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>
41 #include <asm/byteorder.h>
42 #include <uapi/linux/fs.h>
44 struct backing_dev_info
;
47 struct export_operations
;
52 struct pipe_inode_info
;
53 struct poll_table_struct
;
55 struct vm_area_struct
;
58 struct swap_info_struct
;
60 struct workqueue_struct
;
63 struct fscrypt_operations
;
65 extern void __init
inode_init(void);
66 extern void __init
inode_init_early(void);
67 extern void __init
files_init(void);
68 extern void __init
files_maxfiles_init(void);
70 extern struct files_stat_struct files_stat
;
71 extern unsigned long get_max_files(void);
72 extern unsigned int sysctl_nr_open
;
73 extern struct inodes_stat_t inodes_stat
;
74 extern int leases_enable
, lease_break_time
;
75 extern int sysctl_protected_symlinks
;
76 extern int sysctl_protected_hardlinks
;
77 extern int sysctl_protected_fifos
;
78 extern int sysctl_protected_regular
;
80 typedef __kernel_rwf_t rwf_t
;
83 typedef int (get_block_t
)(struct inode
*inode
, sector_t iblock
,
84 struct buffer_head
*bh_result
, int create
);
85 typedef int (dio_iodone_t
)(struct kiocb
*iocb
, loff_t offset
,
86 ssize_t bytes
, void *private);
88 #define MAY_EXEC 0x00000001
89 #define MAY_WRITE 0x00000002
90 #define MAY_READ 0x00000004
91 #define MAY_APPEND 0x00000008
92 #define MAY_ACCESS 0x00000010
93 #define MAY_OPEN 0x00000020
94 #define MAY_CHDIR 0x00000040
95 /* called from RCU mode, don't block */
96 #define MAY_NOT_BLOCK 0x00000080
99 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
100 * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
103 /* file is open for reading */
104 #define FMODE_READ ((__force fmode_t)0x1)
105 /* file is open for writing */
106 #define FMODE_WRITE ((__force fmode_t)0x2)
107 /* file is seekable */
108 #define FMODE_LSEEK ((__force fmode_t)0x4)
109 /* file can be accessed using pread */
110 #define FMODE_PREAD ((__force fmode_t)0x8)
111 /* file can be accessed using pwrite */
112 #define FMODE_PWRITE ((__force fmode_t)0x10)
113 /* File is opened for execution with sys_execve / sys_uselib */
114 #define FMODE_EXEC ((__force fmode_t)0x20)
115 /* File is opened with O_NDELAY (only set for block devices) */
116 #define FMODE_NDELAY ((__force fmode_t)0x40)
117 /* File is opened with O_EXCL (only set for block devices) */
118 #define FMODE_EXCL ((__force fmode_t)0x80)
119 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
120 (specialy hack for floppy.c) */
121 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
122 /* 32bit hashes as llseek() offset (for directories) */
123 #define FMODE_32BITHASH ((__force fmode_t)0x200)
124 /* 64bit hashes as llseek() offset (for directories) */
125 #define FMODE_64BITHASH ((__force fmode_t)0x400)
128 * Don't update ctime and mtime.
130 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
131 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
133 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
135 /* Expect random access pattern */
136 #define FMODE_RANDOM ((__force fmode_t)0x1000)
138 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
139 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
141 /* File is opened with O_PATH; almost nothing can be done with it */
142 #define FMODE_PATH ((__force fmode_t)0x4000)
144 /* File needs atomic accesses to f_pos */
145 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
146 /* Write access to underlying fs */
147 #define FMODE_WRITER ((__force fmode_t)0x10000)
148 /* Has read method(s) */
149 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
150 /* Has write method(s) */
151 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
153 #define FMODE_OPENED ((__force fmode_t)0x80000)
154 #define FMODE_CREATED ((__force fmode_t)0x100000)
156 /* File was opened by fanotify and shouldn't generate fanotify events */
157 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
159 /* File is capable of returning -EAGAIN if I/O will block */
160 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
162 /* File does not contribute to nr_files count */
163 #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000)
166 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
167 * that indicates that they should check the contents of the iovec are
168 * valid, but not check the memory that the iovec elements
171 #define CHECK_IOVEC_ONLY -1
174 * Attribute flags. These should be or-ed together to figure out what
177 #define ATTR_MODE (1 << 0)
178 #define ATTR_UID (1 << 1)
179 #define ATTR_GID (1 << 2)
180 #define ATTR_SIZE (1 << 3)
181 #define ATTR_ATIME (1 << 4)
182 #define ATTR_MTIME (1 << 5)
183 #define ATTR_CTIME (1 << 6)
184 #define ATTR_ATIME_SET (1 << 7)
185 #define ATTR_MTIME_SET (1 << 8)
186 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
187 #define ATTR_KILL_SUID (1 << 11)
188 #define ATTR_KILL_SGID (1 << 12)
189 #define ATTR_FILE (1 << 13)
190 #define ATTR_KILL_PRIV (1 << 14)
191 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
192 #define ATTR_TIMES_SET (1 << 16)
193 #define ATTR_TOUCH (1 << 17)
196 * Whiteout is represented by a char device. The following constants define the
197 * mode and device number to use.
199 #define WHITEOUT_MODE 0
200 #define WHITEOUT_DEV 0
203 * This is the Inode Attributes structure, used for notify_change(). It
204 * uses the above definitions as flags, to know which values have changed.
205 * Also, in this manner, a Filesystem can look at only the values it cares
206 * about. Basically, these are the attributes that the VFS layer can
207 * request to change from the FS layer.
209 * Derek Atkins <warlord@MIT.EDU> 94-10-20
212 unsigned int ia_valid
;
217 struct timespec64 ia_atime
;
218 struct timespec64 ia_mtime
;
219 struct timespec64 ia_ctime
;
222 * Not an attribute, but an auxiliary info for filesystems wanting to
223 * implement an ftruncate() like method. NOTE: filesystem should
224 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
226 struct file
*ia_file
;
230 * Includes for diskquotas.
232 #include <linux/quota.h>
235 * Maximum number of layers of fs stack. Needs to be limited to
236 * prevent kernel stack overflow
238 #define FILESYSTEM_MAX_STACK_DEPTH 2
241 * enum positive_aop_returns - aop return codes with specific semantics
243 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
244 * completed, that the page is still locked, and
245 * should be considered active. The VM uses this hint
246 * to return the page to the active list -- it won't
247 * be a candidate for writeback again in the near
248 * future. Other callers must be careful to unlock
249 * the page if they get this return. Returned by
252 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
253 * unlocked it and the page might have been truncated.
254 * The caller should back up to acquiring a new page and
255 * trying again. The aop will be taking reasonable
256 * precautions not to livelock. If the caller held a page
257 * reference, it should drop it before retrying. Returned
260 * address_space_operation functions return these large constants to indicate
261 * special semantics to the caller. These are much larger than the bytes in a
262 * page to allow for functions that return the number of bytes operated on in a
266 enum positive_aop_returns
{
267 AOP_WRITEPAGE_ACTIVATE
= 0x80000,
268 AOP_TRUNCATED_PAGE
= 0x80001,
271 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
272 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
273 * helper code (eg buffer layer)
274 * to clear GFP_FS from alloc */
277 * oh the beauties of C type declarations.
280 struct address_space
;
281 struct writeback_control
;
284 * Write life time hint values.
285 * Stored in struct inode as u8.
288 WRITE_LIFE_NOT_SET
= 0,
289 WRITE_LIFE_NONE
= RWH_WRITE_LIFE_NONE
,
290 WRITE_LIFE_SHORT
= RWH_WRITE_LIFE_SHORT
,
291 WRITE_LIFE_MEDIUM
= RWH_WRITE_LIFE_MEDIUM
,
292 WRITE_LIFE_LONG
= RWH_WRITE_LIFE_LONG
,
293 WRITE_LIFE_EXTREME
= RWH_WRITE_LIFE_EXTREME
,
296 #define IOCB_EVENTFD (1 << 0)
297 #define IOCB_APPEND (1 << 1)
298 #define IOCB_DIRECT (1 << 2)
299 #define IOCB_HIPRI (1 << 3)
300 #define IOCB_DSYNC (1 << 4)
301 #define IOCB_SYNC (1 << 5)
302 #define IOCB_WRITE (1 << 6)
303 #define IOCB_NOWAIT (1 << 7)
306 struct file
*ki_filp
;
308 void (*ki_complete
)(struct kiocb
*iocb
, long ret
, long ret2
);
312 u16 ki_ioprio
; /* See linux/ioprio.h */
313 } __randomize_layout
;
315 static inline bool is_sync_kiocb(struct kiocb
*kiocb
)
317 return kiocb
->ki_complete
== NULL
;
321 * "descriptor" for what we're up to with a read.
322 * This allows us to use the same read code yet
323 * have multiple different users of the data that
324 * we read from a file.
326 * The simplest case just copies the data to user
339 typedef int (*read_actor_t
)(read_descriptor_t
*, struct page
*,
340 unsigned long, unsigned long);
342 struct address_space_operations
{
343 int (*writepage
)(struct page
*page
, struct writeback_control
*wbc
);
344 int (*readpage
)(struct file
*, struct page
*);
346 /* Write back some dirty pages from this mapping. */
347 int (*writepages
)(struct address_space
*, struct writeback_control
*);
349 /* Set a page dirty. Return true if this dirtied it */
350 int (*set_page_dirty
)(struct page
*page
);
353 * Reads in the requested pages. Unlike ->readpage(), this is
354 * PURELY used for read-ahead!.
356 int (*readpages
)(struct file
*filp
, struct address_space
*mapping
,
357 struct list_head
*pages
, unsigned nr_pages
);
359 int (*write_begin
)(struct file
*, struct address_space
*mapping
,
360 loff_t pos
, unsigned len
, unsigned flags
,
361 struct page
**pagep
, void **fsdata
);
362 int (*write_end
)(struct file
*, struct address_space
*mapping
,
363 loff_t pos
, unsigned len
, unsigned copied
,
364 struct page
*page
, void *fsdata
);
366 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
367 sector_t (*bmap
)(struct address_space
*, sector_t
);
368 void (*invalidatepage
) (struct page
*, unsigned int, unsigned int);
369 int (*releasepage
) (struct page
*, gfp_t
);
370 void (*freepage
)(struct page
*);
371 ssize_t (*direct_IO
)(struct kiocb
*, struct iov_iter
*iter
);
373 * migrate the contents of a page to the specified target. If
374 * migrate_mode is MIGRATE_ASYNC, it must not block.
376 int (*migratepage
) (struct address_space
*,
377 struct page
*, struct page
*, enum migrate_mode
);
378 bool (*isolate_page
)(struct page
*, isolate_mode_t
);
379 void (*putback_page
)(struct page
*);
380 int (*launder_page
) (struct page
*);
381 int (*is_partially_uptodate
) (struct page
*, unsigned long,
383 void (*is_dirty_writeback
) (struct page
*, bool *, bool *);
384 int (*error_remove_page
)(struct address_space
*, struct page
*);
386 /* swapfile support */
387 int (*swap_activate
)(struct swap_info_struct
*sis
, struct file
*file
,
389 void (*swap_deactivate
)(struct file
*file
);
392 extern const struct address_space_operations empty_aops
;
395 * pagecache_write_begin/pagecache_write_end must be used by general code
396 * to write into the pagecache.
398 int pagecache_write_begin(struct file
*, struct address_space
*mapping
,
399 loff_t pos
, unsigned len
, unsigned flags
,
400 struct page
**pagep
, void **fsdata
);
402 int pagecache_write_end(struct file
*, struct address_space
*mapping
,
403 loff_t pos
, unsigned len
, unsigned copied
,
404 struct page
*page
, void *fsdata
);
407 * struct address_space - Contents of a cacheable, mappable object.
408 * @host: Owner, either the inode or the block_device.
409 * @i_pages: Cached pages.
410 * @gfp_mask: Memory allocation flags to use for allocating pages.
411 * @i_mmap_writable: Number of VM_SHARED mappings.
412 * @i_mmap: Tree of private and shared mappings.
413 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
414 * @nrpages: Number of page entries, protected by the i_pages lock.
415 * @nrexceptional: Shadow or DAX entries, protected by the i_pages lock.
416 * @writeback_index: Writeback starts here.
418 * @flags: Error bits and flags (AS_*).
419 * @wb_err: The most recent error which has occurred.
420 * @private_lock: For use by the owner of the address_space.
421 * @private_list: For use by the owner of the address_space.
422 * @private_data: For use by the owner of the address_space.
424 struct address_space
{
426 struct xarray i_pages
;
428 atomic_t i_mmap_writable
;
429 struct rb_root_cached i_mmap
;
430 struct rw_semaphore i_mmap_rwsem
;
431 unsigned long nrpages
;
432 unsigned long nrexceptional
;
433 pgoff_t writeback_index
;
434 const struct address_space_operations
*a_ops
;
437 spinlock_t private_lock
;
438 struct list_head private_list
;
440 } __attribute__((aligned(sizeof(long)))) __randomize_layout
;
442 * On most architectures that alignment is already the case; but
443 * must be enforced here for CRIS, to let the least significant bit
444 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
446 struct request_queue
;
448 struct block_device
{
449 dev_t bd_dev
; /* not a kdev_t - it's a search key */
451 struct inode
* bd_inode
; /* will die */
452 struct super_block
* bd_super
;
453 struct mutex bd_mutex
; /* open/close mutex */
457 bool bd_write_holder
;
459 struct list_head bd_holder_disks
;
461 struct block_device
* bd_contains
;
462 unsigned bd_block_size
;
464 struct hd_struct
* bd_part
;
465 /* number of times partitions within this device have been opened. */
466 unsigned bd_part_count
;
468 struct gendisk
* bd_disk
;
469 struct request_queue
* bd_queue
;
470 struct backing_dev_info
*bd_bdi
;
471 struct list_head bd_list
;
473 * Private data. You must have bd_claim'ed the block_device
474 * to use this. NOTE: bd_claim allows an owner to claim
475 * the same device multiple times, the owner must take special
476 * care to not mess up bd_private for that case.
