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 /* number of shadow or DAX exceptional entries */
437 unsigned long nrexceptional
;
438 pgoff_t writeback_index
;/* writeback starts here */
439 const struct address_space_operations
*a_ops
; /* methods */
440 unsigned long flags
; /* error bits/gfp mask */
441 spinlock_t private_lock
; /* for use by the address_space */
442 struct list_head private_list
; /* ditto */
443 void *private_data
; /* ditto */
444 } __attribute__((aligned(sizeof(long))));
446 * On most architectures that alignment is already the case; but
447 * must be enforced here for CRIS, to let the least significant bit
448 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
450 struct request_queue
;
452 struct block_device
{
453 dev_t bd_dev
; /* not a kdev_t - it's a search key */
455 struct inode
* bd_inode
; /* will die */
456 struct super_block
* bd_super
;
457 struct mutex bd_mutex
; /* open/close mutex */
458 struct list_head bd_inodes
;
462 bool bd_write_holder
;
464 struct list_head bd_holder_disks
;
466 struct block_device
* bd_contains
;
467 unsigned bd_block_size
;
468 struct hd_struct
* bd_part
;
469 /* number of times partitions within this device have been opened. */
470 unsigned bd_part_count
;
472 struct gendisk
* bd_disk
;
473 struct request_queue
* bd_queue
;
474 struct list_head bd_list
;
476 * Private data. You must have bd_claim'ed the block_device
477 * to use this. NOTE: bd_claim allows an owner to claim
478 * the same device multiple times, the owner must take special
479 * care to not mess up bd_private for that case.
481 unsigned long bd_private
;
483 /* The counter of freeze processes */
484 int bd_fsfreeze_count
;
485 /* Mutex for freeze */
486 struct mutex bd_fsfreeze_mutex
;
490 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
493 #define PAGECACHE_TAG_DIRTY 0
494 #define PAGECACHE_TAG_WRITEBACK 1
495 #define PAGECACHE_TAG_TOWRITE 2
497 int mapping_tagged(struct address_space
*mapping
, int tag
);
499 static inline void i_mmap_lock_write(struct address_space
*mapping
)
501 down_write(&mapping
->i_mmap_rwsem
);
504 static inline void i_mmap_unlock_write(struct address_space
*mapping
)
506 up_write(&mapping
->i_mmap_rwsem
);
509 static inline void i_mmap_lock_read(struct address_space
*mapping
)
511 down_read(&mapping
->i_mmap_rwsem
);
514 static inline void i_mmap_unlock_read(struct address_space
*mapping
)
516 up_read(&mapping
->i_mmap_rwsem
);
520 * Might pages of this file be mapped into userspace?
522 static inline int mapping_mapped(struct address_space
*mapping
)
524 return !RB_EMPTY_ROOT(&mapping
->i_mmap
);
528 * Might pages of this file have been modified in userspace?
529 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
530 * marks vma as VM_SHARED if it is shared, and the file was opened for
531 * writing i.e. vma may be mprotected writable even if now readonly.
533 * If i_mmap_writable is negative, no new writable mappings are allowed. You
534 * can only deny writable mappings, if none exists right now.
536 static inline int mapping_writably_mapped(struct address_space
*mapping
)
538 return atomic_read(&mapping
->i_mmap_writable
) > 0;
541 static inline int mapping_map_writable(struct address_space
*mapping
)
543 return atomic_inc_unless_negative(&mapping
->i_mmap_writable
) ?
547 static inline void mapping_unmap_writable(struct address_space
*mapping
)
549 atomic_dec(&mapping
->i_mmap_writable
);
552 static inline int mapping_deny_writable(struct address_space
*mapping
)
554 return atomic_dec_unless_positive(&mapping
->i_mmap_writable
) ?
558 static inline void mapping_allow_writable(struct address_space
*mapping
)
560 atomic_inc(&mapping
->i_mmap_writable
);
564 * Use sequence counter to get consistent i_size on 32-bit processors.
