1 /* SPDX-License-Identifier: GPL-2.0 */
5 #include <linux/linkage.h>
6 #include <linux/wait_bit.h>
7 #include <linux/kdev_t.h>
8 #include <linux/dcache.h>
9 #include <linux/path.h>
10 #include <linux/stat.h>
11 #include <linux/cache.h>
12 #include <linux/list.h>
13 #include <linux/list_lru.h>
14 #include <linux/llist.h>
15 #include <linux/radix-tree.h>
16 #include <linux/xarray.h>
17 #include <linux/rbtree.h>
18 #include <linux/init.h>
19 #include <linux/pid.h>
20 #include <linux/bug.h>
21 #include <linux/mutex.h>
22 #include <linux/rwsem.h>
23 #include <linux/mm_types.h>
24 #include <linux/capability.h>
25 #include <linux/semaphore.h>
26 #include <linux/fcntl.h>
27 #include <linux/fiemap.h>
28 #include <linux/rculist_bl.h>
29 #include <linux/atomic.h>
30 #include <linux/shrinker.h>
31 #include <linux/migrate_mode.h>
32 #include <linux/uidgid.h>
33 #include <linux/lockdep.h>
34 #include <linux/percpu-rwsem.h>
35 #include <linux/workqueue.h>
36 #include <linux/delayed_call.h>
37 #include <linux/uuid.h>
38 #include <linux/errseq.h>
39 #include <linux/ioprio.h>
40 #include <linux/fs_types.h>
41 #include <linux/build_bug.h>
42 #include <linux/stddef.h>
44 #include <asm/byteorder.h>
45 #include <uapi/linux/fs.h>
47 struct backing_dev_info
;
50 struct export_operations
;
55 struct pipe_inode_info
;
56 struct poll_table_struct
;
58 struct vm_area_struct
;
61 struct swap_info_struct
;
63 struct workqueue_struct
;
66 struct fscrypt_operations
;
68 struct fsverity_operations
;
70 struct fs_parameter_description
;
72 extern void __init
inode_init(void);
73 extern void __init
inode_init_early(void);
74 extern void __init
files_init(void);
75 extern void __init
files_maxfiles_init(void);
77 extern struct files_stat_struct files_stat
;
78 extern unsigned long get_max_files(void);
79 extern unsigned int sysctl_nr_open
;
80 extern struct inodes_stat_t inodes_stat
;
81 extern int leases_enable
, lease_break_time
;
82 extern int sysctl_protected_symlinks
;
83 extern int sysctl_protected_hardlinks
;
84 extern int sysctl_protected_fifos
;
85 extern int sysctl_protected_regular
;
87 typedef __kernel_rwf_t rwf_t
;
90 typedef int (get_block_t
)(struct inode
*inode
, sector_t iblock
,
91 struct buffer_head
*bh_result
, int create
);
92 typedef int (dio_iodone_t
)(struct kiocb
*iocb
, loff_t offset
,
93 ssize_t bytes
, void *private);
95 #define MAY_EXEC 0x00000001
96 #define MAY_WRITE 0x00000002
97 #define MAY_READ 0x00000004
98 #define MAY_APPEND 0x00000008
99 #define MAY_ACCESS 0x00000010
100 #define MAY_OPEN 0x00000020
101 #define MAY_CHDIR 0x00000040
102 /* called from RCU mode, don't block */
103 #define MAY_NOT_BLOCK 0x00000080
106 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
107 * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
110 /* file is open for reading */
111 #define FMODE_READ ((__force fmode_t)0x1)
112 /* file is open for writing */
113 #define FMODE_WRITE ((__force fmode_t)0x2)
114 /* file is seekable */
115 #define FMODE_LSEEK ((__force fmode_t)0x4)
116 /* file can be accessed using pread */
117 #define FMODE_PREAD ((__force fmode_t)0x8)
118 /* file can be accessed using pwrite */
119 #define FMODE_PWRITE ((__force fmode_t)0x10)
120 /* File is opened for execution with sys_execve / sys_uselib */
121 #define FMODE_EXEC ((__force fmode_t)0x20)
122 /* File is opened with O_NDELAY (only set for block devices) */
123 #define FMODE_NDELAY ((__force fmode_t)0x40)
124 /* File is opened with O_EXCL (only set for block devices) */
125 #define FMODE_EXCL ((__force fmode_t)0x80)
126 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
127 (specialy hack for floppy.c) */
128 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
129 /* 32bit hashes as llseek() offset (for directories) */
130 #define FMODE_32BITHASH ((__force fmode_t)0x200)
131 /* 64bit hashes as llseek() offset (for directories) */
132 #define FMODE_64BITHASH ((__force fmode_t)0x400)
135 * Don't update ctime and mtime.
137 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
138 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
140 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
142 /* Expect random access pattern */
143 #define FMODE_RANDOM ((__force fmode_t)0x1000)
145 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
146 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
148 /* File is opened with O_PATH; almost nothing can be done with it */
149 #define FMODE_PATH ((__force fmode_t)0x4000)
151 /* File needs atomic accesses to f_pos */
152 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
153 /* Write access to underlying fs */
154 #define FMODE_WRITER ((__force fmode_t)0x10000)
155 /* Has read method(s) */
156 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
157 /* Has write method(s) */
158 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
160 #define FMODE_OPENED ((__force fmode_t)0x80000)
161 #define FMODE_CREATED ((__force fmode_t)0x100000)
163 /* File is stream-like */
164 #define FMODE_STREAM ((__force fmode_t)0x200000)
166 /* File was opened by fanotify and shouldn't generate fanotify events */
167 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
169 /* File is capable of returning -EAGAIN if I/O will block */
170 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
172 /* File represents mount that needs unmounting */
173 #define FMODE_NEED_UNMOUNT ((__force fmode_t)0x10000000)
175 /* File does not contribute to nr_files count */
176 #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000)
179 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
180 * that indicates that they should check the contents of the iovec are
181 * valid, but not check the memory that the iovec elements
184 #define CHECK_IOVEC_ONLY -1
187 * Attribute flags. These should be or-ed together to figure out what
190 #define ATTR_MODE (1 << 0)
191 #define ATTR_UID (1 << 1)
192 #define ATTR_GID (1 << 2)
193 #define ATTR_SIZE (1 << 3)
194 #define ATTR_ATIME (1 << 4)
195 #define ATTR_MTIME (1 << 5)
196 #define ATTR_CTIME (1 << 6)
197 #define ATTR_ATIME_SET (1 << 7)
198 #define ATTR_MTIME_SET (1 << 8)
199 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
200 #define ATTR_KILL_SUID (1 << 11)
201 #define ATTR_KILL_SGID (1 << 12)
202 #define ATTR_FILE (1 << 13)
203 #define ATTR_KILL_PRIV (1 << 14)
204 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
205 #define ATTR_TIMES_SET (1 << 16)
206 #define ATTR_TOUCH (1 << 17)
209 * Whiteout is represented by a char device. The following constants define the
210 * mode and device number to use.
212 #define WHITEOUT_MODE 0
213 #define WHITEOUT_DEV 0
216 * This is the Inode Attributes structure, used for notify_change(). It
217 * uses the above definitions as flags, to know which values have changed.
218 * Also, in this manner, a Filesystem can look at only the values it cares
219 * about. Basically, these are the attributes that the VFS layer can
220 * request to change from the FS layer.
222 * Derek Atkins <warlord@MIT.EDU> 94-10-20
225 unsigned int ia_valid
;
230 struct timespec64 ia_atime
;
231 struct timespec64 ia_mtime
;
232 struct timespec64 ia_ctime
;
235 * Not an attribute, but an auxiliary info for filesystems wanting to
236 * implement an ftruncate() like method. NOTE: filesystem should
237 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
239 struct file
*ia_file
;
243 * Includes for diskquotas.
245 #include <linux/quota.h>
248 * Maximum number of layers of fs stack. Needs to be limited to
249 * prevent kernel stack overflow
251 #define FILESYSTEM_MAX_STACK_DEPTH 2
254 * enum positive_aop_returns - aop return codes with specific semantics
256 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
257 * completed, that the page is still locked, and
258 * should be considered active. The VM uses this hint
259 * to return the page to the active list -- it won't
260 * be a candidate for writeback again in the near
261 * future. Other callers must be careful to unlock
262 * the page if they get this return. Returned by
265 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
266 * unlocked it and the page might have been truncated.
267 * The caller should back up to acquiring a new page and
268 * trying again. The aop will be taking reasonable
269 * precautions not to livelock. If the caller held a page
270 * reference, it should drop it before retrying. Returned
273 * address_space_operation functions return these large constants to indicate
274 * special semantics to the caller. These are much larger than the bytes in a
275 * page to allow for functions that return the number of bytes operated on in a
279 enum positive_aop_returns
{
280 AOP_WRITEPAGE_ACTIVATE
= 0x80000,
281 AOP_TRUNCATED_PAGE
= 0x80001,
284 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
285 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
286 * helper code (eg buffer layer)
287 * to clear GFP_FS from alloc */
290 * oh the beauties of C type declarations.
293 struct address_space
;
294 struct writeback_control
;
297 * Write life time hint values.
298 * Stored in struct inode as u8.
301 WRITE_LIFE_NOT_SET
= 0,
302 WRITE_LIFE_NONE
= RWH_WRITE_LIFE_NONE
,
303 WRITE_LIFE_SHORT
= RWH_WRITE_LIFE_SHORT
,
304 WRITE_LIFE_MEDIUM
= RWH_WRITE_LIFE_MEDIUM
,
305 WRITE_LIFE_LONG
= RWH_WRITE_LIFE_LONG
,
306 WRITE_LIFE_EXTREME
= RWH_WRITE_LIFE_EXTREME
,
309 #define IOCB_EVENTFD (1 << 0)
310 #define IOCB_APPEND (1 << 1)
311 #define IOCB_DIRECT (1 << 2)
312 #define IOCB_HIPRI (1 << 3)
313 #define IOCB_DSYNC (1 << 4)
314 #define IOCB_SYNC (1 << 5)
315 #define IOCB_WRITE (1 << 6)
316 #define IOCB_NOWAIT (1 << 7)
319 struct file
*ki_filp
;
321 /* The 'ki_filp' pointer is shared in a union for aio */
322 randomized_struct_fields_start
325 void (*ki_complete
)(struct kiocb
*iocb
, long ret
, long ret2
);
329 u16 ki_ioprio
; /* See linux/ioprio.h */
330 unsigned int ki_cookie
; /* for ->iopoll */
332 randomized_struct_fields_end
335 static inline bool is_sync_kiocb(struct kiocb
*kiocb
)
337 return kiocb
->ki_complete
== NULL
;
341 * "descriptor" for what we're up to with a read.
342 * This allows us to use the same read code yet
343 * have multiple different users of the data that
344 * we read from a file.
346 * The simplest case just copies the data to user
359 typedef int (*read_actor_t
)(read_descriptor_t
*, struct page
*,
360 unsigned long, unsigned long);
362 struct address_space_operations
{
363 int (*writepage
)(struct page
*page
, struct writeback_control
*wbc
);
364 int (*readpage
)(struct file
*, struct page
*);
366 /* Write back some dirty pages from this mapping. */
367 int (*writepages
)(struct address_space
*, struct writeback_control
*);
369 /* Set a page dirty. Return true if this dirtied it */
370 int (*set_page_dirty
)(struct page
*page
);
373 * Reads in the requested pages. Unlike ->readpage(), this is
374 * PURELY used for read-ahead!.
376 int (*readpages
)(struct file
*filp
, struct address_space
*mapping
,
377 struct list_head
*pages
, unsigned nr_pages
);
379 int (*write_begin
)(struct file
*, struct address_space
*mapping
,
380 loff_t pos
, unsigned len
, unsigned flags
,
381 struct page
**pagep
, void **fsdata
);
382 int (*write_end
)(struct file
*, struct address_space
*mapping
,
383 loff_t pos
, unsigned len
, unsigned copied
,
384 struct page
*page
, void *fsdata
);
386 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
387 sector_t (*bmap
)(struct address_space
*, sector_t
);
388 void (*invalidatepage
) (struct page
*, unsigned int, unsigned int);
389 int (*releasepage
) (struct page
*, gfp_t
);
390 void (*freepage
)(struct page
*);
391 ssize_t (*direct_IO
)(struct kiocb
*, struct iov_iter
*iter
);
393 * migrate the contents of a page to the specified target. If
394 * migrate_mode is MIGRATE_ASYNC, it must not block.
396 int (*migratepage
) (struct address_space
*,
397 struct page
*, struct page
*, enum migrate_mode
);
398 bool (*isolate_page
)(struct page
*, isolate_mode_t
);
399 void (*putback_page
)(struct page
*);
400 int (*launder_page
) (struct page
*);
401 int (*is_partially_uptodate
) (struct page
*, unsigned long,
403 void (*is_dirty_writeback
) (struct page
*, bool *, bool *);
404 int (*error_remove_page
)(struct address_space
*, struct page
*);
406 /* swapfile support */
407 int (*swap_activate
)(struct swap_info_struct
*sis
, struct file
*file
,
409 void (*swap_deactivate
)(struct file
*file
);
412 extern const struct address_space_operations empty_aops
;
415 * pagecache_write_begin/pagecache_write_end must be used by general code
416 * to write into the pagecache.
418 int pagecache_write_begin(struct file
*, struct address_space
*mapping
,
419 loff_t pos
, unsigned len
, unsigned flags
,
420 struct page
**pagep
, void **fsdata
);
422 int pagecache_write_end(struct file
*, struct address_space
*mapping
,
423 loff_t pos
, unsigned len
, unsigned copied
,
424 struct page
*page
, void *fsdata
);
427 * struct address_space - Contents of a cacheable, mappable object.
428 * @host: Owner, either the inode or the block_device.
429 * @i_pages: Cached pages.
430 * @gfp_mask: Memory allocation flags to use for allocating pages.
431 * @i_mmap_writable: Number of VM_SHARED mappings.
432 * @nr_thps: Number of THPs in the pagecache (non-shmem only).
433 * @i_mmap: Tree of private and shared mappings.
434 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
435 * @nrpages: Number of page entries, protected by the i_pages lock.
436 * @nrexceptional: Shadow or DAX entries, protected by the i_pages lock.
437 * @writeback_index: Writeback starts here.
439 * @flags: Error bits and flags (AS_*).
440 * @wb_err: The most recent error which has occurred.
441 * @private_lock: For use by the owner of the address_space.
442 * @private_list: For use by the owner of the address_space.
