1 /* SPDX-License-Identifier: GPL-2.0 */
5 #include <linux/linkage.h>
6 #include <linux/wait_bit.h>
7 #include <linux/kdev_t.h>
8 #include <linux/dcache.h>
9 #include <linux/path.h>
10 #include <linux/stat.h>
11 #include <linux/cache.h>
12 #include <linux/list.h>
13 #include <linux/list_lru.h>
14 #include <linux/llist.h>
15 #include <linux/radix-tree.h>
16 #include <linux/xarray.h>
17 #include <linux/rbtree.h>
18 #include <linux/init.h>
19 #include <linux/pid.h>
20 #include <linux/bug.h>
21 #include <linux/mutex.h>
22 #include <linux/rwsem.h>
23 #include <linux/mm_types.h>
24 #include <linux/capability.h>
25 #include <linux/semaphore.h>
26 #include <linux/fcntl.h>
27 #include <linux/fiemap.h>
28 #include <linux/rculist_bl.h>
29 #include <linux/atomic.h>
30 #include <linux/shrinker.h>
31 #include <linux/migrate_mode.h>
32 #include <linux/uidgid.h>
33 #include <linux/lockdep.h>
34 #include <linux/percpu-rwsem.h>
35 #include <linux/workqueue.h>
36 #include <linux/delayed_call.h>
37 #include <linux/uuid.h>
38 #include <linux/errseq.h>
39 #include <linux/ioprio.h>
41 #include <asm/byteorder.h>
42 #include <uapi/linux/fs.h>
44 struct backing_dev_info
;
47 struct export_operations
;
52 struct pipe_inode_info
;
53 struct poll_table_struct
;
55 struct vm_area_struct
;
58 struct swap_info_struct
;
60 struct workqueue_struct
;
63 struct fscrypt_operations
;
65 extern void __init
inode_init(void);
66 extern void __init
inode_init_early(void);
67 extern void __init
files_init(void);
68 extern void __init
files_maxfiles_init(void);
70 extern struct files_stat_struct files_stat
;
71 extern unsigned long get_max_files(void);
72 extern unsigned int sysctl_nr_open
;
73 extern struct inodes_stat_t inodes_stat
;
74 extern int leases_enable
, lease_break_time
;
75 extern int sysctl_protected_symlinks
;
76 extern int sysctl_protected_hardlinks
;
78 typedef __kernel_rwf_t rwf_t
;
81 typedef int (get_block_t
)(struct inode
*inode
, sector_t iblock
,
82 struct buffer_head
*bh_result
, int create
);
83 typedef int (dio_iodone_t
)(struct kiocb
*iocb
, loff_t offset
,
84 ssize_t bytes
, void *private);
86 #define MAY_EXEC 0x00000001
87 #define MAY_WRITE 0x00000002
88 #define MAY_READ 0x00000004
89 #define MAY_APPEND 0x00000008
90 #define MAY_ACCESS 0x00000010
91 #define MAY_OPEN 0x00000020
92 #define MAY_CHDIR 0x00000040
93 /* called from RCU mode, don't block */
94 #define MAY_NOT_BLOCK 0x00000080
97 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
98 * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
101 /* file is open for reading */
102 #define FMODE_READ ((__force fmode_t)0x1)
103 /* file is open for writing */
104 #define FMODE_WRITE ((__force fmode_t)0x2)
105 /* file is seekable */
106 #define FMODE_LSEEK ((__force fmode_t)0x4)
107 /* file can be accessed using pread */
108 #define FMODE_PREAD ((__force fmode_t)0x8)
109 /* file can be accessed using pwrite */
110 #define FMODE_PWRITE ((__force fmode_t)0x10)
111 /* File is opened for execution with sys_execve / sys_uselib */
112 #define FMODE_EXEC ((__force fmode_t)0x20)
113 /* File is opened with O_NDELAY (only set for block devices) */
114 #define FMODE_NDELAY ((__force fmode_t)0x40)
115 /* File is opened with O_EXCL (only set for block devices) */
116 #define FMODE_EXCL ((__force fmode_t)0x80)
117 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
118 (specialy hack for floppy.c) */
119 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
120 /* 32bit hashes as llseek() offset (for directories) */
121 #define FMODE_32BITHASH ((__force fmode_t)0x200)
122 /* 64bit hashes as llseek() offset (for directories) */
123 #define FMODE_64BITHASH ((__force fmode_t)0x400)
126 * Don't update ctime and mtime.
128 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
129 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
131 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
133 /* Expect random access pattern */
134 #define FMODE_RANDOM ((__force fmode_t)0x1000)
136 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
137 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
139 /* File is opened with O_PATH; almost nothing can be done with it */
140 #define FMODE_PATH ((__force fmode_t)0x4000)
142 /* File needs atomic accesses to f_pos */
143 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
144 /* Write access to underlying fs */
145 #define FMODE_WRITER ((__force fmode_t)0x10000)
146 /* Has read method(s) */
147 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
148 /* Has write method(s) */
149 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
151 /* File was opened by fanotify and shouldn't generate fanotify events */
152 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
154 /* File is capable of returning -EAGAIN if I/O will block */
155 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
158 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
159 * that indicates that they should check the contents of the iovec are
160 * valid, but not check the memory that the iovec elements
163 #define CHECK_IOVEC_ONLY -1
166 * Attribute flags. These should be or-ed together to figure out what
169 #define ATTR_MODE (1 << 0)
170 #define ATTR_UID (1 << 1)
171 #define ATTR_GID (1 << 2)
172 #define ATTR_SIZE (1 << 3)
173 #define ATTR_ATIME (1 << 4)
174 #define ATTR_MTIME (1 << 5)
175 #define ATTR_CTIME (1 << 6)
176 #define ATTR_ATIME_SET (1 << 7)
177 #define ATTR_MTIME_SET (1 << 8)
178 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
179 #define ATTR_ATTR_FLAG (1 << 10)
180 #define ATTR_KILL_SUID (1 << 11)
181 #define ATTR_KILL_SGID (1 << 12)
182 #define ATTR_FILE (1 << 13)
183 #define ATTR_KILL_PRIV (1 << 14)
184 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
185 #define ATTR_TIMES_SET (1 << 16)
186 #define ATTR_TOUCH (1 << 17)
189 * Whiteout is represented by a char device. The following constants define the
190 * mode and device number to use.
192 #define WHITEOUT_MODE 0
193 #define WHITEOUT_DEV 0
196 * This is the Inode Attributes structure, used for notify_change(). It
197 * uses the above definitions as flags, to know which values have changed.
198 * Also, in this manner, a Filesystem can look at only the values it cares
199 * about. Basically, these are the attributes that the VFS layer can
200 * request to change from the FS layer.
202 * Derek Atkins <warlord@MIT.EDU> 94-10-20
205 unsigned int ia_valid
;
210 struct timespec64 ia_atime
;
211 struct timespec64 ia_mtime
;
212 struct timespec64 ia_ctime
;
215 * Not an attribute, but an auxiliary info for filesystems wanting to
216 * implement an ftruncate() like method. NOTE: filesystem should
217 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
219 struct file
*ia_file
;
223 * Includes for diskquotas.
225 #include <linux/quota.h>
228 * Maximum number of layers of fs stack. Needs to be limited to
229 * prevent kernel stack overflow
231 #define FILESYSTEM_MAX_STACK_DEPTH 2
234 * enum positive_aop_returns - aop return codes with specific semantics
236 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
237 * completed, that the page is still locked, and
238 * should be considered active. The VM uses this hint
239 * to return the page to the active list -- it won't
240 * be a candidate for writeback again in the near
241 * future. Other callers must be careful to unlock
242 * the page if they get this return. Returned by
245 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
246 * unlocked it and the page might have been truncated.
247 * The caller should back up to acquiring a new page and
248 * trying again. The aop will be taking reasonable
249 * precautions not to livelock. If the caller held a page
250 * reference, it should drop it before retrying. Returned
253 * address_space_operation functions return these large constants to indicate
254 * special semantics to the caller. These are much larger than the bytes in a
255 * page to allow for functions that return the number of bytes operated on in a
259 enum positive_aop_returns
{
260 AOP_WRITEPAGE_ACTIVATE
= 0x80000,
261 AOP_TRUNCATED_PAGE
= 0x80001,
264 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
265 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
266 * helper code (eg buffer layer)
267 * to clear GFP_FS from alloc */
270 * oh the beauties of C type declarations.
273 struct address_space
;
274 struct writeback_control
;
277 * Write life time hint values.
280 WRITE_LIFE_NOT_SET
= 0,
281 WRITE_LIFE_NONE
= RWH_WRITE_LIFE_NONE
,
282 WRITE_LIFE_SHORT
= RWH_WRITE_LIFE_SHORT
,
283 WRITE_LIFE_MEDIUM
= RWH_WRITE_LIFE_MEDIUM
,
284 WRITE_LIFE_LONG
= RWH_WRITE_LIFE_LONG
,
285 WRITE_LIFE_EXTREME
= RWH_WRITE_LIFE_EXTREME
,
288 #define IOCB_EVENTFD (1 << 0)
289 #define IOCB_APPEND (1 << 1)
290 #define IOCB_DIRECT (1 << 2)
291 #define IOCB_HIPRI (1 << 3)
292 #define IOCB_DSYNC (1 << 4)
293 #define IOCB_SYNC (1 << 5)
294 #define IOCB_WRITE (1 << 6)
295 #define IOCB_NOWAIT (1 << 7)
298 struct file
*ki_filp
;
300 void (*ki_complete
)(struct kiocb
*iocb
, long ret
, long ret2
);
304 u16 ki_ioprio
; /* See linux/ioprio.h */
305 } __randomize_layout
;
307 static inline bool is_sync_kiocb(struct kiocb
*kiocb
)
309 return kiocb
->ki_complete
== NULL
;
313 * "descriptor" for what we're up to with a read.
314 * This allows us to use the same read code yet
315 * have multiple different users of the data that
316 * we read from a file.
318 * The simplest case just copies the data to user
331 typedef int (*read_actor_t
)(read_descriptor_t
*, struct page
*,
332 unsigned long, unsigned long);
334 struct address_space_operations
{
335 int (*writepage
)(struct page
*page
, struct writeback_control
*wbc
);
336 int (*readpage
)(struct file
*, struct page
*);
338 /* Write back some dirty pages from this mapping. */
339 int (*writepages
)(struct address_space
*, struct writeback_control
*);
341 /* Set a page dirty. Return true if this dirtied it */
342 int (*set_page_dirty
)(struct page
*page
);
344 int (*readpages
)(struct file
*filp
, struct address_space
*mapping
,
345 struct list_head
*pages
, unsigned nr_pages
);
347 int (*write_begin
)(struct file
*, struct address_space
*mapping
,
348 loff_t pos
, unsigned len
, unsigned flags
,
349 struct page
**pagep
, void **fsdata
);
350 int (*write_end
)(struct file
*, struct address_space
*mapping
,
351 loff_t pos
, unsigned len
, unsigned copied
,
352 struct page
*page
, void *fsdata
);
354 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
355 sector_t (*bmap
)(struct address_space
*, sector_t
);
356 void (*invalidatepage
) (struct page
*, unsigned int, unsigned int);
357 int (*releasepage
) (struct page
*, gfp_t
);
358 void (*freepage
)(struct page
*);
359 ssize_t (*direct_IO
)(struct kiocb
*, struct iov_iter
*iter
);
361 * migrate the contents of a page to the specified target. If
362 * migrate_mode is MIGRATE_ASYNC, it must not block.
364 int (*migratepage
) (struct address_space
*,
365 struct page
*, struct page
*, enum migrate_mode
);
366 bool (*isolate_page
)(struct page
*, isolate_mode_t
);
367 void (*putback_page
)(struct page
*);
368 int (*launder_page
) (struct page
*);
369 int (*is_partially_uptodate
) (struct page
*, unsigned long,
371 void (*is_dirty_writeback
) (struct page
*, bool *, bool *);
372 int (*error_remove_page
)(struct address_space
*, struct page
*);
374 /* swapfile support */
375 int (*swap_activate
)(struct swap_info_struct
*sis
, struct file
*file
,
377 void (*swap_deactivate
)(struct file
*file
);
380 extern const struct address_space_operations empty_aops
;
383 * pagecache_write_begin/pagecache_write_end must be used by general code
384 * to write into the pagecache.
386 int pagecache_write_begin(struct file
*, struct address_space
*mapping
,
387 loff_t pos
, unsigned len
, unsigned flags
,
388 struct page
**pagep
, void **fsdata
);
390 int pagecache_write_end(struct file
*, struct address_space
*mapping
,
391 loff_t pos
, unsigned len
, unsigned copied
,
392 struct page
*page
, void *fsdata
);
394 struct address_space
{
395 struct inode
*host
; /* owner: inode, block_device */
396 struct radix_tree_root i_pages
; /* cached pages */
397 atomic_t i_mmap_writable
;/* count VM_SHARED mappings */
398 struct rb_root_cached i_mmap
; /* tree of private and shared mappings */
399 struct rw_semaphore i_mmap_rwsem
; /* protect tree, count, list */
400 /* Protected by the i_pages lock */
401 unsigned long nrpages
; /* number of total pages */
402 /* number of shadow or DAX exceptional entries */
403 unsigned long nrexceptional
;
404 pgoff_t writeback_index
;/* writeback starts here */
405 const struct address_space_operations
*a_ops
; /* methods */
406 unsigned long flags
; /* error bits */
407 spinlock_t private_lock
; /* for use by the address_space */
408 gfp_t gfp_mask
; /* implicit gfp mask for allocations */
409 struct list_head private_list
; /* for use by the address_space */
410 void *private_data
; /* ditto */
412 } __attribute__((aligned(sizeof(long)))) __randomize_layout
;
414 * On most architectures that alignment is already the case; but
415 * must be enforced here for CRIS, to let the least significant bit
416 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
418 struct request_queue
;
420 struct block_device
{
421 dev_t bd_dev
; /* not a kdev_t - it's a search key */
423 struct inode
* bd_inode
; /* will die */
424 struct super_block
* bd_super
;
425 struct mutex bd_mutex
; /* open/close mutex */
429 bool bd_write_holder
;
431 struct list_head bd_holder_disks
;
433 struct block_device
* bd_contains
;
434 unsigned bd_block_size
;
436 struct hd_struct
* bd_part
;
437 /* number of times partitions within this device have been opened. */
438 unsigned bd_part_count
;
440 struct gendisk
* bd_disk
;
441 struct request_queue
* bd_queue
;
442 struct backing_dev_info
*bd_bdi
;
443 struct list_head bd_list
;
445 * Private data. You must have bd_claim'ed the block_device
446 * to use this. NOTE: bd_claim allows an owner to claim
447 * the same device multiple times, the owner must take special
448 * care to not mess up bd_private for that case.
