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/rbtree.h>
17 #include <linux/init.h>
18 #include <linux/pid.h>
19 #include <linux/bug.h>
20 #include <linux/mutex.h>
21 #include <linux/rwsem.h>
22 #include <linux/mm_types.h>
23 #include <linux/capability.h>
24 #include <linux/semaphore.h>
25 #include <linux/fcntl.h>
26 #include <linux/fiemap.h>
27 #include <linux/rculist_bl.h>
28 #include <linux/atomic.h>
29 #include <linux/shrinker.h>
30 #include <linux/migrate_mode.h>
31 #include <linux/uidgid.h>
32 #include <linux/lockdep.h>
33 #include <linux/percpu-rwsem.h>
34 #include <linux/workqueue.h>
35 #include <linux/delayed_call.h>
36 #include <linux/uuid.h>
37 #include <linux/errseq.h>
39 #include <asm/byteorder.h>
40 #include <uapi/linux/fs.h>
42 struct backing_dev_info
;
45 struct export_operations
;
50 struct pipe_inode_info
;
51 struct poll_table_struct
;
53 struct vm_area_struct
;
56 struct swap_info_struct
;
58 struct workqueue_struct
;
61 struct fscrypt_operations
;
63 extern void __init
inode_init(void);
64 extern void __init
inode_init_early(void);
65 extern void __init
files_init(void);
66 extern void __init
files_maxfiles_init(void);
68 extern struct files_stat_struct files_stat
;
69 extern unsigned long get_max_files(void);
70 extern unsigned int sysctl_nr_open
;
71 extern struct inodes_stat_t inodes_stat
;
72 extern int leases_enable
, lease_break_time
;
73 extern int sysctl_protected_symlinks
;
74 extern int sysctl_protected_hardlinks
;
76 typedef __kernel_rwf_t rwf_t
;
79 typedef int (get_block_t
)(struct inode
*inode
, sector_t iblock
,
80 struct buffer_head
*bh_result
, int create
);
81 typedef int (dio_iodone_t
)(struct kiocb
*iocb
, loff_t offset
,
82 ssize_t bytes
, void *private);
84 #define MAY_EXEC 0x00000001
85 #define MAY_WRITE 0x00000002
86 #define MAY_READ 0x00000004
87 #define MAY_APPEND 0x00000008
88 #define MAY_ACCESS 0x00000010
89 #define MAY_OPEN 0x00000020
90 #define MAY_CHDIR 0x00000040
91 /* called from RCU mode, don't block */
92 #define MAY_NOT_BLOCK 0x00000080
95 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
96 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
99 /* file is open for reading */
100 #define FMODE_READ ((__force fmode_t)0x1)
101 /* file is open for writing */
102 #define FMODE_WRITE ((__force fmode_t)0x2)
103 /* file is seekable */
104 #define FMODE_LSEEK ((__force fmode_t)0x4)
105 /* file can be accessed using pread */
106 #define FMODE_PREAD ((__force fmode_t)0x8)
107 /* file can be accessed using pwrite */
108 #define FMODE_PWRITE ((__force fmode_t)0x10)
109 /* File is opened for execution with sys_execve / sys_uselib */
110 #define FMODE_EXEC ((__force fmode_t)0x20)
111 /* File is opened with O_NDELAY (only set for block devices) */
112 #define FMODE_NDELAY ((__force fmode_t)0x40)
113 /* File is opened with O_EXCL (only set for block devices) */
114 #define FMODE_EXCL ((__force fmode_t)0x80)
115 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
116 (specialy hack for floppy.c) */
117 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
118 /* 32bit hashes as llseek() offset (for directories) */
119 #define FMODE_32BITHASH ((__force fmode_t)0x200)
120 /* 64bit hashes as llseek() offset (for directories) */
121 #define FMODE_64BITHASH ((__force fmode_t)0x400)
124 * Don't update ctime and mtime.
126 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
127 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
129 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
131 /* Expect random access pattern */
132 #define FMODE_RANDOM ((__force fmode_t)0x1000)
134 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
135 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
137 /* File is opened with O_PATH; almost nothing can be done with it */
138 #define FMODE_PATH ((__force fmode_t)0x4000)
140 /* File needs atomic accesses to f_pos */
141 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
142 /* Write access to underlying fs */
143 #define FMODE_WRITER ((__force fmode_t)0x10000)
144 /* Has read method(s) */
145 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
146 /* Has write method(s) */
147 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
149 /* File was opened by fanotify and shouldn't generate fanotify events */
150 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
152 /* File is capable of returning -EAGAIN if I/O will block */
153 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
156 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
157 * that indicates that they should check the contents of the iovec are
158 * valid, but not check the memory that the iovec elements
161 #define CHECK_IOVEC_ONLY -1
164 * Attribute flags. These should be or-ed together to figure out what
167 #define ATTR_MODE (1 << 0)
168 #define ATTR_UID (1 << 1)
169 #define ATTR_GID (1 << 2)
170 #define ATTR_SIZE (1 << 3)
171 #define ATTR_ATIME (1 << 4)
172 #define ATTR_MTIME (1 << 5)
173 #define ATTR_CTIME (1 << 6)
174 #define ATTR_ATIME_SET (1 << 7)
175 #define ATTR_MTIME_SET (1 << 8)
176 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
177 #define ATTR_ATTR_FLAG (1 << 10)
178 #define ATTR_KILL_SUID (1 << 11)
179 #define ATTR_KILL_SGID (1 << 12)
180 #define ATTR_FILE (1 << 13)
181 #define ATTR_KILL_PRIV (1 << 14)
182 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
183 #define ATTR_TIMES_SET (1 << 16)
184 #define ATTR_TOUCH (1 << 17)
187 * Whiteout is represented by a char device. The following constants define the
188 * mode and device number to use.
190 #define WHITEOUT_MODE 0
191 #define WHITEOUT_DEV 0
194 * This is the Inode Attributes structure, used for notify_change(). It
195 * uses the above definitions as flags, to know which values have changed.
196 * Also, in this manner, a Filesystem can look at only the values it cares
197 * about. Basically, these are the attributes that the VFS layer can
198 * request to change from the FS layer.
200 * Derek Atkins <warlord@MIT.EDU> 94-10-20
203 unsigned int ia_valid
;
208 struct timespec ia_atime
;
209 struct timespec ia_mtime
;
210 struct timespec ia_ctime
;
213 * Not an attribute, but an auxiliary info for filesystems wanting to
214 * implement an ftruncate() like method. NOTE: filesystem should
215 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
217 struct file
*ia_file
;
221 * Includes for diskquotas.
223 #include <linux/quota.h>
226 * Maximum number of layers of fs stack. Needs to be limited to
227 * prevent kernel stack overflow
229 #define FILESYSTEM_MAX_STACK_DEPTH 2
232 * enum positive_aop_returns - aop return codes with specific semantics
234 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
235 * completed, that the page is still locked, and
236 * should be considered active. The VM uses this hint
237 * to return the page to the active list -- it won't
238 * be a candidate for writeback again in the near
239 * future. Other callers must be careful to unlock
240 * the page if they get this return. Returned by
243 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
244 * unlocked it and the page might have been truncated.
245 * The caller should back up to acquiring a new page and
246 * trying again. The aop will be taking reasonable
247 * precautions not to livelock. If the caller held a page
248 * reference, it should drop it before retrying. Returned
251 * address_space_operation functions return these large constants to indicate
252 * special semantics to the caller. These are much larger than the bytes in a
253 * page to allow for functions that return the number of bytes operated on in a
257 enum positive_aop_returns
{
258 AOP_WRITEPAGE_ACTIVATE
= 0x80000,
259 AOP_TRUNCATED_PAGE
= 0x80001,
262 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
263 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
264 * helper code (eg buffer layer)
265 * to clear GFP_FS from alloc */
268 * oh the beauties of C type declarations.
271 struct address_space
;
272 struct writeback_control
;
275 * Write life time hint values.
278 WRITE_LIFE_NOT_SET
= 0,
279 WRITE_LIFE_NONE
= RWH_WRITE_LIFE_NONE
,
280 WRITE_LIFE_SHORT
= RWH_WRITE_LIFE_SHORT
,
281 WRITE_LIFE_MEDIUM
= RWH_WRITE_LIFE_MEDIUM
,
282 WRITE_LIFE_LONG
= RWH_WRITE_LIFE_LONG
,
283 WRITE_LIFE_EXTREME
= RWH_WRITE_LIFE_EXTREME
,
286 #define IOCB_EVENTFD (1 << 0)
287 #define IOCB_APPEND (1 << 1)
288 #define IOCB_DIRECT (1 << 2)
289 #define IOCB_HIPRI (1 << 3)
290 #define IOCB_DSYNC (1 << 4)
291 #define IOCB_SYNC (1 << 5)
292 #define IOCB_WRITE (1 << 6)
293 #define IOCB_NOWAIT (1 << 7)
296 struct file
*ki_filp
;
298 void (*ki_complete
)(struct kiocb
*iocb
, long ret
, long ret2
);
301 enum rw_hint ki_hint
;
302 } __randomize_layout
;
304 static inline bool is_sync_kiocb(struct kiocb
*kiocb
)
306 return kiocb
->ki_complete
== NULL
;
310 * "descriptor" for what we're up to with a read.
311 * This allows us to use the same read code yet
312 * have multiple different users of the data that
313 * we read from a file.
315 * The simplest case just copies the data to user
328 typedef int (*read_actor_t
)(read_descriptor_t
*, struct page
*,
329 unsigned long, unsigned long);
331 struct address_space_operations
{
332 int (*writepage
)(struct page
*page
, struct writeback_control
*wbc
);
333 int (*readpage
)(struct file
*, struct page
*);
335 /* Write back some dirty pages from this mapping. */
336 int (*writepages
)(struct address_space
*, struct writeback_control
*);
338 /* Set a page dirty. Return true if this dirtied it */
339 int (*set_page_dirty
)(struct page
*page
);
341 int (*readpages
)(struct file
*filp
, struct address_space
*mapping
,
342 struct list_head
*pages
, unsigned nr_pages
);
344 int (*write_begin
)(struct file
*, struct address_space
*mapping
,
345 loff_t pos
, unsigned len
, unsigned flags
,
346 struct page
**pagep
, void **fsdata
);
347 int (*write_end
)(struct file
*, struct address_space
*mapping
,
348 loff_t pos
, unsigned len
, unsigned copied
,
349 struct page
*page
, void *fsdata
);
351 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
352 sector_t (*bmap
)(struct address_space
*, sector_t
);
353 void (*invalidatepage
) (struct page
*, unsigned int, unsigned int);
354 int (*releasepage
) (struct page
*, gfp_t
);
355 void (*freepage
)(struct page
*);
356 ssize_t (*direct_IO
)(struct kiocb
*, struct iov_iter
*iter
);
358 * migrate the contents of a page to the specified target. If
359 * migrate_mode is MIGRATE_ASYNC, it must not block.
361 int (*migratepage
) (struct address_space
*,
362 struct page
*, struct page
*, enum migrate_mode
);
363 bool (*isolate_page
)(struct page
*, isolate_mode_t
);
364 void (*putback_page
)(struct page
*);
365 int (*launder_page
) (struct page
*);
366 int (*is_partially_uptodate
) (struct page
*, unsigned long,
368 void (*is_dirty_writeback
) (struct page
*, bool *, bool *);
369 int (*error_remove_page
)(struct address_space
*, struct page
*);
371 /* swapfile support */
372 int (*swap_activate
)(struct swap_info_struct
*sis
, struct file
*file
,
374 void (*swap_deactivate
)(struct file
*file
);
377 extern const struct address_space_operations empty_aops
;
380 * pagecache_write_begin/pagecache_write_end must be used by general code
381 * to write into the pagecache.
383 int pagecache_write_begin(struct file
*, struct address_space
*mapping
,
384 loff_t pos
, unsigned len
, unsigned flags
,
385 struct page
**pagep
, void **fsdata
);
387 int pagecache_write_end(struct file
*, struct address_space
*mapping
,
388 loff_t pos
, unsigned len
, unsigned copied
,
389 struct page
*page
, void *fsdata
);
391 struct address_space
{
392 struct inode
*host
; /* owner: inode, block_device */
393 struct radix_tree_root page_tree
; /* radix tree of all pages */
394 spinlock_t tree_lock
; /* and lock protecting it */
395 atomic_t i_mmap_writable
;/* count VM_SHARED mappings */
396 struct rb_root_cached i_mmap
; /* tree of private and shared mappings */
397 struct rw_semaphore i_mmap_rwsem
; /* protect tree, count, list */
398 /* Protected by tree_lock together with the radix tree */
399 unsigned long nrpages
; /* number of total pages */
400 /* number of shadow or DAX exceptional entries */
401 unsigned long nrexceptional
;
402 pgoff_t writeback_index
;/* writeback starts here */
403 const struct address_space_operations
*a_ops
; /* methods */
404 unsigned long flags
; /* error bits */
405 spinlock_t private_lock
; /* for use by the address_space */
406 gfp_t gfp_mask
; /* implicit gfp mask for allocations */
407 struct list_head private_list
; /* for use by the address_space */
408 void *private_data
; /* ditto */
410 } __attribute__((aligned(sizeof(long)))) __randomize_layout
;
412 * On most architectures that alignment is already the case; but
413 * must be enforced here for CRIS, to let the least significant bit
414 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
416 struct request_queue
;
418 struct block_device
{
419 dev_t bd_dev
; /* not a kdev_t - it's a search key */
421 struct inode
* bd_inode
; /* will die */
422 struct super_block
* bd_super
;
423 struct mutex bd_mutex
; /* open/close mutex */
427 bool bd_write_holder
;
429 struct list_head bd_holder_disks
;
431 struct block_device
* bd_contains
;
432 unsigned bd_block_size
;
434 struct hd_struct
* bd_part
;
435 /* number of times partitions within this device have been opened. */
436 unsigned bd_part_count
;
438 struct gendisk
* bd_disk
;
439 struct request_queue
* bd_queue
;
440 struct backing_dev_info
*bd_bdi
;
441 struct list_head bd_list
;
443 * Private data. You must have bd_claim'ed the block_device
444 * to use this. NOTE: bd_claim allows an owner to claim
445 * the same device multiple times, the owner must take special
446 * care to not mess up bd_private for that case.