443 * @private_data: For use by the owner of the address_space.
445 struct address_space
{
447 struct xarray i_pages
;
449 atomic_t i_mmap_writable
;
450 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
451 /* number of thp, only for non-shmem files */
454 struct rb_root_cached i_mmap
;
455 struct rw_semaphore i_mmap_rwsem
;
456 unsigned long nrpages
;
457 unsigned long nrexceptional
;
458 pgoff_t writeback_index
;
459 const struct address_space_operations
*a_ops
;
462 spinlock_t private_lock
;
463 struct list_head private_list
;
465 } __attribute__((aligned(sizeof(long)))) __randomize_layout
;
467 * On most architectures that alignment is already the case; but
468 * must be enforced here for CRIS, to let the least significant bit
469 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
471 struct request_queue
;
473 struct block_device
{
474 dev_t bd_dev
; /* not a kdev_t - it's a search key */
476 struct inode
* bd_inode
; /* will die */
477 struct super_block
* bd_super
;
478 struct mutex bd_mutex
; /* open/close mutex */
482 bool bd_write_holder
;
484 struct list_head bd_holder_disks
;
486 struct block_device
* bd_contains
;
487 unsigned bd_block_size
;
489 struct hd_struct
* bd_part
;
490 /* number of times partitions within this device have been opened. */
491 unsigned bd_part_count
;
493 struct gendisk
* bd_disk
;
494 struct request_queue
* bd_queue
;
495 struct backing_dev_info
*bd_bdi
;
496 struct list_head bd_list
;
498 * Private data. You must have bd_claim'ed the block_device
499 * to use this. NOTE: bd_claim allows an owner to claim
500 * the same device multiple times, the owner must take special
501 * care to not mess up bd_private for that case.
503 unsigned long bd_private
;
505 /* The counter of freeze processes */
506 int bd_fsfreeze_count
;
507 /* Mutex for freeze */
508 struct mutex bd_fsfreeze_mutex
;
509 } __randomize_layout
;
511 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
512 #define PAGECACHE_TAG_DIRTY XA_MARK_0
513 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
514 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
517 * Returns true if any of the pages in the mapping are marked with the tag.
519 static inline bool mapping_tagged(struct address_space
*mapping
, xa_mark_t tag
)
521 return xa_marked(&mapping
->i_pages
, tag
);
524 static inline void i_mmap_lock_write(struct address_space
*mapping
)
526 down_write(&mapping
->i_mmap_rwsem
);
529 static inline void i_mmap_unlock_write(struct address_space
*mapping
)
531 up_write(&mapping
->i_mmap_rwsem
);
534 static inline void i_mmap_lock_read(struct address_space
*mapping
)
536 down_read(&mapping
->i_mmap_rwsem
);
539 static inline void i_mmap_unlock_read(struct address_space
*mapping
)
541 up_read(&mapping
->i_mmap_rwsem
);
545 * Might pages of this file be mapped into userspace?
547 static inline int mapping_mapped(struct address_space
*mapping
)
549 return !RB_EMPTY_ROOT(&mapping
->i_mmap
.rb_root
);
553 * Might pages of this file have been modified in userspace?
554 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
555 * marks vma as VM_SHARED if it is shared, and the file was opened for
556 * writing i.e. vma may be mprotected writable even if now readonly.
558 * If i_mmap_writable is negative, no new writable mappings are allowed. You
559 * can only deny writable mappings, if none exists right now.
561 static inline int mapping_writably_mapped(struct address_space
*mapping
)
563 return atomic_read(&mapping
->i_mmap_writable
) > 0;
566 static inline int mapping_map_writable(struct address_space
*mapping
)
568 return atomic_inc_unless_negative(&mapping
->i_mmap_writable
) ?
572 static inline void mapping_unmap_writable(struct address_space
*mapping
)
574 atomic_dec(&mapping
->i_mmap_writable
);
577 static inline int mapping_deny_writable(struct address_space
*mapping
)
579 return atomic_dec_unless_positive(&mapping
->i_mmap_writable
) ?
583 static inline void mapping_allow_writable(struct address_space
*mapping
)
585 atomic_inc(&mapping
->i_mmap_writable
);
589 * Use sequence counter to get consistent i_size on 32-bit processors.
591 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
592 #include <linux/seqlock.h>
593 #define __NEED_I_SIZE_ORDERED
594 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
596 #define i_size_ordered_init(inode) do { } while (0)
600 #define ACL_NOT_CACHED ((void *)(-1))
601 #define ACL_DONT_CACHE ((void *)(-3))
603 static inline struct posix_acl
*
604 uncached_acl_sentinel(struct task_struct
*task
)
606 return (void *)task
+ 1;
610 is_uncached_acl(struct posix_acl
*acl
)
612 return (long)acl
& 1;
615 #define IOP_FASTPERM 0x0001
616 #define IOP_LOOKUP 0x0002
617 #define IOP_NOFOLLOW 0x0004
618 #define IOP_XATTR 0x0008
619 #define IOP_DEFAULT_READLINK 0x0010
621 struct fsnotify_mark_connector
;
624 * Keep mostly read-only and often accessed (especially for
625 * the RCU path lookup and 'stat' data) fields at the beginning
626 * of the 'struct inode'
630 unsigned short i_opflags
;
633 unsigned int i_flags
;
635 #ifdef CONFIG_FS_POSIX_ACL
636 struct posix_acl
*i_acl
;
637 struct posix_acl
*i_default_acl
;
640 const struct inode_operations
*i_op
;
641 struct super_block
*i_sb
;
642 struct address_space
*i_mapping
;
644 #ifdef CONFIG_SECURITY
648 /* Stat data, not accessed from path walking */
651 * Filesystems may only read i_nlink directly. They shall use the
652 * following functions for modification:
654 * (set|clear|inc|drop)_nlink
655 * inode_(inc|dec)_link_count
658 const unsigned int i_nlink
;
659 unsigned int __i_nlink
;
663 struct timespec64 i_atime
;
664 struct timespec64 i_mtime
;
665 struct timespec64 i_ctime
;
666 spinlock_t i_lock
; /* i_blocks, i_bytes, maybe i_size */
667 unsigned short i_bytes
;
672 #ifdef __NEED_I_SIZE_ORDERED
673 seqcount_t i_size_seqcount
;
677 unsigned long i_state
;
678 struct rw_semaphore i_rwsem
;
680 unsigned long dirtied_when
; /* jiffies of first dirtying */
681 unsigned long dirtied_time_when
;
683 struct hlist_node i_hash
;
684 struct list_head i_io_list
; /* backing dev IO list */
685 #ifdef CONFIG_CGROUP_WRITEBACK
686 struct bdi_writeback
*i_wb
; /* the associated cgroup wb */
688 /* foreign inode detection, see wbc_detach_inode() */
690 u16 i_wb_frn_avg_time
;
691 u16 i_wb_frn_history
;
693 struct list_head i_lru
; /* inode LRU list */
694 struct list_head i_sb_list
;
695 struct list_head i_wb_list
; /* backing dev writeback list */
697 struct hlist_head i_dentry
;
698 struct rcu_head i_rcu
;
700 atomic64_t i_version
;
702 atomic_t i_dio_count
;
703 atomic_t i_writecount
;
704 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
705 atomic_t i_readcount
; /* struct files open RO */
708 const struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
709 void (*free_inode
)(struct inode
*);
711 struct file_lock_context
*i_flctx
;
712 struct address_space i_data
;
713 struct list_head i_devices
;
715 struct pipe_inode_info
*i_pipe
;
716 struct block_device
*i_bdev
;
724 #ifdef CONFIG_FSNOTIFY
725 __u32 i_fsnotify_mask
; /* all events this inode cares about */
726 struct fsnotify_mark_connector __rcu
*i_fsnotify_marks
;
729 #ifdef CONFIG_FS_ENCRYPTION
730 struct fscrypt_info
*i_crypt_info
;
733 #ifdef CONFIG_FS_VERITY
734 struct fsverity_info
*i_verity_info
;
737 void *i_private
; /* fs or device private pointer */
738 } __randomize_layout
;
740 struct timespec64
timestamp_truncate(struct timespec64 t
, struct inode
*inode
);
742 static inline unsigned int i_blocksize(const struct inode
*node
)
744 return (1 << node
->i_blkbits
);
747 static inline int inode_unhashed(struct inode
*inode
)
749 return hlist_unhashed(&inode
->i_hash
);
753 * __mark_inode_dirty expects inodes to be hashed. Since we don't
754 * want special inodes in the fileset inode space, we make them
755 * appear hashed, but do not put on any lists. hlist_del()
756 * will work fine and require no locking.
758 static inline void inode_fake_hash(struct inode
*inode
)
760 hlist_add_fake(&inode
->i_hash
);
764 * inode->i_mutex nesting subclasses for the lock validator:
766 * 0: the object of the current VFS operation
770 * 4: second non-directory
771 * 5: second parent (when locking independent directories in rename)
773 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
774 * non-directories at once.
776 * The locking order between these classes is
777 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
779 enum inode_i_mutex_lock_class
789 static inline void inode_lock(struct inode
*inode
)
791 down_write(&inode
->i_rwsem
);
794 static inline void inode_unlock(struct inode
*inode
)
796 up_write(&inode
->i_rwsem
);
799 static inline void inode_lock_shared(struct inode
*inode
)
801 down_read(&inode
->i_rwsem
);
804 static inline void inode_unlock_shared(struct inode
*inode
)
806 up_read(&inode
->i_rwsem
);
809 static inline int inode_trylock(struct inode
*inode
)
811 return down_write_trylock(&inode
->i_rwsem
);
814 static inline int inode_trylock_shared(struct inode
*inode
)
816 return down_read_trylock(&inode
->i_rwsem
);
819 static inline int inode_is_locked(struct inode
*inode
)
821 return rwsem_is_locked(&inode
->i_rwsem
);
824 static inline void inode_lock_nested(struct inode
*inode
, unsigned subclass
)
826 down_write_nested(&inode
->i_rwsem
, subclass
);
829 static inline void inode_lock_shared_nested(struct inode
*inode
, unsigned subclass
)
831 down_read_nested(&inode
->i_rwsem
, subclass
);
834 void lock_two_nondirectories(struct inode
*, struct inode
*);
835 void unlock_two_nondirectories(struct inode
*, struct inode
*);
838 * NOTE: in a 32bit arch with a preemptable kernel and
839 * an UP compile the i_size_read/write must be atomic
840 * with respect to the local cpu (unlike with preempt disabled),
841 * but they don't need to be atomic with respect to other cpus like in
842 * true SMP (so they need either to either locally disable irq around
843 * the read or for example on x86 they can be still implemented as a
844 * cmpxchg8b without the need of the lock prefix). For SMP compiles
845 * and 64bit archs it makes no difference if preempt is enabled or not.
847 static inline loff_t
i_size_read(const struct inode
*inode
)
849 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
854 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
855 i_size
= inode
->i_size
;
856 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
858 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
862 i_size
= inode
->i_size
;
866 return inode
->i_size
;
871 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
872 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
873 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
875 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
877 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
879 write_seqcount_begin(&inode
->i_size_seqcount
);
880 inode
->i_size
= i_size
;
881 write_seqcount_end(&inode
->i_size_seqcount
);
883 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
885 inode
->i_size
= i_size
;
888 inode
->i_size
= i_size
;
892 static inline unsigned iminor(const struct inode
*inode
)
894 return MINOR(inode
->i_rdev
);
897 static inline unsigned imajor(const struct inode
*inode
)
899 return MAJOR(inode
->i_rdev
);
902 extern struct block_device
*I_BDEV(struct inode
*inode
);
905 rwlock_t lock
; /* protects pid, uid, euid fields */
906 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
907 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
908 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
909 int signum
; /* posix.1b rt signal to be delivered on IO */
913 * Track a single file's readahead state
915 struct file_ra_state
{
916 pgoff_t start
; /* where readahead started */
917 unsigned int size
; /* # of readahead pages */
918 unsigned int async_size
; /* do asynchronous readahead when
919 there are only # of pages ahead */
921 unsigned int ra_pages
; /* Maximum readahead window */
922 unsigned int mmap_miss
; /* Cache miss stat for mmap accesses */
923 loff_t prev_pos
; /* Cache last read() position */
927 * Check if @index falls in the readahead windows.
929 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
931 return (index
>= ra
->start
&&
932 index
< ra
->start
+ ra
->size
);
937 struct llist_node fu_llist
;
938 struct rcu_head fu_rcuhead
;
941 struct inode
*f_inode
; /* cached value */
942 const struct file_operations
*f_op
;
945 * Protects f_ep_links, f_flags.
946 * Must not be taken from IRQ context.
949 enum rw_hint f_write_hint
;
950 atomic_long_t f_count
;
951 unsigned int f_flags
;
953 struct mutex f_pos_lock
;
955 struct fown_struct f_owner
;
956 const struct cred
*f_cred
;
957 struct file_ra_state f_ra
;
960 #ifdef CONFIG_SECURITY
963 /* needed for tty driver, and maybe others */
967 /* Used by fs/eventpoll.c to link all the hooks to this file */
968 struct list_head f_ep_links
;
969 struct list_head f_tfile_llink
;
970 #endif /* #ifdef CONFIG_EPOLL */
971 struct address_space
*f_mapping
;
974 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
979 /* file identifier */
980 unsigned char f_handle
[0];
983 static inline struct file
*get_file(struct file
*f
)
985 atomic_long_inc(&f
->f_count
);
988 #define get_file_rcu_many(x, cnt) \
989 atomic_long_add_unless(&(x)->f_count, (cnt), 0)
990 #define get_file_rcu(x) get_file_rcu_many((x), 1)
991 #define file_count(x) atomic_long_read(&(x)->f_count)
993 #define MAX_NON_LFS ((1UL<<31) - 1)
995 /* Page cache limit. The filesystems should put that into their s_maxbytes
996 limits, otherwise bad things can happen in VM. */
997 #if BITS_PER_LONG==32
998 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
999 #elif BITS_PER_LONG==64
1000 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
1005 #define FL_DELEG 4 /* NFSv4 delegation */
1006 #define FL_ACCESS 8 /* not trying to lock, just looking */
1007 #define FL_EXISTS 16 /* when unlocking, test for existence */
1008 #define FL_LEASE 32 /* lease held on this file */
1009 #define FL_CLOSE 64 /* unlock on close */
1010 #define FL_SLEEP 128 /* A blocking lock */
1011 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
1012 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
1013 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
1014 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
1016 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
1019 * Special return value from posix_lock_file() and vfs_lock_file() for
1020 * asynchronous locking.
