in the middle of setting up.
2169 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2171 #define I_DIRTY_SYNC (1 << 0)
2172 #define I_DIRTY_DATASYNC (1 << 1)
2173 #define I_DIRTY_PAGES (1 << 2)
2175 #define I_NEW (1 << __I_NEW)
2176 #define I_WILL_FREE (1 << 4)
2177 #define I_FREEING (1 << 5)
2178 #define I_CLEAR (1 << 6)
2180 #define I_SYNC (1 << __I_SYNC)
2181 #define I_REFERENCED (1 << 8)
2182 #define __I_DIO_WAKEUP 9
2183 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2184 #define I_LINKABLE (1 << 10)
2185 #define I_DIRTY_TIME (1 << 11)
2186 #define __I_DIRTY_TIME_EXPIRED 12
2187 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
2188 #define I_WB_SWITCH (1 << 13)
2189 #define I_OVL_INUSE (1 << 14)
2190 #define I_CREATING (1 << 15)
2192 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2193 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2194 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2196 extern void __mark_inode_dirty(struct inode
*, int);
2197 static inline void mark_inode_dirty(struct inode
*inode
)
2199 __mark_inode_dirty(inode
, I_DIRTY
);
2202 static inline void mark_inode_dirty_sync(struct inode
*inode
)
2204 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
2207 extern void inc_nlink(struct inode
*inode
);
2208 extern void drop_nlink(struct inode
*inode
);
2209 extern void clear_nlink(struct inode
*inode
);
2210 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
2212 static inline void inode_inc_link_count(struct inode
*inode
)
2215 mark_inode_dirty(inode
);
2218 static inline void inode_dec_link_count(struct inode
*inode
)
2221 mark_inode_dirty(inode
);
2224 enum file_time_flags
{
2231 extern bool atime_needs_update(const struct path
*, struct inode
*);
2232 extern void touch_atime(const struct path
*);
2233 static inline void file_accessed(struct file
*file
)
2235 if (!(file
->f_flags
& O_NOATIME
))
2236 touch_atime(&file
->f_path
);
2239 extern int file_modified(struct file
*file
);
2241 int sync_inode(struct inode
*inode
, struct writeback_control
*wbc
);
2242 int sync_inode_metadata(struct inode
*inode
, int wait
);
2244 struct file_system_type
{
2247 #define FS_REQUIRES_DEV 1
2248 #define FS_BINARY_MOUNTDATA 2
2249 #define FS_HAS_SUBTYPE 4
2250 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2251 #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */
2252 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2253 int (*init_fs_context
)(struct fs_context
*);
2254 const struct fs_parameter_spec
*parameters
;
2255 struct dentry
*(*mount
) (struct file_system_type
*, int,
2256 const char *, void *);
2257 void (*kill_sb
) (struct super_block
*);
2258 struct module
*owner
;
2259 struct file_system_type
* next
;
2260 struct hlist_head fs_supers
;
2262 struct lock_class_key s_lock_key
;
2263 struct lock_class_key s_umount_key
;
2264 struct lock_class_key s_vfs_rename_key
;
2265 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
2267 struct lock_class_key i_lock_key
;
2268 struct lock_class_key i_mutex_key
;
2269 struct lock_class_key i_mutex_dir_key
;
2272 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2275 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
2276 int flags
, const char *dev_name
, void *data
,
2277 int (*fill_super
)(struct super_block
*, void *, int));
2279 static inline struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
2280 int flags
, const char *dev_name
, void *data
,
2281 int (*fill_super
)(struct super_block
*, void *, int))
2283 return ERR_PTR(-ENODEV
);
2286 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
2287 int flags
, void *data
,
2288 int (*fill_super
)(struct super_block
*, void *, int));
2289 extern struct dentry
*mount_nodev(struct file_system_type
*fs_type
,
2290 int flags
, void *data
,
2291 int (*fill_super
)(struct super_block
*, void *, int));
2292 extern struct dentry
*mount_subtree(struct vfsmount
*mnt
, const char *path
);
2293 void generic_shutdown_super(struct super_block
*sb
);
2295 void kill_block_super(struct super_block
*sb
);
2297 static inline void kill_block_super(struct super_block
*sb
)
2302 void kill_anon_super(struct super_block
*sb
);
2303 void kill_litter_super(struct super_block
*sb
);
2304 void deactivate_super(struct super_block
*sb
);
2305 void deactivate_locked_super(struct super_block
*sb
);
2306 int set_anon_super(struct super_block
*s
, void *data
);
2307 int set_anon_super_fc(struct super_block
*s
, struct fs_context
*fc
);
2308 int get_anon_bdev(dev_t
*);
2309 void free_anon_bdev(dev_t
);
2310 struct super_block
*sget_fc(struct fs_context
*fc
,
2311 int (*test
)(struct super_block
*, struct fs_context
*),
2312 int (*set
)(struct super_block
*, struct fs_context
*));
2313 struct super_block
*sget(struct file_system_type
*type
,
2314 int (*test
)(struct super_block
*,void *),
2315 int (*set
)(struct super_block
*,void *),
2316 int flags
, void *data
);
2318 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2319 #define fops_get(fops) \
2320 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2321 #define fops_put(fops) \
2322 do { if (fops) module_put((fops)->owner); } while(0)
2324 * This one is to be used *ONLY* from ->open() instances.
2325 * fops must be non-NULL, pinned down *and* module dependencies
2326 * should be sufficient to pin the caller down as well.