448 unsigned long bd_private
;
450 /* The counter of freeze processes */
451 int bd_fsfreeze_count
;
452 /* Mutex for freeze */
453 struct mutex bd_fsfreeze_mutex
;
454 } __randomize_layout
;
457 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
460 #define PAGECACHE_TAG_DIRTY 0
461 #define PAGECACHE_TAG_WRITEBACK 1
462 #define PAGECACHE_TAG_TOWRITE 2
464 int mapping_tagged(struct address_space
*mapping
, int tag
);
466 static inline void i_mmap_lock_write(struct address_space
*mapping
)
468 down_write(&mapping
->i_mmap_rwsem
);
471 static inline void i_mmap_unlock_write(struct address_space
*mapping
)
473 up_write(&mapping
->i_mmap_rwsem
);
476 static inline void i_mmap_lock_read(struct address_space
*mapping
)
478 down_read(&mapping
->i_mmap_rwsem
);
481 static inline void i_mmap_unlock_read(struct address_space
*mapping
)
483 up_read(&mapping
->i_mmap_rwsem
);
487 * Might pages of this file be mapped into userspace?
489 static inline int mapping_mapped(struct address_space
*mapping
)
491 return !RB_EMPTY_ROOT(&mapping
->i_mmap
.rb_root
);
495 * Might pages of this file have been modified in userspace?
496 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
497 * marks vma as VM_SHARED if it is shared, and the file was opened for
498 * writing i.e. vma may be mprotected writable even if now readonly.
500 * If i_mmap_writable is negative, no new writable mappings are allowed. You
501 * can only deny writable mappings, if none exists right now.
503 static inline int mapping_writably_mapped(struct address_space
*mapping
)
505 return atomic_read(&mapping
->i_mmap_writable
) > 0;
508 static inline int mapping_map_writable(struct address_space
*mapping
)
510 return atomic_inc_unless_negative(&mapping
->i_mmap_writable
) ?
514 static inline void mapping_unmap_writable(struct address_space
*mapping
)
516 atomic_dec(&mapping
->i_mmap_writable
);
519 static inline int mapping_deny_writable(struct address_space
*mapping
)
521 return atomic_dec_unless_positive(&mapping
->i_mmap_writable
) ?
525 static inline void mapping_allow_writable(struct address_space
*mapping
)
527 atomic_inc(&mapping
->i_mmap_writable
);
531 * Use sequence counter to get consistent i_size on 32-bit processors.
533 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
534 #include <linux/seqlock.h>
535 #define __NEED_I_SIZE_ORDERED
536 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
538 #define i_size_ordered_init(inode) do { } while (0)
542 #define ACL_NOT_CACHED ((void *)(-1))
543 #define ACL_DONT_CACHE ((void *)(-3))
545 static inline struct posix_acl
*
546 uncached_acl_sentinel(struct task_struct
*task
)
548 return (void *)task
+ 1;
552 is_uncached_acl(struct posix_acl
*acl
)
554 return (long)acl
& 1;
557 #define IOP_FASTPERM 0x0001
558 #define IOP_LOOKUP 0x0002
559 #define IOP_NOFOLLOW 0x0004
560 #define IOP_XATTR 0x0008
561 #define IOP_DEFAULT_READLINK 0x0010
563 struct fsnotify_mark_connector
;
566 * Keep mostly read-only and often accessed (especially for
567 * the RCU path lookup and 'stat' data) fields at the beginning
568 * of the 'struct inode'
572 unsigned short i_opflags
;
575 unsigned int i_flags
;
577 #ifdef CONFIG_FS_POSIX_ACL
578 struct posix_acl
*i_acl
;
579 struct posix_acl
*i_default_acl
;
582 const struct inode_operations
*i_op
;
583 struct super_block
*i_sb
;
584 struct address_space
*i_mapping
;
586 #ifdef CONFIG_SECURITY
590 /* Stat data, not accessed from path walking */
593 * Filesystems may only read i_nlink directly. They shall use the
594 * following functions for modification:
596 * (set|clear|inc|drop)_nlink
597 * inode_(inc|dec)_link_count
600 const unsigned int i_nlink
;
601 unsigned int __i_nlink
;
605 struct timespec i_atime
;
606 struct timespec i_mtime
;
607 struct timespec i_ctime
;
608 spinlock_t i_lock
; /* i_blocks, i_bytes, maybe i_size */
609 unsigned short i_bytes
;
610 unsigned int i_blkbits
;
611 enum rw_hint i_write_hint
;
614 #ifdef __NEED_I_SIZE_ORDERED
615 seqcount_t i_size_seqcount
;
619 unsigned long i_state
;
620 struct rw_semaphore i_rwsem
;
622 unsigned long dirtied_when
; /* jiffies of first dirtying */
623 unsigned long dirtied_time_when
;
625 struct hlist_node i_hash
;
626 struct list_head i_io_list
; /* backing dev IO list */
627 #ifdef CONFIG_CGROUP_WRITEBACK
628 struct bdi_writeback
*i_wb
; /* the associated cgroup wb */
630 /* foreign inode detection, see wbc_detach_inode() */
632 u16 i_wb_frn_avg_time
;
633 u16 i_wb_frn_history
;
635 struct list_head i_lru
; /* inode LRU list */
636 struct list_head i_sb_list
;
637 struct list_head i_wb_list
; /* backing dev writeback list */
639 struct hlist_head i_dentry
;
640 struct rcu_head i_rcu
;
644 atomic_t i_dio_count
;
645 atomic_t i_writecount
;
647 atomic_t i_readcount
; /* struct files open RO */
649 const struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
650 struct file_lock_context
*i_flctx
;
651 struct address_space i_data
;
652 struct list_head i_devices
;
654 struct pipe_inode_info
*i_pipe
;
655 struct block_device
*i_bdev
;
663 #ifdef CONFIG_FSNOTIFY
664 __u32 i_fsnotify_mask
; /* all events this inode cares about */
665 struct fsnotify_mark_connector __rcu
*i_fsnotify_marks
;
668 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
669 struct fscrypt_info
*i_crypt_info
;
672 void *i_private
; /* fs or device private pointer */
673 } __randomize_layout
;
675 static inline unsigned int i_blocksize(const struct inode
*node
)
677 return (1 << node
->i_blkbits
);
680 static inline int inode_unhashed(struct inode
*inode
)
682 return hlist_unhashed(&inode
->i_hash
);
686 * inode->i_mutex nesting subclasses for the lock validator:
688 * 0: the object of the current VFS operation
692 * 4: second non-directory
693 * 5: second parent (when locking independent directories in rename)
695 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
696 * non-directories at once.
698 * The locking order between these classes is
699 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
701 enum inode_i_mutex_lock_class
711 static inline void inode_lock(struct inode
*inode
)
713 down_write(&inode
->i_rwsem
);
716 static inline void inode_unlock(struct inode
*inode
)
718 up_write(&inode
->i_rwsem
);
721 static inline void inode_lock_shared(struct inode
*inode
)
723 down_read(&inode
->i_rwsem
);
726 static inline void inode_unlock_shared(struct inode
*inode
)
728 up_read(&inode
->i_rwsem
);
731 static inline int inode_trylock(struct inode
*inode
)
733 return down_write_trylock(&inode
->i_rwsem
);
736 static inline int inode_trylock_shared(struct inode
*inode
)
738 return down_read_trylock(&inode
->i_rwsem
);
741 static inline int inode_is_locked(struct inode
*inode
)
743 return rwsem_is_locked(&inode
->i_rwsem
);
746 static inline void inode_lock_nested(struct inode
*inode
, unsigned subclass
)
748 down_write_nested(&inode
->i_rwsem
, subclass
);
751 void lock_two_nondirectories(struct inode
*, struct inode
*);
752 void unlock_two_nondirectories(struct inode
*, struct inode
*);
755 * NOTE: in a 32bit arch with a preemptable kernel and
756 * an UP compile the i_size_read/write must be atomic
757 * with respect to the local cpu (unlike with preempt disabled),
758 * but they don't need to be atomic with respect to other cpus like in
759 * true SMP (so they need either to either locally disable irq around
760 * the read or for example on x86 they can be still implemented as a
761 * cmpxchg8b without the need of the lock prefix). For SMP compiles
762 * and 64bit archs it makes no difference if preempt is enabled or not.
764 static inline loff_t
i_size_read(const struct inode
*inode
)
766 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
771 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
772 i_size
= inode
->i_size
;
773 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
775 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
779 i_size
= inode
->i_size
;
783 return inode
->i_size
;
788 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
789 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
790 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
792 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
794 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
796 write_seqcount_begin(&inode
->i_size_seqcount
);
797 inode
->i_size
= i_size
;
798 write_seqcount_end(&inode
->i_size_seqcount
);
800 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
802 inode
->i_size
= i_size
;
805 inode
->i_size
= i_size
;
809 static inline unsigned iminor(const struct inode
*inode
)
811 return MINOR(inode
->i_rdev
);
814 static inline unsigned imajor(const struct inode
*inode
)
816 return MAJOR(inode
->i_rdev
);
819 extern struct block_device
*I_BDEV(struct inode
*inode
);
822 rwlock_t lock
; /* protects pid, uid, euid fields */
823 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
824 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
825 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
826 int signum
; /* posix.1b rt signal to be delivered on IO */
830 * Track a single file's readahead state
832 struct file_ra_state
{
833 pgoff_t start
; /* where readahead started */
834 unsigned int size
; /* # of readahead pages */
835 unsigned int async_size
; /* do asynchronous readahead when
836 there are only # of pages ahead */
838 unsigned int ra_pages
; /* Maximum readahead window */
839 unsigned int mmap_miss
; /* Cache miss stat for mmap accesses */
840 loff_t prev_pos
; /* Cache last read() position */
844 * Check if @index falls in the readahead windows.
846 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
848 return (index
>= ra
->start
&&
849 index
< ra
->start
+ ra
->size
);
854 struct llist_node fu_llist
;
855 struct rcu_head fu_rcuhead
;
858 struct inode
*f_inode
; /* cached value */
859 const struct file_operations
*f_op
;
862 * Protects f_ep_links, f_flags.
863 * Must not be taken from IRQ context.
866 enum rw_hint f_write_hint
;
867 atomic_long_t f_count
;
868 unsigned int f_flags
;
870 struct mutex f_pos_lock
;
872 struct fown_struct f_owner
;
873 const struct cred
*f_cred
;
874 struct file_ra_state f_ra
;
877 #ifdef CONFIG_SECURITY
880 /* needed for tty driver, and maybe others */
884 /* Used by fs/eventpoll.c to link all the hooks to this file */
885 struct list_head f_ep_links
;
886 struct list_head f_tfile_llink
;
887 #endif /* #ifdef CONFIG_EPOLL */
888 struct address_space
*f_mapping
;
891 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
896 /* file identifier */
897 unsigned char f_handle
[0];
900 static inline struct file
*get_file(struct file
*f
)
902 atomic_long_inc(&f
->f_count
);
905 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
906 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
907 #define file_count(x) atomic_long_read(&(x)->f_count)
909 #define MAX_NON_LFS ((1UL<<31) - 1)
911 /* Page cache limit. The filesystems should put that into their s_maxbytes
912 limits, otherwise bad things can happen in VM. */
913 #if BITS_PER_LONG==32
914 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
915 #elif BITS_PER_LONG==64
916 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
921 #define FL_DELEG 4 /* NFSv4 delegation */
922 #define FL_ACCESS 8 /* not trying to lock, just looking */
923 #define FL_EXISTS 16 /* when unlocking, test for existence */
924 #define FL_LEASE 32 /* lease held on this file */
925 #define FL_CLOSE 64 /* unlock on close */
926 #define FL_SLEEP 128 /* A blocking lock */
927 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
928 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
929 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
930 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
932 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
935 * Special return value from posix_lock_file() and vfs_lock_file() for
936 * asynchronous locking.
938 #define FILE_LOCK_DEFERRED 1
940 /* legacy typedef, should eventually be removed */
941 typedef void *fl_owner_t
;
945 struct file_lock_operations
{
946 void (*fl_copy_lock
)(struct file_lock
*, struct file_lock
*);
947 void (*fl_release_private
)(struct file_lock
*);
950 struct lock_manager_operations
{
951 int (*lm_compare_owner
)(struct file_lock
*, struct file_lock
*);
952 unsigned long (*lm_owner_key
)(struct file_lock
*);
953 fl_owner_t (*lm_get_owner
)(fl_owner_t
);
954 void (*lm_put_owner
)(fl_owner_t
);
955 void (*lm_notify
)(struct file_lock
*); /* unblock callback */
956 int (*lm_grant
)(struct file_lock
*, int);
957 bool (*lm_break
)(struct file_lock
*);
958 int (*lm_change
)(struct file_lock
*, int, struct list_head
*);
959 void (*lm_setup
)(struct file_lock
*, void **);
962 struct lock_manager
{
963 struct list_head list
;
965 * NFSv4 and up also want opens blocked during the grace period;
972 void locks_start_grace(struct net
*, struct lock_manager
*);
973 void locks_end_grace(struct lock_manager
*);
974 bool locks_in_grace(struct net
*);
975 bool opens_in_grace(struct net
*);
977 /* that will die - we need it for nfs_lock_info */
978 #include <linux/nfs_fs_i.h>
981 * struct file_lock represents a generic "file lock". It's used to represent
982 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
983 * note that the same struct is used to represent both a request for a lock and
984 * the lock itself, but the same object is never used for both.