1022 #define FILE_LOCK_DEFERRED 1
1024 /* legacy typedef, should eventually be removed */
1025 typedef void *fl_owner_t
;
1029 struct file_lock_operations
{
1030 void (*fl_copy_lock
)(struct file_lock
*, struct file_lock
*);
1031 void (*fl_release_private
)(struct file_lock
*);
1034 struct lock_manager_operations
{
1035 fl_owner_t (*lm_get_owner
)(fl_owner_t
);
1036 void (*lm_put_owner
)(fl_owner_t
);
1037 void (*lm_notify
)(struct file_lock
*); /* unblock callback */
1038 int (*lm_grant
)(struct file_lock
*, int);
1039 bool (*lm_break
)(struct file_lock
*);
1040 int (*lm_change
)(struct file_lock
*, int, struct list_head
*);
1041 void (*lm_setup
)(struct file_lock
*, void **);
1044 struct lock_manager
{
1045 struct list_head list
;
1047 * NFSv4 and up also want opens blocked during the grace period;
1054 void locks_start_grace(struct net
*, struct lock_manager
*);
1055 void locks_end_grace(struct lock_manager
*);
1056 bool locks_in_grace(struct net
*);
1057 bool opens_in_grace(struct net
*);
1059 /* that will die - we need it for nfs_lock_info */
1060 #include <linux/nfs_fs_i.h>
1063 * struct file_lock represents a generic "file lock". It's used to represent
1064 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1065 * note that the same struct is used to represent both a request for a lock and
1066 * the lock itself, but the same object is never used for both.
1068 * FIXME: should we create a separate "struct lock_request" to help distinguish
1071 * The varous i_flctx lists are ordered by:
1074 * 2) lock range start
1077 * Obviously, the last two criteria only matter for POSIX locks.
1080 struct file_lock
*fl_blocker
; /* The lock, that is blocking us */
1081 struct list_head fl_list
; /* link into file_lock_context */
1082 struct hlist_node fl_link
; /* node in global lists */
1083 struct list_head fl_blocked_requests
; /* list of requests with
1084 * ->fl_blocker pointing here
1086 struct list_head fl_blocked_member
; /* node in
1087 * ->fl_blocker->fl_blocked_requests
1089 fl_owner_t fl_owner
;
1090 unsigned int fl_flags
;
1091 unsigned char fl_type
;
1092 unsigned int fl_pid
;
1093 int fl_link_cpu
; /* what cpu's list is this on? */
1094 wait_queue_head_t fl_wait
;
1095 struct file
*fl_file
;
1099 struct fasync_struct
* fl_fasync
; /* for lease break notifications */
1100 /* for lease breaks: */
1101 unsigned long fl_break_time
;
1102 unsigned long fl_downgrade_time
;
1104 const struct file_lock_operations
*fl_ops
; /* Callbacks for filesystems */
1105 const struct lock_manager_operations
*fl_lmops
; /* Callbacks for lockmanagers */
1107 struct nfs_lock_info nfs_fl
;
1108 struct nfs4_lock_info nfs4_fl
;
1110 struct list_head link
; /* link in AFS vnode's pending_locks list */
1111 int state
; /* state of grant or error if -ve */
1112 unsigned int debug_id
;
1115 } __randomize_layout
;
1117 struct file_lock_context
{
1118 spinlock_t flc_lock
;
1119 struct list_head flc_flock
;
1120 struct list_head flc_posix
;
1121 struct list_head flc_lease
;
1124 /* The following constant reflects the upper bound of the file/locking space */
1126 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1127 #define OFFSET_MAX INT_LIMIT(loff_t)
1128 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1131 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
1133 #define locks_inode(f) file_inode(f)
1135 #ifdef CONFIG_FILE_LOCKING
1136 extern int fcntl_getlk(struct file
*, unsigned int, struct flock
*);
1137 extern int fcntl_setlk(unsigned int, struct file
*, unsigned int,
1140 #if BITS_PER_LONG == 32
1141 extern int fcntl_getlk64(struct file
*, unsigned int, struct flock64
*);
1142 extern int fcntl_setlk64(unsigned int, struct file
*, unsigned int,
1146 extern int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
);
1147 extern int fcntl_getlease(struct file
*filp
);
1150 void locks_free_lock_context(struct inode
*inode
);
1151 void locks_free_lock(struct file_lock
*fl
);
1152 extern void locks_init_lock(struct file_lock
*);
1153 extern struct file_lock
* locks_alloc_lock(void);
1154 extern void locks_copy_lock(struct file_lock
*, struct file_lock
*);
1155 extern void locks_copy_conflock(struct file_lock
*, struct file_lock
*);
1156 extern void locks_remove_posix(struct file
*, fl_owner_t
);
1157 extern void locks_remove_file(struct file
*);
1158 extern void locks_release_private(struct file_lock
*);
1159 extern void posix_test_lock(struct file
*, struct file_lock
*);
1160 extern int posix_lock_file(struct file
*, struct file_lock
*, struct file_lock
*);
1161 extern int locks_delete_block(struct file_lock
*);
1162 extern int vfs_test_lock(struct file
*, struct file_lock
*);
1163 extern int vfs_lock_file(struct file
*, unsigned int, struct file_lock
*, struct file_lock
*);
1164 extern int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
);
1165 extern int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
);
1166 extern int __break_lease(struct inode
*inode
, unsigned int flags
, unsigned int type
);
1167 extern void lease_get_mtime(struct inode
*, struct timespec64
*time
);
1168 extern int generic_setlease(struct file
*, long, struct file_lock
**, void **priv
);
1169 extern int vfs_setlease(struct file
*, long, struct file_lock
**, void **);
1170 extern int lease_modify(struct file_lock
*, int, struct list_head
*);
1171 struct files_struct
;
1172 extern void show_fd_locks(struct seq_file
*f
,
1173 struct file
*filp
, struct files_struct
*files
);
1174 #else /* !CONFIG_FILE_LOCKING */
1175 static inline int fcntl_getlk(struct file
*file
, unsigned int cmd
,
1176 struct flock __user
*user
)
1181 static inline int fcntl_setlk(unsigned int fd
, struct file
*file
,
1182 unsigned int cmd
, struct flock __user
*user
)
1187 #if BITS_PER_LONG == 32
1188 static inline int fcntl_getlk64(struct file
*file
, unsigned int cmd
,
1189 struct flock64 __user
*user
)
1194 static inline int fcntl_setlk64(unsigned int fd
, struct file
*file
,
1195 unsigned int cmd
, struct flock64 __user
*user
)
1200 static inline int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
)
1205 static inline int fcntl_getlease(struct file
*filp
)
1211 locks_free_lock_context(struct inode
*inode
)
1215 static inline void locks_init_lock(struct file_lock
*fl
)
1220 static inline void locks_copy_conflock(struct file_lock
*new, struct file_lock
*fl
)
1225 static inline void locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
1230 static inline void locks_remove_posix(struct file
*filp
, fl_owner_t owner
)
1235 static inline void locks_remove_file(struct file
*filp
)
1240 static inline void posix_test_lock(struct file
*filp
, struct file_lock
*fl
)
1245 static inline int posix_lock_file(struct file
*filp
, struct file_lock
*fl
,
1246 struct file_lock
*conflock
)
1251 static inline int locks_delete_block(struct file_lock
*waiter
)
1256 static inline int vfs_test_lock(struct file
*filp
, struct file_lock
*fl
)
1261 static inline int vfs_lock_file(struct file
*filp
, unsigned int cmd
,
1262 struct file_lock
*fl
, struct file_lock
*conf
)
1267 static inline int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
)
1272 static inline int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
)
1277 static inline int __break_lease(struct inode
*inode
, unsigned int mode
, unsigned int type
)
1282 static inline void lease_get_mtime(struct inode
*inode
,
1283 struct timespec64
*time
)
1288 static inline int generic_setlease(struct file
*filp
, long arg
,
1289 struct file_lock
**flp
, void **priv
)
1294 static inline int vfs_setlease(struct file
*filp
, long arg
,
1295 struct file_lock
**lease
, void **priv
)
1300 static inline int lease_modify(struct file_lock
*fl
, int arg
,
1301 struct list_head
*dispose
)
1306 struct files_struct
;
1307 static inline void show_fd_locks(struct seq_file
*f
,
1308 struct file
*filp
, struct files_struct
*files
) {}
1309 #endif /* !CONFIG_FILE_LOCKING */
1311 static inline struct inode
*file_inode(const struct file
*f
)
1316 static inline struct dentry
*file_dentry(const struct file
*file
)
1318 return d_real(file
->f_path
.dentry
, file_inode(file
));
1321 static inline int locks_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1323 return locks_lock_inode_wait(locks_inode(filp
), fl
);
1326 struct fasync_struct
{
1330 struct fasync_struct
*fa_next
; /* singly linked list */
1331 struct file
*fa_file
;
1332 struct rcu_head fa_rcu
;
1335 #define FASYNC_MAGIC 0x4601
1337 /* SMP safe fasync helpers: */
1338 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1339 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1340 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1341 extern struct fasync_struct
*fasync_alloc(void);
1342 extern void fasync_free(struct fasync_struct
*);
1344 /* can be called from interrupts */
1345 extern void kill_fasync(struct fasync_struct
**, int, int);
1347 extern void __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1348 extern int f_setown(struct file
*filp
, unsigned long arg
, int force
);
1349 extern void f_delown(struct file
*filp
);
1350 extern pid_t
f_getown(struct file
*filp
);
1351 extern int send_sigurg(struct fown_struct
*fown
);
1354 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1355 * represented in both.
1357 #define SB_RDONLY 1 /* Mount read-only */
1358 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1359 #define SB_NODEV 4 /* Disallow access to device special files */
1360 #define SB_NOEXEC 8 /* Disallow program execution */
1361 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1362 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1363 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1364 #define SB_NOATIME 1024 /* Do not update access times. */
1365 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1366 #define SB_SILENT 32768
1367 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1368 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1369 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1370 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1372 /* These sb flags are internal to the kernel */
1373 #define SB_SUBMOUNT (1<<26)
1374 #define SB_FORCE (1<<27)
1375 #define SB_NOSEC (1<<28)
1376 #define SB_BORN (1<<29)
1377 #define SB_ACTIVE (1<<30)
1378 #define SB_NOUSER (1<<31)
1384 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1385 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1386 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1387 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1388 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1391 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1392 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1393 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1394 #define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */
1396 /* sb->s_iflags to limit user namespace mounts */
1397 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1398 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1399 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1401 /* Possible states of 'frozen' field */
1403 SB_UNFROZEN
= 0, /* FS is unfrozen */
1404 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1405 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1406 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1407 * internal threads if needed) */
1408 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1411 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1414 int frozen
; /* Is sb frozen? */
1415 wait_queue_head_t wait_unfrozen
; /* for get_super_thawed() */
1416 struct percpu_rw_semaphore rw_sem
[SB_FREEZE_LEVELS
];
1419 struct super_block
{
1420 struct list_head s_list
; /* Keep this first */
1421 dev_t s_dev
; /* search index; _not_ kdev_t */
1422 unsigned char s_blocksize_bits
;
1423 unsigned long s_blocksize
;
1424 loff_t s_maxbytes
; /* Max file size */
1425 struct file_system_type
*s_type
;
1426 const struct super_operations
*s_op
;
1427 const struct dquot_operations
*dq_op
;
1428 const struct quotactl_ops
*s_qcop
;
1429 const struct export_operations
*s_export_op
;
1430 unsigned long s_flags
;
1431 unsigned long s_iflags
; /* internal SB_I_* flags */
1432 unsigned long s_magic
;
1433 struct dentry
*s_root
;
1434 struct rw_semaphore s_umount
;
1437 #ifdef CONFIG_SECURITY
1440 const struct xattr_handler
**s_xattr
;
1441 #ifdef CONFIG_FS_ENCRYPTION
1442 const struct fscrypt_operations
*s_cop
;
1443 struct key
*s_master_keys
; /* master crypto keys in use */
1445 #ifdef CONFIG_FS_VERITY
1446 const struct fsverity_operations
*s_vop
;
1448 struct hlist_bl_head s_roots
; /* alternate root dentries for NFS */
1449 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1450 struct block_device
*s_bdev
;
1451 struct backing_dev_info
*s_bdi
;
1452 struct mtd_info
*s_mtd
;
1453 struct hlist_node s_instances
;
1454 unsigned int s_quota_types
; /* Bitmask of supported quota types */
1455 struct quota_info s_dquot
; /* Diskquota specific options */
1457 struct sb_writers s_writers
;
1460 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1461 * s_fsnotify_marks together for cache efficiency. They are frequently
1462 * accessed and rarely modified.
1464 void *s_fs_info
; /* Filesystem private info */
1466 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1468 /* Time limits for c/m/atime in seconds */
1469 time64_t s_time_min
;
1470 time64_t s_time_max
;
1471 #ifdef CONFIG_FSNOTIFY
1472 __u32 s_fsnotify_mask
;
1473 struct fsnotify_mark_connector __rcu
*s_fsnotify_marks
;
1476 char s_id
[32]; /* Informational name */
1477 uuid_t s_uuid
; /* UUID */
1479 unsigned int s_max_links
;
1483 * The next field is for VFS *only*. No filesystems have any business
1484 * even looking at it. You had been warned.
1486 struct mutex s_vfs_rename_mutex
; /* Kludge */
1489 * Filesystem subtype. If non-empty the filesystem type field
1490 * in /proc/mounts will be "type.subtype"
1492 const char *s_subtype
;
1494 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1497 * Saved pool identifier for cleancache (-1 means none)
1499 int cleancache_poolid
;
1501 struct shrinker s_shrink
; /* per-sb shrinker handle */
1503 /* Number of inodes with nlink == 0 but still referenced */
1504 atomic_long_t s_remove_count
;
1506 /* Pending fsnotify inode refs */
1507 atomic_long_t s_fsnotify_inode_refs
;
1509 /* Being remounted read-only */
1510 int s_readonly_remount
;
1512 /* AIO completions deferred from interrupt context */
1513 struct workqueue_struct
*s_dio_done_wq
;
1514 struct hlist_head s_pins
;
1517 * Owning user namespace and default context in which to
1518 * interpret filesystem uids, gids, quotas, device nodes,
1519 * xattrs and security labels.
1521 struct user_namespace
*s_user_ns
;
1524 * The list_lru structure is essentially just a pointer to a table
1525 * of per-node lru lists, each of which has its own spinlock.