2328 #define replace_fops(f, fops) \
2330 struct file *__file = (f); \
2331 fops_put(__file->f_op); \
2332 BUG_ON(!(__file->f_op = (fops))); \
2335 extern int register_filesystem(struct file_system_type
*);
2336 extern int unregister_filesystem(struct file_system_type
*);
2337 extern struct vfsmount
*kern_mount(struct file_system_type
*);
2338 extern void kern_unmount(struct vfsmount
*mnt
);
2339 extern int may_umount_tree(struct vfsmount
*);
2340 extern int may_umount(struct vfsmount
*);
2341 extern long do_mount(const char *, const char __user
*,
2342 const char *, unsigned long, void *);
2343 extern struct vfsmount
*collect_mounts(const struct path
*);
2344 extern void drop_collected_mounts(struct vfsmount
*);
2345 extern int iterate_mounts(int (*)(struct vfsmount
*, void *), void *,
2347 extern int vfs_statfs(const struct path
*, struct kstatfs
*);
2348 extern int user_statfs(const char __user
*, struct kstatfs
*);
2349 extern int fd_statfs(int, struct kstatfs
*);
2350 extern int freeze_super(struct super_block
*super
);
2351 extern int thaw_super(struct super_block
*super
);
2352 extern bool our_mnt(struct vfsmount
*mnt
);
2353 extern __printf(2, 3)
2354 int super_setup_bdi_name(struct super_block
*sb
, char *fmt
, ...);
2355 extern int super_setup_bdi(struct super_block
*sb
);
2357 extern int current_umask(void);
2359 extern void ihold(struct inode
* inode
);
2360 extern void iput(struct inode
*);
2361 extern int generic_update_time(struct inode
*, struct timespec64
*, int);
2364 extern struct kobject
*fs_kobj
;
2366 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2368 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2369 extern int locks_mandatory_locked(struct file
*);
2370 extern int locks_mandatory_area(struct inode
*, struct file
*, loff_t
, loff_t
, unsigned char);
2373 * Candidates for mandatory locking have the setgid bit set
2374 * but no group execute bit - an otherwise meaningless combination.
2377 static inline int __mandatory_lock(struct inode
*ino
)
2379 return (ino
->i_mode
& (S_ISGID
| S_IXGRP
)) == S_ISGID
;
2383 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2384 * otherwise these will be advisory locks
2387 static inline int mandatory_lock(struct inode
*ino
)
2389 return IS_MANDLOCK(ino
) && __mandatory_lock(ino
);
2392 static inline int locks_verify_locked(struct file
*file
)
2394 if (mandatory_lock(locks_inode(file
)))
2395 return locks_mandatory_locked(file
);
2399 static inline int locks_verify_truncate(struct inode
*inode
,
2403 if (!inode
->i_flctx
|| !mandatory_lock(inode
))
2406 if (size
< inode
->i_size
) {
2407 return locks_mandatory_area(inode
, f
, size
, inode
->i_size
- 1,
2410 return locks_mandatory_area(inode
, f
, inode
->i_size
, size
- 1,
2415 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2417 static inline int locks_mandatory_locked(struct file
*file
)
2422 static inline int locks_mandatory_area(struct inode
*inode
, struct file
*filp
,
2423 loff_t start
, loff_t end
, unsigned char type
)
2428 static inline int __mandatory_lock(struct inode
*inode
)
2433 static inline int mandatory_lock(struct inode
*inode
)
2438 static inline int locks_verify_locked(struct file
*file
)
2443 static inline int locks_verify_truncate(struct inode
*inode
, struct file
*filp
,
2449 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2452 #ifdef CONFIG_FILE_LOCKING
2453 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2456 * Since this check is lockless, we must ensure that any refcounts
2457 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2458 * could end up racing with tasks trying to set a new lease on this
2462 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2463 return __break_lease(inode
, mode
, FL_LEASE
);
2467 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2470 * Since this check is lockless, we must ensure that any refcounts
2471 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2472 * could end up racing with tasks trying to set a new lease on this
2476 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2477 return __break_lease(inode
, mode
, FL_DELEG
);
2481 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2485 ret
= break_deleg(inode
, O_WRONLY
|O_NONBLOCK
);
2486 if (ret
== -EWOULDBLOCK
&& delegated_inode
) {
2487 *delegated_inode
= inode
;
2493 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2497 ret
= break_deleg(*delegated_inode
, O_WRONLY
);
2498 iput(*delegated_inode
);
2499 *delegated_inode
= NULL
;
2503 static inline int break_layout(struct inode
*inode
, bool wait
)
2506 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2507 return __break_lease(inode
,
2508 wait
? O_WRONLY
: O_WRONLY
| O_NONBLOCK
,
2513 #else /* !CONFIG_FILE_LOCKING */
2514 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2519 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2524 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2529 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2535 static inline int break_layout(struct inode
*inode
, bool wait
)
2540 #endif /* CONFIG_FILE_LOCKING */
2545 const char *name
; /* pointer to actual string */
2546 const __user
char *uptr
; /* original userland pointer */
2548 struct audit_names
*aname
;
2551 static_assert(offsetof(struct filename
, iname
) % sizeof(long) == 0);
2553 extern long vfs_truncate(const struct path
*, loff_t
);
2554 extern int do_truncate(struct dentry
*, loff_t start
, unsigned int time_attrs
,
2556 extern int vfs_fallocate(struct file
*file
, int mode
, loff_t offset
,
2558 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
2560 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
2561 extern struct file
*filp_open(const char *, int, umode_t
);
2562 extern struct file
*file_open_root(struct dentry
*, struct vfsmount
*,
2563 const char *, int, umode_t
);
2564 extern struct file
* dentry_open(const struct path
*, int, const struct cred
*);
2565 extern struct file
* open_with_fake_path(const struct path
*, int,
2566 struct inode
*, const struct cred
*);
2567 static inline struct file
*file_clone_open(struct file
*file
)
2569 return dentry_open(&file
->f_path
, file
->f_flags
, file
->f_cred
);
2571 extern int filp_close(struct file
*, fl_owner_t id
);
2573 extern struct filename
*getname_flags(const char __user
*, int, int *);
2574 extern struct filename
*getname(const char __user
*);
2575 extern struct filename
*getname_kernel(const char *);
2576 extern void putname(struct filename
*name
);
2578 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
2579 int (*open
)(struct inode
*, struct file
*));
2580 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
2583 extern void __init
vfs_caches_init_early(void);
2584 extern void __init
vfs_caches_init(void);
2586 extern struct kmem_cache
*names_cachep
;
2588 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2589 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2592 extern int register_blkdev(unsigned int, const char *);
2593 extern void unregister_blkdev(unsigned int, const char *);
2594 extern struct block_device
*bdget(dev_t
);
2595 extern struct block_device
*bdgrab(struct block_device
*bdev
);
2596 extern void bd_set_size(struct block_device
*, loff_t size
);
2597 extern void bd_forget(struct inode
*inode
);
2598 extern void bdput(struct block_device
