1 /* SPDX-License-Identifier: GPL-2.0 */
5 #include <linux/linkage.h>
6 #include <linux/wait_bit.h>
7 #include <linux/kdev_t.h>
8 #include <linux/dcache.h>
9 #include <linux/path.h>
10 #include <linux/stat.h>
11 #include <linux/cache.h>
12 #include <linux/list.h>
13 #include <linux/list_lru.h>
14 #include <linux/llist.h>
15 #include <linux/radix-tree.h>
16 #include <linux/xarray.h>
17 #include <linux/rbtree.h>
18 #include <linux/init.h>
19 #include <linux/pid.h>
20 #include <linux/bug.h>
21 #include <linux/mutex.h>
22 #include <linux/rwsem.h>
23 #include <linux/mm_types.h>
24 #include <linux/capability.h>
25 #include <linux/semaphore.h>
26 #include <linux/fcntl.h>
27 #include <linux/fiemap.h>
28 #include <linux/rculist_bl.h>
29 #include <linux/atomic.h>
30 #include <linux/shrinker.h>
31 #include <linux/migrate_mode.h>
32 #include <linux/uidgid.h>
33 #include <linux/lockdep.h>
34 #include <linux/percpu-rwsem.h>
35 #include <linux/workqueue.h>
36 #include <linux/delayed_call.h>
37 #include <linux/uuid.h>
38 #include <linux/errseq.h>
39 #include <linux/ioprio.h>
40 #include <linux/fs_types.h>
41 #include <linux/build_bug.h>
42 #include <linux/stddef.h>
44 #include <asm/byteorder.h>
45 #include <uapi/linux/fs.h>
47 struct backing_dev_info
;
50 struct export_operations
;
55 struct pipe_inode_info
;
56 struct poll_table_struct
;
58 struct vm_area_struct
;
61 struct swap_info_struct
;
63 struct workqueue_struct
;
66 struct fscrypt_operations
;
68 struct fs_parameter_description
;
70 extern void __init
inode_init(void);
71 extern void __init
inode_init_early(void);
72 extern void __init
files_init(void);
73 extern void __init
files_maxfiles_init(void);
75 extern struct files_stat_struct files_stat
;
76 extern unsigned long get_max_files(void);
77 extern unsigned int sysctl_nr_open
;
78 extern struct inodes_stat_t inodes_stat
;
79 extern int leases_enable
, lease_break_time
;
80 extern int sysctl_protected_symlinks
;
81 extern int sysctl_protected_hardlinks
;
82 extern int sysctl_protected_fifos
;
83 extern int sysctl_protected_regular
;
85 typedef __kernel_rwf_t rwf_t
;
88 typedef int (get_block_t
)(struct inode
*inode
, sector_t iblock
,
89 struct buffer_head
*bh_result
, int create
);
90 typedef int (dio_iodone_t
)(struct kiocb
*iocb
, loff_t offset
,
91 ssize_t bytes
, void *private);
93 #define MAY_EXEC 0x00000001
94 #define MAY_WRITE 0x00000002
95 #define MAY_READ 0x00000004
96 #define MAY_APPEND 0x00000008
97 #define MAY_ACCESS 0x00000010
98 #define MAY_OPEN 0x00000020
99 #define MAY_CHDIR 0x00000040
100 /* called from RCU mode, don't block */
101 #define MAY_NOT_BLOCK 0x00000080
104 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
105 * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
108 /* file is open for reading */
109 #define FMODE_READ ((__force fmode_t)0x1)
110 /* file is open for writing */
111 #define FMODE_WRITE ((__force fmode_t)0x2)
112 /* file is seekable */
113 #define FMODE_LSEEK ((__force fmode_t)0x4)
114 /* file can be accessed using pread */
115 #define FMODE_PREAD ((__force fmode_t)0x8)
116 /* file can be accessed using pwrite */
117 #define FMODE_PWRITE ((__force fmode_t)0x10)
118 /* File is opened for execution with sys_execve / sys_uselib */
119 #define FMODE_EXEC ((__force fmode_t)0x20)
120 /* File is opened with O_NDELAY (only set for block devices) */
121 #define FMODE_NDELAY ((__force fmode_t)0x40)
122 /* File is opened with O_EXCL (only set for block devices) */
123 #define FMODE_EXCL ((__force fmode_t)0x80)
124 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
125 (specialy hack for floppy.c) */
126 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
127 /* 32bit hashes as llseek() offset (for directories) */
128 #define FMODE_32BITHASH ((__force fmode_t)0x200)
129 /* 64bit hashes as llseek() offset (for directories) */
130 #define FMODE_64BITHASH ((__force fmode_t)0x400)
133 * Don't update ctime and mtime.
135 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
136 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
138 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
140 /* Expect random access pattern */
141 #define FMODE_RANDOM ((__force fmode_t)0x1000)
143 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
144 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
146 /* File is opened with O_PATH; almost nothing can be done with it */
147 #define FMODE_PATH ((__force fmode_t)0x4000)
149 /* File needs atomic accesses to f_pos */
150 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
151 /* Write access to underlying fs */
152 #define FMODE_WRITER ((__force fmode_t)0x10000)
153 /* Has read method(s) */
154 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
155 /* Has write method(s) */
156 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
158 #define FMODE_OPENED ((__force fmode_t)0x80000)
159 #define FMODE_CREATED ((__force fmode_t)0x100000)
161 /* File is stream-like */
162 #define FMODE_STREAM ((__force fmode_t)0x200000)
164 /* File was opened by fanotify and shouldn't generate fanotify events */
165 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
167 /* File is capable of returning -EAGAIN if I/O will block */
168 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
170 /* File represents mount that needs unmounting */
171 #define FMODE_NEED_UNMOUNT ((__force fmode_t)0x10000000)
173 /* File does not contribute to nr_files count */
174 #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000)
177 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
178 * that indicates that they should check the contents of the iovec are
179 * valid, but not check the memory that the iovec elements
182 #define CHECK_IOVEC_ONLY -1
185 * Attribute flags. These should be or-ed together to figure out what
188 #define ATTR_MODE (1 << 0)
189 #define ATTR_UID (1 << 1)
190 #define ATTR_GID (1 << 2)
191 #define ATTR_SIZE (1 << 3)
192 #define ATTR_ATIME (1 << 4)
193 #define ATTR_MTIME (1 << 5)
194 #define ATTR_CTIME (1 << 6)
195 #define ATTR_ATIME_SET (1 << 7)
196 #define ATTR_MTIME_SET (1 << 8)
197 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
198 #define ATTR_KILL_SUID (1 << 11)
199 #define ATTR_KILL_SGID (1 << 12)
200 #define ATTR_FILE (1 << 13)
201 #define ATTR_KILL_PRIV (1 << 14)
202 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
203 #define ATTR_TIMES_SET (1 << 16)
204 #define ATTR_TOUCH (1 << 17)
207 * Whiteout is represented by a char device. The following constants define the
208 * mode and device number to use.
210 #define WHITEOUT_MODE 0
211 #define WHITEOUT_DEV 0
214 * This is the Inode Attributes structure, used for notify_change(). It
215 * uses the above definitions as flags, to know which values have changed.
216 * Also, in this manner, a Filesystem can look at only the values it cares
217 * about. Basically, these are the attributes that the VFS layer can
218 * request to change from the FS layer.
220 * Derek Atkins <warlord@MIT.EDU> 94-10-20
223 unsigned int ia_valid
;
228 struct timespec64 ia_atime
;
229 struct timespec64 ia_mtime
;
230 struct timespec64 ia_ctime
;
233 * Not an attribute, but an auxiliary info for filesystems wanting to
234 * implement an ftruncate() like method. NOTE: filesystem should
235 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
237 struct file
*ia_file
;
241 * Includes for diskquotas.
243 #include <linux/quota.h>
246 * Maximum number of layers of fs stack. Needs to be limited to
247 * prevent kernel stack overflow
249 #define FILESYSTEM_MAX_STACK_DEPTH 2
252 * enum positive_aop_returns - aop return codes with specific semantics
254 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
255 * completed, that the page is still locked, and
256 * should be considered active. The VM uses this hint
257 * to return the page to the active list -- it won't
258 * be a candidate for writeback again in the near
259 * future. Other callers must be careful to unlock
260 * the page if they get this return. Returned by
263 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
264 * unlocked it and the page might have been truncated.
265 * The caller should back up to acquiring a new page and
266 * trying again. The aop will be taking reasonable
267 * precautions not to livelock. If the caller held a page
268 * reference, it should drop it before retrying. Returned
271 * address_space_operation functions return these large constants to indicate
272 * special semantics to the caller. These are much larger than the bytes in a
273 * page to allow for functions that return the number of bytes operated on in a
277 enum positive_aop_returns
{
278 AOP_WRITEPAGE_ACTIVATE
= 0x80000,
279 AOP_TRUNCATED_PAGE
= 0x80001,
282 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
283 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
284 * helper code (eg buffer layer)
285 * to clear GFP_FS from alloc */
288 * oh the beauties of C type declarations.
291 struct address_space
;
292 struct writeback_control
;
295 * Write life time hint values.
296 * Stored in struct inode as u8.
299 WRITE_LIFE_NOT_SET
= 0,
300 WRITE_LIFE_NONE
= RWH_WRITE_LIFE_NONE
,
301 WRITE_LIFE_SHORT
= RWH_WRITE_LIFE_SHORT
,
302 WRITE_LIFE_MEDIUM
= RWH_WRITE_LIFE_MEDIUM
,
303 WRITE_LIFE_LONG
= RWH_WRITE_LIFE_LONG
,
304 WRITE_LIFE_EXTREME
= RWH_WRITE_LIFE_EXTREME
,
307 #define IOCB_EVENTFD (1 << 0)
308 #define IOCB_APPEND (1 << 1)
309 #define IOCB_DIRECT (1 << 2)
310 #define IOCB_HIPRI (1 << 3)
311 #define IOCB_DSYNC (1 << 4)
312 #define IOCB_SYNC (1 << 5)
313 #define IOCB_WRITE (1 << 6)
314 #define IOCB_NOWAIT (1 << 7)
317 struct file
*ki_filp
;
319 /* The 'ki_filp' pointer is shared in a union for aio */
320 randomized_struct_fields_start
323 void (*ki_complete
)(struct kiocb
*iocb
, long ret
, long ret2
);
327 u16 ki_ioprio
; /* See linux/ioprio.h */
328 unsigned int ki_cookie
; /* for ->iopoll */
330 randomized_struct_fields_end
333 static inline bool is_sync_kiocb(struct kiocb
*kiocb
)
335 return kiocb
->ki_complete
== NULL
;
339 * "descriptor" for what we're up to with a read.
340 * This allows us to use the same read code yet
341 * have multiple different users of the data that
342 * we read from a file.
344 * The simplest case just copies the data to user
357 typedef int (*read_actor_t
)(read_descriptor_t
*, struct page
*,
358 unsigned long, unsigned long);
360 struct address_space_operations
{
361 int (*writepage
)(struct page
*page
, struct writeback_control
*wbc
);
362 int (*readpage
)(struct file
*, struct page
*);
364 /* Write back some dirty pages from this mapping. */
365 int (*writepages
)(struct address_space
*, struct writeback_control
*);
367 /* Set a page dirty. Return true if this dirtied it */
368 int (*set_page_dirty
)(struct page
*page
);
371 * Reads in the requested pages. Unlike ->readpage(), this is
372 * PURELY used for read-ahead!.
374 int (*readpages
)(struct file
*filp
, struct address_space
*mapping
,
375 struct list_head
*pages
, unsigned nr_pages
);
377 int (*write_begin
)(struct file
*, struct address_space
*mapping
,
378 loff_t pos
, unsigned len
, unsigned flags
,
379 struct page
**pagep
, void **fsdata
);
380 int (*write_end
)(struct file
*, struct address_space
*mapping
,
381 loff_t pos
, unsigned len
, unsigned copied
,
382 struct page
*page
, void *fsdata
);
384 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
385 sector_t (*bmap
)(struct address_space
*, sector_t
);
386 void (*invalidatepage
) (struct page
*, unsigned int, unsigned int);
387 int (*releasepage
) (struct page
*, gfp_t
);
388 void (*freepage
)(struct page
*);
389 ssize_t (*direct_IO
)(struct kiocb
*, struct iov_iter
*iter
);
391 * migrate the contents of a page to the specified target. If
392 * migrate_mode is MIGRATE_ASYNC, it must not block.
394 int (*migratepage
) (struct address_space
*,
395 struct page
*, struct page
*, enum migrate_mode
);
396 bool (*isolate_page
)(struct page
*, isolate_mode_t
);
397 void (*putback_page
)(struct page
*);
398 int (*launder_page
) (struct page
*);
399 int (*is_partially_uptodate
) (struct page
*, unsigned long,
401 void (*is_dirty_writeback
) (struct page
*, bool *, bool *);
402 int (*error_remove_page
)(struct address_space
*, struct page
*);
404 /* swapfile support */
405 int (*swap_activate
)(struct swap_info_struct
*sis
, struct file
*file
,
407 void (*swap_deactivate
)(struct file
*file
);
410 extern const struct address_space_operations empty_aops
;
413 * pagecache_write_begin/pagecache_write_end must be used by general code
414 * to write into the pagecache.
416 int pagecache_write_begin(struct file
*, struct address_space
*mapping
,
417 loff_t pos
, unsigned len
, unsigned flags
,
418 struct page
**pagep
, void **fsdata
);
420 int pagecache_write_end(struct file
*, struct address_space
*mapping
,
421 loff_t pos
, unsigned len
, unsigned copied
,
422 struct page
*page
, void *fsdata
);
425 * struct address_space - Contents of a cacheable, mappable object.
426 * @host: Owner, either the inode or the block_device.
427 * @i_pages: Cached pages.
428 * @gfp_mask: Memory allocation flags to use for allocating pages.
429 * @i_mmap_writable: Number of VM_SHARED mappings.
430 * @i_mmap: Tree of private and shared mappings.
431 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
432 * @nrpages: Number of page entries, protected by the i_pages lock.
433 * @nrexceptional: Shadow or DAX entries, protected by the i_pages lock.
434 * @writeback_index: Writeback starts here.
436 * @flags: Error bits and flags (AS_*).
