5 #include <linux/linkage.h>
6 #include <linux/wait.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/capability.h>
23 #include <linux/semaphore.h>
24 #include <linux/fiemap.h>
25 #include <linux/rculist_bl.h>
26 #include <linux/atomic.h>
27 #include <linux/shrinker.h>
28 #include <linux/migrate_mode.h>
29 #include <linux/uidgid.h>
30 #include <linux/lockdep.h>
31 #include <linux/percpu-rwsem.h>
32 #include <linux/blk_types.h>
34 #include <asm/byteorder.h>
35 #include <uapi/linux/fs.h>
37 struct backing_dev_info
;
39 struct export_operations
;
44 struct pipe_inode_info
;
45 struct poll_table_struct
;
47 struct vm_area_struct
;
50 struct swap_info_struct
;
52 struct workqueue_struct
;
56 extern void __init
inode_init(void);
57 extern void __init
inode_init_early(void);
58 extern void __init
files_init(void);
59 extern void __init
files_maxfiles_init(void);
61 extern struct files_stat_struct files_stat
;
62 extern unsigned long get_max_files(void);
63 extern int sysctl_nr_open
;
64 extern struct inodes_stat_t inodes_stat
;
65 extern int leases_enable
, lease_break_time
;
66 extern int sysctl_protected_symlinks
;
67 extern int sysctl_protected_hardlinks
;
70 typedef int (get_block_t
)(struct inode
*inode
, sector_t iblock
,
71 struct buffer_head
*bh_result
, int create
);
72 typedef void (dio_iodone_t
)(struct kiocb
*iocb
, loff_t offset
,
73 ssize_t bytes
, void *private);
74 typedef void (dax_iodone_t
)(struct buffer_head
*bh_map
, int uptodate
);
76 #define MAY_EXEC 0x00000001
77 #define MAY_WRITE 0x00000002
78 #define MAY_READ 0x00000004
79 #define MAY_APPEND 0x00000008
80 #define MAY_ACCESS 0x00000010
81 #define MAY_OPEN 0x00000020
82 #define MAY_CHDIR 0x00000040
83 /* called from RCU mode, don't block */
84 #define MAY_NOT_BLOCK 0x00000080
87 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
88 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
91 /* file is open for reading */
92 #define FMODE_READ ((__force fmode_t)0x1)
93 /* file is open for writing */
94 #define FMODE_WRITE ((__force fmode_t)0x2)
95 /* file is seekable */
96 #define FMODE_LSEEK ((__force fmode_t)0x4)
97 /* file can be accessed using pread */
98 #define FMODE_PREAD ((__force fmode_t)0x8)
99 /* file can be accessed using pwrite */
100 #define FMODE_PWRITE ((__force fmode_t)0x10)
101 /* File is opened for execution with sys_execve / sys_uselib */
102 #define FMODE_EXEC ((__force fmode_t)0x20)
103 /* File is opened with O_NDELAY (only set for block devices) */
104 #define FMODE_NDELAY ((__force fmode_t)0x40)
105 /* File is opened with O_EXCL (only set for block devices) */
106 #define FMODE_EXCL ((__force fmode_t)0x80)
107 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
108 (specialy hack for floppy.c) */
109 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
110 /* 32bit hashes as llseek() offset (for directories) */
111 #define FMODE_32BITHASH ((__force fmode_t)0x200)
112 /* 64bit hashes as llseek() offset (for directories) */
113 #define FMODE_64BITHASH ((__force fmode_t)0x400)
116 * Don't update ctime and mtime.
118 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
119 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
121 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
123 /* Expect random access pattern */
124 #define FMODE_RANDOM ((__force fmode_t)0x1000)
126 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
127 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
129 /* File is opened with O_PATH; almost nothing can be done with it */
130 #define FMODE_PATH ((__force fmode_t)0x4000)
132 /* File needs atomic accesses to f_pos */
133 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
134 /* Write access to underlying fs */
135 #define FMODE_WRITER ((__force fmode_t)0x10000)
136 /* Has read method(s) */
137 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
138 /* Has write method(s) */
139 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
141 /* File was opened by fanotify and shouldn't generate fanotify events */
142 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
145 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
146 * that indicates that they should check the contents of the iovec are
147 * valid, but not check the memory that the iovec elements
150 #define CHECK_IOVEC_ONLY -1
153 * The below are the various read and write types that we support. Some of
154 * them include behavioral modifiers that send information down to the
155 * block layer and IO scheduler. Terminology:
157 * The block layer uses device plugging to defer IO a little bit, in
158 * the hope that we will see more IO very shortly. This increases
159 * coalescing of adjacent IO and thus reduces the number of IOs we
160 * have to send to the device. It also allows for better queuing,
161 * if the IO isn't mergeable. If the caller is going to be waiting
162 * for the IO, then he must ensure that the device is unplugged so
163 * that the IO is dispatched to the driver.
165 * All IO is handled async in Linux. This is fine for background
166 * writes, but for reads or writes that someone waits for completion
167 * on, we want to notify the block layer and IO scheduler so that they
168 * know about it. That allows them to make better scheduling
169 * decisions. So when the below references 'sync' and 'async', it
170 * is referencing this priority hint.
172 * With that in mind, the available types are:
174 * READ A normal read operation. Device will be plugged.
175 * READ_SYNC A synchronous read. Device is not plugged, caller can
176 * immediately wait on this read without caring about
178 * READA Used for read-ahead operations. Lower priority, and the
179 * block layer could (in theory) choose to ignore this
180 * request if it runs into resource problems.
181 * WRITE A normal async write. Device will be plugged.
182 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
183 * the hint that someone will be waiting on this IO
184 * shortly. The write equivalent of READ_SYNC.
185 * WRITE_ODIRECT Special case write for O_DIRECT only.
186 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
187 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
188 * non-volatile media on completion.
189 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
190 * by a cache flush and data is guaranteed to be on
191 * non-volatile media on completion.
194 #define RW_MASK REQ_WRITE
195 #define RWA_MASK REQ_RAHEAD
198 #define WRITE RW_MASK
199 #define READA RWA_MASK
201 #define READ_SYNC (READ | REQ_SYNC)
202 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
203 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
204 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
205 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
206 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
209 * Attribute flags. These should be or-ed together to figure out what
212 #define ATTR_MODE (1 << 0)
213 #define ATTR_UID (1 << 1)
214 #define ATTR_GID (1 << 2)
215 #define ATTR_SIZE (1 << 3)
216 #define ATTR_ATIME (1 << 4)
217 #define ATTR_MTIME (1 << 5)
218 #define ATTR_CTIME (1 << 6)
219 #define ATTR_ATIME_SET (1 << 7)
220 #define ATTR_MTIME_SET (1 << 8)
221 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
222 #define ATTR_ATTR_FLAG (1 << 10)
223 #define ATTR_KILL_SUID (1 << 11)
224 #define ATTR_KILL_SGID (1 << 12)
225 #define ATTR_FILE (1 << 13)
226 #define ATTR_KILL_PRIV (1 << 14)
227 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
228 #define ATTR_TIMES_SET (1 << 16)
231 * Whiteout is represented by a char device. The following constants define the
232 * mode and device number to use.
234 #define WHITEOUT_MODE 0
235 #define WHITEOUT_DEV 0
238 * This is the Inode Attributes structure, used for notify_change(). It
239 * uses the above definitions as flags, to know which values have changed.
240 * Also, in this manner, a Filesystem can look at only the values it cares
241 * about. Basically, these are the attributes that the VFS layer can
242 * request to change from the FS layer.
244 * Derek Atkins <warlord@MIT.EDU> 94-10-20
247 unsigned int ia_valid
;
252 struct timespec ia_atime
;
253 struct timespec ia_mtime
;
254 struct timespec ia_ctime
;
257 * Not an attribute, but an auxiliary info for filesystems wanting to
258 * implement an ftruncate() like method. NOTE: filesystem should
259 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
261 struct file
*ia_file
;
265 * Includes for diskquotas.
267 #include <linux/quota.h>
270 * Maximum number of layers of fs stack. Needs to be limited to
271 * prevent kernel stack overflow
273 #define FILESYSTEM_MAX_STACK_DEPTH 2
276 * enum positive_aop_returns - aop return codes with specific semantics
278 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
279 * completed, that the page is still locked, and
280 * should be considered active. The VM uses this hint
281 * to return the page to the active list -- it won't
282 * be a candidate for writeback again in the near
283 * future. Other callers must be careful to unlock
284 * the page if they get this return. Returned by
287 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
288 * unlocked it and the page might have been truncated.
289 * The caller should back up to acquiring a new page and
290 * trying again. The aop will be taking reasonable
291 * precautions not to livelock. If the caller held a page
292 * reference, it should drop it before retrying. Returned
295 * address_space_operation functions return these large constants to indicate
296 * special semantics to the caller. These are much larger than the bytes in a
297 * page to allow for functions that return the number of bytes operated on in a
301 enum positive_aop_returns
{
302 AOP_WRITEPAGE_ACTIVATE
= 0x80000,
303 AOP_TRUNCATED_PAGE
= 0x80001,
306 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
307 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
308 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
309 * helper code (eg buffer layer)
310 * to clear GFP_FS from alloc */
313 * oh the beauties of C type declarations.
316 struct address_space
;
317 struct writeback_control
;
319 #define IOCB_EVENTFD (1 << 0)
320 #define IOCB_APPEND (1 << 1)
321 #define IOCB_DIRECT (1 << 2)
324 struct file
*ki_filp
;
326 void (*ki_complete
)(struct kiocb
*iocb
, long ret
, long ret2
);
331 static inline bool is_sync_kiocb(struct kiocb
*kiocb
)
333 return kiocb
->ki_complete
== NULL
;
336 static inline int iocb_flags(struct file
*file
);
338 static inline void init_sync_kiocb(struct kiocb
*kiocb
, struct file
*filp
)
340 *kiocb
= (struct kiocb
) {
342 .ki_flags
= iocb_flags(filp
),
347 * "descriptor" for what we're up to with a read.
348 * This allows us to use the same read code yet
349 * have multiple different users of the data that
350 * we read from a file.
352 * The simplest case just copies the data to user
365 typedef int (*read_actor_t
)(read_descriptor_t
*, struct page
*,
366 unsigned long, unsigned long);
368 struct address_space_operations
{
369 int (*writepage
)(struct page
*page
, struct writeback_control
*wbc
);
370 int (*readpage
)(struct file
*, struct page
*);
372 /* Write back some dirty pages from this mapping. */
373 int (*writepages
)(struct address_space
*, struct writeback_control
*);
375 /* Set a page dirty. Return true if this dirtied it */
376 int (*set_page_dirty
)(struct page
*page
);
378 int (*readpages
)(struct file
*filp
, struct address_space
*mapping
,
379 struct list_head
*pages
, unsigned nr_pages
);
381 int (*write_begin
)(struct file
*, struct address_space
*mapping
,
382 loff_t pos
, unsigned len
, unsigned flags
,
383 struct page
**pagep
, void **fsdata
);
384 int (*write_end
)(struct file
*, struct address_space
*mapping
,
385 loff_t pos
, unsigned len
, unsigned copied
,
386 struct page
*page
, void *fsdata
);
388 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
389 sector_t (*bmap
)(struct address_space
*, sector_t
);
390 void (*invalidatepage
) (struct page
*, unsigned int, unsigned int);
391 int (*releasepage
) (struct page
*, gfp_t
);
392 void (*freepage
)(struct page
*);
393 ssize_t (*direct_IO
)(struct kiocb
*, struct iov_iter
*iter
, loff_t offset
);
395 * migrate the contents of a page to the specified target. If
396 * migrate_mode is MIGRATE_ASYNC, it must not block.