478 unsigned long bd_private
;
480 /* The counter of freeze processes */
481 int bd_fsfreeze_count
;
482 /* Mutex for freeze */
483 struct mutex bd_fsfreeze_mutex
;
484 } __randomize_layout
;
486 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
487 #define PAGECACHE_TAG_DIRTY XA_MARK_0
488 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
489 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
492 * Returns true if any of the pages in the mapping are marked with the tag.
494 static inline bool mapping_tagged(struct address_space
*mapping
, xa_mark_t tag
)
496 return xa_marked(&mapping
->i_pages
, tag
);
499 static inline void i_mmap_lock_write(struct address_space
*mapping
)
501 down_write(&mapping
->i_mmap_rwsem
);
504 static inline void i_mmap_unlock_write(struct address_space
*mapping
)
506 up_write(&mapping
->i_mmap_rwsem
);
509 static inline void i_mmap_lock_read(struct address_space
*mapping
)
511 down_read(&mapping
->i_mmap_rwsem
);
514 static inline void i_mmap_unlock_read(struct address_space
*mapping
)
516 up_read(&mapping
->i_mmap_rwsem
);
520 * Might pages of this file be mapped into userspace?
522 static inline int mapping_mapped(struct address_space
*mapping
)
524 return !RB_EMPTY_ROOT(&mapping
->i_mmap
.rb_root
);
528 * Might pages of this file have been modified in userspace?
529 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
530 * marks vma as VM_SHARED if it is shared, and the file was opened for
531 * writing i.e. vma may be mprotected writable even if now readonly.
533 * If i_mmap_writable is negative, no new writable mappings are allowed. You
534 * can only deny writable mappings, if none exists right now.
536 static inline int mapping_writably_mapped(struct address_space
*mapping
)
538 return atomic_read(&mapping
->i_mmap_writable
) > 0;
541 static inline int mapping_map_writable(struct address_space
*mapping
)
543 return atomic_inc_unless_negative(&mapping
->i_mmap_writable
) ?
547 static inline void mapping_unmap_writable(struct address_space
*mapping
)
549 atomic_dec(&mapping
->i_mmap_writable
);
552 static inline int mapping_deny_writable(struct address_space
*mapping
)
554 return atomic_dec_unless_positive(&mapping
->i_mmap_writable
) ?
558 static inline void mapping_allow_writable(struct address_space
*mapping
)
560 atomic_inc(&mapping
->i_mmap_writable
);
564 * Use sequence counter to get consistent i_size on 32-bit processors.
566 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
567 #include <linux/seqlock.h>
568 #define __NEED_I_SIZE_ORDERED
569 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
571 #define i_size_ordered_init(inode) do { } while (0)
575 #define ACL_NOT_CACHED ((void *)(-1))
576 #define ACL_DONT_CACHE ((void *)(-3))
578 static inline struct posix_acl
*
579 uncached_acl_sentinel(struct task_struct
*task
)
581 return (void *)task
+ 1;
585 is_uncached_acl(struct posix_acl
*acl
)
587 return (long)acl
& 1;
590 #define IOP_FASTPERM 0x0001
591 #define IOP_LOOKUP 0x0002
592 #define IOP_NOFOLLOW 0x0004
593 #define IOP_XATTR 0x0008
594 #define IOP_DEFAULT_READLINK 0x0010
596 struct fsnotify_mark_connector
;
599 * Keep mostly read-only and often accessed (especially for
600 * the RCU path lookup and 'stat' data) fields at the beginning
601 * of the 'struct inode'
605 unsigned short i_opflags
;
608 unsigned int i_flags
;
610 #ifdef CONFIG_FS_POSIX_ACL
611 struct posix_acl
*i_acl
;
612 struct posix_acl
*i_default_acl
;
615 const struct inode_operations
*i_op
;
616 struct super_block
*i_sb
;
617 struct address_space
*i_mapping
;
619 #ifdef CONFIG_SECURITY
623 /* Stat data, not accessed from path walking */
626 * Filesystems may only read i_nlink directly. They shall use the
627 * following functions for modification:
629 * (set|clear|inc|drop)_nlink
630 * inode_(inc|dec)_link_count
633 const unsigned int i_nlink
;
634 unsigned int __i_nlink
;
638 struct timespec64 i_atime
;
639 struct timespec64 i_mtime
;
640 struct timespec64 i_ctime
;
641 spinlock_t i_lock
; /* i_blocks, i_bytes, maybe i_size */
642 unsigned short i_bytes
;
647 #ifdef __NEED_I_SIZE_ORDERED
648 seqcount_t i_size_seqcount
;
652 unsigned long i_state
;
653 struct rw_semaphore i_rwsem
;
655 unsigned long dirtied_when
; /* jiffies of first dirtying */
656 unsigned long dirtied_time_when
;
658 struct hlist_node i_hash
;
659 struct list_head i_io_list
; /* backing dev IO list */
660 #ifdef CONFIG_CGROUP_WRITEBACK
661 struct bdi_writeback
*i_wb
; /* the associated cgroup wb */
663 /* foreign inode detection, see wbc_detach_inode() */
665 u16 i_wb_frn_avg_time
;
666 u16 i_wb_frn_history
;
668 struct list_head i_lru
; /* inode LRU list */
669 struct list_head i_sb_list
;
670 struct list_head i_wb_list
; /* backing dev writeback list */
672 struct hlist_head i_dentry
;
673 struct rcu_head i_rcu
;
675 atomic64_t i_version
;
677 atomic_t i_dio_count
;
678 atomic_t i_writecount
;
680 atomic_t i_readcount
; /* struct files open RO */
682 const struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
683 struct file_lock_context
*i_flctx
;
684 struct address_space i_data
;
685 struct list_head i_devices
;
687 struct pipe_inode_info
*i_pipe
;
688 struct block_device
*i_bdev
;
696 #ifdef CONFIG_FSNOTIFY
697 __u32 i_fsnotify_mask
; /* all events this inode cares about */
698 struct fsnotify_mark_connector __rcu
*i_fsnotify_marks
;
701 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
702 struct fscrypt_info
*i_crypt_info
;
705 void *i_private
; /* fs or device private pointer */
706 } __randomize_layout
;
708 static inline unsigned int i_blocksize(const struct inode
*node
)
710 return (1 << node
->i_blkbits
);
713 static inline int inode_unhashed(struct inode
*inode
)
715 return hlist_unhashed(&inode
->i_hash
);
719 * __mark_inode_dirty expects inodes to be hashed. Since we don't
720 * want special inodes in the fileset inode space, we make them
721 * appear hashed, but do not put on any lists. hlist_del()
722 * will work fine and require no locking.
724 static inline void inode_fake_hash(struct inode
*inode
)
726 hlist_add_fake(&inode
->i_hash
);
730 * inode->i_mutex nesting subclasses for the lock validator:
732 * 0: the object of the current VFS operation
736 * 4: second non-directory
737 * 5: second parent (when locking independent directories in rename)
739 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
740 * non-directories at once.
742 * The locking order between these classes is
743 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
745 enum inode_i_mutex_lock_class
755 static inline void inode_lock(struct inode
*inode
)
757 down_write(&inode
->i_rwsem
);
760 static inline void inode_unlock(struct inode
*inode
)
762 up_write(&inode
->i_rwsem
);
765 static inline void inode_lock_shared(struct inode
*inode
)
767 down_read(&inode
->i_rwsem
);
770 static inline void inode_unlock_shared(struct inode
*inode
)
772 up_read(&inode
->i_rwsem
);
775 static inline int inode_trylock(struct inode
*inode
)
777 return down_write_trylock(&inode
->i_rwsem
);
780 static inline int inode_trylock_shared(struct inode
*inode
)
782 return down_read_trylock(&inode
->i_rwsem
);
785 static inline int inode_is_locked(struct inode
*inode
)
787 return rwsem_is_locked(&inode
->i_rwsem
);
790 static inline void inode_lock_nested(struct inode
*inode
, unsigned subclass
)
792 down_write_nested(&inode
->i_rwsem
, subclass
);
795 static inline void inode_lock_shared_nested(struct inode
*inode
, unsigned subclass
)
797 down_read_nested(&inode
->i_rwsem
, subclass
);
800 void lock_two_nondirectories(struct inode
*, struct inode
*);
801 void unlock_two_nondirectories(struct inode
*, struct inode
*);
804 * NOTE: in a 32bit arch with a preemptable kernel and
805 * an UP compile the i_size_read/write must be atomic
806 * with respect to the local cpu (unlike with preempt disabled),
807 * but they don't need to be atomic with respect to other cpus like in
808 * true SMP (so they need either to either locally disable irq around
809 * the read or for example on x86 they can be still implemented as a
810 * cmpxchg8b without the need of the lock prefix). For SMP compiles
811 * and 64bit archs it makes no difference if preempt is enabled or not.
813 static inline loff_t
i_size_read(const struct inode
*inode
)
815 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
820 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
821 i_size
= inode
->i_size
;
822 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
824 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
828 i_size
= inode
->i_size
;
832 return inode
->i_size
;
837 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
838 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
839 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
841 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
843 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
845 write_seqcount_begin(&inode
->i_size_seqcount
);
846 inode
->i_size
= i_size
;
847 write_seqcount_end(&inode
->i_size_seqcount
);
849 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
851 inode
->i_size
= i_size
;
854 inode
->i_size
= i_size
;
858 static inline unsigned iminor(const struct inode
*inode
)
860 return MINOR(inode
->i_rdev
);
863 static inline unsigned imajor(const struct inode
*inode
)
865 return MAJOR(inode
->i_rdev
);
868 extern struct block_device
*I_BDEV(struct inode
*inode
);
871 rwlock_t lock
; /* protects pid, uid, euid fields */
872 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
873 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
874 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
875 int signum
; /* posix.1b rt signal to be delivered on IO */
879 * Track a single file's readahead state
881 struct file_ra_state
{
882 pgoff_t start
; /* where readahead started */
883 unsigned int size
; /* # of readahead pages */
884 unsigned int async_size
; /* do asynchronous readahead when
885 there are only # of pages ahead */
887 unsigned int ra_pages
; /* Maximum readahead window */
888 unsigned int mmap_miss
; /* Cache miss stat for mmap accesses */
889 loff_t prev_pos
; /* Cache last read() position */
893 * Check if @index falls in the readahead windows.
895 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
897 return (index
>= ra
->start
&&
898 index
< ra
->start
+ ra
->size
);
903 struct llist_node fu_llist
;
904 struct rcu_head fu_rcuhead
;
907 struct inode
*f_inode
; /* cached value */
908 const struct file_operations
*f_op
;
911 * Protects f_ep_links, f_flags.
912 * Must not be taken from IRQ context.
915 enum rw_hint f_write_hint
;
916 atomic_long_t f_count
;
917 unsigned int f_flags
;
919 struct mutex f_pos_lock
;
921 struct fown_struct f_owner
;
922 const struct cred
*f_cred
;
923 struct file_ra_state f_ra
;
926 #ifdef CONFIG_SECURITY
929 /* needed for tty driver, and maybe others */
933 /* Used by fs/eventpoll.c to link all the hooks to this file */
934 struct list_head f_ep_links
;
935 struct list_head f_tfile_llink
;
936 #endif /* #ifdef CONFIG_EPOLL */
937 struct address_space
*f_mapping
;
940 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
945 /* file identifier */
946 unsigned char f_handle
[0];
949 static inline struct file
*get_file(struct file
*f
)
951 atomic_long_inc(&f
->f_count
);
954 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
955 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
956 #define file_count(x) atomic_long_read(&(x)->f_count)
958 #define MAX_NON_LFS ((1UL<<31) - 1)
960 /* Page cache limit. The filesystems should put that into their s_maxbytes
961 limits, otherwise bad things can happen in VM. */
962 #if BITS_PER_LONG==32
963 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
964 #elif BITS_PER_LONG==64
965 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
970 #define FL_DELEG 4 /* NFSv4 delegation */
971 #define FL_ACCESS 8 /* not trying to lock, just looking */
972 #define FL_EXISTS 16 /* when unlocking, test for existence */
973 #define FL_LEASE 32 /* lease held on this file */
974 #define FL_CLOSE 64 /* unlock on close */
975 #define FL_SLEEP 128 /* A blocking lock */
976 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
977 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
978 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
979 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
981 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
984 * Special return value from posix_lock_file() and vfs_lock_file() for
985 * asynchronous locking.
987 #define FILE_LOCK_DEFERRED 1
989 /* legacy typedef, should eventually be removed */
990 typedef void *fl_owner_t
;
994 struct file_lock_operations
{
995 void (*fl_copy_lock
)(struct file_lock
*, struct file_lock
*);
996 void (*fl_release_private
)(struct file_lock
*);
999 struct lock_manager_operations
{
1000 int (*lm_compare_owner
)(struct file_lock
*, struct file_lock
*);
1001 unsigned long (*lm_owner_key
)(struct file_lock
*);
1002 fl_owner_t (*lm_get_owner
)(fl_owner_t
);
1003 void (*lm_put_owner
)(fl_owner_t
);
1004 void (*lm_notify
)(struct file_lock
*); /* unblock callback */
1005 int (*lm_grant
)(struct file_lock
*, int);
1006 bool (*lm_break
)(struct file_lock
*);
1007 int (*lm_change
)(struct file_lock
*, int, struct list_head
*);
1008 void (*lm_setup
)(struct file_lock
*, void **);
1011 struct lock_manager
{
1012 struct list_head list
;
1014 * NFSv4 and up also want opens blocked during the grace period;
1021 void locks_start_grace(struct net
*, struct lock_manager
*);
1022 void locks_end_grace(struct lock_manager
*);
1023 bool locks_in_grace(struct net
*);
1024 bool opens_in_grace(struct net
*);
1026 /* that will die - we need it for nfs_lock_info */
1027 #include <linux/nfs_fs_i.h>
1030 * struct file_lock represents a generic "file lock". It's used to represent
1031 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1032 * note that the same struct is used to represent both a request for a lock and
1033 * the lock itself, but the same object is never used for both.
