4 #include <linux/linkage.h>
5 #include <linux/wait.h>
6 #include <linux/kdev_t.h>
7 #include <linux/dcache.h>
8 #include <linux/path.h>
9 #include <linux/stat.h>
10 #include <linux/cache.h>
11 #include <linux/list.h>
12 #include <linux/list_lru.h>
13 #include <linux/llist.h>
14 #include <linux/radix-tree.h>
15 #include <linux/rbtree.h>
16 #include <linux/init.h>
17 #include <linux/pid.h>
18 #include <linux/bug.h>
19 #include <linux/mutex.h>
20 #include <linux/rwsem.h>
21 #include <linux/capability.h>
22 #include <linux/semaphore.h>
23 #include <linux/fiemap.h>
24 #include <linux/rculist_bl.h>
25 #include <linux/atomic.h>
26 #include <linux/shrinker.h>
27 #include <linux/migrate_mode.h>
28 #include <linux/uidgid.h>
29 #include <linux/lockdep.h>
30 #include <linux/percpu-rwsem.h>
31 #include <linux/blk_types.h>
32 #include <linux/workqueue.h>
33 #include <linux/percpu-rwsem.h>
34 #include <linux/delayed_call.h>
36 #include <asm/byteorder.h>
37 #include <uapi/linux/fs.h>
39 struct backing_dev_info
;
41 struct export_operations
;
46 struct pipe_inode_info
;
47 struct poll_table_struct
;
49 struct vm_area_struct
;
52 struct swap_info_struct
;
54 struct workqueue_struct
;
57 extern void __init
inode_init(void);
58 extern void __init
inode_init_early(void);
59 extern void __init
files_init(void);
60 extern void __init
files_maxfiles_init(void);
62 extern struct files_stat_struct files_stat
;
63 extern unsigned long get_max_files(void);
64 extern int sysctl_nr_open
;
65 extern struct inodes_stat_t inodes_stat
;
66 extern int leases_enable
, lease_break_time
;
67 extern int sysctl_protected_symlinks
;
68 extern int sysctl_protected_hardlinks
;
71 typedef int (get_block_t
)(struct inode
*inode
, sector_t iblock
,
72 struct buffer_head
*bh_result
, int create
);
73 typedef void (dio_iodone_t
)(struct kiocb
*iocb
, loff_t offset
,
74 ssize_t bytes
, void *private);
75 typedef void (dax_iodone_t
)(struct buffer_head
*bh_map
, int uptodate
);
77 #define MAY_EXEC 0x00000001
78 #define MAY_WRITE 0x00000002
79 #define MAY_READ 0x00000004
80 #define MAY_APPEND 0x00000008
81 #define MAY_ACCESS 0x00000010
82 #define MAY_OPEN 0x00000020
83 #define MAY_CHDIR 0x00000040
84 /* called from RCU mode, don't block */
85 #define MAY_NOT_BLOCK 0x00000080
88 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
89 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
92 /* file is open for reading */
93 #define FMODE_READ ((__force fmode_t)0x1)
94 /* file is open for writing */
95 #define FMODE_WRITE ((__force fmode_t)0x2)
96 /* file is seekable */
97 #define FMODE_LSEEK ((__force fmode_t)0x4)
98 /* file can be accessed using pread */
99 #define FMODE_PREAD ((__force fmode_t)0x8)
100 /* file can be accessed using pwrite */
101 #define FMODE_PWRITE ((__force fmode_t)0x10)
102 /* File is opened for execution with sys_execve / sys_uselib */
103 #define FMODE_EXEC ((__force fmode_t)0x20)
104 /* File is opened with O_NDELAY (only set for block devices) */
105 #define FMODE_NDELAY ((__force fmode_t)0x40)
106 /* File is opened with O_EXCL (only set for block devices) */
107 #define FMODE_EXCL ((__force fmode_t)0x80)
108 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
109 (specialy hack for floppy.c) */
110 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
111 /* 32bit hashes as llseek() offset (for directories) */
112 #define FMODE_32BITHASH ((__force fmode_t)0x200)
113 /* 64bit hashes as llseek() offset (for directories) */
114 #define FMODE_64BITHASH ((__force fmode_t)0x400)
117 * Don't update ctime and mtime.
119 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
120 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
122 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
124 /* Expect random access pattern */
125 #define FMODE_RANDOM ((__force fmode_t)0x1000)
127 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
128 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
130 /* File is opened with O_PATH; almost nothing can be done with it */
131 #define FMODE_PATH ((__force fmode_t)0x4000)
133 /* File needs atomic accesses to f_pos */
134 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
135 /* Write access to underlying fs */
136 #define FMODE_WRITER ((__force fmode_t)0x10000)
137 /* Has read method(s) */
138 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
139 /* Has write method(s) */
140 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
142 /* File was opened by fanotify and shouldn't generate fanotify events */
143 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
146 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
147 * that indicates that they should check the contents of the iovec are
148 * valid, but not check the memory that the iovec elements
151 #define CHECK_IOVEC_ONLY -1
154 * The below are the various read and write types that we support. Some of
155 * them include behavioral modifiers that send information down to the
156 * block layer and IO scheduler. Terminology:
158 * The block layer uses device plugging to defer IO a little bit, in
159 * the hope that we will see more IO very shortly. This increases
160 * coalescing of adjacent IO and thus reduces the number of IOs we
161 * have to send to the device. It also allows for better queuing,
162 * if the IO isn't mergeable. If the caller is going to be waiting
163 * for the IO, then he must ensure that the device is unplugged so
164 * that the IO is dispatched to the driver.
166 * All IO is handled async in Linux. This is fine for background
167 * writes, but for reads or writes that someone waits for completion
168 * on, we want to notify the block layer and IO scheduler so that they
169 * know about it. That allows them to make better scheduling
170 * decisions. So when the below references 'sync' and 'async', it
171 * is referencing this priority hint.
173 * With that in mind, the available types are:
175 * READ A normal read operation. Device will be plugged.
176 * READ_SYNC A synchronous read. Device is not plugged, caller can
177 * immediately wait on this read without caring about
179 * READA Used for read-ahead operations. Lower priority, and the
180 * block layer could (in theory) choose to ignore this
181 * request if it runs into resource problems.
182 * WRITE A normal async write. Device will be plugged.
183 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
184 * the hint that someone will be waiting on this IO
185 * shortly. The write equivalent of READ_SYNC.
186 * WRITE_ODIRECT Special case write for O_DIRECT only.
187 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
188 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
189 * non-volatile media on completion.
190 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
191 * by a cache flush and data is guaranteed to be on
192 * non-volatile media on completion.
195 #define RW_MASK REQ_WRITE
196 #define RWA_MASK REQ_RAHEAD
199 #define WRITE RW_MASK
200 #define READA RWA_MASK
202 #define READ_SYNC (READ | REQ_SYNC)
203 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
204 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
205 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
206 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
207 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
210 * Attribute flags. These should be or-ed together to figure out what
213 #define ATTR_MODE (1 << 0)
214 #define ATTR_UID (1 << 1)
215 #define ATTR_GID (1 << 2)
216 #define ATTR_SIZE (1 << 3)
217 #define ATTR_ATIME (1 << 4)
218 #define ATTR_MTIME (1 << 5)
219 #define ATTR_CTIME (1 << 6)
220 #define ATTR_ATIME_SET (1 << 7)
221 #define ATTR_MTIME_SET (1 << 8)
222 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
223 #define ATTR_ATTR_FLAG (1 << 10)
224 #define ATTR_KILL_SUID (1 << 11)
225 #define ATTR_KILL_SGID (1 << 12)
226 #define ATTR_FILE (1 << 13)
227 #define ATTR_KILL_PRIV (1 << 14)
228 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
229 #define ATTR_TIMES_SET (1 << 16)
232 * Whiteout is represented by a char device. The following constants define the
233 * mode and device number to use.
235 #define WHITEOUT_MODE 0
236 #define WHITEOUT_DEV 0
239 * This is the Inode Attributes structure, used for notify_change(). It
240 * uses the above definitions as flags, to know which values have changed.
241 * Also, in this manner, a Filesystem can look at only the values it cares
242 * about. Basically, these are the attributes that the VFS layer can
243 * request to change from the FS layer.
245 * Derek Atkins <warlord@MIT.EDU> 94-10-20
248 unsigned int ia_valid
;
253 struct timespec ia_atime
;
254 struct timespec ia_mtime
;
255 struct timespec ia_ctime
;
258 * Not an attribute, but an auxiliary info for filesystems wanting to
259 * implement an ftruncate() like method. NOTE: filesystem should
260 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
262 struct file
*ia_file
;
266 * Includes for diskquotas.
268 #include <linux/quota.h>
271 * Maximum number of layers of fs stack. Needs to be limited to
272 * prevent kernel stack overflow
274 #define FILESYSTEM_MAX_STACK_DEPTH 2
277 * enum positive_aop_returns - aop return codes with specific semantics
279 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
280 * completed, that the page is still locked, and
281 * should be considered active. The VM uses this hint
282 * to return the page to the active list -- it won't
283 * be a candidate for writeback again in the near
284 * future. Other callers must be careful to unlock
285 * the page if they get this return. Returned by
288 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
289 * unlocked it and the page might have been truncated.
290 * The caller should back up to acquiring a new page and
291 * trying again. The aop will be taking reasonable
292 * precautions not to livelock. If the caller held a page
293 * reference, it should drop it before retrying. Returned
296 * address_space_operation functions return these large constants to indicate
297 * special semantics to the caller. These are much larger than the bytes in a
298 * page to allow for functions that return the number of bytes operated on in a
302 enum positive_aop_returns
{
303 AOP_WRITEPAGE_ACTIVATE
= 0x80000,
304 AOP_TRUNCATED_PAGE
= 0x80001,
307 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
308 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
309 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
310 * helper code (eg buffer layer)
311 * to clear GFP_FS from alloc */
314 * oh the beauties of C type declarations.
317 struct address_space
;
318 struct writeback_control
;
320 #define IOCB_EVENTFD (1 << 0)
321 #define IOCB_APPEND (1 << 1)
322 #define IOCB_DIRECT (1 << 2)
325 struct file
*ki_filp
;
327 void (*ki_complete
)(struct kiocb
*iocb
, long ret
, long ret2
);
332 static inline bool is_sync_kiocb(struct kiocb
*kiocb
)
334 return kiocb
->ki_complete
== NULL
;
337 static inline int iocb_flags(struct file
*file
);
339 static inline void init_sync_kiocb(struct kiocb
*kiocb
, struct file
*filp
)
341 *kiocb
= (struct kiocb
) {
343 .ki_flags
= iocb_flags(filp
),
348 * "descriptor" for what we're up to with a read.
349 * This allows us to use the same read code yet
350 * have multiple different users of the data that
351 * we read from a file.
353 * The simplest case just copies the data to user
366 typedef int (*read_actor_t
)(read_descriptor_t
*, struct page
*,
367 unsigned long, unsigned long);
369 struct address_space_operations
{
370 int (*writepage
)(struct page
*page
, struct writeback_control
*wbc
);
371 int (*readpage
)(struct file
*, struct page
*);
373 /* Write back some dirty pages from this mapping. */
374 int (*writepages
)(struct address_space
*, struct writeback_control
*);
376 /* Set a page dirty. Return true if this dirtied it */
377 int (*set_page_dirty
)(struct page
*page
);
379 int (*readpages
)(struct file
*filp
, struct address_space
*mapping
,
380 struct list_head
*pages
, unsigned nr_pages
);
382 int (*write_begin
)(struct file
*, struct address_space
*mapping
,
383 loff_t pos
, unsigned len
, unsigned flags
,
384 struct page
**pagep
, void **fsdata
);
385 int (*write_end
)(struct file
*, struct address_space
*mapping
,
386 loff_t pos
, unsigned len
, unsigned copied
,
387 struct page
*page
, void *fsdata
);
389 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
390 sector_t (*bmap
)(struct address_space
*, sector_t
);
391 void (*invalidatepage
) (struct page
*, unsigned int, unsigned int);
392 int (*releasepage
) (struct page
*, gfp_t
);
393 void (*freepage
)(struct page
*);
394 ssize_t (*direct_IO
)(struct kiocb
*, struct iov_iter
*iter
, loff_t offset
);
396 * migrate the contents of a page to the specified target. If
397 * migrate_mode is MIGRATE_ASYNC, it must not block.