*);
2599 extern void invalidate_bdev(struct block_device
*);
2600 extern void iterate_bdevs(void (*)(struct block_device
*, void *), void *);
2601 extern int sync_blockdev(struct block_device
*bdev
);
2602 extern void kill_bdev(struct block_device
*);
2603 extern struct super_block
*freeze_bdev(struct block_device
*);
2604 extern void emergency_thaw_all(void);
2605 extern void emergency_thaw_bdev(struct super_block
*sb
);
2606 extern int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
);
2607 extern int fsync_bdev(struct block_device
*);
2609 extern struct super_block
*blockdev_superblock
;
2611 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2613 return sb
== blockdev_superblock
;
2616 static inline void bd_forget(struct inode
*inode
) {}
2617 static inline int sync_blockdev(struct block_device
*bdev
) { return 0; }
2618 static inline void kill_bdev(struct block_device
*bdev
) {}
2619 static inline void invalidate_bdev(struct block_device
*bdev
) {}
2621 static inline struct super_block
*freeze_bdev(struct block_device
*sb
)
2626 static inline int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
)
2631 static inline int emergency_thaw_bdev(struct super_block
*sb
)
2636 static inline void iterate_bdevs(void (*f
)(struct block_device
*, void *), void *arg
)
2640 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2645 extern int sync_filesystem(struct super_block
*);
2646 extern const struct file_operations def_blk_fops
;
2647 extern const struct file_operations def_chr_fops
;
2649 extern int blkdev_ioctl(struct block_device
*, fmode_t
, unsigned, unsigned long);
2650 extern long compat_blkdev_ioctl(struct file
*, unsigned, unsigned long);
2651 extern int blkdev_get(struct block_device
*bdev
, fmode_t mode
, void *holder
);
2652 extern struct block_device
*blkdev_get_by_path(const char *path
, fmode_t mode
,
2654 extern struct block_device
*blkdev_get_by_dev(dev_t dev
, fmode_t mode
,
2656 extern struct block_device
*bd_start_claiming(struct block_device
*bdev
,
2658 extern void bd_finish_claiming(struct block_device
*bdev
,
2659 struct block_device
*whole
, void *holder
);
2660 extern void bd_abort_claiming(struct block_device
*bdev
,
2661 struct block_device
*whole
, void *holder
);
2662 extern void blkdev_put(struct block_device
*bdev
, fmode_t mode
);
2665 extern int bd_link_disk_holder(struct block_device
*bdev
, struct gendisk
*disk
);
2666 extern void bd_unlink_disk_holder(struct block_device
*bdev
,
2667 struct gendisk
*disk
);
2669 static inline int bd_link_disk_holder(struct block_device
*bdev
,
2670 struct gendisk
*disk
)
2674 static inline void bd_unlink_disk_holder(struct block_device
*bdev
,
2675 struct gendisk
*disk
)
2682 #define CHRDEV_MAJOR_MAX 512
2683 /* Marks the bottom of the first segment of free char majors */
2684 #define CHRDEV_MAJOR_DYN_END 234
2685 /* Marks the top and bottom of the second segment of free char majors */
2686 #define CHRDEV_MAJOR_DYN_EXT_START 511
2687 #define CHRDEV_MAJOR_DYN_EXT_END 384
2689 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
2690 extern int register_chrdev_region(dev_t
, unsigned, const char *);
2691 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
2692 unsigned int count
, const char *name
,
2693 const struct file_operations
*fops
);
2694 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
2695 unsigned int count
, const char *name
);
2696 extern void unregister_chrdev_region(dev_t
, unsigned);
2697 extern void chrdev_show(struct seq_file
*,off_t
);
2699 static inline int register_chrdev(unsigned int major
, const char *name
,
2700 const struct file_operations
*fops
)
2702 return __register_chrdev(major
, 0, 256, name
, fops
);
2705 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2707 __unregister_chrdev(major
, 0, 256, name
);
2710 /* fs/block_dev.c */
2711 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2712 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2715 #define BLKDEV_MAJOR_MAX 512
2716 extern const char *bdevname(struct block_device
*bdev
, char *buffer
);
2717 extern struct block_device
*lookup_bdev(const char *);
2718 extern void blkdev_show(struct seq_file
*,off_t
);
2721 #define BLKDEV_MAJOR_MAX 0
2724 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2726 /* Invalid inode operations -- fs/bad_inode.c */
2727 extern void make_bad_inode(struct inode
*);
2728 extern bool is_bad_inode(struct inode
*);
2731 extern int revalidate_disk(struct gendisk
*);
2732 extern int check_disk_change(struct block_device
*);
2733 extern int __invalidate_device(struct block_device
*, bool);
2735 unsigned long invalidate_mapping_pages(struct address_space
*mapping
,
2736 pgoff_t start
, pgoff_t end
);
2738 static inline void invalidate_remote_inode(struct inode
*inode
)
2740 if (S_ISREG(inode
->i_mode
) || S_ISDIR(inode
->i_mode
) ||
2741 S_ISLNK(inode
->i_mode
))
2742 invalidate_mapping_pages(inode
->i_mapping
, 0, -1);
2744 extern int invalidate_inode_pages2(struct address_space
*mapping
);
2745 extern int invalidate_inode_pages2_range(struct address_space
*mapping
,
2746 pgoff_t start
, pgoff_t end
);
2747 extern int write_inode_now(struct inode
*, int);
2748 extern int filemap_fdatawrite(struct address_space
*);
2749 extern int filemap_flush(struct address_space
*);
2750 extern int filemap_fdatawait_keep_errors(struct address_space
*mapping
);
2751 extern int filemap_fdatawait_range(struct address_space
*, loff_t lstart
,
2753 extern int filemap_fdatawait_range_keep_errors(struct address_space
*mapping
,
2754 loff_t start_byte
, loff_t end_byte
);
2756 static inline int filemap_fdatawait(struct address_space
*mapping
)
2758 return filemap_fdatawait_range(mapping
, 0, LLONG_MAX
);
2761 extern bool filemap_range_has_page(struct address_space
*, loff_t lstart
,
2763 extern int filemap_write_and_wait_range(struct address_space
*mapping
,
2764 loff_t lstart
, loff_t lend
);
2765 extern int __filemap_fdatawrite_range(struct address_space
*mapping
,
2766 loff_t start
, loff_t end
, int sync_mode
);
2767 extern int filemap_fdatawrite_range(struct address_space
*mapping
,
2768 loff_t start
, loff_t end
);
2769 extern int filemap_check_errors(struct address_space
*mapping
);
2770 extern void __filemap_set_wb_err(struct address_space
*mapping
, int err
);
2772 static inline int filemap_write_and_wait(struct address_space
*mapping
)
2774 return filemap_write_and_wait_range(mapping
, 0, LLONG_MAX
);
2777 extern int __must_check
file_fdatawait_range(struct file
*file
, loff_t lstart
,
2779 extern int __must_check
file_check_and_advance_wb_err(struct file
*file
);
2780 extern int __must_check
file_write_and_wait_range(struct file
*file
,
2781 loff_t start
, loff_t end
);
2783 static inline int file_write_and_wait(struct file
*file
)
2785 return file_write_and_wait_range(file
, 0, LLONG_MAX
);
2789 * filemap_set_wb_err - set a writeback error on an address_space
2790 * @mapping: mapping in which to set writeback error
2791 * @err: error to be set in mapping
2793 * When writeback fails in some way, we must record that error so that
2794 * userspace can be informed when fsync and the like are called. We endeavor
2795 * to report errors on any file that was open at the time of the error. Some
2796 * internal callers also need to know when writeback errors have occurred.