986 * FIXME: should we create a separate "struct lock_request" to help distinguish
989 * The varous i_flctx lists are ordered by:
992 * 2) lock range start
995 * Obviously, the last two criteria only matter for POSIX locks.
998 struct file_lock
*fl_next
; /* singly linked list for this inode */
999 struct list_head fl_list
; /* link into file_lock_context */
1000 struct hlist_node fl_link
; /* node in global lists */
1001 struct list_head fl_block
; /* circular list of blocked processes */
1002 fl_owner_t fl_owner
;
1003 unsigned int fl_flags
;
1004 unsigned char fl_type
;
1005 unsigned int fl_pid
;
1006 int fl_link_cpu
; /* what cpu's list is this on? */
1007 wait_queue_head_t fl_wait
;
1008 struct file
*fl_file
;
1012 struct fasync_struct
* fl_fasync
; /* for lease break notifications */
1013 /* for lease breaks: */
1014 unsigned long fl_break_time
;
1015 unsigned long fl_downgrade_time
;
1017 const struct file_lock_operations
*fl_ops
; /* Callbacks for filesystems */
1018 const struct lock_manager_operations
*fl_lmops
; /* Callbacks for lockmanagers */
1020 struct nfs_lock_info nfs_fl
;
1021 struct nfs4_lock_info nfs4_fl
;
1023 struct list_head link
; /* link in AFS vnode's pending_locks list */
1024 int state
; /* state of grant or error if -ve */
1027 } __randomize_layout
;
1029 struct file_lock_context
{
1030 spinlock_t flc_lock
;
1031 struct list_head flc_flock
;
1032 struct list_head flc_posix
;
1033 struct list_head flc_lease
;
1036 /* The following constant reflects the upper bound of the file/locking space */
1038 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1039 #define OFFSET_MAX INT_LIMIT(loff_t)
1040 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1043 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
1046 * Return the inode to use for locking
1048 * For overlayfs this should be the overlay inode, not the real inode returned
1049 * by file_inode(). For any other fs file_inode(filp) and locks_inode(filp) are
1052 static inline struct inode
*locks_inode(const struct file
*f
)
1054 return f
->f_path
.dentry
->d_inode
;
1057 #ifdef CONFIG_FILE_LOCKING
1058 extern int fcntl_getlk(struct file
*, unsigned int, struct flock
*);
1059 extern int fcntl_setlk(unsigned int, struct file
*, unsigned int,
1062 #if BITS_PER_LONG == 32
1063 extern int fcntl_getlk64(struct file
*, unsigned int, struct flock64
*);
1064 extern int fcntl_setlk64(unsigned int, struct file
*, unsigned int,
1068 extern int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
);
1069 extern int fcntl_getlease(struct file
*filp
);
1072 void locks_free_lock_context(struct inode
*inode
);
1073 void locks_free_lock(struct file_lock
*fl
);
1074 extern void locks_init_lock(struct file_lock
*);
1075 extern struct file_lock
* locks_alloc_lock(void);
1076 extern void locks_copy_lock(struct file_lock
*, struct file_lock
*);
1077 extern void locks_copy_conflock(struct file_lock
*, struct file_lock
*);
1078 extern void locks_remove_posix(struct file
*, fl_owner_t
);
1079 extern void locks_remove_file(struct file
*);
1080 extern void locks_release_private(struct file_lock
*);
1081 extern void posix_test_lock(struct file
*, struct file_lock
*);
1082 extern int posix_lock_file(struct file
*, struct file_lock
*, struct file_lock
*);
1083 extern int posix_unblock_lock(struct file_lock
*);
1084 extern int vfs_test_lock(struct file
*, struct file_lock
*);
1085 extern int vfs_lock_file(struct file
*, unsigned int, struct file_lock
*, struct file_lock
*);
1086 extern int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
);
1087 extern int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
);
1088 extern int __break_lease(struct inode
*inode
, unsigned int flags
, unsigned int type
);
1089 extern void lease_get_mtime(struct inode
*, struct timespec
*time
);
1090 extern int generic_setlease(struct file
*, long, struct file_lock
**, void **priv
);
1091 extern int vfs_setlease(struct file
*, long, struct file_lock
**, void **);
1092 extern int lease_modify(struct file_lock
*, int, struct list_head
*);
1093 struct files_struct
;
1094 extern void show_fd_locks(struct seq_file
*f
,
1095 struct file
*filp
, struct files_struct
*files
);
1096 #else /* !CONFIG_FILE_LOCKING */
1097 static inline int fcntl_getlk(struct file
*file
, unsigned int cmd
,
1098 struct flock __user
*user
)
1103 static inline int fcntl_setlk(unsigned int fd
, struct file
*file
,
1104 unsigned int cmd
, struct flock __user
*user
)
1109 #if BITS_PER_LONG == 32
1110 static inline int fcntl_getlk64(struct file
*file
, unsigned int cmd
,
1111 struct flock64 __user
*user
)
1116 static inline int fcntl_setlk64(unsigned int fd
, struct file
*file
,
1117 unsigned int cmd
, struct flock64 __user
*user
)
1122 static inline int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
)
1127 static inline int fcntl_getlease(struct file
*filp
)
1133 locks_free_lock_context(struct inode
*inode
)
1137 static inline void locks_init_lock(struct file_lock
*fl
)
1142 static inline void locks_copy_conflock(struct file_lock
*new, struct file_lock
*fl
)
1147 static inline void locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
1152 static inline void locks_remove_posix(struct file
*filp
, fl_owner_t owner
)
1157 static inline void locks_remove_file(struct file
*filp
)
1162 static inline void posix_test_lock(struct file
*filp
, struct file_lock
*fl
)
1167 static inline int posix_lock_file(struct file
*filp
, struct file_lock
*fl
,
1168 struct file_lock
*conflock
)
1173 static inline int posix_unblock_lock(struct file_lock
*waiter
)
1178 static inline int vfs_test_lock(struct file
*filp
, struct file_lock
*fl
)
1183 static inline int vfs_lock_file(struct file
*filp
, unsigned int cmd
,
1184 struct file_lock
*fl
, struct file_lock
*conf
)
1189 static inline int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
)
1194 static inline int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
)
1199 static inline int __break_lease(struct inode
*inode
, unsigned int mode
, unsigned int type
)
1204 static inline void lease_get_mtime(struct inode
*inode
, struct timespec
*time
)
1209 static inline int generic_setlease(struct file
*filp
, long arg
,
1210 struct file_lock
**flp
, void **priv
)
1215 static inline int vfs_setlease(struct file
*filp
, long arg
,
1216 struct file_lock
**lease
, void **priv
)
1221 static inline int lease_modify(struct file_lock
*fl
, int arg
,
1222 struct list_head
*dispose
)
1227 struct files_struct
;
1228 static inline void show_fd_locks(struct seq_file
*f
,
1229 struct file
*filp
, struct files_struct
*files
) {}
1230 #endif /* !CONFIG_FILE_LOCKING */
1232 static inline struct inode
*file_inode(const struct file
*f
)
1237 static inline struct dentry
*file_dentry(const struct file
*file
)
1239 return d_real(file
->f_path
.dentry
, file_inode(file
), 0, 0);
1242 static inline int locks_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1244 return locks_lock_inode_wait(locks_inode(filp
), fl
);
1247 struct fasync_struct
{
1251 struct fasync_struct
*fa_next
; /* singly linked list */
1252 struct file
*fa_file
;
1253 struct rcu_head fa_rcu
;
1256 #define FASYNC_MAGIC 0x4601
1258 /* SMP safe fasync helpers: */
1259 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1260 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1261 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1262 extern struct fasync_struct
*fasync_alloc(void);
1263 extern void fasync_free(struct fasync_struct
*);
1265 /* can be called from interrupts */
1266 extern void kill_fasync(struct fasync_struct
**, int, int);
1268 extern int setfl(int fd
, struct file
* filp
, unsigned long arg
);
1269 extern void __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1270 extern int f_setown(struct file
*filp
, unsigned long arg
, int force
);
1271 extern void f_delown(struct file
*filp
);
1272 extern pid_t
f_getown(struct file
*filp
);
1273 extern int send_sigurg(struct fown_struct
*fown
);
1276 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1277 * represented in both.
1279 #define SB_RDONLY 1 /* Mount read-only */
1280 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1281 #define SB_NODEV 4 /* Disallow access to device special files */
1282 #define SB_NOEXEC 8 /* Disallow program execution */
1283 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1284 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1285 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1286 #define SB_NOATIME 1024 /* Do not update access times. */
1287 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1288 #define SB_SILENT 32768
1289 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1290 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1291 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1292 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1294 /* These sb flags are internal to the kernel */
1295 #define SB_SUBMOUNT (1<<26)
1296 #define SB_NOREMOTELOCK (1<<27)
1297 #define SB_NOSEC (1<<28)
1298 #define SB_BORN (1<<29)
1299 #define SB_ACTIVE (1<<30)
1300 #define SB_NOUSER (1<<31)
1306 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1307 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1308 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1309 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1310 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1313 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1314 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1315 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1316 #define SB_I_NOSUID 0x00000008 /* Ignore suid on this fs */
1318 /* sb->s_iflags to limit user namespace mounts */
1319 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1321 /* Possible states of 'frozen' field */
1323 SB_UNFROZEN
= 0, /* FS is unfrozen */
1324 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1325 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1326 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1327 * internal threads if needed) */
1328 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1331 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1334 int frozen
; /* Is sb frozen? */
1335 wait_queue_head_t wait_unfrozen
; /* for get_super_thawed() */
1336 struct percpu_rw_semaphore rw_sem
[SB_FREEZE_LEVELS
];
1339 struct super_block
{
1340 struct list_head s_list
; /* Keep this first */
1341 dev_t s_dev
; /* search index; _not_ kdev_t */
1342 unsigned char s_blocksize_bits
;
1343 unsigned long s_blocksize
;
1344 loff_t s_maxbytes
; /* Max file size */
1345 struct file_system_type
*s_type
;
1346 const struct super_operations
*s_op
;
1347 const struct dquot_operations
*dq_op
;
1348 const struct quotactl_ops
*s_qcop
;
1349 const struct export_operations
*s_export_op
;
1350 unsigned long s_flags
;
1351 unsigned long s_iflags
; /* internal SB_I_* flags */
1352 unsigned long s_magic
;
1353 struct dentry
*s_root
;
1354 struct rw_semaphore s_umount
;
1357 #ifdef CONFIG_SECURITY
1360 const struct xattr_handler
**s_xattr
;
1362 const struct fscrypt_operations
*s_cop
;
1364 struct hlist_bl_head s_anon
; /* anonymous dentries for (nfs) exporting */
1365 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1366 struct block_device
*s_bdev
;
1367 struct backing_dev_info
*s_bdi
;
1368 struct mtd_info
*s_mtd
;
1369 struct hlist_node s_instances
;
1370 unsigned int s_quota_types
; /* Bitmask of supported quota types */
1371 struct quota_info s_dquot
; /* Diskquota specific options */
1373 struct sb_writers s_writers
;
1375 char s_id
[32]; /* Informational name */
1376 uuid_t s_uuid
; /* UUID */
1378 void *s_fs_info
; /* Filesystem private info */
1379 unsigned int s_max_links
;
1382 /* Granularity of c/m/atime in ns.
1383 Cannot be worse than a second */
1387 * The next field is for VFS *only*. No filesystems have any business
1388 * even looking at it. You had been warned.
1390 struct mutex s_vfs_rename_mutex
; /* Kludge */
1393 * Filesystem subtype. If non-empty the filesystem type field
1394 * in /proc/mounts will be "type.subtype"
1398 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1401 * Saved pool identifier for cleancache (-1 means none)
1403 int cleancache_poolid
;
1405 struct shrinker s_shrink
; /* per-sb shrinker handle */
1407 /* Number of inodes with nlink == 0 but still referenced */
1408 atomic_long_t s_remove_count
;
1410 /* Being remounted read-only */
1411 int s_readonly_remount
;
1413 /* AIO completions deferred from interrupt context */
1414 struct workqueue_struct
*s_dio_done_wq
;
1415 struct hlist_head s_pins
;
1418 * Owning user namespace and default context in which to
1419 * interpret filesystem uids, gids, quotas, device nodes,
1420 * xattrs and security labels.
1422 struct user_namespace
*s_user_ns
;
1425 * Keep the lru lists last in the structure so they always sit on their
1426 * own individual cachelines.
1428 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp
;
1429 struct list_lru s_inode_lru ____cacheline_aligned_in_smp
;
1430 struct rcu_head rcu
;
1431 struct work_struct destroy_work
;
1433 struct mutex s_sync_lock
; /* sync serialisation lock */
1436 * Indicates how deep in a filesystem stack this SB is
1440 /* s_inode_list_lock protects s_inodes */
1441 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp
;
1442 struct list_head s_inodes
; /* all inodes */
1444 spinlock_t s_inode_wblist_lock
;
1445 struct list_head s_inodes_wb
; /* writeback inodes */
1446 } __randomize_layout
;
1448 /* Helper functions so that in most cases filesystems will
1449 * not need to deal directly with kuid_t and kgid_t and can
1450 * instead deal with the raw numeric values that are stored
1451 * in the filesystem.