450 unsigned long bd_private
;
452 /* The counter of freeze processes */
453 int bd_fsfreeze_count
;
454 /* Mutex for freeze */
455 struct mutex bd_fsfreeze_mutex
;
456 } __randomize_layout
;
459 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
462 #define PAGECACHE_TAG_DIRTY 0
463 #define PAGECACHE_TAG_WRITEBACK 1
464 #define PAGECACHE_TAG_TOWRITE 2
466 int mapping_tagged(struct address_space
*mapping
, int tag
);
468 static inline void i_mmap_lock_write(struct address_space
*mapping
)
470 down_write(&mapping
->i_mmap_rwsem
);
473 static inline void i_mmap_unlock_write(struct address_space
*mapping
)
475 up_write(&mapping
->i_mmap_rwsem
);
478 static inline void i_mmap_lock_read(struct address_space
*mapping
)
480 down_read(&mapping
->i_mmap_rwsem
);
483 static inline void i_mmap_unlock_read(struct address_space
*mapping
)
485 up_read(&mapping
->i_mmap_rwsem
);
489 * Might pages of this file be mapped into userspace?
491 static inline int mapping_mapped(struct address_space
*mapping
)
493 return !RB_EMPTY_ROOT(&mapping
->i_mmap
.rb_root
);
497 * Might pages of this file have been modified in userspace?
498 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
499 * marks vma as VM_SHARED if it is shared, and the file was opened for
500 * writing i.e. vma may be mprotected writable even if now readonly.
502 * If i_mmap_writable is negative, no new writable mappings are allowed. You
503 * can only deny writable mappings, if none exists right now.
505 static inline int mapping_writably_mapped(struct address_space
*mapping
)
507 return atomic_read(&mapping
->i_mmap_writable
) > 0;
510 static inline int mapping_map_writable(struct address_space
*mapping
)
512 return atomic_inc_unless_negative(&mapping
->i_mmap_writable
) ?
516 static inline void mapping_unmap_writable(struct address_space
*mapping
)
518 atomic_dec(&mapping
->i_mmap_writable
);
521 static inline int mapping_deny_writable(struct address_space
*mapping
)
523 return atomic_dec_unless_positive(&mapping
->i_mmap_writable
) ?
527 static inline void mapping_allow_writable(struct address_space
*mapping
)
529 atomic_inc(&mapping
->i_mmap_writable
);
533 * Use sequence counter to get consistent i_size on 32-bit processors.
535 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
536 #include <linux/seqlock.h>
537 #define __NEED_I_SIZE_ORDERED
538 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
540 #define i_size_ordered_init(inode) do { } while (0)
544 #define ACL_NOT_CACHED ((void *)(-1))
545 #define ACL_DONT_CACHE ((void *)(-3))
547 static inline struct posix_acl
*
548 uncached_acl_sentinel(struct task_struct
*task
)
550 return (void *)task
+ 1;
554 is_uncached_acl(struct posix_acl
*acl
)
556 return (long)acl
& 1;
559 #define IOP_FASTPERM 0x0001
560 #define IOP_LOOKUP 0x0002
561 #define IOP_NOFOLLOW 0x0004
562 #define IOP_XATTR 0x0008
563 #define IOP_DEFAULT_READLINK 0x0010
565 struct fsnotify_mark_connector
;
568 * Keep mostly read-only and often accessed (especially for
569 * the RCU path lookup and 'stat' data) fields at the beginning
570 * of the 'struct inode'
574 unsigned short i_opflags
;
577 unsigned int i_flags
;
579 #ifdef CONFIG_FS_POSIX_ACL
580 struct posix_acl
*i_acl
;
581 struct posix_acl
*i_default_acl
;
584 const struct inode_operations
*i_op
;
585 struct super_block
*i_sb
;
586 struct address_space
*i_mapping
;
588 #ifdef CONFIG_SECURITY
592 /* Stat data, not accessed from path walking */
595 * Filesystems may only read i_nlink directly. They shall use the
596 * following functions for modification:
598 * (set|clear|inc|drop)_nlink
599 * inode_(inc|dec)_link_count
602 const unsigned int i_nlink
;
603 unsigned int __i_nlink
;
607 struct timespec64 i_atime
;
608 struct timespec64 i_mtime
;
609 struct timespec64 i_ctime
;
610 spinlock_t i_lock
; /* i_blocks, i_bytes, maybe i_size */
611 unsigned short i_bytes
;
612 unsigned int i_blkbits
;
613 enum rw_hint i_write_hint
;
616 #ifdef __NEED_I_SIZE_ORDERED
617 seqcount_t i_size_seqcount
;
621 unsigned long i_state
;
622 struct rw_semaphore i_rwsem
;
624 unsigned long dirtied_when
; /* jiffies of first dirtying */
625 unsigned long dirtied_time_when
;
627 struct hlist_node i_hash
;
628 struct list_head i_io_list
; /* backing dev IO list */
629 #ifdef CONFIG_CGROUP_WRITEBACK
630 struct bdi_writeback
*i_wb
; /* the associated cgroup wb */
632 /* foreign inode detection, see wbc_detach_inode() */
634 u16 i_wb_frn_avg_time
;
635 u16 i_wb_frn_history
;
637 struct list_head i_lru
; /* inode LRU list */
638 struct list_head i_sb_list
;
639 struct list_head i_wb_list
; /* backing dev writeback list */
641 struct hlist_head i_dentry
;
642 struct rcu_head i_rcu
;
644 atomic64_t i_version
;
646 atomic_t i_dio_count
;
647 atomic_t i_writecount
;
649 atomic_t i_readcount
; /* struct files open RO */
651 const struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
652 struct file_lock_context
*i_flctx
;
653 struct address_space i_data
;
654 struct list_head i_devices
;
656 struct pipe_inode_info
*i_pipe
;
657 struct block_device
*i_bdev
;
665 #ifdef CONFIG_FSNOTIFY
666 __u32 i_fsnotify_mask
; /* all events this inode cares about */
667 struct fsnotify_mark_connector __rcu
*i_fsnotify_marks
;
670 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
671 struct fscrypt_info
*i_crypt_info
;
674 void *i_private
; /* fs or device private pointer */
675 } __randomize_layout
;
677 static inline unsigned int i_blocksize(const struct inode
*node
)
679 return (1 << node
->i_blkbits
);
682 static inline int inode_unhashed(struct inode
*inode
)
684 return hlist_unhashed(&inode
->i_hash
);
688 * inode->i_mutex nesting subclasses for the lock validator:
690 * 0: the object of the current VFS operation
694 * 4: second non-directory
695 * 5: second parent (when locking independent directories in rename)
697 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
698 * non-directories at once.
700 * The locking order between these classes is
701 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
703 enum inode_i_mutex_lock_class
713 static inline void inode_lock(struct inode
*inode
)
715 down_write(&inode
->i_rwsem
);
718 static inline void inode_unlock(struct inode
*inode
)
720 up_write(&inode
->i_rwsem
);
723 static inline void inode_lock_shared(struct inode
*inode
)
725 down_read(&inode
->i_rwsem
);
728 static inline void inode_unlock_shared(struct inode
*inode
)
730 up_read(&inode
->i_rwsem
);
733 static inline int inode_trylock(struct inode
*inode
)
735 return down_write_trylock(&inode
->i_rwsem
);
738 static inline int inode_trylock_shared(struct inode
*inode
)
740 return down_read_trylock(&inode
->i_rwsem
);
743 static inline int inode_is_locked(struct inode
*inode
)
745 return rwsem_is_locked(&inode
->i_rwsem
);
748 static inline void inode_lock_nested(struct inode
*inode
, unsigned subclass
)
750 down_write_nested(&inode
->i_rwsem
, subclass
);
753 static inline void inode_lock_shared_nested(struct inode
*inode
, unsigned subclass
)
755 down_read_nested(&inode
->i_rwsem
, subclass
);
758 void lock_two_nondirectories(struct inode
*, struct inode
*);
759 void unlock_two_nondirectories(struct inode
*, struct inode
*);
762 * NOTE: in a 32bit arch with a preemptable kernel and
763 * an UP compile the i_size_read/write must be atomic
764 * with respect to the local cpu (unlike with preempt disabled),
765 * but they don't need to be atomic with respect to other cpus like in
766 * true SMP (so they need either to either locally disable irq around
767 * the read or for example on x86 they can be still implemented as a
768 * cmpxchg8b without the need of the lock prefix). For SMP compiles
769 * and 64bit archs it makes no difference if preempt is enabled or not.
771 static inline loff_t
i_size_read(const struct inode
*inode
)
773 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
778 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
779 i_size
= inode
->i_size
;
780 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
782 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
786 i_size
= inode
->i_size
;
790 return inode
->i_size
;
795 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
796 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
797 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
799 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
801 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
803 write_seqcount_begin(&inode
->i_size_seqcount
);
804 inode
->i_size
= i_size
;
805 write_seqcount_end(&inode
->i_size_seqcount
);
807 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
809 inode
->i_size
= i_size
;
812 inode
->i_size
= i_size
;
816 static inline unsigned iminor(const struct inode
*inode
)
818 return MINOR(inode
->i_rdev
);
821 static inline unsigned imajor(const struct inode
*inode
)
823 return MAJOR(inode
->i_rdev
);
826 extern struct block_device
*I_BDEV(struct inode
*inode
);
829 rwlock_t lock
; /* protects pid, uid, euid fields */
830 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
831 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
832 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
833 int signum
; /* posix.1b rt signal to be delivered on IO */
837 * Track a single file's readahead state
839 struct file_ra_state
{
840 pgoff_t start
; /* where readahead started */
841 unsigned int size
; /* # of readahead pages */
842 unsigned int async_size
; /* do asynchronous readahead when
843 there are only # of pages ahead */
845 unsigned int ra_pages
; /* Maximum readahead window */
846 unsigned int mmap_miss
; /* Cache miss stat for mmap accesses */
847 loff_t prev_pos
; /* Cache last read() position */
851 * Check if @index falls in the readahead windows.
853 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
855 return (index
>= ra
->start
&&
856 index
< ra
->start
+ ra
->size
);
861 struct llist_node fu_llist
;
862 struct rcu_head fu_rcuhead
;
865 struct inode
*f_inode
; /* cached value */
866 const struct file_operations
*f_op
;
869 * Protects f_ep_links, f_flags.
870 * Must not be taken from IRQ context.
873 enum rw_hint f_write_hint
;
874 atomic_long_t f_count
;
875 unsigned int f_flags
;
877 struct mutex f_pos_lock
;
879 struct fown_struct f_owner
;
880 const struct cred
*f_cred
;
881 struct file_ra_state f_ra
;
884 #ifdef CONFIG_SECURITY
887 /* needed for tty driver, and maybe others */
891 /* Used by fs/eventpoll.c to link all the hooks to this file */
892 struct list_head f_ep_links
;
893 struct list_head f_tfile_llink
;
894 #endif /* #ifdef CONFIG_EPOLL */
895 struct address_space
*f_mapping
;
898 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
903 /* file identifier */
904 unsigned char f_handle
[0];
907 static inline struct file
*get_file(struct file
*f
)
909 atomic_long_inc(&f
->f_count
);
912 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
913 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
914 #define file_count(x) atomic_long_read(&(x)->f_count)
916 #define MAX_NON_LFS ((1UL<<31) - 1)
918 /* Page cache limit. The filesystems should put that into their s_maxbytes
919 limits, otherwise bad things can happen in VM. */
920 #if BITS_PER_LONG==32
921 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
922 #elif BITS_PER_LONG==64
923 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
928 #define FL_DELEG 4 /* NFSv4 delegation */
929 #define FL_ACCESS 8 /* not trying to lock, just looking */
930 #define FL_EXISTS 16 /* when unlocking, test for existence */
931 #define FL_LEASE 32 /* lease held on this file */
932 #define FL_CLOSE 64 /* unlock on close */
933 #define FL_SLEEP 128 /* A blocking lock */
934 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
935 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
936 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
937 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
939 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
942 * Special return value from posix_lock_file() and vfs_lock_file() for
943 * asynchronous locking.