437 * @wb_err: The most recent error which has occurred.
438 * @private_lock: For use by the owner of the address_space.
439 * @private_list: For use by the owner of the address_space.
440 * @private_data: For use by the owner of the address_space.
442 struct address_space
{
444 struct xarray i_pages
;
446 atomic_t i_mmap_writable
;
447 struct rb_root_cached i_mmap
;
448 struct rw_semaphore i_mmap_rwsem
;
449 unsigned long nrpages
;
450 unsigned long nrexceptional
;
451 pgoff_t writeback_index
;
452 const struct address_space_operations
*a_ops
;
455 spinlock_t private_lock
;
456 struct list_head private_list
;
458 } __attribute__((aligned(sizeof(long)))) __randomize_layout
;
460 * On most architectures that alignment is already the case; but
461 * must be enforced here for CRIS, to let the least significant bit
462 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
464 struct request_queue
;
466 struct block_device
{
467 dev_t bd_dev
; /* not a kdev_t - it's a search key */
469 struct inode
* bd_inode
; /* will die */
470 struct super_block
* bd_super
;
471 struct mutex bd_mutex
; /* open/close mutex */
475 bool bd_write_holder
;
477 struct list_head bd_holder_disks
;
479 struct block_device
* bd_contains
;
480 unsigned bd_block_size
;
482 struct hd_struct
* bd_part
;
483 /* number of times partitions within this device have been opened. */
484 unsigned bd_part_count
;
486 struct gendisk
* bd_disk
;
487 struct request_queue
* bd_queue
;
488 struct backing_dev_info
*bd_bdi
;
489 struct list_head bd_list
;
491 * Private data. You must have bd_claim'ed the block_device
492 * to use this. NOTE: bd_claim allows an owner to claim
493 * the same device multiple times, the owner must take special
494 * care to not mess up bd_private for that case.
496 unsigned long bd_private
;
498 /* The counter of freeze processes */
499 int bd_fsfreeze_count
;
500 /* Mutex for freeze */
501 struct mutex bd_fsfreeze_mutex
;
502 } __randomize_layout
;
504 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
505 #define PAGECACHE_TAG_DIRTY XA_MARK_0
506 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
507 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
510 * Returns true if any of the pages in the mapping are marked with the tag.
512 static inline bool mapping_tagged(struct address_space
*mapping
, xa_mark_t tag
)
514 return xa_marked(&mapping
->i_pages
, tag
);
517 static inline void i_mmap_lock_write(struct address_space
*mapping
)
519 down_write(&mapping
->i_mmap_rwsem
);
522 static inline void i_mmap_unlock_write(struct address_space
*mapping
)
524 up_write(&mapping
->i_mmap_rwsem
);
527 static inline void i_mmap_lock_read(struct address_space
*mapping
)
529 down_read(&mapping
->i_mmap_rwsem
);
532 static inline void i_mmap_unlock_read(struct address_space
*mapping
)
534 up_read(&mapping
->i_mmap_rwsem
);
538 * Might pages of this file be mapped into userspace?
540 static inline int mapping_mapped(struct address_space
*mapping
)
542 return !RB_EMPTY_ROOT(&mapping
->i_mmap
.rb_root
);
546 * Might pages of this file have been modified in userspace?
547 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
548 * marks vma as VM_SHARED if it is shared, and the file was opened for
549 * writing i.e. vma may be mprotected writable even if now readonly.
551 * If i_mmap_writable is negative, no new writable mappings are allowed. You
552 * can only deny writable mappings, if none exists right now.
554 static inline int mapping_writably_mapped(struct address_space
*mapping
)
556 return atomic_read(&mapping
->i_mmap_writable
) > 0;
559 static inline int mapping_map_writable(struct address_space
*mapping
)
561 return atomic_inc_unless_negative(&mapping
->i_mmap_writable
) ?
565 static inline void mapping_unmap_writable(struct address_space
*mapping
)
567 atomic_dec(&mapping
->i_mmap_writable
);
570 static inline int mapping_deny_writable(struct address_space
*mapping
)
572 return atomic_dec_unless_positive(&mapping
->i_mmap_writable
) ?
576 static inline void mapping_allow_writable(struct address_space
*mapping
)
578 atomic_inc(&mapping
->i_mmap_writable
);
582 * Use sequence counter to get consistent i_size on 32-bit processors.
584 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
585 #include <linux/seqlock.h>
586 #define __NEED_I_SIZE_ORDERED
587 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
589 #define i_size_ordered_init(inode) do { } while (0)
593 #define ACL_NOT_CACHED ((void *)(-1))
594 #define ACL_DONT_CACHE ((void *)(-3))
596 static inline struct posix_acl
*
597 uncached_acl_sentinel(struct task_struct
*task
)
599 return (void *)task
+ 1;
603 is_uncached_acl(struct posix_acl
*acl
)
605 return (long)acl
& 1;
608 #define IOP_FASTPERM 0x0001
609 #define IOP_LOOKUP 0x0002
610 #define IOP_NOFOLLOW 0x0004
611 #define IOP_XATTR 0x0008
612 #define IOP_DEFAULT_READLINK 0x0010
614 struct fsnotify_mark_connector
;
617 * Keep mostly read-only and often accessed (especially for
618 * the RCU path lookup and 'stat' data) fields at the beginning
619 * of the 'struct inode'
623 unsigned short i_opflags
;
626 unsigned int i_flags
;
628 #ifdef CONFIG_FS_POSIX_ACL
629 struct posix_acl
*i_acl
;
630 struct posix_acl
*i_default_acl
;
633 const struct inode_operations
*i_op
;
634 struct super_block
*i_sb
;
635 struct address_space
*i_mapping
;
637 #ifdef CONFIG_SECURITY
641 /* Stat data, not accessed from path walking */
644 * Filesystems may only read i_nlink directly. They shall use the
645 * following functions for modification:
647 * (set|clear|inc|drop)_nlink
648 * inode_(inc|dec)_link_count
651 const unsigned int i_nlink
;
652 unsigned int __i_nlink
;
656 struct timespec64 i_atime
;
657 struct timespec64 i_mtime
;
658 struct timespec64 i_ctime
;
659 spinlock_t i_lock
; /* i_blocks, i_bytes, maybe i_size */
660 unsigned short i_bytes
;
665 #ifdef __NEED_I_SIZE_ORDERED
666 seqcount_t i_size_seqcount
;
670 unsigned long i_state
;
671 struct rw_semaphore i_rwsem
;
673 unsigned long dirtied_when
; /* jiffies of first dirtying */
674 unsigned long dirtied_time_when
;
676 struct hlist_node i_hash
;
677 struct list_head i_io_list
; /* backing dev IO list */
678 #ifdef CONFIG_CGROUP_WRITEBACK
679 struct bdi_writeback
*i_wb
; /* the associated cgroup wb */
681 /* foreign inode detection, see wbc_detach_inode() */
683 u16 i_wb_frn_avg_time
;
684 u16 i_wb_frn_history
;
686 struct list_head i_lru
; /* inode LRU list */
687 struct list_head i_sb_list
;
688 struct list_head i_wb_list
; /* backing dev writeback list */
690 struct hlist_head i_dentry
;
691 struct rcu_head i_rcu
;
693 atomic64_t i_version
;
695 atomic_t i_dio_count
;
696 atomic_t i_writecount
;
697 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
698 atomic_t i_readcount
; /* struct files open RO */
701 const struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
702 void (*free_inode
)(struct inode
*);
704 struct file_lock_context
*i_flctx
;
705 struct address_space i_data
;
706 struct list_head i_devices
;
708 struct pipe_inode_info
*i_pipe
;
709 struct block_device
*i_bdev
;
717 #ifdef CONFIG_FSNOTIFY
718 __u32 i_fsnotify_mask
; /* all events this inode cares about */
719 struct fsnotify_mark_connector __rcu
*i_fsnotify_marks
;
722 #ifdef CONFIG_FS_ENCRYPTION
723 struct fscrypt_info
*i_crypt_info
;
726 void *i_private
; /* fs or device private pointer */
727 } __randomize_layout
;
729 static inline unsigned int i_blocksize(const struct inode
*node
)
731 return (1 << node
->i_blkbits
);
734 static inline int inode_unhashed(struct inode
*inode
)
736 return hlist_unhashed(&inode
->i_hash
);
740 * __mark_inode_dirty expects inodes to be hashed. Since we don't
741 * want special inodes in the fileset inode space, we make them
742 * appear hashed, but do not put on any lists. hlist_del()
743 * will work fine and require no locking.
745 static inline void inode_fake_hash(struct inode
*inode
)
747 hlist_add_fake(&inode
->i_hash
);
751 * inode->i_mutex nesting subclasses for the lock validator:
753 * 0: the object of the current VFS operation
757 * 4: second non-directory
758 * 5: second parent (when locking independent directories in rename)
760 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
761 * non-directories at once.
763 * The locking order between these classes is
764 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
766 enum inode_i_mutex_lock_class
776 static inline void inode_lock(struct inode
*inode
)
778 down_write(&inode
->i_rwsem
);
781 static inline void inode_unlock(struct inode
*inode
)
783 up_write(&inode
->i_rwsem
);
786 static inline void inode_lock_shared(struct inode
*inode
)
788 down_read(&inode
->i_rwsem
);
791 static inline void inode_unlock_shared(struct inode
*inode
)
793 up_read(&inode
->i_rwsem
);
796 static inline int inode_trylock(struct inode
*inode
)
798 return down_write_trylock(&inode
->i_rwsem
);
801 static inline int inode_trylock_shared(struct inode
*inode
)
803 return down_read_trylock(&inode
->i_rwsem
);
806 static inline int inode_is_locked(struct inode
*inode
)
808 return rwsem_is_locked(&inode
->i_rwsem
);
811 static inline void inode_lock_nested(struct inode
*inode
, unsigned subclass
)
813 down_write_nested(&inode
->i_rwsem
, subclass
);
816 static inline void inode_lock_shared_nested(struct inode
*inode
, unsigned subclass
)
818 down_read_nested(&inode
->i_rwsem
, subclass
);
821 void lock_two_nondirectories(struct inode
*, struct inode
*);
822 void unlock_two_nondirectories(struct inode
*, struct inode
*);
825 * NOTE: in a 32bit arch with a preemptable kernel and
826 * an UP compile the i_size_read/write must be atomic
827 * with respect to the local cpu (unlike with preempt disabled),
828 * but they don't need to be atomic with respect to other cpus like in
829 * true SMP (so they need either to either locally disable irq around
830 * the read or for example on x86 they can be still implemented as a
831 * cmpxchg8b without the need of the lock prefix). For SMP compiles
832 * and 64bit archs it makes no difference if preempt is enabled or not.
834 static inline loff_t
i_size_read(const struct inode
*inode
)
836 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
841 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
842 i_size
= inode
->i_size
;
843 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
845 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
849 i_size
= inode
->i_size
;
853 return inode
->i_size
;
858 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
859 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
860 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
862 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
864 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
866 write_seqcount_begin(&inode
->i_size_seqcount
);
867 inode
->i_size
= i_size
;
868 write_seqcount_end(&inode
->i_size_seqcount
);
870 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
872 inode
->i_size
= i_size
;
875 inode
->i_size
= i_size
;
879 static inline unsigned iminor(const struct inode
*inode
)
881 return MINOR(inode
->i_rdev
);
884 static inline unsigned imajor(const struct inode
*inode
)
886 return MAJOR(inode
->i_rdev
);
889 extern struct block_device
*I_BDEV(struct inode
*inode
);
892 rwlock_t lock
; /* protects pid, uid, euid fields */
893 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
894 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
895 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
896 int signum
; /* posix.1b rt signal to be delivered on IO */
900 * Track a single file's readahead state
902 struct file_ra_state
{
903 pgoff_t start
; /* where readahead started */
904 unsigned int size
; /* # of readahead pages */
905 unsigned int async_size
; /* do asynchronous readahead when
906 there are only # of pages ahead */
908 unsigned int ra_pages
; /* Maximum readahead window */
909 unsigned int mmap_miss
; /* Cache miss stat for mmap accesses */
910 loff_t prev_pos
; /* Cache last read() position */
914 * Check if @index falls in the readahead windows.
916 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
918 return (index
>= ra
->start
&&
919 index
< ra
->start
+ ra
->size
);
924 struct llist_node fu_llist
;
925 struct rcu_head fu_rcuhead
;
928 struct inode
*f_inode
; /* cached value */
929 const struct file_operations
*f_op
;
932 * Protects f_ep_links, f_flags.
933 * Must not be taken from IRQ context.
936 enum rw_hint f_write_hint
;
937 atomic_long_t f_count
;
938 unsigned int f_flags
;
940 struct mutex f_pos_lock
;
942 struct fown_struct f_owner
;
943 const struct cred
*f_cred
;
944 struct file_ra_state f_ra
;
947 #ifdef CONFIG_SECURITY
950 /* needed for tty driver, and maybe others */
954 /* Used by fs/eventpoll.c to link all the hooks to this file */
955 struct list_head f_ep_links
;
956 struct list_head f_tfile_llink
;
957 #endif /* #ifdef CONFIG_EPOLL */
958 struct address_space
*f_mapping
;
961 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
966 /* file identifier */
967 unsigned char f_handle
[0];
970 static inline struct file
*get_file(struct file
*f
)
972 atomic_long_inc(&f
->f_count
);
975 #define get_file_rcu_many(x, cnt) \
976 atomic_long_add_unless(&(x)->f_count, (cnt), 0)
977 #define get_file_rcu(x) get_file_rcu_many((x), 1)
978 #define file_count(x) atomic_long_read(&(x)->f_count)
980 #define MAX_NON_LFS ((1UL<<31) - 1)
982 /* Page cache limit. The filesystems should put that into their s_maxbytes
983 limits, otherwise bad things can happen in VM. */
984 #if BITS_PER_LONG==32
985 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
986 #elif BITS_PER_LONG==64
987 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
992 #define FL_DELEG 4 /* NFSv4 delegation */
993 #define FL_ACCESS 8 /* not trying to lock, just looking */
994 #define FL_EXISTS 16 /* when unlocking, test for existence */
995 #define FL_LEASE 32 /* lease held on this file */
996 #define FL_CLOSE 64 /* unlock on close */
997 #define FL_SLEEP 128 /* A blocking lock */
998 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
999 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
1000 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
1001 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
1003 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
1006 * Special return value from posix_lock_file() and vfs_lock_file() for
1007 * asynchronous locking.