1035 * FIXME: should we create a separate "struct lock_request" to help distinguish
1038 * The varous i_flctx lists are ordered by:
1041 * 2) lock range start
1044 * Obviously, the last two criteria only matter for POSIX locks.
1047 struct file_lock
*fl_next
; /* singly linked list for this inode */
1048 struct list_head fl_list
; /* link into file_lock_context */
1049 struct hlist_node fl_link
; /* node in global lists */
1050 struct list_head fl_block
; /* circular list of blocked processes */
1051 fl_owner_t fl_owner
;
1052 unsigned int fl_flags
;
1053 unsigned char fl_type
;
1054 unsigned int fl_pid
;
1055 int fl_link_cpu
; /* what cpu's list is this on? */
1056 wait_queue_head_t fl_wait
;
1057 struct file
*fl_file
;
1061 struct fasync_struct
* fl_fasync
; /* for lease break notifications */
1062 /* for lease breaks: */
1063 unsigned long fl_break_time
;
1064 unsigned long fl_downgrade_time
;
1066 const struct file_lock_operations
*fl_ops
; /* Callbacks for filesystems */
1067 const struct lock_manager_operations
*fl_lmops
; /* Callbacks for lockmanagers */
1069 struct nfs_lock_info nfs_fl
;
1070 struct nfs4_lock_info nfs4_fl
;
1072 struct list_head link
; /* link in AFS vnode's pending_locks list */
1073 int state
; /* state of grant or error if -ve */
1076 } __randomize_layout
;
1078 struct file_lock_context
{
1079 spinlock_t flc_lock
;
1080 struct list_head flc_flock
;
1081 struct list_head flc_posix
;
1082 struct list_head flc_lease
;
1085 /* The following constant reflects the upper bound of the file/locking space */
1087 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1088 #define OFFSET_MAX INT_LIMIT(loff_t)
1089 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1092 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
1094 #define locks_inode(f) file_inode(f)
1096 #ifdef CONFIG_FILE_LOCKING
1097 extern int fcntl_getlk(struct file
*, unsigned int, struct flock
*);
1098 extern int fcntl_setlk(unsigned int, struct file
*, unsigned int,
1101 #if BITS_PER_LONG == 32
1102 extern int fcntl_getlk64(struct file
*, unsigned int, struct flock64
*);
1103 extern int fcntl_setlk64(unsigned int, struct file
*, unsigned int,
1107 extern int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
);
1108 extern int fcntl_getlease(struct file
*filp
);
1111 void locks_free_lock_context(struct inode
*inode
);
1112 void locks_free_lock(struct file_lock
*fl
);
1113 extern void locks_init_lock(struct file_lock
*);
1114 extern struct file_lock
* locks_alloc_lock(void);
1115 extern void locks_copy_lock(struct file_lock
*, struct file_lock
*);
1116 extern void locks_copy_conflock(struct file_lock
*, struct file_lock
*);
1117 extern void locks_remove_posix(struct file
*, fl_owner_t
);
1118 extern void locks_remove_file(struct file
*);
1119 extern void locks_release_private(struct file_lock
*);
1120 extern void posix_test_lock(struct file
*, struct file_lock
*);
1121 extern int posix_lock_file(struct file
*, struct file_lock
*, struct file_lock
*);
1122 extern int posix_unblock_lock(struct file_lock
*);
1123 extern int vfs_test_lock(struct file
*, struct file_lock
*);
1124 extern int vfs_lock_file(struct file
*, unsigned int, struct file_lock
*, struct file_lock
*);
1125 extern int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
);
1126 extern int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
);
1127 extern int __break_lease(struct inode
*inode
, unsigned int flags
, unsigned int type
);
1128 extern void lease_get_mtime(struct inode
*, struct timespec64
*time
);
1129 extern int generic_setlease(struct file
*, long, struct file_lock
**, void **priv
);
1130 extern int vfs_setlease(struct file
*, long, struct file_lock
**, void **);
1131 extern int lease_modify(struct file_lock
*, int, struct list_head
*);
1132 struct files_struct
;
1133 extern void show_fd_locks(struct seq_file
*f
,
1134 struct file
*filp
, struct files_struct
*files
);
1135 #else /* !CONFIG_FILE_LOCKING */
1136 static inline int fcntl_getlk(struct file
*file
, unsigned int cmd
,
1137 struct flock __user
*user
)
1142 static inline int fcntl_setlk(unsigned int fd
, struct file
*file
,
1143 unsigned int cmd
, struct flock __user
*user
)
1148 #if BITS_PER_LONG == 32
1149 static inline int fcntl_getlk64(struct file
*file
, unsigned int cmd
,
1150 struct flock64 __user
*user
)
1155 static inline int fcntl_setlk64(unsigned int fd
, struct file
*file
,
1156 unsigned int cmd
, struct flock64 __user
*user
)
1161 static inline int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
)
1166 static inline int fcntl_getlease(struct file
*filp
)
1172 locks_free_lock_context(struct inode
*inode
)
1176 static inline void locks_init_lock(struct file_lock
*fl
)
1181 static inline void locks_copy_conflock(struct file_lock
*new, struct file_lock
*fl
)
1186 static inline void locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
1191 static inline void locks_remove_posix(struct file
*filp
, fl_owner_t owner
)
1196 static inline void locks_remove_file(struct file
*filp
)
1201 static inline void posix_test_lock(struct file
*filp
, struct file_lock
*fl
)
1206 static inline int posix_lock_file(struct file
*filp
, struct file_lock
*fl
,
1207 struct file_lock
*conflock
)
1212 static inline int posix_unblock_lock(struct file_lock
*waiter
)
1217 static inline int vfs_test_lock(struct file
*filp
, struct file_lock
*fl
)
1222 static inline int vfs_lock_file(struct file
*filp
, unsigned int cmd
,
1223 struct file_lock
*fl
, struct file_lock
*conf
)
1228 static inline int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
)
1233 static inline int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
)
1238 static inline int __break_lease(struct inode
*inode
, unsigned int mode
, unsigned int type
)
1243 static inline void lease_get_mtime(struct inode
*inode
,
1244 struct timespec64
*time
)
1249 static inline int generic_setlease(struct file
*filp
, long arg
,
1250 struct file_lock
**flp
, void **priv
)
1255 static inline int vfs_setlease(struct file
*filp
, long arg
,
1256 struct file_lock
**lease
, void **priv
)
1261 static inline int lease_modify(struct file_lock
*fl
, int arg
,
1262 struct list_head
*dispose
)
1267 struct files_struct
;
1268 static inline void show_fd_locks(struct seq_file
*f
,
1269 struct file
*filp
, struct files_struct
*files
) {}
1270 #endif /* !CONFIG_FILE_LOCKING */
1272 static inline struct inode
*file_inode(const struct file
*f
)
1277 static inline struct dentry
*file_dentry(const struct file
*file
)
1279 return d_real(file
->f_path
.dentry
, file_inode(file
));
1282 static inline int locks_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1284 return locks_lock_inode_wait(locks_inode(filp
), fl
);
1287 struct fasync_struct
{
1291 struct fasync_struct
*fa_next
; /* singly linked list */
1292 struct file
*fa_file
;
1293 struct rcu_head fa_rcu
;
1296 #define FASYNC_MAGIC 0x4601
1298 /* SMP safe fasync helpers: */
1299 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1300 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1301 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1302 extern struct fasync_struct
*fasync_alloc(void);
1303 extern void fasync_free(struct fasync_struct
*);
1305 /* can be called from interrupts */
1306 extern void kill_fasync(struct fasync_struct
**, int, int);
1308 extern void __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1309 extern int f_setown(struct file
*filp
, unsigned long arg
, int force
);
1310 extern void f_delown(struct file
*filp
);
1311 extern pid_t
f_getown(struct file
*filp
);
1312 extern int send_sigurg(struct fown_struct
*fown
);
1315 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1316 * represented in both.
1318 #define SB_RDONLY 1 /* Mount read-only */
1319 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1320 #define SB_NODEV 4 /* Disallow access to device special files */
1321 #define SB_NOEXEC 8 /* Disallow program execution */
1322 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1323 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1324 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1325 #define SB_NOATIME 1024 /* Do not update access times. */
1326 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1327 #define SB_SILENT 32768
1328 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1329 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1330 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1331 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1333 /* These sb flags are internal to the kernel */
1334 #define SB_SUBMOUNT (1<<26)
1335 #define SB_NOSEC (1<<28)
1336 #define SB_BORN (1<<29)
1337 #define SB_ACTIVE (1<<30)
1338 #define SB_NOUSER (1<<31)
1344 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1345 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1346 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1347 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1348 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1351 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1352 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1353 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1354 #define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */
1356 /* sb->s_iflags to limit user namespace mounts */
1357 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1358 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1359 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1361 /* Possible states of 'frozen' field */
1363 SB_UNFROZEN
= 0, /* FS is unfrozen */
1364 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1365 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1366 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1367 * internal threads if needed) */
1368 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1371 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1374 int frozen
; /* Is sb frozen? */
1375 wait_queue_head_t wait_unfrozen
; /* for get_super_thawed() */
1376 struct percpu_rw_semaphore rw_sem
[SB_FREEZE_LEVELS
];
1379 struct super_block
{
1380 struct list_head s_list
; /* Keep this first */
1381 dev_t s_dev
; /* search index; _not_ kdev_t */
1382 unsigned char s_blocksize_bits
;
1383 unsigned long s_blocksize
;
1384 loff_t s_maxbytes
; /* Max file size */
1385 struct file_system_type
*s_type
;
1386 const struct super_operations
*s_op
;
1387 const struct dquot_operations
*dq_op
;
1388 const struct quotactl_ops
*s_qcop
;
1389 const struct export_operations
*s_export_op
;
1390 unsigned long s_flags
;
1391 unsigned long s_iflags
; /* internal SB_I_* flags */
1392 unsigned long s_magic
;
1393 struct dentry
*s_root
;
1394 struct rw_semaphore s_umount
;
1397 #ifdef CONFIG_SECURITY
1400 const struct xattr_handler
**s_xattr
;
1401 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
1402 const struct fscrypt_operations
*s_cop
;
1404 struct hlist_bl_head s_roots
; /* alternate root dentries for NFS */
1405 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1406 struct block_device
*s_bdev
;
1407 struct backing_dev_info
*s_bdi
;
1408 struct mtd_info
*s_mtd
;
1409 struct hlist_node s_instances
;
1410 unsigned int s_quota_types
; /* Bitmask of supported quota types */
1411 struct quota_info s_dquot
; /* Diskquota specific options */
1413 struct sb_writers s_writers
;
1416 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1417 * s_fsnotify_marks together for cache efficiency. They are frequently
1418 * accessed and rarely modified.
1420 void *s_fs_info
; /* Filesystem private info */
1422 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1424 #ifdef CONFIG_FSNOTIFY
1425 __u32 s_fsnotify_mask
;
1426 struct fsnotify_mark_connector __rcu
*s_fsnotify_marks
;
1429 char s_id
[32]; /* Informational name */
1430 uuid_t s_uuid
; /* UUID */
1432 unsigned int s_max_links
;
1436 * The next field is for VFS *only*. No filesystems have any business
1437 * even looking at it. You had been warned.
1439 struct mutex s_vfs_rename_mutex
; /* Kludge */
1442 * Filesystem subtype. If non-empty the filesystem type field
1443 * in /proc/mounts will be "type.subtype"
1447 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1450 * Saved pool identifier for cleancache (-1 means none)
1452 int cleancache_poolid
;
1454 struct shrinker s_shrink
; /* per-sb shrinker handle */
1456 /* Number of inodes with nlink == 0 but still referenced */
1457 atomic_long_t s_remove_count
;
1459 /* Pending fsnotify inode refs */
1460 atomic_long_t s_fsnotify_inode_refs
;
1462 /* Being remounted read-only */
1463 int s_readonly_remount
;
1465 /* AIO completions deferred from interrupt context */
1466 struct workqueue_struct
*s_dio_done_wq
;
1467 struct hlist_head s_pins
;
1470 * Owning user namespace and default context in which to
1471 * interpret filesystem uids, gids, quotas, device nodes,
1472 * xattrs and security labels.
1474 struct user_namespace
*s_user_ns
;
1477 * Keep the lru lists last in the structure so they always sit on their
1478 * own individual cachelines.
1480 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp
;
1481 struct list_lru s_inode_lru ____cacheline_aligned_in_smp
;
1482 struct rcu_head rcu
;
1483 struct work_struct destroy_work
;
1485 struct mutex s_sync_lock
; /* sync serialisation lock */
1488 * Indicates how deep in a filesystem stack this SB is
1492 /* s_inode_list_lock protects s_inodes */
1493 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp
;
1494 struct list_head s_inodes
; /* all inodes */
1496 spinlock_t s_inode_wblist_lock
;
1497 struct list_head s_inodes_wb
; /* writeback inodes */
1498 } __randomize_layout
;
1500 /* Helper functions so that in most cases filesystems will
1501 * not need to deal directly with kuid_t and kgid_t and can
1502 * instead deal with the raw numeric values that are stored
1503 * in the filesystem.