945 #define FILE_LOCK_DEFERRED 1
947 /* legacy typedef, should eventually be removed */
948 typedef void *fl_owner_t
;
952 struct file_lock_operations
{
953 void (*fl_copy_lock
)(struct file_lock
*, struct file_lock
*);
954 void (*fl_release_private
)(struct file_lock
*);
957 struct lock_manager_operations
{
958 int (*lm_compare_owner
)(struct file_lock
*, struct file_lock
*);
959 unsigned long (*lm_owner_key
)(struct file_lock
*);
960 fl_owner_t (*lm_get_owner
)(fl_owner_t
);
961 void (*lm_put_owner
)(fl_owner_t
);
962 void (*lm_notify
)(struct file_lock
*); /* unblock callback */
963 int (*lm_grant
)(struct file_lock
*, int);
964 bool (*lm_break
)(struct file_lock
*);
965 int (*lm_change
)(struct file_lock
*, int, struct list_head
*);
966 void (*lm_setup
)(struct file_lock
*, void **);
969 struct lock_manager
{
970 struct list_head list
;
972 * NFSv4 and up also want opens blocked during the grace period;
979 void locks_start_grace(struct net
*, struct lock_manager
*);
980 void locks_end_grace(struct lock_manager
*);
981 bool locks_in_grace(struct net
*);
982 bool opens_in_grace(struct net
*);
984 /* that will die - we need it for nfs_lock_info */
985 #include <linux/nfs_fs_i.h>
988 * struct file_lock represents a generic "file lock". It's used to represent
989 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
990 * note that the same struct is used to represent both a request for a lock and
991 * the lock itself, but the same object is never used for both.
993 * FIXME: should we create a separate "struct lock_request" to help distinguish
996 * The varous i_flctx lists are ordered by:
999 * 2) lock range start
1002 * Obviously, the last two criteria only matter for POSIX locks.
1005 struct file_lock
*fl_next
; /* singly linked list for this inode */
1006 struct list_head fl_list
; /* link into file_lock_context */
1007 struct hlist_node fl_link
; /* node in global lists */
1008 struct list_head fl_block
; /* circular list of blocked processes */
1009 fl_owner_t fl_owner
;
1010 unsigned int fl_flags
;
1011 unsigned char fl_type
;
1012 unsigned int fl_pid
;
1013 int fl_link_cpu
; /* what cpu's list is this on? */
1014 wait_queue_head_t fl_wait
;
1015 struct file
*fl_file
;
1019 struct fasync_struct
* fl_fasync
; /* for lease break notifications */
1020 /* for lease breaks: */
1021 unsigned long fl_break_time
;
1022 unsigned long fl_downgrade_time
;
1024 const struct file_lock_operations
*fl_ops
; /* Callbacks for filesystems */
1025 const struct lock_manager_operations
*fl_lmops
; /* Callbacks for lockmanagers */
1027 struct nfs_lock_info nfs_fl
;
1028 struct nfs4_lock_info nfs4_fl
;
1030 struct list_head link
; /* link in AFS vnode's pending_locks list */
1031 int state
; /* state of grant or error if -ve */
1034 } __randomize_layout
;
1036 struct file_lock_context
{
1037 spinlock_t flc_lock
;
1038 struct list_head flc_flock
;
1039 struct list_head flc_posix
;
1040 struct list_head flc_lease
;
1043 /* The following constant reflects the upper bound of the file/locking space */
1045 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1046 #define OFFSET_MAX INT_LIMIT(loff_t)
1047 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1050 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
1053 * Return the inode to use for locking
1055 * For overlayfs this should be the overlay inode, not the real inode returned
1056 * by file_inode(). For any other fs file_inode(filp) and locks_inode(filp) are
1059 static inline struct inode
*locks_inode(const struct file
*f
)
1061 return f
->f_path
.dentry
->d_inode
;
1064 #ifdef CONFIG_FILE_LOCKING
1065 extern int fcntl_getlk(struct file
*, unsigned int, struct flock
*);
1066 extern int fcntl_setlk(unsigned int, struct file
*, unsigned int,
1069 #if BITS_PER_LONG == 32
1070 extern int fcntl_getlk64(struct file
*, unsigned int, struct flock64
*);
1071 extern int fcntl_setlk64(unsigned int, struct file
*, unsigned int,
1075 extern int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
);
1076 extern int fcntl_getlease(struct file
*filp
);
1079 void locks_free_lock_context(struct inode
*inode
);
1080 void locks_free_lock(struct file_lock
*fl
);
1081 extern void locks_init_lock(struct file_lock
*);
1082 extern struct file_lock
* locks_alloc_lock(void);
1083 extern void locks_copy_lock(struct file_lock
*, struct file_lock
*);
1084 extern void locks_copy_conflock(struct file_lock
*, struct file_lock
*);
1085 extern void locks_remove_posix(struct file
*, fl_owner_t
);
1086 extern void locks_remove_file(struct file
*);
1087 extern void locks_release_private(struct file_lock
*);
1088 extern void posix_test_lock(struct file
*, struct file_lock
*);
1089 extern int posix_lock_file(struct file
*, struct file_lock
*, struct file_lock
*);
1090 extern int posix_unblock_lock(struct file_lock
*);
1091 extern int vfs_test_lock(struct file
*, struct file_lock
*);
1092 extern int vfs_lock_file(struct file
*, unsigned int, struct file_lock
*, struct file_lock
*);
1093 extern int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
);
1094 extern int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
);
1095 extern int __break_lease(struct inode
*inode
, unsigned int flags
, unsigned int type
);
1096 extern void lease_get_mtime(struct inode
*, struct timespec64
*time
);
1097 extern int generic_setlease(struct file
*, long, struct file_lock
**, void **priv
);
1098 extern int vfs_setlease(struct file
*, long, struct file_lock
**, void **);
1099 extern int lease_modify(struct file_lock
*, int, struct list_head
*);
1100 struct files_struct
;
1101 extern void show_fd_locks(struct seq_file
*f
,
1102 struct file
*filp
, struct files_struct
*files
);
1103 #else /* !CONFIG_FILE_LOCKING */
1104 static inline int fcntl_getlk(struct file
*file
, unsigned int cmd
,
1105 struct flock __user
*user
)
1110 static inline int fcntl_setlk(unsigned int fd
, struct file
*file
,
1111 unsigned int cmd
, struct flock __user
*user
)
1116 #if BITS_PER_LONG == 32
1117 static inline int fcntl_getlk64(struct file
*file
, unsigned int cmd
,
1118 struct flock64 __user
*user
)
1123 static inline int fcntl_setlk64(unsigned int fd
, struct file
*file
,
1124 unsigned int cmd
, struct flock64 __user
*user
)
1129 static inline int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
)
1134 static inline int fcntl_getlease(struct file
*filp
)
1140 locks_free_lock_context(struct inode
*inode
)
1144 static inline void locks_init_lock(struct file_lock
*fl
)
1149 static inline void locks_copy_conflock(struct file_lock
*new, struct file_lock
*fl
)
1154 static inline void locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
1159 static inline void locks_remove_posix(struct file
*filp
, fl_owner_t owner
)
1164 static inline void locks_remove_file(struct file
*filp
)
1169 static inline void posix_test_lock(struct file
*filp
, struct file_lock
*fl
)
1174 static inline int posix_lock_file(struct file
*filp
, struct file_lock
*fl
,
1175 struct file_lock
*conflock
)
1180 static inline int posix_unblock_lock(struct file_lock
*waiter
)
1185 static inline int vfs_test_lock(struct file
*filp
, struct file_lock
*fl
)
1190 static inline int vfs_lock_file(struct file
*filp
, unsigned int cmd
,
1191 struct file_lock
*fl
, struct file_lock
*conf
)
1196 static inline int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
)
1201 static inline int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
)
1206 static inline int __break_lease(struct inode
*inode
, unsigned int mode
, unsigned int type
)
1211 static inline void lease_get_mtime(struct inode
*inode
,
1212 struct timespec64
*time
)
1217 static inline int generic_setlease(struct file
*filp
, long arg
,
1218 struct file_lock
**flp
, void **priv
)
1223 static inline int vfs_setlease(struct file
*filp
, long arg
,
1224 struct file_lock
**lease
, void **priv
)
1229 static inline int lease_modify(struct file_lock
*fl
, int arg
,
1230 struct list_head
*dispose
)
1235 struct files_struct
;
1236 static inline void show_fd_locks(struct seq_file
*f
,
1237 struct file
*filp
, struct files_struct
*files
) {}
1238 #endif /* !CONFIG_FILE_LOCKING */
1240 static inline struct inode
*file_inode(const struct file
*f
)
1245 static inline struct dentry
*file_dentry(const struct file
*file
)
1247 return d_real(file
->f_path
.dentry
, file_inode(file
), 0, 0);
1250 static inline int locks_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1252 return locks_lock_inode_wait(locks_inode(filp
), fl
);
1255 struct fasync_struct
{
1259 struct fasync_struct
*fa_next
; /* singly linked list */
1260 struct file
*fa_file
;
1261 struct rcu_head fa_rcu
;
1264 #define FASYNC_MAGIC 0x4601
1266 /* SMP safe fasync helpers: */
1267 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1268 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1269 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1270 extern struct fasync_struct
*fasync_alloc(void);
1271 extern void fasync_free(struct fasync_struct
*);
1273 /* can be called from interrupts */
1274 extern void kill_fasync(struct fasync_struct
**, int, int);
1276 extern void __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1277 extern int f_setown(struct file
*filp
, unsigned long arg
, int force
);
1278 extern void f_delown(struct file
*filp
);
1279 extern pid_t
f_getown(struct file
*filp
);
1280 extern int send_sigurg(struct fown_struct
*fown
);
1283 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1284 * represented in both.
1286 #define SB_RDONLY 1 /* Mount read-only */
1287 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1288 #define SB_NODEV 4 /* Disallow access to device special files */
1289 #define SB_NOEXEC 8 /* Disallow program execution */
1290 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1291 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1292 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1293 #define SB_NOATIME 1024 /* Do not update access times. */
1294 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1295 #define SB_SILENT 32768
1296 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1297 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1298 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1299 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1301 /* These sb flags are internal to the kernel */
1302 #define SB_SUBMOUNT (1<<26)
1303 #define SB_NOREMOTELOCK (1<<27)
1304 #define SB_NOSEC (1<<28)
1305 #define SB_BORN (1<<29)
1306 #define SB_ACTIVE (1<<30)
1307 #define SB_NOUSER (1<<31)
1313 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1314 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1315 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1316 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1317 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1320 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1321 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1322 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1323 #define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */
1325 /* sb->s_iflags to limit user namespace mounts */
1326 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1327 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1328 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1330 /* Possible states of 'frozen' field */
1332 SB_UNFROZEN
= 0, /* FS is unfrozen */
1333 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1334 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1335 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1336 * internal threads if needed) */
1337 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1340 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1343 int frozen
; /* Is sb frozen? */
1344 wait_queue_head_t wait_unfrozen
; /* for get_super_thawed() */
1345 struct percpu_rw_semaphore rw_sem
[SB_FREEZE_LEVELS
];
1348 struct super_block
{
1349 struct list_head s_list
; /* Keep this first */
1350 dev_t s_dev
; /* search index; _not_ kdev_t */
1351 unsigned char s_blocksize_bits
;
1352 unsigned long s_blocksize
;
1353 loff_t s_maxbytes
; /* Max file size */
1354 struct file_system_type
*s_type
;
1355 const struct super_operations
*s_op
;
1356 const struct dquot_operations
*dq_op
;
1357 const struct quotactl_ops
*s_qcop
;
1358 const struct export_operations
*s_export_op
;
1359 unsigned long s_flags
;
1360 unsigned long s_iflags
; /* internal SB_I_* flags */
1361 unsigned long s_magic
;
1362 struct dentry
*s_root
;
1363 struct rw_semaphore s_umount
;
1366 #ifdef CONFIG_SECURITY
1369 const struct xattr_handler
**s_xattr
;
1370 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
1371 const struct fscrypt_operations
*s_cop
;
1373 struct hlist_bl_head s_roots
; /* alternate root dentries for NFS */
1374 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1375 struct block_device
*s_bdev
;
1376 struct backing_dev_info
*s_bdi
;
1377 struct mtd_info
*s_mtd
;
1378 struct hlist_node s_instances
;
1379 unsigned int s_quota_types
; /* Bitmask of supported quota types */
1380 struct quota_info s_dquot
; /* Diskquota specific options */
1382 struct sb_writers s_writers
;
1384 char s_id
[32]; /* Informational name */
1385 uuid_t s_uuid
; /* UUID */
1387 void *s_fs_info
; /* Filesystem private info */
1388 unsigned int s_max_links
;
1391 /* Granularity of c/m/atime in ns.
1392 Cannot be worse than a second */
1396 * The next field is for VFS *only*. No filesystems have any business
1397 * even looking at it. You had been warned.
1399 struct mutex s_vfs_rename_mutex
; /* Kludge */
1402 * Filesystem subtype. If non-empty the filesystem type field
1403 * in /proc/mounts will be "type.subtype"
1407 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1410 * Saved pool identifier for cleancache (-1 means none)
1412 int cleancache_poolid
;
1414 struct shrinker s_shrink
; /* per-sb shrinker handle */
1416 /* Number of inodes with nlink == 0 but still referenced */
1417 atomic_long_t s_remove_count
;
1419 /* Being remounted read-only */
1420 int s_readonly_remount
;
1422 /* AIO completions deferred from interrupt context */
1423 struct workqueue_struct
*s_dio_done_wq
;
1424 struct hlist_head s_pins
;
1427 * Owning user namespace and default context in which to
1428 * interpret filesystem uids, gids, quotas, device nodes,
1429 * xattrs and security labels.
1431 struct user_namespace
*s_user_ns
;
1434 * Keep the lru lists last in the structure so they always sit on their
1435 * own individual cachelines.