566 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
567 #include <linux/seqlock.h>
568 #define __NEED_I_SIZE_ORDERED
569 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
571 #define i_size_ordered_init(inode) do { } while (0)
575 #define ACL_NOT_CACHED ((void *)(-1))
577 #define IOP_FASTPERM 0x0001
578 #define IOP_LOOKUP 0x0002
579 #define IOP_NOFOLLOW 0x0004
582 * Keep mostly read-only and often accessed (especially for
583 * the RCU path lookup and 'stat' data) fields at the beginning
584 * of the 'struct inode'
588 unsigned short i_opflags
;
591 unsigned int i_flags
;
593 #ifdef CONFIG_FS_POSIX_ACL
594 struct posix_acl
*i_acl
;
595 struct posix_acl
*i_default_acl
;
598 const struct inode_operations
*i_op
;
599 struct super_block
*i_sb
;
600 struct address_space
*i_mapping
;
602 #ifdef CONFIG_SECURITY
606 /* Stat data, not accessed from path walking */
609 * Filesystems may only read i_nlink directly. They shall use the
610 * following functions for modification:
612 * (set|clear|inc|drop)_nlink
613 * inode_(inc|dec)_link_count
616 const unsigned int i_nlink
;
617 unsigned int __i_nlink
;
621 struct timespec i_atime
;
622 struct timespec i_mtime
;
623 struct timespec i_ctime
;
624 spinlock_t i_lock
; /* i_blocks, i_bytes, maybe i_size */
625 unsigned short i_bytes
;
626 unsigned int i_blkbits
;
629 #ifdef __NEED_I_SIZE_ORDERED
630 seqcount_t i_size_seqcount
;
634 unsigned long i_state
;
635 struct mutex i_mutex
;
637 unsigned long dirtied_when
; /* jiffies of first dirtying */
638 unsigned long dirtied_time_when
;
640 struct hlist_node i_hash
;
641 struct list_head i_io_list
; /* backing dev IO list */
642 #ifdef CONFIG_CGROUP_WRITEBACK
643 struct bdi_writeback
*i_wb
; /* the associated cgroup wb */
645 /* foreign inode detection, see wbc_detach_inode() */
647 u16 i_wb_frn_avg_time
;
648 u16 i_wb_frn_history
;
650 struct list_head i_lru
; /* inode LRU list */
651 struct list_head i_sb_list
;
653 struct hlist_head i_dentry
;
654 struct rcu_head i_rcu
;
658 atomic_t i_dio_count
;
659 atomic_t i_writecount
;
661 atomic_t i_readcount
; /* struct files open RO */
663 const struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
664 struct file_lock_context
*i_flctx
;
665 struct address_space i_data
;
666 struct list_head i_devices
;
668 struct pipe_inode_info
*i_pipe
;
669 struct block_device
*i_bdev
;
676 #ifdef CONFIG_FSNOTIFY
677 __u32 i_fsnotify_mask
; /* all events this inode cares about */
678 struct hlist_head i_fsnotify_marks
;
681 void *i_private
; /* fs or device private pointer */
684 static inline int inode_unhashed(struct inode
*inode
)
686 return hlist_unhashed(&inode
->i_hash
);
690 * inode->i_mutex nesting subclasses for the lock validator:
692 * 0: the object of the current VFS operation
696 * 4: second non-directory
697 * 5: second parent (when locking independent directories in rename)
699 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
700 * non-directories at once.
702 * The locking order between these classes is
703 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
705 enum inode_i_mutex_lock_class
715 static inline void inode_lock(struct inode
*inode
)
717 mutex_lock(&inode
->i_mutex
);
720 static inline void inode_unlock(struct inode
*inode
)
722 mutex_unlock(&inode
->i_mutex
);
725 static inline int inode_trylock(struct inode
*inode
)
727 return mutex_trylock(&inode
->i_mutex
);
730 static inline int inode_is_locked(struct inode
*inode
)
732 return mutex_is_locked(&inode
->i_mutex
);
735 static inline void inode_lock_nested(struct inode
*inode
, unsigned subclass
)
737 mutex_lock_nested(&inode
->i_mutex
, subclass
);
740 void lock_two_nondirectories(struct inode
*, struct inode
*);
741 void unlock_two_nondirectories(struct inode
*, struct inode
*);
744 * NOTE: in a 32bit arch with a preemptable kernel and
745 * an UP compile the i_size_read/write must be atomic
746 * with respect to the local cpu (unlike with preempt disabled),
747 * but they don't need to be atomic with respect to other cpus like in
748 * true SMP (so they need either to either locally disable irq around
749 * the read or for example on x86 they can be still implemented as a
750 * cmpxchg8b without the need of the lock prefix). For SMP compiles
751 * and 64bit archs it makes no difference if preempt is enabled or not.