398 int (*migratepage
) (struct address_space
*,
399 struct page
*, struct page
*, enum migrate_mode
);
400 int (*launder_page
) (struct page
*);
401 int (*is_partially_uptodate
) (struct page
*, unsigned long,
403 void (*is_dirty_writeback
) (struct page
*, bool *, bool *);
404 int (*error_remove_page
)(struct address_space
*, struct page
*);
406 /* swapfile support */
407 int (*swap_activate
)(struct swap_info_struct
*sis
, struct file
*file
,
409 void (*swap_deactivate
)(struct file
*file
);
412 extern const struct address_space_operations empty_aops
;
415 * pagecache_write_begin/pagecache_write_end must be used by general code
416 * to write into the pagecache.
418 int pagecache_write_begin(struct file
*, struct address_space
*mapping
,
419 loff_t pos
, unsigned len
, unsigned flags
,
420 struct page
**pagep
, void **fsdata
);
422 int pagecache_write_end(struct file
*, struct address_space
*mapping
,
423 loff_t pos
, unsigned len
, unsigned copied
,
424 struct page
*page
, void *fsdata
);
426 struct address_space
{
427 struct inode
*host
; /* owner: inode, block_device */
428 struct radix_tree_root page_tree
; /* radix tree of all pages */
429 spinlock_t tree_lock
; /* and lock protecting it */
430 atomic_t i_mmap_writable
;/* count VM_SHARED mappings */
431 struct rb_root i_mmap
; /* tree of private and shared mappings */
432 struct rw_semaphore i_mmap_rwsem
; /* protect tree, count, list */
433 /* Protected by tree_lock together with the radix tree */
434 unsigned long nrpages
; /* number of total pages */
435 unsigned long nrshadows
; /* number of shadow entries */
436 pgoff_t writeback_index
;/* writeback starts here */
437 const struct address_space_operations
*a_ops
; /* methods */
438 unsigned long flags
; /* error bits/gfp mask */
439 spinlock_t private_lock
; /* for use by the address_space */
440 struct list_head private_list
; /* ditto */
441 void *private_data
; /* ditto */
442 } __attribute__((aligned(sizeof(long))));
444 * On most architectures that alignment is already the case; but
445 * must be enforced here for CRIS, to let the least significant bit
446 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
448 struct request_queue
;
450 struct block_device
{
451 dev_t bd_dev
; /* not a kdev_t - it's a search key */
453 struct inode
* bd_inode
; /* will die */
454 struct super_block
* bd_super
;
455 struct mutex bd_mutex
; /* open/close mutex */
456 struct list_head bd_inodes
;
460 bool bd_write_holder
;
462 struct list_head bd_holder_disks
;
464 struct block_device
* bd_contains
;
465 unsigned bd_block_size
;
466 struct hd_struct
* bd_part
;
467 /* number of times partitions within this device have been opened. */
468 unsigned bd_part_count
;
470 struct gendisk
* bd_disk
;
471 struct request_queue
* bd_queue
;
472 struct list_head bd_list
;
474 * Private data. You must have bd_claim'ed the block_device
475 * to use this. NOTE: bd_claim allows an owner to claim
476 * the same device multiple times, the owner must take special
477 * care to not mess up bd_private for that case.
479 unsigned long bd_private
;
481 /* The counter of freeze processes */
482 int bd_fsfreeze_count
;
483 /* Mutex for freeze */
484 struct mutex bd_fsfreeze_mutex
;
488 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
491 #define PAGECACHE_TAG_DIRTY 0
492 #define PAGECACHE_TAG_WRITEBACK 1
493 #define PAGECACHE_TAG_TOWRITE 2
495 int mapping_tagged(struct address_space
*mapping
, int tag
);
497 static inline void i_mmap_lock_write(struct address_space
*mapping
)
499 down_write(&mapping
->i_mmap_rwsem
);
502 static inline void i_mmap_unlock_write(struct address_space
*mapping
)
504 up_write(&mapping
->i_mmap_rwsem
);
507 static inline void i_mmap_lock_read(struct address_space
*mapping
)
509 down_read(&mapping
->i_mmap_rwsem
);
512 static inline void i_mmap_unlock_read(struct address_space
*mapping
)
514 up_read(&mapping
->i_mmap_rwsem
);
518 * Might pages of this file be mapped into userspace?
520 static inline int mapping_mapped(struct address_space
*mapping
)
522 return !RB_EMPTY_ROOT(&mapping
->i_mmap
);
526 * Might pages of this file have been modified in userspace?
527 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
528 * marks vma as VM_SHARED if it is shared, and the file was opened for
529 * writing i.e. vma may be mprotected writable even if now readonly.
531 * If i_mmap_writable is negative, no new writable mappings are allowed. You
532 * can only deny writable mappings, if none exists right now.
534 static inline int mapping_writably_mapped(struct address_space
*mapping
)
536 return atomic_read(&mapping
->i_mmap_writable
) > 0;
539 static inline int mapping_map_writable(struct address_space
*mapping
)
541 return atomic_inc_unless_negative(&mapping
->i_mmap_writable
) ?
545 static inline void mapping_unmap_writable(struct address_space
*mapping
)
547 atomic_dec(&mapping
->i_mmap_writable
);
550 static inline int mapping_deny_writable(struct address_space
*mapping
)
552 return atomic_dec_unless_positive(&mapping
->i_mmap_writable
) ?
556 static inline void mapping_allow_writable(struct address_space
*mapping
)
558 atomic_inc(&mapping
->i_mmap_writable
);
562 * Use sequence counter to get consistent i_size on 32-bit processors.
564 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
565 #include <linux/seqlock.h>
566 #define __NEED_I_SIZE_ORDERED
567 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
569 #define i_size_ordered_init(inode) do { } while (0)
573 #define ACL_NOT_CACHED ((void *)(-1))
575 #define IOP_FASTPERM 0x0001
576 #define IOP_LOOKUP 0x0002
577 #define IOP_NOFOLLOW 0x0004
580 * Keep mostly read-only and often accessed (especially for
581 * the RCU path lookup and 'stat' data) fields at the beginning
582 * of the 'struct inode'
586 unsigned short i_opflags
;
589 unsigned int i_flags
;
591 #ifdef CONFIG_FS_POSIX_ACL
592 struct posix_acl
*i_acl
;
593 struct posix_acl
*i_default_acl
;
596 const struct inode_operations
*i_op
;
597 struct super_block
*i_sb
;
598 struct address_space
*i_mapping
;
600 #ifdef CONFIG_SECURITY
604 /* Stat data, not accessed from path walking */
607 * Filesystems may only read i_nlink directly. They shall use the
608 * following functions for modification:
610 * (set|clear|inc|drop)_nlink
611 * inode_(inc|dec)_link_count
614 const unsigned int i_nlink
;
615 unsigned int __i_nlink
;
619 struct timespec i_atime
;
620 struct timespec i_mtime
;
621 struct timespec i_ctime
;
622 spinlock_t i_lock
; /* i_blocks, i_bytes, maybe i_size */
623 unsigned short i_bytes
;
624 unsigned int i_blkbits
;
627 #ifdef __NEED_I_SIZE_ORDERED
628 seqcount_t i_size_seqcount
;
632 unsigned long i_state
;
633 struct mutex i_mutex
;
635 unsigned long dirtied_when
; /* jiffies of first dirtying */
636 unsigned long dirtied_time_when
;
638 struct hlist_node i_hash
;
639 struct list_head i_wb_list
; /* backing dev IO list */
640 #ifdef CONFIG_CGROUP_WRITEBACK
641 struct bdi_writeback
*i_wb
; /* the associated cgroup wb */
643 /* foreign inode detection, see wbc_detach_inode() */
645 u16 i_wb_frn_avg_time
;
646 u16 i_wb_frn_history
;
648 struct list_head i_lru
; /* inode LRU list */
649 struct list_head i_sb_list
;
651 struct hlist_head i_dentry
;
652 struct rcu_head i_rcu
;
656 atomic_t i_dio_count
;
657 atomic_t i_writecount
;
659 atomic_t i_readcount
; /* struct files open RO */
661 const struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
662 struct file_lock_context
*i_flctx
;
663 struct address_space i_data
;
664 struct list_head i_devices
;
666 struct pipe_inode_info
*i_pipe
;
667 struct block_device
*i_bdev
;
674 #ifdef CONFIG_FSNOTIFY
675 __u32 i_fsnotify_mask
; /* all events this inode cares about */
676 struct hlist_head i_fsnotify_marks
;
679 void *i_private
; /* fs or device private pointer */
682 static inline int inode_unhashed(struct inode
*inode
)
684 return hlist_unhashed(&inode
->i_hash
);
688 * inode->i_mutex nesting subclasses for the lock validator:
690 * 0: the object of the current VFS operation
694 * 4: second non-directory
695 * 5: second parent (when locking independent directories in rename)
697 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
698 * non-directories at once.
700 * The locking order between these classes is
701 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
703 enum inode_i_mutex_lock_class
713 void lock_two_nondirectories(struct inode
*, struct inode
*);
714 void unlock_two_nondirectories(struct inode
*, struct inode
*);
717 * NOTE: in a 32bit arch with a preemptable kernel and
718 * an UP compile the i_size_read/write must be atomic
719 * with respect to the local cpu (unlike with preempt disabled),
720 * but they don't need to be atomic with respect to other cpus like in
721 * true SMP (so they need either to either locally disable irq around
722 * the read or for example on x86 they can be still implemented as a
723 * cmpxchg8b without the need of the lock prefix). For SMP compiles
724 * and 64bit archs it makes no difference if preempt is enabled or not.
726 static inline loff_t
i_size_read(const struct inode
*inode
)
728 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
733 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
734 i_size
= inode
->i_size
;
735 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
737 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
741 i_size
= inode
->i_size
;
745 return inode
->i_size
;
750 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
751 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
752 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
754 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
756 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
758 write_seqcount_begin(&inode
->i_size_seqcount
);
759 inode
->i_size
= i_size
;
760 write_seqcount_end(&inode
->i_size_seqcount
);
762 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
764 inode
->i_size
= i_size
;
767 inode
->i_size
= i_size
;
771 /* Helper functions so that in most cases filesystems will
772 * not need to deal directly with kuid_t and kgid_t and can
773 * instead deal with the raw numeric values that are stored
776 static inline uid_t
i_uid_read(const struct inode
*inode
)
778 return from_kuid(&init_user_ns
, inode
->i_uid
);
781 static inline gid_t
i_gid_read(const struct inode
*inode
)
783 return from_kgid(&init_user_ns
, inode
->i_gid
);
786 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
788 inode
->i_uid
= make_kuid(&init_user_ns
, uid
);
791 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
793 inode
->i_gid
= make_kgid(&init_user_ns
, gid
);
796 static inline unsigned iminor(const struct inode
*inode
)
798 return MINOR(inode
->i_rdev
);
801 static inline unsigned imajor(const struct inode
*inode
)
803 return MAJOR(inode
->i_rdev
);
806 extern struct block_device
*I_BDEV(struct inode
*inode
);
809 rwlock_t lock
; /* protects pid, uid, euid fields */
810 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
811 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
812 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
813 int signum
; /* posix.1b rt signal to be delivered on IO */
817 * Track a single file's readahead state
819 struct file_ra_state
{
820 pgoff_t start
; /* where readahead started */
821 unsigned int size
; /* # of readahead pages */
822 unsigned int async_size
; /* do asynchronous readahead when
823 there are only # of pages ahead */
825 unsigned int ra_pages
; /* Maximum readahead window */
826 unsigned int mmap_miss
; /* Cache miss stat for mmap accesses */
827 loff_t prev_pos
; /* Cache last read() position */
831 * Check if @index falls in the readahead windows.
833 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
835 return (index
>= ra
->start
&&
836 index
< ra
->start
+ ra
->size
);
841 struct llist_node fu_llist
;
842 struct rcu_head fu_rcuhead
;
845 struct inode
*f_inode
; /* cached value */
846 const struct file_operations
*f_op
;
849 * Protects f_ep_links, f_flags.
850 * Must not be taken from IRQ context.