1526 * There is no need to put them into separate cachelines.
1528 struct list_lru s_dentry_lru
;
1529 struct list_lru s_inode_lru
;
1530 struct rcu_head rcu
;
1531 struct work_struct destroy_work
;
1533 struct mutex s_sync_lock
; /* sync serialisation lock */
1536 * Indicates how deep in a filesystem stack this SB is
1540 /* s_inode_list_lock protects s_inodes */
1541 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp
;
1542 struct list_head s_inodes
; /* all inodes */
1544 spinlock_t s_inode_wblist_lock
;
1545 struct list_head s_inodes_wb
; /* writeback inodes */
1546 } __randomize_layout
;
1548 /* Helper functions so that in most cases filesystems will
1549 * not need to deal directly with kuid_t and kgid_t and can
1550 * instead deal with the raw numeric values that are stored
1551 * in the filesystem.
1553 static inline uid_t
i_uid_read(const struct inode
*inode
)
1555 return from_kuid(inode
->i_sb
->s_user_ns
, inode
->i_uid
);
1558 static inline gid_t
i_gid_read(const struct inode
*inode
)
1560 return from_kgid(inode
->i_sb
->s_user_ns
, inode
->i_gid
);
1563 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
1565 inode
->i_uid
= make_kuid(inode
->i_sb
->s_user_ns
, uid
);
1568 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
1570 inode
->i_gid
= make_kgid(inode
->i_sb
->s_user_ns
, gid
);
1573 extern struct timespec64
timespec64_trunc(struct timespec64 t
, unsigned gran
);
1574 extern struct timespec64
current_time(struct inode
*inode
);
1577 * Snapshotting support.
1580 void __sb_end_write(struct super_block
*sb
, int level
);
1581 int __sb_start_write(struct super_block
*sb
, int level
, bool wait
);
1583 #define __sb_writers_acquired(sb, lev) \
1584 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1585 #define __sb_writers_release(sb, lev) \
1586 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1589 * sb_end_write - drop write access to a superblock
1590 * @sb: the super we wrote to
1592 * Decrement number of writers to the filesystem. Wake up possible waiters
1593 * wanting to freeze the filesystem.
1595 static inline void sb_end_write(struct super_block
*sb
)
1597 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1601 * sb_end_pagefault - drop write access to a superblock from a page fault
1602 * @sb: the super we wrote to
1604 * Decrement number of processes handling write page fault to the filesystem.
1605 * Wake up possible waiters wanting to freeze the filesystem.
1607 static inline void sb_end_pagefault(struct super_block
*sb
)
1609 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1613 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1614 * @sb: the super we wrote to
1616 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1617 * waiters wanting to freeze the filesystem.
1619 static inline void sb_end_intwrite(struct super_block
*sb
)
1621 __sb_end_write(sb
, SB_FREEZE_FS
);
1625 * sb_start_write - get write access to a superblock
1626 * @sb: the super we write to
1628 * When a process wants to write data or metadata to a file system (i.e. dirty
1629 * a page or an inode), it should embed the operation in a sb_start_write() -
1630 * sb_end_write() pair to get exclusion against file system freezing. This
1631 * function increments number of writers preventing freezing. If the file
1632 * system is already frozen, the function waits until the file system is
1635 * Since freeze protection behaves as a lock, users have to preserve
1636 * ordering of freeze protection and other filesystem locks. Generally,
1637 * freeze protection should be the outermost lock. In particular, we have:
1640 * -> i_mutex (write path, truncate, directory ops, ...)
1641 * -> s_umount (freeze_super, thaw_super)
1643 static inline void sb_start_write(struct super_block
*sb
)
1645 __sb_start_write(sb
, SB_FREEZE_WRITE
, true);
1648 static inline int sb_start_write_trylock(struct super_block
*sb
)
1650 return __sb_start_write(sb
, SB_FREEZE_WRITE
, false);
1654 * sb_start_pagefault - get write access to a superblock from a page fault
1655 * @sb: the super we write to
1657 * When a process starts handling write page fault, it should embed the
1658 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1659 * exclusion against file system freezing. This is needed since the page fault
1660 * is going to dirty a page. This function increments number of running page
1661 * faults preventing freezing. If the file system is already frozen, the
1662 * function waits until the file system is thawed.
1664 * Since page fault freeze protection behaves as a lock, users have to preserve
1665 * ordering of freeze protection and other filesystem locks. It is advised to
1666 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1667 * handling code implies lock dependency:
1670 * -> sb_start_pagefault
1672 static inline void sb_start_pagefault(struct super_block
*sb
)
1674 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
, true);
1678 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1679 * @sb: the super we write to
1681 * This is the third level of protection against filesystem freezing. It is
1682 * free for use by a filesystem. The only requirement is that it must rank
1683 * below sb_start_pagefault.
1685 * For example filesystem can call sb_start_intwrite() when starting a
1686 * transaction which somewhat eases handling of freezing for internal sources
1687 * of filesystem changes (internal fs threads, discarding preallocation on file
1690 static inline void sb_start_intwrite(struct super_block
*sb
)
1692 __sb_start_write(sb
, SB_FREEZE_FS
, true);
1695 static inline int sb_start_intwrite_trylock(struct super_block
*sb
)
1697 return __sb_start_write(sb
, SB_FREEZE_FS
, false);
1701 extern bool inode_owner_or_capable(const struct inode
*inode
);
1704 * VFS helper functions..
1706 extern int vfs_create(struct inode
*, struct dentry
*, umode_t
, bool);
1707 extern int vfs_mkdir(struct inode
*, struct dentry
*, umode_t
);
1708 extern int vfs_mknod(struct inode
*, struct dentry
*, umode_t
, dev_t
);
1709 extern int vfs_symlink(struct inode
*, struct dentry
*, const char *);
1710 extern int vfs_link(struct dentry
*, struct inode
*, struct dentry
*, struct inode
**);
1711 extern int vfs_rmdir(struct inode
*, struct dentry
*);
1712 extern int vfs_unlink(struct inode
*, struct dentry
*, struct inode
**);
1713 extern int vfs_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*, struct inode
**, unsigned int);
1714 extern int vfs_whiteout(struct inode
*, struct dentry
*);
1716 extern struct dentry
*vfs_tmpfile(struct dentry
*dentry
, umode_t mode
,
1719 int vfs_mkobj(struct dentry
*, umode_t
,
1720 int (*f
)(struct dentry
*, umode_t
, void *),
1723 extern long vfs_ioctl(struct file
*file
, unsigned int cmd
, unsigned long arg
);
1726 * VFS file helper functions.
1728 extern void inode_init_owner(struct inode
*inode
, const struct inode
*dir
,
1730 extern bool may_open_dev(const struct path
*path
);
1732 * VFS FS_IOC_FIEMAP helper definitions.
1734 struct fiemap_extent_info
{
1735 unsigned int fi_flags
; /* Flags as passed from user */
1736 unsigned int fi_extents_mapped
; /* Number of mapped extents */
1737 unsigned int fi_extents_max
; /* Size of fiemap_extent array */
1738 struct fiemap_extent __user
*fi_extents_start
; /* Start of
1739 fiemap_extent array */
1741 int fiemap_fill_next_extent(struct fiemap_extent_info
*info
, u64 logical
,
1742 u64 phys
, u64 len
, u32 flags
);
1743 int fiemap_check_flags(struct fiemap_extent_info
*fieinfo
, u32 fs_flags
);
1746 * This is the "filldir" function type, used by readdir() to let
1747 * the kernel specify what kind of dirent layout it wants to have.
1748 * This allows the kernel to read directories into kernel space or
1749 * to have different dirent layouts depending on the binary type.
1752 typedef int (*filldir_t
)(struct dir_context
*, const char *, int, loff_t
, u64
,
1755 struct dir_context
{
1760 struct block_device_operations
;
1762 /* These macros are for out of kernel modules to test that
1763 * the kernel supports the unlocked_ioctl and compat_ioctl
1764 * fields in struct file_operations. */
1765 #define HAVE_COMPAT_IOCTL 1
1766 #define HAVE_UNLOCKED_IOCTL 1
1769 * These flags let !MMU mmap() govern direct device mapping vs immediate
1770 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1772 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1773 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1774 * NOMMU_MAP_READ: Can be mapped for reading
1775 * NOMMU_MAP_WRITE: Can be mapped for writing
1776 * NOMMU_MAP_EXEC: Can be mapped for execution
1778 #define NOMMU_MAP_COPY 0x00000001
1779 #define NOMMU_MAP_DIRECT 0x00000008
1780 #define NOMMU_MAP_READ VM_MAYREAD
1781 #define NOMMU_MAP_WRITE VM_MAYWRITE
1782 #define NOMMU_MAP_EXEC VM_MAYEXEC
1784 #define NOMMU_VMFLAGS \
1785 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1788 * These flags control the behavior of the remap_file_range function pointer.
1789 * If it is called with len == 0 that means "remap to end of source file".
1790 * See Documentation/filesystems/vfs.rst for more details about this call.
1792 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
1793 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
1795 #define REMAP_FILE_DEDUP (1 << 0)
1796 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
1799 * These flags signal that the caller is ok with altering various aspects of
1800 * the behavior of the remap operation. The changes must be made by the
1801 * implementation; the vfs remap helper functions can take advantage of them.
1802 * Flags in this category exist to preserve the quirky behavior of the hoisted
1803 * btrfs clone/dedupe ioctls.
1805 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
1809 struct file_operations
{
1810 struct module
*owner
;
1811 loff_t (*llseek
) (struct file
*, loff_t
, int);
1812 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
1813 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
1814 ssize_t (*read_iter
) (struct kiocb
*, struct iov_iter
*);
1815 ssize_t (*write_iter
) (struct kiocb
*, struct iov_iter
*);
1816 int (*iopoll
)(struct kiocb
*kiocb
, bool spin
);
1817 int (*iterate
) (struct file
*, struct dir_context
*);
1818 int (*iterate_shared
) (struct file
*, struct dir_context
*);
1819 __poll_t (*poll
) (struct file
*, struct poll_table_struct
*);
1820 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
1821 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
1822 int (*mmap
) (struct file
*, struct vm_area_struct
*);
1823 unsigned long mmap_supported_flags
;
1824 int (*open
) (struct inode
*, struct file
*);
1825 int (*flush
) (struct file
*, fl_owner_t id
);
1826 int (*release
) (struct inode
*, struct file
*);
1827 int (*fsync
) (struct file
*, loff_t
, loff_t
, int datasync
);
1828 int (*fasync
) (int, struct file
*, int);
1829 int (*lock
) (struct file
*, int, struct file_lock
*);
1830 ssize_t (*sendpage
) (struct file
*, struct page
*, int, size_t, loff_t
*, int);
1831 unsigned long (*get_unmapped_area
)(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1832 int (*check_flags
)(int);
1833 int (*flock
) (struct file
*, int, struct file_lock
*);
1834 ssize_t (*splice_write
)(struct pipe_inode_info
*, struct file
*, loff_t
*, size_t, unsigned int);
1835 ssize_t (*splice_read
)(struct file
*, loff_t
*, struct pipe_inode_info
*, size_t, unsigned int);
1836 int (*setlease
)(struct file
*, long, struct file_lock
**, void **);
1837 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
1839 void (*show_fdinfo
)(struct seq_file
*m
, struct file
*f
);
1841 unsigned (*mmap_capabilities
)(struct file
*);
1843 ssize_t (*copy_file_range
)(struct file
*, loff_t
, struct file
*,
1844 loff_t
, size_t, unsigned int);
1845 loff_t (*remap_file_range
)(struct file
*file_in
, loff_t pos_in
,
1846 struct file
*file_out
, loff_t pos_out
,
1847 loff_t len
, unsigned int remap_flags
);
1848 int (*fadvise
)(struct file
*, loff_t
, loff_t
, int);
1849 } __randomize_layout
;
1851 struct inode_operations
{
1852 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*, unsigned int);
1853 const char * (*get_link
) (struct dentry
*, struct inode
*, struct delayed_call
*);
1854 int (*permission
) (struct inode
*, int);
1855 struct posix_acl
* (*get_acl
)(struct inode
*, int);
1857 int (*readlink
) (struct dentry
*, char __user
*,int);
1859 int (*create
) (struct inode
*,struct dentry
*, umode_t
, bool);
1860 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
1861 int (*unlink
) (struct inode
*,struct dentry
*);
1862 int (*symlink
) (struct inode
*,struct dentry
*,const char *);
1863 int (*mkdir
) (struct inode
*,struct dentry
*,umode_t
);
1864 int (*rmdir
) (struct inode
*,struct dentry
*);
1865 int (*mknod
) (struct inode
*,struct dentry
*,umode_t
,dev_t
);
1866 int (*rename
) (struct inode
*, struct dentry
*,
1867 struct inode
*, struct dentry
*, unsigned int);
1868 int (*setattr
) (struct dentry
*, struct iattr
*);
1869 int (*getattr
) (const struct path
*, struct kstat
*, u32
, unsigned int);
1870 ssize_t (*listxattr
) (struct dentry
*, char *, size_t);
1871 int (*fiemap
)(struct inode
*, struct fiemap_extent_info
*, u64 start
,
1873 int (*update_time
)(struct inode
*, struct timespec64
*, int);
1874 int (*atomic_open
)(struct inode
*, struct dentry
*,
1875 struct file
*, unsigned open_flag
,
1876 umode_t create_mode
);
1877 int (*tmpfile
) (struct inode
*, struct dentry
*, umode_t
);
1878 int (*set_acl
)(struct inode
*, struct posix_acl
*, int);
1879 } ____cacheline_aligned
;
1881 static inline ssize_t
call_read_iter(struct file
*file
, struct kiocb
*kio
,
1882 struct iov_iter
*iter
)
1884 return file
->f_op
->read_iter(kio
, iter
);
1887 static inline ssize_t
call_write_iter(struct file
*file
, struct kiocb
*kio
,
1888 struct iov_iter
*iter
)
1890 return file
->f_op
->write_iter(kio
, iter
);
1893 static inline int call_mmap(struct file
*file
, struct vm_area_struct
*vma
)
1895 return file
->f_op
->mmap(file
, vma
);
1898 ssize_t
rw_copy_check_uvector(int type
, const struct iovec __user
* uvector
,
1899 unsigned long nr_segs
, unsigned long fast_segs
,
1900 struct iovec
*fast_pointer
,
1901 struct iovec
**ret_pointer
);
1903 extern ssize_t
__vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1904 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1905 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1906 extern ssize_t
vfs_readv(struct file
*, const struct iovec __user
*,
1907 unsigned long, loff_t
*, rwf_t
);
1908 extern ssize_t
vfs_copy_file_range(struct file
*, loff_t
, struct file
*,
1909 loff_t
, size_t, unsigned int);
1910 extern ssize_t
generic_copy_file_range(struct file
*file_in
, loff_t pos_in
,
1911 struct file
*file_out
, loff_t pos_out
,
1912 size_t len
, unsigned int flags
);
1913 extern int generic_remap_file_range_prep(struct file
*file_in
, loff_t pos_in
,
1914 struct file
*file_out
, loff_t pos_out
,
1916 unsigned int remap_flags
);
1917 extern loff_t
do_clone_file_range(struct file
*file_in
, loff_t pos_in
,
1918 struct file
*file_out
, loff_t pos_out
,
1919 loff_t len
, unsigned int remap_flags
);
1920 extern loff_t
vfs_clone_file_range(struct file
*file_in
, loff_t pos_in
,
1921 struct file
*file_out
, loff_t pos_out
,
1922 loff_t len
, unsigned int remap_flags
);
1923 extern int vfs_dedupe_file_range(struct file
*file
,
1924 struct file_dedupe_range
*same
);
1925 extern loff_t
vfs_dedupe_file_range_one(struct file
*src_file
, loff_t src_pos
,
1926 struct file
*dst_file
, loff_t dst_pos
,
1927 loff_t len
, unsigned int remap_flags
);
1930 struct super_operations
{
1931 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
1932 void (*destroy_inode
)(struct inode
*);
1933 void (*free_inode
)(struct inode
*);
1935 void (*dirty_inode
) (struct inode
*, int flags
);
1936 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
1937 int (*drop_inode
) (struct inode
*);
1938 void (*evict_inode
) (struct inode
*);
1939 void (*put_super
) (struct super_block
*);
1940 int (*sync_fs
)(struct super_block
*sb
, int wait
);
1941 int (*freeze_super
) (struct super_block
*);
1942 int (*freeze_fs
) (struct super_block
*);
1943 int (*thaw_super
) (struct super_block
*);
1944 int (*unfreeze_fs
) (struct super_block
*);
1945 int (*statfs
) (struct dentry
*, struct kstatfs
*);
1946 int (*remount_fs
) (struct super_block
*, int *, char *);
1947 void (*umount_begin
) (struct super_block
*);
1949 int (*show_options
)(struct seq_file
*, struct dentry
*);
1950 int (*show_devname
)(struct seq_file
*, struct dentry
*);
1951 int (*show_path
)(struct seq_file
*, struct dentry
*);
1952 int (*show_stats
)(struct seq_file
*, struct dentry
*);
1954 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
1955 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
1956 struct dquot
**(*get_dquots
)(struct inode
*);
1958 int (*bdev_try_to_free_page
)(struct super_block
*, struct page
*, gfp_t
);
1959 long (*nr_cached_objects
)(struct super_block
*,
1960 struct shrink_control
*);
1961 long (*free_cached_objects
)(struct super_block
*,
1962 struct shrink_control
*);
1966 * Inode flags - they have no relation to superblock flags now
1968 #define S_SYNC 1 /* Writes are synced at once */
1969 #define S_NOATIME 2 /* Do not update access times */
1970 #define S_APPEND 4 /* Append-only file */
1971 #define S_IMMUTABLE 8 /* Immutable file */
1972 #define S_DEAD 16 /* removed, but still open directory */
1973 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1974 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1975 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1976 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1977 #define S_PRIVATE 512 /* Inode is fs-internal */
1978 #define S_IMA 1024 /* Inode has an associated IMA struct */
1979 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1980 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1981 #ifdef CONFIG_FS_DAX
1982 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1984 #define S_DAX 0 /* Make all the DAX code disappear */
1986 #define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */
1987 #define S_CASEFOLD 32768 /* Casefolded file */
1988 #define S_VERITY 65536 /* Verity file (using fs/verity/) */
1991 * Note that nosuid etc flags are inode-specific: setting some file-system
1992 * flags just means all the inodes inherit those flags by default. It might be
1993 * possible to override it selectively if you really wanted to with some
1994 * ioctl() that is not currently implemented.