1009 #define FILE_LOCK_DEFERRED 1
1011 /* legacy typedef, should eventually be removed */
1012 typedef void *fl_owner_t
;
1016 struct file_lock_operations
{
1017 void (*fl_copy_lock
)(struct file_lock
*, struct file_lock
*);
1018 void (*fl_release_private
)(struct file_lock
*);
1021 struct lock_manager_operations
{
1022 int (*lm_compare_owner
)(struct file_lock
*, struct file_lock
*);
1023 unsigned long (*lm_owner_key
)(struct file_lock
*);
1024 fl_owner_t (*lm_get_owner
)(fl_owner_t
);
1025 void (*lm_put_owner
)(fl_owner_t
);
1026 void (*lm_notify
)(struct file_lock
*); /* unblock callback */
1027 int (*lm_grant
)(struct file_lock
*, int);
1028 bool (*lm_break
)(struct file_lock
*);
1029 int (*lm_change
)(struct file_lock
*, int, struct list_head
*);
1030 void (*lm_setup
)(struct file_lock
*, void **);
1033 struct lock_manager
{
1034 struct list_head list
;
1036 * NFSv4 and up also want opens blocked during the grace period;
1043 void locks_start_grace(struct net
*, struct lock_manager
*);
1044 void locks_end_grace(struct lock_manager
*);
1045 bool locks_in_grace(struct net
*);
1046 bool opens_in_grace(struct net
*);
1048 /* that will die - we need it for nfs_lock_info */
1049 #include <linux/nfs_fs_i.h>
1052 * struct file_lock represents a generic "file lock". It's used to represent
1053 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1054 * note that the same struct is used to represent both a request for a lock and
1055 * the lock itself, but the same object is never used for both.
1057 * FIXME: should we create a separate "struct lock_request" to help distinguish
1060 * The varous i_flctx lists are ordered by:
1063 * 2) lock range start
1066 * Obviously, the last two criteria only matter for POSIX locks.
1069 struct file_lock
*fl_blocker
; /* The lock, that is blocking us */
1070 struct list_head fl_list
; /* link into file_lock_context */
1071 struct hlist_node fl_link
; /* node in global lists */
1072 struct list_head fl_blocked_requests
; /* list of requests with
1073 * ->fl_blocker pointing here
1075 struct list_head fl_blocked_member
; /* node in
1076 * ->fl_blocker->fl_blocked_requests
1078 fl_owner_t fl_owner
;
1079 unsigned int fl_flags
;
1080 unsigned char fl_type
;
1081 unsigned int fl_pid
;
1082 int fl_link_cpu
; /* what cpu's list is this on? */
1083 wait_queue_head_t fl_wait
;
1084 struct file
*fl_file
;
1088 struct fasync_struct
* fl_fasync
; /* for lease break notifications */
1089 /* for lease breaks: */
1090 unsigned long fl_break_time
;
1091 unsigned long fl_downgrade_time
;
1093 const struct file_lock_operations
*fl_ops
; /* Callbacks for filesystems */
1094 const struct lock_manager_operations
*fl_lmops
; /* Callbacks for lockmanagers */
1096 struct nfs_lock_info nfs_fl
;
1097 struct nfs4_lock_info nfs4_fl
;
1099 struct list_head link
; /* link in AFS vnode's pending_locks list */
1100 int state
; /* state of grant or error if -ve */
1101 unsigned int debug_id
;
1104 } __randomize_layout
;
1106 struct file_lock_context
{
1107 spinlock_t flc_lock
;
1108 struct list_head flc_flock
;
1109 struct list_head flc_posix
;
1110 struct list_head flc_lease
;
1113 /* The following constant reflects the upper bound of the file/locking space */
1115 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1116 #define OFFSET_MAX INT_LIMIT(loff_t)
1117 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1120 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
1122 #define locks_inode(f) file_inode(f)
1124 #ifdef CONFIG_FILE_LOCKING
1125 extern int fcntl_getlk(struct file
*, unsigned int, struct flock
*);
1126 extern int fcntl_setlk(unsigned int, struct file
*, unsigned int,
1129 #if BITS_PER_LONG == 32
1130 extern int fcntl_getlk64(struct file
*, unsigned int, struct flock64
*);
1131 extern int fcntl_setlk64(unsigned int, struct file
*, unsigned int,
1135 extern int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
);
1136 extern int fcntl_getlease(struct file
*filp
);
1139 void locks_free_lock_context(struct inode
*inode
);
1140 void locks_free_lock(struct file_lock
*fl
);
1141 extern void locks_init_lock(struct file_lock
*);
1142 extern struct file_lock
* locks_alloc_lock(void);
1143 extern void locks_copy_lock(struct file_lock
*, struct file_lock
*);
1144 extern void locks_copy_conflock(struct file_lock
*, struct file_lock
*);
1145 extern void locks_remove_posix(struct file
*, fl_owner_t
);
1146 extern void locks_remove_file(struct file
*);
1147 extern void locks_release_private(struct file_lock
*);
1148 extern void posix_test_lock(struct file
*, struct file_lock
*);
1149 extern int posix_lock_file(struct file
*, struct file_lock
*, struct file_lock
*);
1150 extern int locks_delete_block(struct file_lock
*);
1151 extern int vfs_test_lock(struct file
*, struct file_lock
*);
1152 extern int vfs_lock_file(struct file
*, unsigned int, struct file_lock
*, struct file_lock
*);
1153 extern int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
);
1154 extern int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
);
1155 extern int __break_lease(struct inode
*inode
, unsigned int flags
, unsigned int type
);
1156 extern void lease_get_mtime(struct inode
*, struct timespec64
*time
);
1157 extern int generic_setlease(struct file
*, long, struct file_lock
**, void **priv
);
1158 extern int vfs_setlease(struct file
*, long, struct file_lock
**, void **);
1159 extern int lease_modify(struct file_lock
*, int, struct list_head
*);
1160 struct files_struct
;
1161 extern void show_fd_locks(struct seq_file
*f
,
1162 struct file
*filp
, struct files_struct
*files
);
1163 #else /* !CONFIG_FILE_LOCKING */
1164 static inline int fcntl_getlk(struct file
*file
, unsigned int cmd
,
1165 struct flock __user
*user
)
1170 static inline int fcntl_setlk(unsigned int fd
, struct file
*file
,
1171 unsigned int cmd
, struct flock __user
*user
)
1176 #if BITS_PER_LONG == 32
1177 static inline int fcntl_getlk64(struct file
*file
, unsigned int cmd
,
1178 struct flock64 __user
*user
)
1183 static inline int fcntl_setlk64(unsigned int fd
, struct file
*file
,
1184 unsigned int cmd
, struct flock64 __user
*user
)
1189 static inline int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
)
1194 static inline int fcntl_getlease(struct file
*filp
)
1200 locks_free_lock_context(struct inode
*inode
)
1204 static inline void locks_init_lock(struct file_lock
*fl
)
1209 static inline void locks_copy_conflock(struct file_lock
*new, struct file_lock
*fl
)
1214 static inline void locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
1219 static inline void locks_remove_posix(struct file
*filp
, fl_owner_t owner
)
1224 static inline void locks_remove_file(struct file
*filp
)
1229 static inline void posix_test_lock(struct file
*filp
, struct file_lock
*fl
)
1234 static inline int posix_lock_file(struct file
*filp
, struct file_lock
*fl
,
1235 struct file_lock
*conflock
)
1240 static inline int locks_delete_block(struct file_lock
*waiter
)
1245 static inline int vfs_test_lock(struct file
*filp
, struct file_lock
*fl
)
1250 static inline int vfs_lock_file(struct file
*filp
, unsigned int cmd
,
1251 struct file_lock
*fl
, struct file_lock
*conf
)
1256 static inline int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
)
1261 static inline int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
)
1266 static inline int __break_lease(struct inode
*inode
, unsigned int mode
, unsigned int type
)
1271 static inline void lease_get_mtime(struct inode
*inode
,
1272 struct timespec64
*time
)
1277 static inline int generic_setlease(struct file
*filp
, long arg
,
1278 struct file_lock
**flp
, void **priv
)
1283 static inline int vfs_setlease(struct file
*filp
, long arg
,
1284 struct file_lock
**lease
, void **priv
)
1289 static inline int lease_modify(struct file_lock
*fl
, int arg
,
1290 struct list_head
*dispose
)
1295 struct files_struct
;
1296 static inline void show_fd_locks(struct seq_file
*f
,
1297 struct file
*filp
, struct files_struct
*files
) {}
1298 #endif /* !CONFIG_FILE_LOCKING */
1300 static inline struct inode
*file_inode(const struct file
*f
)
1305 static inline struct dentry
*file_dentry(const struct file
*file
)
1307 return d_real(file
->f_path
.dentry
, file_inode(file
));
1310 static inline int locks_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1312 return locks_lock_inode_wait(locks_inode(filp
), fl
);
1315 struct fasync_struct
{
1319 struct fasync_struct
*fa_next
; /* singly linked list */
1320 struct file
*fa_file
;
1321 struct rcu_head fa_rcu
;
1324 #define FASYNC_MAGIC 0x4601
1326 /* SMP safe fasync helpers: */
1327 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1328 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1329 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1330 extern struct fasync_struct
*fasync_alloc(void);
1331 extern void fasync_free(struct fasync_struct
*);
1333 /* can be called from interrupts */
1334 extern void kill_fasync(struct fasync_struct
**, int, int);
1336 extern void __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1337 extern int f_setown(struct file
*filp
, unsigned long arg
, int force
);
1338 extern void f_delown(struct file
*filp
);
1339 extern pid_t
f_getown(struct file
*filp
);
1340 extern int send_sigurg(struct fown_struct
*fown
);
1343 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1344 * represented in both.
1346 #define SB_RDONLY 1 /* Mount read-only */
1347 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1348 #define SB_NODEV 4 /* Disallow access to device special files */
1349 #define SB_NOEXEC 8 /* Disallow program execution */
1350 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1351 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1352 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1353 #define SB_NOATIME 1024 /* Do not update access times. */
1354 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1355 #define SB_SILENT 32768
1356 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1357 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1358 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1359 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1361 /* These sb flags are internal to the kernel */
1362 #define SB_SUBMOUNT (1<<26)
1363 #define SB_FORCE (1<<27)
1364 #define SB_NOSEC (1<<28)
1365 #define SB_BORN (1<<29)
1366 #define SB_ACTIVE (1<<30)
1367 #define SB_NOUSER (1<<31)
1373 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1374 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1375 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1376 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1377 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1380 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1381 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1382 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1383 #define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */
1385 /* sb->s_iflags to limit user namespace mounts */
1386 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1387 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1388 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1390 /* Possible states of 'frozen' field */
1392 SB_UNFROZEN
= 0, /* FS is unfrozen */
1393 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1394 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1395 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1396 * internal threads if needed) */
1397 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1400 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1403 int frozen
; /* Is sb frozen? */
1404 wait_queue_head_t wait_unfrozen
; /* for get_super_thawed() */
1405 struct percpu_rw_semaphore rw_sem
[SB_FREEZE_LEVELS
];
1408 struct super_block
{
1409 struct list_head s_list
; /* Keep this first */
1410 dev_t s_dev
; /* search index; _not_ kdev_t */
1411 unsigned char s_blocksize_bits
;
1412 unsigned long s_blocksize
;
1413 loff_t s_maxbytes
; /* Max file size */
1414 struct file_system_type
*s_type
;
1415 const struct super_operations
*s_op
;
1416 const struct dquot_operations
*dq_op
;
1417 const struct quotactl_ops
*s_qcop
;
1418 const struct export_operations
*s_export_op
;
1419 unsigned long s_flags
;
1420 unsigned long s_iflags
; /* internal SB_I_* flags */
1421 unsigned long s_magic
;
1422 struct dentry
*s_root
;
1423 struct rw_semaphore s_umount
;
1426 #ifdef CONFIG_SECURITY
1429 const struct xattr_handler
**s_xattr
;
1430 #ifdef CONFIG_FS_ENCRYPTION
1431 const struct fscrypt_operations
*s_cop
;
1433 struct hlist_bl_head s_roots
; /* alternate root dentries for NFS */
1434 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1435 struct block_device
*s_bdev
;
1436 struct backing_dev_info
*s_bdi
;
1437 struct mtd_info
*s_mtd
;
1438 struct hlist_node s_instances
;
1439 unsigned int s_quota_types
; /* Bitmask of supported quota types */
1440 struct quota_info s_dquot
; /* Diskquota specific options */
1442 struct sb_writers s_writers
;
1445 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1446 * s_fsnotify_marks together for cache efficiency. They are frequently
1447 * accessed and rarely modified.
1449 void *s_fs_info
; /* Filesystem private info */
1451 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1453 #ifdef CONFIG_FSNOTIFY
1454 __u32 s_fsnotify_mask
;
1455 struct fsnotify_mark_connector __rcu
*s_fsnotify_marks
;
1458 char s_id
[32]; /* Informational name */
1459 uuid_t s_uuid
; /* UUID */
1461 unsigned int s_max_links
;
1465 * The next field is for VFS *only*. No filesystems have any business
1466 * even looking at it. You had been warned.
1468 struct mutex s_vfs_rename_mutex
; /* Kludge */
1471 * Filesystem subtype. If non-empty the filesystem type field
1472 * in /proc/mounts will be "type.subtype"
1474 const char *s_subtype
;
1476 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1479 * Saved pool identifier for cleancache (-1 means none)
1481 int cleancache_poolid
;
1483 struct shrinker s_shrink
; /* per-sb shrinker handle */
1485 /* Number of inodes with nlink == 0 but still referenced */
1486 atomic_long_t s_remove_count
;
1488 /* Pending fsnotify inode refs */
1489 atomic_long_t s_fsnotify_inode_refs
;
1491 /* Being remounted read-only */
1492 int s_readonly_remount
;
1494 /* AIO completions deferred from interrupt context */
1495 struct workqueue_struct
*s_dio_done_wq
;
1496 struct hlist_head s_pins
;
1499 * Owning user namespace and default context in which to
1500 * interpret filesystem uids, gids, quotas, device nodes,
1501 * xattrs and security labels.