853 atomic_long_t f_count
;
854 unsigned int f_flags
;
856 struct mutex f_pos_lock
;
858 struct fown_struct f_owner
;
859 const struct cred
*f_cred
;
860 struct file_ra_state f_ra
;
863 #ifdef CONFIG_SECURITY
866 /* needed for tty driver, and maybe others */
870 /* Used by fs/eventpoll.c to link all the hooks to this file */
871 struct list_head f_ep_links
;
872 struct list_head f_tfile_llink
;
873 #endif /* #ifdef CONFIG_EPOLL */
874 struct address_space
*f_mapping
;
875 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
880 /* file identifier */
881 unsigned char f_handle
[0];
884 static inline struct file
*get_file(struct file
*f
)
886 atomic_long_inc(&f
->f_count
);
889 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
890 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
891 #define file_count(x) atomic_long_read(&(x)->f_count)
893 #define MAX_NON_LFS ((1UL<<31) - 1)
895 /* Page cache limit. The filesystems should put that into their s_maxbytes
896 limits, otherwise bad things can happen in VM. */
897 #if BITS_PER_LONG==32
898 #define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
899 #elif BITS_PER_LONG==64
900 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
905 #define FL_DELEG 4 /* NFSv4 delegation */
906 #define FL_ACCESS 8 /* not trying to lock, just looking */
907 #define FL_EXISTS 16 /* when unlocking, test for existence */
908 #define FL_LEASE 32 /* lease held on this file */
909 #define FL_CLOSE 64 /* unlock on close */
910 #define FL_SLEEP 128 /* A blocking lock */
911 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
912 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
913 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
914 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
917 * Special return value from posix_lock_file() and vfs_lock_file() for
918 * asynchronous locking.
920 #define FILE_LOCK_DEFERRED 1
922 /* legacy typedef, should eventually be removed */
923 typedef void *fl_owner_t
;
927 struct file_lock_operations
{
928 void (*fl_copy_lock
)(struct file_lock
*, struct file_lock
*);
929 void (*fl_release_private
)(struct file_lock
*);
932 struct lock_manager_operations
{
933 int (*lm_compare_owner
)(struct file_lock
*, struct file_lock
*);
934 unsigned long (*lm_owner_key
)(struct file_lock
*);
935 fl_owner_t (*lm_get_owner
)(fl_owner_t
);
936 void (*lm_put_owner
)(fl_owner_t
);
937 void (*lm_notify
)(struct file_lock
*); /* unblock callback */
938 int (*lm_grant
)(struct file_lock
*, int);
939 bool (*lm_break
)(struct file_lock
*);
940 int (*lm_change
)(struct file_lock
*, int, struct list_head
*);
941 void (*lm_setup
)(struct file_lock
*, void **);
944 struct lock_manager
{
945 struct list_head list
;
949 void locks_start_grace(struct net
*, struct lock_manager
*);
950 void locks_end_grace(struct lock_manager
*);
951 int locks_in_grace(struct net
*);
953 /* that will die - we need it for nfs_lock_info */
954 #include <linux/nfs_fs_i.h>
957 * struct file_lock represents a generic "file lock". It's used to represent
958 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
959 * note that the same struct is used to represent both a request for a lock and
960 * the lock itself, but the same object is never used for both.
962 * FIXME: should we create a separate "struct lock_request" to help distinguish
965 * The varous i_flctx lists are ordered by:
968 * 2) lock range start
971 * Obviously, the last two criteria only matter for POSIX locks.
974 struct file_lock
*fl_next
; /* singly linked list for this inode */
975 struct list_head fl_list
; /* link into file_lock_context */
976 struct hlist_node fl_link
; /* node in global lists */
977 struct list_head fl_block
; /* circular list of blocked processes */
979 unsigned int fl_flags
;
980 unsigned char fl_type
;
982 int fl_link_cpu
; /* what cpu's list is this on? */
983 struct pid
*fl_nspid
;
984 wait_queue_head_t fl_wait
;
985 struct file
*fl_file
;
989 struct fasync_struct
* fl_fasync
; /* for lease break notifications */
990 /* for lease breaks: */
991 unsigned long fl_break_time
;
992 unsigned long fl_downgrade_time
;
994 const struct file_lock_operations
*fl_ops
; /* Callbacks for filesystems */
995 const struct lock_manager_operations
*fl_lmops
; /* Callbacks for lockmanagers */
997 struct nfs_lock_info nfs_fl
;
998 struct nfs4_lock_info nfs4_fl
;
1000 struct list_head link
; /* link in AFS vnode's pending_locks list */
1001 int state
; /* state of grant or error if -ve */
1006 struct file_lock_context
{
1007 spinlock_t flc_lock
;
1008 struct list_head flc_flock
;
1009 struct list_head flc_posix
;
1010 struct list_head flc_lease
;
1013 /* The following constant reflects the upper bound of the file/locking space */
1015 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1016 #define OFFSET_MAX INT_LIMIT(loff_t)
1017 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1020 #include <linux/fcntl.h>
1022 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
1024 #ifdef CONFIG_FILE_LOCKING
1025 extern int fcntl_getlk(struct file
*, unsigned int, struct flock __user
*);
1026 extern int fcntl_setlk(unsigned int, struct file
*, unsigned int,
1027 struct flock __user
*);
1029 #if BITS_PER_LONG == 32
1030 extern int fcntl_getlk64(struct file
*, unsigned int, struct flock64 __user
*);
1031 extern int fcntl_setlk64(unsigned int, struct file
*, unsigned int,
1032 struct flock64 __user
*);
1035 extern int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
);
1036 extern int fcntl_getlease(struct file
*filp
);
1039 void locks_free_lock_context(struct file_lock_context
*ctx
);
1040 void locks_free_lock(struct file_lock
*fl
);
1041 extern void locks_init_lock(struct file_lock
*);
1042 extern struct file_lock
* locks_alloc_lock(void);
1043 extern void locks_copy_lock(struct file_lock
*, struct file_lock
*);
1044 extern void locks_copy_conflock(struct file_lock
*, struct file_lock
*);
1045 extern void locks_remove_posix(struct file
*, fl_owner_t
);
1046 extern void locks_remove_file(struct file
*);
1047 extern void locks_release_private(struct file_lock
*);
1048 extern void posix_test_lock(struct file
*, struct file_lock
*);
1049 extern int posix_lock_file(struct file
*, struct file_lock
*, struct file_lock
*);
1050 extern int posix_lock_inode_wait(struct inode
*, struct file_lock
*);
1051 extern int posix_unblock_lock(struct file_lock
*);
1052 extern int vfs_test_lock(struct file
*, struct file_lock
*);
1053 extern int vfs_lock_file(struct file
*, unsigned int, struct file_lock
*, struct file_lock
*);
1054 extern int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
);
1055 extern int flock_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
);
1056 extern int __break_lease(struct inode
*inode
, unsigned int flags
, unsigned int type
);
1057 extern void lease_get_mtime(struct inode
*, struct timespec
*time
);
1058 extern int generic_setlease(struct file
*, long, struct file_lock
**, void **priv
);
1059 extern int vfs_setlease(struct file
*, long, struct file_lock
**, void **);
1060 extern int lease_modify(struct file_lock
*, int, struct list_head
*);
1061 struct files_struct
;
1062 extern void show_fd_locks(struct seq_file
*f
,
1063 struct file
*filp
, struct files_struct
*files
);
1064 #else /* !CONFIG_FILE_LOCKING */
1065 static inline int fcntl_getlk(struct file
*file
, unsigned int cmd
,
1066 struct flock __user
*user
)
1071 static inline int fcntl_setlk(unsigned int fd
, struct file
*file
,
1072 unsigned int cmd
, struct flock __user
*user
)
1077 #if BITS_PER_LONG == 32
1078 static inline int fcntl_getlk64(struct file
*file
, unsigned int cmd
,
1079 struct flock64 __user
*user
)
1084 static inline int fcntl_setlk64(unsigned int fd
, struct file
*file
,
1085 unsigned int cmd
, struct flock64 __user
*user
)
1090 static inline int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
)
1095 static inline int fcntl_getlease(struct file
*filp
)
1101 locks_free_lock_context(struct file_lock_context
*ctx
)
1105 static inline void locks_init_lock(struct file_lock
*fl
)
1110 static inline void locks_copy_conflock(struct file_lock
*new, struct file_lock
*fl
)
1115 static inline void locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
1120 static inline void locks_remove_posix(struct file
*filp
, fl_owner_t owner
)
1125 static inline void locks_remove_file(struct file
*filp
)
1130 static inline void posix_test_lock(struct file
*filp
, struct file_lock
*fl
)
1135 static inline int posix_lock_file(struct file
*filp
, struct file_lock
*fl
,
1136 struct file_lock
*conflock
)
1141 static inline int posix_lock_inode_wait(struct inode
*inode
,
1142 struct file_lock
*fl
)
1147 static inline int posix_unblock_lock(struct file_lock
*waiter
)
1152 static inline int vfs_test_lock(struct file
*filp
, struct file_lock
*fl
)
1157 static inline int vfs_lock_file(struct file
*filp
, unsigned int cmd
,
1158 struct file_lock
*fl
, struct file_lock
*conf
)
1163 static inline int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
)
1168 static inline int flock_lock_inode_wait(struct inode
*inode
,
1169 struct file_lock
*request
)
1174 static inline int __break_lease(struct inode
*inode
, unsigned int mode
, unsigned int type
)
1179 static inline void lease_get_mtime(struct inode
*inode
, struct timespec
*time
)
1184 static inline int generic_setlease(struct file
*filp
, long arg
,
1185 struct file_lock
**flp
, void **priv
)
1190 static inline int vfs_setlease(struct file
*filp
, long arg
,
1191 struct file_lock
**lease
, void **priv
)
1196 static inline int lease_modify(struct file_lock
*fl
, int arg
,
1197 struct list_head
*dispose
)
1202 struct files_struct
;
1203 static inline void show_fd_locks(struct seq_file
*f
,
1204 struct file
*filp
, struct files_struct
*files
) {}
1205 #endif /* !CONFIG_FILE_LOCKING */
1207 static inline struct inode
*file_inode(const struct file
*f
)
1212 static inline int posix_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1214 return posix_lock_inode_wait(file_inode(filp
), fl
);
1217 static inline int flock_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1219 return flock_lock_inode_wait(file_inode(filp
), fl
);
1222 struct fasync_struct
{
1226 struct fasync_struct
*fa_next
; /* singly linked list */
1227 struct file
*fa_file
;
1228 struct rcu_head fa_rcu
;
1231 #define FASYNC_MAGIC 0x4601
1233 /* SMP safe fasync helpers: */
1234 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1235 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1236 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1237 extern struct fasync_struct
*fasync_alloc(void);
1238 extern void fasync_free(struct fasync_struct
*);
1240 /* can be called from interrupts */
1241 extern void kill_fasync(struct fasync_struct
**, int, int);
1243 extern void __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1244 extern void f_setown(struct file
*filp
, unsigned long arg
, int force
);
1245 extern void f_delown(struct file
*filp
);
1246 extern pid_t
f_getown(struct file
*filp
);
1247 extern int send_sigurg(struct fown_struct
*fown
);
1255 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1256 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1257 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1258 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1259 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1262 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1263 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1265 /* Possible states of 'frozen' field */
1267 SB_UNFROZEN
= 0, /* FS is unfrozen */
1268 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1269 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1270 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1271 * internal threads if needed) */
1272 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1275 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1278 /* Counters for counting writers at each level */
1279 struct percpu_counter counter
[SB_FREEZE_LEVELS
];
1280 wait_queue_head_t wait
; /* queue for waiting for
1281 writers / faults to finish */
1282 int frozen
; /* Is sb frozen? */
1283 wait_queue_head_t wait_unfrozen
; /* queue for waiting for
1285 #ifdef CONFIG_DEBUG_LOCK_ALLOC
1286 struct lockdep_map lock_map
[SB_FREEZE_LEVELS
];
1290 struct super_block
{
1291 struct list_head s_list
; /* Keep this first */
1292 dev_t s_dev
; /* search index; _not_ kdev_t */
1293 unsigned char s_blocksize_bits
;
1294 unsigned long s_blocksize
;
1295 loff_t s_maxbytes
; /* Max file size */
1296 struct file_system_type
*s_type
;
1297 const struct super_operations
*s_op
;
1298 const struct dquot_operations
*dq_op
;
1299 const struct quotactl_ops
*s_qcop
;
1300 const struct export_operations
*s_export_op
;
1301 unsigned long s_flags
;
1302 unsigned long s_iflags
; /* internal SB_I_* flags */
1303 unsigned long s_magic
;
1304 struct dentry
*s_root
;
1305 struct rw_semaphore s_umount
;
1308 #ifdef CONFIG_SECURITY
1311 const struct xattr_handler
**s_xattr
;
1313 struct list_head s_inodes
; /* all inodes */
1314 struct hlist_bl_head s_anon
; /* anonymous dentries for (nfs) exporting */
1315 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1316 struct block_device
*s_bdev
;
1317 struct backing_dev_info
*s_bdi
;
1318 struct mtd_info
*s_mtd
;
1319 struct hlist_node s_instances
;
1320 unsigned int s_quota_types
; /* Bitmask of supported quota types */
1321 struct quota_info s_dquot
; /* Diskquota specific options */
1323 struct sb_writers s_writers
;
1325 char s_id
[32]; /* Informational name */
1326 u8 s_uuid
[16]; /* UUID */
1328 void *s_fs_info
; /* Filesystem private info */
1329 unsigned int s_max_links
;
1332 /* Granularity of c/m/atime in ns.