1453 static inline uid_t
i_uid_read(const struct inode
*inode
)
1455 return from_kuid(inode
->i_sb
->s_user_ns
, inode
->i_uid
);
1458 static inline gid_t
i_gid_read(const struct inode
*inode
)
1460 return from_kgid(inode
->i_sb
->s_user_ns
, inode
->i_gid
);
1463 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
1465 inode
->i_uid
= make_kuid(inode
->i_sb
->s_user_ns
, uid
);
1468 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
1470 inode
->i_gid
= make_kgid(inode
->i_sb
->s_user_ns
, gid
);
1473 extern struct timespec
current_time(struct inode
*inode
);
1476 * Snapshotting support.
1479 void __sb_end_write(struct super_block
*sb
, int level
);
1480 int __sb_start_write(struct super_block
*sb
, int level
, bool wait
);
1482 #define __sb_writers_acquired(sb, lev) \
1483 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1484 #define __sb_writers_release(sb, lev) \
1485 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1488 * sb_end_write - drop write access to a superblock
1489 * @sb: the super we wrote to
1491 * Decrement number of writers to the filesystem. Wake up possible waiters
1492 * wanting to freeze the filesystem.
1494 static inline void sb_end_write(struct super_block
*sb
)
1496 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1500 * sb_end_pagefault - drop write access to a superblock from a page fault
1501 * @sb: the super we wrote to
1503 * Decrement number of processes handling write page fault to the filesystem.
1504 * Wake up possible waiters wanting to freeze the filesystem.
1506 static inline void sb_end_pagefault(struct super_block
*sb
)
1508 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1512 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1513 * @sb: the super we wrote to
1515 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1516 * waiters wanting to freeze the filesystem.
1518 static inline void sb_end_intwrite(struct super_block
*sb
)
1520 __sb_end_write(sb
, SB_FREEZE_FS
);
1524 * sb_start_write - get write access to a superblock
1525 * @sb: the super we write to
1527 * When a process wants to write data or metadata to a file system (i.e. dirty
1528 * a page or an inode), it should embed the operation in a sb_start_write() -
1529 * sb_end_write() pair to get exclusion against file system freezing. This
1530 * function increments number of writers preventing freezing. If the file
1531 * system is already frozen, the function waits until the file system is
1534 * Since freeze protection behaves as a lock, users have to preserve
1535 * ordering of freeze protection and other filesystem locks. Generally,
1536 * freeze protection should be the outermost lock. In particular, we have:
1539 * -> i_mutex (write path, truncate, directory ops, ...)
1540 * -> s_umount (freeze_super, thaw_super)
1542 static inline void sb_start_write(struct super_block
*sb
)
1544 __sb_start_write(sb
, SB_FREEZE_WRITE
, true);
1547 static inline int sb_start_write_trylock(struct super_block
*sb
)
1549 return __sb_start_write(sb
, SB_FREEZE_WRITE
, false);
1553 * sb_start_pagefault - get write access to a superblock from a page fault
1554 * @sb: the super we write to
1556 * When a process starts handling write page fault, it should embed the
1557 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1558 * exclusion against file system freezing. This is needed since the page fault
1559 * is going to dirty a page. This function increments number of running page
1560 * faults preventing freezing. If the file system is already frozen, the
1561 * function waits until the file system is thawed.
1563 * Since page fault freeze protection behaves as a lock, users have to preserve
1564 * ordering of freeze protection and other filesystem locks. It is advised to
1565 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1566 * handling code implies lock dependency:
1569 * -> sb_start_pagefault
1571 static inline void sb_start_pagefault(struct super_block
*sb
)
1573 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
, true);
1577 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1578 * @sb: the super we write to
1580 * This is the third level of protection against filesystem freezing. It is
1581 * free for use by a filesystem. The only requirement is that it must rank
1582 * below sb_start_pagefault.
1584 * For example filesystem can call sb_start_intwrite() when starting a
1585 * transaction which somewhat eases handling of freezing for internal sources
1586 * of filesystem changes (internal fs threads, discarding preallocation on file
1589 static inline void sb_start_intwrite(struct super_block
*sb
)
1591 __sb_start_write(sb
, SB_FREEZE_FS
, true);
1595 extern bool inode_owner_or_capable(const struct inode
*inode
);
1598 * VFS helper functions..
1600 extern int vfs_create(struct inode
*, struct dentry
*, umode_t
, bool);
1601 extern int vfs_mkdir(struct inode
*, struct dentry
*, umode_t
);
1602 extern int vfs_mknod(struct inode
*, struct dentry
*, umode_t
, dev_t
);
1603 extern int vfs_symlink(struct inode
*, struct dentry
*, const char *);
1604 extern int vfs_link(struct dentry
*, struct inode
*, struct dentry
*, struct inode
**);
1605 extern int vfs_rmdir(struct inode
*, struct dentry
*);
1606 extern int vfs_unlink(struct inode
*, struct dentry
*, struct inode
**);
1607 extern int vfs_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*, struct inode
**, unsigned int);
1608 extern int vfs_whiteout(struct inode
*, struct dentry
*);
1610 extern struct dentry
*vfs_tmpfile(struct dentry
*dentry
, umode_t mode
,
1614 * VFS file helper functions.
1616 extern void inode_init_owner(struct inode
*inode
, const struct inode
*dir
,
1618 extern bool may_open_dev(const struct path
*path
);
1620 * VFS FS_IOC_FIEMAP helper definitions.
1622 struct fiemap_extent_info
{
1623 unsigned int fi_flags
; /* Flags as passed from user */
1624 unsigned int fi_extents_mapped
; /* Number of mapped extents */
1625 unsigned int fi_extents_max
; /* Size of fiemap_extent array */
1626 struct fiemap_extent __user
*fi_extents_start
; /* Start of
1627 fiemap_extent array */
1629 int fiemap_fill_next_extent(struct fiemap_extent_info
*info
, u64 logical
,
1630 u64 phys
, u64 len
, u32 flags
);
1631 int fiemap_check_flags(struct fiemap_extent_info
*fieinfo
, u32 fs_flags
);
1636 * NOTE! These match bits 12..15 of stat.st_mode
1637 * (ie "(i_mode >> 12) & 15").
1639 #define DT_UNKNOWN 0
1650 * This is the "filldir" function type, used by readdir() to let
1651 * the kernel specify what kind of dirent layout it wants to have.
1652 * This allows the kernel to read directories into kernel space or
1653 * to have different dirent layouts depending on the binary type.
1656 typedef int (*filldir_t
)(struct dir_context
*, const char *, int, loff_t
, u64
,
1659 struct dir_context
{
1660 const filldir_t actor
;
1664 struct block_device_operations
;
1666 /* These macros are for out of kernel modules to test that
1667 * the kernel supports the unlocked_ioctl and compat_ioctl
1668 * fields in struct file_operations. */
1669 #define HAVE_COMPAT_IOCTL 1
1670 #define HAVE_UNLOCKED_IOCTL 1
1673 * These flags let !MMU mmap() govern direct device mapping vs immediate
1674 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1676 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1677 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1678 * NOMMU_MAP_READ: Can be mapped for reading
1679 * NOMMU_MAP_WRITE: Can be mapped for writing
1680 * NOMMU_MAP_EXEC: Can be mapped for execution
1682 #define NOMMU_MAP_COPY 0x00000001
1683 #define NOMMU_MAP_DIRECT 0x00000008
1684 #define NOMMU_MAP_READ VM_MAYREAD
1685 #define NOMMU_MAP_WRITE VM_MAYWRITE
1686 #define NOMMU_MAP_EXEC VM_MAYEXEC
1688 #define NOMMU_VMFLAGS \
1689 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1694 struct file_operations
{
1695 struct module
*owner
;
1696 loff_t (*llseek
) (struct file
*, loff_t
, int);
1697 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
1698 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
1699 ssize_t (*read_iter
) (struct kiocb
*, struct iov_iter
*);
1700 ssize_t (*write_iter
) (struct kiocb
*, struct iov_iter
*);
1701 int (*iterate
) (struct file
*, struct dir_context
*);
1702 int (*iterate_shared
) (struct file
*, struct dir_context
*);
1703 unsigned int (*poll
) (struct file
*, struct poll_table_struct
*);
1704 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
1705 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
1706 int (*mmap
) (struct file
*, struct vm_area_struct
*);
1707 unsigned long mmap_supported_flags
;
1708 int (*open
) (struct inode
*, struct file
*);
1709 int (*flush
) (struct file
*, fl_owner_t id
);
1710 int (*release
) (struct inode
*, struct file
*);
1711 int (*fsync
) (struct file
*, loff_t
, loff_t
, int datasync
);
1712 int (*fasync
) (int, struct file
*, int);
1713 int (*lock
) (struct file
*, int, struct file_lock
*);
1714 ssize_t (*sendpage
) (struct file
*, struct page
*, int, size_t, loff_t
*, int);
1715 unsigned long (*get_unmapped_area
)(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1716 int (*check_flags
)(int);
1717 int (*setfl
)(struct file
*, unsigned long);
1718 int (*flock
) (struct file
*, int, struct file_lock
*);
1719 ssize_t (*splice_write
)(struct pipe_inode_info
*, struct file
*, loff_t
*, size_t, unsigned int);
1720 ssize_t (*splice_read
)(struct file
*, loff_t
*, struct pipe_inode_info
*, size_t, unsigned int);
1721 int (*setlease
)(struct file
*, long, struct file_lock
**, void **);
1722 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
1724 void (*show_fdinfo
)(struct seq_file
*m
, struct file
*f
);
1726 unsigned (*mmap_capabilities
)(struct file
*);
1728 ssize_t (*copy_file_range
)(struct file
*, loff_t
, struct file
*,
1729 loff_t
, size_t, unsigned int);
1730 int (*clone_file_range
)(struct file
*, loff_t
, struct file
*, loff_t
,
1732 ssize_t (*dedupe_file_range
)(struct file
*, u64
, u64
, struct file
*,
1734 } __randomize_layout
;
1736 struct inode_operations
{
1737 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*, unsigned int);
1738 const char * (*get_link
) (struct dentry
*, struct inode
*, struct delayed_call
*);
1739 int (*permission
) (struct inode
*, int);
1740 struct posix_acl
* (*get_acl
)(struct inode
*, int);
1742 int (*readlink
) (struct dentry
*, char __user
*,int);
1744 int (*create
) (struct inode
*,struct dentry
*, umode_t
, bool);
1745 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
1746 int (*unlink
) (struct inode
*,struct dentry
*);
1747 int (*symlink
) (struct inode
*,struct dentry
*,const char *);
1748 int (*mkdir
) (struct inode
*,struct dentry
*,umode_t
);
1749 int (*rmdir
) (struct inode
*,struct dentry
*);
1750 int (*mknod
) (struct inode
*,struct dentry
*,umode_t
,dev_t
);
1751 int (*rename
) (struct inode
*, struct dentry
*,
1752 struct inode
*, struct dentry
*, unsigned int);
1753 int (*setattr
) (struct dentry
*, struct iattr
*);
1754 int (*getattr
) (const struct path
*, struct kstat
*, u32
, unsigned int);
1755 ssize_t (*listxattr
) (struct dentry
*, char *, size_t);
1756 int (*fiemap
)(struct inode
*, struct fiemap_extent_info
*, u64 start
,
1758 int (*update_time
)(struct inode
*, struct timespec
*, int);
1759 int (*atomic_open
)(struct inode
*, struct dentry
*,
1760 struct file
*, unsigned open_flag
,
1761 umode_t create_mode
, int *opened
);
1762 int (*tmpfile
) (struct inode
*, struct dentry
*, umode_t
);
1763 int (*set_acl
)(struct inode
*, struct posix_acl
*, int);
1764 } ____cacheline_aligned
;
1766 static inline ssize_t
call_read_iter(struct file
*file
, struct kiocb
*kio
,
1767 struct iov_iter
*iter
)
1769 return file
->f_op
->read_iter(kio
, iter
);
1772 static inline ssize_t
call_write_iter(struct file
*file
, struct kiocb
*kio
,
1773 struct iov_iter
*iter
)
1775 return file
->f_op
->write_iter(kio
, iter
);
1778 static inline int call_mmap(struct file
*file
, struct vm_area_struct
*vma
)
1780 return file
->f_op
->mmap(file
, vma
);
1783 ssize_t
rw_copy_check_uvector(int type
, const struct iovec __user
* uvector
,
1784 unsigned long nr_segs
, unsigned long fast_segs
,
1785 struct iovec
*fast_pointer
,
1786 struct iovec
**ret_pointer
);
1788 typedef ssize_t (*vfs_readf_t
)(struct file
*, char __user
*, size_t, loff_t
*);
1789 typedef ssize_t (*vfs_writef_t
)(struct file
*, const char __user
*, size_t,
1791 vfs_readf_t
vfs_readf(struct file
*file
);
1792 vfs_writef_t
vfs_writef(struct file
*file
);
1794 extern ssize_t
__vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1795 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1796 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1797 extern ssize_t
vfs_readv(struct file
*, const struct iovec __user
*,
1798 unsigned long, loff_t
*, rwf_t
);
1799 extern ssize_t
vfs_copy_file_range(struct file
*, loff_t
, struct file
*,
1800 loff_t
, size_t, unsigned int);
1801 extern int vfs_clone_file_prep_inodes(struct inode
*inode_in
, loff_t pos_in
,
1802 struct inode
*inode_out
, loff_t pos_out
,
1803 u64
*len
, bool is_dedupe
);
1804 extern int vfs_clone_file_range(struct file
*file_in
, loff_t pos_in
,
1805 struct file
*file_out
, loff_t pos_out
, u64 len
);
1806 extern int vfs_dedupe_file_range_compare(struct inode
*src
, loff_t srcoff
,
1807 struct inode
*dest
, loff_t destoff
,
1808 loff_t len
, bool *is_same
);
1809 extern int vfs_dedupe_file_range(struct file
*file
,
1810 struct file_dedupe_range
*same
);
1812 struct super_operations
{
1813 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
1814 void (*destroy_inode
)(struct inode
*);
1816 void (*dirty_inode
) (struct inode
*, int flags
);
1817 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
1818 int (*drop_inode
) (struct inode
*);
1819 void (*evict_inode
) (struct inode
*);
1820 void (*put_super
) (struct super_block
*);
1821 int (*sync_fs
)(struct super_block
*sb
, int wait
);
1822 int (*freeze_super
) (struct super_block
*);
1823 int (*freeze_fs
) (struct super_block
*);
1824 int (*thaw_super
) (struct super_block
*);
1825 int (*unfreeze_fs
) (struct super_block
*);
1826 int (*statfs
) (struct dentry
*, struct kstatfs
*);
1827 int (*remount_fs
) (struct super_block
*, int *, char *);
1828 void (*umount_begin
) (struct super_block
*);
1830 int (*show_options
)(struct seq_file
*, struct dentry
*);
1831 int (*show_devname
)(struct seq_file
*, struct dentry
*);
1832 int (*show_path
)(struct seq_file
*, struct dentry
*);
1833 int (*show_stats
)(struct seq_file
*, struct dentry
*);
1835 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
1836 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
1837 struct dquot
**(*get_dquots
)(struct inode
*);
1839 int (*bdev_try_to_free_page
)(struct super_block
*, struct page
*, gfp_t
);
1840 long (*nr_cached_objects
)(struct super_block
*,
1841 struct shrink_control
*);
1842 long (*free_cached_objects
)(struct super_block
*,
1843 struct shrink_control
*);
1844 #if defined(CONFIG_BLK_DEV_LOOP) || defined(CONFIG_BLK_DEV_LOOP_MODULE)
1846 struct file
*(*real_loop
)(struct file
*);
1851 * Inode flags - they have no relation to superblock flags now
1853 #define S_SYNC 1 /* Writes are synced at once */
1854 #define S_NOATIME 2 /* Do not update access times */
1855 #define S_APPEND 4 /* Append-only file */
1856 #define S_IMMUTABLE 8 /* Immutable file */
1857 #define S_DEAD 16 /* removed, but still open directory */
1858 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1859 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1860 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1861 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1862 #define S_PRIVATE 512 /* Inode is fs-internal */
1863 #define S_IMA 1024 /* Inode has an associated IMA struct */
1864 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1865 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1866 #ifdef CONFIG_FS_DAX
1867 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1869 #define S_DAX 0 /* Make all the DAX code disappear */
1871 #define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */
1874 * Note that nosuid etc flags are inode-specific: setting some file-system
1875 * flags just means all the inodes inherit those flags by default. It might be
1876 * possible to override it selectively if you really wanted to with some
1877 * ioctl() that is not currently implemented.