1996 * Exception: SB_RDONLY is always applied to the entire file system.
1998 * Unfortunately, it is possible to change a filesystems flags with it mounted
1999 * with files in use. This means that all of the inodes will not have their
2000 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
2001 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
2003 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
2005 static inline bool sb_rdonly(const struct super_block
*sb
) { return sb
->s_flags
& SB_RDONLY
; }
2006 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
2007 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
2008 ((inode)->i_flags & S_SYNC))
2009 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
2010 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
2011 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
2012 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
2013 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
2015 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
2016 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
2017 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
2018 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
2020 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
2021 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
2022 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
2023 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
2024 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
2025 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2026 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2027 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2028 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2029 #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD)
2030 #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY)
2032 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2033 (inode)->i_rdev == WHITEOUT_DEV)
2035 static inline bool HAS_UNMAPPED_ID(struct inode
*inode
)
2037 return !uid_valid(inode
->i_uid
) || !gid_valid(inode
->i_gid
);
2040 static inline enum rw_hint
file_write_hint(struct file
*file
)
2042 if (file
->f_write_hint
!= WRITE_LIFE_NOT_SET
)
2043 return file
->f_write_hint
;
2045 return file_inode(file
)->i_write_hint
;
2048 static inline int iocb_flags(struct file
*file
);
2050 static inline u16
ki_hint_validate(enum rw_hint hint
)
2052 typeof(((struct kiocb
*)0)->ki_hint
) max_hint
= -1;
2054 if (hint
<= max_hint
)
2059 static inline void init_sync_kiocb(struct kiocb
*kiocb
, struct file
*filp
)
2061 *kiocb
= (struct kiocb
) {
2063 .ki_flags
= iocb_flags(filp
),
2064 .ki_hint
= ki_hint_validate(file_write_hint(filp
)),
2065 .ki_ioprio
= get_current_ioprio(),
2070 * Inode state bits. Protected by inode->i_lock
2072 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
2073 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
2075 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2076 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2077 * various stages of removing an inode.
2079 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2081 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2082 * fdatasync(). i_atime is the usual cause.
2083 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2084 * these changes separately from I_DIRTY_SYNC so that we
2085 * don't have to write inode on fdatasync() when only
2086 * mtime has changed in it.
2087 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2088 * I_NEW Serves as both a mutex and completion notification.
2089 * New inodes set I_NEW. If two processes both create
2090 * the same inode, one of them will release its inode and
2091 * wait for I_NEW to be released before returning.
2092 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2093 * also cause waiting on I_NEW, without I_NEW actually
2094 * being set. find_inode() uses this to prevent returning
2095 * nearly-dead inodes.
2096 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2097 * is zero. I_FREEING must be set when I_WILL_FREE is
2099 * I_FREEING Set when inode is about to be freed but still has dirty
2100 * pages or buffers attached or the inode itself is still
2102 * I_CLEAR Added by clear_inode(). In this state the inode is
2103 * clean and can be destroyed. Inode keeps I_FREEING.
2105 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2106 * prohibited for many purposes. iget() must wait for
2107 * the inode to be completely released, then create it
2108 * anew. Other functions will just ignore such inodes,
2109 * if appropriate. I_NEW is used for waiting.
2111 * I_SYNC Writeback of inode is running. The bit is set during
2112 * data writeback, and cleared with a wakeup on the bit
2113 * address once it is done. The bit is also used to pin
2114 * the inode in memory for flusher thread.
2116 * I_REFERENCED Marks the inode as recently references on the LRU list.
2118 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2120 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2121 * synchronize competing switching instances and to tell
2122 * wb stat updates to grab the i_pages lock. See
2123 * inode_switch_wbs_work_fn() for details.
2125 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2126 * and work dirs among overlayfs mounts.
2128 * I_CREATING New object's inode in the middle of setting up.
2130 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2132 #define I_DIRTY_SYNC (1 << 0)
2133 #define I_DIRTY_DATASYNC (1 << 1)
2134 #define I_DIRTY_PAGES (1 << 2)
2136 #define I_NEW (1 << __I_NEW)
2137 #define I_WILL_FREE (1 << 4)
2138 #define I_FREEING (1 << 5)
2139 #define I_CLEAR (1 << 6)
2141 #define I_SYNC (1 << __I_SYNC)
2142 #define I_REFERENCED (1 << 8)
2143 #define __I_DIO_WAKEUP 9
2144 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2145 #define I_LINKABLE (1 << 10)
2146 #define I_DIRTY_TIME (1 << 11)
2147 #define __I_DIRTY_TIME_EXPIRED 12
2148 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
2149 #define I_WB_SWITCH (1 << 13)
2150 #define I_OVL_INUSE (1 << 14)
2151 #define I_CREATING (1 << 15)
2153 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2154 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2155 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2157 extern void __mark_inode_dirty(struct inode
*, int);
2158 static inline void mark_inode_dirty(struct inode
*inode
)
2160 __mark_inode_dirty(inode
, I_DIRTY
);
2163 static inline void mark_inode_dirty_sync(struct inode
*inode
)
2165 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
2168 extern void inc_nlink(struct inode
*inode
);
2169 extern void drop_nlink(struct inode
*inode
);
2170 extern void clear_nlink(struct inode
*inode
);
2171 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
2173 static inline void inode_inc_link_count(struct inode
*inode
)
2176 mark_inode_dirty(inode
);
2179 static inline void inode_dec_link_count(struct inode
*inode
)
2182 mark_inode_dirty(inode
);
2185 enum file_time_flags
{
2192 extern bool atime_needs_update(const struct path
*, struct inode
*);
2193 extern void touch_atime(const struct path
*);
2194 static inline void file_accessed(struct file
*file
)
2196 if (!(file
->f_flags
& O_NOATIME
))
2197 touch_atime(&file
->f_path
);
2200 extern int file_modified(struct file
*file
);
2202 int sync_inode(struct inode
*inode
, struct writeback_control
*wbc
);
2203 int sync_inode_metadata(struct inode
*inode
, int wait
);
2205 struct file_system_type
{
2208 #define FS_REQUIRES_DEV 1
2209 #define FS_BINARY_MOUNTDATA 2
2210 #define FS_HAS_SUBTYPE 4
2211 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2212 #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */
2213 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2214 int (*init_fs_context
)(struct fs_context
*);
2215 const struct fs_parameter_description
*parameters
;
2216 struct dentry
*(*mount
) (struct file_system_type
*, int,
2217 const char *, void *);
2218 void (*kill_sb
) (struct super_block
*);
2219 struct module
*owner
;
2220 struct file_system_type
* next
;
2221 struct hlist_head fs_supers
;
2223 struct lock_class_key s_lock_key
;
2224 struct lock_class_key s_umount_key
;
2225 struct lock_class_key s_vfs_rename_key
;
2226 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
2228 struct lock_class_key i_lock_key
;
2229 struct lock_class_key i_mutex_key
;
2230 struct lock_class_key i_mutex_dir_key
;
2233 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2236 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
2237 int flags
, const char *dev_name
, void *data
,
2238 int (*fill_super
)(struct super_block
*, void *, int));
2240 static inline struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
2241 int flags
, const char *dev_name
, void *data
,
2242 int (*fill_super
)(struct super_block
*, void *, int))
2244 return ERR_PTR(-ENODEV
);
2247 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
2248 int flags
, void *data
,
2249 int (*fill_super
)(struct super_block
*, void *, int));
2250 extern struct dentry
*mount_nodev(struct file_system_type
*fs_type
,
2251 int flags
, void *data
,
2252 int (*fill_super
)(struct super_block
*, void *, int));
2253 extern struct dentry
*mount_subtree(struct vfsmount
*mnt
, const char *path
);
2254 void generic_shutdown_super(struct super_block
*sb
);
2256 void kill_block_super(struct super_block
*sb
);
2258 static inline void kill_block_super(struct super_block
*sb
)
2263 void kill_anon_super(struct super_block
*sb
);
2264 void kill_litter_super(struct super_block
*sb
);
2265 void deactivate_super(struct super_block
*sb
);
2266 void deactivate_locked_super(struct super_block
*sb
);
2267 int set_anon_super(struct super_block
*s
, void *data
);
2268 int set_anon_super_fc(struct super_block
*s
, struct fs_context
*fc
);
2269 int get_anon_bdev(dev_t
*);
2270 void free_anon_bdev(dev_t
);
2271 struct super_block
*sget_fc(struct fs_context
*fc
,
2272 int (*test
)(struct super_block
*, struct fs_context
*),
2273 int (*set
)(struct super_block
*, struct fs_context
*));
2274 struct super_block
*sget(struct file_system_type
*type
,
2275 int (*test
)(struct super_block
*,void *),
2276 int (*set
)(struct super_block
*,void *),
2277 int flags
, void *data
);
2279 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2280 #define fops_get(fops) \
2281 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2282 #define fops_put(fops) \
2283 do { if (fops) module_put((fops)->owner); } while(0)
2285 * This one is to be used *ONLY* from ->open() instances.
2286 * fops must be non-NULL, pinned down *and* module dependencies
2287 * should be sufficient to pin the caller down as well.
2289 #define replace_fops(f, fops) \
2291 struct file *__file = (f); \
2292 fops_put(__file->f_op); \
2293 BUG_ON(!(__file->f_op = (fops))); \
2296 extern int register_filesystem(struct file_system_type
*);
2297 extern int unregister_filesystem(struct file_system_type
*);
2298 extern struct vfsmount
*kern_mount(struct file_system_type
*);
2299 extern void kern_unmount(struct vfsmount
*mnt
);
2300 extern int may_umount_tree(struct vfsmount
*);
2301 extern int may_umount(struct vfsmount
*);
2302 extern long do_mount(const char *, const char __user
*,
2303 const char *, unsigned long, void *);
2304 extern struct vfsmount
*collect_mounts(const struct path
*);
2305 extern void drop_collected_mounts(struct vfsmount
*);
2306 extern int iterate_mounts(int (*)(struct vfsmount
*, void *), void *,
2308 extern int vfs_statfs(const struct path
*, struct kstatfs
*);
2309 extern int user_statfs(const char __user
*, struct kstatfs
*);
2310 extern int fd_statfs(int, struct kstatfs
*);
2311 extern int freeze_super(struct super_block
*super
);
2312 extern int thaw_super(struct super_block
*super
);
2313 extern bool our_mnt(struct vfsmount
*mnt
);
2314 extern __printf(2, 3)
2315 int super_setup_bdi_name(struct super_block
*sb
, char *fmt
, ...);
2316 extern int super_setup_bdi(struct super_block
*sb
);
2318 extern int current_umask(void);
2320 extern void ihold(struct inode
* inode
);
2321 extern void iput(struct inode
*);
2322 extern int generic_update_time(struct inode
*, struct timespec64
*, int);
2325 extern struct kobject
*fs_kobj
;
2327 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2329 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2330 extern int locks_mandatory_locked(struct file
*);
2331 extern int locks_mandatory_area(struct inode
*, struct file
*, loff_t
, loff_t
, unsigned char);
2334 * Candidates for mandatory locking have the setgid bit set
2335 * but no group execute bit - an otherwise meaningless combination.