1503 struct user_namespace
*s_user_ns
;
1506 * The list_lru structure is essentially just a pointer to a table
1507 * of per-node lru lists, each of which has its own spinlock.
1508 * There is no need to put them into separate cachelines.
1510 struct list_lru s_dentry_lru
;
1511 struct list_lru s_inode_lru
;
1512 struct rcu_head rcu
;
1513 struct work_struct destroy_work
;
1515 struct mutex s_sync_lock
; /* sync serialisation lock */
1518 * Indicates how deep in a filesystem stack this SB is
1522 /* s_inode_list_lock protects s_inodes */
1523 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp
;
1524 struct list_head s_inodes
; /* all inodes */
1526 spinlock_t s_inode_wblist_lock
;
1527 struct list_head s_inodes_wb
; /* writeback inodes */
1528 } __randomize_layout
;
1530 /* Helper functions so that in most cases filesystems will
1531 * not need to deal directly with kuid_t and kgid_t and can
1532 * instead deal with the raw numeric values that are stored
1533 * in the filesystem.
1535 static inline uid_t
i_uid_read(const struct inode
*inode
)
1537 return from_kuid(inode
->i_sb
->s_user_ns
, inode
->i_uid
);
1540 static inline gid_t
i_gid_read(const struct inode
*inode
)
1542 return from_kgid(inode
->i_sb
->s_user_ns
, inode
->i_gid
);
1545 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
1547 inode
->i_uid
= make_kuid(inode
->i_sb
->s_user_ns
, uid
);
1550 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
1552 inode
->i_gid
= make_kgid(inode
->i_sb
->s_user_ns
, gid
);
1555 extern struct timespec64
timespec64_trunc(struct timespec64 t
, unsigned gran
);
1556 extern struct timespec64
current_time(struct inode
*inode
);
1559 * Snapshotting support.
1562 void __sb_end_write(struct super_block
*sb
, int level
);
1563 int __sb_start_write(struct super_block
*sb
, int level
, bool wait
);
1565 #define __sb_writers_acquired(sb, lev) \
1566 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1567 #define __sb_writers_release(sb, lev) \
1568 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1571 * sb_end_write - drop write access to a superblock
1572 * @sb: the super we wrote to
1574 * Decrement number of writers to the filesystem. Wake up possible waiters
1575 * wanting to freeze the filesystem.
1577 static inline void sb_end_write(struct super_block
*sb
)
1579 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1583 * sb_end_pagefault - drop write access to a superblock from a page fault
1584 * @sb: the super we wrote to
1586 * Decrement number of processes handling write page fault to the filesystem.
1587 * Wake up possible waiters wanting to freeze the filesystem.
1589 static inline void sb_end_pagefault(struct super_block
*sb
)
1591 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1595 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1596 * @sb: the super we wrote to
1598 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1599 * waiters wanting to freeze the filesystem.
1601 static inline void sb_end_intwrite(struct super_block
*sb
)
1603 __sb_end_write(sb
, SB_FREEZE_FS
);
1607 * sb_start_write - get write access to a superblock
1608 * @sb: the super we write to
1610 * When a process wants to write data or metadata to a file system (i.e. dirty
1611 * a page or an inode), it should embed the operation in a sb_start_write() -
1612 * sb_end_write() pair to get exclusion against file system freezing. This
1613 * function increments number of writers preventing freezing. If the file
1614 * system is already frozen, the function waits until the file system is
1617 * Since freeze protection behaves as a lock, users have to preserve
1618 * ordering of freeze protection and other filesystem locks. Generally,
1619 * freeze protection should be the outermost lock. In particular, we have:
1622 * -> i_mutex (write path, truncate, directory ops, ...)
1623 * -> s_umount (freeze_super, thaw_super)
1625 static inline void sb_start_write(struct super_block
*sb
)
1627 __sb_start_write(sb
, SB_FREEZE_WRITE
, true);
1630 static inline int sb_start_write_trylock(struct super_block
*sb
)
1632 return __sb_start_write(sb
, SB_FREEZE_WRITE
, false);
1636 * sb_start_pagefault - get write access to a superblock from a page fault
1637 * @sb: the super we write to
1639 * When a process starts handling write page fault, it should embed the
1640 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1641 * exclusion against file system freezing. This is needed since the page fault
1642 * is going to dirty a page. This function increments number of running page
1643 * faults preventing freezing. If the file system is already frozen, the
1644 * function waits until the file system is thawed.
1646 * Since page fault freeze protection behaves as a lock, users have to preserve
1647 * ordering of freeze protection and other filesystem locks. It is advised to
1648 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1649 * handling code implies lock dependency:
1652 * -> sb_start_pagefault
1654 static inline void sb_start_pagefault(struct super_block
*sb
)
1656 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
, true);
1660 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1661 * @sb: the super we write to
1663 * This is the third level of protection against filesystem freezing. It is
1664 * free for use by a filesystem. The only requirement is that it must rank
1665 * below sb_start_pagefault.
1667 * For example filesystem can call sb_start_intwrite() when starting a
1668 * transaction which somewhat eases handling of freezing for internal sources
1669 * of filesystem changes (internal fs threads, discarding preallocation on file
1672 static inline void sb_start_intwrite(struct super_block
*sb
)
1674 __sb_start_write(sb
, SB_FREEZE_FS
, true);
1677 static inline int sb_start_intwrite_trylock(struct super_block
*sb
)
1679 return __sb_start_write(sb
, SB_FREEZE_FS
, false);
1683 extern bool inode_owner_or_capable(const struct inode
*inode
);
1686 * VFS helper functions..
1688 extern int vfs_create(struct inode
*, struct dentry
*, umode_t
, bool);
1689 extern int vfs_mkdir(struct inode
*, struct dentry
*, umode_t
);
1690 extern int vfs_mknod(struct inode
*, struct dentry
*, umode_t
, dev_t
);
1691 extern int vfs_symlink(struct inode
*, struct dentry
*, const char *);
1692 extern int vfs_link(struct dentry
*, struct inode
*, struct dentry
*, struct inode
**);
1693 extern int vfs_rmdir(struct inode
*, struct dentry
*);
1694 extern int vfs_unlink(struct inode
*, struct dentry
*, struct inode
**);
1695 extern int vfs_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*, struct inode
**, unsigned int);
1696 extern int vfs_whiteout(struct inode
*, struct dentry
*);
1698 extern struct dentry
*vfs_tmpfile(struct dentry
*dentry
, umode_t mode
,
1701 int vfs_mkobj(struct dentry
*, umode_t
,
1702 int (*f
)(struct dentry
*, umode_t
, void *),
1705 extern long vfs_ioctl(struct file
*file
, unsigned int cmd
, unsigned long arg
);
1708 * VFS file helper functions.
1710 extern void inode_init_owner(struct inode
*inode
, const struct inode
*dir
,
1712 extern bool may_open_dev(const struct path
*path
);
1714 * VFS FS_IOC_FIEMAP helper definitions.
1716 struct fiemap_extent_info
{
1717 unsigned int fi_flags
; /* Flags as passed from user */
1718 unsigned int fi_extents_mapped
; /* Number of mapped extents */
1719 unsigned int fi_extents_max
; /* Size of fiemap_extent array */
1720 struct fiemap_extent __user
*fi_extents_start
; /* Start of
1721 fiemap_extent array */
1723 int fiemap_fill_next_extent(struct fiemap_extent_info
*info
, u64 logical
,
1724 u64 phys
, u64 len
, u32 flags
);
1725 int fiemap_check_flags(struct fiemap_extent_info
*fieinfo
, u32 fs_flags
);
1728 * This is the "filldir" function type, used by readdir() to let
1729 * the kernel specify what kind of dirent layout it wants to have.
1730 * This allows the kernel to read directories into kernel space or
1731 * to have different dirent layouts depending on the binary type.
1734 typedef int (*filldir_t
)(struct dir_context
*, const char *, int, loff_t
, u64
,
1737 struct dir_context
{
1742 struct block_device_operations
;
1744 /* These macros are for out of kernel modules to test that
1745 * the kernel supports the unlocked_ioctl and compat_ioctl
1746 * fields in struct file_operations. */
1747 #define HAVE_COMPAT_IOCTL 1
1748 #define HAVE_UNLOCKED_IOCTL 1
1751 * These flags let !MMU mmap() govern direct device mapping vs immediate
1752 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1754 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1755 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1756 * NOMMU_MAP_READ: Can be mapped for reading
1757 * NOMMU_MAP_WRITE: Can be mapped for writing
1758 * NOMMU_MAP_EXEC: Can be mapped for execution
1760 #define NOMMU_MAP_COPY 0x00000001
1761 #define NOMMU_MAP_DIRECT 0x00000008
1762 #define NOMMU_MAP_READ VM_MAYREAD
1763 #define NOMMU_MAP_WRITE VM_MAYWRITE
1764 #define NOMMU_MAP_EXEC VM_MAYEXEC
1766 #define NOMMU_VMFLAGS \
1767 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1770 * These flags control the behavior of the remap_file_range function pointer.
1771 * If it is called with len == 0 that means "remap to end of source file".
1772 * See Documentation/filesystems/vfs.txt for more details about this call.
1774 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
1775 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
1777 #define REMAP_FILE_DEDUP (1 << 0)
1778 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
1781 * These flags signal that the caller is ok with altering various aspects of
1782 * the behavior of the remap operation. The changes must be made by the
1783 * implementation; the vfs remap helper functions can take advantage of them.
1784 * Flags in this category exist to preserve the quirky behavior of the hoisted
1785 * btrfs clone/dedupe ioctls.
1787 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
1791 struct file_operations
{
1792 struct module
*owner
;
1793 loff_t (*llseek
) (struct file
*, loff_t
, int);
1794 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
1795 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
1796 ssize_t (*read_iter
) (struct kiocb
*, struct iov_iter
*);
1797 ssize_t (*write_iter
) (struct kiocb
*, struct iov_iter
*);
1798 int (*iopoll
)(struct kiocb
*kiocb
, bool spin
);
1799 int (*iterate
) (struct file
*, struct dir_context
*);
1800 int (*iterate_shared
) (struct file
*, struct dir_context
*);
1801 __poll_t (*poll
) (struct file
*, struct poll_table_struct
*);
1802 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
1803 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
1804 int (*mmap
) (struct file
*, struct vm_area_struct
*);
1805 unsigned long mmap_supported_flags
;
1806 int (*open
) (struct inode
*, struct file
*);
1807 int (*flush
) (struct file
*, fl_owner_t id
);
1808 int (*release
) (struct inode
*, struct file
*);
1809 int (*fsync
) (struct file
*, loff_t
, loff_t
, int datasync
);
1810 int (*fasync
) (int, struct file
*, int);
1811 int (*lock
) (struct file
*, int, struct file_lock
*);
1812 ssize_t (*sendpage
) (struct file
*, struct page
*, int, size_t, loff_t
*, int);
1813 unsigned long (*get_unmapped_area
)(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1814 int (*check_flags
)(int);
1815 int (*flock
) (struct file
*, int, struct file_lock
*);
1816 ssize_t (*splice_write
)(struct pipe_inode_info
*, struct file
*, loff_t
*, size_t, unsigned int);
1817 ssize_t (*splice_read
)(struct file
*, loff_t
*, struct pipe_inode_info
*, size_t, unsigned int);
1818 int (*setlease
)(struct file
*, long, struct file_lock
**, void **);
1819 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
1821 void (*show_fdinfo
)(struct seq_file
*m
, struct file
*f
);
1823 unsigned (*mmap_capabilities
)(struct file
*);
1825 ssize_t (*copy_file_range
)(struct file
*, loff_t
, struct file
*,
1826 loff_t
, size_t, unsigned int);
1827 loff_t (*remap_file_range
)(struct file
*file_in
, loff_t pos_in
,
1828 struct file
*file_out
, loff_t pos_out
,
1829 loff_t len
, unsigned int remap_flags
);
1830 int (*fadvise
)(struct file
*, loff_t
, loff_t
, int);
1831 } __randomize_layout
;
1833 struct inode_operations
{
1834 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*, unsigned int);
1835 const char * (*get_link
) (struct dentry
*, struct inode
*, struct delayed_call
*);
1836 int (*permission
) (struct inode
*, int);
1837 struct posix_acl
* (*get_acl
)(struct inode
*, int);
1839 int (*readlink
) (struct dentry
*, char __user
*,int);
1841 int (*create
) (struct inode
*,struct dentry
*, umode_t
, bool);
1842 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
1843 int (*unlink
) (struct inode
*,struct dentry
*);
1844 int (*symlink
) (struct inode
*,struct dentry
*,const char *);
1845 int (*mkdir
) (struct inode
*,struct dentry
*,umode_t
);
1846 int (*rmdir
) (struct inode
*,struct dentry
*);
1847 int (*mknod
) (struct inode
*,struct dentry
*,umode_t
,dev_t
);
1848 int (*rename
) (struct inode
*, struct dentry
*,
1849 struct inode
*, struct dentry
*, unsigned int);
1850 int (*setattr
) (struct dentry
*, struct iattr
*);
1851 int (*getattr
) (const struct path
*, struct kstat
*, u32
, unsigned int);
1852 ssize_t (*listxattr
) (struct dentry
*, char *, size_t);
1853 int (*fiemap
)(struct inode
*, struct fiemap_extent_info
*, u64 start
,
1855 int (*update_time
)(struct inode
*, struct timespec64
*, int);
1856 int (*atomic_open
)(struct inode
*, struct dentry
*,
1857 struct file
*, unsigned open_flag
,
1858 umode_t create_mode
);
1859 int (*tmpfile
) (struct inode
*, struct dentry
*, umode_t
);
1860 int (*set_acl
)(struct inode
*, struct posix_acl
*, int);
1861 } ____cacheline_aligned
;
1863 static inline ssize_t
call_read_iter(struct file
*file
, struct kiocb
*kio
,
1864 struct iov_iter
*iter
)
1866 return file
->f_op
->read_iter(kio
, iter
);
1869 static inline ssize_t
call_write_iter(struct file
*file
, struct kiocb
*kio
,
1870 struct iov_iter
*iter
)
1872 return file
->f_op
->write_iter(kio
, iter
);
1875 static inline int call_mmap(struct file
*file
, struct vm_area_struct
*vma
)
1877 return file
->f_op
->mmap(file
, vma
);
1880 ssize_t
rw_copy_check_uvector(int type
, const struct iovec __user
* uvector
,
1881 unsigned long nr_segs
, unsigned long fast_segs
,
1882 struct iovec
*fast_pointer
,
1883 struct iovec
**ret_pointer
);
1885 extern ssize_t
__vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1886 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1887 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1888 extern ssize_t
vfs_readv(struct file
*, const struct iovec __user
*,
1889 unsigned long, loff_t
*, rwf_t
);
1890 extern ssize_t
vfs_copy_file_range(struct file
*, loff_t
, struct file
*,
1891 loff_t
, size_t, unsigned int);
1892 extern int generic_remap_file_range_prep(struct file