1437 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp
;
1438 struct list_lru s_inode_lru ____cacheline_aligned_in_smp
;
1439 struct rcu_head rcu
;
1440 struct work_struct destroy_work
;
1442 struct mutex s_sync_lock
; /* sync serialisation lock */
1445 * Indicates how deep in a filesystem stack this SB is
1449 /* s_inode_list_lock protects s_inodes */
1450 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp
;
1451 struct list_head s_inodes
; /* all inodes */
1453 spinlock_t s_inode_wblist_lock
;
1454 struct list_head s_inodes_wb
; /* writeback inodes */
1455 } __randomize_layout
;
1457 /* Helper functions so that in most cases filesystems will
1458 * not need to deal directly with kuid_t and kgid_t and can
1459 * instead deal with the raw numeric values that are stored
1460 * in the filesystem.
1462 static inline uid_t
i_uid_read(const struct inode
*inode
)
1464 return from_kuid(inode
->i_sb
->s_user_ns
, inode
->i_uid
);
1467 static inline gid_t
i_gid_read(const struct inode
*inode
)
1469 return from_kgid(inode
->i_sb
->s_user_ns
, inode
->i_gid
);
1472 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
1474 inode
->i_uid
= make_kuid(inode
->i_sb
->s_user_ns
, uid
);
1477 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
1479 inode
->i_gid
= make_kgid(inode
->i_sb
->s_user_ns
, gid
);
1482 extern struct timespec64
timespec64_trunc(struct timespec64 t
, unsigned gran
);
1483 extern struct timespec64
current_time(struct inode
*inode
);
1486 * Snapshotting support.
1489 void __sb_end_write(struct super_block
*sb
, int level
);
1490 int __sb_start_write(struct super_block
*sb
, int level
, bool wait
);
1492 #define __sb_writers_acquired(sb, lev) \
1493 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1494 #define __sb_writers_release(sb, lev) \
1495 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1498 * sb_end_write - drop write access to a superblock
1499 * @sb: the super we wrote to
1501 * Decrement number of writers to the filesystem. Wake up possible waiters
1502 * wanting to freeze the filesystem.
1504 static inline void sb_end_write(struct super_block
*sb
)
1506 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1510 * sb_end_pagefault - drop write access to a superblock from a page fault
1511 * @sb: the super we wrote to
1513 * Decrement number of processes handling write page fault to the filesystem.
1514 * Wake up possible waiters wanting to freeze the filesystem.
1516 static inline void sb_end_pagefault(struct super_block
*sb
)
1518 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1522 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1523 * @sb: the super we wrote to
1525 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1526 * waiters wanting to freeze the filesystem.
1528 static inline void sb_end_intwrite(struct super_block
*sb
)
1530 __sb_end_write(sb
, SB_FREEZE_FS
);
1534 * sb_start_write - get write access to a superblock
1535 * @sb: the super we write to
1537 * When a process wants to write data or metadata to a file system (i.e. dirty
1538 * a page or an inode), it should embed the operation in a sb_start_write() -
1539 * sb_end_write() pair to get exclusion against file system freezing. This
1540 * function increments number of writers preventing freezing. If the file
1541 * system is already frozen, the function waits until the file system is
1544 * Since freeze protection behaves as a lock, users have to preserve
1545 * ordering of freeze protection and other filesystem locks. Generally,
1546 * freeze protection should be the outermost lock. In particular, we have:
1549 * -> i_mutex (write path, truncate, directory ops, ...)
1550 * -> s_umount (freeze_super, thaw_super)
1552 static inline void sb_start_write(struct super_block
*sb
)
1554 __sb_start_write(sb
, SB_FREEZE_WRITE
, true);
1557 static inline int sb_start_write_trylock(struct super_block
*sb
)
1559 return __sb_start_write(sb
, SB_FREEZE_WRITE
, false);
1563 * sb_start_pagefault - get write access to a superblock from a page fault
1564 * @sb: the super we write to
1566 * When a process starts handling write page fault, it should embed the
1567 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1568 * exclusion against file system freezing. This is needed since the page fault
1569 * is going to dirty a page. This function increments number of running page
1570 * faults preventing freezing. If the file system is already frozen, the
1571 * function waits until the file system is thawed.
1573 * Since page fault freeze protection behaves as a lock, users have to preserve
1574 * ordering of freeze protection and other filesystem locks. It is advised to
1575 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1576 * handling code implies lock dependency:
1579 * -> sb_start_pagefault
1581 static inline void sb_start_pagefault(struct super_block
*sb
)
1583 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
, true);
1587 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1588 * @sb: the super we write to
1590 * This is the third level of protection against filesystem freezing. It is
1591 * free for use by a filesystem. The only requirement is that it must rank
1592 * below sb_start_pagefault.
1594 * For example filesystem can call sb_start_intwrite() when starting a
1595 * transaction which somewhat eases handling of freezing for internal sources
1596 * of filesystem changes (internal fs threads, discarding preallocation on file
1599 static inline void sb_start_intwrite(struct super_block
*sb
)
1601 __sb_start_write(sb
, SB_FREEZE_FS
, true);
1604 static inline int sb_start_intwrite_trylock(struct super_block
*sb
)
1606 return __sb_start_write(sb
, SB_FREEZE_FS
, false);
1610 extern bool inode_owner_or_capable(const struct inode
*inode
);
1613 * VFS helper functions..
1615 extern int vfs_create(struct inode
*, struct dentry
*, umode_t
, bool);
1616 extern int vfs_mkdir(struct inode
*, struct dentry
*, umode_t
);
1617 extern int vfs_mknod(struct inode
*, struct dentry
*, umode_t
, dev_t
);
1618 extern int vfs_symlink(struct inode
*, struct dentry
*, const char *);
1619 extern int vfs_link(struct dentry
*, struct inode
*, struct dentry
*, struct inode
**);
1620 extern int vfs_rmdir(struct inode
*, struct dentry
*);
1621 extern int vfs_unlink(struct inode
*, struct dentry
*, struct inode
**);
1622 extern int vfs_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*, struct inode
**, unsigned int);
1623 extern int vfs_whiteout(struct inode
*, struct dentry
*);
1625 extern struct dentry
*vfs_tmpfile(struct dentry
*dentry
, umode_t mode
,
1628 int vfs_mkobj(struct dentry
*, umode_t
,
1629 int (*f
)(struct dentry
*, umode_t
, void *),
1633 * VFS file helper functions.
1635 extern void inode_init_owner(struct inode
*inode
, const struct inode
*dir
,
1637 extern bool may_open_dev(const struct path
*path
);
1639 * VFS FS_IOC_FIEMAP helper definitions.
1641 struct fiemap_extent_info
{
1642 unsigned int fi_flags
; /* Flags as passed from user */
1643 unsigned int fi_extents_mapped
; /* Number of mapped extents */
1644 unsigned int fi_extents_max
; /* Size of fiemap_extent array */
1645 struct fiemap_extent __user
*fi_extents_start
; /* Start of
1646 fiemap_extent array */
1648 int fiemap_fill_next_extent(struct fiemap_extent_info
*info
, u64 logical
,
1649 u64 phys
, u64 len
, u32 flags
);
1650 int fiemap_check_flags(struct fiemap_extent_info
*fieinfo
, u32 fs_flags
);
1655 * NOTE! These match bits 12..15 of stat.st_mode
1656 * (ie "(i_mode >> 12) & 15").
1658 #define DT_UNKNOWN 0
1669 * This is the "filldir" function type, used by readdir() to let
1670 * the kernel specify what kind of dirent layout it wants to have.
1671 * This allows the kernel to read directories into kernel space or
1672 * to have different dirent layouts depending on the binary type.
1675 typedef int (*filldir_t
)(struct dir_context
*, const char *, int, loff_t
, u64
,
1678 struct dir_context
{
1683 struct block_device_operations
;
1685 /* These macros are for out of kernel modules to test that
1686 * the kernel supports the unlocked_ioctl and compat_ioctl
1687 * fields in struct file_operations. */
1688 #define HAVE_COMPAT_IOCTL 1
1689 #define HAVE_UNLOCKED_IOCTL 1
1692 * These flags let !MMU mmap() govern direct device mapping vs immediate
1693 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1695 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1696 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1697 * NOMMU_MAP_READ: Can be mapped for reading
1698 * NOMMU_MAP_WRITE: Can be mapped for writing
1699 * NOMMU_MAP_EXEC: Can be mapped for execution
1701 #define NOMMU_MAP_COPY 0x00000001
1702 #define NOMMU_MAP_DIRECT 0x00000008
1703 #define NOMMU_MAP_READ VM_MAYREAD
1704 #define NOMMU_MAP_WRITE VM_MAYWRITE
1705 #define NOMMU_MAP_EXEC VM_MAYEXEC
1707 #define NOMMU_VMFLAGS \
1708 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1713 struct file_operations
{
1714 struct module
*owner
;
1715 loff_t (*llseek
) (struct file
*, loff_t
, int);
1716 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
1717 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
1718 ssize_t (*read_iter
) (struct kiocb
*, struct iov_iter
*);
1719 ssize_t (*write_iter
) (struct kiocb
*, struct iov_iter
*);
1720 int (*iterate
) (struct file
*, struct dir_context
*);
1721 int (*iterate_shared
) (struct file
*, struct dir_context
*);
1722 __poll_t (*poll
) (struct file
*, struct poll_table_struct
*);
1723 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
1724 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
1725 int (*mmap
) (struct file
*, struct vm_area_struct
*);
1726 unsigned long mmap_supported_flags
;
1727 int (*open
) (struct inode
*, struct file
*);
1728 int (*flush
) (struct file
*, fl_owner_t id
);
1729 int (*release
) (struct inode
*, struct file
*);
1730 int (*fsync
) (struct file
*, loff_t
, loff_t
, int datasync
);
1731 int (*fasync
) (int, struct file
*, int);
1732 int (*lock
) (struct file
*, int, struct file_lock
*);
1733 ssize_t (*sendpage
) (struct file
*, struct page
*, int, size_t, loff_t
*, int);
1734 unsigned long (*get_unmapped_area
)(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1735 int (*check_flags
)(int);
1736 int (*flock
) (struct file
*, int, struct file_lock
*);
1737 ssize_t (*splice_write
)(struct pipe_inode_info
*, struct file
*, loff_t
*, size_t, unsigned int);
1738 ssize_t (*splice_read
)(struct file
*, loff_t
*, struct pipe_inode_info
*, size_t, unsigned int);
1739 int (*setlease
)(struct file
*, long, struct file_lock
**, void **);
1740 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
1742 void (*show_fdinfo
)(struct seq_file
*m
, struct file
*f
);
1744 unsigned (*mmap_capabilities
)(struct file
*);
1746 ssize_t (*copy_file_range
)(struct file
*, loff_t
, struct file
*,
1747 loff_t
, size_t, unsigned int);
1748 int (*clone_file_range
)(struct file
*, loff_t
, struct file
*, loff_t
,
1750 ssize_t (*dedupe_file_range
)(struct file
*, u64
, u64
, struct file
*,
1752 } __randomize_layout
;
1754 struct inode_operations
{
1755 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*, unsigned int);
1756 const char * (*get_link
) (struct dentry
*, struct inode
*, struct delayed_call
*);
1757 int (*permission
) (struct inode
*, int);
1758 struct posix_acl
* (*get_acl
)(struct inode
*, int);
1760 int (*readlink
) (struct dentry
*, char __user
*,int);
1762 int (*create
) (struct inode
*,struct dentry
*, umode_t
, bool);
1763 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
1764 int (*unlink
) (struct inode
*,struct dentry
*);
1765 int (*symlink
) (struct inode
*,struct dentry
*,const char *);
1766 int (*mkdir
) (struct inode
*,struct dentry
*,umode_t
);
1767 int (*rmdir
) (struct inode
*,struct dentry
*);
1768 int (*mknod
) (struct inode
*,struct dentry
*,umode_t
,dev_t
);
1769 int (*rename
) (struct inode
*, struct dentry
*,
1770 struct inode
*, struct dentry
*, unsigned int);
1771 int (*setattr
) (struct dentry
*, struct iattr
*);
1772 int (*getattr
) (const struct path
*, struct kstat
*, u32
, unsigned int);
1773 ssize_t (*listxattr
) (struct dentry
*, char *, size_t);
1774 int (*fiemap
)(struct inode
*, struct fiemap_extent_info
*, u64 start
,
1776 int (*update_time
)(struct inode
*, struct timespec64
*, int);
1777 int (*atomic_open
)(struct inode
*, struct dentry
*,
1778 struct file
*, unsigned open_flag
,
1779 umode_t create_mode
, int *opened
);
1780 int (*tmpfile
) (struct inode
*, struct dentry
*, umode_t
);
1781 int (*set_acl
)(struct inode
*, struct posix_acl
*, int);
1782 } ____cacheline_aligned
;
1784 static inline ssize_t
call_read_iter(struct file
*file
, struct kiocb
*kio
,
1785 struct iov_iter
*iter
)
1787 return file
->f_op
->read_iter(kio
, iter
);
1790 static inline ssize_t
call_write_iter(struct file
*file
, struct kiocb
*kio
,
1791 struct iov_iter
*iter
)
1793 return file
->f_op
->write_iter(kio
, iter
);
1796 static inline int call_mmap(struct file
*file
, struct vm_area_struct
*vma
)
1798 return file
->f_op
->mmap(file
, vma
);
1801 ssize_t
rw_copy_check_uvector(int type
, const struct iovec __user
* uvector
,
1802 unsigned long nr_segs
, unsigned long fast_segs
,
1803 struct iovec
*fast_pointer
,
1804 struct iovec
**ret_pointer
);
1806 extern ssize_t
__vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1807 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1808 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1809 extern ssize_t
vfs_readv(struct file
*, const struct iovec __user
*,
1810 unsigned long, loff_t
*, rwf_t
);
1811 extern ssize_t
vfs_copy_file_range(struct file
*, loff_t
, struct file
*,
1812 loff_t
, size_t, unsigned int);
1813 extern int vfs_clone_file_prep_inodes(struct inode
*inode_in
, loff_t pos_in
,
1814 struct inode
*inode_out
, loff_t pos_out
,
1815 u64
*len
, bool is_dedupe
);
1816 extern int vfs_clone_file_range(struct file
*file_in
, loff_t pos_in
,
1817 struct file
*file_out
, loff_t pos_out
, u64 len
);
1818 extern int vfs_dedupe_file_range_compare(struct inode
*src
, loff_t srcoff
,
1819 struct inode
*dest
, loff_t destoff
,
1820 loff_t len
, bool *is_same
);
1821 extern int vfs_dedupe_file_range(struct file
*file
,
1822 struct file_dedupe_range
*same
);
1824 struct super_operations
{
1825 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
1826 void (*destroy_inode
)(struct inode
*);
1828 void (*dirty_inode
) (struct inode
*, int flags
);
1829 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
1830 int (*drop_inode
) (struct inode
*);
1831 void (*evict_inode
) (struct inode
*);
1832 void (*put_super
) (struct super_block
*);
1833 int (*sync_fs
)(struct super_block
*sb
, int wait
);
1834 int (*freeze_super
) (struct super_block
*);
1835 int (*freeze_fs
) (struct super_block
*);
1836 int (*thaw_super
) (struct super_block
*);
1837 int (*unfreeze_fs
) (struct super_block
*);
1838 int (*statfs
) (struct dentry
*, struct kstatfs
*);
1839 int (*remount_fs
) (struct super_block
*, int *, char *);
1840 void (*umount_begin
) (struct super_block
*);
1842 int (*show_options
)(struct seq_file
*, struct dentry
*);
1843 int (*show_devname
)(struct seq_file
*, struct dentry
*);
1844 int (*show_path
)(struct seq_file
*, struct dentry
*);
1845 int (*show_stats
)(struct seq_file
*, struct dentry
*);
1847 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
1848 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
1849 struct dquot
**(*get_dquots
)(struct inode
*);
1851 int (*bdev_try_to_free_page
)(struct super_block
*, struct page
*, gfp_t
);
1852 long (*nr_cached_objects
)(struct super_block
*,
1853 struct shrink_control
*);
1854 long (*free_cached_objects
)(struct super_block
*,
1855 struct shrink_control
*);
1859 * Inode flags - they have no relation to superblock flags now
1861 #define S_SYNC 1 /* Writes are synced at once */
1862 #define S_NOATIME 2 /* Do not update access times */
1863 #define S_APPEND 4 /* Append-only file */
1864 #define S_IMMUTABLE 8 /* Immutable file */
1865 #define S_DEAD 16 /* removed, but still open directory */
1866 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1867 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1868 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1869 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1870 #define S_PRIVATE 512 /* Inode is fs-internal */
1871 #define S_IMA 1024 /* Inode has an associated IMA struct */
1872 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1873 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1874 #ifdef CONFIG_FS_DAX
1875 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1877 #define S_DAX 0 /* Make all the DAX code disappear */
1879 #define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */
1882 * Note that nosuid etc flags are inode-specific: setting some file-system
1883 * flags just means all the inodes inherit those flags by default. It might be
1884 * possible to override it selectively if you really wanted to with some
1885 * ioctl() that is not currently implemented.