753 static inline loff_t
i_size_read(const struct inode
*inode
)
755 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
760 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
761 i_size
= inode
->i_size
;
762 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
764 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
768 i_size
= inode
->i_size
;
772 return inode
->i_size
;
777 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
778 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
779 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
781 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
783 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
785 write_seqcount_begin(&inode
->i_size_seqcount
);
786 inode
->i_size
= i_size
;
787 write_seqcount_end(&inode
->i_size_seqcount
);
789 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
791 inode
->i_size
= i_size
;
794 inode
->i_size
= i_size
;
798 /* Helper functions so that in most cases filesystems will
799 * not need to deal directly with kuid_t and kgid_t and can
800 * instead deal with the raw numeric values that are stored
803 static inline uid_t
i_uid_read(const struct inode
*inode
)
805 return from_kuid(&init_user_ns
, inode
->i_uid
);
808 static inline gid_t
i_gid_read(const struct inode
*inode
)
810 return from_kgid(&init_user_ns
, inode
->i_gid
);
813 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
815 inode
->i_uid
= make_kuid(&init_user_ns
, uid
);
818 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
820 inode
->i_gid
= make_kgid(&init_user_ns
, gid
);
823 static inline unsigned iminor(const struct inode
*inode
)
825 return MINOR(inode
->i_rdev
);
828 static inline unsigned imajor(const struct inode
*inode
)
830 return MAJOR(inode
->i_rdev
);
833 extern struct block_device
*I_BDEV(struct inode
*inode
);
836 rwlock_t lock
; /* protects pid, uid, euid fields */
837 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
838 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
839 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
840 int signum
; /* posix.1b rt signal to be delivered on IO */
844 * Track a single file's readahead state
846 struct file_ra_state
{
847 pgoff_t start
; /* where readahead started */
848 unsigned int size
; /* # of readahead pages */
849 unsigned int async_size
; /* do asynchronous readahead when
850 there are only # of pages ahead */
852 unsigned int ra_pages
; /* Maximum readahead window */
853 unsigned int mmap_miss
; /* Cache miss stat for mmap accesses */
854 loff_t prev_pos
; /* Cache last read() position */
858 * Check if @index falls in the readahead windows.
860 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
862 return (index
>= ra
->start
&&
863 index
< ra
->start
+ ra
->size
);
868 struct llist_node fu_llist
;
869 struct rcu_head fu_rcuhead
;
872 struct inode
*f_inode
; /* cached value */
873 const struct file_operations
*f_op
;
876 * Protects f_ep_links, f_flags.
877 * Must not be taken from IRQ context.
880 atomic_long_t f_count
;
881 unsigned int f_flags
;
883 struct mutex f_pos_lock
;
885 struct fown_struct f_owner
;
886 const struct cred
*f_cred
;
887 struct file_ra_state f_ra
;
890 #ifdef CONFIG_SECURITY
893 /* needed for tty driver, and maybe others */
897 /* Used by fs/eventpoll.c to link all the hooks to this file */
898 struct list_head f_ep_links
;
899 struct list_head f_tfile_llink
;
900 #endif /* #ifdef CONFIG_EPOLL */
901 struct address_space
*f_mapping
;
902 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
907 /* file identifier */
908 unsigned char f_handle
[0];
911 static inline struct file
*get_file(struct file
*f
)
913 atomic_long_inc(&f
->f_count
);
916 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
917 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
918 #define file_count(x) atomic_long_read(&(x)->f_count)
920 #define MAX_NON_LFS ((1UL<<31) - 1)
922 /* Page cache limit. The filesystems should put that into their s_maxbytes
923 limits, otherwise bad things can happen in VM. */
924 #if BITS_PER_LONG==32
925 #define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
926 #elif BITS_PER_LONG==64
927 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
932 #define FL_DELEG 4 /* NFSv4 delegation */
933 #define FL_ACCESS 8 /* not trying to lock, just looking */
934 #define FL_EXISTS 16 /* when unlocking, test for existence */
935 #define FL_LEASE 32 /* lease held on this file */
936 #define FL_CLOSE 64 /* unlock on close */
937 #define FL_SLEEP 128 /* A blocking lock */
938 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
939 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
940 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
941 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
944 * Special return value from posix_lock_file() and vfs_lock_file() for
945 * asynchronous locking.