*file_in
, loff_t pos_in
,
1893 struct file
*file_out
, loff_t pos_out
,
1895 unsigned int remap_flags
);
1896 extern loff_t
do_clone_file_range(struct file
*file_in
, loff_t pos_in
,
1897 struct file
*file_out
, loff_t pos_out
,
1898 loff_t len
, unsigned int remap_flags
);
1899 extern loff_t
vfs_clone_file_range(struct file
*file_in
, loff_t pos_in
,
1900 struct file
*file_out
, loff_t pos_out
,
1901 loff_t len
, unsigned int remap_flags
);
1902 extern int vfs_dedupe_file_range(struct file
*file
,
1903 struct file_dedupe_range
*same
);
1904 extern loff_t
vfs_dedupe_file_range_one(struct file
*src_file
, loff_t src_pos
,
1905 struct file
*dst_file
, loff_t dst_pos
,
1906 loff_t len
, unsigned int remap_flags
);
1909 struct super_operations
{
1910 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
1911 void (*destroy_inode
)(struct inode
*);
1912 void (*free_inode
)(struct inode
*);
1914 void (*dirty_inode
) (struct inode
*, int flags
);
1915 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
1916 int (*drop_inode
) (struct inode
*);
1917 void (*evict_inode
) (struct inode
*);
1918 void (*put_super
) (struct super_block
*);
1919 int (*sync_fs
)(struct super_block
*sb
, int wait
);
1920 int (*freeze_super
) (struct super_block
*);
1921 int (*freeze_fs
) (struct super_block
*);
1922 int (*thaw_super
) (struct super_block
*);
1923 int (*unfreeze_fs
) (struct super_block
*);
1924 int (*statfs
) (struct dentry
*, struct kstatfs
*);
1925 int (*remount_fs
) (struct super_block
*, int *, char *);
1926 void (*umount_begin
) (struct super_block
*);
1928 int (*show_options
)(struct seq_file
*, struct dentry
*);
1929 int (*show_devname
)(struct seq_file
*, struct dentry
*);
1930 int (*show_path
)(struct seq_file
*, struct dentry
*);
1931 int (*show_stats
)(struct seq_file
*, struct dentry
*);
1933 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
1934 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
1935 struct dquot
**(*get_dquots
)(struct inode
*);
1937 int (*bdev_try_to_free_page
)(struct super_block
*, struct page
*, gfp_t
);
1938 long (*nr_cached_objects
)(struct super_block
*,
1939 struct shrink_control
*);
1940 long (*free_cached_objects
)(struct super_block
*,
1941 struct shrink_control
*);
1945 * Inode flags - they have no relation to superblock flags now
1947 #define S_SYNC 1 /* Writes are synced at once */
1948 #define S_NOATIME 2 /* Do not update access times */
1949 #define S_APPEND 4 /* Append-only file */
1950 #define S_IMMUTABLE 8 /* Immutable file */
1951 #define S_DEAD 16 /* removed, but still open directory */
1952 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1953 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1954 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1955 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1956 #define S_PRIVATE 512 /* Inode is fs-internal */
1957 #define S_IMA 1024 /* Inode has an associated IMA struct */
1958 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1959 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1960 #ifdef CONFIG_FS_DAX
1961 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1963 #define S_DAX 0 /* Make all the DAX code disappear */
1965 #define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */
1966 #define S_CASEFOLD 32768 /* Casefolded file */
1969 * Note that nosuid etc flags are inode-specific: setting some file-system
1970 * flags just means all the inodes inherit those flags by default. It might be
1971 * possible to override it selectively if you really wanted to with some
1972 * ioctl() that is not currently implemented.
1974 * Exception: SB_RDONLY is always applied to the entire file system.
1976 * Unfortunately, it is possible to change a filesystems flags with it mounted
1977 * with files in use. This means that all of the inodes will not have their
1978 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1979 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1981 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1983 static inline bool sb_rdonly(const struct super_block
*sb
) { return sb
->s_flags
& SB_RDONLY
; }
1984 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
1985 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
1986 ((inode)->i_flags & S_SYNC))
1987 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
1988 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1989 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
1990 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
1991 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
1993 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1994 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1995 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1996 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
1998 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1999 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
2000 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
2001 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
2002 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
2003 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2004 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2005 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2006 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2007 #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD)
2009 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2010 (inode)->i_rdev == WHITEOUT_DEV)
2012 static inline bool HAS_UNMAPPED_ID(struct inode
*inode
)
2014 return !uid_valid(inode
->i_uid
) || !gid_valid(inode
->i_gid
);
2017 static inline enum rw_hint
file_write_hint(struct file
*file
)
2019 if (file
->f_write_hint
!= WRITE_LIFE_NOT_SET
)
2020 return file
->f_write_hint
;
2022 return file_inode(file
)->i_write_hint
;
2025 static inline int iocb_flags(struct file
*file
);
2027 static inline u16
ki_hint_validate(enum rw_hint hint
)
2029 typeof(((struct kiocb
*)0)->ki_hint
) max_hint
= -1;
2031 if (hint
<= max_hint
)
2036 static inline void init_sync_kiocb(struct kiocb
*kiocb
, struct file
*filp
)
2038 *kiocb
= (struct kiocb
) {
2040 .ki_flags
= iocb_flags(filp
),
2041 .ki_hint
= ki_hint_validate(file_write_hint(filp
)),
2042 .ki_ioprio
= get_current_ioprio(),
2047 * Inode state bits. Protected by inode->i_lock
2049 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
2050 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
2052 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2053 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2054 * various stages of removing an inode.
2056 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2058 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2059 * fdatasync(). i_atime is the usual cause.
2060 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2061 * these changes separately from I_DIRTY_SYNC so that we
2062 * don't have to write inode on fdatasync() when only
2063 * mtime has changed in it.
2064 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2065 * I_NEW Serves as both a mutex and completion notification.
2066 * New inodes set I_NEW. If two processes both create
2067 * the same inode, one of them will release its inode and
2068 * wait for I_NEW to be released before returning.
2069 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2070 * also cause waiting on I_NEW, without I_NEW actually
2071 * being set. find_inode() uses this to prevent returning
2072 * nearly-dead inodes.
2073 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2074 * is zero. I_FREEING must be set when I_WILL_FREE is
2076 * I_FREEING Set when inode is about to be freed but still has dirty
2077 * pages or buffers attached or the inode itself is still
2079 * I_CLEAR Added by clear_inode(). In this state the inode is
2080 * clean and can be destroyed. Inode keeps I_FREEING.
2082 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2083 * prohibited for many purposes. iget() must wait for
2084 * the inode to be completely released, then create it
2085 * anew. Other functions will just ignore such inodes,
2086 * if appropriate. I_NEW is used for waiting.
2088 * I_SYNC Writeback of inode is running. The bit is set during
2089 * data writeback, and cleared with a wakeup on the bit
2090 * address once it is done. The bit is also used to pin
2091 * the inode in memory for flusher thread.
2093 * I_REFERENCED Marks the inode as recently references on the LRU list.
2095 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2097 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2098 * synchronize competing switching instances and to tell
2099 * wb stat updates to grab the i_pages lock. See
2100 * inode_switch_wbs_work_fn() for details.
2102 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2103 * and work dirs among overlayfs mounts.
2105 * I_CREATING New object's inode in the middle of setting up.
2107 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2109 #define I_DIRTY_SYNC (1 << 0)
2110 #define I_DIRTY_DATASYNC (1 << 1)
2111 #define I_DIRTY_PAGES (1 << 2)
2113 #define I_NEW (1 << __I_NEW)
2114 #define I_WILL_FREE (1 << 4)
2115 #define I_FREEING (1 << 5)
2116 #define I_CLEAR (1 << 6)
2118 #define I_SYNC (1 << __I_SYNC)
2119 #define I_REFERENCED (1 << 8)
2120 #define __I_DIO_WAKEUP 9
2121 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2122 #define I_LINKABLE (1 << 10)
2123 #define I_DIRTY_TIME (1 << 11)
2124 #define __I_DIRTY_TIME_EXPIRED 12
2125 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
2126 #define I_WB_SWITCH (1 << 13)
2127 #define I_OVL_INUSE (1 << 14)
2128 #define I_CREATING (1 << 15)
2130 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2131 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2132 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2134 extern void __mark_inode_dirty(struct inode
*, int);
2135 static inline void mark_inode_dirty(struct inode
*inode
)
2137 __mark_inode_dirty(inode
, I_DIRTY
);
2140 static inline void mark_inode_dirty_sync(struct inode
*inode
)
2142 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
2145 extern void inc_nlink(struct inode
*inode
);
2146 extern void drop_nlink(struct inode
*inode
);
2147 extern void clear_nlink(struct inode
*inode
);
2148 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
2150 static inline void inode_inc_link_count(struct inode
*inode
)
2153 mark_inode_dirty(inode
);
2156 static inline void inode_dec_link_count(struct inode
*inode
)
2159 mark_inode_dirty(inode
);
2162 enum file_time_flags
{
2169 extern bool atime_needs_update(const struct path
*, struct inode
*);
2170 extern void touch_atime(const struct path
*);
2171 static inline void file_accessed(struct file
*file
)
2173 if (!(file
->f_flags
& O_NOATIME
))
2174 touch_atime(&file
->f_path
);
2177 int sync_inode(struct inode
*inode
, struct writeback_control
*wbc
);
2178 int sync_inode_metadata(struct inode
*inode
, int wait
);
2180 struct file_system_type
{
2183 #define FS_REQUIRES_DEV 1
2184 #define FS_BINARY_MOUNTDATA 2
2185 #define FS_HAS_SUBTYPE 4
2186 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2187 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2188 int (*init_fs_context
)(struct fs_context
*);
2189 const struct fs_parameter_description
*parameters
;
2190 struct dentry
*(*mount
) (struct file_system_type
*, int,
2191 const char *, void *);
2192 void (*kill_sb
) (struct super_block
*);
2193 struct module
*owner
;
2194 struct file_system_type
* next
;
2195 struct hlist_head fs_supers
;
2197 struct lock_class_key s_lock_key
;
2198 struct lock_class_key s_umount_key
;
2199 struct lock_class_key s_vfs_rename_key
;
2200 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
2202 struct lock_class_key i_lock_key
;
2203 struct lock_class_key i_mutex_key
;
2204 struct lock_class_key i_mutex_dir_key
;
2207 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2209 extern struct dentry
*mount_ns(struct file_system_type
*fs_type
,
2210 int flags
, void *data
, void *ns
, struct user_namespace
*user_ns
,
2211 int (*fill_super
)(struct super_block
*, void *, int));
2213 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
2214 int flags
, const char *dev_name
, void *data
,
2215 int (*fill_super
)(struct super_block
*, void *, int));
2217 static inline struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
2218 int flags
, const char *dev_name
, void *data
,
2219 int (*fill_super
)(struct super_block
*, void *, int))
2221 return ERR_PTR(-ENODEV
);
2224 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
2225 int flags
, void *data
,
2226 int (*fill_super
)(struct super_block
*, void *, int));
2227 extern struct dentry
*mount_nodev(struct file_system_type
*fs_type
,
2228 int flags
, void *data
,
2229 int (*fill_super
)(struct super_block
*, void *, int));
2230 extern struct dentry
*mount_subtree(struct vfsmount
*mnt
, const char *path
);
2231 void generic_shutdown_super(struct super_block
*sb
);
2233 void kill_block_super(struct super_block
*sb
);
2235 static inline void kill_block_super(struct super_block
*sb
)
2240 void kill_anon_super(struct super_block
*sb
);
2241 void kill_litter_super(struct super_block
*sb
);
2242 void deactivate_super(struct super_block
*sb
);
2243 void deactivate_locked_super(struct super_block
*sb
);
2244 int set_anon_super(struct super_block
*s
, void *data
);
2245 int set_anon_super_fc(struct super_block
*s
, struct fs_context
*fc
);
2246 int get_anon_bdev(dev_t
*);
2247 void free_anon_bdev(dev_t
);
2248 struct super_block
*sget_fc(struct fs_context
*fc
,
2249 int (*test
)(struct super_block
*, struct fs_context
*),
2250 int (*set
)(struct super_block
*, struct fs_context
*));
2251 struct super_block
*sget_userns(struct file_system_type
*type
,
2252 int (*test
)(struct super_block
*,void *),
2253 int (*set
)(struct super_block
*,void *),
2254 int flags
, struct user_namespace
*user_ns
,
2256 struct super_block
*sget(struct file_system_type
*type
,
2257 int (*test
)(struct super_block
*,void *),
2258 int (*set
)(struct super_block
*,void *),
2259 int flags
, void *data
);
2260 extern struct dentry
*mount_pseudo_xattr(struct file_system_type
*, char *,
2261 const struct super_operations
*ops
,
2262 const struct xattr_handler
**xattr
,
2263 const struct dentry_operations
*dops
,
2266 static inline struct dentry
*
2267 mount_pseudo(struct file_system_type
*fs_type
, char *name
,
2268 const struct super_operations
*ops
,
2269 const struct dentry_operations
*dops
, unsigned long magic
)
2271 return mount_pseudo_xattr(fs_type
, name
, ops
, NULL
, dops
, magic
);
2274 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2275 #define fops_get(fops) \
2276 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2277 #define fops_put(fops) \
2278 do { if (fops) module_put((fops)->owner); } while(0)
2280 * This one is to be used *ONLY* from ->open() instances.