2798 * When a writeback error occurs, most filesystems will want to call
2799 * filemap_set_wb_err to record the error in the mapping so that it will be
2800 * automatically reported whenever fsync is called on the file.
2802 static inline void filemap_set_wb_err(struct address_space
*mapping
, int err
)
2804 /* Fastpath for common case of no error */
2806 __filemap_set_wb_err(mapping
, err
);
2810 * filemap_check_wb_error - has an error occurred since the mark was sampled?
2811 * @mapping: mapping to check for writeback errors
2812 * @since: previously-sampled errseq_t
2814 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2815 * the given value was sampled.
2817 * If it has then report the latest error set, otherwise return 0.
2819 static inline int filemap_check_wb_err(struct address_space
*mapping
,
2822 return errseq_check(&mapping
->wb_err
, since
);
2826 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2827 * @mapping: mapping to be sampled
2829 * Writeback errors are always reported relative to a particular sample point
2830 * in the past. This function provides those sample points.
2832 static inline errseq_t
filemap_sample_wb_err(struct address_space
*mapping
)
2834 return errseq_sample(&mapping
->wb_err
);
2838 * file_sample_sb_err - sample the current errseq_t to test for later errors
2839 * @mapping: mapping to be sampled
2841 * Grab the most current superblock-level errseq_t value for the given
2844 static inline errseq_t
file_sample_sb_err(struct file
*file
)
2846 return errseq_sample(&file
->f_path
.dentry
->d_sb
->s_wb_err
);
2849 static inline int filemap_nr_thps(struct address_space
*mapping
)
2851 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
2852 return atomic_read(&mapping
->nr_thps
);
2858 static inline void filemap_nr_thps_inc(struct address_space
*mapping
)
2860 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
2861 atomic_inc(&mapping
->nr_thps
);
2867 static inline void filemap_nr_thps_dec(struct address_space
*mapping
)
2869 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
2870 atomic_dec(&mapping
->nr_thps
);
2876 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
2878 extern int vfs_fsync(struct file
*file
, int datasync
);
2880 extern int sync_file_range(struct file
*file
, loff_t offset
, loff_t nbytes
,
2881 unsigned int flags
);
2884 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2885 * to already be updated for the write, and will return either the amount
2886 * of bytes passed in, or an error if syncing the file failed.
2888 static inline ssize_t
generic_write_sync(struct kiocb
*iocb
, ssize_t count
)
2890 if (iocb
->ki_flags
& IOCB_DSYNC
) {
2891 int ret
= vfs_fsync_range(iocb
->ki_filp
,
2892 iocb
->ki_pos
- count
, iocb
->ki_pos
- 1,
2893 (iocb
->ki_flags
& IOCB_SYNC
) ? 0 : 1);
2901 extern void emergency_sync(void);
2902 extern void emergency_remount(void);
2905 extern int bmap(struct inode
*inode
, sector_t
*block
);
2907 static inline int bmap(struct inode
*inode
, sector_t
*block
)
2913 extern int notify_change(struct dentry
*, struct iattr
*, struct inode
**);
2914 extern int inode_permission(struct inode
*, int);
2915 extern int generic_permission(struct inode
*, int);
2916 extern int __check_sticky(struct inode
*dir
, struct inode
*inode
);
2918 static inline bool execute_ok(struct inode
*inode
)
2920 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
2923 static inline void file_start_write(struct file
*file
)
2925 if (!S_ISREG(file_inode(file
)->i_mode
))
2927 __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, true);
2930 static inline bool file_start_write_trylock(struct file
*file
)
2932 if (!S_ISREG(file_inode(file
)->i_mode
))
2934 return __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, false);
2937 static inline void file_end_write(struct file
*file
)
2939 if (!S_ISREG(file_inode(file
)->i_mode
))
2941 __sb_end_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
);
2945 * get_write_access() gets write permission for a file.
2946 * put_write_access() releases this write permission.
2947 * This is used for regular files.
2948 * We cannot support write (and maybe mmap read-write shared) accesses and
2949 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2950 * can have the following values:
2951 * 0: no writers, no VM_DENYWRITE mappings
2952 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2953 * > 0: (i_writecount) users are writing to the file.
2955 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2956 * except for the cases where we don't hold i_writecount yet. Then we need to
2957 * use {get,deny}_write_access() - these functions check the sign and refuse
2958 * to do the change if sign is wrong.