1333 Cannot be worse than a second */
1337 * The next field is for VFS *only*. No filesystems have any business
1338 * even looking at it. You had been warned.
1340 struct mutex s_vfs_rename_mutex
; /* Kludge */
1343 * Filesystem subtype. If non-empty the filesystem type field
1344 * in /proc/mounts will be "type.subtype"
1349 * Saved mount options for lazy filesystems using
1350 * generic_show_options()
1352 char __rcu
*s_options
;
1353 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1356 * Saved pool identifier for cleancache (-1 means none)
1358 int cleancache_poolid
;
1360 struct shrinker s_shrink
; /* per-sb shrinker handle */
1362 /* Number of inodes with nlink == 0 but still referenced */
1363 atomic_long_t s_remove_count
;
1365 /* Being remounted read-only */
1366 int s_readonly_remount
;
1368 /* AIO completions deferred from interrupt context */
1369 struct workqueue_struct
*s_dio_done_wq
;
1370 struct hlist_head s_pins
;
1373 * Keep the lru lists last in the structure so they always sit on their
1374 * own individual cachelines.
1376 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp
;
1377 struct list_lru s_inode_lru ____cacheline_aligned_in_smp
;
1378 struct rcu_head rcu
;
1381 * Indicates how deep in a filesystem stack this SB is
1386 extern struct timespec
current_fs_time(struct super_block
*sb
);
1389 * Snapshotting support.
1392 void __sb_end_write(struct super_block
*sb
, int level
);
1393 int __sb_start_write(struct super_block
*sb
, int level
, bool wait
);
1396 * sb_end_write - drop write access to a superblock
1397 * @sb: the super we wrote to
1399 * Decrement number of writers to the filesystem. Wake up possible waiters
1400 * wanting to freeze the filesystem.
1402 static inline void sb_end_write(struct super_block
*sb
)
1404 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1408 * sb_end_pagefault - drop write access to a superblock from a page fault
1409 * @sb: the super we wrote to
1411 * Decrement number of processes handling write page fault to the filesystem.
1412 * Wake up possible waiters wanting to freeze the filesystem.
1414 static inline void sb_end_pagefault(struct super_block
*sb
)
1416 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1420 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1421 * @sb: the super we wrote to
1423 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1424 * waiters wanting to freeze the filesystem.
1426 static inline void sb_end_intwrite(struct super_block
*sb
)
1428 __sb_end_write(sb
, SB_FREEZE_FS
);
1432 * sb_start_write - get write access to a superblock
1433 * @sb: the super we write to
1435 * When a process wants to write data or metadata to a file system (i.e. dirty
1436 * a page or an inode), it should embed the operation in a sb_start_write() -
1437 * sb_end_write() pair to get exclusion against file system freezing. This
1438 * function increments number of writers preventing freezing. If the file
1439 * system is already frozen, the function waits until the file system is
1442 * Since freeze protection behaves as a lock, users have to preserve
1443 * ordering of freeze protection and other filesystem locks. Generally,
1444 * freeze protection should be the outermost lock. In particular, we have:
1447 * -> i_mutex (write path, truncate, directory ops, ...)
1448 * -> s_umount (freeze_super, thaw_super)
1450 static inline void sb_start_write(struct super_block
*sb
)
1452 __sb_start_write(sb
, SB_FREEZE_WRITE
, true);
1455 static inline int sb_start_write_trylock(struct super_block
*sb
)
1457 return __sb_start_write(sb
, SB_FREEZE_WRITE
, false);
1461 * sb_start_pagefault - get write access to a superblock from a page fault
1462 * @sb: the super we write to
1464 * When a process starts handling write page fault, it should embed the
1465 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1466 * exclusion against file system freezing. This is needed since the page fault
1467 * is going to dirty a page. This function increments number of running page
1468 * faults preventing freezing. If the file system is already frozen, the
1469 * function waits until the file system is thawed.
1471 * Since page fault freeze protection behaves as a lock, users have to preserve
1472 * ordering of freeze protection and other filesystem locks. It is advised to
1473 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1474 * handling code implies lock dependency:
1477 * -> sb_start_pagefault
1479 static inline void sb_start_pagefault(struct super_block
*sb
)
1481 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
, true);
1485 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1486 * @sb: the super we write to
1488 * This is the third level of protection against filesystem freezing. It is
1489 * free for use by a filesystem. The only requirement is that it must rank
1490 * below sb_start_pagefault.
1492 * For example filesystem can call sb_start_intwrite() when starting a
1493 * transaction which somewhat eases handling of freezing for internal sources
1494 * of filesystem changes (internal fs threads, discarding preallocation on file
1497 static inline void sb_start_intwrite(struct super_block
*sb
)
1499 __sb_start_write(sb
, SB_FREEZE_FS
, true);
1503 extern bool inode_owner_or_capable(const struct inode
*inode
);
1506 * VFS helper functions..
1508 extern int vfs_create(struct inode
*, struct dentry
*, umode_t
, bool);
1509 extern int vfs_mkdir(struct inode
*, struct dentry
*, umode_t
);
1510 extern int vfs_mknod(struct inode
*, struct dentry
*, umode_t
, dev_t
);
1511 extern int vfs_symlink(struct inode
*, struct dentry
*, const char *);
1512 extern int vfs_link(struct dentry
*, struct inode
*, struct dentry
*, struct inode
**);
1513 extern int vfs_rmdir(struct inode
*, struct dentry
*);
1514 extern int vfs_unlink(struct inode
*, struct dentry
*, struct inode
**);
1515 extern int vfs_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*, struct inode
**, unsigned int);
1516 extern int vfs_whiteout(struct inode
*, struct dentry
*);
1519 * VFS dentry helper functions.
1521 extern void dentry_unhash(struct dentry
*dentry
);
1524 * VFS file helper functions.
1526 extern void inode_init_owner(struct inode
*inode
, const struct inode
*dir
,
1529 * VFS FS_IOC_FIEMAP helper definitions.
1531 struct fiemap_extent_info
{
1532 unsigned int fi_flags
; /* Flags as passed from user */
1533 unsigned int fi_extents_mapped
; /* Number of mapped extents */
1534 unsigned int fi_extents_max
; /* Size of fiemap_extent array */
1535 struct fiemap_extent __user
*fi_extents_start
; /* Start of
1536 fiemap_extent array */
1538 int fiemap_fill_next_extent(struct fiemap_extent_info
*info
, u64 logical
,
1539 u64 phys
, u64 len
, u32 flags
);
1540 int fiemap_check_flags(struct fiemap_extent_info
*fieinfo
, u32 fs_flags
);
1545 * NOTE! These match bits 12..15 of stat.st_mode
1546 * (ie "(i_mode >> 12) & 15").
1548 #define DT_UNKNOWN 0
1559 * This is the "filldir" function type, used by readdir() to let
1560 * the kernel specify what kind of dirent layout it wants to have.
1561 * This allows the kernel to read directories into kernel space or
1562 * to have different dirent layouts depending on the binary type.