1887 * Exception: SB_RDONLY is always applied to the entire file system.
1889 * Unfortunately, it is possible to change a filesystems flags with it mounted
1890 * with files in use. This means that all of the inodes will not have their
1891 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1892 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1894 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1896 static inline bool sb_rdonly(const struct super_block
*sb
) { return sb
->s_flags
& SB_RDONLY
; }
1897 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
1898 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
1899 ((inode)->i_flags & S_SYNC))
1900 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
1901 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1902 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
1903 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
1904 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
1906 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1907 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1908 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1909 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
1911 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1912 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1913 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1914 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1915 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1916 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1917 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1918 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1919 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
1921 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1922 (inode)->i_rdev == WHITEOUT_DEV)
1924 static inline bool HAS_UNMAPPED_ID(struct inode
*inode
)
1926 return !uid_valid(inode
->i_uid
) || !gid_valid(inode
->i_gid
);
1929 static inline enum rw_hint
file_write_hint(struct file
*file
)
1931 if (file
->f_write_hint
!= WRITE_LIFE_NOT_SET
)
1932 return file
->f_write_hint
;
1934 return file_inode(file
)->i_write_hint
;
1937 static inline int iocb_flags(struct file
*file
);
1939 static inline u16
ki_hint_validate(enum rw_hint hint
)
1941 typeof(((struct kiocb
*)0)->ki_hint
) max_hint
= -1;
1943 if (hint
<= max_hint
)
1948 static inline void init_sync_kiocb(struct kiocb
*kiocb
, struct file
*filp
)
1950 *kiocb
= (struct kiocb
) {
1952 .ki_flags
= iocb_flags(filp
),
1953 .ki_hint
= ki_hint_validate(file_write_hint(filp
)),
1954 .ki_ioprio
= IOPRIO_PRIO_VALUE(IOPRIO_CLASS_NONE
, 0),
1959 * Inode state bits. Protected by inode->i_lock
1961 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1962 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1964 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1965 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1966 * various stages of removing an inode.
1968 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1970 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1971 * fdatasync(). i_atime is the usual cause.
1972 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1973 * these changes separately from I_DIRTY_SYNC so that we
1974 * don't have to write inode on fdatasync() when only
1975 * mtime has changed in it.
1976 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1977 * I_NEW Serves as both a mutex and completion notification.
1978 * New inodes set I_NEW. If two processes both create
1979 * the same inode, one of them will release its inode and
1980 * wait for I_NEW to be released before returning.
1981 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1982 * also cause waiting on I_NEW, without I_NEW actually
1983 * being set. find_inode() uses this to prevent returning
1984 * nearly-dead inodes.
1985 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1986 * is zero. I_FREEING must be set when I_WILL_FREE is
1988 * I_FREEING Set when inode is about to be freed but still has dirty
1989 * pages or buffers attached or the inode itself is still
1991 * I_CLEAR Added by clear_inode(). In this state the inode is
1992 * clean and can be destroyed. Inode keeps I_FREEING.
1994 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1995 * prohibited for many purposes. iget() must wait for
1996 * the inode to be completely released, then create it
1997 * anew. Other functions will just ignore such inodes,
1998 * if appropriate. I_NEW is used for waiting.
2000 * I_SYNC Writeback of inode is running. The bit is set during
2001 * data writeback, and cleared with a wakeup on the bit
2002 * address once it is done. The bit is also used to pin
2003 * the inode in memory for flusher thread.
2005 * I_REFERENCED Marks the inode as recently references on the LRU list.
2007 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2009 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2010 * synchronize competing switching instances and to tell
2011 * wb stat updates to grab the i_pages lock. See
2012 * inode_switch_wb_work_fn() for details.
2014 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2015 * and work dirs among overlayfs mounts.
2017 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2019 #define I_DIRTY_SYNC (1 << 0)
2020 #define I_DIRTY_DATASYNC (1 << 1)
2021 #define I_DIRTY_PAGES (1 << 2)
2023 #define I_NEW (1 << __I_NEW)
2024 #define I_WILL_FREE (1 << 4)
2025 #define I_FREEING (1 << 5)
2026 #define I_CLEAR (1 << 6)
2028 #define I_SYNC (1 << __I_SYNC)
2029 #define I_REFERENCED (1 << 8)
2030 #define __I_DIO_WAKEUP 9
2031 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2032 #define I_LINKABLE (1 << 10)
2033 #define I_DIRTY_TIME (1 << 11)
2034 #define __I_DIRTY_TIME_EXPIRED 12
2035 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
2036 #define I_WB_SWITCH (1 << 13)
2037 #define I_OVL_INUSE (1 << 14)
2039 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2040 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2041 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2043 extern void __mark_inode_dirty(struct inode
*, int);
2044 static inline void mark_inode_dirty(struct inode
*inode
)
2046 __mark_inode_dirty(inode
, I_DIRTY
);
2049 static inline void mark_inode_dirty_sync(struct inode
*inode
)
2051 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
2054 extern void inc_nlink(struct inode
*inode
);
2055 extern void drop_nlink(struct inode
*inode
);
2056 extern void clear_nlink(struct inode
*inode
);
2057 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
2059 static inline void inode_inc_link_count(struct inode
*inode
)
2062 mark_inode_dirty(inode
);
2065 static inline void inode_dec_link_count(struct inode
*inode
)
2068 mark_inode_dirty(inode
);
2071 enum file_time_flags
{
2078 extern void touch_atime(const struct path
*);
2079 static inline void file_accessed(struct file
*file
)
2081 if (!(file
->f_flags
& O_NOATIME
))
2082 touch_atime(&file
->f_path
);
2085 int sync_inode(struct inode
*inode
, struct writeback_control
*wbc
);
2086 int sync_inode_metadata(struct inode
*inode
, int wait
);
2088 struct file_system_type
{
2091 #define FS_REQUIRES_DEV 1
2092 #define FS_BINARY_MOUNTDATA 2
2093 #define FS_HAS_SUBTYPE 4
2094 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2095 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2096 struct dentry
*(*mount
) (struct file_system_type
*, int,
2097 const char *, void *);
2098 void (*kill_sb
) (struct super_block
*);
2099 struct module
*owner
;
2100 struct file_system_type
* next
;
2101 struct hlist_head fs_supers
;
2103 struct lock_class_key s_lock_key
;
2104 struct lock_class_key s_umount_key
;
2105 struct lock_class_key s_vfs_rename_key
;
2106 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
2108 struct lock_class_key i_lock_key
;
2109 struct lock_class_key i_mutex_key
;
2110 struct lock_class_key i_mutex_dir_key
;
2113 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2115 extern struct dentry
*mount_ns(struct file_system_type
*fs_type
,
2116 int flags
, void *data
, void *ns
, struct user_namespace
*user_ns
,
2117 int (*fill_super
)(struct super_block
*, void *, int));
2119 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
2120 int flags
, const char *dev_name
, void *data
,
2121 int (*fill_super
)(struct super_block
*, void *, int));
2123 static inline struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
2124 int flags
, const char *dev_name
, void *data
,
2125 int (*fill_super
)(struct super_block
*, void *, int))
2127 return ERR_PTR(-ENODEV
);
2130 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
2131 int flags
, void *data
,
2132 int (*fill_super
)(struct super_block
*, void *, int));
2133 extern struct dentry
*mount_nodev(struct file_system_type
*fs_type
,
2134 int flags
, void *data
,
2135 int (*fill_super
)(struct super_block
*, void *, int));
2136 extern struct dentry
*mount_subtree(struct vfsmount
*mnt
, const char *path
);
2137 void generic_shutdown_super(struct super_block
*sb
);
2139 void kill_block_super(struct super_block
*sb
);
2141 static inline void kill_block_super(struct super_block
*sb
)
2146 void kill_anon_super(struct super_block
*sb
);
2147 void kill_litter_super(struct super_block
*sb
);
2148 void deactivate_super(struct super_block
*sb
);
2149 void deactivate_locked_super(struct super_block
*sb
);
2150 int set_anon_super(struct super_block
*s
, void *data
);
2151 int get_anon_bdev(dev_t
*);
2152 void free_anon_bdev(dev_t
);
2153 struct super_block
*sget_userns(struct file_system_type
*type
,
2154 int (*test
)(struct super_block
*,void *),
2155 int (*set
)(struct super_block
*,void *),
2156 int flags
, struct user_namespace
*user_ns
,
2158 struct super_block
*sget(struct file_system_type
*type
,
2159 int (*test
)(struct super_block
*,void *),
2160 int (*set
)(struct super_block
*,void *),
2161 int flags
, void *data
);
2162 extern struct dentry
*mount_pseudo_xattr(struct file_system_type
*, char *,
2163 const struct super_operations
*ops
,
2164 const struct xattr_handler
**xattr
,
2165 const struct dentry_operations
*dops
,
2168 static inline struct dentry
*
2169 mount_pseudo(struct file_system_type
*fs_type
, char *name
,
2170 const struct super_operations
*ops
,
2171 const struct dentry_operations
*dops
, unsigned long magic
)
2173 return mount_pseudo_xattr(fs_type
, name
, ops
, NULL
, dops
, magic
);
2176 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2177 #define fops_get(fops) \
2178 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2179 #define fops_put(fops) \
2180 do { if (fops) module_put((fops)->owner); } while(0)
2182 * This one is to be used *ONLY* from ->open() instances.
2183 * fops must be non-NULL, pinned down *and* module dependencies
2184 * should be sufficient to pin the caller down as well.