2960 static inline int get_write_access(struct inode
*inode
)
2962 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2964 static inline int deny_write_access(struct file
*file
)
2966 struct inode
*inode
= file_inode(file
);
2967 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2969 static inline void put_write_access(struct inode
* inode
)
2971 atomic_dec(&inode
->i_writecount
);
2973 static inline void allow_write_access(struct file
*file
)
2976 atomic_inc(&file_inode(file
)->i_writecount
);
2978 static inline bool inode_is_open_for_write(const struct inode
*inode
)
2980 return atomic_read(&inode
->i_writecount
) > 0;
2983 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
2984 static inline void i_readcount_dec(struct inode
*inode
)
2986 BUG_ON(!atomic_read(&inode
->i_readcount
));
2987 atomic_dec(&inode
->i_readcount
);
2989 static inline void i_readcount_inc(struct inode
*inode
)
2991 atomic_inc(&inode
->i_readcount
);
2994 static inline void i_readcount_dec(struct inode
*inode
)
2998 static inline void i_readcount_inc(struct inode
*inode
)
3003 extern int do_pipe_flags(int *, int);
3005 #define __kernel_read_file_id(id) \
3006 id(UNKNOWN, unknown) \
3007 id(FIRMWARE, firmware) \
3008 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
3009 id(FIRMWARE_EFI_EMBEDDED, firmware) \
3010 id(MODULE, kernel-module) \
3011 id(KEXEC_IMAGE, kexec-image) \
3012 id(KEXEC_INITRAMFS, kexec-initramfs) \
3013 id(POLICY, security-policy) \
3014 id(X509_CERTIFICATE, x509-certificate) \
3017 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
3018 #define __fid_stringify(dummy, str) #str,
3020 enum kernel_read_file_id
{
3021 __kernel_read_file_id(__fid_enumify
)
3024 static const char * const kernel_read_file_str
[] = {
3025 __kernel_read_file_id(__fid_stringify
)
3028 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id
)
3030 if ((unsigned)id
>= READING_MAX_ID
)
3031 return kernel_read_file_str
[READING_UNKNOWN
];
3033 return kernel_read_file_str
[id
];
3036 extern int kernel_read_file(struct file
*, void **, loff_t
*, loff_t
,
3037 enum kernel_read_file_id
);
3038 extern int kernel_read_file_from_path(const char *, void **, loff_t
*, loff_t
,
3039 enum kernel_read_file_id
);
3040 extern int kernel_read_file_from_path_initns(const char *, void **, loff_t
*, loff_t
,
3041 enum kernel_read_file_id
);
3042 extern int kernel_read_file_from_fd(int, void **, loff_t
*, loff_t
,
3043 enum kernel_read_file_id
);
3044 extern ssize_t
kernel_read(struct file
*, void *, size_t, loff_t
*);
3045 extern ssize_t
kernel_write(struct file
*, const void *, size_t, loff_t
*);
3046 extern ssize_t
__kernel_write(struct file
*, const void *, size_t, loff_t
*);
3047 extern struct file
* open_exec(const char *);
3049 /* fs/dcache.c -- generic fs support functions */
3050 extern bool is_subdir(struct dentry
*, struct dentry
*);
3051 extern bool path_is_under(const struct path
*, const struct path
*);
3053 extern char *file_path(struct file
*, char *, int);
3055 #include <linux/err.h>
3057 /* needed for stackable file system support */
3058 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int whence
);
3060 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int whence
);
3062 extern int inode_init_always(struct super_block
*, struct inode
*);
3063 extern void inode_init_once(struct inode
*);
3064 extern void address_space_init_once(struct address_space
*mapping
);
3065 extern struct inode
* igrab(struct inode
*);
3066 extern ino_t
iunique(struct super_block
*, ino_t
);
3067 extern int inode_needs_sync(struct inode
*inode
);
3068 extern int generic_delete_inode(struct inode
*inode
);
3069 static inline int generic_drop_inode(struct inode
*inode
)
3071 return !inode
->i_nlink
|| inode_unhashed(inode
);
3074 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
3075 unsigned long hashval
, int (*test
)(struct inode
*, void *),
3077 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
3078 int (*test
)(struct inode
*, void *), void *data
);
3079 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
3081 extern struct inode
*inode_insert5(struct inode
*inode
, unsigned long hashval
,
3082 int (*test
)(struct inode
*, void *),
3083 int (*set
)(struct inode
*, void *),
3085 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
3086 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
3087 extern struct inode
*find_inode_nowait(struct super_block
*,
3089 int (*match
)(struct inode
*,
3090 unsigned long, void *),
3092 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
3093 extern int insert_inode_locked(struct inode
*);
3094 #ifdef CONFIG_DEBUG_LOCK_ALLOC
3095 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
3097 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
3099 extern void unlock_new_inode(struct inode
*);
3100 extern void discard_new_inode(struct inode
*);
3101 extern unsigned int get_next_ino(void);
3102 extern void evict_inodes(struct super_block
*sb
);
3104 extern void __iget(struct inode
* inode
);
3105 extern void iget_failed(struct inode
*);
3106 extern void clear_inode(struct inode
*);
3107 extern void __destroy_inode(struct inode
*);
3108 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
3109 extern struct inode
*new_inode(struct super_block
*sb
);
3110 extern void free_inode_nonrcu(struct inode
*inode
);
3111 extern int should_remove_suid(struct dentry
*);
3112 extern int file_remove_privs(struct file
*);
3114 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
3115 static inline void insert_inode_hash(struct inode
*inode
)
3117 __insert_inode_hash(inode
, inode
->i_ino
);
3120 extern void __remove_inode_hash(struct inode
*);
3121 static inline void remove_inode_hash(struct inode
*inode
)
3123 if (!inode_unhashed(inode
) && !hlist_fake(&inode
->i_hash
))
3124 __remove_inode_hash(inode
);
3127 extern void inode_sb_list_add(struct inode
*inode
);
3130 extern int bdev_read_only(struct block_device
*);
3132 extern int set_blocksize(struct block_device
*, int);
3133 extern int sb_set_blocksize(struct super_block
*, int);
3134 extern int sb_min_blocksize(struct super_block
*, int);
3136 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
3137 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
3138 extern ssize_t
generic_write_checks(struct kiocb
*, struct iov_iter
*);
3139 extern int generic_remap_checks(struct file
*file_in
, loff_t pos_in
,
3140 struct file
*file_out
, loff_t pos_out
,
3141 loff_t
*count
, unsigned int remap_flags
);
3142 extern int generic_file_rw_checks(struct file
*file_in
, struct file
*file_out
);
3143 extern int generic_copy_file_checks(struct file
*file_in
, loff_t pos_in
,
3144 struct file
*file_out
, loff_t pos_out
,