2338 static inline int __mandatory_lock(struct inode
*ino
)
2340 return (ino
->i_mode
& (S_ISGID
| S_IXGRP
)) == S_ISGID
;
2344 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2345 * otherwise these will be advisory locks
2348 static inline int mandatory_lock(struct inode
*ino
)
2350 return IS_MANDLOCK(ino
) && __mandatory_lock(ino
);
2353 static inline int locks_verify_locked(struct file
*file
)
2355 if (mandatory_lock(locks_inode(file
)))
2356 return locks_mandatory_locked(file
);
2360 static inline int locks_verify_truncate(struct inode
*inode
,
2364 if (!inode
->i_flctx
|| !mandatory_lock(inode
))
2367 if (size
< inode
->i_size
) {
2368 return locks_mandatory_area(inode
, f
, size
, inode
->i_size
- 1,
2371 return locks_mandatory_area(inode
, f
, inode
->i_size
, size
- 1,
2376 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2378 static inline int locks_mandatory_locked(struct file
*file
)
2383 static inline int locks_mandatory_area(struct inode
*inode
, struct file
*filp
,
2384 loff_t start
, loff_t end
, unsigned char type
)
2389 static inline int __mandatory_lock(struct inode
*inode
)
2394 static inline int mandatory_lock(struct inode
*inode
)
2399 static inline int locks_verify_locked(struct file
*file
)
2404 static inline int locks_verify_truncate(struct inode
*inode
, struct file
*filp
,
2410 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2413 #ifdef CONFIG_FILE_LOCKING
2414 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2417 * Since this check is lockless, we must ensure that any refcounts
2418 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2419 * could end up racing with tasks trying to set a new lease on this
2423 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2424 return __break_lease(inode
, mode
, FL_LEASE
);
2428 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2431 * Since this check is lockless, we must ensure that any refcounts
2432 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2433 * could end up racing with tasks trying to set a new lease on this
2437 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2438 return __break_lease(inode
, mode
, FL_DELEG
);
2442 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2446 ret
= break_deleg(inode
, O_WRONLY
|O_NONBLOCK
);
2447 if (ret
== -EWOULDBLOCK
&& delegated_inode
) {
2448 *delegated_inode
= inode
;
2454 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2458 ret
= break_deleg(*delegated_inode
, O_WRONLY
);
2459 iput(*delegated_inode
);
2460 *delegated_inode
= NULL
;
2464 static inline int break_layout(struct inode
*inode
, bool wait
)
2467 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2468 return __break_lease(inode
,
2469 wait
? O_WRONLY
: O_WRONLY
| O_NONBLOCK
,
2474 #else /* !CONFIG_FILE_LOCKING */
2475 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2480 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2485 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2490 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2496 static inline int break_layout(struct inode
*inode
, bool wait
)
2501 #endif /* CONFIG_FILE_LOCKING */
2506 const char *name
; /* pointer to actual string */
2507 const __user
char *uptr
; /* original userland pointer */
2509 struct audit_names
*aname
;
2512 static_assert(offsetof(struct filename
, iname
) % sizeof(long) == 0);
2514 extern long vfs_truncate(const struct path
*, loff_t
);
2515 extern int do_truncate(struct dentry
*, loff_t start
, unsigned int time_attrs
,
2517 extern int vfs_fallocate(struct file
*file
, int mode
, loff_t offset
,
2519 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
2521 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
2522 extern struct file
*filp_open(const char *, int, umode_t
);
2523 extern struct file
*file_open_root(struct dentry
*, struct vfsmount
*,
2524 const char *, int, umode_t
);
2525 extern struct file
* dentry_open(const struct path
*, int, const struct cred
*);
2526 extern struct file
* open_with_fake_path(const struct path
*, int,
2527 struct inode
*, const struct cred
*);
2528 static inline struct file
*file_clone_open(struct file
*file
)
2530 return dentry_open(&file
->f_path
, file
->f_flags
, file
->f_cred
);
2532 extern int filp_close(struct file
*, fl_owner_t id
);
2534 extern struct filename
*getname_flags(const char __user
*, int, int *);
2535 extern struct filename
*getname(const char __user
*);
2536 extern struct filename
*getname_kernel(const char *);
2537 extern void putname(struct filename
*name
);
2539 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
2540 int (*open
)(struct inode
*, struct file
*));
2541 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
2545 extern int ioctl_preallocate(struct file
*filp
, void __user
*argp
);
2548 extern void __init
vfs_caches_init_early(void);
2549 extern void __init
vfs_caches_init(void);
2551 extern struct kmem_cache
*names_cachep
;
2553 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2554 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2557 extern int register_blkdev(unsigned int, const char *);
2558 extern void unregister_blkdev(unsigned int, const char *);
2559 extern void bdev_unhash_inode(dev_t dev
);
2560 extern struct block_device
*bdget(dev_t
);
2561 extern struct block_device
*bdgrab(struct block_device
*bdev
);
2562 extern void bd_set_size(struct block_device
*, loff_t size
);
2563 extern void bd_forget(struct inode
*inode
);
2564 extern void bdput(struct block_device
*);
2565 extern void invalidate_bdev(struct block_device
*);
2566 extern void iterate_bdevs(void (*)(struct block_device
*, void *), void *);
2567 extern int sync_blockdev(struct block_device
*bdev
);
2568 extern void kill_bdev(struct block_device
*);
2569 extern struct super_block
*freeze_bdev(struct block_device
*);
2570 extern void emergency_thaw_all(void);
2571 extern void emergency_thaw_bdev(struct super_block
*sb
);
2572 extern int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
);
2573 extern int fsync_bdev(struct block_device
*);
2575 extern struct super_block
*blockdev_superblock
;
2577 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2579 return sb
== blockdev_superblock
;
2582 static inline void bd_forget(struct inode
*inode
) {}
2583 static inline int sync_blockdev(struct block_device
*bdev
) { return 0; }
2584 static inline void kill_bdev(struct block_device
*bdev
) {}
2585 static inline void invalidate_bdev(struct block_device
*bdev
) {}
2587 static inline struct super_block
*freeze_bdev(struct block_device
*sb
)
2592 static inline int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
)
2597 static inline int emergency_thaw_bdev(struct super_block
*sb
)
2602 static inline void iterate_bdevs(void (*f
)(struct block_device
*, void *), void *arg
)
2606 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2611 extern int sync_filesystem(struct super_block
*);
2612 extern const struct file_operations def_blk_fops
;
2613 extern const struct file_operations def_chr_fops
;
2615 extern int ioctl_by_bdev(struct block_device
*, unsigned, unsigned long);
2616 extern int blkdev_ioctl(struct block_device
*, fmode_t
, unsigned, unsigned long);
2617 extern long compat_blkdev_ioctl(struct file
*, unsigned, unsigned long);
2618 extern int blkdev_get(struct block_device
*bdev
, fmode_t mode
, void *holder
);
2619 extern struct block_device
*blkdev_get_by_path(const char *path
, fmode_t mode
,
2621 extern struct block_device
*blkdev_get_by_dev(dev_t dev
, fmode_t mode
,
2623 extern struct block_device
*bd_start_claiming(struct block_device
*bdev
,
2625 extern void bd_finish_claiming(struct block_device
*bdev
,
2626 struct block_device
*whole
, void *holder
);
2627 extern void bd_abort_claiming(struct block_device
*bdev
,
2628 struct block_device
*whole
, void *holder
);
2629 extern void blkdev_put(struct block_device
*bdev
, fmode_t mode
);
2630 extern int __blkdev_reread_part(struct block_device
*bdev
);
2631 extern int blkdev_reread_part(struct block_device
*bdev
);
2634 extern int bd_link_disk_holder(struct block_device
*bdev
, struct gendisk
*disk
);
2635 extern void bd_unlink_disk_holder(struct block_device
*bdev
,
2636 struct gendisk
*disk
);
2638 static inline int bd_link_disk_holder(struct block_device
*bdev
,
2639 struct gendisk
*disk
)
2643 static inline void bd_unlink_disk_holder(struct block_device
*bdev
,
2644 struct gendisk
*disk
)
2651 #define CHRDEV_MAJOR_MAX 512
2652 /* Marks the bottom of the first segment of free char majors */
2653 #define CHRDEV_MAJOR_DYN_END 234
2654 /* Marks the top and bottom of the second segment of free char majors */
2655 #define CHRDEV_MAJOR_DYN_EXT_START 511
2656 #define CHRDEV_MAJOR_DYN_EXT_END 384
2658 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
2659 extern int register_chrdev_region(dev_t
, unsigned, const char *);
2660 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
2661 unsigned int count
, const char *name
,
2662 const struct file_operations
*fops
);
2663 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
2664 unsigned int count
, const char *name
);
2665 extern void unregister_chrdev_region(dev_t
, unsigned);
2666 extern void chrdev_show(struct seq_file
*,off_t
);
2668 static inline int register_chrdev(unsigned int major
, const char *name
,
2669 const struct file_operations
*fops
)
2671 return __register_chrdev(major
, 0, 256, name
, fops
);
2674 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2676 __unregister_chrdev(major
, 0, 256, name
);
2679 /* fs/block_dev.c */
2680 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2681 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2684 #define BLKDEV_MAJOR_MAX 512
2685 extern const char *__bdevname(dev_t
, char *buffer
);
2686 extern const char *bdevname(struct block_device
*bdev
, char *buffer
);
2687 extern struct block_device
*lookup_bdev(const char *);
2688 extern void blkdev_show(struct seq_file
*,off_t
);
2691 #define BLKDEV_MAJOR_MAX 0
2694 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2696 /* Invalid inode operations -- fs/bad_inode.c */
2697 extern void make_bad_inode(struct inode
*);
2698 extern bool is_bad_inode(struct inode
*);
2701 extern void check_disk_size_change(struct gendisk
*disk
,
2702 struct block_device
*bdev
, bool verbose
);
2703 extern int revalidate_disk(struct gendisk
*);
2704 extern int check_disk_change(struct block_device
*);
2705 extern int __invalidate_device(struct block_device
*, bool);
2706 extern int invalidate_partition(struct gendisk
*, int);
2708 unsigned long invalidate_mapping_pages(struct address_space
*mapping
,
2709 pgoff_t start
, pgoff_t end
);
2711 static inline void invalidate_remote_inode(struct inode
*inode
)
2713 if (S_ISREG(inode
->i_mode
) || S_ISDIR(inode
->i_mode
) ||
2714 S_ISLNK(inode
->i_mode
))
2715 invalidate_mapping_pages(inode
->i_mapping
, 0, -1);
2717 extern int invalidate_inode_pages2(struct address_space
*mapping
);
2718 extern int invalidate_inode_pages2_range(struct address_space
*mapping
,
2719 pgoff_t start
, pgoff_t end
);
2720 extern int write_inode_now(struct inode
*, int);
2721 extern int filemap_fdatawrite(struct address_space
*);
2722 extern int filemap_flush(struct address_space
*);
2723 extern int filemap_fdatawait_keep_errors(struct address_space
*mapping
);
2724 extern int filemap_fdatawait_range(struct address_space
*, loff_t lstart
,
2726 extern int filemap_fdatawait_range_keep_errors(struct address_space
*mapping
,
2727 loff_t start_byte
, loff_t end_byte
);
2729 static inline int filemap_fdatawait(struct address_space
*mapping
)
2731 return filemap_fdatawait_range(mapping
, 0, LLONG_MAX
);
2734 extern bool filemap_range_has_page(struct address_space
*, loff_t lstart
,
2736 extern int filemap_write_and_wait(struct address_space
*mapping
);
2737 extern int filemap_write_and_wait_range(struct address_space
*mapping
,
2738 loff_t lstart
, loff_t lend
);
2739 extern int __filemap_fdatawrite_range(struct address_space
*mapping
,
2740 loff_t start
, loff_t end
, int sync_mode
);
2741 extern int filemap_fdatawrite_range(struct address_space
*mapping
,
2742 loff_t start
, loff_t end
);
2743 extern int filemap_check_errors(struct address_space
*mapping
);
2744 extern void __filemap_set_wb_err(struct address_space
*mapping
, int err
);
2746 extern int __must_check
file_fdatawait_range(struct file
*file
, loff_t lstart
,
2748 extern int __must_check
file_check_and_advance_wb_err(struct file
*file
);
2749 extern int __must_check
file_write_and_wait_range(struct file
*file
,
2750 loff_t start
, loff_t end
);
2752 static inline int file_write_and_wait(struct file
*file
)
2754 return file_write_and_wait_range(file
, 0, LLONG_MAX
);
2758 * filemap_set_wb_err - set a writeback error on an address_space
2759 * @mapping: mapping in which to set writeback error
2760 * @err: error to be set in mapping
2762 * When writeback fails in some way, we must record that error so that
2763 * userspace can be informed when fsync and the like are called. We endeavor
2764 * to report errors on any file that was open at the time of the error. Some
2765 * internal callers also need to know when writeback errors have occurred.
2767 * When a writeback error occurs, most filesystems will want to call
2768 * filemap_set_wb_err to record the error in the mapping so that it will be
2769 * automatically reported whenever fsync is called on the file.
2771 static inline void filemap_set_wb_err(struct address_space
*mapping
, int err
)
2773 /* Fastpath for common case of no error */
2775 __filemap_set_wb_err(mapping
, err
);
2779 * filemap_check_wb_error - has an error occurred since the mark was sampled?
2780 * @mapping: mapping to check for writeback errors
2781 * @since: previously-sampled errseq_t
2783 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2784 * the given value was sampled.
2786 * If it has then report the latest error set, otherwise return 0.
2788 static inline int filemap_check_wb_err(struct address_space
*mapping
,
2791 return errseq_check(&mapping
->wb_err
, since
);
2795 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2796 * @mapping: mapping to be sampled
2798 * Writeback errors are always reported relative to a particular sample point
2799 * in the past. This function provides those sample points.