2186 #define replace_fops(f, fops) \
2188 struct file *__file = (f); \
2189 fops_put(__file->f_op); \
2190 BUG_ON(!(__file->f_op = (fops))); \
2193 extern int register_filesystem(struct file_system_type
*);
2194 extern int unregister_filesystem(struct file_system_type
*);
2195 extern struct vfsmount
*kern_mount_data(struct file_system_type
*, void *data
);
2196 #define kern_mount(type) kern_mount_data(type, NULL)
2197 extern void kern_unmount(struct vfsmount
*mnt
);
2198 extern int may_umount_tree(struct vfsmount
*);
2199 extern int may_umount(struct vfsmount
*);
2200 extern long do_mount(const char *, const char __user
*,
2201 const char *, unsigned long, void *);
2202 extern struct vfsmount
*collect_mounts(const struct path
*);
2203 extern void drop_collected_mounts(struct vfsmount
*);
2204 extern int iterate_mounts(int (*)(struct vfsmount
*, void *), void *,
2206 extern int vfs_statfs(const struct path
*, struct kstatfs
*);
2207 extern int user_statfs(const char __user
*, struct kstatfs
*);
2208 extern int fd_statfs(int, struct kstatfs
*);
2209 extern int freeze_super(struct super_block
*super
);
2210 extern int thaw_super(struct super_block
*super
);
2211 extern bool our_mnt(struct vfsmount
*mnt
);
2212 extern __printf(2, 3)
2213 int super_setup_bdi_name(struct super_block
*sb
, char *fmt
, ...);
2214 extern int super_setup_bdi(struct super_block
*sb
);
2216 extern int current_umask(void);
2218 extern void ihold(struct inode
* inode
);
2219 extern void iput(struct inode
*);
2220 extern int generic_update_time(struct inode
*, struct timespec64
*, int);
2223 extern struct kobject
*fs_kobj
;
2225 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2227 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2228 extern int locks_mandatory_locked(struct file
*);
2229 extern int locks_mandatory_area(struct inode
*, struct file
*, loff_t
, loff_t
, unsigned char);
2232 * Candidates for mandatory locking have the setgid bit set
2233 * but no group execute bit - an otherwise meaningless combination.
2236 static inline int __mandatory_lock(struct inode
*ino
)
2238 return (ino
->i_mode
& (S_ISGID
| S_IXGRP
)) == S_ISGID
;
2242 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2243 * otherwise these will be advisory locks
2246 static inline int mandatory_lock(struct inode
*ino
)
2248 return IS_MANDLOCK(ino
) && __mandatory_lock(ino
);
2251 static inline int locks_verify_locked(struct file
*file
)
2253 if (mandatory_lock(locks_inode(file
)))
2254 return locks_mandatory_locked(file
);
2258 static inline int locks_verify_truncate(struct inode
*inode
,
2262 if (!inode
->i_flctx
|| !mandatory_lock(inode
))
2265 if (size
< inode
->i_size
) {
2266 return locks_mandatory_area(inode
, f
, size
, inode
->i_size
- 1,
2269 return locks_mandatory_area(inode
, f
, inode
->i_size
, size
- 1,
2274 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2276 static inline int locks_mandatory_locked(struct file
*file
)
2281 static inline int locks_mandatory_area(struct inode
*inode
, struct file
*filp
,
2282 loff_t start
, loff_t end
, unsigned char type
)
2287 static inline int __mandatory_lock(struct inode
*inode
)
2292 static inline int mandatory_lock(struct inode
*inode
)
2297 static inline int locks_verify_locked(struct file
*file
)
2302 static inline int locks_verify_truncate(struct inode
*inode
, struct file
*filp
,
2308 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2311 #ifdef CONFIG_FILE_LOCKING
2312 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2315 * Since this check is lockless, we must ensure that any refcounts
2316 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2317 * could end up racing with tasks trying to set a new lease on this
2321 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2322 return __break_lease(inode
, mode
, FL_LEASE
);
2326 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2329 * Since this check is lockless, we must ensure that any refcounts
2330 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2331 * could end up racing with tasks trying to set a new lease on this
2335 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2336 return __break_lease(inode
, mode
, FL_DELEG
);
2340 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2344 ret
= break_deleg(inode
, O_WRONLY
|O_NONBLOCK
);
2345 if (ret
== -EWOULDBLOCK
&& delegated_inode
) {
2346 *delegated_inode
= inode
;
2352 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2356 ret
= break_deleg(*delegated_inode
, O_WRONLY
);
2357 iput(*delegated_inode
);
2358 *delegated_inode
= NULL
;
2362 static inline int break_layout(struct inode
*inode
, bool wait
)
2365 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2366 return __break_lease(inode
,
2367 wait
? O_WRONLY
: O_WRONLY
| O_NONBLOCK
,
2372 #else /* !CONFIG_FILE_LOCKING */
2373 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2378 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2383 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2388 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2394 static inline int break_layout(struct inode
*inode
, bool wait
)
2399 #endif /* CONFIG_FILE_LOCKING */
2404 const char *name
; /* pointer to actual string */
2405 const __user
char *uptr
; /* original userland pointer */
2407 struct audit_names
*aname
;
2411 extern long vfs_truncate(const struct path
*, loff_t
);
2412 extern int do_truncate(struct dentry
*, loff_t start
, unsigned int time_attrs
,
2414 extern int vfs_fallocate(struct file
*file
, int mode
, loff_t offset
,
2416 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
2418 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
2419 extern struct file
*filp_open(const char *, int, umode_t
);
2420 extern struct file
*file_open_root(struct dentry
*, struct vfsmount
*,
2421 const char *, int, umode_t
);
2422 extern struct file
* dentry_open(const struct path
*, int, const struct cred
*);
2423 extern int filp_close(struct file
*, fl_owner_t id
);
2425 extern struct filename
*getname_flags(const char __user
*, int, int *);
2426 extern struct filename
*getname(const char __user
*);
2427 extern struct filename
*getname_kernel(const char *);
2428 extern void putname(struct filename
*name
);
2434 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
2435 int (*open
)(struct inode
*, struct file
*),
2437 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
2441 extern int ioctl_preallocate(struct file
*filp
, void __user
*argp
);
2444 extern void __init
vfs_caches_init_early(void);
2445 extern void __init
vfs_caches_init(void);
2447 extern struct kmem_cache
*names_cachep
;
2449 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2450 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2453 extern int register_blkdev(unsigned int, const char *);
2454 extern void unregister_blkdev(unsigned int, const char *);
2455 extern void bdev_unhash_inode(dev_t dev
);
2456 extern struct block_device
*bdget(dev_t
);
2457 extern struct block_device
*bdgrab(struct block_device
*bdev
);
2458 extern void bd_set_size(struct block_device
*, loff_t size
);
2459 extern void bd_forget(struct inode
*inode
);
2460 extern void bdput(struct block_device
*);
2461 extern void invalidate_bdev(struct block_device
*);
2462 extern void iterate_bdevs(void (*)(struct block_device
*, void *), void *);
2463 extern int sync_blockdev(struct block_device
*bdev
);
2464 extern void kill_bdev(struct block_device
*);
2465 extern struct super_block
*freeze_bdev(struct block_device
*);
2466 extern void emergency_thaw_all(void);
2467 extern void emergency_thaw_bdev(struct super_block
*sb
);
2468 extern int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
);
2469 extern int fsync_bdev(struct block_device
*);
2471 extern struct super_block
*blockdev_superblock
;
2473 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2475 return sb
== blockdev_superblock
;
2478 static inline void bd_forget(struct inode
*inode
) {}
2479 static inline int sync_blockdev(struct block_device
*bdev
) { return 0; }
2480 static inline void kill_bdev(struct block_device
*bdev
) {}
2481 static inline void invalidate_bdev(struct block_device
*bdev
) {}
2483 static inline struct super_block
*freeze_bdev(struct block_device
*sb
)
2488 static inline int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
)
2493 static inline int emergency_thaw_bdev(struct super_block
*sb
)
2498 static inline void iterate_bdevs(void (*f
)(struct block_device
*, void *), void *arg
)
2502 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2507 extern int sync_filesystem(struct super_block
*);
2508 extern const struct file_operations def_blk_fops
;
2509 extern const struct file_operations def_chr_fops
;
2511 extern int ioctl_by_bdev(struct block_device
*, unsigned, unsigned long);
2512 extern int blkdev_ioctl(struct block_device
*, fmode_t
, unsigned, unsigned long);
2513 extern long compat_blkdev_ioctl(struct file
*, unsigned, unsigned long);
2514 extern int blkdev_get(struct block_device
*bdev
, fmode_t mode
, void *holder
);
2515 extern struct block_device
*blkdev_get_by_path(const char *path
, fmode_t mode
,
2517 extern struct block_device
*blkdev_get_by_dev(dev_t dev
, fmode_t mode
,
2519 extern void blkdev_put(struct block_device
*bdev
, fmode_t mode
);
2520 extern int __blkdev_reread_part(struct block_device
*bdev
);
2521 extern int blkdev_reread_part(struct block_device
*bdev
);
2524 extern int bd_link_disk_holder(struct block_device
*bdev
, struct gendisk
*disk
);
2525 extern void bd_unlink_disk_holder(struct block_device
*bdev
,
2526 struct gendisk
*disk
);
2528 static inline int bd_link_disk_holder(struct block_device
*bdev
,
2529 struct gendisk
*disk
)
2533 static inline void bd_unlink_disk_holder(struct block_device
*bdev
,
2534 struct gendisk
*disk
)
2541 #define CHRDEV_MAJOR_MAX 512
2542 /* Marks the bottom of the first segment of free char majors */
2543 #define CHRDEV_MAJOR_DYN_END 234
2544 /* Marks the top and bottom of the second segment of free char majors */
2545 #define CHRDEV_MAJOR_DYN_EXT_START 511
2546 #define CHRDEV_MAJOR_DYN_EXT_END 384
2548 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
2549 extern int register_chrdev_region(dev_t
, unsigned, const char *);
2550 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
2551 unsigned int count
, const char *name
,
2552 const struct file_operations
*fops
);
2553 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
2554 unsigned int count
, const char *name
);
2555 extern void unregister_chrdev_region(dev_t
, unsigned);
2556 extern void chrdev_show(struct seq_file
*,off_t
);
2558 static inline int register_chrdev(unsigned int major
, const char *name
,
2559 const struct file_operations
*fops
)
2561 return __register_chrdev(major
, 0, 256, name
, fops
);
2564 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2566 __unregister_chrdev(major
, 0, 256, name
);
2569 /* fs/block_dev.c */
2570 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2571 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2574 #define BLKDEV_MAJOR_MAX 512
2575 extern const char *__bdevname(dev_t
, char *buffer
);
2576 extern const char *bdevname(struct block_device
*bdev
, char *buffer
);
2577 extern struct block_device
*lookup_bdev(const char *);
2578 extern void blkdev_show(struct seq_file
*,off_t
);
2581 #define BLKDEV_MAJOR_MAX 0
2584 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2586 /* Invalid inode operations -- fs/bad_inode.c */
2587 extern void make_bad_inode(struct inode
*);
2588 extern bool is_bad_inode(struct inode
*);
2591 extern void check_disk_size_change(struct gendisk
*disk
,
2592 struct block_device
*bdev
, bool verbose
);
2593 extern int revalidate_disk(struct gendisk
*);
2594 extern int check_disk_change(struct block_device
*);
2595 extern int __invalidate_device(struct block_device
*, bool);
2596 extern int invalidate_partition(struct gendisk
*, int);
2598 unsigned long invalidate_mapping_pages(struct address_space
*mapping
,
2599 pgoff_t start
, pgoff_t end
);
2601 static inline void invalidate_remote_inode(struct inode
*inode
)
2603 if (S_ISREG(inode
->i_mode
) || S_ISDIR(inode
->i_mode
) ||
2604 S_ISLNK(inode
->i_mode
))
2605 invalidate_mapping_pages(inode
->i_mapping
, 0, -1);
2607 extern int invalidate_inode_pages2(struct address_space
*mapping
);
2608 extern int invalidate_inode_pages2_range(struct address_space
*mapping
,
2609 pgoff_t start
, pgoff_t end
);
2610 extern int write_inode_now(struct inode
*, int);
2611 extern int filemap_fdatawrite(struct address_space
*);
2612 extern int filemap_flush(struct address_space
*);
2613 extern int filemap_fdatawait_keep_errors(struct address_space
*mapping
);
2614 extern int filemap_fdatawait_range(struct address_space
*, loff_t lstart
,
2617 static inline int filemap_fdatawait(struct address_space
*mapping
)
2619 return filemap_fdatawait_range(mapping
, 0, LLONG_MAX
);
2622 extern bool filemap_range_has_page(struct address_space
*, loff_t lstart
,
2624 extern int __must_check
file_fdatawait_range(struct file
*file
, loff_t lstart
,
2626 extern int filemap_write_and_wait(struct address_space
*mapping
);
2627 extern int filemap_write_and_wait_range(struct address_space
*mapping
,
2628 loff_t lstart
, loff_t lend
);
2629 extern int __filemap_fdatawrite_range(struct address_space
*mapping
,
2630 loff_t start
, loff_t end
, int sync_mode
);
2631 extern int filemap_fdatawrite_range(struct address_space
*mapping
,
2632 loff_t start
, loff_t end
);
2633 extern int filemap_check_errors(struct address_space
*mapping
);
2634 extern void __filemap_set_wb_err(struct address_space
*mapping
, int err
);
2636 extern int __must_check
file_fdatawait_range(struct file
*file
, loff_t lstart
,
2638 extern int __must_check
file_check_and_advance_wb_err(struct file
*file
);
2639 extern int __must_check
file_write_and_wait_range(struct file
*file
,
2640 loff_t start
, loff_t end
);
2642 static inline int file_write_and_wait(struct file
*file
)
2644 return file_write_and_wait_range(file
, 0, LLONG_MAX
);
2648 * filemap_set_wb_err - set a writeback error on an address_space
2649 * @mapping: mapping in which to set writeback error
2650 * @err: error to be set in mapping
2652 * When writeback fails in some way, we must record that error so that
2653 * userspace can be informed when fsync and the like are called. We endeavor
2654 * to report errors on any file that was open at the time of the error. Some
2655 * internal callers also need to know when writeback errors have occurred.
2657 * When a writeback error occurs, most filesystems will want to call
2658 * filemap_set_wb_err to record the error in the mapping so that it will be
2659 * automatically reported whenever fsync is called on the file.