3145 size_t *count
, unsigned int flags
);
3146 extern ssize_t
generic_file_read_iter(struct kiocb
*, struct iov_iter
*);
3147 extern ssize_t
__generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
3148 extern ssize_t
generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
3149 extern ssize_t
generic_file_direct_write(struct kiocb
*, struct iov_iter
*);
3150 extern ssize_t
generic_perform_write(struct file
*, struct iov_iter
*, loff_t
);
3152 ssize_t
vfs_iter_read(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
3154 ssize_t
vfs_iter_write(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
3156 ssize_t
vfs_iocb_iter_read(struct file
*file
, struct kiocb
*iocb
,
3157 struct iov_iter
*iter
);
3158 ssize_t
vfs_iocb_iter_write(struct file
*file
, struct kiocb
*iocb
,
3159 struct iov_iter
*iter
);
3161 /* fs/block_dev.c */
3162 extern ssize_t
blkdev_read_iter(struct kiocb
*iocb
, struct iov_iter
*to
);
3163 extern ssize_t
blkdev_write_iter(struct kiocb
*iocb
, struct iov_iter
*from
);
3164 extern int blkdev_fsync(struct file
*filp
, loff_t start
, loff_t end
,
3166 extern void block_sync_page(struct page
*page
);
3169 extern ssize_t
generic_file_splice_read(struct file
*, loff_t
*,
3170 struct pipe_inode_info
*, size_t, unsigned int);
3171 extern ssize_t
iter_file_splice_write(struct pipe_inode_info
*,
3172 struct file
*, loff_t
*, size_t, unsigned int);
3173 extern ssize_t
generic_splice_sendpage(struct pipe_inode_info
*pipe
,
3174 struct file
*out
, loff_t
*, size_t len
, unsigned int flags
);
3175 extern long do_splice_direct(struct file
*in
, loff_t
*ppos
, struct file
*out
,
3176 loff_t
*opos
, size_t len
, unsigned int flags
);
3180 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
3181 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int whence
);
3182 extern loff_t
no_llseek(struct file
*file
, loff_t offset
, int whence
);
3183 extern loff_t
vfs_setpos(struct file
*file
, loff_t offset
, loff_t maxsize
);
3184 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int whence
);
3185 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
3186 int whence
, loff_t maxsize
, loff_t eof
);
3187 extern loff_t
fixed_size_llseek(struct file
*file
, loff_t offset
,
3188 int whence
, loff_t size
);
3189 extern loff_t
no_seek_end_llseek_size(struct file
*, loff_t
, int, loff_t
);
3190 extern loff_t
no_seek_end_llseek(struct file
*, loff_t
, int);
3191 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
3192 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
3193 extern int stream_open(struct inode
* inode
, struct file
* filp
);
3196 typedef void (dio_submit_t
)(struct bio
*bio
, struct inode
*inode
,
3197 loff_t file_offset
);
3200 /* need locking between buffered and direct access */
3203 /* filesystem does not support filling holes */
3204 DIO_SKIP_HOLES
= 0x02,
3207 void dio_end_io(struct bio
*bio
);
3209 ssize_t
__blockdev_direct_IO(struct kiocb
*iocb
, struct inode
*inode
,
3210 struct block_device
*bdev
, struct iov_iter
*iter
,
3211 get_block_t get_block
,
3212 dio_iodone_t end_io
, dio_submit_t submit_io
,
3215 static inline ssize_t
blockdev_direct_IO(struct kiocb
*iocb
,
3216 struct inode
*inode
,
3217 struct iov_iter
*iter
,
3218 get_block_t get_block
)
3220 return __blockdev_direct_IO(iocb
, inode
, inode
->i_sb
->s_bdev
, iter
,
3221 get_block
, NULL
, NULL
, DIO_LOCKING
| DIO_SKIP_HOLES
);
3225 void inode_dio_wait(struct inode
*inode
);
3228 * inode_dio_begin - signal start of a direct I/O requests
3229 * @inode: inode the direct I/O happens on
3231 * This is called once we've finished processing a direct I/O request,
3232 * and is used to wake up callers waiting for direct I/O to be quiesced.
3234 static inline void inode_dio_begin(struct inode
*inode
)
3236 atomic_inc(&inode
->i_dio_count
);
3240 * inode_dio_end - signal finish of a direct I/O requests
3241 * @inode: inode the direct I/O happens on
3243 * This is called once we've finished processing a direct I/O request,
3244 * and is used to wake up callers waiting for direct I/O to be quiesced.
3246 static inline void inode_dio_end(struct inode
*inode
)
3248 if (atomic_dec_and_test(&inode
->i_dio_count
))
3249 wake_up_bit(&inode
->i_state
, __I_DIO_WAKEUP
);
3253 * Warn about a page cache invalidation failure diring a direct I/O write.
3255 void dio_warn_stale_pagecache(struct file
*filp
);
3257 extern void inode_set_flags(struct inode
*inode
, unsigned int flags
,
3260 extern const struct file_operations generic_ro_fops
;
3262 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3264 extern int readlink_copy(char __user
*, int, const char *);
3265 extern int page_readlink(struct dentry
*, char __user
*, int);
3266 extern const char *page_get_link(struct dentry
*, struct inode
*,
3267 struct delayed_call
*);
3268 extern void page_put_link(void *);
3269 extern int __page_symlink(struct inode
*inode
, const char *symname
, int len
,
3271 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
3272 extern const struct inode_operations page_symlink_inode_operations
;
3273 extern void kfree_link(void *);
3274 extern void generic_fillattr(struct inode
*, struct kstat
*);
3275 extern int vfs_getattr_nosec(const struct path
*, struct kstat
*, u32
, unsigned int);
3276 extern int vfs_getattr(const struct path
*, struct kstat
*, u32
, unsigned int);
3277 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
3278 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
3279 void __inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
3280 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
3281 static inline loff_t
__inode_get_bytes(struct inode
*inode
)
3283 return (((loff_t
)inode
->i_blocks
) << 9) + inode
->i_bytes
;
3285 loff_t
inode_get_bytes(struct inode
*inode
);
3286 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
3287 const char *simple_get_link(struct dentry
*, struct inode
*,
3288 struct delayed_call
*);
3289 extern const struct inode_operations simple_symlink_inode_operations
;
3291 extern int iterate_dir(struct file
*, struct dir_context
*);
3293 extern int vfs_statx(int, const char __user
*, int, struct kstat
*, u32
);
3294 extern int vfs_statx_fd(unsigned int, struct kstat
*, u32
, unsigned int);
3296 static inline int vfs_stat(const char __user
*filename
, struct kstat
*stat
)
3298 return vfs_statx(AT_FDCWD
, filename
, AT_NO_AUTOMOUNT
,
3299 stat
, STATX_BASIC_STATS
);
3301 static inline int vfs_lstat(const char __user
*name
, struct kstat
*stat
)
3303 return vfs_statx(AT_FDCWD
, name
, AT_SYMLINK_NOFOLLOW
| AT_NO_AUTOMOUNT
,
3304 stat
, STATX_BASIC_STATS
);
3306 static inline int vfs_fstatat(int dfd
, const char __user
*filename
,
3307 struct kstat
*stat