1565 typedef int (*filldir_t
)(struct dir_context
*, const char *, int, loff_t
, u64
,
1568 struct dir_context
{
1569 const filldir_t actor
;
1573 struct block_device_operations
;
1575 /* These macros are for out of kernel modules to test that
1576 * the kernel supports the unlocked_ioctl and compat_ioctl
1577 * fields in struct file_operations. */
1578 #define HAVE_COMPAT_IOCTL 1
1579 #define HAVE_UNLOCKED_IOCTL 1
1582 * These flags let !MMU mmap() govern direct device mapping vs immediate
1583 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1585 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1586 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1587 * NOMMU_MAP_READ: Can be mapped for reading
1588 * NOMMU_MAP_WRITE: Can be mapped for writing
1589 * NOMMU_MAP_EXEC: Can be mapped for execution
1591 #define NOMMU_MAP_COPY 0x00000001
1592 #define NOMMU_MAP_DIRECT 0x00000008
1593 #define NOMMU_MAP_READ VM_MAYREAD
1594 #define NOMMU_MAP_WRITE VM_MAYWRITE
1595 #define NOMMU_MAP_EXEC VM_MAYEXEC
1597 #define NOMMU_VMFLAGS \
1598 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1603 struct file_operations
{
1604 struct module
*owner
;
1605 loff_t (*llseek
) (struct file
*, loff_t
, int);
1606 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
1607 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
1608 ssize_t (*read_iter
) (struct kiocb
*, struct iov_iter
*);
1609 ssize_t (*write_iter
) (struct kiocb
*, struct iov_iter
*);
1610 int (*iterate
) (struct file
*, struct dir_context
*);
1611 unsigned int (*poll
) (struct file
*, struct poll_table_struct
*);
1612 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
1613 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
1614 int (*mmap
) (struct file
*, struct vm_area_struct
*);
1615 int (*open
) (struct inode
*, struct file
*);
1616 int (*flush
) (struct file
*, fl_owner_t id
);
1617 int (*release
) (struct inode
*, struct file
*);
1618 int (*fsync
) (struct file
*, loff_t
, loff_t
, int datasync
);
1619 int (*aio_fsync
) (struct kiocb
*, int datasync
);
1620 int (*fasync
) (int, struct file
*, int);
1621 int (*lock
) (struct file
*, int, struct file_lock
*);
1622 ssize_t (*sendpage
) (struct file
*, struct page
*, int, size_t, loff_t
*, int);
1623 unsigned long (*get_unmapped_area
)(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1624 int (*check_flags
)(int);
1625 int (*flock
) (struct file
*, int, struct file_lock
*);
1626 ssize_t (*splice_write
)(struct pipe_inode_info
*, struct file
*, loff_t
*, size_t, unsigned int);
1627 ssize_t (*splice_read
)(struct file
*, loff_t
*, struct pipe_inode_info
*, size_t, unsigned int);
1628 int (*setlease
)(struct file
*, long, struct file_lock
**, void **);
1629 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
1631 void (*show_fdinfo
)(struct seq_file
*m
, struct file
*f
);
1633 unsigned (*mmap_capabilities
)(struct file
*);
1637 struct inode_operations
{
1638 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*, unsigned int);
1639 const char * (*follow_link
) (struct dentry
*, void **);
1640 int (*permission
) (struct inode
*, int);
1641 struct posix_acl
* (*get_acl
)(struct inode
*, int);
1643 int (*readlink
) (struct dentry
*, char __user
*,int);
1644 void (*put_link
) (struct inode
*, void *);
1646 int (*create
) (struct inode
*,struct dentry
*, umode_t
, bool);
1647 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
1648 int (*unlink
) (struct inode
*,struct dentry
*);
1649 int (*symlink
) (struct inode
*,struct dentry
*,const char *);
1650 int (*mkdir
) (struct inode
*,struct dentry
*,umode_t
);
1651 int (*rmdir
) (struct inode
*,struct dentry
*);
1652 int (*mknod
) (struct inode
*,struct dentry
*,umode_t
,dev_t
);
1653 int (*rename
) (struct inode
*, struct dentry
*,
1654 struct inode
*, struct dentry
*);
1655 int (*rename2
) (struct inode
*, struct dentry
*,
1656 struct inode
*, struct dentry
*, unsigned int);
1657 int (*setattr
) (struct dentry
*, struct iattr
*);
1658 int (*getattr
) (struct vfsmount
*mnt
, struct dentry
*, struct kstat
*);
1659 int (*setxattr
) (struct dentry
*, const char *,const void *,size_t,int);
1660 ssize_t (*getxattr
) (struct dentry
*, const char *, void *, size_t);
1661 ssize_t (*listxattr
) (struct dentry
*, char *, size_t);
1662 int (*removexattr
) (struct dentry
*, const char *);
1663 int (*fiemap
)(struct inode
*, struct fiemap_extent_info
*, u64 start
,
1665 int (*update_time
)(struct inode
*, struct timespec
*, int);
1666 int (*atomic_open
)(struct inode
*, struct dentry
*,
1667 struct file
*, unsigned open_flag
,
1668 umode_t create_mode
, int *opened
);
1669 int (*tmpfile
) (struct inode
*, struct dentry
*, umode_t
);
1670 int (*set_acl
)(struct inode
*, struct posix_acl
*, int);
1672 /* WARNING: probably going away soon, do not use! */
1673 } ____cacheline_aligned
;
1675 ssize_t
rw_copy_check_uvector(int type
, const struct iovec __user
* uvector
,
1676 unsigned long nr_segs
, unsigned long fast_segs
,
1677 struct iovec
*fast_pointer
,
1678 struct iovec
**ret_pointer
);
1680 extern ssize_t
__vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1681 extern ssize_t
__vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1682 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1683 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1684 extern ssize_t
vfs_readv(struct file
*, const struct iovec __user
*,
1685 unsigned long, loff_t
*);
1686 extern ssize_t
vfs_writev(struct file
*, const struct iovec __user
*,
1687 unsigned long, loff_t
*);
1689 struct super_operations
{
1690 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
1691 void (*destroy_inode
)(struct inode
*);
1693 void (*dirty_inode
) (struct inode
*, int flags
);
1694 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
1695 int (*drop_inode
) (struct inode
*);
1696 void (*evict_inode
) (struct inode
*);
1697 void (*put_super
) (struct super_block
*);
1698 int (*sync_fs
)(struct super_block
*sb
, int wait
);
1699 int (*freeze_super
) (struct super_block
*);
1700 int (*freeze_fs
) (struct super_block
*);
1701 int (*thaw_super
) (struct super_block
*);
1702 int (*unfreeze_fs
) (struct super_block
*);
1703 int (*statfs
) (struct dentry
*, struct kstatfs
*);
1704 int (*remount_fs
) (struct super_block
*, int *, char *);
1705 void (*umount_begin
) (struct super_block
*);
1707 int (*show_options
)(struct seq_file
*, struct dentry
*);
1708 int (*show_devname
)(struct seq_file
*, struct dentry
*);
1709 int (*show_path
)(struct seq_file
*, struct dentry
*);
1710 int (*show_stats
)(struct seq_file
*, struct dentry
*);
1712 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
1713 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
1714 struct dquot
**(*get_dquots
)(struct inode
*);
1716 int (*bdev_try_to_free_page
)(struct super_block
*, struct page
*, gfp_t
);
1717 long (*nr_cached_objects
)(struct super_block
*,
1718 struct shrink_control
*);
1719 long (*free_cached_objects
)(struct super_block
*,
1720 struct shrink_control
*);
1724 * Inode flags - they have no relation to superblock flags now
1726 #define S_SYNC 1 /* Writes are synced at once */
1727 #define S_NOATIME 2 /* Do not update access times */
1728 #define S_APPEND 4 /* Append-only file */
1729 #define S_IMMUTABLE 8 /* Immutable file */
1730 #define S_DEAD 16 /* removed, but still open directory */
1731 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1732 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1733 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1734 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1735 #define S_PRIVATE 512 /* Inode is fs-internal */
1736 #define S_IMA 1024 /* Inode has an associated IMA struct */
1737 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1738 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1739 #ifdef CONFIG_FS_DAX
1740 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1742 #define S_DAX 0 /* Make all the DAX code disappear */
1746 * Note that nosuid etc flags are inode-specific: setting some file-system
1747 * flags just means all the inodes inherit those flags by default. It might be
1748 * possible to override it selectively if you really wanted to with some
1749 * ioctl() that is not currently implemented.
1751 * Exception: MS_RDONLY is always applied to the entire file system.
1753 * Unfortunately, it is possible to change a filesystems flags with it mounted
1754 * with files in use. This means that all of the inodes will not have their
1755 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1756 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1758 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1760 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1761 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1762 ((inode)->i_flags & S_SYNC))
1763 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1764 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1765 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1766 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1767 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1769 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1770 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1771 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1772 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1774 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1775 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1776 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1777 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1778 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1779 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1780 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1781 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1783 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1784 (inode)->i_rdev == WHITEOUT_DEV)
1787 * Inode state bits. Protected by inode->i_lock
1789 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1790 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1792 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1793 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1794 * various stages of removing an inode.
1796 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1798 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1799 * fdatasync(). i_atime is the usual cause.
1800 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1801 * these changes separately from I_DIRTY_SYNC so that we
1802 * don't have to write inode on fdatasync() when only
1803 * mtime has changed in it.
1804 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1805 * I_NEW Serves as both a mutex and completion notification.
1806 * New inodes set I_NEW. If two processes both create
1807 * the same inode, one of them will release its inode and
1808 * wait for I_NEW to be released before returning.
1809 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1810 * also cause waiting on I_NEW, without I_NEW actually
1811 * being set. find_inode() uses this to prevent returning
1812 * nearly-dead inodes.
1813 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1814 * is zero. I_FREEING must be set when I_WILL_FREE is
1816 * I_FREEING Set when inode is about to be freed but still has dirty
1817 * pages or buffers attached or the inode itself is still
1819 * I_CLEAR Added by clear_inode(). In this state the inode is
1820 * clean and can be destroyed. Inode keeps I_FREEING.
1822 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1823 * prohibited for many purposes. iget() must wait for
1824 * the inode to be completely released, then create it
1825 * anew. Other functions will just ignore such inodes,
1826 * if appropriate. I_NEW is used for waiting.
1828 * I_SYNC Writeback of inode is running. The bit is set during
1829 * data writeback, and cleared with a wakeup on the bit
1830 * address once it is done. The bit is also used to pin
1831 * the inode in memory for flusher thread.
1833 * I_REFERENCED Marks the inode as recently references on the LRU list.
1835 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1837 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1838 * synchronize competing switching instances and to tell
1839 * wb stat updates to grab mapping->tree_lock. See
1840 * inode_switch_wb_work_fn() for details.
1842 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1844 #define I_DIRTY_SYNC (1 << 0)
1845 #define I_DIRTY_DATASYNC (1 << 1)
1846 #define I_DIRTY_PAGES (1 << 2)
1848 #define I_NEW (1 << __I_NEW)
1849 #define I_WILL_FREE (1 << 4)
1850 #define I_FREEING (1 << 5)
1851 #define I_CLEAR (1 << 6)
1853 #define I_SYNC (1 << __I_SYNC)
1854 #define I_REFERENCED (1 << 8)
1855 #define __I_DIO_WAKEUP 9
1856 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1857 #define I_LINKABLE (1 << 10)
1858 #define I_DIRTY_TIME (1 << 11)
1859 #define __I_DIRTY_TIME_EXPIRED 12
1860 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1861 #define I_WB_SWITCH (1 << 13)
1863 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1864 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1866 extern void __mark_inode_dirty(struct inode
*, int);
1867 static inline void mark_inode_dirty(struct inode
*inode
)
1869 __mark_inode_dirty(inode
, I_DIRTY
);
1872 static inline void mark_inode_dirty_sync(struct inode
*inode
)
1874 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
1877 extern void inc_nlink(struct inode
*inode
);
1878 extern void drop_nlink(struct inode
*inode
);
1879 extern void clear_nlink(struct inode
*inode
);
1880 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
1882 static inline void inode_inc_link_count(struct inode
*inode
)
1885 mark_inode_dirty(inode
);
1888 static inline void inode_dec_link_count(struct inode
*inode
)
1891 mark_inode_dirty(inode
);
1895 * inode_inc_iversion - increments i_version
1896 * @inode: inode that need to be updated
1898 * Every time the inode is modified, the i_version field will be incremented.
1899 * The filesystem has to be mounted with i_version flag
1902 static inline void inode_inc_iversion(struct inode
*inode
)
1904 spin_lock(&inode
->i_lock
);
1906 spin_unlock(&inode
->i_lock
);
1909 enum file_time_flags
{
1916 extern bool atime_needs_update(const struct path
*, struct inode
*);
1917 extern void touch_atime(const struct path
*);
1918 static inline void file_accessed(struct file
*file
)
1920 if (!(file
->f_flags
& O_NOATIME
))
1921 touch_atime(&file
->f_path
);
1924 int sync_inode(struct inode
*inode
, struct writeback_control
*wbc
);
1925 int sync_inode_metadata(struct inode
*inode
, int wait
);
1927 struct file_system_type
{
1930 #define FS_REQUIRES_DEV 1
1931 #define FS_BINARY_MOUNTDATA 2
1932 #define FS_HAS_SUBTYPE 4
1933 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
1934 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
1935 #define FS_USERNS_VISIBLE 32 /* FS must already be visible */
1936 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
1937 struct dentry
*(*mount
) (struct file_system_type
*, int,
1938 const char *, void *);
1939 void (*kill_sb
) (struct super_block
*);
1940 struct module
*owner
;
1941 struct file_system_type
* next
;
1942 struct hlist_head fs_supers
;
1944 struct lock_class_key s_lock_key
;
1945 struct lock_class_key s_umount_key
;
1946 struct lock_class_key s_vfs_rename_key
;
1947 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
1949 struct lock_class_key i_lock_key
;
1950 struct lock_class_key i_mutex_key
;
1951 struct lock_class_key i_mutex_dir_key
;
1954 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1956 extern struct dentry
*mount_ns(struct file_system_type
*fs_type
, int flags
,
1957 void *data
, int (*fill_super
)(struct super_block
*, void *, int));
1958 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
1959 int flags
, const char *dev_name
, void *data
,
1960 int (*fill_super
)(struct super_block
*, void *, int));
1961 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
1962 int flags
, void *data
,
1963 int (*fill_super
)(struct super_block
*, void *, int));
1964 extern struct dentry
*mount_nodev(struct file_system_type
*fs_type
,
1965 int flags
, void *data
,
1966 int (*fill_super
)(struct super_block
*, void *, int));
1967 extern struct dentry
*mount_subtree(struct vfsmount
*mnt
, const char *path
);
1968 void generic_shutdown_super(struct super_block
*sb
);
1969 void kill_block_super(struct super_block
*sb
);
1970 void kill_anon_super(struct super_block
*sb
);
1971 void kill_litter_super(struct super_block
*sb
);
1972 void deactivate_super(struct super_block
*sb
);
1973 void deactivate_locked_super(struct super_block
*sb
);
1974 int set_anon_super(struct super_block
*s
, void *data
);
1975 int get_anon_bdev(dev_t
*);
1976 void free_anon_bdev(dev_t
);
1977 struct super_block
*sget(struct file_system_type
*type
,
1978 int (*test
)(struct super_block
*,void *),
1979 int (*set
)(struct super_block
*,void *),
1980 int flags
, void *data
);
1981 extern struct dentry
*mount_pseudo(struct file_system_type
*, char *,
1982 const struct super_operations
*ops
,
1983 const struct dentry_operations
*dops
,
1986 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1987 #define fops_get(fops) \
1988 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1989 #define fops_put(fops) \
1990 do { if (fops) module_put((fops)->owner); } while(0)
1992 * This one is to be used *ONLY* from ->open() instances.