2661 static inline void filemap_set_wb_err(struct address_space
*mapping
, int err
)
2663 /* Fastpath for common case of no error */
2665 __filemap_set_wb_err(mapping
, err
);
2669 * filemap_check_wb_error - has an error occurred since the mark was sampled?
2670 * @mapping: mapping to check for writeback errors
2671 * @since: previously-sampled errseq_t
2673 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2674 * the given value was sampled.
2676 * If it has then report the latest error set, otherwise return 0.
2678 static inline int filemap_check_wb_err(struct address_space
*mapping
,
2681 return errseq_check(&mapping
->wb_err
, since
);
2685 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2686 * @mapping: mapping to be sampled
2688 * Writeback errors are always reported relative to a particular sample point
2689 * in the past. This function provides those sample points.
2691 static inline errseq_t
filemap_sample_wb_err(struct address_space
*mapping
)
2693 return errseq_sample(&mapping
->wb_err
);
2696 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
2698 extern int vfs_fsync(struct file
*file
, int datasync
);
2701 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2702 * to already be updated for the write, and will return either the amount
2703 * of bytes passed in, or an error if syncing the file failed.
2705 static inline ssize_t
generic_write_sync(struct kiocb
*iocb
, ssize_t count
)
2707 if (iocb
->ki_flags
& IOCB_DSYNC
) {
2708 int ret
= vfs_fsync_range(iocb
->ki_filp
,
2709 iocb
->ki_pos
- count
, iocb
->ki_pos
- 1,
2710 (iocb
->ki_flags
& IOCB_SYNC
) ? 0 : 1);
2718 extern void emergency_sync(void);
2719 extern void emergency_remount(void);
2721 extern sector_t
bmap(struct inode
*, sector_t
);
2723 extern int notify_change(struct dentry
*, struct iattr
*, struct inode
**);
2724 extern int inode_permission(struct inode
*, int);
2725 extern int generic_permission(struct inode
*, int);
2726 extern int __check_sticky(struct inode
*dir
, struct inode
*inode
);
2728 static inline bool execute_ok(struct inode
*inode
)
2730 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
2733 static inline void file_start_write(struct file
*file
)
2735 if (!S_ISREG(file_inode(file
)->i_mode
))
2737 __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, true);
2740 static inline bool file_start_write_trylock(struct file
*file
)
2742 if (!S_ISREG(file_inode(file
)->i_mode
))
2744 return __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, false);
2747 static inline void file_end_write(struct file
*file
)
2749 if (!S_ISREG(file_inode(file
)->i_mode
))
2751 __sb_end_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
);
2754 static inline int do_clone_file_range(struct file
*file_in
, loff_t pos_in
,
2755 struct file
*file_out
, loff_t pos_out
,
2760 file_start_write(file_out
);
2761 ret
= vfs_clone_file_range(file_in
, pos_in
, file_out
, pos_out
, len
);
2762 file_end_write(file_out
);
2768 * get_write_access() gets write permission for a file.
2769 * put_write_access() releases this write permission.
2770 * This is used for regular files.
2771 * We cannot support write (and maybe mmap read-write shared) accesses and
2772 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2773 * can have the following values:
2774 * 0: no writers, no VM_DENYWRITE mappings
2775 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2776 * > 0: (i_writecount) users are writing to the file.
2778 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2779 * except for the cases where we don't hold i_writecount yet. Then we need to
2780 * use {get,deny}_write_access() - these functions check the sign and refuse
2781 * to do the change if sign is wrong.
2783 static inline int get_write_access(struct inode
*inode
)
2785 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2787 static inline int deny_write_access(struct file
*file
)
2789 struct inode
*inode
= file_inode(file
);
2790 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2792 static inline void put_write_access(struct inode
* inode
)
2794 atomic_dec(&inode
->i_writecount
);
2796 static inline void allow_write_access(struct file
*file
)
2799 atomic_inc(&file_inode(file
)->i_writecount
);
2801 static inline bool inode_is_open_for_write(const struct inode
*inode
)
2803 return atomic_read(&inode
->i_writecount
) > 0;
2807 static inline void i_readcount_dec(struct inode
*inode
)
2809 BUG_ON(!atomic_read(&inode
->i_readcount
));
2810 atomic_dec(&inode
->i_readcount
);
2812 static inline void i_readcount_inc(struct inode
*inode
)
2814 atomic_inc(&inode
->i_readcount
);
2817 static inline void i_readcount_dec(struct inode
*inode
)
2821 static inline void i_readcount_inc(struct inode
*inode
)
2826 extern int do_pipe_flags(int *, int);
2828 #define __kernel_read_file_id(id) \
2829 id(UNKNOWN, unknown) \
2830 id(FIRMWARE, firmware) \
2831 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2832 id(MODULE, kernel-module) \
2833 id(KEXEC_IMAGE, kexec-image) \
2834 id(KEXEC_INITRAMFS, kexec-initramfs) \
2835 id(POLICY, security-policy) \
2836 id(X509_CERTIFICATE, x509-certificate) \
2839 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2840 #define __fid_stringify(dummy, str) #str,
2842 enum kernel_read_file_id
{
2843 __kernel_read_file_id(__fid_enumify
)
2846 static const char * const kernel_read_file_str
[] = {
2847 __kernel_read_file_id(__fid_stringify
)
2850 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id
)
2852 if ((unsigned)id
>= READING_MAX_ID
)
2853 return kernel_read_file_str
[READING_UNKNOWN
];
2855 return kernel_read_file_str
[id
];
2858 extern int kernel_read_file(struct file
*, void **, loff_t
*, loff_t
,
2859 enum kernel_read_file_id
);
2860 extern int kernel_read_file_from_path(const char *, void **, loff_t
*, loff_t
,
2861 enum kernel_read_file_id
);
2862 extern int kernel_read_file_from_fd(int, void **, loff_t
*, loff_t
,
2863 enum kernel_read_file_id
);
2864 extern ssize_t
kernel_read(struct file
*, void *, size_t, loff_t
*);
2865 extern ssize_t
kernel_write(struct file
*, const void *, size_t, loff_t
*);
2866 extern ssize_t
__kernel_write(struct file
*, const void *, size_t, loff_t
*);
2867 extern struct file
* open_exec(const char *);
2869 /* fs/dcache.c -- generic fs support functions */
2870 extern bool is_subdir(struct dentry
*, struct dentry
*);
2871 extern bool path_is_under(const struct path
*, const struct path
*);
2873 extern char *file_path(struct file
*, char *, int);
2875 #include <linux/err.h>
2877 /* needed for stackable file system support */
2878 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int whence
);
2880 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int whence
);
2882 extern int inode_init_always(struct super_block
*, struct inode
*);
2883 extern void inode_init_once(struct inode
*);
2884 extern void address_space_init_once(struct address_space
*mapping
);
2885 extern struct inode
* igrab(struct inode
*);
2886 extern ino_t
iunique(struct super_block
*, ino_t
);
2887 extern int inode_needs_sync(struct inode
*inode
);
2888 extern int generic_delete_inode(struct inode
*inode
);
2889 static inline int generic_drop_inode(struct inode
*inode
)
2891 return !inode
->i_nlink
|| inode_unhashed(inode
);
2894 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
2895 unsigned long hashval
, int (*test
)(struct inode
*, void *),
2897 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
2898 int (*test
)(struct inode
*, void *), void *data
);
2899 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
2901 extern struct inode
*inode_insert5(struct inode
*inode
, unsigned long hashval
,
2902 int (*test
)(struct inode
*, void *),
2903 int (*set
)(struct inode
*, void *),
2905 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
2906 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
2907 extern struct inode
*find_inode_nowait(struct super_block
*,
2909 int (*match
)(struct inode
*,
2910 unsigned long, void *),
2912 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
2913 extern int insert_inode_locked(struct inode
*);
2914 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2915 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
2917 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
2919 extern void unlock_new_inode(struct inode
*);
2920 extern unsigned int get_next_ino(void);
2921 extern void evict_inodes(struct super_block
*sb
);
2923 extern void __iget(struct inode
* inode
);
2924 extern void iget_failed(struct inode
*);
2925 extern void clear_inode(struct inode
*);
2926 extern void __destroy_inode(struct inode
*);
2927 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
2928 extern struct inode
*new_inode(struct super_block
*sb
);
2929 extern void free_inode_nonrcu(struct inode
*inode
);
2930 extern int should_remove_suid(struct dentry
*);
2931 extern int file_remove_privs(struct file
*);
2933 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
2934 static inline void insert_inode_hash(struct inode
*inode
)
2936 __insert_inode_hash(inode
, inode
->i_ino
);
2939 extern void __remove_inode_hash(struct inode
*);
2940 static inline void remove_inode_hash(struct inode
*inode
)
2942 if (!inode_unhashed(inode
) && !hlist_fake(&inode
->i_hash
))
2943 __remove_inode_hash(inode
);
2946 extern void inode_sb_list_add(struct inode
*inode
);
2949 extern int bdev_read_only(struct block_device
*);
2951 extern int set_blocksize(struct block_device
*, int);
2952 extern int sb_set_blocksize(struct super_block
*, int);
2953 extern int sb_min_blocksize(struct super_block
*, int);
2955 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
2956 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
2957 extern ssize_t
generic_write_checks(struct kiocb
*, struct iov_iter
*);
2958 extern ssize_t
generic_file_read_iter(struct kiocb
*, struct iov_iter
*);
2959 extern ssize_t
__generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2960 extern ssize_t
generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2961 extern ssize_t
generic_file_direct_write(struct kiocb
*, struct iov_iter
*);
2962 extern ssize_t
generic_perform_write(struct file
*, struct iov_iter
*, loff_t
);
2964 ssize_t
vfs_iter_read(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
2966 ssize_t
vfs_iter_write(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
2969 /* fs/block_dev.c */
2970 extern ssize_t
blkdev_read_iter(struct kiocb
*iocb
, struct iov_iter
*to
);
2971 extern ssize_t
blkdev_write_iter(struct kiocb
*iocb
, struct iov_iter
*from
);
2972 extern int blkdev_fsync(struct file
*filp
, loff_t start
, loff_t end
,
2974 extern void block_sync_page(struct page
*page
);
2977 extern ssize_t
generic_file_splice_read(struct file
*, loff_t
*,
2978 struct pipe_inode_info
*, size_t, unsigned int);
2979 extern ssize_t
iter_file_splice_write(struct pipe_inode_info
*,
2980 struct file
*, loff_t
*, size_t, unsigned int);
2981 extern ssize_t
generic_splice_sendpage(struct pipe_inode_info
*pipe
,
2982 struct file
*out
, loff_t
*, size_t len
, unsigned int flags
);
2983 extern long do_splice_direct(struct file
*in
, loff_t
*ppos
, struct file
*out
,
2984 loff_t
*opos
, size_t len
, unsigned int flags
);
2988 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
2989 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int whence
);
2990 extern loff_t
no_llseek(struct file
*file
, loff_t offset
, int whence
);
2991 extern loff_t
vfs_setpos(struct file
*file
, loff_t offset
, loff_t maxsize
);
2992 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int whence
);
2993 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
2994 int whence
, loff_t maxsize
, loff_t eof
);
2995 extern loff_t
fixed_size_llseek(struct file
*file
, loff_t offset
,
2996 int whence
, loff_t size
);
2997 extern loff_t
no_seek_end_llseek_size(struct file
*, loff_t
, int, loff_t
);
2998 extern loff_t
no_seek_end_llseek(struct file
*, loff_t
, int);
2999 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
3000 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
3003 typedef void (dio_submit_t
)(struct bio
*bio
, struct inode
*inode
,
3004 loff_t file_offset
);
3007 /* need locking between buffered and direct access */
3010 /* filesystem does not support filling holes */
3011 DIO_SKIP_HOLES
= 0x02,
3014 void dio_end_io(struct bio
*bio
);
3015 void dio_warn_stale_pagecache(struct file
*filp
);
3017 ssize_t
__blockdev_direct_IO(struct kiocb
*iocb
, struct inode
*inode
,
3018 struct block_device
*bdev
, struct iov_iter
*iter
,
3019 get_block_t get_block
,
3020 dio_iodone_t end_io
, dio_submit_t submit_io
,
3023 static inline ssize_t
blockdev_direct_IO(struct kiocb
*iocb
,
3024 struct inode
*inode
,
3025 struct iov_iter
*iter
,
3026 get_block_t get_block
)
3028 return __blockdev_direct_IO(iocb
, inode
, inode
->i_sb
->s_bdev
, iter
,
3029 get_block
, NULL
, NULL
, DIO_LOCKING
| DIO_SKIP_HOLES
);
3033 void inode_dio_wait(struct inode
*inode
);
3036 * inode_dio_begin - signal start of a direct I/O requests
3037 * @inode: inode the direct I/O happens on
3039 * This is called once we've finished processing a direct I/O request,
3040 * and is used to wake up callers waiting for direct I/O to be quiesced.
3042 static inline void inode_dio_begin(struct inode
*inode
)
3044 atomic_inc(&inode
->i_dio_count
);
3048 * inode_dio_end - signal finish of a direct I/O requests
3049 * @inode: inode the direct I/O happens on
3051 * This is called once we've finished processing a direct I/O request,
3052 * and is used to wake up callers waiting for direct I/O to be quiesced.