, int flags
)
3309 return vfs_statx(dfd
, filename
, flags
| AT_NO_AUTOMOUNT
,
3310 stat
, STATX_BASIC_STATS
);
3312 static inline int vfs_fstat(int fd
, struct kstat
*stat
)
3314 return vfs_statx_fd(fd
, stat
, STATX_BASIC_STATS
, 0);
3318 extern const char *vfs_get_link(struct dentry
*, struct delayed_call
*);
3319 extern int vfs_readlink(struct dentry
*, char __user
*, int);
3321 extern int __generic_block_fiemap(struct inode
*inode
,
3322 struct fiemap_extent_info
*fieinfo
,
3323 loff_t start
, loff_t len
,
3324 get_block_t
*get_block
);
3325 extern int generic_block_fiemap(struct inode
*inode
,
3326 struct fiemap_extent_info
*fieinfo
, u64 start
,
3327 u64 len
, get_block_t
*get_block
);
3329 extern struct file_system_type
*get_filesystem(struct file_system_type
*fs
);
3330 extern void put_filesystem(struct file_system_type
*fs
);
3331 extern struct file_system_type
*get_fs_type(const char *name
);
3332 extern struct super_block
*get_super(struct block_device
*);
3333 extern struct super_block
*get_super_thawed(struct block_device
*);
3334 extern struct super_block
*get_super_exclusive_thawed(struct block_device
*bdev
);
3335 extern struct super_block
*get_active_super(struct block_device
*bdev
);
3336 extern void drop_super(struct super_block
*sb
);
3337 extern void drop_super_exclusive(struct super_block
*sb
);
3338 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
3339 extern void iterate_supers_type(struct file_system_type
*,
3340 void (*)(struct super_block
*, void *), void *);
3342 extern int dcache_dir_open(struct inode
*, struct file
*);
3343 extern int dcache_dir_close(struct inode
*, struct file
*);
3344 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
3345 extern int dcache_readdir(struct file
*, struct dir_context
*);
3346 extern int simple_setattr(struct dentry
*, struct iattr
*);
3347 extern int simple_getattr(const struct path
*, struct kstat
*, u32
, unsigned int);
3348 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
3349 extern int simple_open(struct inode
*inode
, struct file
*file
);
3350 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
3351 extern int simple_unlink(struct inode
*, struct dentry
*);
3352 extern int simple_rmdir(struct inode
*, struct dentry
*);
3353 extern int simple_rename(struct inode
*, struct dentry
*,
3354 struct inode
*, struct dentry
*, unsigned int);
3355 extern void simple_recursive_removal(struct dentry
*,
3356 void (*callback
)(struct dentry
*));
3357 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
3358 extern int noop_set_page_dirty(struct page
*page
);
3359 extern void noop_invalidatepage(struct page
*page
, unsigned int offset
,
3360 unsigned int length
);
3361 extern ssize_t
noop_direct_IO(struct kiocb
*iocb
, struct iov_iter
*iter
);
3362 extern int simple_empty(struct dentry
*);
3363 extern int simple_readpage(struct file
*file
, struct page
*page
);
3364 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
3365 loff_t pos
, unsigned len
, unsigned flags
,
3366 struct page
**pagep
, void **fsdata
);
3367 extern int simple_write_end(struct file
*file
, struct address_space
*mapping
,
3368 loff_t pos
, unsigned len
, unsigned copied
,
3369 struct page
*page
, void *fsdata
);
3370 extern int always_delete_dentry(const struct dentry
*);
3371 extern struct inode
*alloc_anon_inode(struct super_block
*);
3372 extern int simple_nosetlease(struct file
*, long, struct file_lock
**, void **);
3373 extern const struct dentry_operations simple_dentry_operations
;
3375 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
3376 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
3377 extern const struct file_operations simple_dir_operations
;
3378 extern const struct inode_operations simple_dir_inode_operations
;
3379 extern void make_empty_dir_inode(struct inode
*inode
);
3380 extern bool is_empty_dir_inode(struct inode
*inode
);
3381 struct tree_descr
{ const char *name
; const struct file_operations
*ops
; int mode
; };
3382 struct dentry
*d_alloc_name(struct dentry
*, const char *);
3383 extern int simple_fill_super(struct super_block
*, unsigned long,
3384 const struct tree_descr
*);
3385 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
3386 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
3388 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
3389 loff_t
*ppos
, const void *from
, size_t available
);
3390 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
3391 const void __user
*from
, size_t count
);
3393 extern int __generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3394 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3396 extern int generic_check_addressable(unsigned, u64
);
3398 #ifdef CONFIG_MIGRATION
3399 extern int buffer_migrate_page(struct address_space
*,
3400 struct page
*, struct page
*,
3402 extern int buffer_migrate_page_norefs(struct address_space
*,
3403 struct page
*, struct page
*,
3406 #define buffer_migrate_page NULL
3407 #define buffer_migrate_page_norefs NULL
3410 extern int setattr_prepare(struct dentry
*, struct iattr
*);
3411 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
3412 extern void setattr_copy(struct inode
*inode
, const struct iattr
*attr
);
3414 extern int file_update_time(struct file
*file
);
3416 static inline bool io_is_direct(struct file
*filp
)
3418 return (filp
->f_flags
& O_DIRECT
) || IS_DAX(filp
->f_mapping
->host
);
3421 static inline bool vma_is_dax(const struct vm_area_struct
*vma
)
3423 return vma
->vm_file
&& IS_DAX(vma
->vm_file
->f_mapping
->host
);
3426 static inline bool vma_is_fsdax(struct vm_area_struct
*vma
)
3428 struct inode
*inode
;
3432 if (!vma_is_dax(vma
))
3434 inode
= file_inode(vma
->vm_file
);
3435 if (S_ISCHR(inode
->i_mode
))
3436 return false; /* device-dax */
3440 static inline int iocb_flags(struct file
*file
)
3443 if (file
->f_flags
& O_APPEND
)
3445 if (io_is_direct(file
))
3447 if ((file
->f_flags
& O_DSYNC
) || IS_SYNC(file
->f_mapping
->host
))
3449 if (file
->f_flags
& __O_SYNC
)
3454 static inline int kiocb_set_rw_flags(struct kiocb
*ki
, rwf_t flags
)
3456 if (unlikely(flags
& ~RWF_SUPPORTED
))
3459 if (flags
& RWF_NOWAIT
) {
3460 if (!(ki
->ki_filp
->f_mode
& FMODE_NOWAIT
))
3462 ki
->ki_flags
|= IOCB_NOWAIT
;
3464 if (flags
& RWF_HIPRI
)
3465 ki
->ki_flags
|= IOCB_HIPRI
;
3466 if (flags
& RWF_DSYNC
)
3467 ki
->ki_flags
|= IOCB_DSYNC
;
3468 if (flags
& RWF_SYNC
)
3469 ki
->ki_flags
|= (IOCB_DSYNC
| IOCB_SYNC
);
3470 if (flags
& RWF_APPEND
)
3471 ki
->ki_flags
|= IOCB_APPEND
;
3475 static inline ino_t
parent_ino(struct dentry
*dentry
)
3480 * Don't strictly need d_lock here? If the parent ino could change
3481 * then surely we'd have a deeper race in the caller?