1993 * fops must be non-NULL, pinned down *and* module dependencies
1994 * should be sufficient to pin the caller down as well.
1996 #define replace_fops(f, fops) \
1998 struct file *__file = (f); \
1999 fops_put(__file->f_op); \
2000 BUG_ON(!(__file->f_op = (fops))); \
2003 extern int register_filesystem(struct file_system_type
*);
2004 extern int unregister_filesystem(struct file_system_type
*);
2005 extern struct vfsmount
*kern_mount_data(struct file_system_type
*, void *data
);
2006 #define kern_mount(type) kern_mount_data(type, NULL)
2007 extern void kern_unmount(struct vfsmount
*mnt
);
2008 extern int may_umount_tree(struct vfsmount
*);
2009 extern int may_umount(struct vfsmount
*);
2010 extern long do_mount(const char *, const char __user
*,
2011 const char *, unsigned long, void *);
2012 extern struct vfsmount
*collect_mounts(struct path
*);
2013 extern void drop_collected_mounts(struct vfsmount
*);
2014 extern int iterate_mounts(int (*)(struct vfsmount
*, void *), void *,
2016 extern int vfs_statfs(struct path
*, struct kstatfs
*);
2017 extern int user_statfs(const char __user
*, struct kstatfs
*);
2018 extern int fd_statfs(int, struct kstatfs
*);
2019 extern int vfs_ustat(dev_t
, struct kstatfs
*);
2020 extern int freeze_super(struct super_block
*super
);
2021 extern int thaw_super(struct super_block
*super
);
2022 extern bool our_mnt(struct vfsmount
*mnt
);
2024 extern int current_umask(void);
2026 extern void ihold(struct inode
* inode
);
2027 extern void iput(struct inode
*);
2028 extern int generic_update_time(struct inode
*, struct timespec
*, int);
2031 extern struct kobject
*fs_kobj
;
2033 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
2035 #define FLOCK_VERIFY_READ 1
2036 #define FLOCK_VERIFY_WRITE 2
2038 #ifdef CONFIG_FILE_LOCKING
2039 extern int locks_mandatory_locked(struct file
*);
2040 extern int locks_mandatory_area(int, struct inode
*, struct file
*, loff_t
, size_t);
2043 * Candidates for mandatory locking have the setgid bit set
2044 * but no group execute bit - an otherwise meaningless combination.
2047 static inline int __mandatory_lock(struct inode
*ino
)
2049 return (ino
->i_mode
& (S_ISGID
| S_IXGRP
)) == S_ISGID
;
2053 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2054 * otherwise these will be advisory locks
2057 static inline int mandatory_lock(struct inode
*ino
)
2059 return IS_MANDLOCK(ino
) && __mandatory_lock(ino
);
2062 static inline int locks_verify_locked(struct file
*file
)
2064 if (mandatory_lock(file_inode(file
)))
2065 return locks_mandatory_locked(file
);
2069 static inline int locks_verify_truncate(struct inode
*inode
,
2073 if (inode
->i_flctx
&& mandatory_lock(inode
))
2074 return locks_mandatory_area(
2075 FLOCK_VERIFY_WRITE
, inode
, filp
,
2076 size
< inode
->i_size
? size
: inode
->i_size
,
2077 (size
< inode
->i_size
? inode
->i_size
- size
2078 : size
- inode
->i_size
)
2083 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2086 * Since this check is lockless, we must ensure that any refcounts
2087 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2088 * could end up racing with tasks trying to set a new lease on this
2092 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2093 return __break_lease(inode
, mode
, FL_LEASE
);
2097 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2100 * Since this check is lockless, we must ensure that any refcounts
2101 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2102 * could end up racing with tasks trying to set a new lease on this
2106 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2107 return __break_lease(inode
, mode
, FL_DELEG
);
2111 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2115 ret
= break_deleg(inode
, O_WRONLY
|O_NONBLOCK
);
2116 if (ret
== -EWOULDBLOCK
&& delegated_inode
) {
2117 *delegated_inode
= inode
;
2123 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2127 ret
= break_deleg(*delegated_inode
, O_WRONLY
);
2128 iput(*delegated_inode
);
2129 *delegated_inode
= NULL
;
2133 static inline int break_layout(struct inode
*inode
, bool wait
)
2136 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2137 return __break_lease(inode
,
2138 wait
? O_WRONLY
: O_WRONLY
| O_NONBLOCK
,
2143 #else /* !CONFIG_FILE_LOCKING */
2144 static inline int locks_mandatory_locked(struct file
*file
)
2149 static inline int locks_mandatory_area(int rw
, struct inode
*inode
,
2150 struct file
*filp
, loff_t offset
,
2156 static inline int __mandatory_lock(struct inode
*inode
)
2161 static inline int mandatory_lock(struct inode
*inode
)
2166 static inline int locks_verify_locked(struct file
*file
)
2171 static inline int locks_verify_truncate(struct inode
*inode
, struct file
*filp
,
2177 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2182 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2187 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2192 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2198 static inline int break_layout(struct inode
*inode
, bool wait
)
2203 #endif /* CONFIG_FILE_LOCKING */
2208 const char *name
; /* pointer to actual string */
2209 const __user
char *uptr
; /* original userland pointer */
2210 struct audit_names
*aname
;
2215 extern long vfs_truncate(struct path
*, loff_t
);
2216 extern int do_truncate(struct dentry
*, loff_t start
, unsigned int time_attrs
,
2218 extern int vfs_fallocate(struct file
*file
, int mode
, loff_t offset
,
2220 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
2222 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
2223 extern struct file
*filp_open(const char *, int, umode_t
);
2224 extern struct file
*file_open_root(struct dentry
*, struct vfsmount
*,
2226 extern struct file
* dentry_open(const struct path
*, int, const struct cred
*);
2227 extern int filp_close(struct file
*, fl_owner_t id
);
2229 extern struct filename
*getname_flags(const char __user
*, int, int *);
2230 extern struct filename
*getname(const char __user
*);
2231 extern struct filename
*getname_kernel(const char *);
2232 extern void putname(struct filename
*name
);
2238 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
2239 int (*open
)(struct inode
*, struct file
*),
2241 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
2245 extern int ioctl_preallocate(struct file
*filp
, void __user
*argp
);
2248 extern void __init
vfs_caches_init_early(void);
2249 extern void __init
vfs_caches_init(void);
2251 extern struct kmem_cache
*names_cachep
;
2253 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2254 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2257 extern int register_blkdev(unsigned int, const char *);
2258 extern void unregister_blkdev(unsigned int, const char *);
2259 extern struct block_device
*bdget(dev_t
);
2260 extern struct block_device
*bdgrab(struct block_device
*bdev
);
2261 extern void bd_set_size(struct block_device
*, loff_t size
);
2262 extern void bd_forget(struct inode
*inode
);
2263 extern void bdput(struct block_device
*);
2264 extern void invalidate_bdev(struct block_device
*);
2265 extern void iterate_bdevs(void (*)(struct block_device
*, void *), void *);
2266 extern int sync_blockdev(struct block_device
*bdev
);
2267 extern void kill_bdev(struct block_device
*);
2268 extern struct super_block
*freeze_bdev(struct block_device
*);
2269 extern void emergency_thaw_all(void);
2270 extern int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
);
2271 extern int fsync_bdev(struct block_device
*);
2273 extern struct super_block
*blockdev_superblock
;
2275 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2277 return sb
== blockdev_superblock
;
2280 static inline void bd_forget(struct inode
*inode
) {}
2281 static inline int sync_blockdev(struct block_device
*bdev
) { return 0; }
2282 static inline void kill_bdev(struct block_device
*bdev
) {}
2283 static inline void invalidate_bdev(struct block_device
*bdev
) {}
2285 static inline struct super_block
*freeze_bdev(struct block_device
*sb
)
2290 static inline int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
)
2295 static inline void iterate_bdevs(void (*f
)(struct block_device
*, void *), void *arg
)
2299 static inline int sb_is_blkdev_sb(struct super_block
*sb
)
2304 extern int sync_filesystem(struct super_block
*);
2305 extern const struct file_operations def_blk_fops
;
2306 extern const struct file_operations def_chr_fops
;
2308 extern int ioctl_by_bdev(struct block_device
*, unsigned, unsigned long);
2309 extern int blkdev_ioctl(struct block_device
*, fmode_t
, unsigned, unsigned long);
2310 extern long compat_blkdev_ioctl(struct file
*, unsigned, unsigned long);
2311 extern int blkdev_get(struct block_device
*bdev
, fmode_t mode
, void *holder
);
2312 extern struct block_device
*blkdev_get_by_path(const char *path
, fmode_t mode
,
2314 extern struct block_device
*blkdev_get_by_dev(dev_t dev
, fmode_t mode
,
2316 extern void blkdev_put(struct block_device
*bdev
, fmode_t mode
);
2317 extern int __blkdev_reread_part(struct block_device
*bdev
);
2318 extern int blkdev_reread_part(struct block_device
*bdev
);
2321 extern int bd_link_disk_holder(struct block_device
*bdev
, struct gendisk
*disk
);
2322 extern void bd_unlink_disk_holder(struct block_device
*bdev
,
2323 struct gendisk
*disk
);
2325 static inline int bd_link_disk_holder(struct block_device
*bdev
,
2326 struct gendisk
*disk
)
2330 static inline void bd_unlink_disk_holder(struct block_device
*bdev
,
2331 struct gendisk
*disk
)
2338 #define CHRDEV_MAJOR_HASH_SIZE 255
2339 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
2340 extern int register_chrdev_region(dev_t
, unsigned, const char *);
2341 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
2342 unsigned int count
, const char *name
,
2343 const struct file_operations
*fops
);
2344 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
2345 unsigned int count
, const char *name
);
2346 extern void unregister_chrdev_region(dev_t
, unsigned);
2347 extern void chrdev_show(struct seq_file
*,off_t
);
2349 static inline int register_chrdev(unsigned int major
, const char *name
,
2350 const struct file_operations
*fops
)
2352 return __register_chrdev(major
, 0, 256, name
, fops
);
2355 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2357 __unregister_chrdev(major
, 0, 256, name
);
2360 /* fs/block_dev.c */
2361 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2362 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2365 #define BLKDEV_MAJOR_HASH_SIZE 255
2366 extern const char *__bdevname(dev_t
, char *buffer
);
2367 extern const char *bdevname(struct block_device
*bdev
, char *buffer
);
2368 extern struct block_device
*lookup_bdev(const char *);
2369 extern void blkdev_show(struct seq_file
*,off_t
);
2372 #define BLKDEV_MAJOR_HASH_SIZE 0
2375 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2377 /* Invalid inode operations -- fs/bad_inode.c */
2378 extern void make_bad_inode(struct inode
*);
2379 extern int is_bad_inode(struct inode
*);
2383 * return READ, READA, or WRITE
2385 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2388 * return data direction, READ or WRITE
2390 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2392 extern void check_disk_size_change(struct gendisk
*disk
,
2393 struct block_device
*bdev
);
2394 extern int revalidate_disk(struct gendisk
*);
2395 extern int check_disk_change(struct block_device
*);
2396 extern int __invalidate_device(struct block_device
*, bool);
2397 extern int invalidate_partition(struct gendisk
*, int);
2399 unsigned long invalidate_mapping_pages(struct address_space
*mapping
,
2400 pgoff_t start
, pgoff_t end
);
2402 static inline void invalidate_remote_inode(struct inode
*inode
)
2404 if (S_ISREG(inode
->i_mode
) || S_ISDIR(inode
->i_mode
) ||
2405 S_ISLNK(inode
->i_mode
))
2406 invalidate_mapping_pages(inode
->i_mapping
, 0, -1);
2408 extern int invalidate_inode_pages2(struct address_space
*mapping
);
2409 extern int invalidate_inode_pages2_range(struct address_space
*mapping
,
2410 pgoff_t start
, pgoff_t end
);
2411 extern int write_inode_now(struct inode
*, int);
2412 extern int filemap_fdatawrite(struct address_space
*);
2413 extern int filemap_flush(struct address_space
*);
2414 extern int filemap_fdatawait(struct address_space
*);
2415 extern int filemap_fdatawait_range(struct address_space
*, loff_t lstart
,
2417 extern int filemap_write_and_wait(struct address_space
*mapping
);
2418 extern int filemap_write_and_wait_range(struct address_space
*mapping
,
2419 loff_t lstart
, loff_t lend
);
2420 extern int __filemap_fdatawrite_range(struct address_space
*mapping
,
2421 loff_t start
, loff_t end
, int sync_mode
);
2422 extern int filemap_fdatawrite_range(struct address_space
*mapping
,
2423 loff_t start
, loff_t end
);
2425 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
2427 extern int vfs_fsync(struct file
*file
, int datasync
);
2428 static inline int generic_write_sync(struct file
*file
, loff_t pos
, loff_t count
)
2430 if (!(file
->f_flags
& O_DSYNC
) && !IS_SYNC(file
->f_mapping
->host
))
2432 return vfs_fsync_range(file
, pos
, pos
+ count
- 1,
2433 (file
->f_flags
& __O_SYNC
) ? 0 : 1);
2435 extern void emergency_sync(void);
2436 extern void emergency_remount(void);
2438 extern sector_t
bmap(struct inode
*, sector_t
);
2440 extern int notify_change(struct dentry
*, struct iattr
*, struct inode
**);
2441 extern int inode_permission(struct inode
*, int);
2442 extern int __inode_permission(struct inode
*, int);
2443 extern int generic_permission(struct inode
*, int);
2444 extern int __check_sticky(struct inode
*dir
, struct inode
*inode
);
2446 static inline bool execute_ok(struct inode
*inode
)
2448 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
2451 static inline void file_start_write(struct file
*file
)
2453 if (!S_ISREG(file_inode(file
)->i_mode
))
2455 __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, true);
2458 static inline bool file_start_write_trylock(struct file
*file
)
2460 if (!S_ISREG(file_inode(file
)->i_mode
))
2462 return __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, false);
2465 static inline void file_end_write(struct file
*file
)
2467 if (!S_ISREG(file_inode(file
)->i_mode
))
2469 __sb_end_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
);
2473 * get_write_access() gets write permission for a file.