1879 * Exception: SB_RDONLY is always applied to the entire file system.
1881 * Unfortunately, it is possible to change a filesystems flags with it mounted
1882 * with files in use. This means that all of the inodes will not have their
1883 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1884 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1886 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1888 static inline bool sb_rdonly(const struct super_block
*sb
) { return sb
->s_flags
& SB_RDONLY
; }
1889 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
1890 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
1891 ((inode)->i_flags & S_SYNC))
1892 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
1893 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1894 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
1895 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
1896 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
1898 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1899 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1900 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1901 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
1903 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1904 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1905 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1906 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1907 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1908 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1909 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1910 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1911 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
1913 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1914 (inode)->i_rdev == WHITEOUT_DEV)
1916 static inline bool HAS_UNMAPPED_ID(struct inode
*inode
)
1918 return !uid_valid(inode
->i_uid
) || !gid_valid(inode
->i_gid
);
1921 static inline enum rw_hint
file_write_hint(struct file
*file
)
1923 if (file
->f_write_hint
!= WRITE_LIFE_NOT_SET
)
1924 return file
->f_write_hint
;
1926 return file_inode(file
)->i_write_hint
;
1929 static inline int iocb_flags(struct file
*file
);
1931 static inline void init_sync_kiocb(struct kiocb
*kiocb
, struct file
*filp
)
1933 *kiocb
= (struct kiocb
) {
1935 .ki_flags
= iocb_flags(filp
),
1936 .ki_hint
= file_write_hint(filp
),
1941 * Inode state bits. Protected by inode->i_lock
1943 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1944 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1946 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1947 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1948 * various stages of removing an inode.
1950 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1952 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1953 * fdatasync(). i_atime is the usual cause.
1954 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1955 * these changes separately from I_DIRTY_SYNC so that we
1956 * don't have to write inode on fdatasync() when only
1957 * mtime has changed in it.
1958 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1959 * I_NEW Serves as both a mutex and completion notification.
1960 * New inodes set I_NEW. If two processes both create
1961 * the same inode, one of them will release its inode and
1962 * wait for I_NEW to be released before returning.
1963 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1964 * also cause waiting on I_NEW, without I_NEW actually
1965 * being set. find_inode() uses this to prevent returning
1966 * nearly-dead inodes.
1967 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1968 * is zero. I_FREEING must be set when I_WILL_FREE is
1970 * I_FREEING Set when inode is about to be freed but still has dirty
1971 * pages or buffers attached or the inode itself is still
1973 * I_CLEAR Added by clear_inode(). In this state the inode is
1974 * clean and can be destroyed. Inode keeps I_FREEING.
1976 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1977 * prohibited for many purposes. iget() must wait for
1978 * the inode to be completely released, then create it
1979 * anew. Other functions will just ignore such inodes,
1980 * if appropriate. I_NEW is used for waiting.
1982 * I_SYNC Writeback of inode is running. The bit is set during
1983 * data writeback, and cleared with a wakeup on the bit
1984 * address once it is done. The bit is also used to pin
1985 * the inode in memory for flusher thread.
1987 * I_REFERENCED Marks the inode as recently references on the LRU list.
1989 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1991 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1992 * synchronize competing switching instances and to tell
1993 * wb stat updates to grab mapping->tree_lock. See
1994 * inode_switch_wb_work_fn() for details.
1996 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
1997 * and work dirs among overlayfs mounts.
1999 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2001 #define I_DIRTY_SYNC (1 << 0)
2002 #define I_DIRTY_DATASYNC (1 << 1)
2003 #define I_DIRTY_PAGES (1 << 2)
2005 #define I_NEW (1 << __I_NEW)
2006 #define I_WILL_FREE (1 << 4)
2007 #define I_FREEING (1 << 5)
2008 #define I_CLEAR (1 << 6)
2010 #define I_SYNC (1 << __I_SYNC)
2011 #define I_REFERENCED (1 << 8)
2012 #define __I_DIO_WAKEUP 9
2013 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2014 #define I_LINKABLE (1 << 10)
2015 #define I_DIRTY_TIME (1 << 11)
2016 #define __I_DIRTY_TIME_EXPIRED 12
2017 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
2018 #define I_WB_SWITCH (1 << 13)
2019 #define I_OVL_INUSE (1 << 14)
2021 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
2022 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2024 extern void __mark_inode_dirty(struct inode
*, int);
2025 static inline void mark_inode_dirty(struct inode
*inode
)
2027 __mark_inode_dirty(inode
, I_DIRTY
);
2030 static inline void mark_inode_dirty_sync(struct inode
*inode
)
2032 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
2035 extern void inc_nlink(struct inode
*inode
);
2036 extern void drop_nlink(struct inode
*inode
);
2037 extern void clear_nlink(struct inode
*inode
);
2038 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
2040 static inline void inode_inc_link_count(struct inode
*inode
)
2043 mark_inode_dirty(inode
);
2046 static inline void inode_dec_link_count(struct inode
*inode
)
2049 mark_inode_dirty(inode
);
2053 * inode_inc_iversion - increments i_version
2054 * @inode: inode that need to be updated
2056 * Every time the inode is modified, the i_version field will be incremented.
2057 * The filesystem has to be mounted with i_version flag
2060 static inline void inode_inc_iversion(struct inode
*inode
)
2062 spin_lock(&inode
->i_lock
);
2064 spin_unlock(&inode
->i_lock
);
2067 enum file_time_flags
{
2074 extern void touch_atime(const struct path
*);
2075 static inline void file_accessed(struct file
*file
)
2077 if (!(file
->f_flags
& O_NOATIME
))
2078 touch_atime(&file
->f_path
);
2081 int sync_inode(struct inode
*inode
, struct writeback_control
*wbc
);
2082 int sync_inode_metadata(struct inode
*inode
, int wait
);
2084 struct file_system_type
{
2087 #define FS_REQUIRES_DEV 1
2088 #define FS_BINARY_MOUNTDATA 2
2089 #define FS_HAS_SUBTYPE 4
2090 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2091 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2092 struct dentry
*(*mount
) (struct file_system_type
*, int,
2093 const char *, void *);
2094 void (*kill_sb
) (struct super_block
*);
2095 struct module
*owner
;
2096 struct file_system_type
* next
;
2097 struct hlist_head fs_supers
;
2099 struct lock_class_key s_lock_key
;
2100 struct lock_class_key s_umount_key
;
2101 struct lock_class_key s_vfs_rename_key
;
2102 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
2104 struct lock_class_key i_lock_key
;
2105 struct lock_class_key i_mutex_key
;
2106 struct lock_class_key i_mutex_dir_key
;
2109 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2111 extern struct dentry
*mount_ns(struct file_system_type
*fs_type
,
2112 int flags
, void *data
, void *ns
, struct user_namespace
*user_ns
,
2113 int (*fill_super
)(struct super_block
*, void *, int));
2115 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
2116 int flags
, const char *dev_name
, void *data
,
2117 int (*fill_super
)(struct super_block
*, void *, int));
2119 static inline 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 return ERR_PTR(-ENODEV
);
2126 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
2127 int flags
, void *data
,
2128 int (*fill_super
)(struct super_block
*, void *, int));
2129 extern struct dentry
*mount_nodev(struct file_system_type
*fs_type
,
2130 int flags
, void *data
,
2131 int (*fill_super
)(struct super_block
*, void *, int));
2132 extern struct dentry
*mount_subtree(struct vfsmount
*mnt
, const char *path
);
2133 void generic_shutdown_super(struct super_block
*sb
);
2135 void kill_block_super(struct super_block
*sb
);
2137 static inline void kill_block_super(struct super_block
*sb
)
2142 void kill_anon_super(struct super_block
*sb
);
2143 void kill_litter_super(struct super_block
*sb
);
2144 void deactivate_super(struct super_block
*sb
);
2145 void deactivate_locked_super(struct super_block
*sb
);
2146 int set_anon_super(struct super_block
*s
, void *data
);
2147 int get_anon_bdev(dev_t
*);
2148 void free_anon_bdev(dev_t
);
2149 struct super_block
*sget_userns(struct file_system_type
*type
,
2150 int (*test
)(struct super_block
*,void *),
2151 int (*set
)(struct super_block
*,void *),
2152 int flags
, struct user_namespace
*user_ns
,
2154 struct super_block
*sget(struct file_system_type
*type
,
2155 int (*test
)(struct super_block
*,void *),
2156 int (*set
)(struct super_block
*,void *),
2157 int flags
, void *data
);
2158 extern struct dentry
*mount_pseudo_xattr(struct file_system_type
*, char *,
2159 const struct super_operations
*ops
,
2160 const struct xattr_handler
**xattr
,
2161 const struct dentry_operations
*dops
,
2164 static inline struct dentry
*
2165 mount_pseudo(struct file_system_type
*fs_type
, char *name
,
2166 const struct super_operations
*ops
,
2167 const struct dentry_operations
*dops
, unsigned long magic
)
2169 return mount_pseudo_xattr(fs_type
, name
, ops
, NULL
, dops
, magic
);
2172 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2173 #define fops_get(fops) \
2174 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2175 #define fops_put(fops) \
2176 do { if (fops) module_put((fops)->owner); } while(0)
2178 * This one is to be used *ONLY* from ->open() instances.
2179 * fops must be non-NULL, pinned down *and* module dependencies
2180 * should be sufficient to pin the caller down as well.