2281 * fops must be non-NULL, pinned down *and* module dependencies
2282 * should be sufficient to pin the caller down as well.
2284 #define replace_fops(f, fops) \
2286 struct file *__file = (f); \
2287 fops_put(__file->f_op); \
2288 BUG_ON(!(__file->f_op = (fops))); \
2291 extern int register_filesystem(struct file_system_type
*);
2292 extern int unregister_filesystem(struct file_system_type
*);
2293 extern struct vfsmount
*kern_mount(struct file_system_type
*);
2294 extern void kern_unmount(struct vfsmount
*mnt
);
2295 extern int may_umount_tree(struct vfsmount
*);
2296 extern int may_umount(struct vfsmount
*);
2297 extern long do_mount(const char *, const char __user
*,
2298 const char *, unsigned long, void *);
2299 extern struct vfsmount
*collect_mounts(const struct path
*);
2300 extern void drop_collected_mounts(struct vfsmount
*);
2301 extern int iterate_mounts(int (*)(struct vfsmount
*, void *), void *,
2303 extern int vfs_statfs(const struct path
*, struct kstatfs
*);
2304 extern int user_statfs(const char __user
*, struct kstatfs
*);
2305 extern int fd_statfs(int, struct kstatfs
*);
2306 extern int freeze_super(struct super_block
*super
);
2307 extern int thaw_super(struct super_block
*super
);
2308 extern bool our_mnt(struct vfsmount
*mnt
);
2309 extern __printf(2, 3)
2310 int super_setup_bdi_name(struct super_block
*sb
, char *fmt
, ...);
2311 extern int super_setup_bdi(struct super_block
*sb
);
2313 extern int current_umask(void);
2315 extern void ihold(struct inode
* inode
);
2316 extern void iput(struct inode
*);
2317 extern int generic_update_time(struct inode
*, struct timespec64
*, int);
2320 extern struct kobject
*fs_kobj
;
2322 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2324 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2325 extern int locks_mandatory_locked(struct file
*);
2326 extern int locks_mandatory_area(struct inode
*, struct file
*, loff_t
, loff_t
, unsigned char);
2329 * Candidates for mandatory locking have the setgid bit set
2330 * but no group execute bit - an otherwise meaningless combination.
2333 static inline int __mandatory_lock(struct inode
*ino
)
2335 return (ino
->i_mode
& (S_ISGID
| S_IXGRP
)) == S_ISGID
;
2339 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2340 * otherwise these will be advisory locks
2343 static inline int mandatory_lock(struct inode
*ino
)
2345 return IS_MANDLOCK(ino
) && __mandatory_lock(ino
);
2348 static inline int locks_verify_locked(struct file
*file
)
2350 if (mandatory_lock(locks_inode(file
)))
2351 return locks_mandatory_locked(file
);
2355 static inline int locks_verify_truncate(struct inode
*inode
,
2359 if (!inode
->i_flctx
|| !mandatory_lock(inode
))
2362 if (size
< inode
->i_size
) {
2363 return locks_mandatory_area(inode
, f
, size
, inode
->i_size
- 1,
2366 return locks_mandatory_area(inode
, f
, inode
->i_size
, size
- 1,
2371 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2373 static inline int locks_mandatory_locked(struct file
*file
)
2378 static inline int locks_mandatory_area(struct inode
*inode
, struct file
*filp
,
2379 loff_t start
, loff_t end
, unsigned char type
)
2384 static inline int __mandatory_lock(struct inode
*inode
)
2389 static inline int mandatory_lock(struct inode
*inode
)
2394 static inline int locks_verify_locked(struct file
*file
)
2399 static inline int locks_verify_truncate(struct inode
*inode
, struct file
*filp
,
2405 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2408 #ifdef CONFIG_FILE_LOCKING
2409 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2412 * Since this check is lockless, we must ensure that any refcounts
2413 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2414 * could end up racing with tasks trying to set a new lease on this
2418 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2419 return __break_lease(inode
, mode
, FL_LEASE
);
2423 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2426 * Since this check is lockless, we must ensure that any refcounts
2427 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2428 * could end up racing with tasks trying to set a new lease on this
2432 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2433 return __break_lease(inode
, mode
, FL_DELEG
);
2437 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2441 ret
= break_deleg(inode
, O_WRONLY
|O_NONBLOCK
);
2442 if (ret
== -EWOULDBLOCK
&& delegated_inode
) {
2443 *delegated_inode
= inode
;
2449 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2453 ret
= break_deleg(*delegated_inode
, O_WRONLY
);
2454 iput(*delegated_inode
);
2455 *delegated_inode
= NULL
;
2459 static inline int break_layout(struct inode
*inode
, bool wait
)
2462 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2463 return __break_lease(inode
,
2464 wait
? O_WRONLY
: O_WRONLY
| O_NONBLOCK
,
2469 #else /* !CONFIG_FILE_LOCKING */
2470 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2475 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2480 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2485 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2491 static inline int break_layout(struct inode
*inode
, bool wait
)
2496 #endif /* CONFIG_FILE_LOCKING */
2501 const char *name
; /* pointer to actual string */
2502 const __user
char *uptr
; /* original userland pointer */
2504 struct audit_names
*aname
;
2507 static_assert(offsetof(struct filename
, iname
) % sizeof(long) == 0);
2509 extern long vfs_truncate(const struct path
*, loff_t
);
2510 extern int do_truncate(struct dentry
*, loff_t start
, unsigned int time_attrs
,
2512 extern int vfs_fallocate(struct file
*file
, int mode
, loff_t offset
,
2514 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
2516 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
2517 extern struct file
*filp_open(const char *, int, umode_t
);
2518 extern struct file
*file_open_root(struct dentry
*, struct vfsmount
*,
2519 const char *, int, umode_t
);
2520 extern struct file
* dentry_open(const struct path
*, int, const struct cred
*);
2521 extern struct file
* open_with_fake_path(const struct path
*, int,
2522 struct inode
*, const struct cred
*);
2523 static inline struct file
*file_clone_open(struct file
*file
)
2525 return dentry_open(&file
->f_path
, file
->f_flags
, file
->f_cred
);
2527 extern int filp_close(struct file
*, fl_owner_t id
);
2529 extern struct filename
*getname_flags(const char __user
*, int, int *);
2530 extern struct filename
*getname(const char __user
*);
2531 extern struct filename
*getname_kernel(const char *);
2532 extern void putname(struct filename
*name
);
2534 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
2535 int (*open
)(struct inode
*, struct file
*));
2536 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
2540 extern int ioctl_preallocate(struct file
*filp
, void __user
*argp
);
2543 extern void __init
vfs_caches_init_early(void);
2544 extern void __init
vfs_caches_init(void);
2546 extern struct kmem_cache
*names_cachep
;
2548 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2549 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2552 extern int register_blkdev(unsigned int, const char *);
2553 extern void unregister_blkdev(unsigned int, const char *);
2554 extern void bdev_unhash_inode(dev_t dev
);
2555 extern struct block_device
*bdget(dev_t
);
2556 extern struct block_device
*bdgrab(struct block_device
*bdev
);
2557 extern void bd_set_size(struct block_device
*, loff_t size
);
2558 extern void bd_forget(struct inode
*inode
);
2559 extern void bdput(struct block_device
*);
2560 extern void invalidate_bdev(struct block_device
*);
2561 extern void iterate_bdevs(void (*)(struct block_device
*, void *), void *);
2562 extern int sync_blockdev(struct block_device
*bdev
);
2563 extern void kill_bdev(struct block_device
*);
2564 extern struct super_block
*freeze_bdev(struct block_device
*);
2565 extern void emergency_thaw_all(void);
2566 extern void emergency_thaw_bdev(struct super_block
*sb
);
2567 extern int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
);
2568 extern int fsync_bdev(struct block_device
*);
2570 extern struct super_block
*blockdev_superblock
;
2572 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2574 return sb
== blockdev_superblock
;
2577 static inline void bd_forget(struct inode
*inode
) {}
2578 static inline int sync_blockdev(struct block_device
*bdev
) { return 0; }
2579 static inline void kill_bdev(struct block_device
*bdev
) {}
2580 static inline void invalidate_bdev(struct block_device
*bdev
) {}
2582 static inline struct super_block
*freeze_bdev(struct block_device
*sb
)
2587 static inline int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
)
2592 static inline int emergency_thaw_bdev(struct super_block
*sb
)
2597 static inline void iterate_bdevs(void (*f
)(struct block_device
*, void *), void *arg
)
2601 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2606 extern int sync_filesystem(struct super_block
*);
2607 extern const struct file_operations def_blk_fops
;
2608 extern const struct file_operations def_chr_fops
;
2610 extern int ioctl_by_bdev(struct block_device
*, unsigned, unsigned long);
2611 extern int blkdev_ioctl(struct block_device
*, fmode_t
, unsigned, unsigned long);
2612 extern long compat_blkdev_ioctl(struct file
*, unsigned, unsigned long);
2613 extern int blkdev_get(struct block_device
*bdev
, fmode_t mode
, void *holder
);
2614 extern struct block_device
*blkdev_get_by_path(const char *path
, fmode_t mode
,
2616 extern struct block_device
*blkdev_get_by_dev(dev_t dev
, fmode_t mode
,
2618 extern void blkdev_put(struct block_device
*bdev
, fmode_t mode
);
2619 extern int __blkdev_reread_part(struct block_device
*bdev
);
2620 extern int blkdev_reread_part(struct block_device
*bdev
);
2623 extern int bd_link_disk_holder(struct block_device
*bdev
, struct gendisk
*disk
);
2624 extern void bd_unlink_disk_holder(struct block_device
*bdev
,
2625 struct gendisk
*disk
);
2627 static inline int bd_link_disk_holder(struct block_device
*bdev
,
2628 struct gendisk
*disk
)
2632 static inline void bd_unlink_disk_holder(struct block_device
*bdev
,
2633 struct gendisk
*disk
)
2640 #define CHRDEV_MAJOR_MAX 512
2641 /* Marks the bottom of the first segment of free char majors */
2642 #define CHRDEV_MAJOR_DYN_END 234
2643 /* Marks the top and bottom of the second segment of free char majors */
2644 #define CHRDEV_MAJOR_DYN_EXT_START 511
2645 #define CHRDEV_MAJOR_DYN_EXT_END 384
2647 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
2648 extern int register_chrdev_region(dev_t
, unsigned, const char *);
2649 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
2650 unsigned int count
, const char *name
,
2651 const struct file_operations
*fops
);
2652 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
2653 unsigned int count
, const char *name
);
2654 extern void unregister_chrdev_region(dev_t
, unsigned);
2655 extern void chrdev_show(struct seq_file
*,off_t
);
2657 static inline int register_chrdev(unsigned int major
, const char *name
,
2658 const struct file_operations
*fops
)
2660 return __register_chrdev(major
, 0, 256, name
, fops
);
2663 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2665 __unregister_chrdev(major
, 0, 256, name
);
2668 /* fs/block_dev.c */
2669 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2670 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2673 #define BLKDEV_MAJOR_MAX 512
2674 extern const char *__bdevname(dev_t
, char *buffer
);
2675 extern const char *bdevname(struct block_device
*bdev
, char *buffer
);
2676 extern struct block_device
*lookup_bdev(const char *);
2677 extern void blkdev_show(struct seq_file
*,off_t
);
2680 #define BLKDEV_MAJOR_MAX 0
2683 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2685 /* Invalid inode operations -- fs/bad_inode.c */
2686 extern void make_bad_inode(struct inode
*);
2687 extern bool is_bad_inode(struct inode
*);
2690 extern void check_disk_size_change(struct gendisk
*disk
,
2691 struct block_device
*bdev
, bool verbose
);
2692 extern int revalidate_disk(struct gendisk
*);
2693 extern int check_disk_change(struct block_device
*);
2694 extern int __invalidate_device(struct block_device
*, bool);
2695 extern int invalidate_partition(struct gendisk
*, int);
2697 unsigned long invalidate_mapping_pages(struct address_space
*mapping
,
2698 pgoff_t start
, pgoff_t end
);
2700 static inline void invalidate_remote_inode(struct inode
*inode
)
2702 if (S_ISREG(inode
->i_mode
) || S_ISDIR(inode
->i_mode
) ||
2703 S_ISLNK(inode
->i_mode
))
2704 invalidate_mapping_pages(inode
->i_mapping
, 0, -1);
2706 extern int invalidate_inode_pages2(struct address_space
*mapping
);
2707 extern int invalidate_inode_pages2_range(struct address_space
*mapping
,
2708 pgoff_t start
, pgoff_t end
);
2709 extern int write_inode_now(struct inode
*, int);
2710 extern int filemap_fdatawrite(struct address_space
*);
2711 extern int filemap_flush(struct address_space
*);
2712 extern int filemap_fdatawait_keep_errors(struct address_space
*mapping
);
2713 extern int filemap_fdatawait_range(struct address_space
*, loff_t lstart
,
2716 static inline int filemap_fdatawait(struct address_space
*mapping
)
2718 return filemap_fdatawait_range(mapping
, 0, LLONG_MAX
);
2721 extern bool filemap_range_has_page(struct address_space
*, loff_t lstart
,
2723 extern int filemap_write_and_wait(struct address_space
*mapping
);
2724 extern int filemap_write_and_wait_range(struct address_space
*mapping
,
2725 loff_t lstart
, loff_t lend
);
2726 extern int __filemap_fdatawrite_range(struct address_space
*mapping
,
2727 loff_t start
, loff_t end
, int sync_mode
);
2728 extern int filemap_fdatawrite_range(struct address_space
*mapping
,
2729 loff_t start
, loff_t end
);
2730 extern int filemap_check_errors(struct address_space
*mapping
);
2731 extern void __filemap_set_wb_err(struct address_space
*mapping
, int err
);
2733 extern int __must_check
file_fdatawait_range(struct file
*file
, loff_t lstart
,
2735 extern int __must_check
file_check_and_advance_wb_err(struct file
*file
);
2736 extern int __must_check
file_write_and_wait_range(struct file
*file
,
2737 loff_t start
, loff_t end
);
2739 static inline int file_write_and_wait(struct file
*file
)
2741 return file_write_and_wait_range(file
, 0, LLONG_MAX
);
2745 * filemap_set_wb_err - set a writeback error on an address_space
2746 * @mapping: mapping in which to set writeback error
2747 * @err: error to be set in mapping
2749 * When writeback fails in some way, we must record that error so that
2750 * userspace can be informed when fsync and the like are called. We endeavor
2751 * to report errors on any file that was open at the time of the error. Some
2752 * internal callers also need to know when writeback errors have occurred.