399 int (*migratepage
) (struct address_space
*,
400 struct page
*, struct page
*, enum migrate_mode
);
401 int (*launder_page
) (struct page
*);
402 int (*is_partially_uptodate
) (struct page
*, unsigned long,
404 void (*is_dirty_writeback
) (struct page
*, bool *, bool *);
405 int (*error_remove_page
)(struct address_space
*, struct page
*);
407 /* swapfile support */
408 int (*swap_activate
)(struct swap_info_struct
*sis
, struct file
*file
,
410 void (*swap_deactivate
)(struct file
*file
);
413 extern const struct address_space_operations empty_aops
;
416 * pagecache_write_begin/pagecache_write_end must be used by general code
417 * to write into the pagecache.
419 int pagecache_write_begin(struct file
*, struct address_space
*mapping
,
420 loff_t pos
, unsigned len
, unsigned flags
,
421 struct page
**pagep
, void **fsdata
);
423 int pagecache_write_end(struct file
*, struct address_space
*mapping
,
424 loff_t pos
, unsigned len
, unsigned copied
,
425 struct page
*page
, void *fsdata
);
427 struct address_space
{
428 struct inode
*host
; /* owner: inode, block_device */
429 struct radix_tree_root page_tree
; /* radix tree of all pages */
430 spinlock_t tree_lock
; /* and lock protecting it */
431 atomic_t i_mmap_writable
;/* count VM_SHARED mappings */
432 struct rb_root i_mmap
; /* tree of private and shared mappings */
433 struct rw_semaphore i_mmap_rwsem
; /* protect tree, count, list */
434 /* Protected by tree_lock together with the radix tree */
435 unsigned long nrpages
; /* number of total pages */
436 unsigned long nrshadows
; /* number of shadow entries */
437 pgoff_t writeback_index
;/* writeback starts here */
438 const struct address_space_operations
*a_ops
; /* methods */
439 unsigned long flags
; /* error bits/gfp mask */
440 spinlock_t private_lock
; /* for use by the address_space */
441 struct list_head private_list
; /* ditto */
442 void *private_data
; /* ditto */
443 } __attribute__((aligned(sizeof(long))));
445 * On most architectures that alignment is already the case; but
446 * must be enforced here for CRIS, to let the least significant bit
447 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
449 struct request_queue
;
451 struct block_device
{
452 dev_t bd_dev
; /* not a kdev_t - it's a search key */
454 struct inode
* bd_inode
; /* will die */
455 struct super_block
* bd_super
;
456 struct mutex bd_mutex
; /* open/close mutex */
457 struct list_head bd_inodes
;
461 bool bd_write_holder
;
463 struct list_head bd_holder_disks
;
465 struct block_device
* bd_contains
;
466 unsigned bd_block_size
;
467 struct hd_struct
* bd_part
;
468 /* number of times partitions within this device have been opened. */
469 unsigned bd_part_count
;
471 struct gendisk
* bd_disk
;
472 struct request_queue
* bd_queue
;
473 struct list_head bd_list
;
475 * Private data. You must have bd_claim'ed the block_device
476 * to use this. NOTE: bd_claim allows an owner to claim
477 * the same device multiple times, the owner must take special
478 * care to not mess up bd_private for that case.
480 unsigned long bd_private
;
482 /* The counter of freeze processes */
483 int bd_fsfreeze_count
;
484 /* Mutex for freeze */
485 struct mutex bd_fsfreeze_mutex
;
492 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
495 #define PAGECACHE_TAG_DIRTY 0
496 #define PAGECACHE_TAG_WRITEBACK 1
497 #define PAGECACHE_TAG_TOWRITE 2
499 int mapping_tagged(struct address_space
*mapping
, int tag
);
501 static inline void i_mmap_lock_write(struct address_space
*mapping
)
503 down_write(&mapping
->i_mmap_rwsem
);
506 static inline void i_mmap_unlock_write(struct address_space
*mapping
)
508 up_write(&mapping
->i_mmap_rwsem
);
511 static inline void i_mmap_lock_read(struct address_space
*mapping
)
513 down_read(&mapping
->i_mmap_rwsem
);
516 static inline void i_mmap_unlock_read(struct address_space
*mapping
)
518 up_read(&mapping
->i_mmap_rwsem
);
522 * Might pages of this file be mapped into userspace?
524 static inline int mapping_mapped(struct address_space
*mapping
)
526 return !RB_EMPTY_ROOT(&mapping
->i_mmap
);
530 * Might pages of this file have been modified in userspace?
531 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
532 * marks vma as VM_SHARED if it is shared, and the file was opened for
533 * writing i.e. vma may be mprotected writable even if now readonly.
535 * If i_mmap_writable is negative, no new writable mappings are allowed. You
536 * can only deny writable mappings, if none exists right now.
538 static inline int mapping_writably_mapped(struct address_space
*mapping
)
540 return atomic_read(&mapping
->i_mmap_writable
) > 0;
543 static inline int mapping_map_writable(struct address_space
*mapping
)
545 return atomic_inc_unless_negative(&mapping
->i_mmap_writable
) ?
549 static inline void mapping_unmap_writable(struct address_space
*mapping
)
551 atomic_dec(&mapping
->i_mmap_writable
);
554 static inline int mapping_deny_writable(struct address_space
*mapping
)
556 return atomic_dec_unless_positive(&mapping
->i_mmap_writable
) ?
560 static inline void mapping_allow_writable(struct address_space
*mapping
)
562 atomic_inc(&mapping
->i_mmap_writable
);
566 * Use sequence counter to get consistent i_size on 32-bit processors.
568 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
569 #include <linux/seqlock.h>
570 #define __NEED_I_SIZE_ORDERED
571 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
573 #define i_size_ordered_init(inode) do { } while (0)
577 #define ACL_NOT_CACHED ((void *)(-1))
579 #define IOP_FASTPERM 0x0001
580 #define IOP_LOOKUP 0x0002
581 #define IOP_NOFOLLOW 0x0004
584 * Keep mostly read-only and often accessed (especially for
585 * the RCU path lookup and 'stat' data) fields at the beginning
586 * of the 'struct inode'
590 unsigned short i_opflags
;
593 unsigned int i_flags
;
595 #ifdef CONFIG_FS_POSIX_ACL
596 struct posix_acl
*i_acl
;
597 struct posix_acl
*i_default_acl
;
600 const struct inode_operations
*i_op
;
601 struct super_block
*i_sb
;
602 struct address_space
*i_mapping
;
604 #ifdef CONFIG_SECURITY
608 /* Stat data, not accessed from path walking */
611 * Filesystems may only read i_nlink directly. They shall use the
612 * following functions for modification:
614 * (set|clear|inc|drop)_nlink
615 * inode_(inc|dec)_link_count
618 const unsigned int i_nlink
;
619 unsigned int __i_nlink
;
623 struct timespec i_atime
;
624 struct timespec i_mtime
;
625 struct timespec i_ctime
;
626 spinlock_t i_lock
; /* i_blocks, i_bytes, maybe i_size */
627 unsigned short i_bytes
;
628 unsigned int i_blkbits
;
631 #ifdef __NEED_I_SIZE_ORDERED
632 seqcount_t i_size_seqcount
;
636 unsigned long i_state
;
637 struct mutex i_mutex
;
639 unsigned long dirtied_when
; /* jiffies of first dirtying */
640 unsigned long dirtied_time_when
;
642 struct hlist_node i_hash
;
643 struct list_head i_io_list
; /* backing dev IO list */
644 #ifdef CONFIG_CGROUP_WRITEBACK
645 struct bdi_writeback
*i_wb
; /* the associated cgroup wb */
647 /* foreign inode detection, see wbc_detach_inode() */
649 u16 i_wb_frn_avg_time
;
650 u16 i_wb_frn_history
;
652 struct list_head i_lru
; /* inode LRU list */
653 struct list_head i_sb_list
;
655 struct hlist_head i_dentry
;
656 struct rcu_head i_rcu
;
660 atomic_t i_dio_count
;
661 atomic_t i_writecount
;
663 atomic_t i_readcount
; /* struct files open RO */
665 const struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
666 struct file_lock_context
*i_flctx
;
667 struct address_space i_data
;
668 struct list_head i_devices
;
670 struct pipe_inode_info
*i_pipe
;
671 struct block_device
*i_bdev
;
678 #ifdef CONFIG_FSNOTIFY
679 __u32 i_fsnotify_mask
; /* all events this inode cares about */
680 struct hlist_head i_fsnotify_marks
;
683 void *i_private
; /* fs or device private pointer */
686 static inline int inode_unhashed(struct inode
*inode
)
688 return hlist_unhashed(&inode
->i_hash
);
692 * inode->i_mutex nesting subclasses for the lock validator:
694 * 0: the object of the current VFS operation
698 * 4: second non-directory
699 * 5: second parent (when locking independent directories in rename)
701 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
702 * non-directories at once.
704 * The locking order between these classes is
705 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
707 enum inode_i_mutex_lock_class
717 void lock_two_nondirectories(struct inode
*, struct inode
*);
718 void unlock_two_nondirectories(struct inode
*, struct inode
*);
721 * NOTE: in a 32bit arch with a preemptable kernel and
722 * an UP compile the i_size_read/write must be atomic
723 * with respect to the local cpu (unlike with preempt disabled),
724 * but they don't need to be atomic with respect to other cpus like in
725 * true SMP (so they need either to either locally disable irq around
726 * the read or for example on x86 they can be still implemented as a
727 * cmpxchg8b without the need of the lock prefix). For SMP compiles
728 * and 64bit archs it makes no difference if preempt is enabled or not.
730 static inline loff_t
i_size_read(const struct inode
*inode
)
732 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
737 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
738 i_size
= inode
->i_size
;
739 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
741 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
745 i_size
= inode
->i_size
;
749 return inode
->i_size
;
754 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
755 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
756 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
758 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
760 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
762 write_seqcount_begin(&inode
->i_size_seqcount
);
763 inode
->i_size
= i_size
;
764 write_seqcount_end(&inode
->i_size_seqcount
);
766 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
768 inode
->i_size
= i_size
;
771 inode
->i_size
= i_size
;
775 /* Helper functions so that in most cases filesystems will
776 * not need to deal directly with kuid_t and kgid_t and can
777 * instead deal with the raw numeric values that are stored
780 static inline uid_t
i_uid_read(const struct inode
*inode
)
782 return from_kuid(&init_user_ns
, inode
->i_uid
);
785 static inline gid_t
i_gid_read(const struct inode
*inode
)
787 return from_kgid(&init_user_ns
, inode
->i_gid
);
790 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
792 inode
->i_uid
= make_kuid(&init_user_ns
, uid
);
795 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
797 inode
->i_gid
= make_kgid(&init_user_ns
, gid
);
800 static inline unsigned iminor(const struct inode
*inode
)
802 return MINOR(inode
->i_rdev
);
805 static inline unsigned imajor(const struct inode
*inode
)
807 return MAJOR(inode
->i_rdev
);
810 extern struct block_device
*I_BDEV(struct inode
*inode
);
813 rwlock_t lock
; /* protects pid, uid, euid fields */
814 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
815 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
816 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
817 int signum
; /* posix.1b rt signal to be delivered on IO */
821 * Track a single file's readahead state
823 struct file_ra_state
{
824 pgoff_t start
; /* where readahead started */
825 unsigned int size
; /* # of readahead pages */
826 unsigned int async_size
; /* do asynchronous readahead when
827 there are only # of pages ahead */
829 unsigned int ra_pages
; /* Maximum readahead window */
830 unsigned int mmap_miss
; /* Cache miss stat for mmap accesses */
831 loff_t prev_pos
; /* Cache last read() position */
835 * Check if @index falls in the readahead windows.