2182 #define replace_fops(f, fops) \
2184 struct file *__file = (f); \
2185 fops_put(__file->f_op); \
2186 BUG_ON(!(__file->f_op = (fops))); \
2189 extern int register_filesystem(struct file_system_type
*);
2190 extern int unregister_filesystem(struct file_system_type
*);
2191 extern struct vfsmount
*kern_mount_data(struct file_system_type
*, void *data
);
2192 #define kern_mount(type) kern_mount_data(type, NULL)
2193 extern void kern_unmount(struct vfsmount
*mnt
);
2194 extern int may_umount_tree(struct vfsmount
*);
2195 extern int may_umount(struct vfsmount
*);
2196 extern long do_mount(const char *, const char __user
*,
2197 const char *, unsigned long, void *);
2198 extern struct vfsmount
*collect_mounts(const struct path
*);
2199 extern void drop_collected_mounts(struct vfsmount
*);
2200 extern int iterate_mounts(int (*)(struct vfsmount
*, void *), void *,
2202 extern int vfs_statfs(const struct path
*, struct kstatfs
*);
2203 extern int user_statfs(const char __user
*, struct kstatfs
*);
2204 extern int fd_statfs(int, struct kstatfs
*);
2205 extern int freeze_super(struct super_block
*super
);
2206 extern int thaw_super(struct super_block
*super
);
2207 extern bool our_mnt(struct vfsmount
*mnt
);
2208 extern __printf(2, 3)
2209 int super_setup_bdi_name(struct super_block
*sb
, char *fmt
, ...);
2210 extern int super_setup_bdi(struct super_block
*sb
);
2212 extern int current_umask(void);
2214 extern void ihold(struct inode
* inode
);
2215 extern void iput(struct inode
*);
2216 extern int generic_update_time(struct inode
*, struct timespec
*, int);
2217 extern int update_time(struct inode
*, struct timespec
*, int);
2220 extern struct kobject
*fs_kobj
;
2222 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2224 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2225 extern int locks_mandatory_locked(struct file
*);
2226 extern int locks_mandatory_area(struct inode
*, struct file
*, loff_t
, loff_t
, unsigned char);
2229 * Candidates for mandatory locking have the setgid bit set
2230 * but no group execute bit - an otherwise meaningless combination.
2233 static inline int __mandatory_lock(struct inode
*ino
)
2235 return (ino
->i_mode
& (S_ISGID
| S_IXGRP
)) == S_ISGID
;
2239 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2240 * otherwise these will be advisory locks
2243 static inline int mandatory_lock(struct inode
*ino
)
2245 return IS_MANDLOCK(ino
) && __mandatory_lock(ino
);
2248 static inline int locks_verify_locked(struct file
*file
)
2250 if (mandatory_lock(locks_inode(file
)))
2251 return locks_mandatory_locked(file
);
2255 static inline int locks_verify_truncate(struct inode
*inode
,
2259 if (!inode
->i_flctx
|| !mandatory_lock(inode
))
2262 if (size
< inode
->i_size
) {
2263 return locks_mandatory_area(inode
, f
, size
, inode
->i_size
- 1,
2266 return locks_mandatory_area(inode
, f
, inode
->i_size
, size
- 1,
2271 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2273 static inline int locks_mandatory_locked(struct file
*file
)
2278 static inline int locks_mandatory_area(struct inode
*inode
, struct file
*filp
,
2279 loff_t start
, loff_t end
, unsigned char type
)
2284 static inline int __mandatory_lock(struct inode
*inode
)
2289 static inline int mandatory_lock(struct inode
*inode
)
2294 static inline int locks_verify_locked(struct file
*file
)
2299 static inline int locks_verify_truncate(struct inode
*inode
, struct file
*filp
,
2305 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2308 #ifdef CONFIG_FILE_LOCKING
2309 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2312 * Since this check is lockless, we must ensure that any refcounts
2313 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2314 * could end up racing with tasks trying to set a new lease on this
2318 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2319 return __break_lease(inode
, mode
, FL_LEASE
);
2323 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2326 * Since this check is lockless, we must ensure that any refcounts
2327 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2328 * could end up racing with tasks trying to set a new lease on this
2332 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2333 return __break_lease(inode
, mode
, FL_DELEG
);
2337 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2341 ret
= break_deleg(inode
, O_WRONLY
|O_NONBLOCK
);
2342 if (ret
== -EWOULDBLOCK
&& delegated_inode
) {
2343 *delegated_inode
= inode
;
2349 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2353 ret
= break_deleg(*delegated_inode
, O_WRONLY
);
2354 iput(*delegated_inode
);
2355 *delegated_inode
= NULL
;
2359 static inline int break_layout(struct inode
*inode
, bool wait
)
2362 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2363 return __break_lease(inode
,
2364 wait
? O_WRONLY
: O_WRONLY
| O_NONBLOCK
,
2369 #else /* !CONFIG_FILE_LOCKING */
2370 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2375 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2380 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2385 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2391 static inline int break_layout(struct inode
*inode
, bool wait
)
2396 #endif /* CONFIG_FILE_LOCKING */
2401 const char *name
; /* pointer to actual string */
2402 const __user
char *uptr
; /* original userland pointer */
2403 struct audit_names
*aname
;
2408 extern long vfs_truncate(const struct path
*, loff_t
);
2409 extern int do_truncate(struct dentry
*, loff_t start
, unsigned int time_attrs
,
2411 extern int vfs_fallocate(struct file
*file
, int mode
, loff_t offset
,
2413 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
2415 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
2416 extern struct file
*filp_open(const char *, int, umode_t
);
2417 extern struct file
*file_open_root(struct dentry
*, struct vfsmount
*,
2418 const char *, int, umode_t
);
2419 extern struct file
* dentry_open(const struct path
*, int, const struct cred
*);
2420 extern int filp_close(struct file
*, fl_owner_t id
);
2422 extern struct filename
*getname_flags(const char __user
*, int, int *);
2423 extern struct filename
*getname(const char __user
*);
2424 extern struct filename
*getname_kernel(const char *);
2425 extern void putname(struct filename
*name
);
2431 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
2432 int (*open
)(struct inode
*, struct file
*),
2434 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
2438 extern int ioctl_preallocate(struct file
*filp
, void __user
*argp
);
2441 extern void __init
vfs_caches_init_early(void);
2442 extern void __init
vfs_caches_init(void);
2444 extern struct kmem_cache
*names_cachep
;
2446 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2447 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2450 extern int register_blkdev(unsigned int, const char *);
2451 extern void unregister_blkdev(unsigned int, const char *);
2452 extern void bdev_unhash_inode(dev_t dev
);
2453 extern struct block_device
*bdget(dev_t
);
2454 extern struct block_device
*bdgrab(struct block_device
*bdev
);
2455 extern void bd_set_size(struct block_device
*, loff_t size
);
2456 extern void bd_forget(struct inode
*inode
);
2457 extern void bdput(struct block_device
*);
2458 extern void invalidate_bdev(struct block_device
*);
2459 extern void iterate_bdevs(void (*)(struct block_device
*, void *), void *);
2460 extern int sync_blockdev(struct block_device
*bdev
);
2461 extern void kill_bdev(struct block_device
*);
2462 extern struct super_block
*freeze_bdev(struct block_device
*);
2463 extern void emergency_thaw_all(void);
2464 extern int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
);
2465 extern int fsync_bdev(struct block_device
*);
2467 extern struct super_block
*blockdev_superblock
;
2469 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2471 return sb
== blockdev_superblock
;
2474 static inline void bd_forget(struct inode
*inode
) {}
2475 static inline int sync_blockdev(struct block_device
*bdev
) { return 0; }
2476 static inline void kill_bdev(struct block_device
*bdev
) {}
2477 static inline void invalidate_bdev(struct block_device
*bdev
) {}
2479 static inline struct super_block
*freeze_bdev(struct block_device
*sb
)
2484 static inline int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
)
2489 static inline void iterate_bdevs(void (*f
)(struct block_device
*, void *), void *arg
)
2493 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2498 extern int __sync_filesystem(struct super_block
*, int);
2499 extern int sync_filesystem(struct super_block
*);
2500 extern const struct file_operations def_blk_fops
;
2501 extern const struct file_operations def_chr_fops
;
2503 extern int ioctl_by_bdev(struct block_device
*, unsigned, unsigned long);
2504 extern int blkdev_ioctl(struct block_device
*, fmode_t
, unsigned, unsigned long);
2505 extern long compat_blkdev_ioctl(struct file
*, unsigned, unsigned long);
2506 extern int blkdev_get(struct block_device
*bdev
, fmode_t mode
, void *holder
);
2507 extern struct block_device
*blkdev_get_by_path(const char *path
, fmode_t mode
,
2509 extern struct block_device
*blkdev_get_by_dev(dev_t dev
, fmode_t mode
,
2511 extern void blkdev_put(struct block_device
*bdev
, fmode_t mode
);
2512 extern int __blkdev_reread_part(struct block_device
*bdev
);
2513 extern int blkdev_reread_part(struct block_device
*bdev
);
2516 extern int bd_link_disk_holder(struct block_device
*bdev
, struct gendisk
*disk
);
2517 extern void bd_unlink_disk_holder(struct block_device
*bdev
,
2518 struct gendisk
*disk
);
2520 static inline int bd_link_disk_holder(struct block_device
*bdev
,
2521 struct gendisk
*disk
)
2525 static inline void bd_unlink_disk_holder(struct block_device
*bdev
,
2526 struct gendisk
*disk
)
2533 #define CHRDEV_MAJOR_MAX 512
2534 /* Marks the bottom of the first segment of free char majors */
2535 #define CHRDEV_MAJOR_DYN_END 234
2536 /* Marks the top and bottom of the second segment of free char majors */
2537 #define CHRDEV_MAJOR_DYN_EXT_START 511
2538 #define CHRDEV_MAJOR_DYN_EXT_END 384
2540 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
2541 extern int register_chrdev_region(dev_t
, unsigned, const char *);
2542 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
2543 unsigned int count
, const char *name
,
2544 const struct file_operations
*fops
);
2545 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
2546 unsigned int count
, const char *name
);
2547 extern void unregister_chrdev_region(dev_t
, unsigned);
2548 extern void chrdev_show(struct seq_file
*,off_t
);
2550 static inline int register_chrdev(unsigned int major
, const char *name
,
2551 const struct file_operations
*fops
)
2553 return __register_chrdev(major
, 0, 256, name
, fops
);
2556 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2558 __unregister_chrdev(major
, 0, 256, name
);
2561 /* fs/block_dev.c */
2562 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2563 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2566 #define BLKDEV_MAJOR_MAX 512
2567 extern const char *__bdevname(dev_t
, char *buffer
);
2568 extern const char *bdevname(struct block_device
*bdev
, char *buffer
);
2569 extern struct block_device
*lookup_bdev(const char *, int mask
);
2570 extern void blkdev_show(struct seq_file
*,off_t
);
2573 #define BLKDEV_MAJOR_MAX 0
2576 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2578 /* Invalid inode operations -- fs/bad_inode.c */
2579 extern void make_bad_inode(struct inode
*);
2580 extern bool is_bad_inode(struct inode
*);
2583 extern void check_disk_size_change(struct gendisk
*disk
,
2584 struct block_device
*bdev
);
2585 extern int revalidate_disk(struct gendisk
*);
2586 extern int check_disk_change(struct block_device
*);
2587 extern int __invalidate_device(struct block_device
*, bool);
2588 extern int invalidate_partition(struct gendisk
*, int);
2590 unsigned long invalidate_mapping_pages(struct address_space
*mapping
,
2591 pgoff_t start
, pgoff_t end
);
2593 static inline void invalidate_remote_inode(struct inode
*inode
)
2595 if (S_ISREG(inode
->i_mode
) || S_ISDIR(inode
->i_mode
) ||
2596 S_ISLNK(inode
->i_mode
))
2597 invalidate_mapping_pages(inode
->i_mapping
, 0, -1);
2599 extern int invalidate_inode_pages2(struct address_space
*mapping
);
2600 extern int invalidate_inode_pages2_range(struct address_space
*mapping
,
2601 pgoff_t start
, pgoff_t end
);
2602 extern int write_inode_now(struct inode
*, int);
2603 extern int filemap_fdatawrite(struct address_space
*);
2604 extern int filemap_flush(struct address_space
*);
2605 extern int filemap_fdatawait_keep_errors(struct address_space
*mapping
);
2606 extern int filemap_fdatawait_range(struct address_space
*, loff_t lstart
,
2609 static inline int filemap_fdatawait(struct address_space
*mapping
)
2611 return filemap_fdatawait_range(mapping
, 0, LLONG_MAX
);
2614 extern bool filemap_range_has_page(struct address_space
*, loff_t lstart
,
2616 extern int __must_check
file_fdatawait_range(struct file
*file
, loff_t lstart
,
2618 extern int filemap_write_and_wait(struct address_space
*mapping
);
2619 extern int filemap_write_and_wait_range(struct address_space
*mapping
,
2620 loff_t lstart
, loff_t lend
);
2621 extern int __filemap_fdatawrite_range(struct address_space
*mapping
,
2622 loff_t start
, loff_t end
, int sync_mode
);
2623 extern int filemap_fdatawrite_range(struct address_space
*mapping
,
2624 loff_t start
, loff_t end
);
2625 extern int filemap_check_errors(struct address_space
*mapping
);
2626 extern void __filemap_set_wb_err(struct address_space
*mapping
, int err
);
2628 extern int __must_check
file_fdatawait_range(struct file
*file
, loff_t lstart
,
2630 extern int __must_check
file_check_and_advance_wb_err(struct file
*file
);
2631 extern int __must_check
file_write_and_wait_range(struct file
*file
,
2632 loff_t start
, loff_t end
);
2634 static inline int file_write_and_wait(struct file
*file
)
2636 return file_write_and_wait_range(file
, 0, LLONG_MAX
);
2640 * filemap_set_wb_err - set a writeback error on an address_space
2641 * @mapping: mapping in which to set writeback error
2642 * @err: error to be set in mapping
2644 * When writeback fails in some way, we must record that error so that
2645 * userspace can be informed when fsync and the like are called. We endeavor
2646 * to report errors on any file that was open at the time of the error. Some
2647 * internal callers also need to know when writeback errors have occurred.
2649 * When a writeback error occurs, most filesystems will want to call
2650 * filemap_set_wb_err to record the error in the mapping so that it will be
2651 * automatically reported whenever fsync is called on the file.