2801 static inline errseq_t
filemap_sample_wb_err(struct address_space
*mapping
)
2803 return errseq_sample(&mapping
->wb_err
);
2806 static inline int filemap_nr_thps(struct address_space
*mapping
)
2808 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
2809 return atomic_read(&mapping
->nr_thps
);
2815 static inline void filemap_nr_thps_inc(struct address_space
*mapping
)
2817 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
2818 atomic_inc(&mapping
->nr_thps
);
2824 static inline void filemap_nr_thps_dec(struct address_space
*mapping
)
2826 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
2827 atomic_dec(&mapping
->nr_thps
);
2833 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
2835 extern int vfs_fsync(struct file
*file
, int datasync
);
2837 extern int sync_file_range(struct file
*file
, loff_t offset
, loff_t nbytes
,
2838 unsigned int flags
);
2841 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2842 * to already be updated for the write, and will return either the amount
2843 * of bytes passed in, or an error if syncing the file failed.
2845 static inline ssize_t
generic_write_sync(struct kiocb
*iocb
, ssize_t count
)
2847 if (iocb
->ki_flags
& IOCB_DSYNC
) {
2848 int ret
= vfs_fsync_range(iocb
->ki_filp
,
2849 iocb
->ki_pos
- count
, iocb
->ki_pos
- 1,
2850 (iocb
->ki_flags
& IOCB_SYNC
) ? 0 : 1);
2858 extern void emergency_sync(void);
2859 extern void emergency_remount(void);
2861 extern sector_t
bmap(struct inode
*, sector_t
);
2863 extern int notify_change(struct dentry
*, struct iattr
*, struct inode
**);
2864 extern int inode_permission(struct inode
*, int);
2865 extern int generic_permission(struct inode
*, int);
2866 extern int __check_sticky(struct inode
*dir
, struct inode
*inode
);
2868 static inline bool execute_ok(struct inode
*inode
)
2870 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
2873 static inline void file_start_write(struct file
*file
)
2875 if (!S_ISREG(file_inode(file
)->i_mode
))
2877 __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, true);
2880 static inline bool file_start_write_trylock(struct file
*file
)
2882 if (!S_ISREG(file_inode(file
)->i_mode
))
2884 return __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, false);
2887 static inline void file_end_write(struct file
*file
)
2889 if (!S_ISREG(file_inode(file
)->i_mode
))
2891 __sb_end_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
);
2895 * get_write_access() gets write permission for a file.
2896 * put_write_access() releases this write permission.
2897 * This is used for regular files.
2898 * We cannot support write (and maybe mmap read-write shared) accesses and
2899 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2900 * can have the following values:
2901 * 0: no writers, no VM_DENYWRITE mappings
2902 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2903 * > 0: (i_writecount) users are writing to the file.
2905 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2906 * except for the cases where we don't hold i_writecount yet. Then we need to
2907 * use {get,deny}_write_access() - these functions check the sign and refuse
2908 * to do the change if sign is wrong.
2910 static inline int get_write_access(struct inode
*inode
)
2912 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2914 static inline int deny_write_access(struct file
*file
)
2916 struct inode
*inode
= file_inode(file
);
2917 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2919 static inline void put_write_access(struct inode
* inode
)
2921 atomic_dec(&inode
->i_writecount
);
2923 static inline void allow_write_access(struct file
*file
)
2926 atomic_inc(&file_inode(file
)->i_writecount
);
2928 static inline bool inode_is_open_for_write(const struct inode
*inode
)
2930 return atomic_read(&inode
->i_writecount
) > 0;
2933 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
2934 static inline void i_readcount_dec(struct inode
*inode
)
2936 BUG_ON(!atomic_read(&inode
->i_readcount
));
2937 atomic_dec(&inode
->i_readcount
);
2939 static inline void i_readcount_inc(struct inode
*inode
)
2941 atomic_inc(&inode
->i_readcount
);
2944 static inline void i_readcount_dec(struct inode
*inode
)
2948 static inline void i_readcount_inc(struct inode
*inode
)
2953 extern int do_pipe_flags(int *, int);
2955 #define __kernel_read_file_id(id) \
2956 id(UNKNOWN, unknown) \
2957 id(FIRMWARE, firmware) \
2958 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2959 id(MODULE, kernel-module) \
2960 id(KEXEC_IMAGE, kexec-image) \
2961 id(KEXEC_INITRAMFS, kexec-initramfs) \
2962 id(POLICY, security-policy) \
2963 id(X509_CERTIFICATE, x509-certificate) \
2966 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2967 #define __fid_stringify(dummy, str) #str,
2969 enum kernel_read_file_id
{
2970 __kernel_read_file_id(__fid_enumify
)
2973 static const char * const kernel_read_file_str
[] = {
2974 __kernel_read_file_id(__fid_stringify
)
2977 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id
)
2979 if ((unsigned)id
>= READING_MAX_ID
)
2980 return kernel_read_file_str
[READING_UNKNOWN
];
2982 return kernel_read_file_str
[id
];
2985 extern int kernel_read_file(struct file
*, void **, loff_t
*, loff_t
,
2986 enum kernel_read_file_id
);
2987 extern int kernel_read_file_from_path(const char *, void **, loff_t
*, loff_t
,
2988 enum kernel_read_file_id
);
2989 extern int kernel_read_file_from_fd(int, void **, loff_t
*, loff_t
,
2990 enum kernel_read_file_id
);
2991 extern ssize_t
kernel_read(struct file
*, void *, size_t, loff_t
*);
2992 extern ssize_t
kernel_write(struct file
*, const void *, size_t, loff_t
*);
2993 extern ssize_t
__kernel_write(struct file
*, const void *, size_t, loff_t
*);
2994 extern struct file
* open_exec(const char *);
2996 /* fs/dcache.c -- generic fs support functions */
2997 extern bool is_subdir(struct dentry
*, struct dentry
*);
2998 extern bool path_is_under(const struct path
*, const struct path
*);
3000 extern char *file_path(struct file
*, char *, int);
3002 #include <linux/err.h>
3004 /* needed for stackable file system support */
3005 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int whence
);
3007 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int whence
);
3009 extern int inode_init_always(struct super_block
*, struct inode
*);
3010 extern void inode_init_once(struct inode
*);
3011 extern void address_space_init_once(struct address_space
*mapping
);
3012 extern struct inode
* igrab(struct inode
*);
3013 extern ino_t
iunique(struct super_block
*, ino_t
);
3014 extern int inode_needs_sync(struct inode
*inode
);
3015 extern int generic_delete_inode(struct inode
*inode
);
3016 static inline int generic_drop_inode(struct inode
*inode
)
3018 return !inode
->i_nlink
|| inode_unhashed(inode
);
3021 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
3022 unsigned long hashval
, int (*test
)(struct inode
*, void *),
3024 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
3025 int (*test
)(struct inode
*, void *), void *data
);
3026 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
3028 extern struct inode
*inode_insert5(struct inode
*inode
, unsigned long hashval
,
3029 int (*test
)(struct inode
*, void *),
3030 int (*set
)(struct inode
*, void *),
3032 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
3033 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
3034 extern struct inode
*find_inode_nowait(struct super_block
*,
3036 int (*match
)(struct inode
*,
3037 unsigned long, void *),
3039 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
3040 extern int insert_inode_locked(struct inode
*);
3041 #ifdef CONFIG_DEBUG_LOCK_ALLOC
3042 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
3044 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
3046 extern void unlock_new_inode(struct inode
*);
3047 extern void discard_new_inode(struct inode
*);
3048 extern unsigned int get_next_ino(void);
3049 extern void evict_inodes(struct super_block
*sb
);
3051 extern void __iget(struct inode
* inode
);
3052 extern void iget_failed(struct inode
*);
3053 extern void clear_inode(struct inode
*);
3054 extern void __destroy_inode(struct inode
*);
3055 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
3056 extern struct inode
*new_inode(struct super_block
*sb
);
3057 extern void free_inode_nonrcu(struct inode
*inode
);
3058 extern int should_remove_suid(struct dentry
*);
3059 extern int file_remove_privs(struct file
*);
3061 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
3062 static inline void insert_inode_hash(struct inode
*inode
)
3064 __insert_inode_hash(inode
, inode
->i_ino
);
3067 extern void __remove_inode_hash(struct inode
*);
3068 static inline void remove_inode_hash(struct inode
*inode
)
3070 if (!inode_unhashed(inode
) && !hlist_fake(&inode
->i_hash
))
3071 __remove_inode_hash(inode
);
3074 extern void inode_sb_list_add(struct inode
*inode
);
3077 extern int bdev_read_only(struct block_device
*);
3079 extern int set_blocksize(struct block_device
*, int);
3080 extern int sb_set_blocksize(struct super_block
*, int);
3081 extern int sb_min_blocksize(struct super_block
*, int);
3083 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
3084 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
3085 extern ssize_t
generic_write_checks(struct kiocb
*, struct iov_iter
*);
3086 extern int generic_remap_checks(struct file
*file_in
, loff_t pos_in
,
3087 struct file
*file_out
, loff_t pos_out
,
3088 loff_t
*count
, unsigned int remap_flags
);
3089 extern int generic_file_rw_checks(struct file
*file_in
, struct file
*file_out
);
3090 extern int generic_copy_file_checks(struct file
*file_in
, loff_t pos_in
,
3091 struct file
*file_out
, loff_t pos_out
,
3092 size_t *count
, unsigned int flags
);
3093 extern ssize_t
generic_file_read_iter(struct kiocb
*, struct iov_iter
*);
3094 extern ssize_t
__generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
3095 extern ssize_t
generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
3096 extern ssize_t
generic_file_direct_write(struct kiocb
*, struct iov_iter
*);
3097 extern ssize_t
generic_perform_write(struct file
*, struct iov_iter
*, loff_t
);
3099 ssize_t
vfs_iter_read(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
3101 ssize_t
vfs_iter_write(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
3104 /* fs/block_dev.c */
3105 extern ssize_t
blkdev_read_iter(struct kiocb
*iocb
, struct iov_iter
*to
);
3106 extern ssize_t
blkdev_write_iter(struct kiocb
*iocb
, struct iov_iter
*from
);
3107 extern int blkdev_fsync(struct file
*filp
, loff_t start
, loff_t end
,
3109 extern void block_sync_page(struct page
*page
);
3112 extern ssize_t
generic_file_splice_read(struct file
*, loff_t
*,
3113 struct pipe_inode_info
*, size_t, unsigned int);
3114 extern ssize_t
iter_file_splice_write(struct pipe_inode_info
*,
3115 struct file
*, loff_t
*, size_t, unsigned int);
3116 extern ssize_t
generic_splice_sendpage(struct pipe_inode_info
*pipe
,
3117 struct file
*out
, loff_t
*, size_t len
, unsigned int flags
);
3118 extern long do_splice_direct(struct file
*in
, loff_t
*ppos
, struct file
*out
,
3119 loff_t
*opos
, size_t len
, unsigned int flags
);
3123 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
3124 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int whence
);
3125 extern loff_t
no_llseek(struct file
*file
, loff_t offset
, int whence
);
3126 extern loff_t
vfs_setpos(struct file
*file
, loff_t offset
, loff_t maxsize
);
3127 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int whence
);
3128 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
3129 int whence
, loff_t maxsize
, loff_t eof
);
3130 extern loff_t
fixed_size_llseek(struct file
*file
, loff_t offset
,
3131 int whence
, loff_t size
);
3132 extern loff_t
no_seek_end_llseek_size(struct file
*, loff_t
, int, loff_t
);
3133 extern loff_t
no_seek_end_llseek(struct file
*, loff_t
, int);
3134 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
3135 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
3136 extern int stream_open(struct inode
* inode
, struct file
* filp
);
3139 typedef void (dio_submit_t
)(struct bio
*bio
, struct inode
*inode
,
3140 loff_t file_offset
);
3143 /* need locking between buffered and direct access */
3146 /* filesystem does not support filling holes */
3147 DIO_SKIP_HOLES
= 0x02,
3150 void dio_end_io(struct bio
*bio
);
3151 void dio_warn_stale_pagecache(struct file
*filp
);
3153 ssize_t
__blockdev_direct_IO(struct kiocb
*iocb
, struct inode
*inode
,
3154 struct block_device
*bdev
, struct iov_iter
*iter
,
3155 get_block_t get_block
,
3156 dio_iodone_t end_io
, dio_submit_t submit_io
,
3159 static inline ssize_t
blockdev_direct_IO(struct kiocb
*iocb
,
3160 struct inode
*inode
,
3161 struct iov_iter
*iter
,
3162 get_block_t get_block
)
3164 return __blockdev_direct_IO(iocb
, inode
, inode
->i_sb
->s_bdev
, iter
,
3165 get_block
, NULL
, NULL
, DIO_LOCKING
| DIO_SKIP_HOLES
);
3169 void inode_dio_wait(struct inode
*inode
);
3172 * inode_dio_begin - signal start of a direct I/O requests
3173 * @inode: inode the direct I/O happens on
3175 * This is called once we've finished processing a direct I/O request,
3176 * and is used to wake up callers waiting for direct I/O to be quiesced.
3178 static inline void inode_dio_begin(struct inode
*inode
)
3180 atomic_inc(&inode
->i_dio_count
);
3184 * inode_dio_end - signal finish of a direct I/O requests
3185 * @inode: inode the direct I/O happens on
3187 * This is called once we've finished processing a direct I/O request,
3188 * and is used to wake up callers waiting for direct I/O to be quiesced.