2474 * put_write_access() releases this write permission.
2475 * This is used for regular files.
2476 * We cannot support write (and maybe mmap read-write shared) accesses and
2477 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2478 * can have the following values:
2479 * 0: no writers, no VM_DENYWRITE mappings
2480 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2481 * > 0: (i_writecount) users are writing to the file.
2483 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2484 * except for the cases where we don't hold i_writecount yet. Then we need to
2485 * use {get,deny}_write_access() - these functions check the sign and refuse
2486 * to do the change if sign is wrong.
2488 static inline int get_write_access(struct inode
*inode
)
2490 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2492 static inline int deny_write_access(struct file
*file
)
2494 struct inode
*inode
= file_inode(file
);
2495 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2497 static inline void put_write_access(struct inode
* inode
)
2499 atomic_dec(&inode
->i_writecount
);
2501 static inline void allow_write_access(struct file
*file
)
2504 atomic_inc(&file_inode(file
)->i_writecount
);
2506 static inline bool inode_is_open_for_write(const struct inode
*inode
)
2508 return atomic_read(&inode
->i_writecount
) > 0;
2512 static inline void i_readcount_dec(struct inode
*inode
)
2514 BUG_ON(!atomic_read(&inode
->i_readcount
));
2515 atomic_dec(&inode
->i_readcount
);
2517 static inline void i_readcount_inc(struct inode
*inode
)
2519 atomic_inc(&inode
->i_readcount
);
2522 static inline void i_readcount_dec(struct inode
*inode
)
2526 static inline void i_readcount_inc(struct inode
*inode
)
2531 extern int do_pipe_flags(int *, int);
2533 extern int kernel_read(struct file
*, loff_t
, char *, unsigned long);
2534 extern ssize_t
kernel_write(struct file
*, const char *, size_t, loff_t
);
2535 extern ssize_t
__kernel_write(struct file
*, const char *, size_t, loff_t
*);
2536 extern struct file
* open_exec(const char *);
2538 /* fs/dcache.c -- generic fs support functions */
2539 extern int is_subdir(struct dentry
*, struct dentry
*);
2540 extern int path_is_under(struct path
*, struct path
*);
2542 extern char *file_path(struct file
*, char *, int);
2544 #include <linux/err.h>
2546 /* needed for stackable file system support */
2547 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int whence
);
2549 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int whence
);
2551 extern int inode_init_always(struct super_block
*, struct inode
*);
2552 extern void inode_init_once(struct inode
*);
2553 extern void address_space_init_once(struct address_space
*mapping
);
2554 extern struct inode
* igrab(struct inode
*);
2555 extern ino_t
iunique(struct super_block
*, ino_t
);
2556 extern int inode_needs_sync(struct inode
*inode
);
2557 extern int generic_delete_inode(struct inode
*inode
);
2558 static inline int generic_drop_inode(struct inode
*inode
)
2560 return !inode
->i_nlink
|| inode_unhashed(inode
);
2563 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
2564 unsigned long hashval
, int (*test
)(struct inode
*, void *),
2566 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
2567 int (*test
)(struct inode
*, void *), void *data
);
2568 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
2570 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
2571 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
2572 extern struct inode
*find_inode_nowait(struct super_block
*,
2574 int (*match
)(struct inode
*,
2575 unsigned long, void *),
2577 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
2578 extern int insert_inode_locked(struct inode
*);
2579 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2580 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
2582 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
2584 extern void unlock_new_inode(struct inode
*);
2585 extern unsigned int get_next_ino(void);
2587 extern void __iget(struct inode
* inode
);
2588 extern void iget_failed(struct inode
*);
2589 extern void clear_inode(struct inode
*);
2590 extern void __destroy_inode(struct inode
*);
2591 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
2592 extern struct inode
*new_inode(struct super_block
*sb
);
2593 extern void free_inode_nonrcu(struct inode
*inode
);
2594 extern int should_remove_suid(struct dentry
*);
2595 extern int file_remove_privs(struct file
*);
2596 extern int dentry_needs_remove_privs(struct dentry
*dentry
);
2597 static inline int file_needs_remove_privs(struct file
*file
)
2599 return dentry_needs_remove_privs(file
->f_path
.dentry
);
2602 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
2603 static inline void insert_inode_hash(struct inode
*inode
)
2605 __insert_inode_hash(inode
, inode
->i_ino
);
2608 extern void __remove_inode_hash(struct inode
*);
2609 static inline void remove_inode_hash(struct inode
*inode
)
2611 if (!inode_unhashed(inode
))
2612 __remove_inode_hash(inode
);
2615 extern void inode_sb_list_add(struct inode
*inode
);
2618 extern void submit_bio(int, struct bio
*);
2619 extern int bdev_read_only(struct block_device
*);
2621 extern int set_blocksize(struct block_device
*, int);
2622 extern int sb_set_blocksize(struct super_block
*, int);
2623 extern int sb_min_blocksize(struct super_block
*, int);
2625 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
2626 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
2627 extern ssize_t
generic_write_checks(struct kiocb
*, struct iov_iter
*);
2628 extern ssize_t
generic_file_read_iter(struct kiocb
*, struct iov_iter
*);
2629 extern ssize_t
__generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2630 extern ssize_t
generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2631 extern ssize_t
generic_file_direct_write(struct kiocb
*, struct iov_iter
*, loff_t
);
2632 extern ssize_t
generic_perform_write(struct file
*, struct iov_iter
*, loff_t
);
2634 ssize_t
vfs_iter_read(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
);
2635 ssize_t
vfs_iter_write(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
);
2637 /* fs/block_dev.c */
2638 extern ssize_t
blkdev_read_iter(struct kiocb
*iocb
, struct iov_iter
*to
);
2639 extern ssize_t
blkdev_write_iter(struct kiocb
*iocb
, struct iov_iter
*from
);
2640 extern int blkdev_fsync(struct file
*filp
, loff_t start
, loff_t end
,
2642 extern void block_sync_page(struct page
*page
);
2645 extern ssize_t
generic_file_splice_read(struct file
*, loff_t
*,
2646 struct pipe_inode_info
*, size_t, unsigned int);
2647 extern ssize_t
default_file_splice_read(struct file
*, loff_t
*,
2648 struct pipe_inode_info
*, size_t, unsigned int);
2649 extern ssize_t
iter_file_splice_write(struct pipe_inode_info
*,
2650 struct file
*, loff_t
*, size_t, unsigned int);
2651 extern ssize_t
generic_splice_sendpage(struct pipe_inode_info
*pipe
,
2652 struct file
*out
, loff_t
*, size_t len
, unsigned int flags
);
2653 extern long do_splice_direct(struct file
*in
, loff_t
*ppos
, struct file
*out
,
2654 loff_t
*opos
, size_t len
, unsigned int flags
);
2658 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
2659 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int whence
);
2660 extern loff_t
no_llseek(struct file
*file
, loff_t offset
, int whence
);
2661 extern loff_t
vfs_setpos(struct file
*file
, loff_t offset
, loff_t maxsize
);
2662 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int whence
);
2663 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
2664 int whence
, loff_t maxsize
, loff_t eof
);
2665 extern loff_t
fixed_size_llseek(struct file
*file
, loff_t offset
,
2666 int whence
, loff_t size
);
2667 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
2668 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
2670 ssize_t
dax_do_io(struct kiocb
*, struct inode
*, struct iov_iter
*, loff_t
,
2671 get_block_t
, dio_iodone_t
, int flags
);
2672 int dax_clear_blocks(struct inode
*, sector_t block
, long size
);
2673 int dax_zero_page_range(struct inode
*, loff_t from
, unsigned len
, get_block_t
);
2674 int dax_truncate_page(struct inode
*, loff_t from
, get_block_t
);
2675 int dax_fault(struct vm_area_struct
*, struct vm_fault
*, get_block_t
,
2677 int __dax_fault(struct vm_area_struct
*, struct vm_fault
*, get_block_t
,
2679 int dax_pfn_mkwrite(struct vm_area_struct
*, struct vm_fault
*);
2680 #define dax_mkwrite(vma, vmf, gb, iod) dax_fault(vma, vmf, gb, iod)
2681 #define __dax_mkwrite(vma, vmf, gb, iod) __dax_fault(vma, vmf, gb, iod)
2684 typedef void (dio_submit_t
)(int rw
, struct bio
*bio
, struct inode
*inode
,
2685 loff_t file_offset
);
2688 /* need locking between buffered and direct access */
2691 /* filesystem does not support filling holes */
2692 DIO_SKIP_HOLES
= 0x02,
2694 /* filesystem can handle aio writes beyond i_size */
2695 DIO_ASYNC_EXTEND
= 0x04,
2697 /* inode/fs/bdev does not need truncate protection */
2698 DIO_SKIP_DIO_COUNT
= 0x08,
2701 void dio_end_io(struct bio
*bio
, int error
);
2703 ssize_t
__blockdev_direct_IO(struct kiocb
*iocb
, struct inode
*inode
,
2704 struct block_device
*bdev
, struct iov_iter
*iter
,
2705 loff_t offset
, get_block_t get_block
,
2706 dio_iodone_t end_io
, dio_submit_t submit_io
,
2709 static inline ssize_t
blockdev_direct_IO(struct kiocb
*iocb
,
2710 struct inode
*inode
,
2711 struct iov_iter
*iter
, loff_t offset
,
2712 get_block_t get_block
)
2714 return __blockdev_direct_IO(iocb
, inode
, inode
->i_sb
->s_bdev
, iter
,
2715 offset
, get_block
, NULL
, NULL
,
2716 DIO_LOCKING
| DIO_SKIP_HOLES
);
2720 void inode_dio_wait(struct inode
*inode
);
2723 * inode_dio_begin - signal start of a direct I/O requests
2724 * @inode: inode the direct I/O happens on
2726 * This is called once we've finished processing a direct I/O request,
2727 * and is used to wake up callers waiting for direct I/O to be quiesced.