837 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
839 return (index
>= ra
->start
&&
840 index
< ra
->start
+ ra
->size
);
845 struct llist_node fu_llist
;
846 struct rcu_head fu_rcuhead
;
849 struct inode
*f_inode
; /* cached value */
850 const struct file_operations
*f_op
;
853 * Protects f_ep_links, f_flags.
854 * Must not be taken from IRQ context.
857 atomic_long_t f_count
;
858 unsigned int f_flags
;
860 struct mutex f_pos_lock
;
862 struct fown_struct f_owner
;
863 const struct cred
*f_cred
;
864 struct file_ra_state f_ra
;
867 #ifdef CONFIG_SECURITY
870 /* needed for tty driver, and maybe others */
874 /* Used by fs/eventpoll.c to link all the hooks to this file */
875 struct list_head f_ep_links
;
876 struct list_head f_tfile_llink
;
877 #endif /* #ifdef CONFIG_EPOLL */
878 struct address_space
*f_mapping
;
879 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
884 /* file identifier */
885 unsigned char f_handle
[0];
888 static inline struct file
*get_file(struct file
*f
)
890 atomic_long_inc(&f
->f_count
);
893 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
894 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
895 #define file_count(x) atomic_long_read(&(x)->f_count)
897 #define MAX_NON_LFS ((1UL<<31) - 1)
899 /* Page cache limit. The filesystems should put that into their s_maxbytes
900 limits, otherwise bad things can happen in VM. */
901 #if BITS_PER_LONG==32
902 #define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
903 #elif BITS_PER_LONG==64
904 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
909 #define FL_DELEG 4 /* NFSv4 delegation */
910 #define FL_ACCESS 8 /* not trying to lock, just looking */
911 #define FL_EXISTS 16 /* when unlocking, test for existence */
912 #define FL_LEASE 32 /* lease held on this file */
913 #define FL_CLOSE 64 /* unlock on close */
914 #define FL_SLEEP 128 /* A blocking lock */
915 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
916 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
917 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
918 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
921 * Special return value from posix_lock_file() and vfs_lock_file() for
922 * asynchronous locking.
924 #define FILE_LOCK_DEFERRED 1
926 /* legacy typedef, should eventually be removed */
927 typedef void *fl_owner_t
;
931 struct file_lock_operations
{
932 void (*fl_copy_lock
)(struct file_lock
*, struct file_lock
*);
933 void (*fl_release_private
)(struct file_lock
*);
936 struct lock_manager_operations
{
937 int (*lm_compare_owner
)(struct file_lock
*, struct file_lock
*);
938 unsigned long (*lm_owner_key
)(struct file_lock
*);
939 fl_owner_t (*lm_get_owner
)(fl_owner_t
);
940 void (*lm_put_owner
)(fl_owner_t
);
941 void (*lm_notify
)(struct file_lock
*); /* unblock callback */
942 int (*lm_grant
)(struct file_lock
*, int);
943 bool (*lm_break
)(struct file_lock
*);
944 int (*lm_change
)(struct file_lock
*, int, struct list_head
*);
945 void (*lm_setup
)(struct file_lock
*, void **);
948 struct lock_manager
{
949 struct list_head list
;
951 * NFSv4 and up also want opens blocked during the grace period;
958 void locks_start_grace(struct net
*, struct lock_manager
*);
959 void locks_end_grace(struct lock_manager
*);
960 int locks_in_grace(struct net
*);
961 int opens_in_grace(struct net
*);
963 /* that will die - we need it for nfs_lock_info */
964 #include <linux/nfs_fs_i.h>
967 * struct file_lock represents a generic "file lock". It's used to represent
968 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
969 * note that the same struct is used to represent both a request for a lock and
970 * the lock itself, but the same object is never used for both.
972 * FIXME: should we create a separate "struct lock_request" to help distinguish
975 * The varous i_flctx lists are ordered by:
978 * 2) lock range start
981 * Obviously, the last two criteria only matter for POSIX locks.
984 struct file_lock
*fl_next
; /* singly linked list for this inode */
985 struct list_head fl_list
; /* link into file_lock_context */
986 struct hlist_node fl_link
; /* node in global lists */
987 struct list_head fl_block
; /* circular list of blocked processes */
989 unsigned int fl_flags
;
990 unsigned char fl_type
;
992 int fl_link_cpu
; /* what cpu's list is this on? */
993 struct pid
*fl_nspid
;
994 wait_queue_head_t fl_wait
;
995 struct file
*fl_file
;
999 struct fasync_struct
* fl_fasync
; /* for lease break notifications */
1000 /* for lease breaks: */
1001 unsigned long fl_break_time
;
1002 unsigned long fl_downgrade_time
;
1004 const struct file_lock_operations
*fl_ops
; /* Callbacks for filesystems */
1005 const struct lock_manager_operations
*fl_lmops
; /* Callbacks for lockmanagers */
1007 struct nfs_lock_info nfs_fl
;
1008 struct nfs4_lock_info nfs4_fl
;
1010 struct list_head link
; /* link in AFS vnode's pending_locks list */
1011 int state
; /* state of grant or error if -ve */
1016 struct file_lock_context
{
1017 spinlock_t flc_lock
;
1018 struct list_head flc_flock
;
1019 struct list_head flc_posix
;
1020 struct list_head flc_lease
;
1023 /* The following constant reflects the upper bound of the file/locking space */
1025 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1026 #define OFFSET_MAX INT_LIMIT(loff_t)
1027 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1030 #include <linux/fcntl.h>
1032 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
1034 #ifdef CONFIG_FILE_LOCKING
1035 extern int fcntl_getlk(struct file
*, unsigned int, struct flock __user
*);
1036 extern int fcntl_setlk(unsigned int, struct file
*, unsigned int,
1037 struct flock __user
*);
1039 #if BITS_PER_LONG == 32
1040 extern int fcntl_getlk64(struct file
*, unsigned int, struct flock64 __user
*);
1041 extern int fcntl_setlk64(unsigned int, struct file
*, unsigned int,
1042 struct flock64 __user
*);
1045 extern int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
);
1046 extern int fcntl_getlease(struct file
*filp
);
1049 void locks_free_lock_context(struct inode
*inode
);
1050 void locks_free_lock(struct file_lock
*fl
);
1051 extern void locks_init_lock(struct file_lock
*);
1052 extern struct file_lock
* locks_alloc_lock(void);
1053 extern void locks_copy_lock(struct file_lock
*, struct file_lock
*);
1054 extern void locks_copy_conflock(struct file_lock
*, struct file_lock
*);
1055 extern void locks_remove_posix(struct file
*, fl_owner_t
);
1056 extern void locks_remove_file(struct file
*);
1057 extern void locks_release_private(struct file_lock
*);
1058 extern void posix_test_lock(struct file
*, struct file_lock
*);
1059 extern int posix_lock_file(struct file
*, struct file_lock
*, struct file_lock
*);
1060 extern int posix_unblock_lock(struct file_lock
*);
1061 extern int vfs_test_lock(struct file
*, struct file_lock
*);
1062 extern int vfs_lock_file(struct file
*, unsigned int, struct file_lock
*, struct file_lock
*);
1063 extern int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
);
1064 extern int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
);
1065 extern int __break_lease(struct inode
*inode
, unsigned int flags
, unsigned int type
);
1066 extern void lease_get_mtime(struct inode
*, struct timespec
*time
);
1067 extern int generic_setlease(struct file
*, long, struct file_lock
**, void **priv
);
1068 extern int vfs_setlease(struct file
*, long, struct file_lock
**, void **);
1069 extern int lease_modify(struct file_lock
*, int, struct list_head
*);
1070 struct files_struct
;
1071 extern void show_fd_locks(struct seq_file
*f
,
1072 struct file
*filp
, struct files_struct
*files
);
1073 #else /* !CONFIG_FILE_LOCKING */
1074 static inline int fcntl_getlk(struct file
*file
, unsigned int cmd
,
1075 struct flock __user
*user
)
1080 static inline int fcntl_setlk(unsigned int fd
, struct file
*file
,
1081 unsigned int cmd
, struct flock __user
*user
)
1086 #if BITS_PER_LONG == 32
1087 static inline int fcntl_getlk64(struct file
*file
, unsigned int cmd
,
1088 struct flock64 __user
*user
)
1093 static inline int fcntl_setlk64(unsigned int fd
, struct file
*file
,
1094 unsigned int cmd
, struct flock64 __user
*user
)
1099 static inline int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
)
1104 static inline int fcntl_getlease(struct file
*filp
)
1110 locks_free_lock_context(struct inode
*inode
)
1114 static inline void locks_init_lock(struct file_lock
*fl
)
1119 static inline void locks_copy_conflock(struct file_lock
*new, struct file_lock
*fl
)
1124 static inline void locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
1129 static inline void locks_remove_posix(struct file
*filp
, fl_owner_t owner
)
1134 static inline void locks_remove_file(struct file
*filp
)
1139 static inline void posix_test_lock(struct file
*filp
, struct file_lock
*fl
)
1144 static inline int posix_lock_file(struct file
*filp
, struct file_lock
*fl
,
1145 struct file_lock
*conflock
)
1150 static inline int posix_unblock_lock(struct file_lock
*waiter
)
1155 static inline int vfs_test_lock(struct file
*filp
, struct file_lock
*fl
)
1160 static inline int vfs_lock_file(struct file
*filp
, unsigned int cmd
,
1161 struct file_lock
*fl
, struct file_lock
*conf
)
1166 static inline int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
)
1171 static inline int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
)
1176 static inline int __break_lease(struct inode
*inode
, unsigned int mode
, unsigned int type
)
1181 static inline void lease_get_mtime(struct inode
*inode
, struct timespec
*time
)
1186 static inline int generic_setlease(struct file
*filp
, long arg
,
1187 struct file_lock
**flp
, void **priv
)
1192 static inline int vfs_setlease(struct file
*filp
, long arg
,
1193 struct file_lock
**lease
, void **priv
)
1198 static inline int lease_modify(struct file_lock
*fl
, int arg
,
1199 struct list_head
*dispose
)
1204 struct files_struct
;
1205 static inline void show_fd_locks(struct seq_file
*f
,
1206 struct file
*filp
, struct files_struct
*files
) {}
1207 #endif /* !CONFIG_FILE_LOCKING */
1209 static inline struct inode
*file_inode(const struct file
*f
)
1214 static inline int locks_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1216 return locks_lock_inode_wait(file_inode(filp
), fl
);
1219 struct fasync_struct
{
1223 struct fasync_struct
*fa_next
; /* singly linked list */
1224 struct file
*fa_file
;
1225 struct rcu_head fa_rcu
;
1228 #define FASYNC_MAGIC 0x4601
1230 /* SMP safe fasync helpers: */
1231 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1232 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1233 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1234 extern struct fasync_struct
*fasync_alloc(void);
1235 extern void fasync_free(struct fasync_struct
*);
1237 /* can be called from interrupts */
1238 extern void kill_fasync(struct fasync_struct
**, int, int);
1240 extern void __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1241 extern void f_setown(struct file
*filp
, unsigned long arg
, int force
);
1242 extern void f_delown(struct file
*filp
);
1243 extern pid_t
f_getown(struct file
*filp
);
1244 extern int send_sigurg(struct fown_struct
*fown
);
1252 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1253 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1254 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1255 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1256 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1259 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1260 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1262 /* Possible states of 'frozen' field */
1264 SB_UNFROZEN
= 0, /* FS is unfrozen */
1265 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1266 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1267 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1268 * internal threads if needed) */
1269 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1272 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1275 int frozen
; /* Is sb frozen? */
1276 wait_queue_head_t wait_unfrozen
; /* for get_super_thawed() */
1277 struct percpu_rw_semaphore rw_sem
[SB_FREEZE_LEVELS
];
1280 struct super_block
{
1281 struct list_head s_list
; /* Keep this first */
1282 dev_t s_dev
; /* search index; _not_ kdev_t */
1283 unsigned char s_blocksize_bits
;
1284 unsigned long s_blocksize
;
1285 loff_t s_maxbytes
; /* Max file size */
1286 struct file_system_type
*s_type
;
1287 const struct super_operations
*s_op
;
1288 const struct dquot_operations
*dq_op
;
1289 const struct quotactl_ops
*s_qcop
;
1290 const struct export_operations
*s_export_op
;
1291 unsigned long s_flags
;
1292 unsigned long s_iflags
; /* internal SB_I_* flags */
1293 unsigned long s_magic
;
1294 struct dentry
*s_root
;
1295 struct rw_semaphore s_umount
;
1298 #ifdef CONFIG_SECURITY
1301 const struct xattr_handler
**s_xattr
;
1303 struct hlist_bl_head s_anon
; /* anonymous dentries for (nfs) exporting */
1304 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1305 struct block_device
*s_bdev
;
1306 struct backing_dev_info
*s_bdi
;
1307 struct mtd_info
*s_mtd
;
1308 struct hlist_node s_instances
;
1309 unsigned int s_quota_types
; /* Bitmask of supported quota types */
1310 struct quota_info s_dquot
; /* Diskquota specific options */
1312 struct sb_writers s_writers
;
1314 char s_id
[32]; /* Informational name */
1315 u8 s_uuid
[16]; /* UUID */
1317 void *s_fs_info
; /* Filesystem private info */
1318 unsigned int s_max_links
;
1321 /* Granularity of c/m/atime in ns.