1505 static inline uid_t
i_uid_read(const struct inode
*inode
)
1507 return from_kuid(inode
->i_sb
->s_user_ns
, inode
->i_uid
);
1510 static inline gid_t
i_gid_read(const struct inode
*inode
)
1512 return from_kgid(inode
->i_sb
->s_user_ns
, inode
->i_gid
);
1515 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
1517 inode
->i_uid
= make_kuid(inode
->i_sb
->s_user_ns
, uid
);
1520 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
1522 inode
->i_gid
= make_kgid(inode
->i_sb
->s_user_ns
, gid
);
1525 extern struct timespec64
timespec64_trunc(struct timespec64 t
, unsigned gran
);
1526 extern struct timespec64
current_time(struct inode
*inode
);
1529 * Snapshotting support.
1532 void __sb_end_write(struct super_block
*sb
, int level
);
1533 int __sb_start_write(struct super_block
*sb
, int level
, bool wait
);
1535 #define __sb_writers_acquired(sb, lev) \
1536 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1537 #define __sb_writers_release(sb, lev) \
1538 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1541 * sb_end_write - drop write access to a superblock
1542 * @sb: the super we wrote to
1544 * Decrement number of writers to the filesystem. Wake up possible waiters
1545 * wanting to freeze the filesystem.
1547 static inline void sb_end_write(struct super_block
*sb
)
1549 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1553 * sb_end_pagefault - drop write access to a superblock from a page fault
1554 * @sb: the super we wrote to
1556 * Decrement number of processes handling write page fault to the filesystem.
1557 * Wake up possible waiters wanting to freeze the filesystem.
1559 static inline void sb_end_pagefault(struct super_block
*sb
)
1561 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1565 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1566 * @sb: the super we wrote to
1568 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1569 * waiters wanting to freeze the filesystem.
1571 static inline void sb_end_intwrite(struct super_block
*sb
)
1573 __sb_end_write(sb
, SB_FREEZE_FS
);
1577 * sb_start_write - get write access to a superblock
1578 * @sb: the super we write to
1580 * When a process wants to write data or metadata to a file system (i.e. dirty
1581 * a page or an inode), it should embed the operation in a sb_start_write() -
1582 * sb_end_write() pair to get exclusion against file system freezing. This
1583 * function increments number of writers preventing freezing. If the file
1584 * system is already frozen, the function waits until the file system is
1587 * Since freeze protection behaves as a lock, users have to preserve
1588 * ordering of freeze protection and other filesystem locks. Generally,
1589 * freeze protection should be the outermost lock. In particular, we have:
1592 * -> i_mutex (write path, truncate, directory ops, ...)
1593 * -> s_umount (freeze_super, thaw_super)
1595 static inline void sb_start_write(struct super_block
*sb
)
1597 __sb_start_write(sb
, SB_FREEZE_WRITE
, true);
1600 static inline int sb_start_write_trylock(struct super_block
*sb
)
1602 return __sb_start_write(sb
, SB_FREEZE_WRITE
, false);
1606 * sb_start_pagefault - get write access to a superblock from a page fault
1607 * @sb: the super we write to
1609 * When a process starts handling write page fault, it should embed the
1610 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1611 * exclusion against file system freezing. This is needed since the page fault
1612 * is going to dirty a page. This function increments number of running page
1613 * faults preventing freezing. If the file system is already frozen, the
1614 * function waits until the file system is thawed.
1616 * Since page fault freeze protection behaves as a lock, users have to preserve
1617 * ordering of freeze protection and other filesystem locks. It is advised to
1618 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1619 * handling code implies lock dependency:
1622 * -> sb_start_pagefault
1624 static inline void sb_start_pagefault(struct super_block
*sb
)
1626 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
, true);
1630 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1631 * @sb: the super we write to
1633 * This is the third level of protection against filesystem freezing. It is
1634 * free for use by a filesystem. The only requirement is that it must rank
1635 * below sb_start_pagefault.
1637 * For example filesystem can call sb_start_intwrite() when starting a
1638 * transaction which somewhat eases handling of freezing for internal sources
1639 * of filesystem changes (internal fs threads, discarding preallocation on file
1642 static inline void sb_start_intwrite(struct super_block
*sb
)
1644 __sb_start_write(sb
, SB_FREEZE_FS
, true);
1647 static inline int sb_start_intwrite_trylock(struct super_block
*sb
)
1649 return __sb_start_write(sb
, SB_FREEZE_FS
, false);
1653 extern bool inode_owner_or_capable(const struct inode
*inode
);
1656 * VFS helper functions..
1658 extern int vfs_create(struct inode
*, struct dentry
*, umode_t
, bool);
1659 extern int vfs_mkdir(struct inode
*, struct dentry
*, umode_t
);
1660 extern int vfs_mknod(struct inode
*, struct dentry
*, umode_t
, dev_t
);
1661 extern int vfs_symlink(struct inode
*, struct dentry
*, const char *);
1662 extern int vfs_link(struct dentry
*, struct inode
*, struct dentry
*, struct inode
**);
1663 extern int vfs_rmdir(struct inode
*, struct dentry
*);
1664 extern int vfs_unlink(struct inode
*, struct dentry
*, struct inode
**);
1665 extern int vfs_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*, struct inode
**, unsigned int);
1666 extern int vfs_whiteout(struct inode
*, struct dentry
*);
1668 extern struct dentry
*vfs_tmpfile(struct dentry
*dentry
, umode_t mode
,
1671 int vfs_mkobj(struct dentry
*, umode_t
,
1672 int (*f
)(struct dentry
*, umode_t
, void *),
1675 extern long vfs_ioctl(struct file
*file
, unsigned int cmd
, unsigned long arg
);
1678 * VFS file helper functions.
1680 extern void inode_init_owner(struct inode
*inode
, const struct inode
*dir
,
1682 extern bool may_open_dev(const struct path
*path
);
1684 * VFS FS_IOC_FIEMAP helper definitions.
1686 struct fiemap_extent_info
{
1687 unsigned int fi_flags
; /* Flags as passed from user */
1688 unsigned int fi_extents_mapped
; /* Number of mapped extents */
1689 unsigned int fi_extents_max
; /* Size of fiemap_extent array */
1690 struct fiemap_extent __user
*fi_extents_start
; /* Start of
1691 fiemap_extent array */
1693 int fiemap_fill_next_extent(struct fiemap_extent_info
*info
, u64 logical
,
1694 u64 phys
, u64 len
, u32 flags
);
1695 int fiemap_check_flags(struct fiemap_extent_info
*fieinfo
, u32 fs_flags
);
1700 * NOTE! These match bits 12..15 of stat.st_mode
1701 * (ie "(i_mode >> 12) & 15").
1703 #define DT_UNKNOWN 0
1714 * This is the "filldir" function type, used by readdir() to let
1715 * the kernel specify what kind of dirent layout it wants to have.
1716 * This allows the kernel to read directories into kernel space or
1717 * to have different dirent layouts depending on the binary type.
1720 typedef int (*filldir_t
)(struct dir_context
*, const char *, int, loff_t
, u64
,
1723 struct dir_context
{
1728 struct block_device_operations
;
1730 /* These macros are for out of kernel modules to test that
1731 * the kernel supports the unlocked_ioctl and compat_ioctl
1732 * fields in struct file_operations. */
1733 #define HAVE_COMPAT_IOCTL 1
1734 #define HAVE_UNLOCKED_IOCTL 1
1737 * These flags let !MMU mmap() govern direct device mapping vs immediate
1738 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1740 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1741 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1742 * NOMMU_MAP_READ: Can be mapped for reading
1743 * NOMMU_MAP_WRITE: Can be mapped for writing
1744 * NOMMU_MAP_EXEC: Can be mapped for execution
1746 #define NOMMU_MAP_COPY 0x00000001
1747 #define NOMMU_MAP_DIRECT 0x00000008
1748 #define NOMMU_MAP_READ VM_MAYREAD
1749 #define NOMMU_MAP_WRITE VM_MAYWRITE
1750 #define NOMMU_MAP_EXEC VM_MAYEXEC
1752 #define NOMMU_VMFLAGS \
1753 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1756 * These flags control the behavior of the remap_file_range function pointer.
1757 * If it is called with len == 0 that means "remap to end of source file".
1758 * See Documentation/filesystems/vfs.txt for more details about this call.
1760 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
1761 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
1763 #define REMAP_FILE_DEDUP (1 << 0)
1764 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
1767 * These flags signal that the caller is ok with altering various aspects of
1768 * the behavior of the remap operation. The changes must be made by the
1769 * implementation; the vfs remap helper functions can take advantage of them.
1770 * Flags in this category exist to preserve the quirky behavior of the hoisted
1771 * btrfs clone/dedupe ioctls.
1773 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
1777 struct file_operations
{
1778 struct module
*owner
;
1779 loff_t (*llseek
) (struct file
*, loff_t
, int);
1780 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
1781 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
1782 ssize_t (*read_iter
) (struct kiocb
*, struct iov_iter
*);
1783 ssize_t (*write_iter
) (struct kiocb
*, struct iov_iter
*);
1784 int (*iterate
) (struct file
*, struct dir_context
*);
1785 int (*iterate_shared
) (struct file
*, struct dir_context
*);
1786 __poll_t (*poll
) (struct file
*, struct poll_table_struct
*);
1787 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
1788 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
1789 int (*mmap
) (struct file
*, struct vm_area_struct
*);
1790 unsigned long mmap_supported_flags
;
1791 int (*open
) (struct inode
*, struct file
*);
1792 int (*flush
) (struct file
*, fl_owner_t id
);
1793 int (*release
) (struct inode
*, struct file
*);
1794 int (*fsync
) (struct file
*, loff_t
, loff_t
, int datasync
);
1795 int (*fasync
) (int, struct file
*, int);
1796 int (*lock
) (struct file
*, int, struct file_lock
*);
1797 ssize_t (*sendpage
) (struct file
*, struct page
*, int, size_t, loff_t
*, int);
1798 unsigned long (*get_unmapped_area
)(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1799 int (*check_flags
)(int);
1800 int (*flock
) (struct file
*, int, struct file_lock
*);
1801 ssize_t (*splice_write
)(struct pipe_inode_info
*, struct file
*, loff_t
*, size_t, unsigned int);
1802 ssize_t (*splice_read
)(struct file
*, loff_t
*, struct pipe_inode_info
*, size_t, unsigned int);
1803 int (*setlease
)(struct file
*, long, struct file_lock
**, void **);
1804 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
1806 void (*show_fdinfo
)(struct seq_file
*m
, struct file
*f
);
1808 unsigned (*mmap_capabilities
)(struct file
*);
1810 ssize_t (*copy_file_range
)(struct file
*, loff_t
, struct file
*,
1811 loff_t
, size_t, unsigned int);
1812 loff_t (*remap_file_range
)(struct file
*file_in
, loff_t pos_in
,
1813 struct file
*file_out
, loff_t pos_out
,
1814 loff_t len
, unsigned int remap_flags
);
1815 int (*fadvise
)(struct file
*, loff_t
, loff_t
, int);
1816 } __randomize_layout
;
1818 struct inode_operations
{
1819 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*, unsigned int);
1820 const char * (*get_link
) (struct dentry
*, struct inode
*, struct delayed_call
*);
1821 int (*permission
) (struct inode
*, int);
1822 struct posix_acl
* (*get_acl
)(struct inode
*, int);
1824 int (*readlink
) (struct dentry
*, char __user
*,int);
1826 int (*create
) (struct inode
*,struct dentry
*, umode_t
, bool);
1827 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
1828 int (*unlink
) (struct inode
*,struct dentry
*);
1829 int (*symlink
) (struct inode
*,struct dentry
*,const char *);
1830 int (*mkdir
) (struct inode
*,struct dentry
*,umode_t
);
1831 int (*rmdir
) (struct inode
*,struct dentry
*);
1832 int (*mknod
) (struct inode
*,struct dentry
*,umode_t
,dev_t
);
1833 int (*rename
) (struct inode
*, struct dentry
*,
1834 struct inode
*, struct dentry
*, unsigned int);
1835 int (*setattr
) (struct dentry
*, struct iattr
*);
1836 int (*getattr
) (const struct path
*, struct kstat
*, u32
, unsigned int);
1837 ssize_t (*listxattr
) (struct dentry
*, char *, size_t);
1838 int (*fiemap
)(struct inode
*, struct fiemap_extent_info
*, u64 start
,
1840 int (*update_time
)(struct inode
*, struct timespec64
*, int);
1841 int (*atomic_open
)(struct inode
*, struct dentry
*,
1842 struct file
*, unsigned open_flag
,
1843 umode_t create_mode
);
1844 int (*tmpfile
) (struct inode
*, struct dentry
*, umode_t
);
1845 int (*set_acl
)(struct inode
*, struct posix_acl
*, int);
1846 } ____cacheline_aligned
;
1848 static inline ssize_t
call_read_iter(struct file
*file
, struct kiocb
*kio
,
1849 struct iov_iter
*iter
)
1851 return file
->f_op
->read_iter(kio
, iter
);
1854 static inline ssize_t
call_write_iter(struct file
*file
, struct kiocb
*kio
,
1855 struct iov_iter
*iter
)
1857 return file
->f_op
->write_iter(kio
, iter
);
1860 static inline int call_mmap(struct file
*file
, struct vm_area_struct
*vma
)
1862 return file
->f_op
->mmap(file
, vma
);
1865 ssize_t
rw_copy_check_uvector(int type
, const struct iovec __user
* uvector
,
1866 unsigned long nr_segs
, unsigned long fast_segs
,
1867 struct iovec
*fast_pointer
,
1868 struct iovec
**ret_pointer
);
1870 extern ssize_t
__vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1871 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1872 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1873 extern ssize_t
vfs_readv(struct file
*, const struct iovec __user
*,
1874 unsigned long, loff_t
*, rwf_t
);
1875 extern ssize_t
vfs_copy_file_range(struct file
*, loff_t
, struct file
*,
1876 loff_t
, size_t, unsigned int);
1877 extern int generic_remap_file_range_prep(struct file
*file_in
, loff_t pos_in
,
1878 struct file
*file_out
, loff_t pos_out
,
1880 unsigned int remap_flags
);
1881 extern loff_t
do_clone_file_range(struct file
*file_in
, loff_t pos_in
,
1882 struct file
*file_out
, loff_t pos_out
,
1883 loff_t len
, unsigned int remap_flags
);
1884 extern loff_t
vfs_clone_file_range(struct file
*file_in
, loff_t pos_in
,
1885 struct file
*file_out
, loff_t pos_out
,
1886 loff_t len
, unsigned int remap_flags
);
1887 extern int vfs_dedupe_file_range(struct file
*file
,
1888 struct file_dedupe_range
*same
);
1889 extern loff_t
vfs_dedupe_file_range_one(struct file
*src_file
, loff_t src_pos
,
1890 struct file
*dst_file
, loff_t dst_pos
,
1891 loff_t len
, unsigned int remap_flags
);
1894 struct super_operations
{
1895 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
1896 void (*destroy_inode
)(struct inode
*);
1898 void (*dirty_inode
) (struct inode
*, int flags
);
1899 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
1900 int (*drop_inode
) (struct inode
*);
1901 void (*evict_inode
) (struct inode
*);
1902 void (*put_super
) (struct super_block
*);
1903 int (*sync_fs
)(struct super_block
*sb
, int wait
);
1904 int (*freeze_super
) (struct super_block
*);
1905 int (*freeze_fs
) (struct super_block
*);
1906 int (*thaw_super
) (struct super_block
*);
1907 int (*unfreeze_fs
) (struct super_block
*);
1908 int (*statfs
) (struct dentry
*, struct kstatfs
*);
1909 int (*remount_fs
) (struct super_block
*, int *, char *);
1910 void (*umount_begin
) (struct super_block
*);
1912 int (*show_options
)(struct seq_file
*, struct dentry
*);
1913 int (*show_devname
)(struct seq_file
*, struct dentry
*);
1914 int (*show_path
)(struct seq_file
*, struct dentry
*);
1915 int (*show_stats
)(struct seq_file
*, struct dentry
*);
1917 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
1918 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
1919 struct dquot
**(*get_dquots
)(struct inode
*);
1921 int (*bdev_try_to_free_page
)(struct super_block
*, struct page
*, gfp_t
);
1922 long (*nr_cached_objects
)(struct super_block
*,
1923 struct shrink_control
*);
1924 long (*free_cached_objects
)(struct super_block
*,
1925 struct shrink_control
*);
1929 * Inode flags - they have no relation to superblock flags now
1931 #define S_SYNC 1 /* Writes are synced at once */
1932 #define S_NOATIME 2 /* Do not update access times */
1933 #define S_APPEND 4 /* Append-only file */
1934 #define S_IMMUTABLE 8 /* Immutable file */
1935 #define S_DEAD 16 /* removed, but still open directory */
1936 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1937 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1938 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1939 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1940 #define S_PRIVATE 512 /* Inode is fs-internal */
1941 #define S_IMA 1024 /* Inode has an associated IMA struct */
1942 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1943 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1944 #ifdef CONFIG_FS_DAX
1945 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1947 #define S_DAX 0 /* Make all the DAX code disappear */
1949 #define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */
1952 * Note that nosuid etc flags are inode-specific: setting some file-system
1953 * flags just means all the inodes inherit those flags by default. It might be
1954 * possible to override it selectively if you really wanted to with some
1955 * ioctl() that is not currently implemented.