2754 * When a writeback error occurs, most filesystems will want to call
2755 * filemap_set_wb_err to record the error in the mapping so that it will be
2756 * automatically reported whenever fsync is called on the file.
2758 static inline void filemap_set_wb_err(struct address_space
*mapping
, int err
)
2760 /* Fastpath for common case of no error */
2762 __filemap_set_wb_err(mapping
, err
);
2766 * filemap_check_wb_error - has an error occurred since the mark was sampled?
2767 * @mapping: mapping to check for writeback errors
2768 * @since: previously-sampled errseq_t
2770 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2771 * the given value was sampled.
2773 * If it has then report the latest error set, otherwise return 0.
2775 static inline int filemap_check_wb_err(struct address_space
*mapping
,
2778 return errseq_check(&mapping
->wb_err
, since
);
2782 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2783 * @mapping: mapping to be sampled
2785 * Writeback errors are always reported relative to a particular sample point
2786 * in the past. This function provides those sample points.
2788 static inline errseq_t
filemap_sample_wb_err(struct address_space
*mapping
)
2790 return errseq_sample(&mapping
->wb_err
);
2793 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
2795 extern int vfs_fsync(struct file
*file
, int datasync
);
2797 extern int sync_file_range(struct file
*file
, loff_t offset
, loff_t nbytes
,
2798 unsigned int flags
);
2801 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2802 * to already be updated for the write, and will return either the amount
2803 * of bytes passed in, or an error if syncing the file failed.
2805 static inline ssize_t
generic_write_sync(struct kiocb
*iocb
, ssize_t count
)
2807 if (iocb
->ki_flags
& IOCB_DSYNC
) {
2808 int ret
= vfs_fsync_range(iocb
->ki_filp
,
2809 iocb
->ki_pos
- count
, iocb
->ki_pos
- 1,
2810 (iocb
->ki_flags
& IOCB_SYNC
) ? 0 : 1);
2818 extern void emergency_sync(void);
2819 extern void emergency_remount(void);
2821 extern sector_t
bmap(struct inode
*, sector_t
);
2823 extern int notify_change(struct dentry
*, struct iattr
*, struct inode
**);
2824 extern int inode_permission(struct inode
*, int);
2825 extern int generic_permission(struct inode
*, int);
2826 extern int __check_sticky(struct inode
*dir
, struct inode
*inode
);
2828 static inline bool execute_ok(struct inode
*inode
)
2830 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
2833 static inline void file_start_write(struct file
*file
)
2835 if (!S_ISREG(file_inode(file
)->i_mode
))
2837 __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, true);
2840 static inline bool file_start_write_trylock(struct file
*file
)
2842 if (!S_ISREG(file_inode(file
)->i_mode
))
2844 return __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, false);
2847 static inline void file_end_write(struct file
*file
)
2849 if (!S_ISREG(file_inode(file
)->i_mode
))
2851 __sb_end_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
);
2855 * get_write_access() gets write permission for a file.
2856 * put_write_access() releases this write permission.
2857 * This is used for regular files.
2858 * We cannot support write (and maybe mmap read-write shared) accesses and
2859 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2860 * can have the following values:
2861 * 0: no writers, no VM_DENYWRITE mappings
2862 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2863 * > 0: (i_writecount) users are writing to the file.
2865 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2866 * except for the cases where we don't hold i_writecount yet. Then we need to
2867 * use {get,deny}_write_access() - these functions check the sign and refuse
2868 * to do the change if sign is wrong.
2870 static inline int get_write_access(struct inode
*inode
)
2872 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2874 static inline int deny_write_access(struct file
*file
)
2876 struct inode
*inode
= file_inode(file
);
2877 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2879 static inline void put_write_access(struct inode
* inode
)
2881 atomic_dec(&inode
->i_writecount
);
2883 static inline void allow_write_access(struct file
*file
)
2886 atomic_inc(&file_inode(file
)->i_writecount
);
2888 static inline bool inode_is_open_for_write(const struct inode
*inode
)
2890 return atomic_read(&inode
->i_writecount
) > 0;
2893 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
2894 static inline void i_readcount_dec(struct inode
*inode
)
2896 BUG_ON(!atomic_read(&inode
->i_readcount
));
2897 atomic_dec(&inode
->i_readcount
);
2899 static inline void i_readcount_inc(struct inode
*inode
)
2901 atomic_inc(&inode
->i_readcount
);
2904 static inline void i_readcount_dec(struct inode
*inode
)
2908 static inline void i_readcount_inc(struct inode
*inode
)
2913 extern int do_pipe_flags(int *, int);
2915 #define __kernel_read_file_id(id) \
2916 id(UNKNOWN, unknown) \
2917 id(FIRMWARE, firmware) \
2918 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2919 id(MODULE, kernel-module) \
2920 id(KEXEC_IMAGE, kexec-image) \
2921 id(KEXEC_INITRAMFS, kexec-initramfs) \
2922 id(POLICY, security-policy) \
2923 id(X509_CERTIFICATE, x509-certificate) \
2926 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2927 #define __fid_stringify(dummy, str) #str,
2929 enum kernel_read_file_id
{
2930 __kernel_read_file_id(__fid_enumify
)
2933 static const char * const kernel_read_file_str
[] = {
2934 __kernel_read_file_id(__fid_stringify
)
2937 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id
)
2939 if ((unsigned)id
>= READING_MAX_ID
)
2940 return kernel_read_file_str
[READING_UNKNOWN
];
2942 return kernel_read_file_str
[id
];
2945 extern int kernel_read_file(struct file
*, void **, loff_t
*, loff_t
,
2946 enum kernel_read_file_id
);
2947 extern int kernel_read_file_from_path(const char *, void **, loff_t
*, loff_t
,
2948 enum kernel_read_file_id
);
2949 extern int kernel_read_file_from_fd(int, void **, loff_t
*, loff_t
,
2950 enum kernel_read_file_id
);
2951 extern ssize_t
kernel_read(struct file
*, void *, size_t, loff_t
*);
2952 extern ssize_t
kernel_write(struct file
*, const void *, size_t, loff_t
*);
2953 extern ssize_t
__kernel_write(struct file
*, const void *, size_t, loff_t
*);
2954 extern struct file
* open_exec(const char *);
2956 /* fs/dcache.c -- generic fs support functions */
2957 extern bool is_subdir(struct dentry
*, struct dentry
*);
2958 extern bool path_is_under(const struct path
*, const struct path
*);
2960 extern char *file_path(struct file
*, char *, int);
2962 #include <linux/err.h>
2964 /* needed for stackable file system support */
2965 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int whence
);
2967 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int whence
);
2969 extern int inode_init_always(struct super_block
*, struct inode
*);
2970 extern void inode_init_once(struct inode
*);
2971 extern void address_space_init_once(struct address_space
*mapping
);
2972 extern struct inode
* igrab(struct inode
*);
2973 extern ino_t
iunique(struct super_block
*, ino_t
);
2974 extern int inode_needs_sync(struct inode
*inode
);
2975 extern int generic_delete_inode(struct inode
*inode
);
2976 static inline int generic_drop_inode(struct inode
*inode
)
2978 return !inode
->i_nlink
|| inode_unhashed(inode
);
2981 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
2982 unsigned long hashval
, int (*test
)(struct inode
*, void *),
2984 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
2985 int (*test
)(struct inode
*, void *), void *data
);
2986 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
2988 extern struct inode
*inode_insert5(struct inode
*inode
, unsigned long hashval
,
2989 int (*test
)(struct inode
*, void *),
2990 int (*set
)(struct inode
*, void *),
2992 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
2993 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
2994 extern struct inode
*find_inode_nowait(struct super_block
*,
2996 int (*match
)(struct inode
*,
2997 unsigned long, void *),
2999 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
3000 extern int insert_inode_locked(struct inode
*);
3001 #ifdef CONFIG_DEBUG_LOCK_ALLOC
3002 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
3004 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
3006 extern void unlock_new_inode(struct inode
*);
3007 extern void discard_new_inode(struct inode
*);
3008 extern unsigned int get_next_ino(void);
3009 extern void evict_inodes(struct super_block
*sb
);
3011 extern void __iget(struct inode
* inode
);
3012 extern void iget_failed(struct inode
*);
3013 extern void clear_inode(struct inode
*);
3014 extern void __destroy_inode(struct inode
*);
3015 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
3016 extern struct inode
*new_inode(struct super_block
*sb
);
3017 extern void free_inode_nonrcu(struct inode
*inode
);
3018 extern int should_remove_suid(struct dentry
*);
3019 extern int file_remove_privs(struct file
*);
3021 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
3022 static inline void insert_inode_hash(struct inode
*inode
)
3024 __insert_inode_hash(inode
, inode
->i_ino
);
3027 extern void __remove_inode_hash(struct inode
*);
3028 static inline void remove_inode_hash(struct inode
*inode
)
3030 if (!inode_unhashed(inode
) && !hlist_fake(&inode
->i_hash
))
3031 __remove_inode_hash(inode
);
3034 extern void inode_sb_list_add(struct inode
*inode
);
3037 extern int bdev_read_only(struct block_device
*);
3039 extern int set_blocksize(struct block_device
*, int);
3040 extern int sb_set_blocksize(struct super_block
*, int);
3041 extern int sb_min_blocksize(struct super_block
*, int);
3043 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
3044 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
3045 extern ssize_t
generic_write_checks(struct kiocb
*, struct iov_iter
*);
3046 extern int generic_remap_checks(struct file
*file_in
, loff_t pos_in
,
3047 struct file
*file_out
, loff_t pos_out
,
3048 loff_t
*count
, unsigned int remap_flags
);
3049 extern ssize_t
generic_file_read_iter(struct kiocb
*, struct iov_iter
*);
3050 extern ssize_t
__generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
3051 extern ssize_t
generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
3052 extern ssize_t
generic_file_direct_write(struct kiocb
*, struct iov_iter
*);
3053 extern ssize_t
generic_perform_write(struct file
*, struct iov_iter
*, loff_t
);
3055 ssize_t
vfs_iter_read(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
3057 ssize_t
vfs_iter_write(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
3060 /* fs/block_dev.c */
3061 extern ssize_t
blkdev_read_iter(struct kiocb
*iocb
, struct iov_iter
*to
);
3062 extern ssize_t
blkdev_write_iter(struct kiocb
*iocb
, struct iov_iter
*from
);
3063 extern int blkdev_fsync(struct file
*filp
, loff_t start
, loff_t end
,
3065 extern void block_sync_page(struct page
*page
);
3068 extern ssize_t
generic_file_splice_read(struct file
*, loff_t
*,
3069 struct pipe_inode_info
*, size_t, unsigned int);
3070 extern ssize_t
iter_file_splice_write(struct pipe_inode_info
*,
3071 struct file
*, loff_t
*, size_t, unsigned int);
3072 extern ssize_t
generic_splice_sendpage(struct pipe_inode_info
*pipe
,
3073 struct file
*out
, loff_t
*, size_t len
, unsigned int flags
);
3074 extern long do_splice_direct(struct file
*in
, loff_t
*ppos
, struct file
*out
,
3075 loff_t
*opos
, size_t len
, unsigned int flags
);
3079 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
3080 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int whence
);
3081 extern loff_t
no_llseek(struct file
*file
, loff_t offset
, int whence
);
3082 extern loff_t
vfs_setpos(struct file
*file
, loff_t offset
, loff_t maxsize
);
3083 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int whence
);
3084 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
3085 int whence
, loff_t maxsize
, loff_t eof
);
3086 extern loff_t
fixed_size_llseek(struct file
*file
, loff_t offset
,
3087 int whence
, loff_t size
);
3088 extern loff_t
no_seek_end_llseek_size(struct file
*, loff_t
, int, loff_t
);
3089 extern loff_t
no_seek_end_llseek(struct file
*, loff_t
, int);
3090 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
3091 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
3092 extern int stream_open(struct inode
* inode
, struct file
* filp
);
3095 typedef void (dio_submit_t
)(struct bio
*bio
, struct inode
*inode
,
3096 loff_t file_offset
);
3099 /* need locking between buffered and direct access */
3102 /* filesystem does not support filling holes */
3103 DIO_SKIP_HOLES
= 0x02,
3106 void dio_end_io(struct bio
*bio
);
3107 void dio_warn_stale_pagecache(struct file
*filp
);
3109 ssize_t
__blockdev_direct_IO(struct kiocb
*iocb
, struct inode
*inode
,
3110 struct block_device
*bdev
, struct iov_iter
*iter
,
3111 get_block_t get_block
,
3112 dio_iodone_t end_io
, dio_submit_t submit_io
,
3115 static inline ssize_t
blockdev_direct_IO(struct kiocb
*iocb
,
3116 struct inode
*inode
,
3117 struct iov_iter
*iter
,
3118 get_block_t get_block
)
3120 return __blockdev_direct_IO(iocb
, inode
, inode
->i_sb
->s_bdev
, iter
,
3121 get_block
, NULL
, NULL
, DIO_LOCKING
| DIO_SKIP_HOLES
);
3125 void inode_dio_wait(struct inode
*inode
);
3128 * inode_dio_begin - signal start of a direct I/O requests
3129 * @inode: inode the direct I/O happens on
3131 * This is called once we've finished processing a direct I/O request,
3132 * and is used to wake up callers waiting for direct I/O to be quiesced.
3134 static inline void inode_dio_begin(struct inode
*inode
)
3136 atomic_inc(&inode
->i_dio_count
);
3140 * inode_dio_end - signal finish of a direct I/O requests
3141 * @inode: inode the direct I/O happens on
3143 * This is called once we've finished processing a direct I/O request,
3144 * and is used to wake up callers waiting for direct I/O to be quiesced.