1322 Cannot be worse than a second */
1326 * The next field is for VFS *only*. No filesystems have any business
1327 * even looking at it. You had been warned.
1329 struct mutex s_vfs_rename_mutex
; /* Kludge */
1332 * Filesystem subtype. If non-empty the filesystem type field
1333 * in /proc/mounts will be "type.subtype"
1338 * Saved mount options for lazy filesystems using
1339 * generic_show_options()
1341 char __rcu
*s_options
;
1342 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1345 * Saved pool identifier for cleancache (-1 means none)
1347 int cleancache_poolid
;
1349 struct shrinker s_shrink
; /* per-sb shrinker handle */
1351 /* Number of inodes with nlink == 0 but still referenced */
1352 atomic_long_t s_remove_count
;
1354 /* Being remounted read-only */
1355 int s_readonly_remount
;
1357 /* AIO completions deferred from interrupt context */
1358 struct workqueue_struct
*s_dio_done_wq
;
1359 struct hlist_head s_pins
;
1362 * Keep the lru lists last in the structure so they always sit on their
1363 * own individual cachelines.
1365 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp
;
1366 struct list_lru s_inode_lru ____cacheline_aligned_in_smp
;
1367 struct rcu_head rcu
;
1368 struct work_struct destroy_work
;
1370 struct mutex s_sync_lock
; /* sync serialisation lock */
1373 * Indicates how deep in a filesystem stack this SB is
1377 /* s_inode_list_lock protects s_inodes */
1378 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp
;
1379 struct list_head s_inodes
; /* all inodes */
1382 extern struct timespec
current_fs_time(struct super_block
*sb
);
1385 * Snapshotting support.
1388 void __sb_end_write(struct super_block
*sb
, int level
);
1389 int __sb_start_write(struct super_block
*sb
, int level
, bool wait
);
1391 #define __sb_writers_acquired(sb, lev) \
1392 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1393 #define __sb_writers_release(sb, lev) \
1394 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1397 * sb_end_write - drop write access to a superblock
1398 * @sb: the super we wrote to
1400 * Decrement number of writers to the filesystem. Wake up possible waiters
1401 * wanting to freeze the filesystem.
1403 static inline void sb_end_write(struct super_block
*sb
)
1405 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1409 * sb_end_pagefault - drop write access to a superblock from a page fault
1410 * @sb: the super we wrote to
1412 * Decrement number of processes handling write page fault to the filesystem.
1413 * Wake up possible waiters wanting to freeze the filesystem.
1415 static inline void sb_end_pagefault(struct super_block
*sb
)
1417 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1421 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1422 * @sb: the super we wrote to
1424 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1425 * waiters wanting to freeze the filesystem.
1427 static inline void sb_end_intwrite(struct super_block
*sb
)
1429 __sb_end_write(sb
, SB_FREEZE_FS
);
1433 * sb_start_write - get write access to a superblock
1434 * @sb: the super we write to
1436 * When a process wants to write data or metadata to a file system (i.e. dirty
1437 * a page or an inode), it should embed the operation in a sb_start_write() -
1438 * sb_end_write() pair to get exclusion against file system freezing. This
1439 * function increments number of writers preventing freezing. If the file
1440 * system is already frozen, the function waits until the file system is
1443 * Since freeze protection behaves as a lock, users have to preserve
1444 * ordering of freeze protection and other filesystem locks. Generally,
1445 * freeze protection should be the outermost lock. In particular, we have:
1448 * -> i_mutex (write path, truncate, directory ops, ...)
1449 * -> s_umount (freeze_super, thaw_super)
1451 static inline void sb_start_write(struct super_block
*sb
)
1453 __sb_start_write(sb
, SB_FREEZE_WRITE
, true);
1456 static inline int sb_start_write_trylock(struct super_block
*sb
)
1458 return __sb_start_write(sb
, SB_FREEZE_WRITE
, false);
1462 * sb_start_pagefault - get write access to a superblock from a page fault
1463 * @sb: the super we write to
1465 * When a process starts handling write page fault, it should embed the
1466 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1467 * exclusion against file system freezing. This is needed since the page fault
1468 * is going to dirty a page. This function increments number of running page
1469 * faults preventing freezing. If the file system is already frozen, the
1470 * function waits until the file system is thawed.
1472 * Since page fault freeze protection behaves as a lock, users have to preserve
1473 * ordering of freeze protection and other filesystem locks. It is advised to
1474 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1475 * handling code implies lock dependency:
1478 * -> sb_start_pagefault
1480 static inline void sb_start_pagefault(struct super_block
*sb
)
1482 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
, true);
1486 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1487 * @sb: the super we write to
1489 * This is the third level of protection against filesystem freezing. It is
1490 * free for use by a filesystem. The only requirement is that it must rank
1491 * below sb_start_pagefault.
1493 * For example filesystem can call sb_start_intwrite() when starting a
1494 * transaction which somewhat eases handling of freezing for internal sources
1495 * of filesystem changes (internal fs threads, discarding preallocation on file
1498 static inline void sb_start_intwrite(struct super_block
*sb
)
1500 __sb_start_write(sb
, SB_FREEZE_FS
, true);
1504 extern bool inode_owner_or_capable(const struct inode
*inode
);
1507 * VFS helper functions..
1509 extern int vfs_create(struct inode
*, struct dentry
*, umode_t
, bool);
1510 extern int vfs_mkdir(struct inode
*, struct dentry
*, umode_t
);
1511 extern int vfs_mknod(struct inode
*, struct dentry
*, umode_t
, dev_t
);
1512 extern int vfs_symlink(struct inode
*, struct dentry
*, const char *);
1513 extern int vfs_link(struct dentry
*, struct inode
*, struct dentry
*, struct inode
**);
1514 extern int vfs_rmdir(struct inode
*, struct dentry
*);
1515 extern int vfs_unlink(struct inode
*, struct dentry
*, struct inode
**);
1516 extern int vfs_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*, struct inode
**, unsigned int);
1517 extern int vfs_whiteout(struct inode
*, struct dentry
*);
1520 * VFS dentry helper functions.
1522 extern void dentry_unhash(struct dentry
*dentry
);
1525 * VFS file helper functions.
1527 extern void inode_init_owner(struct inode
*inode
, const struct inode
*dir
,
1530 * VFS FS_IOC_FIEMAP helper definitions.
1532 struct fiemap_extent_info
{
1533 unsigned int fi_flags
; /* Flags as passed from user */
1534 unsigned int fi_extents_mapped
; /* Number of mapped extents */
1535 unsigned int fi_extents_max
; /* Size of fiemap_extent array */
1536 struct fiemap_extent __user
*fi_extents_start
; /* Start of
1537 fiemap_extent array */
1539 int fiemap_fill_next_extent(struct fiemap_extent_info
*info
, u64 logical
,
1540 u64 phys
, u64 len
, u32 flags
);
1541 int fiemap_check_flags(struct fiemap_extent_info
*fieinfo
, u32 fs_flags
);
1546 * NOTE! These match bits 12..15 of stat.st_mode
1547 * (ie "(i_mode >> 12) & 15").
1549 #define DT_UNKNOWN 0
1560 * This is the "filldir" function type, used by readdir() to let
1561 * the kernel specify what kind of dirent layout it wants to have.
1562 * This allows the kernel to read directories into kernel space or
1563 * to have different dirent layouts depending on the binary type.