1957 * Exception: SB_RDONLY is always applied to the entire file system.
1959 * Unfortunately, it is possible to change a filesystems flags with it mounted
1960 * with files in use. This means that all of the inodes will not have their
1961 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1962 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1964 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1966 static inline bool sb_rdonly(const struct super_block
*sb
) { return sb
->s_flags
& SB_RDONLY
; }
1967 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
1968 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
1969 ((inode)->i_flags & S_SYNC))
1970 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
1971 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1972 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
1973 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
1974 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
1976 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1977 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1978 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1979 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
1981 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1982 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1983 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1984 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1985 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1986 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1987 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1988 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1989 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
1991 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1992 (inode)->i_rdev == WHITEOUT_DEV)
1994 static inline bool HAS_UNMAPPED_ID(struct inode
*inode
)
1996 return !uid_valid(inode
->i_uid
) || !gid_valid(inode
->i_gid
);
1999 static inline enum rw_hint
file_write_hint(struct file
*file
)
2001 if (file
->f_write_hint
!= WRITE_LIFE_NOT_SET
)
2002 return file
->f_write_hint
;
2004 return file_inode(file
)->i_write_hint
;
2007 static inline int iocb_flags(struct file
*file
);
2009 static inline u16
ki_hint_validate(enum rw_hint hint
)
2011 typeof(((struct kiocb
*)0)->ki_hint
) max_hint
= -1;
2013 if (hint
<= max_hint
)
2018 static inline void init_sync_kiocb(struct kiocb
*kiocb
, struct file
*filp
)
2020 *kiocb
= (struct kiocb
) {
2022 .ki_flags
= iocb_flags(filp
),
2023 .ki_hint
= ki_hint_validate(file_write_hint(filp
)),
2024 .ki_ioprio
= IOPRIO_PRIO_VALUE(IOPRIO_CLASS_NONE
, 0),
2029 * Inode state bits. Protected by inode->i_lock
2031 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
2032 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
2034 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2035 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2036 * various stages of removing an inode.
2038 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2040 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2041 * fdatasync(). i_atime is the usual cause.
2042 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2043 * these changes separately from I_DIRTY_SYNC so that we
2044 * don't have to write inode on fdatasync() when only
2045 * mtime has changed in it.
2046 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2047 * I_NEW Serves as both a mutex and completion notification.
2048 * New inodes set I_NEW. If two processes both create
2049 * the same inode, one of them will release its inode and
2050 * wait for I_NEW to be released before returning.
2051 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2052 * also cause waiting on I_NEW, without I_NEW actually
2053 * being set. find_inode() uses this to prevent returning
2054 * nearly-dead inodes.
2055 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2056 * is zero. I_FREEING must be set when I_WILL_FREE is
2058 * I_FREEING Set when inode is about to be freed but still has dirty
2059 * pages or buffers attached or the inode itself is still
2061 * I_CLEAR Added by clear_inode(). In this state the inode is
2062 * clean and can be destroyed. Inode keeps I_FREEING.
2064 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2065 * prohibited for many purposes. iget() must wait for
2066 * the inode to be completely released, then create it
2067 * anew. Other functions will just ignore such inodes,
2068 * if appropriate. I_NEW is used for waiting.
2070 * I_SYNC Writeback of inode is running. The bit is set during
2071 * data writeback, and cleared with a wakeup on the bit
2072 * address once it is done. The bit is also used to pin
2073 * the inode in memory for flusher thread.
2075 * I_REFERENCED Marks the inode as recently references on the LRU list.
2077 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2079 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2080 * synchronize competing switching instances and to tell
2081 * wb stat updates to grab the i_pages lock. See
2082 * inode_switch_wb_work_fn() for details.
2084 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2085 * and work dirs among overlayfs mounts.
2087 * I_CREATING New object's inode in the middle of setting up.
2089 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2091 #define I_DIRTY_SYNC (1 << 0)
2092 #define I_DIRTY_DATASYNC (1 << 1)
2093 #define I_DIRTY_PAGES (1 << 2)
2095 #define I_NEW (1 << __I_NEW)
2096 #define I_WILL_FREE (1 << 4)
2097 #define I_FREEING (1 << 5)
2098 #define I_CLEAR (1 << 6)
2100 #define I_SYNC (1 << __I_SYNC)
2101 #define I_REFERENCED (1 << 8)
2102 #define __I_DIO_WAKEUP 9
2103 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2104 #define I_LINKABLE (1 << 10)
2105 #define I_DIRTY_TIME (1 << 11)
2106 #define __I_DIRTY_TIME_EXPIRED 12
2107 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
2108 #define I_WB_SWITCH (1 << 13)
2109 #define I_OVL_INUSE (1 << 14)
2110 #define I_CREATING (1 << 15)
2112 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2113 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2114 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2116 extern void __mark_inode_dirty(struct inode
*, int);
2117 static inline void mark_inode_dirty(struct inode
*inode
)
2119 __mark_inode_dirty(inode
, I_DIRTY
);
2122 static inline void mark_inode_dirty_sync(struct inode
*inode
)
2124 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
2127 extern void inc_nlink(struct inode
*inode
);
2128 extern void drop_nlink(struct inode
*inode
);
2129 extern void clear_nlink(struct inode
*inode
);
2130 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
2132 static inline void inode_inc_link_count(struct inode
*inode
)
2135 mark_inode_dirty(inode
);
2138 static inline void inode_dec_link_count(struct inode
*inode
)
2141 mark_inode_dirty(inode
);
2144 enum file_time_flags
{
2151 extern bool atime_needs_update(const struct path
*, struct inode
*);
2152 extern void touch_atime(const struct path
*);
2153 static inline void file_accessed(struct file
*file
)
2155 if (!(file
->f_flags
& O_NOATIME
))
2156 touch_atime(&file
->f_path
);
2159 int sync_inode(struct inode
*inode
, struct writeback_control
*wbc
);
2160 int sync_inode_metadata(struct inode
*inode
, int wait
);
2162 struct file_system_type
{
2165 #define FS_REQUIRES_DEV 1
2166 #define FS_BINARY_MOUNTDATA 2
2167 #define FS_HAS_SUBTYPE 4
2168 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2169 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2170 struct dentry
*(*mount
) (struct file_system_type
*, int,
2171 const char *, void *);
2172 void (*kill_sb
) (struct super_block
*);
2173 struct module
*owner
;
2174 struct file_system_type
* next
;
2175 struct hlist_head fs_supers
;
2177 struct lock_class_key s_lock_key
;
2178 struct lock_class_key s_umount_key
;
2179 struct lock_class_key s_vfs_rename_key
;
2180 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
2182 struct lock_class_key i_lock_key
;
2183 struct lock_class_key i_mutex_key
;
2184 struct lock_class_key i_mutex_dir_key
;
2187 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2189 extern struct dentry
*mount_ns(struct file_system_type
*fs_type
,
2190 int flags
, void *data
, void *ns
, struct user_namespace
*user_ns
,
2191 int (*fill_super
)(struct super_block
*, void *, int));
2193 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
2194 int flags
, const char *dev_name
, void *data
,
2195 int (*fill_super
)(struct super_block
*, void *, int));
2197 static inline struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
2198 int flags
, const char *dev_name
, void *data
,
2199 int (*fill_super
)(struct super_block
*, void *, int))
2201 return ERR_PTR(-ENODEV
);
2204 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
2205 int flags
, void *data
,
2206 int (*fill_super
)(struct super_block
*, void *, int));
2207 extern struct dentry
*mount_nodev(struct file_system_type
*fs_type
,
2208 int flags
, void *data
,
2209 int (*fill_super
)(struct super_block
*, void *, int));
2210 extern struct dentry
*mount_subtree(struct vfsmount
*mnt
, const char *path
);
2211 void generic_shutdown_super(struct super_block
*sb
);
2213 void kill_block_super(struct super_block
*sb
);
2215 static inline void kill_block_super(struct super_block
*sb
)
2220 void kill_anon_super(struct super_block
*sb
);
2221 void kill_litter_super(struct super_block
*sb
);
2222 void deactivate_super(struct super_block
*sb
);
2223 void deactivate_locked_super(struct super_block
*sb
);
2224 int set_anon_super(struct super_block
*s
, void *data
);
2225 int get_anon_bdev(dev_t
*);
2226 void free_anon_bdev(dev_t
);
2227 struct super_block
*sget_userns(struct file_system_type
*type
,
2228 int (*test
)(struct super_block
*,void *),
2229 int (*set
)(struct super_block
*,void *),
2230 int flags
, struct user_namespace
*user_ns
,
2232 struct super_block
*sget(struct file_system_type
*type
,
2233 int (*test
)(struct super_block
*,void *),
2234 int (*set
)(struct super_block
*,void *),
2235 int flags
, void *data
);
2236 extern struct dentry
*mount_pseudo_xattr(struct file_system_type
*, char *,
2237 const struct super_operations
*ops
,
2238 const struct xattr_handler
**xattr
,
2239 const struct dentry_operations
*dops
,
2242 static inline struct dentry
*
2243 mount_pseudo(struct file_system_type
*fs_type
, char *name
,
2244 const struct super_operations
*ops
,
2245 const struct dentry_operations
*dops
, unsigned long magic
)
2247 return mount_pseudo_xattr(fs_type
, name
, ops
, NULL
, dops
, magic
);
2250 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2251 #define fops_get(fops) \
2252 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2253 #define fops_put(fops) \
2254 do { if (fops) module_put((fops)->owner); } while(0)
2256 * This one is to be used *ONLY* from ->open() instances.
2257 * fops must be non-NULL, pinned down *and* module dependencies
2258 * should be sufficient to pin the caller down as well.