2653 static inline void filemap_set_wb_err(struct address_space
*mapping
, int err
)
2655 /* Fastpath for common case of no error */
2657 __filemap_set_wb_err(mapping
, err
);
2661 * filemap_check_wb_error - has an error occurred since the mark was sampled?
2662 * @mapping: mapping to check for writeback errors
2663 * @since: previously-sampled errseq_t
2665 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2666 * the given value was sampled.
2668 * If it has then report the latest error set, otherwise return 0.
2670 static inline int filemap_check_wb_err(struct address_space
*mapping
,
2673 return errseq_check(&mapping
->wb_err
, since
);
2677 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2678 * @mapping: mapping to be sampled
2680 * Writeback errors are always reported relative to a particular sample point
2681 * in the past. This function provides those sample points.
2683 static inline errseq_t
filemap_sample_wb_err(struct address_space
*mapping
)
2685 return errseq_sample(&mapping
->wb_err
);
2688 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
2690 extern int vfs_fsync(struct file
*file
, int datasync
);
2693 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2694 * to already be updated for the write, and will return either the amount
2695 * of bytes passed in, or an error if syncing the file failed.
2697 static inline ssize_t
generic_write_sync(struct kiocb
*iocb
, ssize_t count
)
2699 if (iocb
->ki_flags
& IOCB_DSYNC
) {
2700 int ret
= vfs_fsync_range(iocb
->ki_filp
,
2701 iocb
->ki_pos
- count
, iocb
->ki_pos
- 1,
2702 (iocb
->ki_flags
& IOCB_SYNC
) ? 0 : 1);
2710 extern void emergency_sync(void);
2711 extern void emergency_remount(void);
2713 extern sector_t
bmap(struct inode
*, sector_t
);
2715 extern int notify_change(struct dentry
*, struct iattr
*, struct inode
**);
2716 extern int inode_permission(struct inode
*, int);
2717 extern int __inode_permission(struct inode
*, int);
2718 extern int generic_permission(struct inode
*, int);
2719 extern int __check_sticky(struct inode
*dir
, struct inode
*inode
);
2721 static inline bool execute_ok(struct inode
*inode
)
2723 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
2726 static inline void file_start_write(struct file
*file
)
2728 if (!S_ISREG(file_inode(file
)->i_mode
))
2730 __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, true);
2733 static inline bool file_start_write_trylock(struct file
*file
)
2735 if (!S_ISREG(file_inode(file
)->i_mode
))
2737 return __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, false);
2740 static inline void file_end_write(struct file
*file
)
2742 if (!S_ISREG(file_inode(file
)->i_mode
))
2744 __sb_end_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
);
2747 static inline int do_clone_file_range(struct file
*file_in
, loff_t pos_in
,
2748 struct file
*file_out
, loff_t pos_out
,
2753 file_start_write(file_out
);
2754 ret
= vfs_clone_file_range(file_in
, pos_in
, file_out
, pos_out
, len
);
2755 file_end_write(file_out
);
2761 * get_write_access() gets write permission for a file.
2762 * put_write_access() releases this write permission.
2763 * This is used for regular files.
2764 * We cannot support write (and maybe mmap read-write shared) accesses and
2765 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2766 * can have the following values:
2767 * 0: no writers, no VM_DENYWRITE mappings
2768 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2769 * > 0: (i_writecount) users are writing to the file.
2771 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2772 * except for the cases where we don't hold i_writecount yet. Then we need to
2773 * use {get,deny}_write_access() - these functions check the sign and refuse
2774 * to do the change if sign is wrong.
2776 static inline int get_write_access(struct inode
*inode
)
2778 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2780 static inline int deny_write_access(struct file
*file
)
2782 struct inode
*inode
= file_inode(file
);
2783 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2785 static inline void put_write_access(struct inode
* inode
)
2787 atomic_dec(&inode
->i_writecount
);
2789 static inline void allow_write_access(struct file
*file
)
2792 atomic_inc(&file_inode(file
)->i_writecount
);
2794 static inline bool inode_is_open_for_write(const struct inode
*inode
)
2796 return atomic_read(&inode
->i_writecount
) > 0;
2800 static inline void i_readcount_dec(struct inode
*inode
)
2802 BUG_ON(!atomic_read(&inode
->i_readcount
));
2803 atomic_dec(&inode
->i_readcount
);
2805 static inline void i_readcount_inc(struct inode
*inode
)
2807 atomic_inc(&inode
->i_readcount
);
2810 static inline void i_readcount_dec(struct inode
*inode
)
2814 static inline void i_readcount_inc(struct inode
*inode
)
2819 extern int do_pipe_flags(int *, int);
2821 #define __kernel_read_file_id(id) \
2822 id(UNKNOWN, unknown) \
2823 id(FIRMWARE, firmware) \
2824 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2825 id(MODULE, kernel-module) \
2826 id(KEXEC_IMAGE, kexec-image) \
2827 id(KEXEC_INITRAMFS, kexec-initramfs) \
2828 id(POLICY, security-policy) \
2829 id(X509_CERTIFICATE, x509-certificate) \
2832 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2833 #define __fid_stringify(dummy, str) #str,
2835 enum kernel_read_file_id
{
2836 __kernel_read_file_id(__fid_enumify
)
2839 static const char * const kernel_read_file_str
[] = {
2840 __kernel_read_file_id(__fid_stringify
)
2843 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id
)
2845 if ((unsigned)id
>= READING_MAX_ID
)
2846 return kernel_read_file_str
[READING_UNKNOWN
];
2848 return kernel_read_file_str
[id
];
2851 extern int kernel_read_file(struct file
*, void **, loff_t
*, loff_t
,
2852 enum kernel_read_file_id
);
2853 extern int kernel_read_file_from_path(const char *, void **, loff_t
*, loff_t
,
2854 enum kernel_read_file_id
);
2855 extern int kernel_read_file_from_fd(int, void **, loff_t
*, loff_t
,
2856 enum kernel_read_file_id
);
2857 extern ssize_t
kernel_read(struct file
*, void *, size_t, loff_t
*);
2858 extern ssize_t
kernel_write(struct file
*, const void *, size_t, loff_t
*);
2859 extern ssize_t
__kernel_write(struct file
*, const void *, size_t, loff_t
*);
2860 extern struct file
* open_exec(const char *);
2862 /* fs/dcache.c -- generic fs support functions */
2863 extern bool is_subdir(struct dentry
*, struct dentry
*);
2864 extern bool path_is_under(const struct path
*, const struct path
*);
2866 extern char *file_path(struct file
*, char *, int);
2868 #include <linux/err.h>
2870 /* needed for stackable file system support */
2871 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int whence
);
2873 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int whence
);
2875 extern int inode_init_always(struct super_block
*, struct inode
*);
2876 extern void inode_init_once(struct inode
*);
2877 extern void address_space_init_once(struct address_space
*mapping
);
2878 extern struct inode
* igrab(struct inode
*);
2879 extern ino_t
iunique(struct super_block
*, ino_t
);
2880 extern int inode_needs_sync(struct inode
*inode
);
2881 extern int generic_delete_inode(struct inode
*inode
);
2882 static inline int generic_drop_inode(struct inode
*inode
)
2884 return !inode
->i_nlink
|| inode_unhashed(inode
);
2887 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
2888 unsigned long hashval
, int (*test
)(struct inode
*, void *),
2890 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
2891 int (*test
)(struct inode
*, void *), void *data
);
2892 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
2894 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
2895 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
2896 extern struct inode
*find_inode_nowait(struct super_block
*,
2898 int (*match
)(struct inode
*,
2899 unsigned long, void *),
2901 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
2902 extern int insert_inode_locked(struct inode
*);
2903 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2904 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
2906 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
2908 extern void unlock_new_inode(struct inode
*);
2909 extern unsigned int get_next_ino(void);
2910 extern void evict_inodes(struct super_block
*sb
);
2912 extern void __iget(struct inode
* inode
);
2913 extern void iget_failed(struct inode
*);
2914 extern void clear_inode(struct inode
*);
2915 extern void __destroy_inode(struct inode
*);
2916 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
2917 extern struct inode
*new_inode(struct super_block
*sb
);
2918 extern void free_inode_nonrcu(struct inode
*inode
);
2919 extern int should_remove_suid(struct dentry
*);
2920 extern int file_remove_privs(struct file
*);
2922 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
2923 static inline void insert_inode_hash(struct inode
*inode
)
2925 __insert_inode_hash(inode
, inode
->i_ino
);
2928 extern void __remove_inode_hash(struct inode
*);
2929 static inline void remove_inode_hash(struct inode
*inode
)
2931 if (!inode_unhashed(inode
) && !hlist_fake(&inode
->i_hash
))
2932 __remove_inode_hash(inode
);
2935 extern void inode_sb_list_add(struct inode
*inode
);
2938 extern int bdev_read_only(struct block_device
*);
2940 extern int set_blocksize(struct block_device
*, int);
2941 extern int sb_set_blocksize(struct super_block
*, int);
2942 extern int sb_min_blocksize(struct super_block
*, int);
2944 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
2945 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
2946 extern ssize_t
generic_write_checks(struct kiocb
*, struct iov_iter
*);
2947 extern ssize_t
generic_file_read_iter(struct kiocb
*, struct iov_iter
*);
2948 extern ssize_t
__generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2949 extern ssize_t
generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2950 extern ssize_t
generic_file_direct_write(struct kiocb
*, struct iov_iter
*);
2951 extern ssize_t
generic_perform_write(struct file
*, struct iov_iter
*, loff_t
);
2953 ssize_t
vfs_iter_read(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
2955 ssize_t
vfs_iter_write(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
2958 /* fs/block_dev.c */
2959 extern ssize_t
blkdev_read_iter(struct kiocb
*iocb
, struct iov_iter
*to
);
2960 extern ssize_t
blkdev_write_iter(struct kiocb
*iocb
, struct iov_iter
*from
);
2961 extern int blkdev_fsync(struct file
*filp
, loff_t start
, loff_t end
,
2963 extern void block_sync_page(struct page
*page
);
2966 extern ssize_t
generic_file_splice_read(struct file
*, loff_t
*,
2967 struct pipe_inode_info
*, size_t, unsigned int);
2968 extern ssize_t
iter_file_splice_write(struct pipe_inode_info
*,
2969 struct file
*, loff_t
*, size_t, unsigned int);
2970 extern ssize_t
generic_splice_sendpage(struct pipe_inode_info
*pipe
,
2971 struct file
*out
, loff_t
*, size_t len
, unsigned int flags
);
2972 extern long do_splice_direct(struct file
*in
, loff_t
*ppos
, struct file
*out
,
2973 loff_t
*opos
, size_t len
, unsigned int flags
);
2977 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
2978 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int whence
);
2979 extern loff_t
no_llseek(struct file
*file
, loff_t offset
, int whence
);
2980 extern loff_t
vfs_setpos(struct file
*file
, loff_t offset
, loff_t maxsize
);
2981 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int whence
);
2982 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
2983 int whence
, loff_t maxsize
, loff_t eof
);
2984 extern loff_t
fixed_size_llseek(struct file
*file
, loff_t offset
,
2985 int whence
, loff_t size
);
2986 extern loff_t
no_seek_end_llseek_size(struct file
*, loff_t
, int, loff_t
);
2987 extern loff_t
no_seek_end_llseek(struct file
*, loff_t
, int);
2988 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
2989 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
2992 typedef void (dio_submit_t
)(struct bio
*bio
, struct inode
*inode
,
2993 loff_t file_offset
);
2996 /* need locking between buffered and direct access */
2999 /* filesystem does not support filling holes */
3000 DIO_SKIP_HOLES
= 0x02,
3002 /* filesystem can handle aio writes beyond i_size */
3003 DIO_ASYNC_EXTEND
= 0x04,
3005 /* inode/fs/bdev does not need truncate protection */
3006 DIO_SKIP_DIO_COUNT
= 0x08,
3009 void dio_end_io(struct bio
*bio
);
3011 ssize_t
__blockdev_direct_IO(struct kiocb
*iocb
, struct inode
*inode
,
3012 struct block_device
*bdev
, struct iov_iter
*iter
,
3013 get_block_t get_block
,
3014 dio_iodone_t end_io
, dio_submit_t submit_io
,
3017 static inline ssize_t
blockdev_direct_IO(struct kiocb
*iocb
,
3018 struct inode
*inode
,
3019 struct iov_iter
*iter
,
3020 get_block_t get_block
)
3022 return __blockdev_direct_IO(iocb
, inode
, inode
->i_sb
->s_bdev
, iter
,
3023 get_block
, NULL
, NULL
, DIO_LOCKING
| DIO_SKIP_HOLES
);
3027 void inode_dio_wait(struct inode
*inode
);
3030 * inode_dio_begin - signal start of a direct I/O requests
3031 * @inode: inode the direct I/O happens on
3033 * This is called once we've finished processing a direct I/O request,
3034 * and is used to wake up callers waiting for direct I/O to be quiesced.
3036 static inline void inode_dio_begin(struct inode
*inode
)
3038 atomic_inc(&inode
->i_dio_count
);
3042 * inode_dio_end - signal finish of a direct I/O requests
3043 * @inode: inode the direct I/O happens on
3045 * This is called once we've finished processing a direct I/O request,
3046 * and is used to wake up callers waiting for direct I/O to be quiesced.