3190 static inline void inode_dio_end(struct inode
*inode
)
3192 if (atomic_dec_and_test(&inode
->i_dio_count
))
3193 wake_up_bit(&inode
->i_state
, __I_DIO_WAKEUP
);
3196 extern void inode_set_flags(struct inode
*inode
, unsigned int flags
,
3199 extern const struct file_operations generic_ro_fops
;
3201 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3203 extern int readlink_copy(char __user
*, int, const char *);
3204 extern int page_readlink(struct dentry
*, char __user
*, int);
3205 extern const char *page_get_link(struct dentry
*, struct inode
*,
3206 struct delayed_call
*);
3207 extern void page_put_link(void *);
3208 extern int __page_symlink(struct inode
*inode
, const char *symname
, int len
,
3210 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
3211 extern const struct inode_operations page_symlink_inode_operations
;
3212 extern void kfree_link(void *);
3213 extern void generic_fillattr(struct inode
*, struct kstat
*);
3214 extern int vfs_getattr_nosec(const struct path
*, struct kstat
*, u32
, unsigned int);
3215 extern int vfs_getattr(const struct path
*, struct kstat
*, u32
, unsigned int);
3216 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
3217 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
3218 void __inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
3219 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
3220 static inline loff_t
__inode_get_bytes(struct inode
*inode
)
3222 return (((loff_t
)inode
->i_blocks
) << 9) + inode
->i_bytes
;
3224 loff_t
inode_get_bytes(struct inode
*inode
);
3225 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
3226 const char *simple_get_link(struct dentry
*, struct inode
*,
3227 struct delayed_call
*);
3228 extern const struct inode_operations simple_symlink_inode_operations
;
3230 extern int iterate_dir(struct file
*, struct dir_context
*);
3232 extern int vfs_statx(int, const char __user
*, int, struct kstat
*, u32
);
3233 extern int vfs_statx_fd(unsigned int, struct kstat
*, u32
, unsigned int);
3235 static inline int vfs_stat(const char __user
*filename
, struct kstat
*stat
)
3237 return vfs_statx(AT_FDCWD
, filename
, AT_NO_AUTOMOUNT
,
3238 stat
, STATX_BASIC_STATS
);
3240 static inline int vfs_lstat(const char __user
*name
, struct kstat
*stat
)
3242 return vfs_statx(AT_FDCWD
, name
, AT_SYMLINK_NOFOLLOW
| AT_NO_AUTOMOUNT
,
3243 stat
, STATX_BASIC_STATS
);
3245 static inline int vfs_fstatat(int dfd
, const char __user
*filename
,
3246 struct kstat
*stat
, int flags
)
3248 return vfs_statx(dfd
, filename
, flags
| AT_NO_AUTOMOUNT
,
3249 stat
, STATX_BASIC_STATS
);
3251 static inline int vfs_fstat(int fd
, struct kstat
*stat
)
3253 return vfs_statx_fd(fd
, stat
, STATX_BASIC_STATS
, 0);
3257 extern const char *vfs_get_link(struct dentry
*, struct delayed_call
*);
3258 extern int vfs_readlink(struct dentry
*, char __user
*, int);
3260 extern int __generic_block_fiemap(struct inode
*inode
,
3261 struct fiemap_extent_info
*fieinfo
,
3262 loff_t start
, loff_t len
,
3263 get_block_t
*get_block
);
3264 extern int generic_block_fiemap(struct inode
*inode
,
3265 struct fiemap_extent_info
*fieinfo
, u64 start
,
3266 u64 len
, get_block_t
*get_block
);
3268 extern struct file_system_type
*get_filesystem(struct file_system_type
*fs
);
3269 extern void put_filesystem(struct file_system_type
*fs
);
3270 extern struct file_system_type
*get_fs_type(const char *name
);
3271 extern struct super_block
*get_super(struct block_device
*);
3272 extern struct super_block
*get_super_thawed(struct block_device
*);
3273 extern struct super_block
*get_super_exclusive_thawed(struct block_device
*bdev
);
3274 extern struct super_block
*get_active_super(struct block_device
*bdev
);
3275 extern void drop_super(struct super_block
*sb
);
3276 extern void drop_super_exclusive(struct super_block
*sb
);
3277 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
3278 extern void iterate_supers_type(struct file_system_type
*,
3279 void (*)(struct super_block
*, void *), void *);
3281 extern int dcache_dir_open(struct inode
*, struct file
*);
3282 extern int dcache_dir_close(struct inode
*, struct file
*);
3283 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
3284 extern int dcache_readdir(struct file
*, struct dir_context
*);
3285 extern int simple_setattr(struct dentry
*, struct iattr
*);
3286 extern int simple_getattr(const struct path
*, struct kstat
*, u32
, unsigned int);
3287 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
3288 extern int simple_open(struct inode
*inode
, struct file
*file
);
3289 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
3290 extern int simple_unlink(struct inode
*, struct dentry
*);
3291 extern int simple_rmdir(struct inode
*, struct dentry
*);
3292 extern int simple_rename(struct inode
*, struct dentry
*,
3293 struct inode
*, struct dentry
*, unsigned int);
3294 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
3295 extern int noop_set_page_dirty(struct page
*page
);
3296 extern void noop_invalidatepage(struct page
*page
, unsigned int offset
,
3297 unsigned int length
);
3298 extern ssize_t
noop_direct_IO(struct kiocb
*iocb
, struct iov_iter
*iter
);
3299 extern int simple_empty(struct dentry
*);
3300 extern int simple_readpage(struct file
*file
, struct page
*page
);
3301 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
3302 loff_t pos
, unsigned len
, unsigned flags
,
3303 struct page
**pagep
, void **fsdata
);
3304 extern int simple_write_end(struct file
*file
, struct address_space
*mapping
,
3305 loff_t pos
, unsigned len
, unsigned copied
,
3306 struct page
*page
, void *fsdata
);
3307 extern int always_delete_dentry(const struct dentry
*);
3308 extern struct inode
*alloc_anon_inode(struct super_block
*);
3309 extern int simple_nosetlease(struct file
*, long, struct file_lock
**, void **);
3310 extern const struct dentry_operations simple_dentry_operations
;
3312 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
3313 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
3314 extern const struct file_operations simple_dir_operations
;
3315 extern const struct inode_operations simple_dir_inode_operations
;
3316 extern void make_empty_dir_inode(struct inode
*inode
);
3317 extern bool is_empty_dir_inode(struct inode
*inode
);
3318 struct tree_descr
{ const char *name
; const struct file_operations
*ops
; int mode
; };
3319 struct dentry
*d_alloc_name(struct dentry
*, const char *);
3320 extern int simple_fill_super(struct super_block
*, unsigned long,
3321 const struct tree_descr
*);
3322 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
3323 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
3325 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
3326 loff_t
*ppos
, const void *from
, size_t available
);
3327 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
3328 const void __user
*from
, size_t count
);
3330 extern int __generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3331 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3333 extern int generic_check_addressable(unsigned, u64
);
3335 #ifdef CONFIG_MIGRATION
3336 extern int buffer_migrate_page(struct address_space
*,
3337 struct page
*, struct page
*,
3339 extern int buffer_migrate_page_norefs(struct address_space
*,
3340 struct page
*, struct page
*,
3343 #define buffer_migrate_page NULL
3344 #define buffer_migrate_page_norefs NULL
3347 extern int setattr_prepare(struct dentry
*, struct iattr
*);
3348 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
3349 extern void setattr_copy(struct inode
*inode
, const struct iattr
*attr
);
3351 extern int file_update_time(struct file
*file
);
3353 static inline bool io_is_direct(struct file
*filp
)
3355 return (filp
->f_flags
& O_DIRECT
) || IS_DAX(filp
->f_mapping
->host
);
3358 static inline bool vma_is_dax(struct vm_area_struct
*vma
)
3360 return vma
->vm_file
&& IS_DAX(vma
->vm_file
->f_mapping
->host
);
3363 static inline bool vma_is_fsdax(struct vm_area_struct
*vma
)
3365 struct inode
*inode
;
3369 if (!vma_is_dax(vma
))
3371 inode
= file_inode(vma
->vm_file
);
3372 if (S_ISCHR(inode
->i_mode
))
3373 return false; /* device-dax */
3377 static inline int iocb_flags(struct file
*file
)
3380 if (file
->f_flags
& O_APPEND
)
3382 if (io_is_direct(file
))
3384 if ((file
->f_flags
& O_DSYNC
) || IS_SYNC(file
->f_mapping
->host
))
3386 if (file
->f_flags
& __O_SYNC
)
3391 static inline int kiocb_set_rw_flags(struct kiocb
*ki
, rwf_t flags
)
3393 if (unlikely(flags
& ~RWF_SUPPORTED
))
3396 if (flags
& RWF_NOWAIT
) {
3397 if (!(ki
->ki_filp
->f_mode
& FMODE_NOWAIT
))
3399 ki
->ki_flags
|= IOCB_NOWAIT
;
3401 if (flags
& RWF_HIPRI
)
3402 ki
->ki_flags
|= IOCB_HIPRI
;
3403 if (flags
& RWF_DSYNC
)
3404 ki
->ki_flags
|= IOCB_DSYNC
;
3405 if (flags
& RWF_SYNC
)
3406 ki
->ki_flags
|= (IOCB_DSYNC
| IOCB_SYNC
);
3407 if (flags
& RWF_APPEND
)
3408 ki
->ki_flags
|= IOCB_APPEND
;
3412 static inline ino_t
parent_ino(struct dentry
*dentry
)
3417 * Don't strictly need d_lock here? If the parent ino could change
3418 * then surely we'd have a deeper race in the caller?
3420 spin_lock(&dentry
->d_lock
);
3421 res
= dentry
->d_parent
->d_inode
->i_ino
;
3422 spin_unlock(&dentry
->d_lock
);
3426 /* Transaction based IO helpers */
3429 * An argresp is stored in an allocated page and holds the
3430 * size of the argument or response, along with its content
3432 struct simple_transaction_argresp
{
3437 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3439 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
3441 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
3442 size_t size
, loff_t
*pos
);
3443 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
3445 void simple_transaction_set(struct file
*file
, size_t n
);
3448 * simple attribute files
3450 * These attributes behave similar to those in sysfs:
3452 * Writing to an attribute immediately sets a value, an open file can be
3453 * written to multiple times.
3455 * Reading from an attribute creates a buffer from the value that might get
3456 * read with multiple read calls. When the attribute has been read
3457 * completely, no further read calls are possible until the file is opened
3460 * All attributes contain a text representation of a numeric value
3461 * that are accessed with the get() and set() functions.
3463 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3464 static int __fops ## _open(struct inode *inode, struct file *file) \
3466 __simple_attr_check_format(__fmt, 0ull); \
3467 return simple_attr_open(inode, file, __get, __set, __fmt); \
3469 static const struct file_operations __fops = { \
3470 .owner = THIS_MODULE, \
3471 .open = __fops ## _open, \
3472 .release = simple_attr_release, \
3473 .read = simple_attr_read, \
3474 .write = simple_attr_write, \
3475 .llseek = generic_file_llseek, \
3478 static inline __printf(1, 2)
3479 void __simple_attr_check_format(const char *fmt
, ...)
3481 /* don't do anything, just let the compiler check the arguments; */
3484 int simple_attr_open(struct inode
*inode
, struct file
*file
,
3485 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
3487 int simple_attr_release(struct inode
*inode
, struct file
*file
);
3488 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
3489 size_t len
, loff_t
*ppos
);
3490 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
3491 size_t len
, loff_t
*ppos
);
3494 int proc_nr_files(struct ctl_table
*table
, int write
,
3495 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3496 int proc_nr_dentry(struct ctl_table
*table
, int write
,
3497 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3498 int proc_nr_inodes(struct ctl_table
*table
, int write
,
3499 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3500 int __init
get_filesystem_list(char *buf
);
3502 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3503 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3505 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3506 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3507 (flag & __FMODE_NONOTIFY)))
3509 static inline bool is_sxid(umode_t mode
)
3511 return (mode
& S_ISUID
) || ((mode
& S_ISGID
) && (mode
& S_IXGRP
));
3514 static inline int check_sticky(struct inode
*dir
, struct inode
*inode
)
3516 if (!(dir
->i_mode
& S_ISVTX
))
3519 return __check_sticky(dir
, inode
);
3522 static inline void inode_has_no_xattr(struct inode
*inode
)
3524 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& SB_NOSEC
))
3525 inode
->i_flags
|= S_NOSEC
;
3528 static inline bool is_root_inode(struct inode
*inode
)
3530 return inode
== inode
->i_sb
->s_root
->d_inode
;
3533 static inline bool dir_emit(struct dir_context
*ctx
,
3534 const char *name
, int namelen
,
3535 u64 ino
, unsigned type
)
3537 return ctx
->actor(ctx
, name
, namelen
, ctx
->pos
, ino
, type
) == 0;
3539 static inline bool dir_emit_dot(struct file
*file
, struct dir_context
*ctx
)
3541 return ctx
->actor(ctx
, ".", 1, ctx
->pos
,
3542 file
->f_path
.dentry
->d_inode
->i_ino
, DT_DIR
) == 0;
3544 static inline bool dir_emit_dotdot(struct file
*file
, struct dir_context
*ctx
)
3546 return ctx
->actor(ctx
, "..", 2, ctx
->pos
,
3547 parent_ino(file
->f_path
.dentry
), DT_DIR
) == 0;
3549 static inline bool dir_emit_dots(struct file
*file
, struct dir_context
*ctx
)
3551 if (ctx
->pos
== 0) {
3552 if (!dir_emit_dot(file
, ctx
))
3556 if (ctx
->pos
== 1) {
3557 if (!dir_emit_dotdot(file
, ctx
))
3563 static inline bool dir_relax(struct inode
*inode
)
3565 inode_unlock(inode
);
3567 return !IS_DEADDIR(inode
);
3570 static inline bool dir_relax_shared(struct inode
*inode
)
3572 inode_unlock_shared(inode
);
3573 inode_lock_shared(inode
);
3574 return !IS_DEADDIR(inode
);
3577 extern bool path_noexec(const struct path
*path
);
3578 extern void inode_nohighmem(struct inode
*inode
);
3581 extern int vfs_fadvise(struct file
*file
, loff_t offset
, loff_t len
,
3583 extern int generic_fadvise(struct file
*file
, loff_t offset
, loff_t len
,
3586 #if defined(CONFIG_IO_URING)
3587 extern struct sock
*io_uring_get_socket(struct file
*file
);
3589 static inline struct sock
*io_uring_get_socket(struct file
*file
)
3595 int vfs_ioc_setflags_prepare(struct inode
*inode
, unsigned int oldflags
,
3596 unsigned int flags
);
3598 int vfs_ioc_fssetxattr_check(struct inode
*inode
, const struct fsxattr
*old_fa
,
3599 struct fsxattr
*fa
);
3601 static inline void simple_fill_fsxattr(struct fsxattr
*fa
, __u32 xflags
)
3603 memset(fa
, 0, sizeof(*fa
));
3604 fa
->fsx_xflags
= xflags
;
3608 * Flush file data before changing attributes. Caller must hold any locks
3609 * required to prevent further writes to this file until we're done setting
3612 static inline int inode_drain_writes(struct inode
*inode
)
3614 inode_dio_wait(inode
);
3615 return filemap_write_and_wait(inode
->i_mapping
);
3618 #endif /* _LINUX_FS_H */