2729 static inline void inode_dio_begin(struct inode
*inode
)
2731 atomic_inc(&inode
->i_dio_count
);
2735 * inode_dio_end - signal finish of a direct I/O requests
2736 * @inode: inode the direct I/O happens on
2738 * This is called once we've finished processing a direct I/O request,
2739 * and is used to wake up callers waiting for direct I/O to be quiesced.
2741 static inline void inode_dio_end(struct inode
*inode
)
2743 if (atomic_dec_and_test(&inode
->i_dio_count
))
2744 wake_up_bit(&inode
->i_state
, __I_DIO_WAKEUP
);
2747 extern void inode_set_flags(struct inode
*inode
, unsigned int flags
,
2750 extern const struct file_operations generic_ro_fops
;
2752 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2754 extern int readlink_copy(char __user
*, int, const char *);
2755 extern int page_readlink(struct dentry
*, char __user
*, int);
2756 extern const char *page_follow_link_light(struct dentry
*, void **);
2757 extern void page_put_link(struct inode
*, void *);
2758 extern int __page_symlink(struct inode
*inode
, const char *symname
, int len
,
2760 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
2761 extern const struct inode_operations page_symlink_inode_operations
;
2762 extern void kfree_put_link(struct inode
*, void *);
2763 extern void free_page_put_link(struct inode
*, void *);
2764 extern int generic_readlink(struct dentry
*, char __user
*, int);
2765 extern void generic_fillattr(struct inode
*, struct kstat
*);
2766 int vfs_getattr_nosec(struct path
*path
, struct kstat
*stat
);
2767 extern int vfs_getattr(struct path
*, struct kstat
*);
2768 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2769 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2770 void __inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2771 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2772 loff_t
inode_get_bytes(struct inode
*inode
);
2773 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
2774 const char *simple_follow_link(struct dentry
*, void **);
2775 extern const struct inode_operations simple_symlink_inode_operations
;
2777 extern int iterate_dir(struct file
*, struct dir_context
*);
2779 extern int vfs_stat(const char __user
*, struct kstat
*);
2780 extern int vfs_lstat(const char __user
*, struct kstat
*);
2781 extern int vfs_fstat(unsigned int, struct kstat
*);
2782 extern int vfs_fstatat(int , const char __user
*, struct kstat
*, int);
2784 extern int do_vfs_ioctl(struct file
*filp
, unsigned int fd
, unsigned int cmd
,
2786 extern int __generic_block_fiemap(struct inode
*inode
,
2787 struct fiemap_extent_info
*fieinfo
,
2788 loff_t start
, loff_t len
,
2789 get_block_t
*get_block
);
2790 extern int generic_block_fiemap(struct inode
*inode
,
2791 struct fiemap_extent_info
*fieinfo
, u64 start
,
2792 u64 len
, get_block_t
*get_block
);
2794 extern void get_filesystem(struct file_system_type
*fs
);
2795 extern void put_filesystem(struct file_system_type
*fs
);
2796 extern struct file_system_type
*get_fs_type(const char *name
);
2797 extern struct super_block
*get_super(struct block_device
*);
2798 extern struct super_block
*get_super_thawed(struct block_device
*);
2799 extern struct super_block
*get_active_super(struct block_device
*bdev
);
2800 extern void drop_super(struct super_block
*sb
);
2801 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
2802 extern void iterate_supers_type(struct file_system_type
*,
2803 void (*)(struct super_block
*, void *), void *);
2805 extern int dcache_dir_open(struct inode
*, struct file
*);
2806 extern int dcache_dir_close(struct inode
*, struct file
*);
2807 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
2808 extern int dcache_readdir(struct file
*, struct dir_context
*);
2809 extern int simple_setattr(struct dentry
*, struct iattr
*);
2810 extern int simple_getattr(struct vfsmount
*, struct dentry
*, struct kstat
*);
2811 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
2812 extern int simple_open(struct inode
*inode
, struct file
*file
);
2813 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
2814 extern int simple_unlink(struct inode
*, struct dentry
*);
2815 extern int simple_rmdir(struct inode
*, struct dentry
*);
2816 extern int simple_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*);
2817 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
2818 extern int simple_empty(struct dentry
*);
2819 extern int simple_readpage(struct file
*file
, struct page
*page
);
2820 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
2821 loff_t pos
, unsigned len
, unsigned flags
,
2822 struct page
**pagep
, void **fsdata
);
2823 extern int simple_write_end(struct file
*file
, struct address_space
*mapping
,
2824 loff_t pos
, unsigned len
, unsigned copied
,
2825 struct page
*page
, void *fsdata
);
2826 extern int always_delete_dentry(const struct dentry
*);
2827 extern struct inode
*alloc_anon_inode(struct super_block
*);
2828 extern int simple_nosetlease(struct file
*, long, struct file_lock
**, void **);
2829 extern const struct dentry_operations simple_dentry_operations
;
2831 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
2832 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
2833 extern const struct file_operations simple_dir_operations
;
2834 extern const struct inode_operations simple_dir_inode_operations
;
2835 extern void make_empty_dir_inode(struct inode
*inode
);
2836 extern bool is_empty_dir_inode(struct inode
*inode
);
2837 struct tree_descr
{ char *name
; const struct file_operations
*ops
; int mode
; };
2838 struct dentry
*d_alloc_name(struct dentry
*, const char *);
2839 extern int simple_fill_super(struct super_block
*, unsigned long, struct tree_descr
*);
2840 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
2841 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
2843 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
2844 loff_t
*ppos
, const void *from
, size_t available
);
2845 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
2846 const void __user
*from
, size_t count
);
2848 extern int __generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
2849 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
2851 extern int generic_check_addressable(unsigned, u64
);
2853 #ifdef CONFIG_MIGRATION
2854 extern int buffer_migrate_page(struct address_space
*,
2855 struct page
*, struct page
*,
2858 #define buffer_migrate_page NULL
2861 extern int inode_change_ok(const struct inode
*, struct iattr
*);
2862 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
2863 extern void setattr_copy(struct inode
*inode
, const struct iattr
*attr
);
2865 extern int file_update_time(struct file
*file
);
2867 extern int generic_show_options(struct seq_file
*m
, struct dentry
*root
);
2868 extern void save_mount_options(struct super_block
*sb
, char *options
);
2869 extern void replace_mount_options(struct super_block
*sb
, char *options
);
2871 static inline bool io_is_direct(struct file
*filp
)
2873 return (filp
->f_flags
& O_DIRECT
) || IS_DAX(file_inode(filp
));
2876 static inline int iocb_flags(struct file
*file
)
2879 if (file
->f_flags
& O_APPEND
)
2881 if (io_is_direct(file
))
2886 static inline ino_t
parent_ino(struct dentry
*dentry
)
2891 * Don't strictly need d_lock here? If the parent ino could change
2892 * then surely we'd have a deeper race in the caller?
2894 spin_lock(&dentry
->d_lock
);
2895 res
= dentry
->d_parent
->d_inode
->i_ino
;
2896 spin_unlock(&dentry
->d_lock
);
2900 /* Transaction based IO helpers */
2903 * An argresp is stored in an allocated page and holds the
2904 * size of the argument or response, along with its content
2906 struct simple_transaction_argresp
{
2911 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2913 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
2915 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
2916 size_t size
, loff_t
*pos
);
2917 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
2919 void simple_transaction_set(struct file
*file
, size_t n
);
2922 * simple attribute files
2924 * These attributes behave similar to those in sysfs:
2926 * Writing to an attribute immediately sets a value, an open file can be
2927 * written to multiple times.
2929 * Reading from an attribute creates a buffer from the value that might get
2930 * read with multiple read calls. When the attribute has been read
2931 * completely, no further read calls are possible until the file is opened
2934 * All attributes contain a text representation of a numeric value
2935 * that are accessed with the get() and set() functions.
2937 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2938 static int __fops ## _open(struct inode *inode, struct file *file) \
2940 __simple_attr_check_format(__fmt, 0ull); \
2941 return simple_attr_open(inode, file, __get, __set, __fmt); \
2943 static const struct file_operations __fops = { \
2944 .owner = THIS_MODULE, \
2945 .open = __fops ## _open, \
2946 .release = simple_attr_release, \
2947 .read = simple_attr_read, \
2948 .write = simple_attr_write, \
2949 .llseek = generic_file_llseek, \
2952 static inline __printf(1, 2)
2953 void __simple_attr_check_format(const char *fmt
, ...)
2955 /* don't do anything, just let the compiler check the arguments; */
2958 int simple_attr_open(struct inode
*inode
, struct file
*file
,
2959 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
2961 int simple_attr_release(struct inode
*inode
, struct file
*file
);
2962 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
2963 size_t len
, loff_t
*ppos
);
2964 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
2965 size_t len
, loff_t
*ppos
);
2968 int proc_nr_files(struct ctl_table
*table
, int write
,
2969 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
2970 int proc_nr_dentry(struct ctl_table
*table
, int write
,
2971 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
2972 int proc_nr_inodes(struct ctl_table
*table
, int write
,
2973 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
2974 int __init
get_filesystem_list(char *buf
);
2976 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2977 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2979 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2980 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2981 (flag & __FMODE_NONOTIFY)))
2983 static inline int is_sxid(umode_t mode
)
2985 return (mode
& S_ISUID
) || ((mode
& S_ISGID
) && (mode
& S_IXGRP
));
2988 static inline int check_sticky(struct inode
*dir
, struct inode
*inode
)
2990 if (!(dir
->i_mode
& S_ISVTX
))
2993 return __check_sticky(dir
, inode
);
2996 static inline void inode_has_no_xattr(struct inode
*inode
)
2998 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& MS_NOSEC
))
2999 inode
->i_flags
|= S_NOSEC
;
3002 static inline bool is_root_inode(struct inode
*inode
)
3004 return inode
== inode
->i_sb
->s_root
->d_inode
;
3007 static inline bool dir_emit(struct dir_context
*ctx
,
3008 const char *name
, int namelen
,
3009 u64 ino
, unsigned type
)
3011 return ctx
->actor(ctx
, name
, namelen
, ctx
->pos
, ino
, type
) == 0;
3013 static inline bool dir_emit_dot(struct file
*file
, struct dir_context
*ctx
)
3015 return ctx
->actor(ctx
, ".", 1, ctx
->pos
,
3016 file
->f_path
.dentry
->d_inode
->i_ino
, DT_DIR
) == 0;
3018 static inline bool dir_emit_dotdot(struct file
*file
, struct dir_context
*ctx
)
3020 return ctx
->actor(ctx
, "..", 2, ctx
->pos
,
3021 parent_ino(file
->f_path
.dentry
), DT_DIR
) == 0;
3023 static inline bool dir_emit_dots(struct file
*file
, struct dir_context
*ctx
)
3025 if (ctx
->pos
== 0) {
3026 if (!dir_emit_dot(file
, ctx
))
3030 if (ctx
->pos
== 1) {
3031 if (!dir_emit_dotdot(file
, ctx
))
3037 static inline bool dir_relax(struct inode
*inode
)
3039 mutex_unlock(&inode
->i_mutex
);
3040 mutex_lock(&inode
->i_mutex
);
3041 return !IS_DEADDIR(inode
);
3044 extern bool path_noexec(const struct path
*path
);
3046 #endif /* _LINUX_FS_H */