3146 static inline void inode_dio_end(struct inode
*inode
)
3148 if (atomic_dec_and_test(&inode
->i_dio_count
))
3149 wake_up_bit(&inode
->i_state
, __I_DIO_WAKEUP
);
3152 extern void inode_set_flags(struct inode
*inode
, unsigned int flags
,
3155 extern const struct file_operations generic_ro_fops
;
3157 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3159 extern int readlink_copy(char __user
*, int, const char *);
3160 extern int page_readlink(struct dentry
*, char __user
*, int);
3161 extern const char *page_get_link(struct dentry
*, struct inode
*,
3162 struct delayed_call
*);
3163 extern void page_put_link(void *);
3164 extern int __page_symlink(struct inode
*inode
, const char *symname
, int len
,
3166 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
3167 extern const struct inode_operations page_symlink_inode_operations
;
3168 extern void kfree_link(void *);
3169 extern void generic_fillattr(struct inode
*, struct kstat
*);
3170 extern int vfs_getattr_nosec(const struct path
*, struct kstat
*, u32
, unsigned int);
3171 extern int vfs_getattr(const struct path
*, struct kstat
*, u32
, unsigned int);
3172 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
3173 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
3174 void __inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
3175 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
3176 static inline loff_t
__inode_get_bytes(struct inode
*inode
)
3178 return (((loff_t
)inode
->i_blocks
) << 9) + inode
->i_bytes
;
3180 loff_t
inode_get_bytes(struct inode
*inode
);
3181 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
3182 const char *simple_get_link(struct dentry
*, struct inode
*,
3183 struct delayed_call
*);
3184 extern const struct inode_operations simple_symlink_inode_operations
;
3186 extern int iterate_dir(struct file
*, struct dir_context
*);
3188 extern int vfs_statx(int, const char __user
*, int, struct kstat
*, u32
);
3189 extern int vfs_statx_fd(unsigned int, struct kstat
*, u32
, unsigned int);
3191 static inline int vfs_stat(const char __user
*filename
, struct kstat
*stat
)
3193 return vfs_statx(AT_FDCWD
, filename
, AT_NO_AUTOMOUNT
,
3194 stat
, STATX_BASIC_STATS
);
3196 static inline int vfs_lstat(const char __user
*name
, struct kstat
*stat
)
3198 return vfs_statx(AT_FDCWD
, name
, AT_SYMLINK_NOFOLLOW
| AT_NO_AUTOMOUNT
,
3199 stat
, STATX_BASIC_STATS
);
3201 static inline int vfs_fstatat(int dfd
, const char __user
*filename
,
3202 struct kstat
*stat
, int flags
)
3204 return vfs_statx(dfd
, filename
, flags
| AT_NO_AUTOMOUNT
,
3205 stat
, STATX_BASIC_STATS
);
3207 static inline int vfs_fstat(int fd
, struct kstat
*stat
)
3209 return vfs_statx_fd(fd
, stat
, STATX_BASIC_STATS
, 0);
3213 extern const char *vfs_get_link(struct dentry
*, struct delayed_call
*);
3214 extern int vfs_readlink(struct dentry
*, char __user
*, int);
3216 extern int __generic_block_fiemap(struct inode
*inode
,
3217 struct fiemap_extent_info
*fieinfo
,
3218 loff_t start
, loff_t len
,
3219 get_block_t
*get_block
);
3220 extern int generic_block_fiemap(struct inode
*inode
,
3221 struct fiemap_extent_info
*fieinfo
, u64 start
,
3222 u64 len
, get_block_t
*get_block
);
3224 extern struct file_system_type
*get_filesystem(struct file_system_type
*fs
);
3225 extern void put_filesystem(struct file_system_type
*fs
);
3226 extern struct file_system_type
*get_fs_type(const char *name
);
3227 extern struct super_block
*get_super(struct block_device
*);
3228 extern struct super_block
*get_super_thawed(struct block_device
*);
3229 extern struct super_block
*get_super_exclusive_thawed(struct block_device
*bdev
);
3230 extern struct super_block
*get_active_super(struct block_device
*bdev
);
3231 extern void drop_super(struct super_block
*sb
);
3232 extern void drop_super_exclusive(struct super_block
*sb
);
3233 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
3234 extern void iterate_supers_type(struct file_system_type
*,
3235 void (*)(struct super_block
*, void *), void *);
3237 extern int dcache_dir_open(struct inode
*, struct file
*);
3238 extern int dcache_dir_close(struct inode
*, struct file
*);
3239 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
3240 extern int dcache_readdir(struct file
*, struct dir_context
*);
3241 extern int simple_setattr(struct dentry
*, struct iattr
*);
3242 extern int simple_getattr(const struct path
*, struct kstat
*, u32
, unsigned int);
3243 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
3244 extern int simple_open(struct inode
*inode
, struct file
*file
);
3245 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
3246 extern int simple_unlink(struct inode
*, struct dentry
*);
3247 extern int simple_rmdir(struct inode
*, struct dentry
*);
3248 extern int simple_rename(struct inode
*, struct dentry
*,
3249 struct inode
*, struct dentry
*, unsigned int);
3250 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
3251 extern int noop_set_page_dirty(struct page
*page
);
3252 extern void noop_invalidatepage(struct page
*page
, unsigned int offset
,
3253 unsigned int length
);
3254 extern ssize_t
noop_direct_IO(struct kiocb
*iocb
, struct iov_iter
*iter
);
3255 extern int simple_empty(struct dentry
*);
3256 extern int simple_readpage(struct file
*file
, struct page
*page
);
3257 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
3258 loff_t pos
, unsigned len
, unsigned flags
,
3259 struct page
**pagep
, void **fsdata
);
3260 extern int simple_write_end(struct file
*file
, struct address_space
*mapping
,
3261 loff_t pos
, unsigned len
, unsigned copied
,
3262 struct page
*page
, void *fsdata
);
3263 extern int always_delete_dentry(const struct dentry
*);
3264 extern struct inode
*alloc_anon_inode(struct super_block
*);
3265 extern int simple_nosetlease(struct file
*, long, struct file_lock
**, void **);
3266 extern const struct dentry_operations simple_dentry_operations
;
3268 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
3269 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
3270 extern const struct file_operations simple_dir_operations
;
3271 extern const struct inode_operations simple_dir_inode_operations
;
3272 extern void make_empty_dir_inode(struct inode
*inode
);
3273 extern bool is_empty_dir_inode(struct inode
*inode
);
3274 struct tree_descr
{ const char *name
; const struct file_operations
*ops
; int mode
; };
3275 struct dentry
*d_alloc_name(struct dentry
*, const char *);
3276 extern int simple_fill_super(struct super_block
*, unsigned long,
3277 const struct tree_descr
*);
3278 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
3279 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
3281 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
3282 loff_t
*ppos
, const void *from
, size_t available
);
3283 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
3284 const void __user
*from
, size_t count
);
3286 extern int __generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3287 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3289 extern int generic_check_addressable(unsigned, u64
);
3291 #ifdef CONFIG_MIGRATION
3292 extern int buffer_migrate_page(struct address_space
*,
3293 struct page
*, struct page
*,
3295 extern int buffer_migrate_page_norefs(struct address_space
*,
3296 struct page
*, struct page
*,
3299 #define buffer_migrate_page NULL
3300 #define buffer_migrate_page_norefs NULL
3303 extern int setattr_prepare(struct dentry
*, struct iattr
*);
3304 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
3305 extern void setattr_copy(struct inode
*inode
, const struct iattr
*attr
);
3307 extern int file_update_time(struct file
*file
);
3309 static inline bool io_is_direct(struct file
*filp
)
3311 return (filp
->f_flags
& O_DIRECT
) || IS_DAX(filp
->f_mapping
->host
);
3314 static inline bool vma_is_dax(struct vm_area_struct
*vma
)
3316 return vma
->vm_file
&& IS_DAX(vma
->vm_file
->f_mapping
->host
);
3319 static inline bool vma_is_fsdax(struct vm_area_struct
*vma
)
3321 struct inode
*inode
;
3325 if (!vma_is_dax(vma
))
3327 inode
= file_inode(vma
->vm_file
);
3328 if (S_ISCHR(inode
->i_mode
))
3329 return false; /* device-dax */
3333 static inline int iocb_flags(struct file
*file
)
3336 if (file
->f_flags
& O_APPEND
)
3338 if (io_is_direct(file
))
3340 if ((file
->f_flags
& O_DSYNC
) || IS_SYNC(file
->f_mapping
->host
))
3342 if (file
->f_flags
& __O_SYNC
)
3347 static inline int kiocb_set_rw_flags(struct kiocb
*ki
, rwf_t flags
)
3349 if (unlikely(flags
& ~RWF_SUPPORTED
))
3352 if (flags
& RWF_NOWAIT
) {
3353 if (!(ki
->ki_filp
->f_mode
& FMODE_NOWAIT
))
3355 ki
->ki_flags
|= IOCB_NOWAIT
;
3357 if (flags
& RWF_HIPRI
)
3358 ki
->ki_flags
|= IOCB_HIPRI
;
3359 if (flags
& RWF_DSYNC
)
3360 ki
->ki_flags
|= IOCB_DSYNC
;
3361 if (flags
& RWF_SYNC
)
3362 ki
->ki_flags
|= (IOCB_DSYNC
| IOCB_SYNC
);
3363 if (flags
& RWF_APPEND
)
3364 ki
->ki_flags
|= IOCB_APPEND
;
3368 static inline ino_t
parent_ino(struct dentry
*dentry
)
3373 * Don't strictly need d_lock here? If the parent ino could change
3374 * then surely we'd have a deeper race in the caller?
3376 spin_lock(&dentry
->d_lock
);
3377 res
= dentry
->d_parent
->d_inode
->i_ino
;
3378 spin_unlock(&dentry
->d_lock
);
3382 /* Transaction based IO helpers */
3385 * An argresp is stored in an allocated page and holds the
3386 * size of the argument or response, along with its content
3388 struct simple_transaction_argresp
{
3393 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3395 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
3397 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
3398 size_t size
, loff_t
*pos
);
3399 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
3401 void simple_transaction_set(struct file
*file
, size_t n
);
3404 * simple attribute files
3406 * These attributes behave similar to those in sysfs:
3408 * Writing to an attribute immediately sets a value, an open file can be
3409 * written to multiple times.
3411 * Reading from an attribute creates a buffer from the value that might get
3412 * read with multiple read calls. When the attribute has been read
3413 * completely, no further read calls are possible until the file is opened
3416 * All attributes contain a text representation of a numeric value
3417 * that are accessed with the get() and set() functions.
3419 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3420 static int __fops ## _open(struct inode *inode, struct file *file) \
3422 __simple_attr_check_format(__fmt, 0ull); \
3423 return simple_attr_open(inode, file, __get, __set, __fmt); \
3425 static const struct file_operations __fops = { \
3426 .owner = THIS_MODULE, \
3427 .open = __fops ## _open, \
3428 .release = simple_attr_release, \
3429 .read = simple_attr_read, \
3430 .write = simple_attr_write, \
3431 .llseek = generic_file_llseek, \
3434 static inline __printf(1, 2)
3435 void __simple_attr_check_format(const char *fmt
, ...)
3437 /* don't do anything, just let the compiler check the arguments; */
3440 int simple_attr_open(struct inode
*inode
, struct file
*file
,
3441 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
3443 int simple_attr_release(struct inode
*inode
, struct file
*file
);
3444 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
3445 size_t len
, loff_t
*ppos
);
3446 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
3447 size_t len
, loff_t
*ppos
);
3450 int proc_nr_files(struct ctl_table
*table
, int write
,
3451 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3452 int proc_nr_dentry(struct ctl_table
*table
, int write
,
3453 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3454 int proc_nr_inodes(struct ctl_table
*table
, int write
,
3455 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3456 int __init
get_filesystem_list(char *buf
);
3458 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3459 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3461 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3462 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3463 (flag & __FMODE_NONOTIFY)))
3465 static inline bool is_sxid(umode_t mode
)
3467 return (mode
& S_ISUID
) || ((mode
& S_ISGID
) && (mode
& S_IXGRP
));
3470 static inline int check_sticky(struct inode
*dir
, struct inode
*inode
)
3472 if (!(dir
->i_mode
& S_ISVTX
))
3475 return __check_sticky(dir
, inode
);
3478 static inline void inode_has_no_xattr(struct inode
*inode
)
3480 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& SB_NOSEC
))
3481 inode
->i_flags
|= S_NOSEC
;
3484 static inline bool is_root_inode(struct inode
*inode
)
3486 return inode
== inode
->i_sb
->s_root
->d_inode
;
3489 static inline bool dir_emit(struct dir_context
*ctx
,
3490 const char *name
, int namelen
,
3491 u64 ino
, unsigned type
)
3493 return ctx
->actor(ctx
, name
, namelen
, ctx
->pos
, ino
, type
) == 0;
3495 static inline bool dir_emit_dot(struct file
*file
, struct dir_context
*ctx
)
3497 return ctx
->actor(ctx
, ".", 1, ctx
->pos
,
3498 file
->f_path
.dentry
->d_inode
->i_ino
, DT_DIR
) == 0;
3500 static inline bool dir_emit_dotdot(struct file
*file
, struct dir_context
*ctx
)
3502 return ctx
->actor(ctx
, "..", 2, ctx
->pos
,
3503 parent_ino(file
->f_path
.dentry
), DT_DIR
) == 0;
3505 static inline bool dir_emit_dots(struct file
*file
, struct dir_context
*ctx
)
3507 if (ctx
->pos
== 0) {
3508 if (!dir_emit_dot(file
, ctx
))
3512 if (ctx
->pos
== 1) {
3513 if (!dir_emit_dotdot(file
, ctx
))
3519 static inline bool dir_relax(struct inode
*inode
)
3521 inode_unlock(inode
);
3523 return !IS_DEADDIR(inode
);
3526 static inline bool dir_relax_shared(struct inode
*inode
)
3528 inode_unlock_shared(inode
);
3529 inode_lock_shared(inode
);
3530 return !IS_DEADDIR(inode
);
3533 extern bool path_noexec(const struct path
*path
);
3534 extern void inode_nohighmem(struct inode
*inode
);
3537 extern int vfs_fadvise(struct file
*file
, loff_t offset
, loff_t len
,
3540 #if defined(CONFIG_IO_URING)
3541 extern struct sock
*io_uring_get_socket(struct file
*file
);
3543 static inline struct sock
*io_uring_get_socket(struct file
*file
)
3549 #endif /* _LINUX_FS_H */