1566 typedef int (*filldir_t
)(struct dir_context
*, const char *, int, loff_t
, u64
,
1569 struct dir_context
{
1570 const filldir_t actor
;
1574 struct block_device_operations
;
1576 /* These macros are for out of kernel modules to test that
1577 * the kernel supports the unlocked_ioctl and compat_ioctl
1578 * fields in struct file_operations. */
1579 #define HAVE_COMPAT_IOCTL 1
1580 #define HAVE_UNLOCKED_IOCTL 1
1583 * These flags let !MMU mmap() govern direct device mapping vs immediate
1584 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1586 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1587 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1588 * NOMMU_MAP_READ: Can be mapped for reading
1589 * NOMMU_MAP_WRITE: Can be mapped for writing
1590 * NOMMU_MAP_EXEC: Can be mapped for execution
1592 #define NOMMU_MAP_COPY 0x00000001
1593 #define NOMMU_MAP_DIRECT 0x00000008
1594 #define NOMMU_MAP_READ VM_MAYREAD
1595 #define NOMMU_MAP_WRITE VM_MAYWRITE
1596 #define NOMMU_MAP_EXEC VM_MAYEXEC
1598 #define NOMMU_VMFLAGS \
1599 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1604 struct file_operations
{
1605 struct module
*owner
;
1606 loff_t (*llseek
) (struct file
*, loff_t
, int);
1607 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
1608 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
1609 ssize_t (*read_iter
) (struct kiocb
*, struct iov_iter
*);
1610 ssize_t (*write_iter
) (struct kiocb
*, struct iov_iter
*);
1611 int (*iterate
) (struct file
*, struct dir_context
*);
1612 unsigned int (*poll
) (struct file
*, struct poll_table_struct
*);
1613 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
1614 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
1615 int (*mmap
) (struct file
*, struct vm_area_struct
*);
1616 int (*open
) (struct inode
*, struct file
*);
1617 int (*flush
) (struct file
*, fl_owner_t id
);
1618 int (*release
) (struct inode
*, struct file
*);
1619 int (*fsync
) (struct file
*, loff_t
, loff_t
, int datasync
);
1620 int (*aio_fsync
) (struct kiocb
*, int datasync
);
1621 int (*fasync
) (int, struct file
*, int);
1622 int (*lock
) (struct file
*, int, struct file_lock
*);
1623 ssize_t (*sendpage
) (struct file
*, struct page
*, int, size_t, loff_t
*, int);
1624 unsigned long (*get_unmapped_area
)(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1625 int (*check_flags
)(int);
1626 int (*flock
) (struct file
*, int, struct file_lock
*);
1627 ssize_t (*splice_write
)(struct pipe_inode_info
*, struct file
*, loff_t
*, size_t, unsigned int);
1628 ssize_t (*splice_read
)(struct file
*, loff_t
*, struct pipe_inode_info
*, size_t, unsigned int);
1629 int (*setlease
)(struct file
*, long, struct file_lock
**, void **);
1630 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
1632 void (*show_fdinfo
)(struct seq_file
*m
, struct file
*f
);
1634 unsigned (*mmap_capabilities
)(struct file
*);
1636 ssize_t (*copy_file_range
)(struct file
*, loff_t
, struct file
*,
1637 loff_t
, size_t, unsigned int);
1638 int (*clone_file_range
)(struct file
*, loff_t
, struct file
*, loff_t
,
1640 ssize_t (*dedupe_file_range
)(struct file
*, u64
, u64
, struct file
*,
1644 struct inode_operations
{
1645 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*, unsigned int);
1646 const char * (*get_link
) (struct dentry
*, struct inode
*, struct delayed_call
*);
1647 int (*permission
) (struct inode
*, int);
1648 struct posix_acl
* (*get_acl
)(struct inode
*, int);
1650 int (*readlink
) (struct dentry
*, char __user
*,int);
1652 int (*create
) (struct inode
*,struct dentry
*, umode_t
, bool);
1653 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
1654 int (*unlink
) (struct inode
*,struct dentry
*);
1655 int (*symlink
) (struct inode
*,struct dentry
*,const char *);
1656 int (*mkdir
) (struct inode
*,struct dentry
*,umode_t
);
1657 int (*rmdir
) (struct inode
*,struct dentry
*);
1658 int (*mknod
) (struct inode
*,struct dentry
*,umode_t
,dev_t
);
1659 int (*rename
) (struct inode
*, struct dentry
*,
1660 struct inode
*, struct dentry
*);
1661 int (*rename2
) (struct inode
*, struct dentry
*,
1662 struct inode
*, struct dentry
*, unsigned int);
1663 int (*setattr
) (struct dentry
*, struct iattr
*);
1664 int (*getattr
) (struct vfsmount
*mnt
, struct dentry
*, struct kstat
*);
1665 int (*setxattr
) (struct dentry
*, const char *,const void *,size_t,int);
1666 ssize_t (*getxattr
) (struct dentry
*, const char *, void *, size_t);
1667 ssize_t (*listxattr
) (struct dentry
*, char *, size_t);
1668 int (*removexattr
) (struct dentry
*, const char *);
1669 int (*fiemap
)(struct inode
*, struct fiemap_extent_info
*, u64 start
,
1671 int (*update_time
)(struct inode
*, struct timespec
*, int);
1672 int (*atomic_open
)(struct inode
*, struct dentry
*,
1673 struct file
*, unsigned open_flag
,
1674 umode_t create_mode
, int *opened
);
1675 int (*tmpfile
) (struct inode
*, struct dentry
*, umode_t
);
1676 int (*set_acl
)(struct inode
*, struct posix_acl
*, int);
1677 } ____cacheline_aligned
;
1679 ssize_t
rw_copy_check_uvector(int type
, const struct iovec __user
* uvector
,
1680 unsigned long nr_segs
, unsigned long fast_segs
,
1681 struct iovec
*fast_pointer
,
1682 struct iovec
**ret_pointer
);
1684 extern ssize_t
__vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1685 extern ssize_t
__vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1686 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1687 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1688 extern ssize_t
vfs_readv(struct file
*, const struct iovec __user
*,
1689 unsigned long, loff_t
*);
1690 extern ssize_t
vfs_writev(struct file
*, const struct iovec __user
*,
1691 unsigned long, loff_t
*);
1692 extern ssize_t
vfs_copy_file_range(struct file
*, loff_t
, struct file
*,
1693 loff_t
, size_t, unsigned int);
1694 extern int vfs_clone_file_range(struct file
*file_in
, loff_t pos_in
,
1695 struct file
*file_out
, loff_t pos_out
, u64 len
);
1696 extern int vfs_dedupe_file_range(struct file
*file
,
1697 struct file_dedupe_range
*same
);
1699 struct super_operations
{
1700 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
1701 void (*destroy_inode
)(struct inode
*);
1703 void (*dirty_inode
) (struct inode
*, int flags
);
1704 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
1705 int (*drop_inode
) (struct inode
*);
1706 void (*evict_inode
) (struct inode
*);
1707 void (*put_super
) (struct super_block
*);
1708 int (*sync_fs
)(struct super_block
*sb
, int wait
);
1709 int (*freeze_super
) (struct super_block
*);
1710 int (*freeze_fs
) (struct super_block
*);
1711 int (*thaw_super
) (struct super_block
*);
1712 int (*unfreeze_fs
) (struct super_block
*);
1713 int (*statfs
) (struct dentry
*, struct kstatfs
*);
1714 int (*remount_fs
) (struct super_block
*, int *, char *);
1715 void (*umount_begin
) (struct super_block
*);
1717 int (*show_options
)(struct seq_file
*, struct dentry
*);
1718 int (*show_devname
)(struct seq_file
*, struct dentry
*);
1719 int (*show_path
)(struct seq_file
*, struct dentry
*);
1720 int (*show_stats
)(struct seq_file
*, struct dentry
*);
1722 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
1723 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
1724 struct dquot
**(*get_dquots
)(struct inode
*);
1726 int (*bdev_try_to_free_page
)(struct super_block
*, struct page
*, gfp_t
);
1727 long (*nr_cached_objects
)(struct super_block
*,
1728 struct shrink_control
*);
1729 long (*free_cached_objects
)(struct super_block
*,
1730 struct shrink_control
*);
1734 * Inode flags - they have no relation to superblock flags now
1736 #define S_SYNC 1 /* Writes are synced at once */
1737 #define S_NOATIME 2 /* Do not update access times */
1738 #define S_APPEND 4 /* Append-only file */
1739 #define S_IMMUTABLE 8 /* Immutable file */
1740 #define S_DEAD 16 /* removed, but still open directory */
1741 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1742 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1743 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1744 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1745 #define S_PRIVATE 512 /* Inode is fs-internal */
1746 #define S_IMA 1024 /* Inode has an associated IMA struct */
1747 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1748 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1749 #ifdef CONFIG_FS_DAX
1750 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1752 #define S_DAX 0 /* Make all the DAX code disappear */
1756 * Note that nosuid etc flags are inode-specific: setting some file-system
1757 * flags just means all the inodes inherit those flags by default. It might be
1758 * possible to override it selectively if you really wanted to with some
1759 * ioctl() that is not currently implemented.
1761 * Exception: MS_RDONLY is always applied to the entire file system.
1763 * Unfortunately, it is possible to change a filesystems flags with it mounted
1764 * with files in use. This means that all of the inodes will not have their
1765 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1766 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1768 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1770 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1771 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1772 ((inode)->i_flags & S_SYNC))
1773 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1774 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1775 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1776 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1777 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1779 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1780 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1781 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1782 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1784 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1785 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1786 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1787 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1788 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1789 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1790 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1791 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1793 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1794 (inode)->i_rdev == WHITEOUT_DEV)
1797 * Inode state bits. Protected by inode->i_lock
1799 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1800 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1802 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1803 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1804 * various stages of removing an inode.
1806 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1808 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1809 * fdatasync(). i_atime is the usual cause.
1810 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1811 * these changes separately from I_DIRTY_SYNC so that we
1812 * don't have to write inode on fdatasync() when only
1813 * mtime has changed in it.
1814 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1815 * I_NEW Serves as both a mutex and completion notification.
1816 * New inodes set I_NEW. If two processes both create
1817 * the same inode, one of them will release its inode and
1818 * wait for I_NEW to be released before returning.
1819 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1820 * also cause waiting on I_NEW, without I_NEW actually
1821 * being set. find_inode() uses this to prevent returning
1822 * nearly-dead inodes.
1823 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1824 * is zero. I_FREEING must be set when I_WILL_FREE is
1826 * I_FREEING Set when inode is about to be freed but still has dirty
1827 * pages or buffers attached or the inode itself is still
1829 * I_CLEAR Added by clear_inode(). In this state the inode is
1830 * clean and can be destroyed. Inode keeps I_FREEING.
1832 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1833 * prohibited for many purposes. iget() must wait for
1834 * the inode to be completely released, then create it
1835 * anew. Other functions will just ignore such inodes,
1836 * if appropriate. I_NEW is used for waiting.
1838 * I_SYNC Writeback of inode is running. The bit is set during
1839 * data writeback, and cleared with a wakeup on the bit
1840 * address once it is done. The bit is also used to pin
1841 * the inode in memory for flusher thread.
1843 * I_REFERENCED Marks the inode as recently references on the LRU list.
1845 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1847 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1848 * synchronize competing switching instances and to tell
1849 * wb stat updates to grab mapping->tree_lock. See
1850 * inode_switch_wb_work_fn() for details.
1852 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1854 #define I_DIRTY_SYNC (1 << 0)
1855 #define I_DIRTY_DATASYNC (1 << 1)
1856 #define I_DIRTY_PAGES (1 << 2)
1858 #define I_NEW (1 << __I_NEW)
1859 #define I_WILL_FREE (1 << 4)
1860 #define I_FREEING (1 << 5)
1861 #define I_CLEAR (1 << 6)
1863 #define I_SYNC (1 << __I_SYNC)
1864 #define I_REFERENCED (1 << 8)
1865 #define __I_DIO_WAKEUP 9
1866 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1867 #define I_LINKABLE (1 << 10)
1868 #define I_DIRTY_TIME (1 << 11)
1869 #define __I_DIRTY_TIME_EXPIRED 12
1870 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1871 #define I_WB_SWITCH (1 << 13)
1873 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1874 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1876 extern void __mark_inode_dirty(struct inode
*, int);
1877 static inline void mark_inode_dirty(struct inode
*inode
)
1879 __mark_inode_dirty(inode
, I_DIRTY
);
1882 static inline void mark_inode_dirty_sync(struct inode
*inode
)
1884 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
1887 extern void inc_nlink(struct inode
*inode
);
1888 extern void drop_nlink(struct inode
*inode
);
1889 extern void clear_nlink(struct inode
*inode
);
1890 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
1892 static inline void inode_inc_link_count(struct inode
*inode
)
1895 mark_inode_dirty(inode
);
1898 static inline void inode_dec_link_count(struct inode
*inode
)
1901 mark_inode_dirty(inode
);
1905 * inode_inc_iversion - increments i_version
1906 * @inode: inode that need to be updated
1908 * Every time the inode is modified, the i_version field will be incremented.
1909 * The filesystem has to be mounted with i_version flag
1912 static inline void inode_inc_iversion(struct inode
*inode
)
1914 spin_lock(&inode
->i_lock
);
1916 spin_unlock(&inode
->i_lock
);
1919 enum file_time_flags
{
1926 extern bool atime_needs_update(const struct path
*, struct inode
*);
1927 extern void touch_atime(const struct path
*);
1928 static inline void file_accessed(struct file
*file
)
1930 if (!(file
->f_flags
& O_NOATIME
))
1931 touch_atime(&file
->f_path
);
1934 int sync_inode(struct inode
*inode
, struct writeback_control
*wbc
);
1935 int sync_inode_metadata(struct inode
*inode
, int wait
);
1937 struct file_system_type
{
1940 #define FS_REQUIRES_DEV 1
1941 #define FS_BINARY_MOUNTDATA 2
1942 #define FS_HAS_SUBTYPE 4
1943 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
1944 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
1945 #define FS_USERNS_VISIBLE 32 /* FS must already be visible */
1946 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
1947 struct dentry
*(*mount
) (struct file_system_type
*, int,
1948 const char *, void *);
1949 void (*kill_sb
) (struct super_block
*);
1950 struct module
*owner
;
1951 struct file_system_type
* next
;
1952 struct hlist_head fs_supers
;
1954 struct lock_class_key s_lock_key
;
1955 struct lock_class_key s_umount_key
;
1956 struct lock_class_key s_vfs_rename_key
;
1957 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
1959 struct lock_class_key i_lock_key
;
1960 struct lock_class_key i_mutex_key
;
1961 struct lock_class_key i_mutex_dir_key
;
1964 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1966 extern struct dentry
*mount_ns(struct file_system_type
*fs_type
, int flags
,
1967 void *data
, int (*fill_super
)(struct super_block
*, void *, int));
1968 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
1969 int flags
, const char *dev_name
, void *data
,
1970 int (*fill_super
)(struct super_block
*, void *, int));
1971 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
1972 int flags
, void *data
,
1973 int (*fill_super
)(struct super_block
*, void *, int));
1974 extern struct dentry
*mount_nodev(struct file_system_type
*fs_type
,
1975 int flags
, void *data
,
1976 int (*fill_super
)(struct super_block
*, void *, int));
1977 extern struct dentry
*mount_subtree(struct vfsmount
*mnt
, const char *path
);
1978 void generic_shutdown_super(struct super_block
*sb
);
1979 void kill_block_super(struct super_block
*sb
);
1980 void kill_anon_super(struct super_block
*sb
);
1981 void kill_litter_super(struct super_block
*sb
);
1982 void deactivate_super(struct super_block
*sb
);
1983 void deactivate_locked_super(struct super_block
*sb
);
1984 int set_anon_super(struct super_block
*s
, void *data
);
1985 int get_anon_bdev(dev_t
*);
1986 void free_anon_bdev(dev_t
);
1987 struct super_block
*sget(struct file_system_type
*type
,
1988 int (*test
)(struct super_block
*,void *),
1989 int (*set
)(struct super_block
*,void *),
1990 int flags
, void *data
);
1991 extern struct dentry
*mount_pseudo(struct file_system_type
*, char *,
1992 const struct super_operations
*ops
,
1993 const struct dentry_operations
*dops
,
1996 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1997 #define fops_get(fops) \
1998 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1999 #define fops_put(fops) \
2000 do { if (fops) module_put((fops)->owner); } while(0)
2002 * This one is to be used *ONLY* from ->open() instances.