2260 #define replace_fops(f, fops) \
2262 struct file *__file = (f); \
2263 fops_put(__file->f_op); \
2264 BUG_ON(!(__file->f_op = (fops))); \
2267 extern int register_filesystem(struct file_system_type
*);
2268 extern int unregister_filesystem(struct file_system_type
*);
2269 extern struct vfsmount
*kern_mount_data(struct file_system_type
*, void *data
);
2270 #define kern_mount(type) kern_mount_data(type, NULL)
2271 extern void kern_unmount(struct vfsmount
*mnt
);
2272 extern int may_umount_tree(struct vfsmount
*);
2273 extern int may_umount(struct vfsmount
*);
2274 extern long do_mount(const char *, const char __user
*,
2275 const char *, unsigned long, void *);
2276 extern struct vfsmount
*collect_mounts(const struct path
*);
2277 extern void drop_collected_mounts(struct vfsmount
*);
2278 extern int iterate_mounts(int (*)(struct vfsmount
*, void *), void *,
2280 extern int vfs_statfs(const struct path
*, struct kstatfs
*);
2281 extern int user_statfs(const char __user
*, struct kstatfs
*);
2282 extern int fd_statfs(int, struct kstatfs
*);
2283 extern int freeze_super(struct super_block
*super
);
2284 extern int thaw_super(struct super_block
*super
);
2285 extern bool our_mnt(struct vfsmount
*mnt
);
2286 extern __printf(2, 3)
2287 int super_setup_bdi_name(struct super_block
*sb
, char *fmt
, ...);
2288 extern int super_setup_bdi(struct super_block
*sb
);
2290 extern int current_umask(void);
2292 extern void ihold(struct inode
* inode
);
2293 extern void iput(struct inode
*);
2294 extern int generic_update_time(struct inode
*, struct timespec64
*, int);
2297 extern struct kobject
*fs_kobj
;
2299 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2301 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2302 extern int locks_mandatory_locked(struct file
*);
2303 extern int locks_mandatory_area(struct inode
*, struct file
*, loff_t
, loff_t
, unsigned char);
2306 * Candidates for mandatory locking have the setgid bit set
2307 * but no group execute bit - an otherwise meaningless combination.
2310 static inline int __mandatory_lock(struct inode
*ino
)
2312 return (ino
->i_mode
& (S_ISGID
| S_IXGRP
)) == S_ISGID
;
2316 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2317 * otherwise these will be advisory locks
2320 static inline int mandatory_lock(struct inode
*ino
)
2322 return IS_MANDLOCK(ino
) && __mandatory_lock(ino
);
2325 static inline int locks_verify_locked(struct file
*file
)
2327 if (mandatory_lock(locks_inode(file
)))
2328 return locks_mandatory_locked(file
);
2332 static inline int locks_verify_truncate(struct inode
*inode
,
2336 if (!inode
->i_flctx
|| !mandatory_lock(inode
))
2339 if (size
< inode
->i_size
) {
2340 return locks_mandatory_area(inode
, f
, size
, inode
->i_size
- 1,
2343 return locks_mandatory_area(inode
, f
, inode
->i_size
, size
- 1,
2348 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2350 static inline int locks_mandatory_locked(struct file
*file
)
2355 static inline int locks_mandatory_area(struct inode
*inode
, struct file
*filp
,
2356 loff_t start
, loff_t end
, unsigned char type
)
2361 static inline int __mandatory_lock(struct inode
*inode
)
2366 static inline int mandatory_lock(struct inode
*inode
)
2371 static inline int locks_verify_locked(struct file
*file
)
2376 static inline int locks_verify_truncate(struct inode
*inode
, struct file
*filp
,
2382 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2385 #ifdef CONFIG_FILE_LOCKING
2386 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2389 * Since this check is lockless, we must ensure that any refcounts
2390 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2391 * could end up racing with tasks trying to set a new lease on this
2395 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2396 return __break_lease(inode
, mode
, FL_LEASE
);
2400 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2403 * Since this check is lockless, we must ensure that any refcounts
2404 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2405 * could end up racing with tasks trying to set a new lease on this
2409 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2410 return __break_lease(inode
, mode
, FL_DELEG
);
2414 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2418 ret
= break_deleg(inode
, O_WRONLY
|O_NONBLOCK
);
2419 if (ret
== -EWOULDBLOCK
&& delegated_inode
) {
2420 *delegated_inode
= inode
;
2426 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2430 ret
= break_deleg(*delegated_inode
, O_WRONLY
);
2431 iput(*delegated_inode
);
2432 *delegated_inode
= NULL
;
2436 static inline int break_layout(struct inode
*inode
, bool wait
)
2439 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2440 return __break_lease(inode
,
2441 wait
? O_WRONLY
: O_WRONLY
| O_NONBLOCK
,
2446 #else /* !CONFIG_FILE_LOCKING */
2447 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2452 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2457 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2462 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2468 static inline int break_layout(struct inode
*inode
, bool wait
)
2473 #endif /* CONFIG_FILE_LOCKING */
2478 const char *name
; /* pointer to actual string */
2479 const __user
char *uptr
; /* original userland pointer */
2481 struct audit_names
*aname
;
2485 extern long vfs_truncate(const struct path
*, loff_t
);
2486 extern int do_truncate(struct dentry
*, loff_t start
, unsigned int time_attrs
,
2488 extern int vfs_fallocate(struct file
*file
, int mode
, loff_t offset
,
2490 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
2492 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
2493 extern struct file
*filp_open(const char *, int, umode_t
);
2494 extern struct file
*file_open_root(struct dentry
*, struct vfsmount
*,
2495 const char *, int, umode_t
);
2496 extern struct file
* dentry_open(const struct path
*, int, const struct cred
*);
2497 extern struct file
* open_with_fake_path(const struct path
*, int,
2498 struct inode
*, const struct cred
*);
2499 static inline struct file
*file_clone_open(struct file
*file
)
2501 return dentry_open(&file
->f_path
, file
->f_flags
, file
->f_cred
);
2503 extern int filp_close(struct file
*, fl_owner_t id
);
2505 extern struct filename
*getname_flags(const char __user
*, int, int *);
2506 extern struct filename
*getname(const char __user
*);
2507 extern struct filename
*getname_kernel(const char *);
2508 extern void putname(struct filename
*name
);
2510 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
2511 int (*open
)(struct inode
*, struct file
*));
2512 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
2516 extern int ioctl_preallocate(struct file
*filp
, void __user
*argp
);
2519 extern void __init
vfs_caches_init_early(void);
2520 extern void __init
vfs_caches_init(void);
2522 extern struct kmem_cache
*names_cachep
;
2524 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2525 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2528 extern int register_blkdev(unsigned int, const char *);
2529 extern void unregister_blkdev(unsigned int, const char *);
2530 extern void bdev_unhash_inode(dev_t dev
);
2531 extern struct block_device
*bdget(dev_t
);
2532 extern struct block_device
*bdgrab(struct block_device
*bdev
);
2533 extern void bd_set_size(struct block_device
*, loff_t size
);
2534 extern void bd_forget(struct inode
*inode
);
2535 extern void bdput(struct block_device
*);
2536 extern void invalidate_bdev(struct block_device
*);
2537 extern void iterate_bdevs(void (*)(struct block_device
*, void *), void *);
2538 extern int sync_blockdev(struct block_device
*bdev
);
2539 extern void kill_bdev(struct block_device
*);
2540 extern struct super_block
*freeze_bdev(struct block_device
*);
2541 extern void emergency_thaw_all(void);
2542 extern void emergency_thaw_bdev(struct super_block
*sb
);
2543 extern int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
);
2544 extern int fsync_bdev(struct block_device
*);
2546 extern struct super_block
*blockdev_superblock
;
2548 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2550 return sb
== blockdev_superblock
;
2553 static inline void bd_forget(struct inode
*inode
) {}
2554 static inline int sync_blockdev(struct block_device
*bdev
) { return 0; }
2555 static inline void kill_bdev(struct block_device
*bdev
) {}
2556 static inline void invalidate_bdev(struct block_device
*bdev
) {}
2558 static inline struct super_block
*freeze_bdev(struct block_device
*sb
)
2563 static inline int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
)
2568 static inline int emergency_thaw_bdev(struct super_block
*sb
)
2573 static inline void iterate_bdevs(void (*f
)(struct block_device
*, void *), void *arg
)
2577 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2582 extern int sync_filesystem(struct super_block
*);
2583 extern const struct file_operations def_blk_fops
;
2584 extern const struct file_operations def_chr_fops
;
2586 extern int ioctl_by_bdev(struct block_device
*, unsigned, unsigned long);
2587 extern int blkdev_ioctl(struct block_device
*, fmode_t
, unsigned, unsigned long);
2588 extern long compat_blkdev_ioctl(struct file
*, unsigned, unsigned long);
2589 extern int blkdev_get(struct block_device
*bdev
, fmode_t mode
, void *holder
);
2590 extern struct block_device
*blkdev_get_by_path(const char *path
, fmode_t mode
,
2592 extern struct block_device
*blkdev_get_by_dev(dev_t dev
, fmode_t mode
,
2594 extern void blkdev_put(struct block_device
*bdev
, fmode_t mode
);
2595 extern int __blkdev_reread_part(struct block_device
*bdev
);
2596 extern int blkdev_reread_part(struct block_device
*bdev
);
2599 extern int bd_link_disk_holder(struct block_device
*bdev
, struct gendisk
*disk
);
2600 extern void bd_unlink_disk_holder(struct block_device
*bdev
,
2601 struct gendisk
*disk
);
2603 static inline int bd_link_disk_holder(struct block_device
*bdev
,
2604 struct gendisk
*disk
)
2608 static inline void bd_unlink_disk_holder(struct block_device
*bdev
,
2609 struct gendisk
*disk
)
2616 #define CHRDEV_MAJOR_MAX 512
2617 /* Marks the bottom of the first segment of free char majors */
2618 #define CHRDEV_MAJOR_DYN_END 234
2619 /* Marks the top and bottom of the second segment of free char majors */
2620 #define CHRDEV_MAJOR_DYN_EXT_START 511
2621 #define CHRDEV_MAJOR_DYN_EXT_END 384
2623 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
2624 extern int register_chrdev_region(dev_t
, unsigned, const char *);
2625 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
2626 unsigned int count
, const char *name
,
2627 const struct file_operations
*fops
);
2628 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
2629 unsigned int count
, const char *name
);
2630 extern void unregister_chrdev_region(dev_t
, unsigned);
2631 extern void chrdev_show(struct seq_file
*,off_t
);
2633 static inline int register_chrdev(unsigned int major
, const char *name
,
2634 const struct file_operations
*fops
)
2636 return __register_chrdev(major
, 0, 256, name
, fops
);
2639 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2641 __unregister_chrdev(major
, 0, 256, name
);
2644 /* fs/block_dev.c */
2645 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2646 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2649 #define BLKDEV_MAJOR_MAX 512
2650 extern const char *__bdevname(dev_t
, char *buffer
);
2651 extern const char *bdevname(struct block_device
*bdev
, char *buffer
);
2652 extern struct block_device
*lookup_bdev(const char *);
2653 extern void blkdev_show(struct seq_file
*,off_t
);
2656 #define BLKDEV_MAJOR_MAX 0
2659 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2661 /* Invalid inode operations -- fs/bad_inode.c */
2662 extern void make_bad_inode(struct inode
*);
2663 extern bool is_bad_inode(struct inode
*);
2666 extern void check_disk_size_change(struct gendisk
*disk
,
2667 struct block_device
*bdev
, bool verbose
);
2668 extern int revalidate_disk(struct gendisk
*);
2669 extern int check_disk_change(struct block_device
*);
2670 extern int __invalidate_device(struct block_device
*, bool);
2671 extern int invalidate_partition(struct gendisk
*, int);
2673 unsigned long invalidate_mapping_pages(struct address_space
*mapping
,
2674 pgoff_t start
, pgoff_t end
);
2676 static inline void invalidate_remote_inode(struct inode
*inode
)
2678 if (S_ISREG(inode
->i_mode
) || S_ISDIR(inode
->i_mode
) ||
2679 S_ISLNK(inode
->i_mode
))
2680 invalidate_mapping_pages(inode
->i_mapping
, 0, -1);
2682 extern int invalidate_inode_pages2(struct address_space
*mapping
);
2683 extern int invalidate_inode_pages2_range(struct address_space
*mapping
,
2684 pgoff_t start
, pgoff_t end
);
2685 extern int write_inode_now(struct inode
*, int);
2686 extern int filemap_fdatawrite(struct address_space
*);
2687 extern int filemap_flush(struct address_space
*);
2688 extern int filemap_fdatawait_keep_errors(struct address_space
*mapping
);
2689 extern int filemap_fdatawait_range(struct address_space
*, loff_t lstart
,
2692 static inline int filemap_fdatawait(struct address_space
*mapping
)
2694 return filemap_fdatawait_range(mapping
, 0, LLONG_MAX
);
2697 extern bool filemap_range_has_page(struct address_space
*, loff_t lstart
,
2699 extern int filemap_write_and_wait(struct address_space
*mapping
);
2700 extern int filemap_write_and_wait_range(struct address_space
*mapping
,
2701 loff_t lstart
, loff_t lend
);
2702 extern int __filemap_fdatawrite_range(struct address_space
*mapping
,
2703 loff_t start
, loff_t end
, int sync_mode
);
2704 extern int filemap_fdatawrite_range(struct address_space
*mapping
,
2705 loff_t start
, loff_t end
);
2706 extern int filemap_check_errors(struct address_space
*mapping
);
2707 extern void __filemap_set_wb_err(struct address_space
*mapping
, int err
);
2709 extern int __must_check
file_fdatawait_range(struct file
*file
, loff_t lstart
,
2711 extern int __must_check
file_check_and_advance_wb_err(struct file
*file
);
2712 extern int __must_check
file_write_and_wait_range(struct file
*file
,
2713 loff_t start
, loff_t end
);
2715 static inline int file_write_and_wait(struct file
*file
)
2717 return file_write_and_wait_range(file
, 0, LLONG_MAX
);
2721 * filemap_set_wb_err - set a writeback error on an address_space
2722 * @mapping: mapping in which to set writeback error
2723 * @err: error to be set in mapping
2725 * When writeback fails in some way, we must record that error so that
2726 * userspace can be informed when fsync and the like are called. We endeavor
2727 * to report errors on any file that was open at the time of the error. Some
2728 * internal callers also need to know when writeback errors have occurred.