3483 spin_lock(&dentry
->d_lock
);
3484 res
= dentry
->d_parent
->d_inode
->i_ino
;
3485 spin_unlock(&dentry
->d_lock
);
3489 /* Transaction based IO helpers */
3492 * An argresp is stored in an allocated page and holds the
3493 * size of the argument or response, along with its content
3495 struct simple_transaction_argresp
{
3500 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3502 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
3504 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
3505 size_t size
, loff_t
*pos
);
3506 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
3508 void simple_transaction_set(struct file
*file
, size_t n
);
3511 * simple attribute files
3513 * These attributes behave similar to those in sysfs:
3515 * Writing to an attribute immediately sets a value, an open file can be
3516 * written to multiple times.
3518 * Reading from an attribute creates a buffer from the value that might get
3519 * read with multiple read calls. When the attribute has been read
3520 * completely, no further read calls are possible until the file is opened
3523 * All attributes contain a text representation of a numeric value
3524 * that are accessed with the get() and set() functions.
3526 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3527 static int __fops ## _open(struct inode *inode, struct file *file) \
3529 __simple_attr_check_format(__fmt, 0ull); \
3530 return simple_attr_open(inode, file, __get, __set, __fmt); \
3532 static const struct file_operations __fops = { \
3533 .owner = THIS_MODULE, \
3534 .open = __fops ## _open, \
3535 .release = simple_attr_release, \
3536 .read = simple_attr_read, \
3537 .write = simple_attr_write, \
3538 .llseek = generic_file_llseek, \
3541 static inline __printf(1, 2)
3542 void __simple_attr_check_format(const char *fmt
, ...)
3544 /* don't do anything, just let the compiler check the arguments; */
3547 int simple_attr_open(struct inode
*inode
, struct file
*file
,
3548 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
3550 int simple_attr_release(struct inode
*inode
, struct file
*file
);
3551 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
3552 size_t len
, loff_t
*ppos
);
3553 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
3554 size_t len
, loff_t
*ppos
);
3557 int proc_nr_files(struct ctl_table
*table
, int write
,
3558 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3559 int proc_nr_dentry(struct ctl_table
*table
, int write
,
3560 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3561 int proc_nr_inodes(struct ctl_table
*table
, int write
,
3562 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3563 int __init
get_filesystem_list(char *buf
);
3565 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3566 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3568 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3569 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3570 (flag & __FMODE_NONOTIFY)))
3572 static inline bool is_sxid(umode_t mode
)
3574 return (mode
& S_ISUID
) || ((mode
& S_ISGID
) && (mode
& S_IXGRP
));
3577 static inline int check_sticky(struct inode
*dir
, struct inode
*inode
)
3579 if (!(dir
->i_mode
& S_ISVTX
))
3582 return __check_sticky(dir
, inode
);
3585 static inline void inode_has_no_xattr(struct inode
*inode
)
3587 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& SB_NOSEC
))
3588 inode
->i_flags
|= S_NOSEC
;
3591 static inline bool is_root_inode(struct inode
*inode
)
3593 return inode
== inode
->i_sb
->s_root
->d_inode
;
3596 static inline bool dir_emit(struct dir_context
*ctx
,
3597 const char *name
, int namelen
,
3598 u64 ino
, unsigned type
)
3600 return ctx
->actor(ctx
, name
, namelen
, ctx
->pos
, ino
, type
) == 0;
3602 static inline bool dir_emit_dot(struct file
*file
, struct dir_context
*ctx
)
3604 return ctx
->actor(ctx
, ".", 1, ctx
->pos
,
3605 file
->f_path
.dentry
->d_inode
->i_ino
, DT_DIR
) == 0;
3607 static inline bool dir_emit_dotdot(struct file
*file
, struct dir_context
*ctx
)
3609 return ctx
->actor(ctx
, "..", 2, ctx
->pos
,
3610 parent_ino(file
->f_path
.dentry
), DT_DIR
) == 0;
3612 static inline bool dir_emit_dots(struct file
*file
, struct dir_context
*ctx
)
3614 if (ctx
->pos
== 0) {
3615 if (!dir_emit_dot(file
, ctx
))
3619 if (ctx
->pos
== 1) {
3620 if (!dir_emit_dotdot(file
, ctx
))
3626 static inline bool dir_relax(struct inode
*inode
)
3628 inode_unlock(inode
);
3630 return !IS_DEADDIR(inode
);
3633 static inline bool dir_relax_shared(struct inode
*inode
)
3635 inode_unlock_shared(inode
);
3636 inode_lock_shared(inode
);
3637 return !IS_DEADDIR(inode
);
3640 extern bool path_noexec(const struct path
*path
);
3641 extern void inode_nohighmem(struct inode
*inode
);
3644 extern int vfs_fadvise(struct file
*file
, loff_t offset
, loff_t len
,
3646 extern int generic_fadvise(struct file
*file
, loff_t offset
, loff_t len
,
3649 #if defined(CONFIG_IO_URING)
3650 extern struct sock
*io_uring_get_socket(struct file
*file
);
3652 static inline struct sock
*io_uring_get_socket(struct file
*file
)
3658 int vfs_ioc_setflags_prepare(struct inode
*inode
, unsigned int oldflags
,
3659 unsigned int flags
);
3661 int vfs_ioc_fssetxattr_check(struct inode
*inode
, const struct fsxattr
*old_fa
,
3662 struct fsxattr
*fa
);
3664 static inline void simple_fill_fsxattr(struct fsxattr
*fa
, __u32 xflags
)
3666 memset(fa
, 0, sizeof(*fa
));
3667 fa
->fsx_xflags
= xflags
;
3671 * Flush file data before changing attributes. Caller must hold any locks
3672 * required to prevent further writes to this file until we're done setting
3675 static inline int inode_drain_writes(struct inode
*inode
)
3677 inode_dio_wait(inode
);
3678 return filemap_write_and_wait(inode
->i_mapping
);
3681 #endif /* _LINUX_FS_H */