2003 * fops must be non-NULL, pinned down *and* module dependencies
2004 * should be sufficient to pin the caller down as well.
2006 #define replace_fops(f, fops) \
2008 struct file *__file = (f); \
2009 fops_put(__file->f_op); \
2010 BUG_ON(!(__file->f_op = (fops))); \
2013 extern int register_filesystem(struct file_system_type
*);
2014 extern int unregister_filesystem(struct file_system_type
*);
2015 extern struct vfsmount
*kern_mount_data(struct file_system_type
*, void *data
);
2016 #define kern_mount(type) kern_mount_data(type, NULL)
2017 extern void kern_unmount(struct vfsmount
*mnt
);
2018 extern int may_umount_tree(struct vfsmount
*);
2019 extern int may_umount(struct vfsmount
*);
2020 extern long do_mount(const char *, const char __user
*,
2021 const char *, unsigned long, void *);
2022 extern struct vfsmount
*collect_mounts(struct path
*);
2023 extern void drop_collected_mounts(struct vfsmount
*);
2024 extern int iterate_mounts(int (*)(struct vfsmount
*, void *), void *,
2026 extern int vfs_statfs(struct path
*, struct kstatfs
*);
2027 extern int user_statfs(const char __user
*, struct kstatfs
*);
2028 extern int fd_statfs(int, struct kstatfs
*);
2029 extern int vfs_ustat(dev_t
, struct kstatfs
*);
2030 extern int freeze_super(struct super_block
*super
);
2031 extern int thaw_super(struct super_block
*super
);
2032 extern bool our_mnt(struct vfsmount
*mnt
);
2034 extern int current_umask(void);
2036 extern void ihold(struct inode
* inode
);
2037 extern void iput(struct inode
*);
2038 extern int generic_update_time(struct inode
*, struct timespec
*, int);
2041 extern struct kobject
*fs_kobj
;
2043 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
2045 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2046 extern int locks_mandatory_locked(struct file
*);
2047 extern int locks_mandatory_area(struct inode
*, struct file
*, loff_t
, loff_t
, unsigned char);
2050 * Candidates for mandatory locking have the setgid bit set
2051 * but no group execute bit - an otherwise meaningless combination.
2054 static inline int __mandatory_lock(struct inode
*ino
)
2056 return (ino
->i_mode
& (S_ISGID
| S_IXGRP
)) == S_ISGID
;
2060 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2061 * otherwise these will be advisory locks
2064 static inline int mandatory_lock(struct inode
*ino
)
2066 return IS_MANDLOCK(ino
) && __mandatory_lock(ino
);
2069 static inline int locks_verify_locked(struct file
*file
)
2071 if (mandatory_lock(file_inode(file
)))
2072 return locks_mandatory_locked(file
);
2076 static inline int locks_verify_truncate(struct inode
*inode
,
2080 if (!inode
->i_flctx
|| !mandatory_lock(inode
))
2083 if (size
< inode
->i_size
) {
2084 return locks_mandatory_area(inode
, f
, size
, inode
->i_size
- 1,
2087 return locks_mandatory_area(inode
, f
, inode
->i_size
, size
- 1,
2092 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2094 static inline int locks_mandatory_locked(struct file
*file
)
2099 static inline int locks_mandatory_area(struct inode
*inode
, struct file
*filp
,
2100 loff_t start
, loff_t end
, unsigned char type
)
2105 static inline int __mandatory_lock(struct inode
*inode
)
2110 static inline int mandatory_lock(struct inode
*inode
)
2115 static inline int locks_verify_locked(struct file
*file
)
2120 static inline int locks_verify_truncate(struct inode
*inode
, struct file
*filp
,
2126 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2129 #ifdef CONFIG_FILE_LOCKING
2130 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2133 * Since this check is lockless, we must ensure that any refcounts
2134 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2135 * could end up racing with tasks trying to set a new lease on this
2139 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2140 return __break_lease(inode
, mode
, FL_LEASE
);
2144 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2147 * Since this check is lockless, we must ensure that any refcounts
2148 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2149 * could end up racing with tasks trying to set a new lease on this
2153 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2154 return __break_lease(inode
, mode
, FL_DELEG
);
2158 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2162 ret
= break_deleg(inode
, O_WRONLY
|O_NONBLOCK
);
2163 if (ret
== -EWOULDBLOCK
&& delegated_inode
) {
2164 *delegated_inode
= inode
;
2170 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2174 ret
= break_deleg(*delegated_inode
, O_WRONLY
);
2175 iput(*delegated_inode
);
2176 *delegated_inode
= NULL
;
2180 static inline int break_layout(struct inode
*inode
, bool wait
)
2183 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2184 return __break_lease(inode
,
2185 wait
? O_WRONLY
: O_WRONLY
| O_NONBLOCK
,
2190 #else /* !CONFIG_FILE_LOCKING */
2191 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2196 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2201 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2206 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2212 static inline int break_layout(struct inode
*inode
, bool wait
)
2217 #endif /* CONFIG_FILE_LOCKING */
2222 const char *name
; /* pointer to actual string */
2223 const __user
char *uptr
; /* original userland pointer */
2224 struct audit_names
*aname
;
2229 extern long vfs_truncate(struct path
*, loff_t
);
2230 extern int do_truncate(struct dentry
*, loff_t start
, unsigned int time_attrs
,
2232 extern int vfs_fallocate(struct file
*file
, int mode
, loff_t offset
,
2234 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
2236 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
2237 extern struct file
*filp_open(const char *, int, umode_t
);
2238 extern struct file
*file_open_root(struct dentry
*, struct vfsmount
*,
2240 extern struct file
* dentry_open(const struct path
*, int, const struct cred
*);
2241 extern int filp_close(struct file
*, fl_owner_t id
);
2243 extern struct filename
*getname_flags(const char __user
*, int, int *);
2244 extern struct filename
*getname(const char __user
*);
2245 extern struct filename
*getname_kernel(const char *);
2246 extern void putname(struct filename
*name
);
2252 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
2253 int (*open
)(struct inode
*, struct file
*),
2255 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
2259 extern int ioctl_preallocate(struct file
*filp
, void __user
*argp
);
2262 extern void __init
vfs_caches_init_early(void);
2263 extern void __init
vfs_caches_init(void);
2265 extern struct kmem_cache
*names_cachep
;
2267 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2268 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2271 extern int register_blkdev(unsigned int, const char *);
2272 extern void unregister_blkdev(unsigned int, const char *);
2273 extern struct block_device
*bdget(dev_t
);
2274 extern struct block_device
*bdgrab(struct block_device
*bdev
);
2275 extern void bd_set_size(struct block_device
*, loff_t size
);
2276 extern void bd_forget(struct inode
*inode
);
2277 extern void bdput(struct block_device
*);
2278 extern void invalidate_bdev(struct block_device
*);
2279 extern void iterate_bdevs(void (*)(struct block_device
*, void *), void *);
2280 extern int sync_blockdev(struct block_device
*bdev
);
2281 extern void kill_bdev(struct block_device
*);
2282 extern struct super_block
*freeze_bdev(struct block_device
*);
2283 extern void emergency_thaw_all(void);
2284 extern int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
);
2285 extern int fsync_bdev(struct block_device
*);
2286 #ifdef CONFIG_FS_DAX
2287 extern bool blkdev_dax_capable(struct block_device
*bdev
);
2289 static inline bool blkdev_dax_capable(struct block_device
*bdev
)
2295 extern struct super_block
*blockdev_superblock
;
2297 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2299 return sb
== blockdev_superblock
;
2302 static inline void bd_forget(struct inode
*inode
) {}
2303 static inline int sync_blockdev(struct block_device
*bdev
) { return 0; }
2304 static inline void kill_bdev(struct block_device
*bdev
) {}
2305 static inline void invalidate_bdev(struct block_device
*bdev
) {}
2307 static inline struct super_block
*freeze_bdev(struct block_device
*sb
)
2312 static inline int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
)
2317 static inline void iterate_bdevs(void (*f
)(struct block_device
*, void *), void *arg
)
2321 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2326 extern int sync_filesystem(struct super_block
*);
2327 extern const struct file_operations def_blk_fops
;
2328 extern const struct file_operations def_chr_fops
;
2330 extern int ioctl_by_bdev(struct block_device
*, unsigned, unsigned long);
2331 extern int blkdev_ioctl(struct block_device
*, fmode_t
, unsigned, unsigned long);
2332 extern long compat_blkdev_ioctl(struct file
*, unsigned, unsigned long);
2333 extern int blkdev_get(struct block_device
*bdev
, fmode_t mode
, void *holder
);
2334 extern struct block_device
*blkdev_get_by_path(const char *path
, fmode_t mode
,
2336 extern struct block_device
*blkdev_get_by_dev(dev_t dev
, fmode_t mode
,
2338 extern void blkdev_put(struct block_device
*bdev
, fmode_t mode
);
2339 extern int __blkdev_reread_part(struct block_device
*bdev
);
2340 extern int blkdev_reread_part(struct block_device
*bdev
);
2343 extern int bd_link_disk_holder(struct block_device
*bdev
, struct gendisk
*disk
);
2344 extern void bd_unlink_disk_holder(struct block_device
*bdev
,
2345 struct gendisk
*disk
);
2347 static inline int bd_link_disk_holder(struct block_device
*bdev
,
2348 struct gendisk
*disk
)
2352 static inline void bd_unlink_disk_holder(struct block_device
*bdev
,
2353 struct gendisk
*disk
)
2360 #define CHRDEV_MAJOR_HASH_SIZE 255
2361 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
2362 extern int register_chrdev_region(dev_t
, unsigned, const char *);
2363 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
2364 unsigned int count
, const char *name
,
2365 const struct file_operations
*fops
);
2366 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
2367 unsigned int count
, const char *name
);
2368 extern void unregister_chrdev_region(dev_t
, unsigned);
2369 extern void chrdev_show(struct seq_file
*,off_t
);
2371 static inline int register_chrdev(unsigned int major
, const char *name
,
2372 const struct file_operations
*fops
)
2374 return __register_chrdev(major
, 0, 256, name
, fops
);
2377 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2379 __unregister_chrdev(major
, 0, 256, name
);
2382 /* fs/block_dev.c */
2383 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2384 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2387 #define BLKDEV_MAJOR_HASH_SIZE 255
2388 extern const char *__bdevname(dev_t
, char *buffer
);
2389 extern const char *bdevname(struct block_device
*bdev
, char *buffer
);
2390 extern struct block_device
*lookup_bdev(const char *);
2391 extern void blkdev_show(struct seq_file
*,off_t
);
2394 #define BLKDEV_MAJOR_HASH_SIZE 0
2397 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2399 /* Invalid inode operations -- fs/bad_inode.c */
2400 extern void make_bad_inode(struct inode
*);
2401 extern bool is_bad_inode(struct inode
*);
2405 * return READ, READA, or WRITE
2407 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2410 * return data direction, READ or WRITE
2412 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2414 extern void check_disk_size_change(struct gendisk
*disk
,
2415 struct block_device
*bdev
);
2416 extern int revalidate_disk(struct gendisk
*);
2417 extern int check_disk_change(struct block_device
*);
2418 extern int __invalidate_device(struct block_device
*, bool);
2419 extern int invalidate_partition(struct gendisk
*, int);
2421 unsigned long invalidate_mapping_pages(struct address_space
*mapping
,
2422 pgoff_t start
, pgoff_t end
);
2424 static inline void invalidate_remote_inode(struct inode
*inode
)
2426 if (S_ISREG(inode
->i_mode
) || S_ISDIR(inode
->i_mode
) ||
2427 S_ISLNK(inode
->i_mode
))
2428 invalidate_mapping_pages(inode