2730 * When a writeback error occurs, most filesystems will want to call
2731 * filemap_set_wb_err to record the error in the mapping so that it will be
2732 * automatically reported whenever fsync is called on the file.
2734 static inline void filemap_set_wb_err(struct address_space
*mapping
, int err
)
2736 /* Fastpath for common case of no error */
2738 __filemap_set_wb_err(mapping
, err
);
2742 * filemap_check_wb_error - has an error occurred since the mark was sampled?
2743 * @mapping: mapping to check for writeback errors
2744 * @since: previously-sampled errseq_t
2746 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2747 * the given value was sampled.
2749 * If it has then report the latest error set, otherwise return 0.
2751 static inline int filemap_check_wb_err(struct address_space
*mapping
,
2754 return errseq_check(&mapping
->wb_err
, since
);
2758 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2759 * @mapping: mapping to be sampled
2761 * Writeback errors are always reported relative to a particular sample point
2762 * in the past. This function provides those sample points.
2764 static inline errseq_t
filemap_sample_wb_err(struct address_space
*mapping
)
2766 return errseq_sample(&mapping
->wb_err
);
2769 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
2771 extern int vfs_fsync(struct file
*file
, int datasync
);
2774 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2775 * to already be updated for the write, and will return either the amount
2776 * of bytes passed in, or an error if syncing the file failed.
2778 static inline ssize_t
generic_write_sync(struct kiocb
*iocb
, ssize_t count
)
2780 if (iocb
->ki_flags
& IOCB_DSYNC
) {
2781 int ret
= vfs_fsync_range(iocb
->ki_filp
,
2782 iocb
->ki_pos
- count
, iocb
->ki_pos
- 1,
2783 (iocb
->ki_flags
& IOCB_SYNC
) ? 0 : 1);
2791 extern void emergency_sync(void);
2792 extern void emergency_remount(void);
2794 extern sector_t
bmap(struct inode
*, sector_t
);
2796 extern int notify_change(struct dentry
*, struct iattr
*, struct inode
**);
2797 extern int inode_permission(struct inode
*, int);
2798 extern int generic_permission(struct inode
*, int);
2799 extern int __check_sticky(struct inode
*dir
, struct inode
*inode
);
2801 static inline bool execute_ok(struct inode
*inode
)
2803 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
2806 static inline void file_start_write(struct file
*file
)
2808 if (!S_ISREG(file_inode(file
)->i_mode
))
2810 __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, true);
2813 static inline bool file_start_write_trylock(struct file
*file
)
2815 if (!S_ISREG(file_inode(file
)->i_mode
))
2817 return __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, false);
2820 static inline void file_end_write(struct file
*file
)
2822 if (!S_ISREG(file_inode(file
)->i_mode
))
2824 __sb_end_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
);
2828 * get_write_access() gets write permission for a file.
2829 * put_write_access() releases this write permission.
2830 * This is used for regular files.
2831 * We cannot support write (and maybe mmap read-write shared) accesses and
2832 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2833 * can have the following values:
2834 * 0: no writers, no VM_DENYWRITE mappings
2835 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2836 * > 0: (i_writecount) users are writing to the file.
2838 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2839 * except for the cases where we don't hold i_writecount yet. Then we need to
2840 * use {get,deny}_write_access() - these functions check the sign and refuse
2841 * to do the change if sign is wrong.
2843 static inline int get_write_access(struct inode
*inode
)
2845 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2847 static inline int deny_write_access(struct file
*file
)
2849 struct inode
*inode
= file_inode(file
);
2850 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2852 static inline void put_write_access(struct inode
* inode
)
2854 atomic_dec(&inode
->i_writecount
);
2856 static inline void allow_write_access(struct file
*file
)
2859 atomic_inc(&file_inode(file
)->i_writecount
);
2861 static inline bool inode_is_open_for_write(const struct inode
*inode
)
2863 return atomic_read(&inode
->i_writecount
) > 0;
2867 static inline void i_readcount_dec(struct inode
*inode
)
2869 BUG_ON(!atomic_read(&inode
->i_readcount
));
2870 atomic_dec(&inode
->i_readcount
);
2872 static inline void i_readcount_inc(struct inode
*inode
)
2874 atomic_inc(&inode
->i_readcount
);
2877 static inline void i_readcount_dec(struct inode
*inode
)
2881 static inline void i_readcount_inc(struct inode
*inode
)
2886 extern int do_pipe_flags(int *, int);
2888 #define __kernel_read_file_id(id) \
2889 id(UNKNOWN, unknown) \
2890 id(FIRMWARE, firmware) \
2891 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2892 id(MODULE, kernel-module) \
2893 id(KEXEC_IMAGE, kexec-image) \
2894 id(KEXEC_INITRAMFS, kexec-initramfs) \
2895 id(POLICY, security-policy) \
2896 id(X509_CERTIFICATE, x509-certificate) \
2899 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2900 #define __fid_stringify(dummy, str) #str,
2902 enum kernel_read_file_id
{
2903 __kernel_read_file_id(__fid_enumify
)
2906 static const char * const kernel_read_file_str
[] = {
2907 __kernel_read_file_id(__fid_stringify
)
2910 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id
)
2912 if ((unsigned)id
>= READING_MAX_ID
)
2913 return kernel_read_file_str
[READING_UNKNOWN
];
2915 return kernel_read_file_str
[id
];
2918 extern int kernel_read_file(struct file
*, void **, loff_t
*, loff_t
,
2919 enum kernel_read_file_id
);
2920 extern int kernel_read_file_from_path(const char *, void **, loff_t
*, loff_t
,
2921 enum kernel_read_file_id
);
2922 extern int kernel_read_file_from_fd(int, void **, loff_t
*, loff_t
,
2923 enum kernel_read_file_id
);
2924 extern ssize_t
kernel_read(struct file
*, void *, size_t, loff_t
*);
2925 extern ssize_t
kernel_write(struct file
*, const void *, size_t, loff_t
*);
2926 extern ssize_t
__kernel_write(struct file
*, const void *, size_t, loff_t
*);
2927 extern struct file
* open_exec(const char *);
2929 /* fs/dcache.c -- generic fs support functions */
2930 extern bool is_subdir(struct dentry
*, struct dentry
*);
2931 extern bool path_is_under(const struct path
*, const struct path
*);
2933 extern char *file_path(struct file
*, char *, int);
2935 #include <linux/err.h>
2937 /* needed for stackable file system support */
2938 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int whence
);
2940 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int whence
);
2942 extern int inode_init_always(struct super_block
*, struct inode
*);
2943 extern void inode_init_once(struct inode
*);
2944 extern void address_space_init_once(struct address_space
*mapping
);
2945 extern struct inode
* igrab(struct inode
*);
2946 extern ino_t
iunique(struct super_block
*, ino_t
);
2947 extern int inode_needs_sync(struct inode
*inode
);
2948 extern int generic_delete_inode(struct inode
*inode
);
2949 static inline int generic_drop_inode(struct inode
*inode
)
2951 return !inode
->i_nlink
|| inode_unhashed(inode
);
2954 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
2955 unsigned long hashval
, int (*test
)(struct inode
*, void *),
2957 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
2958 int (*test
)(struct inode
*, void *), void *data
);
2959 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
2961 extern struct inode
*inode_insert5(struct inode
*inode
, unsigned long hashval
,
2962 int (*test
)(struct inode
*, void *),
2963 int (*set
)(struct inode
*, void *),
2965 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
2966 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
2967 extern struct inode
*find_inode_nowait(struct super_block
*,
2969 int (*match
)(struct inode
*,
2970 unsigned long, void *),
2972 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
2973 extern int insert_inode_locked(struct inode
*);
2974 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2975 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
2977 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
2979 extern void unlock_new_inode(struct inode
*);
2980 extern void discard_new_inode(struct inode
*);
2981 extern unsigned int get_next_ino(void);
2982 extern void evict_inodes(struct super_block
*sb
);
2984 extern void __iget(struct inode
* inode
);
2985 extern void iget_failed(struct inode
*);
2986 extern void clear_inode(struct inode
*);
2987 extern void __destroy_inode(struct inode
*);
2988 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
2989 extern struct inode
*new_inode(struct super_block
*sb
);
2990 extern void free_inode_nonrcu(struct inode
*inode
);
2991 extern int should_remove_suid(struct dentry
*);
2992 extern int file_remove_privs(struct file
*);
2994 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
2995 static inline void insert_inode_hash(struct inode
*inode
)
2997 __insert_inode_hash(inode
, inode
->i_ino
);
3000 extern void __remove_inode_hash(struct inode
*);
3001 static inline void remove_inode_hash(struct inode
*inode
)
3003 if (!inode_unhashed(inode
) && !hlist_fake(&inode
->i_hash
))
3004 __remove_inode_hash(inode
);
3007 extern void inode_sb_list_add(struct inode
*inode
);
3010 extern int bdev_read_only(struct block_device
*);
3012 extern int set_blocksize(struct block_device
*, int);
3013 extern int sb_set_blocksize(struct super_block
*, int);
3014 extern int sb_min_blocksize(struct super_block
*, int);
3016 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
3017 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
3018 extern ssize_t
generic_write_checks(struct kiocb
*, struct iov_iter
*);
3019 extern int generic_remap_checks(struct file
*file_in
, loff_t pos_in
,
3020 struct file
*file_out
, loff_t pos_out
,
3021 loff_t
*count
, unsigned int remap_flags
);
3022 extern ssize_t
generic_file_read_iter(struct kiocb
*, struct iov_iter
*);
3023 extern ssize_t
__generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
3024 extern ssize_t
generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
3025 extern ssize_t
generic_file_direct_write(struct kiocb
*, struct iov_iter
*);
3026 extern ssize_t
generic_perform_write(struct file
*, struct iov_iter
*, loff_t
);
3028 ssize_t
vfs_iter_read(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
3030 ssize_t
vfs_iter_write(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
3033 /* fs/block_dev.c */
3034 extern ssize_t
blkdev_read_iter(struct kiocb
*iocb
, struct iov_iter
*to
);
3035 extern ssize_t
blkdev_write_iter(struct kiocb
*iocb
, struct iov_iter
*from
);
3036 extern int blkdev_fsync(struct file
*filp
, loff_t start
, loff_t end
,
3038 extern void block_sync_page(struct page
*page
);
3041 extern ssize_t
generic_file_splice_read(struct file
*, loff_t
*,
3042 struct pipe_inode_info
*, size_t, unsigned int);
3043 extern ssize_t
iter_file_splice_write(struct pipe_inode_info
*,
3044 struct file
*, loff_t
*, size_t, unsigned int);
3045 extern ssize_t
generic_splice_sendpage(struct pipe_inode_info
*pipe
,
3046 struct file
*out
, loff_t
*, size_t len
, unsigned int flags
);
3047 extern long do_splice_direct(struct file
*in
, loff_t
*ppos
, struct file
*out
,
3048 loff_t
*opos
, size_t len
, unsigned int flags
);
3052 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
3053 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int whence
);
3054 extern loff_t
no_llseek(struct file
*file
, loff_t offset
, int whence
);
3055 extern loff_t
vfs_setpos(struct file
*file
, loff_t offset
, loff_t maxsize
);
3056 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int whence
);
3057 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
3058 int whence
, loff_t maxsize
, loff_t eof
);
3059 extern loff_t
fixed_size_llseek(struct file
*file
, loff_t offset
,
3060 int whence
, loff_t size
);
3061 extern loff_t
no_seek_end_llseek_size(struct file
*, loff_t
, int, loff_t
);
3062 extern loff_t
no_seek_end_llseek(struct file
*, loff_t
, int);
3063 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
3064 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
3067 typedef void (dio_submit_t
)(struct bio
*bio
, struct inode
*inode
,
3068 loff_t file_offset
);
3071 /* need locking between buffered and direct access */
3074 /* filesystem does not support filling holes */
3075 DIO_SKIP_HOLES
= 0x02,
3078 void dio_end_io(struct bio
*bio
);
3079 void dio_warn_stale_pagecache(struct file
*filp
);
3081 ssize_t
__blockdev_direct_IO(struct kiocb
*iocb
, struct inode
*inode
,
3082 struct block_device
*bdev
, struct iov_iter
*iter
,
3083 get_block_t get_block
,
3084 dio_iodone_t end_io
, dio_submit_t submit_io
,
3087 static inline ssize_t
blockdev_direct_IO(struct kiocb
*iocb
,
3088 struct inode
*inode
,
3089 struct iov_iter
*iter
,
3090 get_block_t get_block
)
3092 return __blockdev_direct_IO(iocb
, inode
, inode
->i_sb
->s_bdev
, iter
,
3093 get_block
, NULL
, NULL
, DIO_LOCKING
| DIO_SKIP_HOLES
);
3097 void inode_dio_wait(struct inode
*inode
);
3100 * inode_dio_begin - signal start of a direct I/O requests
3101 * @inode: inode the direct I/O happens on
3103 * This is called once we've finished processing a direct I/O request,
3104 * and is used to wake up callers waiting for direct I/O to be quiesced.
3106 static inline void inode_dio_begin(struct inode
*inode
)
3108 atomic_inc(&inode
->i_dio_count
);
3112 * inode_dio_end - signal finish of a direct I/O requests
3113 * @inode: inode the direct I/O happens on
3115 * This is called once we've finished processing a direct I/O request,
3116 * and is used to wake up callers waiting for direct I/O to be quiesced.