3054 static inline void inode_dio_end(struct inode
*inode
)
3056 if (atomic_dec_and_test(&inode
->i_dio_count
))
3057 wake_up_bit(&inode
->i_state
, __I_DIO_WAKEUP
);
3060 extern void inode_set_flags(struct inode
*inode
, unsigned int flags
,
3063 extern const struct file_operations generic_ro_fops
;
3065 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3067 extern int readlink_copy(char __user
*, int, const char *);
3068 extern int page_readlink(struct dentry
*, char __user
*, int);
3069 extern const char *page_get_link(struct dentry
*, struct inode
*,
3070 struct delayed_call
*);
3071 extern void page_put_link(void *);
3072 extern int __page_symlink(struct inode
*inode
, const char *symname
, int len
,
3074 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
3075 extern const struct inode_operations page_symlink_inode_operations
;
3076 extern void kfree_link(void *);
3077 extern void generic_fillattr(struct inode
*, struct kstat
*);
3078 extern int vfs_getattr_nosec(const struct path
*, struct kstat
*, u32
, unsigned int);
3079 extern int vfs_getattr(const struct path
*, struct kstat
*, u32
, unsigned int);
3080 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
3081 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
3082 void __inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
3083 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
3084 static inline loff_t
__inode_get_bytes(struct inode
*inode
)
3086 return (((loff_t
)inode
->i_blocks
) << 9) + inode
->i_bytes
;
3088 loff_t
inode_get_bytes(struct inode
*inode
);
3089 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
3090 const char *simple_get_link(struct dentry
*, struct inode
*,
3091 struct delayed_call
*);
3092 extern const struct inode_operations simple_symlink_inode_operations
;
3094 extern int iterate_dir(struct file
*, struct dir_context
*);
3096 extern int vfs_statx(int, const char __user
*, int, struct kstat
*, u32
);
3097 extern int vfs_statx_fd(unsigned int, struct kstat
*, u32
, unsigned int);
3099 static inline int vfs_stat(const char __user
*filename
, struct kstat
*stat
)
3101 return vfs_statx(AT_FDCWD
, filename
, AT_NO_AUTOMOUNT
,
3102 stat
, STATX_BASIC_STATS
);
3104 static inline int vfs_lstat(const char __user
*name
, struct kstat
*stat
)
3106 return vfs_statx(AT_FDCWD
, name
, AT_SYMLINK_NOFOLLOW
| AT_NO_AUTOMOUNT
,
3107 stat
, STATX_BASIC_STATS
);
3109 static inline int vfs_fstatat(int dfd
, const char __user
*filename
,
3110 struct kstat
*stat
, int flags
)
3112 return vfs_statx(dfd
, filename
, flags
| AT_NO_AUTOMOUNT
,
3113 stat
, STATX_BASIC_STATS
);
3115 static inline int vfs_fstat(int fd
, struct kstat
*stat
)
3117 return vfs_statx_fd(fd
, stat
, STATX_BASIC_STATS
, 0);
3121 extern const char *vfs_get_link(struct dentry
*, struct delayed_call
*);
3122 extern int vfs_readlink(struct dentry
*, char __user
*, int);
3124 extern int __generic_block_fiemap(struct inode
*inode
,
3125 struct fiemap_extent_info
*fieinfo
,
3126 loff_t start
, loff_t len
,
3127 get_block_t
*get_block
);
3128 extern int generic_block_fiemap(struct inode
*inode
,
3129 struct fiemap_extent_info
*fieinfo
, u64 start
,
3130 u64 len
, get_block_t
*get_block
);
3132 extern struct file_system_type
*get_filesystem(struct file_system_type
*fs
);
3133 extern void put_filesystem(struct file_system_type
*fs
);
3134 extern struct file_system_type
*get_fs_type(const char *name
);
3135 extern struct super_block
*get_super(struct block_device
*);
3136 extern struct super_block
*get_super_thawed(struct block_device
*);
3137 extern struct super_block
*get_super_exclusive_thawed(struct block_device
*bdev
);
3138 extern struct super_block
*get_active_super(struct block_device
*bdev
);
3139 extern void drop_super(struct super_block
*sb
);
3140 extern void drop_super_exclusive(struct super_block
*sb
);
3141 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
3142 extern void iterate_supers_type(struct file_system_type
*,
3143 void (*)(struct super_block
*, void *), void *);
3145 extern int dcache_dir_open(struct inode
*, struct file
*);
3146 extern int dcache_dir_close(struct inode
*, struct file
*);
3147 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
3148 extern int dcache_readdir(struct file
*, struct dir_context
*);
3149 extern int simple_setattr(struct dentry
*, struct iattr
*);
3150 extern int simple_getattr(const struct path
*, struct kstat
*, u32
, unsigned int);
3151 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
3152 extern int simple_open(struct inode
*inode
, struct file
*file
);
3153 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
3154 extern int simple_unlink(struct inode
*, struct dentry
*);
3155 extern int simple_rmdir(struct inode
*, struct dentry
*);
3156 extern int simple_rename(struct inode
*, struct dentry
*,
3157 struct inode
*, struct dentry
*, unsigned int);
3158 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
3159 extern int noop_set_page_dirty(struct page
*page
);
3160 extern void noop_invalidatepage(struct page
*page
, unsigned int offset
,
3161 unsigned int length
);
3162 extern ssize_t
noop_direct_IO(struct kiocb
*iocb
, struct iov_iter
*iter
);
3163 extern int simple_empty(struct dentry
*);
3164 extern int simple_readpage(struct file
*file
, struct page
*page
);
3165 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
3166 loff_t pos
, unsigned len
, unsigned flags
,
3167 struct page
**pagep
, void **fsdata
);
3168 extern int simple_write_end(struct file
*file
, struct address_space
*mapping
,
3169 loff_t pos
, unsigned len
, unsigned copied
,
3170 struct page
*page
, void *fsdata
);
3171 extern int always_delete_dentry(const struct dentry
*);
3172 extern struct inode
*alloc_anon_inode(struct super_block
*);
3173 extern int simple_nosetlease(struct file
*, long, struct file_lock
**, void **);
3174 extern const struct dentry_operations simple_dentry_operations
;
3176 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
3177 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
3178 extern const struct file_operations simple_dir_operations
;
3179 extern const struct inode_operations simple_dir_inode_operations
;
3180 extern void make_empty_dir_inode(struct inode
*inode
);
3181 extern bool is_empty_dir_inode(struct inode
*inode
);
3182 struct tree_descr
{ const char *name
; const struct file_operations
*ops
; int mode
; };
3183 struct dentry
*d_alloc_name(struct dentry
*, const char *);
3184 extern int simple_fill_super(struct super_block
*, unsigned long,
3185 const struct tree_descr
*);
3186 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
3187 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
3189 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
3190 loff_t
*ppos
, const void *from
, size_t available
);
3191 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
3192 const void __user
*from
, size_t count
);
3194 extern int __generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3195 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3197 extern int generic_check_addressable(unsigned, u64
);
3199 #ifdef CONFIG_MIGRATION
3200 extern int buffer_migrate_page(struct address_space
*,
3201 struct page
*, struct page
*,
3204 #define buffer_migrate_page NULL
3207 extern int setattr_prepare(struct dentry
*, struct iattr
*);
3208 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
3209 extern void setattr_copy(struct inode
*inode
, const struct iattr
*attr
);
3211 extern int file_update_time(struct file
*file
);
3213 static inline bool io_is_direct(struct file
*filp
)
3215 return (filp
->f_flags
& O_DIRECT
) || IS_DAX(filp
->f_mapping
->host
);
3218 static inline bool vma_is_dax(struct vm_area_struct
*vma
)
3220 return vma
->vm_file
&& IS_DAX(vma
->vm_file
->f_mapping
->host
);
3223 static inline bool vma_is_fsdax(struct vm_area_struct
*vma
)
3225 struct inode
*inode
;
3229 if (!vma_is_dax(vma
))
3231 inode
= file_inode(vma
->vm_file
);
3232 if (S_ISCHR(inode
->i_mode
))
3233 return false; /* device-dax */
3237 static inline int iocb_flags(struct file
*file
)
3240 if (file
->f_flags
& O_APPEND
)
3242 if (io_is_direct(file
))
3244 if ((file
->f_flags
& O_DSYNC
) || IS_SYNC(file
->f_mapping
->host
))
3246 if (file
->f_flags
& __O_SYNC
)
3251 static inline int kiocb_set_rw_flags(struct kiocb
*ki
, rwf_t flags
)
3253 if (unlikely(flags
& ~RWF_SUPPORTED
))
3256 if (flags
& RWF_NOWAIT
) {
3257 if (!(ki
->ki_filp
->f_mode
& FMODE_NOWAIT
))
3259 ki
->ki_flags
|= IOCB_NOWAIT
;
3261 if (flags
& RWF_HIPRI
)
3262 ki
->ki_flags
|= IOCB_HIPRI
;
3263 if (flags
& RWF_DSYNC
)
3264 ki
->ki_flags
|= IOCB_DSYNC
;
3265 if (flags
& RWF_SYNC
)
3266 ki
->ki_flags
|= (IOCB_DSYNC
| IOCB_SYNC
);
3267 if (flags
& RWF_APPEND
)
3268 ki
->ki_flags
|= IOCB_APPEND
;
3272 static inline ino_t
parent_ino(struct dentry
*dentry
)
3277 * Don't strictly need d_lock here? If the parent ino could change
3278 * then surely we'd have a deeper race in the caller?
3280 spin_lock(&dentry
->d_lock
);
3281 res
= dentry
->d_parent
->d_inode
->i_ino
;
3282 spin_unlock(&dentry
->d_lock
);
3286 /* Transaction based IO helpers */
3289 * An argresp is stored in an allocated page and holds the
3290 * size of the argument or response, along with its content
3292 struct simple_transaction_argresp
{
3297 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3299 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
3301 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
3302 size_t size
, loff_t
*pos
);
3303 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
3305 void simple_transaction_set(struct file
*file
, size_t n
);
3308 * simple attribute files
3310 * These attributes behave similar to those in sysfs:
3312 * Writing to an attribute immediately sets a value, an open file can be
3313 * written to multiple times.
3315 * Reading from an attribute creates a buffer from the value that might get
3316 * read with multiple read calls. When the attribute has been read
3317 * completely, no further read calls are possible until the file is opened
3320 * All attributes contain a text representation of a numeric value
3321 * that are accessed with the get() and set() functions.
3323 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3324 static int __fops ## _open(struct inode *inode, struct file *file) \
3326 __simple_attr_check_format(__fmt, 0ull); \
3327 return simple_attr_open(inode, file, __get, __set, __fmt); \
3329 static const struct file_operations __fops = { \
3330 .owner = THIS_MODULE, \
3331 .open = __fops ## _open, \
3332 .release = simple_attr_release, \
3333 .read = simple_attr_read, \
3334 .write = simple_attr_write, \
3335 .llseek = generic_file_llseek, \
3338 static inline __printf(1, 2)
3339 void __simple_attr_check_format(const char *fmt
, ...)
3341 /* don't do anything, just let the compiler check the arguments; */
3344 int simple_attr_open(struct inode
*inode
, struct file
*file
,
3345 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
3347 int simple_attr_release(struct inode
*inode
, struct file
*file
);
3348 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
3349 size_t len
, loff_t
*ppos
);
3350 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
3351 size_t len
, loff_t
*ppos
);
3354 int proc_nr_files(struct ctl_table
*table
, int write
,
3355 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3356 int proc_nr_dentry(struct ctl_table
*table
, int write
,
3357 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3358 int proc_nr_inodes(struct ctl_table
*table
, int write
,
3359 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3360 int __init
get_filesystem_list(char *buf
);
3362 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3363 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3365 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3366 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3367 (flag & __FMODE_NONOTIFY)))
3369 static inline bool is_sxid(umode_t mode
)
3371 return (mode
& S_ISUID
) || ((mode
& S_ISGID
) && (mode
& S_IXGRP
));
3374 static inline int check_sticky(struct inode
*dir
, struct inode
*inode
)
3376 if (!(dir
->i_mode
& S_ISVTX
))
3379 return __check_sticky(dir
, inode
);
3382 static inline void inode_has_no_xattr(struct inode
*inode
)
3384 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& SB_NOSEC
))
3385 inode
->i_flags
|= S_NOSEC
;
3388 static inline bool is_root_inode(struct inode
*inode
)
3390 return inode
== inode
->i_sb
->s_root
->d_inode
;
3393 static inline bool dir_emit(struct dir_context
*ctx
,
3394 const char *name
, int namelen
,
3395 u64 ino
, unsigned type
)
3397 return ctx
->actor(ctx
, name
, namelen
, ctx
->pos
, ino
, type
) == 0;
3399 static inline bool dir_emit_dot(struct file
*file
, struct dir_context
*ctx
)
3401 return ctx
->actor(ctx
, ".", 1, ctx
->pos
,
3402 file
->f_path
.dentry
->d_inode
->i_ino
, DT_DIR
) == 0;
3404 static inline bool dir_emit_dotdot(struct file
*file
, struct dir_context
*ctx
)
3406 return ctx
->actor(ctx
, "..", 2, ctx
->pos
,
3407 parent_ino(file
->f_path
.dentry
), DT_DIR
) == 0;
3409 static inline bool dir_emit_dots(struct file
*file
, struct dir_context
*ctx
)
3411 if (ctx
->pos
== 0) {
3412 if (!dir_emit_dot(file
, ctx
))
3416 if (ctx
->pos
== 1) {
3417 if (!dir_emit_dotdot(file
, ctx
))
3423 static inline bool dir_relax(struct inode
*inode
)
3425 inode_unlock(inode
);
3427 return !IS_DEADDIR(inode
);
3430 static inline bool dir_relax_shared(struct inode
*inode
)
3432 inode_unlock_shared(inode
);
3433 inode_lock_shared(inode
);
3434 return !IS_DEADDIR(inode
);
3437 extern bool path_noexec(const struct path
*path
);
3438 extern void inode_nohighmem(struct inode
*inode
);
3440 #endif /* _LINUX_FS_H */