->i_mapping
, 0, -1);
2430 extern int invalidate_inode_pages2(struct address_space
*mapping
);
2431 extern int invalidate_inode_pages2_range(struct address_space
*mapping
,
2432 pgoff_t start
, pgoff_t end
);
2433 extern int write_inode_now(struct inode
*, int);
2434 extern int filemap_fdatawrite(struct address_space
*);
2435 extern int filemap_flush(struct address_space
*);
2436 extern int filemap_fdatawait(struct address_space
*);
2437 extern void filemap_fdatawait_keep_errors(struct address_space
*);
2438 extern int filemap_fdatawait_range(struct address_space
*, loff_t lstart
,
2440 extern int filemap_write_and_wait(struct address_space
*mapping
);
2441 extern int filemap_write_and_wait_range(struct address_space
*mapping
,
2442 loff_t lstart
, loff_t lend
);
2443 extern int __filemap_fdatawrite_range(struct address_space
*mapping
,
2444 loff_t start
, loff_t end
, int sync_mode
);
2445 extern int filemap_fdatawrite_range(struct address_space
*mapping
,
2446 loff_t start
, loff_t end
);
2448 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
2450 extern int vfs_fsync(struct file
*file
, int datasync
);
2451 static inline int generic_write_sync(struct file
*file
, loff_t pos
, loff_t count
)
2453 if (!(file
->f_flags
& O_DSYNC
) && !IS_SYNC(file
->f_mapping
->host
))
2455 return vfs_fsync_range(file
, pos
, pos
+ count
- 1,
2456 (file
->f_flags
& __O_SYNC
) ? 0 : 1);
2458 extern void emergency_sync(void);
2459 extern void emergency_remount(void);
2461 extern sector_t
bmap(struct inode
*, sector_t
);
2463 extern int notify_change(struct dentry
*, struct iattr
*, struct inode
**);
2464 extern int inode_permission(struct inode
*, int);
2465 extern int __inode_permission(struct inode
*, int);
2466 extern int generic_permission(struct inode
*, int);
2467 extern int __check_sticky(struct inode
*dir
, struct inode
*inode
);
2469 static inline bool execute_ok(struct inode
*inode
)
2471 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
2474 static inline void file_start_write(struct file
*file
)
2476 if (!S_ISREG(file_inode(file
)->i_mode
))
2478 __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, true);
2481 static inline bool file_start_write_trylock(struct file
*file
)
2483 if (!S_ISREG(file_inode(file
)->i_mode
))
2485 return __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, false);
2488 static inline void file_end_write(struct file
*file
)
2490 if (!S_ISREG(file_inode(file
)->i_mode
))
2492 __sb_end_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
);
2496 * get_write_access() gets write permission for a file.
2497 * put_write_access() releases this write permission.
2498 * This is used for regular files.
2499 * We cannot support write (and maybe mmap read-write shared) accesses and
2500 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2501 * can have the following values:
2502 * 0: no writers, no VM_DENYWRITE mappings
2503 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2504 * > 0: (i_writecount) users are writing to the file.
2506 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2507 * except for the cases where we don't hold i_writecount yet. Then we need to
2508 * use {get,deny}_write_access() - these functions check the sign and refuse
2509 * to do the change if sign is wrong.
2511 static inline int get_write_access(struct inode
*inode
)
2513 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2515 static inline int deny_write_access(struct file
*file
)
2517 struct inode
*inode
= file_inode(file
);
2518 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2520 static inline void put_write_access(struct inode
* inode
)
2522 atomic_dec(&inode
->i_writecount
);
2524 static inline void allow_write_access(struct file
*file
)
2527 atomic_inc(&file_inode(file
)->i_writecount
);
2529 static inline bool inode_is_open_for_write(const struct inode
*inode
)
2531 return atomic_read(&inode
->i_writecount
) > 0;
2535 static inline void i_readcount_dec(struct inode
*inode
)
2537 BUG_ON(!atomic_read(&inode
->i_readcount
));
2538 atomic_dec(&inode
->i_readcount
);
2540 static inline void i_readcount_inc(struct inode
*inode
)
2542 atomic_inc(&inode
->i_readcount
);
2545 static inline void i_readcount_dec(struct inode
*inode
)
2549 static inline void i_readcount_inc(struct inode
*inode
)
2554 extern int do_pipe_flags(int *, int);
2556 extern int kernel_read(struct file
*, loff_t
, char *, unsigned long);
2557 extern ssize_t
kernel_write(struct file
*, const char *, size_t, loff_t
);
2558 extern ssize_t
__kernel_write(struct file
*, const char *, size_t, loff_t
*);
2559 extern struct file
* open_exec(const char *);
2561 /* fs/dcache.c -- generic fs support functions */
2562 extern bool is_subdir(struct dentry
*, struct dentry
*);
2563 extern bool path_is_under(struct path
*, struct path
*);
2565 extern char *file_path(struct file
*, char *, int);
2567 #include <linux/err.h>
2569 /* needed for stackable file system support */
2570 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int whence
);
2572 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int whence
);
2574 extern int inode_init_always(struct super_block
*, struct inode
*);
2575 extern void inode_init_once(struct inode
*);
2576 extern void address_space_init_once(struct address_space
*mapping
);
2577 extern struct inode
* igrab(struct inode
*);
2578 extern ino_t
iunique(struct super_block
*, ino_t
);
2579 extern int inode_needs_sync(struct inode
*inode
);
2580 extern int generic_delete_inode(struct inode
*inode
);
2581 static inline int generic_drop_inode(struct inode
*inode
)
2583 return !inode
->i_nlink
|| inode_unhashed(inode
);
2586 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
2587 unsigned long hashval
, int (*test
)(struct inode
*, void *),
2589 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
2590 int (*test
)(struct inode
*, void *), void *data
);
2591 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
2593 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
2594 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
2595 extern struct inode
*find_inode_nowait(struct super_block
*,
2597 int (*match
)(struct inode
*,
2598 unsigned long, void *),
2600 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
2601 extern int insert_inode_locked(struct inode
*);
2602 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2603 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
2605 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
2607 extern void unlock_new_inode(struct inode
*);
2608 extern unsigned int get_next_ino(void);
2610 extern void __iget(struct inode
* inode
);
2611 extern void iget_failed(struct inode
*);
2612 extern void clear_inode(struct inode
*);
2613 extern void __destroy_inode(struct inode
*);
2614 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
2615 extern struct inode
*new_inode(struct super_block
*sb
);
2616 extern void free_inode_nonrcu(struct inode
*inode
);
2617 extern int should_remove_suid(struct dentry
*);
2618 extern int file_remove_privs(struct file
*);
2619 extern int dentry_needs_remove_privs(struct dentry
*dentry
);
2620 static inline int file_needs_remove_privs(struct file
*file
)
2622 return dentry_needs_remove_privs(file
->f_path
.dentry
);
2625 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
2626 static inline void insert_inode_hash(struct inode
*inode
)
2628 __insert_inode_hash(inode
, inode
->i_ino
);
2631 extern void __remove_inode_hash(struct inode
*);
2632 static inline void remove_inode_hash(struct inode
*inode
)
2634 if (!inode_unhashed(inode
) && !hlist_fake(&inode
->i_hash
))
2635 __remove_inode_hash(inode
);
2638 extern void inode_sb_list_add(struct inode
*inode
);
2641 extern blk_qc_t
submit_bio(int, struct bio
*);
2642 extern int bdev_read_only(struct block_device
*);
2644 extern int set_blocksize(struct block_device
*, int);
2645 extern int sb_set_blocksize(struct super_block
*, int);
2646 extern int sb_min_blocksize(struct super_block
*, int);
2648 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
2649 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
2650 extern ssize_t
generic_write_checks(struct kiocb
*, struct iov_iter
*);
2651 extern ssize_t
generic_file_read_iter(struct kiocb
*, struct iov_iter
*);
2652 extern ssize_t
__generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2653 extern ssize_t
generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2654 extern ssize_t
generic_file_direct_write(struct kiocb
*, struct iov_iter
*, loff_t
);
2655 extern ssize_t
generic_perform_write(struct file
*, struct iov_iter
*, loff_t
);
2657 ssize_t
vfs_iter_read(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
);
2658 ssize_t
vfs_iter_write(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
);
2660 /* fs/block_dev.c */
2661 extern ssize_t
blkdev_read_iter(struct kiocb
*iocb
, struct iov_iter
*to
);
2662 extern ssize_t
blkdev_write_iter(struct kiocb
*iocb
, struct iov_iter
*from
);
2663 extern int blkdev_fsync(struct file
*filp
, loff_t start
, loff_t end
,
2665 extern void block_sync_page(struct page
*page
);
2668 extern ssize_t
generic_file_splice_read(struct file
*, loff_t
*,
2669 struct pipe_inode_info
*, size_t, unsigned int);
2670 extern ssize_t
default_file_splice_read(struct file
*, loff_t
*,
2671 struct pipe_inode_info
*, size_t, unsigned int);
2672 extern ssize_t
iter_file_splice_write(struct pipe_inode_info
*,
2673 struct file
*, loff_t
*, size_t, unsigned int);
2674 extern ssize_t
generic_splice_sendpage(struct pipe_inode_info
*pipe
,
2675 struct file
*out
, loff_t
*, size_t len
, unsigned int flags
);
2676 extern long do_splice_direct(struct file
*in
, loff_t
*ppos
, struct file
*out
,
2677 loff_t
*opos
, size_t len
, unsigned int flags
);
2681 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
2682 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int whence
);
2683 extern loff_t
no_llseek(struct file
*file
, loff_t offset
, int whence
);
2684 extern loff_t
vfs_setpos(struct file
*file
, loff_t offset
, loff_t maxsize
);
2685 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int whence
);
2686 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
2687 int whence
, loff_t maxsize
, loff_t eof
);
2688 extern loff_t
fixed_size_llseek(struct file
*file
, loff_t offset
,
2689 int whence
, loff_t size
);
2690 extern loff_t
no_seek_end_llseek_size(struct file
*, loff_t
, int, loff_t
);
2691 extern loff_t
no_seek_end_llseek(struct file
*, loff_t
, int);
2692 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
2693 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
2696 typedef void (dio_submit_t
)(int rw
, struct bio
*bio
, struct inode
*inode
,
2697 loff_t file_offset
);
2700 /* need locking between buffered and direct access */
2703 /* filesystem does not support filling holes */
2704 DIO_SKIP_HOLES
= 0x02,
2706 /* filesystem can handle aio writes beyond i_size */
2707 DIO_ASYNC_EXTEND
= 0x04,
2709 /* inode/fs/bdev does not need truncate protection */
2710 DIO_SKIP_DIO_COUNT
= 0x08,
2713 void dio_end_io(struct bio
*bio
, int error
);
2715 ssize_t
__blockdev_direct_IO(struct kiocb
*iocb
, struct inode
*inode
,
2716 struct block_device
*bdev
, struct iov_iter
*iter
,
2717 loff_t offset
, get_block_t get_block
,
2718 dio_iodone_t end_io
, dio_submit_t submit_io
,
2721 static inline ssize_t
blockdev_direct_IO(struct kiocb
*iocb
,
2722 struct inode
*inode
,
2723 struct iov_iter
*iter
, loff_t offset
,
2724 get_block_t get_block
)
2726 return __blockdev_direct_IO(iocb
, inode
, inode
->i_sb
->s_bdev
, iter
,
2727 offset
, get_block
, NULL
, NULL
,
2728 DIO_LOCKING
| DIO_SKIP_HOLES
);
2732 void inode_dio_wait(struct inode
*inode
);
2735 * inode_dio_begin - signal start 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_begin(struct inode
*inode
)
2743 atomic_inc(&inode
->i_dio_count
);
2747 * inode_dio_end - signal finish of a direct I/O requests
2748 * @inode: inode the direct I/O happens on
2750 * This is called once we've finished processing a direct I/O request,
2751 * and is used to wake up callers waiting for direct I/O to be quiesced.