3118 static inline void inode_dio_end(struct inode
*inode
)
3120 if (atomic_dec_and_test(&inode
->i_dio_count
))
3121 wake_up_bit(&inode
->i_state
, __I_DIO_WAKEUP
);
3124 extern void inode_set_flags(struct inode
*inode
, unsigned int flags
,
3127 extern const struct file_operations generic_ro_fops
;
3129 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3131 extern int readlink_copy(char __user
*, int, const char *);
3132 extern int page_readlink(struct dentry
*, char __user
*, int);
3133 extern const char *page_get_link(struct dentry
*, struct inode
*,
3134 struct delayed_call
*);
3135 extern void page_put_link(void *);
3136 extern int __page_symlink(struct inode
*inode
, const char *symname
, int len
,
3138 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
3139 extern const struct inode_operations page_symlink_inode_operations
;
3140 extern void kfree_link(void *);
3141 extern void generic_fillattr(struct inode
*, struct kstat
*);
3142 extern int vfs_getattr_nosec(const struct path
*, struct kstat
*, u32
, unsigned int);
3143 extern int vfs_getattr(const struct path
*, struct kstat
*, u32
, unsigned int);
3144 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
3145 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
3146 void __inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
3147 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
3148 static inline loff_t
__inode_get_bytes(struct inode
*inode
)
3150 return (((loff_t
)inode
->i_blocks
) << 9) + inode
->i_bytes
;
3152 loff_t
inode_get_bytes(struct inode
*inode
);
3153 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
3154 const char *simple_get_link(struct dentry
*, struct inode
*,
3155 struct delayed_call
*);
3156 extern const struct inode_operations simple_symlink_inode_operations
;
3158 extern int iterate_dir(struct file
*, struct dir_context
*);
3160 extern int vfs_statx(int, const char __user
*, int, struct kstat
*, u32
);
3161 extern int vfs_statx_fd(unsigned int, struct kstat
*, u32
, unsigned int);
3163 static inline int vfs_stat(const char __user
*filename
, struct kstat
*stat
)
3165 return vfs_statx(AT_FDCWD
, filename
, AT_NO_AUTOMOUNT
,
3166 stat
, STATX_BASIC_STATS
);
3168 static inline int vfs_lstat(const char __user
*name
, struct kstat
*stat
)
3170 return vfs_statx(AT_FDCWD
, name
, AT_SYMLINK_NOFOLLOW
| AT_NO_AUTOMOUNT
,
3171 stat
, STATX_BASIC_STATS
);
3173 static inline int vfs_fstatat(int dfd
, const char __user
*filename
,
3174 struct kstat
*stat
, int flags
)
3176 return vfs_statx(dfd
, filename
, flags
| AT_NO_AUTOMOUNT
,
3177 stat
, STATX_BASIC_STATS
);
3179 static inline int vfs_fstat(int fd
, struct kstat
*stat
)
3181 return vfs_statx_fd(fd
, stat
, STATX_BASIC_STATS
, 0);
3185 extern const char *vfs_get_link(struct dentry
*, struct delayed_call
*);
3186 extern int vfs_readlink(struct dentry
*, char __user
*, int);
3188 extern int __generic_block_fiemap(struct inode
*inode
,
3189 struct fiemap_extent_info
*fieinfo
,
3190 loff_t start
, loff_t len
,
3191 get_block_t
*get_block
);
3192 extern int generic_block_fiemap(struct inode
*inode
,
3193 struct fiemap_extent_info
*fieinfo
, u64 start
,
3194 u64 len
, get_block_t
*get_block
);
3196 extern struct file_system_type
*get_filesystem(struct file_system_type
*fs
);
3197 extern void put_filesystem(struct file_system_type
*fs
);
3198 extern struct file_system_type
*get_fs_type(const char *name
);
3199 extern struct super_block
*get_super(struct block_device
*);
3200 extern struct super_block
*get_super_thawed(struct block_device
*);
3201 extern struct super_block
*get_super_exclusive_thawed(struct block_device
*bdev
);
3202 extern struct super_block
*get_active_super(struct block_device
*bdev
);
3203 extern void drop_super(struct super_block
*sb
);
3204 extern void drop_super_exclusive(struct super_block
*sb
);
3205 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
3206 extern void iterate_supers_type(struct file_system_type
*,
3207 void (*)(struct super_block
*, void *), void *);
3209 extern int dcache_dir_open(struct inode
*, struct file
*);
3210 extern int dcache_dir_close(struct inode
*, struct file
*);
3211 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
3212 extern int dcache_readdir(struct file
*, struct dir_context
*);
3213 extern int simple_setattr(struct dentry
*, struct iattr
*);
3214 extern int simple_getattr(const struct path
*, struct kstat
*, u32
, unsigned int);
3215 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
3216 extern int simple_open(struct inode
*inode
, struct file
*file
);
3217 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
3218 extern int simple_unlink(struct inode
*, struct dentry
*);
3219 extern int simple_rmdir(struct inode
*, struct dentry
*);
3220 extern int simple_rename(struct inode
*, struct dentry
*,
3221 struct inode
*, struct dentry
*, unsigned int);
3222 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
3223 extern int noop_set_page_dirty(struct page
*page
);
3224 extern void noop_invalidatepage(struct page
*page
, unsigned int offset
,
3225 unsigned int length
);
3226 extern ssize_t
noop_direct_IO(struct kiocb
*iocb
, struct iov_iter
*iter
);
3227 extern int simple_empty(struct dentry
*);
3228 extern int simple_readpage(struct file
*file
, struct page
*page
);
3229 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
3230 loff_t pos
, unsigned len
, unsigned flags
,
3231 struct page
**pagep
, void **fsdata
);
3232 extern int simple_write_end(struct file
*file
, struct address_space
*mapping
,
3233 loff_t pos
, unsigned len
, unsigned copied
,
3234 struct page
*page
, void *fsdata
);
3235 extern int always_delete_dentry(const struct dentry
*);
3236 extern struct inode
*alloc_anon_inode(struct super_block
*);
3237 extern int simple_nosetlease(struct file
*, long, struct file_lock
**, void **);
3238 extern const struct dentry_operations simple_dentry_operations
;
3240 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
3241 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
3242 extern const struct file_operations simple_dir_operations
;
3243 extern const struct inode_operations simple_dir_inode_operations
;
3244 extern void make_empty_dir_inode(struct inode
*inode
);
3245 extern bool is_empty_dir_inode(struct inode
*inode
);
3246 struct tree_descr
{ const char *name
; const struct file_operations
*ops
; int mode
; };
3247 struct dentry
*d_alloc_name(struct dentry
*, const char *);
3248 extern int simple_fill_super(struct super_block
*, unsigned long,
3249 const struct tree_descr
*);
3250 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
3251 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
3253 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
3254 loff_t
*ppos
, const void *from
, size_t available
);
3255 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
3256 const void __user
*from
, size_t count
);
3258 extern int __generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3259 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3261 extern int generic_check_addressable(unsigned, u64
);
3263 #ifdef CONFIG_MIGRATION
3264 extern int buffer_migrate_page(struct address_space
*,
3265 struct page
*, struct page
*,
3268 #define buffer_migrate_page NULL
3271 extern int setattr_prepare(struct dentry
*, struct iattr
*);
3272 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
3273 extern void setattr_copy(struct inode
*inode
, const struct iattr
*attr
);
3275 extern int file_update_time(struct file
*file
);
3277 static inline bool io_is_direct(struct file
*filp
)
3279 return (filp
->f_flags
& O_DIRECT
) || IS_DAX(filp
->f_mapping
->host
);
3282 static inline bool vma_is_dax(struct vm_area_struct
*vma
)
3284 return vma
->vm_file
&& IS_DAX(vma
->vm_file
->f_mapping
->host
);
3287 static inline bool vma_is_fsdax(struct vm_area_struct
*vma
)
3289 struct inode
*inode
;
3293 if (!vma_is_dax(vma
))
3295 inode
= file_inode(vma
->vm_file
);
3296 if (S_ISCHR(inode
->i_mode
))
3297 return false; /* device-dax */
3301 static inline int iocb_flags(struct file
*file
)
3304 if (file
->f_flags
& O_APPEND
)
3306 if (io_is_direct(file
))
3308 if ((file
->f_flags
& O_DSYNC
) || IS_SYNC(file
->f_mapping
->host
))
3310 if (file
->f_flags
& __O_SYNC
)
3315 static inline int kiocb_set_rw_flags(struct kiocb
*ki
, rwf_t flags
)
3317 if (unlikely(flags
& ~RWF_SUPPORTED
))
3320 if (flags
& RWF_NOWAIT
) {
3321 if (!(ki
->ki_filp
->f_mode
& FMODE_NOWAIT
))
3323 ki
->ki_flags
|= IOCB_NOWAIT
;
3325 if (flags
& RWF_HIPRI
)
3326 ki
->ki_flags
|= IOCB_HIPRI
;
3327 if (flags
& RWF_DSYNC
)
3328 ki
->ki_flags
|= IOCB_DSYNC
;
3329 if (flags
& RWF_SYNC
)
3330 ki
->ki_flags
|= (IOCB_DSYNC
| IOCB_SYNC
);
3331 if (flags
& RWF_APPEND
)
3332 ki
->ki_flags
|= IOCB_APPEND
;
3336 static inline ino_t
parent_ino(struct dentry
*dentry
)
3341 * Don't strictly need d_lock here? If the parent ino could change
3342 * then surely we'd have a deeper race in the caller?
3344 spin_lock(&dentry
->d_lock
);
3345 res
= dentry
->d_parent
->d_inode
->i_ino
;
3346 spin_unlock(&dentry
->d_lock
);
3350 /* Transaction based IO helpers */
3353 * An argresp is stored in an allocated page and holds the
3354 * size of the argument or response, along with its content
3356 struct simple_transaction_argresp
{
3361 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3363 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
3365 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
3366 size_t size
, loff_t
*pos
);
3367 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
3369 void simple_transaction_set(struct file
*file
, size_t n
);
3372 * simple attribute files
3374 * These attributes behave similar to those in sysfs:
3376 * Writing to an attribute immediately sets a value, an open file can be
3377 * written to multiple times.
3379 * Reading from an attribute creates a buffer from the value that might get
3380 * read with multiple read calls. When the attribute has been read
3381 * completely, no further read calls are possible until the file is opened
3384 * All attributes contain a text representation of a numeric value
3385 * that are accessed with the get() and set() functions.
3387 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3388 static int __fops ## _open(struct inode *inode, struct file *file) \
3390 __simple_attr_check_format(__fmt, 0ull); \
3391 return simple_attr_open(inode, file, __get, __set, __fmt); \
3393 static const struct file_operations __fops = { \
3394 .owner = THIS_MODULE, \
3395 .open = __fops ## _open, \
3396 .release = simple_attr_release, \
3397 .read = simple_attr_read, \
3398 .write = simple_attr_write, \
3399 .llseek = generic_file_llseek, \
3402 static inline __printf(1, 2)
3403 void __simple_attr_check_format(const char *fmt
, ...)
3405 /* don't do anything, just let the compiler check the arguments; */
3408 int simple_attr_open(struct inode
*inode
, struct file
*file
,
3409 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
3411 int simple_attr_release(struct inode
*inode
, struct file
*file
);
3412 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
3413 size_t len
, loff_t
*ppos
);
3414 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
3415 size_t len
, loff_t
*ppos
);
3418 int proc_nr_files(struct ctl_table
*table
, int write
,
3419 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3420 int proc_nr_dentry(struct ctl_table
*table
, int write
,
3421 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3422 int proc_nr_inodes(struct ctl_table
*table
, int write
,
3423 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3424 int __init
get_filesystem_list(char *buf
);
3426 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3427 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3429 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3430 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3431 (flag & __FMODE_NONOTIFY)))
3433 static inline bool is_sxid(umode_t mode
)
3435 return (mode
& S_ISUID
) || ((mode
& S_ISGID
) && (mode
& S_IXGRP
));
3438 static inline int check_sticky(struct inode
*dir
, struct inode
*inode
)
3440 if (!(dir
->i_mode
& S_ISVTX
))
3443 return __check_sticky(dir
, inode
);
3446 static inline void inode_has_no_xattr(struct inode
*inode
)
3448 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& SB_NOSEC
))
3449 inode
->i_flags
|= S_NOSEC
;
3452 static inline bool is_root_inode(struct inode
*inode
)
3454 return inode
== inode
->i_sb
->s_root
->d_inode
;
3457 static inline bool dir_emit(struct dir_context
*ctx
,
3458 const char *name
, int namelen
,
3459 u64 ino
, unsigned type
)
3461 return ctx
->actor(ctx
, name
, namelen
, ctx
->pos
, ino
, type
) == 0;
3463 static inline bool dir_emit_dot(struct file
*file
, struct dir_context
*ctx
)
3465 return ctx
->actor(ctx
, ".", 1, ctx
->pos
,
3466 file
->f_path
.dentry
->d_inode
->i_ino
, DT_DIR
) == 0;
3468 static inline bool dir_emit_dotdot(struct file
*file
, struct dir_context
*ctx
)
3470 return ctx
->actor(ctx
, "..", 2, ctx
->pos
,
3471 parent_ino(file
->f_path
.dentry
), DT_DIR
) == 0;
3473 static inline bool dir_emit_dots(struct file
*file
, struct dir_context
*ctx
)
3475 if (ctx
->pos
== 0) {
3476 if (!dir_emit_dot(file
, ctx
))
3480 if (ctx
->pos
== 1) {
3481 if (!dir_emit_dotdot(file
, ctx
))
3487 static inline bool dir_relax(struct inode
*inode
)
3489 inode_unlock(inode
);
3491 return !IS_DEADDIR(inode
);
3494 static inline bool dir_relax_shared(struct inode
*inode
)
3496 inode_unlock_shared(inode
);
3497 inode_lock_shared(inode
);
3498 return !IS_DEADDIR(inode
);
3501 extern bool path_noexec(const struct path
*path
);
3502 extern void inode_nohighmem(struct inode
*inode
);
3505 extern int vfs_fadvise(struct file
*file
, loff_t offset
, loff_t len
,
3508 #endif /* _LINUX_FS_H */