3048 static inline void inode_dio_end(struct inode
*inode
)
3050 if (atomic_dec_and_test(&inode
->i_dio_count
))
3051 wake_up_bit(&inode
->i_state
, __I_DIO_WAKEUP
);
3054 extern void inode_set_flags(struct inode
*inode
, unsigned int flags
,
3057 extern const struct file_operations generic_ro_fops
;
3059 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3061 extern int readlink_copy(char __user
*, int, const char *);
3062 extern int page_readlink(struct dentry
*, char __user
*, int);
3063 extern const char *page_get_link(struct dentry
*, struct inode
*,
3064 struct delayed_call
*);
3065 extern void page_put_link(void *);
3066 extern int __page_symlink(struct inode
*inode
, const char *symname
, int len
,
3068 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
3069 extern const struct inode_operations page_symlink_inode_operations
;
3070 extern void kfree_link(void *);
3071 extern void generic_fillattr(struct inode
*, struct kstat
*);
3072 extern int vfs_getattr_nosec(const struct path
*, struct kstat
*, u32
, unsigned int);
3073 extern int vfs_getattr(const struct path
*, struct kstat
*, u32
, unsigned int);
3074 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
3075 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
3076 void __inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
3077 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
3078 static inline loff_t
__inode_get_bytes(struct inode
*inode
)
3080 return (((loff_t
)inode
->i_blocks
) << 9) + inode
->i_bytes
;
3082 loff_t
inode_get_bytes(struct inode
*inode
);
3083 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
3084 const char *simple_get_link(struct dentry
*, struct inode
*,
3085 struct delayed_call
*);
3086 extern const struct inode_operations simple_symlink_inode_operations
;
3088 extern int iterate_dir(struct file
*, struct dir_context
*);
3090 extern int vfs_statx(int, const char __user
*, int, struct kstat
*, u32
);
3091 extern int vfs_statx_fd(unsigned int, struct kstat
*, u32
, unsigned int);
3093 static inline int vfs_stat(const char __user
*filename
, struct kstat
*stat
)
3095 return vfs_statx(AT_FDCWD
, filename
, AT_NO_AUTOMOUNT
,
3096 stat
, STATX_BASIC_STATS
);
3098 static inline int vfs_lstat(const char __user
*name
, struct kstat
*stat
)
3100 return vfs_statx(AT_FDCWD
, name
, AT_SYMLINK_NOFOLLOW
| AT_NO_AUTOMOUNT
,
3101 stat
, STATX_BASIC_STATS
);
3103 static inline int vfs_fstatat(int dfd
, const char __user
*filename
,
3104 struct kstat
*stat
, int flags
)
3106 return vfs_statx(dfd
, filename
, flags
| AT_NO_AUTOMOUNT
,
3107 stat
, STATX_BASIC_STATS
);
3109 static inline int vfs_fstat(int fd
, struct kstat
*stat
)
3111 return vfs_statx_fd(fd
, stat
, STATX_BASIC_STATS
, 0);
3115 extern const char *vfs_get_link(struct dentry
*, struct delayed_call
*);
3116 extern int vfs_readlink(struct dentry
*, char __user
*, int);
3118 extern int __generic_block_fiemap(struct inode
*inode
,
3119 struct fiemap_extent_info
*fieinfo
,
3120 loff_t start
, loff_t len
,
3121 get_block_t
*get_block
);
3122 extern int generic_block_fiemap(struct inode
*inode
,
3123 struct fiemap_extent_info
*fieinfo
, u64 start
,
3124 u64 len
, get_block_t
*get_block
);
3126 extern struct file_system_type
*get_filesystem(struct file_system_type
*fs
);
3127 extern void put_filesystem(struct file_system_type
*fs
);
3128 extern struct file_system_type
*get_fs_type(const char *name
);
3129 extern struct super_block
*get_super(struct block_device
*);
3130 extern struct super_block
*get_super_thawed(struct block_device
*);
3131 extern struct super_block
*get_super_exclusive_thawed(struct block_device
*bdev
);
3132 extern struct super_block
*get_active_super(struct block_device
*bdev
);
3133 extern void drop_super(struct super_block
*sb
);
3134 extern void drop_super_exclusive(struct super_block
*sb
);
3135 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
3136 extern void iterate_supers_type(struct file_system_type
*,
3137 void (*)(struct super_block
*, void *), void *);
3139 extern int dcache_dir_open(struct inode
*, struct file
*);
3140 extern int dcache_dir_close(struct inode
*, struct file
*);
3141 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
3142 extern int dcache_readdir(struct file
*, struct dir_context
*);
3143 extern int simple_setattr(struct dentry
*, struct iattr
*);
3144 extern int simple_getattr(const struct path
*, struct kstat
*, u32
, unsigned int);
3145 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
3146 extern int simple_open(struct inode
*inode
, struct file
*file
);
3147 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
3148 extern int simple_unlink(struct inode
*, struct dentry
*);
3149 extern int simple_rmdir(struct inode
*, struct dentry
*);
3150 extern int simple_rename(struct inode
*, struct dentry
*,
3151 struct inode
*, struct dentry
*, unsigned int);
3152 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
3153 extern int simple_empty(struct dentry
*);
3154 extern int simple_readpage(struct file
*file
, struct page
*page
);
3155 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
3156 loff_t pos
, unsigned len
, unsigned flags
,
3157 struct page
**pagep
, void **fsdata
);
3158 extern int simple_write_end(struct file
*file
, struct address_space
*mapping
,
3159 loff_t pos
, unsigned len
, unsigned copied
,
3160 struct page
*page
, void *fsdata
);
3161 extern int always_delete_dentry(const struct dentry
*);
3162 extern struct inode
*alloc_anon_inode(struct super_block
*);
3163 extern int simple_nosetlease(struct file
*, long, struct file_lock
**, void **);
3164 extern const struct dentry_operations simple_dentry_operations
;
3166 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
3167 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
3168 extern const struct file_operations simple_dir_operations
;
3169 extern const struct inode_operations simple_dir_inode_operations
;
3170 extern void make_empty_dir_inode(struct inode
*inode
);
3171 extern bool is_empty_dir_inode(struct inode
*inode
);
3172 struct tree_descr
{ const char *name
; const struct file_operations
*ops
; int mode
; };
3173 struct dentry
*d_alloc_name(struct dentry
*, const char *);
3174 extern int simple_fill_super(struct super_block
*, unsigned long,
3175 const struct tree_descr
*);
3176 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
3177 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
3179 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
3180 loff_t
*ppos
, const void *from
, size_t available
);
3181 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
3182 const void __user
*from
, size_t count
);
3184 extern int __generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3185 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3187 extern int generic_check_addressable(unsigned, u64
);
3189 #ifdef CONFIG_MIGRATION
3190 extern int buffer_migrate_page(struct address_space
*,
3191 struct page
*, struct page
*,
3194 #define buffer_migrate_page NULL
3197 extern int setattr_prepare(struct dentry
*, struct iattr
*);
3198 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
3199 extern void setattr_copy(struct inode
*inode
, const struct iattr
*attr
);
3201 extern int file_update_time(struct file
*file
);
3203 static inline bool io_is_direct(struct file
*filp
)
3205 return (filp
->f_flags
& O_DIRECT
) || IS_DAX(filp
->f_mapping
->host
);
3208 static inline bool vma_is_dax(struct vm_area_struct
*vma
)
3210 return vma
->vm_file
&& IS_DAX(vma
->vm_file
->f_mapping
->host
);
3213 static inline bool vma_is_fsdax(struct vm_area_struct
*vma
)
3215 struct inode
*inode
;
3219 if (!vma_is_dax(vma
))
3221 inode
= file_inode(vma
->vm_file
);
3222 if (S_ISCHR(inode
->i_mode
))
3223 return false; /* device-dax */
3227 static inline int iocb_flags(struct file
*file
)
3230 if (file
->f_flags
& O_APPEND
)
3232 if (io_is_direct(file
))
3234 if ((file
->f_flags
& O_DSYNC
) || IS_SYNC(file
->f_mapping
->host
))
3236 if (file
->f_flags
& __O_SYNC
)
3241 static inline int kiocb_set_rw_flags(struct kiocb
*ki
, rwf_t flags
)
3243 if (unlikely(flags
& ~RWF_SUPPORTED
))
3246 if (flags
& RWF_NOWAIT
) {
3247 if (!(ki
->ki_filp
->f_mode
& FMODE_NOWAIT
))
3249 ki
->ki_flags
|= IOCB_NOWAIT
;
3251 if (flags
& RWF_HIPRI
)
3252 ki
->ki_flags
|= IOCB_HIPRI
;
3253 if (flags
& RWF_DSYNC
)
3254 ki
->ki_flags
|= IOCB_DSYNC
;
3255 if (flags
& RWF_SYNC
)
3256 ki
->ki_flags
|= (IOCB_DSYNC
| IOCB_SYNC
);
3260 static inline ino_t
parent_ino(struct dentry
*dentry
)
3265 * Don't strictly need d_lock here? If the parent ino could change
3266 * then surely we'd have a deeper race in the caller?
3268 spin_lock(&dentry
->d_lock
);
3269 res
= dentry
->d_parent
->d_inode
->i_ino
;
3270 spin_unlock(&dentry
->d_lock
);
3274 /* Transaction based IO helpers */
3277 * An argresp is stored in an allocated page and holds the
3278 * size of the argument or response, along with its content
3280 struct simple_transaction_argresp
{
3285 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3287 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
3289 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
3290 size_t size
, loff_t
*pos
);
3291 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
3293 void simple_transaction_set(struct file
*file
, size_t n
);
3296 * simple attribute files
3298 * These attributes behave similar to those in sysfs:
3300 * Writing to an attribute immediately sets a value, an open file can be
3301 * written to multiple times.
3303 * Reading from an attribute creates a buffer from the value that might get
3304 * read with multiple read calls. When the attribute has been read
3305 * completely, no further read calls are possible until the file is opened
3308 * All attributes contain a text representation of a numeric value
3309 * that are accessed with the get() and set() functions.
3311 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3312 static int __fops ## _open(struct inode *inode, struct file *file) \
3314 __simple_attr_check_format(__fmt, 0ull); \
3315 return simple_attr_open(inode, file, __get, __set, __fmt); \
3317 static const struct file_operations __fops = { \
3318 .owner = THIS_MODULE, \
3319 .open = __fops ## _open, \
3320 .release = simple_attr_release, \
3321 .read = simple_attr_read, \
3322 .write = simple_attr_write, \
3323 .llseek = generic_file_llseek, \
3326 static inline __printf(1, 2)
3327 void __simple_attr_check_format(const char *fmt
, ...)
3329 /* don't do anything, just let the compiler check the arguments; */
3332 int simple_attr_open(struct inode
*inode
, struct file
*file
,
3333 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
3335 int simple_attr_release(struct inode
*inode
, struct file
*file
);
3336 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
3337 size_t len
, loff_t
*ppos
);
3338 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
3339 size_t len
, loff_t
*ppos
);
3342 int proc_nr_files(struct ctl_table
*table
, int write
,
3343 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3344 int proc_nr_dentry(struct ctl_table
*table
, int write
,
3345 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3346 int proc_nr_inodes(struct ctl_table
*table
, int write
,
3347 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3348 int __init
get_filesystem_list(char *buf
);
3350 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3351 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3353 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3354 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3355 (flag & __FMODE_NONOTIFY)))
3357 static inline bool is_sxid(umode_t mode
)
3359 return (mode
& S_ISUID
) || ((mode
& S_ISGID
) && (mode
& S_IXGRP
));
3362 static inline int check_sticky(struct inode
*dir
, struct inode
*inode
)
3364 if (!(dir
->i_mode
& S_ISVTX
))
3367 return __check_sticky(dir
, inode
);
3370 static inline void inode_has_no_xattr(struct inode
*inode
)
3372 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& SB_NOSEC
))
3373 inode
->i_flags
|= S_NOSEC
;
3376 static inline bool is_root_inode(struct inode
*inode
)
3378 return inode
== inode
->i_sb
->s_root
->d_inode
;
3381 static inline bool dir_emit(struct dir_context
*ctx
,
3382 const char *name
, int namelen
,
3383 u64 ino
, unsigned type
)
3385 return ctx
->actor(ctx
, name
, namelen
, ctx
->pos
, ino
, type
) == 0;
3387 static inline bool dir_emit_dot(struct file
*file
, struct dir_context
*ctx
)
3389 return ctx
->actor(ctx
, ".", 1, ctx
->pos
,
3390 file
->f_path
.dentry
->d_inode
->i_ino
, DT_DIR
) == 0;
3392 static inline bool dir_emit_dotdot(struct file
*file
, struct dir_context
*ctx
)
3394 return ctx
->actor(ctx
, "..", 2, ctx
->pos
,
3395 parent_ino(file
->f_path
.dentry
), DT_DIR
) == 0;
3397 static inline bool dir_emit_dots(struct file
*file
, struct dir_context
*ctx
)
3399 if (ctx
->pos
== 0) {
3400 if (!dir_emit_dot(file
, ctx
))
3404 if (ctx
->pos
== 1) {
3405 if (!dir_emit_dotdot(file
, ctx
))
3411 static inline bool dir_relax(struct inode
*inode
)
3413 inode_unlock(inode
);
3415 return !IS_DEADDIR(inode
);
3418 static inline bool dir_relax_shared(struct inode
*inode
)
3420 inode_unlock_shared(inode
);
3421 inode_lock_shared(inode
);
3422 return !IS_DEADDIR(inode
);
3425 extern bool path_noexec(const struct path
*path
);
3426 extern bool path_nosuid(const struct path
*path
);
3427 extern void inode_nohighmem(struct inode
*inode
);
3429 #endif /* _LINUX_FS_H */