2753 static inline void inode_dio_end(struct inode
*inode
)
2755 if (atomic_dec_and_test(&inode
->i_dio_count
))
2756 wake_up_bit(&inode
->i_state
, __I_DIO_WAKEUP
);
2759 extern void inode_set_flags(struct inode
*inode
, unsigned int flags
,
2762 extern const struct file_operations generic_ro_fops
;
2764 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2766 extern int readlink_copy(char __user
*, int, const char *);
2767 extern int page_readlink(struct dentry
*, char __user
*, int);
2768 extern const char *page_get_link(struct dentry
*, struct inode
*,
2769 struct delayed_call
*);
2770 extern void page_put_link(void *);
2771 extern int __page_symlink(struct inode
*inode
, const char *symname
, int len
,
2773 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
2774 extern const struct inode_operations page_symlink_inode_operations
;
2775 extern void kfree_link(void *);
2776 extern int generic_readlink(struct dentry
*, char __user
*, int);
2777 extern void generic_fillattr(struct inode
*, struct kstat
*);
2778 int vfs_getattr_nosec(struct path
*path
, struct kstat
*stat
);
2779 extern int vfs_getattr(struct path
*, struct kstat
*);
2780 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2781 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2782 void __inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2783 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2784 loff_t
inode_get_bytes(struct inode
*inode
);
2785 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
2786 const char *simple_get_link(struct dentry
*, struct inode
*,
2787 struct delayed_call
*);
2788 extern const struct inode_operations simple_symlink_inode_operations
;
2790 extern int iterate_dir(struct file
*, struct dir_context
*);
2792 extern int vfs_stat(const char __user
*, struct kstat
*);
2793 extern int vfs_lstat(const char __user
*, struct kstat
*);
2794 extern int vfs_fstat(unsigned int, struct kstat
*);
2795 extern int vfs_fstatat(int , const char __user
*, struct kstat
*, int);
2797 extern int __generic_block_fiemap(struct inode
*inode
,
2798 struct fiemap_extent_info
*fieinfo
,
2799 loff_t start
, loff_t len
,
2800 get_block_t
*get_block
);
2801 extern int generic_block_fiemap(struct inode
*inode
,
2802 struct fiemap_extent_info
*fieinfo
, u64 start
,
2803 u64 len
, get_block_t
*get_block
);
2805 extern void get_filesystem(struct file_system_type
*fs
);
2806 extern void put_filesystem(struct file_system_type
*fs
);
2807 extern struct file_system_type
*get_fs_type(const char *name
);
2808 extern struct super_block
*get_super(struct block_device
*);
2809 extern struct super_block
*get_super_thawed(struct block_device
*);
2810 extern struct super_block
*get_active_super(struct block_device
*bdev
);
2811 extern void drop_super(struct super_block
*sb
);
2812 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
2813 extern void iterate_supers_type(struct file_system_type
*,
2814 void (*)(struct super_block
*, void *), void *);
2816 extern int dcache_dir_open(struct inode
*, struct file
*);
2817 extern int dcache_dir_close(struct inode
*, struct file
*);
2818 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
2819 extern int dcache_readdir(struct file
*, struct dir_context
*);
2820 extern int simple_setattr(struct dentry
*, struct iattr
*);
2821 extern int simple_getattr(struct vfsmount
*, struct dentry
*, struct kstat
*);
2822 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
2823 extern int simple_open(struct inode
*inode
, struct file
*file
);
2824 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
2825 extern int simple_unlink(struct inode
*, struct dentry
*);
2826 extern int simple_rmdir(struct inode
*, struct dentry
*);
2827 extern int simple_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*);
2828 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
2829 extern int simple_empty(struct dentry
*);
2830 extern int simple_readpage(struct file
*file
, struct page
*page
);
2831 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
2832 loff_t pos
, unsigned len
, unsigned flags
,
2833 struct page
**pagep
, void **fsdata
);
2834 extern int simple_write_end(struct file
*file
, struct address_space
*mapping
,
2835 loff_t pos
, unsigned len
, unsigned copied
,
2836 struct page
*page
, void *fsdata
);
2837 extern int always_delete_dentry(const struct dentry
*);
2838 extern struct inode
*alloc_anon_inode(struct super_block
*);
2839 extern int simple_nosetlease(struct file
*, long, struct file_lock
**, void **);
2840 extern const struct dentry_operations simple_dentry_operations
;
2842 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
2843 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
2844 extern const struct file_operations simple_dir_operations
;
2845 extern const struct inode_operations simple_dir_inode_operations
;
2846 extern void make_empty_dir_inode(struct inode
*inode
);
2847 extern bool is_empty_dir_inode(struct inode
*inode
);
2848 struct tree_descr
{ char *name
; const struct file_operations
*ops
; int mode
; };
2849 struct dentry
*d_alloc_name(struct dentry
*, const char *);
2850 extern int simple_fill_super(struct super_block
*, unsigned long, struct tree_descr
*);
2851 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
2852 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
2854 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
2855 loff_t
*ppos
, const void *from
, size_t available
);
2856 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
2857 const void __user
*from
, size_t count
);
2859 extern int __generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
2860 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
2862 extern int generic_check_addressable(unsigned, u64
);
2864 #ifdef CONFIG_MIGRATION
2865 extern int buffer_migrate_page(struct address_space
*,
2866 struct page
*, struct page
*,
2869 #define buffer_migrate_page NULL
2872 extern int inode_change_ok(const struct inode
*, struct iattr
*);
2873 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
2874 extern void setattr_copy(struct inode
*inode
, const struct iattr
*attr
);
2876 extern int file_update_time(struct file
*file
);
2878 extern int generic_show_options(struct seq_file
*m
, struct dentry
*root
);
2879 extern void save_mount_options(struct super_block
*sb
, char *options
);
2880 extern void replace_mount_options(struct super_block
*sb
, char *options
);
2882 static inline bool io_is_direct(struct file
*filp
)
2884 return (filp
->f_flags
& O_DIRECT
) || IS_DAX(file_inode(filp
));
2887 static inline int iocb_flags(struct file
*file
)
2890 if (file
->f_flags
& O_APPEND
)
2892 if (io_is_direct(file
))
2897 static inline ino_t
parent_ino(struct dentry
*dentry
)
2902 * Don't strictly need d_lock here? If the parent ino could change
2903 * then surely we'd have a deeper race in the caller?
2905 spin_lock(&dentry
->d_lock
);
2906 res
= dentry
->d_parent
->d_inode
->i_ino
;
2907 spin_unlock(&dentry
->d_lock
);
2911 /* Transaction based IO helpers */
2914 * An argresp is stored in an allocated page and holds the
2915 * size of the argument or response, along with its content
2917 struct simple_transaction_argresp
{
2922 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2924 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
2926 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
2927 size_t size
, loff_t
*pos
);
2928 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
2930 void simple_transaction_set(struct file
*file
, size_t n
);
2933 * simple attribute files
2935 * These attributes behave similar to those in sysfs:
2937 * Writing to an attribute immediately sets a value, an open file can be
2938 * written to multiple times.
2940 * Reading from an attribute creates a buffer from the value that might get
2941 * read with multiple read calls. When the attribute has been read
2942 * completely, no further read calls are possible until the file is opened
2945 * All attributes contain a text representation of a numeric value
2946 * that are accessed with the get() and set() functions.
2948 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2949 static int __fops ## _open(struct inode *inode, struct file *file) \
2951 __simple_attr_check_format(__fmt, 0ull); \
2952 return simple_attr_open(inode, file, __get, __set, __fmt); \
2954 static const struct file_operations __fops = { \
2955 .owner = THIS_MODULE, \
2956 .open = __fops ## _open, \
2957 .release = simple_attr_release, \
2958 .read = simple_attr_read, \
2959 .write = simple_attr_write, \
2960 .llseek = generic_file_llseek, \
2963 static inline __printf(1, 2)
2964 void __simple_attr_check_format(const char *fmt
, ...)
2966 /* don't do anything, just let the compiler check the arguments; */
2969 int simple_attr_open(struct inode
*inode
, struct file
*file
,
2970 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
2972 int simple_attr_release(struct inode
*inode
, struct file
*file
);
2973 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
2974 size_t len
, loff_t
*ppos
);
2975 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
2976 size_t len
, loff_t
*ppos
);
2979 int proc_nr_files(struct ctl_table
*table
, int write
,
2980 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
2981 int proc_nr_dentry(struct ctl_table
*table
, int write
,
2982 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
2983 int proc_nr_inodes(struct ctl_table
*table
, int write
,
2984 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
2985 int __init
get_filesystem_list(char *buf
);
2987 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2988 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2990 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2991 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2992 (flag & __FMODE_NONOTIFY)))
2994 static inline bool is_sxid(umode_t mode
)
2996 return (mode
& S_ISUID
) || ((mode
& S_ISGID
) && (mode
& S_IXGRP
));
2999 static inline int check_sticky(struct inode
*dir
, struct inode
*inode
)
3001 if (!(dir
->i_mode
& S_ISVTX
))
3004 return __check_sticky(dir
, inode
);
3007 static inline void inode_has_no_xattr(struct inode
*inode
)
3009 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& MS_NOSEC
))
3010 inode
->i_flags
|= S_NOSEC
;
3013 static inline bool is_root_inode(struct inode
*inode
)
3015 return inode
== inode
->i_sb
->s_root
->d_inode
;
3018 static inline bool dir_emit(struct dir_context
*ctx
,
3019 const char *name
, int namelen
,
3020 u64 ino
, unsigned type
)
3022 return ctx
->actor(ctx
, name
, namelen
, ctx
->pos
, ino
, type
) == 0;
3024 static inline bool dir_emit_dot(struct file
*file
, struct dir_context
*ctx
)
3026 return ctx
->actor(ctx
, ".", 1, ctx
->pos
,
3027 file
->f_path
.dentry
->d_inode
->i_ino
, DT_DIR
) == 0;
3029 static inline bool dir_emit_dotdot(struct file
*file
, struct dir_context
*ctx
)
3031 return ctx
->actor(ctx
, "..", 2, ctx
->pos
,
3032 parent_ino(file
->f_path
.dentry
), DT_DIR
) == 0;
3034 static inline bool dir_emit_dots(struct file
*file
, struct dir_context
*ctx
)
3036 if (ctx
->pos
== 0) {
3037 if (!dir_emit_dot(file
, ctx
))
3041 if (ctx
->pos
== 1) {
3042 if (!dir_emit_dotdot(file
, ctx
))
3048 static inline bool dir_relax(struct inode
*inode
)
3050 mutex_unlock(&inode
->i_mutex
);
3051 mutex_lock(&inode
->i_mutex
);
3052 return !IS_DEADDIR(inode
);
3055 extern bool path_noexec(const struct path
*path
);
3056 extern void inode_nohighmem(struct inode
*inode
);
3058 #endif /* _LINUX_FS_H */