947 #define FILE_LOCK_DEFERRED 1
949 /* legacy typedef, should eventually be removed */
950 typedef void *fl_owner_t
;
954 struct file_lock_operations
{
955 void (*fl_copy_lock
)(struct file_lock
*, struct file_lock
*);
956 void (*fl_release_private
)(struct file_lock
*);
959 struct lock_manager_operations
{
960 int (*lm_compare_owner
)(struct file_lock
*, struct file_lock
*);
961 unsigned long (*lm_owner_key
)(struct file_lock
*);
962 fl_owner_t (*lm_get_owner
)(fl_owner_t
);
963 void (*lm_put_owner
)(fl_owner_t
);
964 void (*lm_notify
)(struct file_lock
*); /* unblock callback */
965 int (*lm_grant
)(struct file_lock
*, int);
966 bool (*lm_break
)(struct file_lock
*);
967 int (*lm_change
)(struct file_lock
*, int, struct list_head
*);
968 void (*lm_setup
)(struct file_lock
*, void **);
971 struct lock_manager
{
972 struct list_head list
;
974 * NFSv4 and up also want opens blocked during the grace period;
981 void locks_start_grace(struct net
*, struct lock_manager
*);
982 void locks_end_grace(struct lock_manager
*);
983 int locks_in_grace(struct net
*);
984 int opens_in_grace(struct net
*);
986 /* that will die - we need it for nfs_lock_info */
987 #include <linux/nfs_fs_i.h>
990 * struct file_lock represents a generic "file lock". It's used to represent
991 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
992 * note that the same struct is used to represent both a request for a lock and
993 * the lock itself, but the same object is never used for both.
995 * FIXME: should we create a separate "struct lock_request" to help distinguish
998 * The varous i_flctx lists are ordered by:
1001 * 2) lock range start
1004 * Obviously, the last two criteria only matter for POSIX locks.
1007 struct file_lock
*fl_next
; /* singly linked list for this inode */
1008 struct list_head fl_list
; /* link into file_lock_context */
1009 struct hlist_node fl_link
; /* node in global lists */
1010 struct list_head fl_block
; /* circular list of blocked processes */
1011 fl_owner_t fl_owner
;
1012 unsigned int fl_flags
;
1013 unsigned char fl_type
;
1014 unsigned int fl_pid
;
1015 int fl_link_cpu
; /* what cpu's list is this on? */
1016 struct pid
*fl_nspid
;
1017 wait_queue_head_t fl_wait
;
1018 struct file
*fl_file
;
1022 struct fasync_struct
* fl_fasync
; /* for lease break notifications */
1023 /* for lease breaks: */
1024 unsigned long fl_break_time
;
1025 unsigned long fl_downgrade_time
;
1027 const struct file_lock_operations
*fl_ops
; /* Callbacks for filesystems */
1028 const struct lock_manager_operations
*fl_lmops
; /* Callbacks for lockmanagers */
1030 struct nfs_lock_info nfs_fl
;
1031 struct nfs4_lock_info nfs4_fl
;
1033 struct list_head link
; /* link in AFS vnode's pending_locks list */
1034 int state
; /* state of grant or error if -ve */
1039 struct file_lock_context
{
1040 spinlock_t flc_lock
;
1041 struct list_head flc_flock
;
1042 struct list_head flc_posix
;
1043 struct list_head flc_lease
;
1046 /* The following constant reflects the upper bound of the file/locking space */
1048 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1049 #define OFFSET_MAX INT_LIMIT(loff_t)
1050 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1053 #include <linux/fcntl.h>
1055 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
1057 #ifdef CONFIG_FILE_LOCKING
1058 extern int fcntl_getlk(struct file
*, unsigned int, struct flock __user
*);
1059 extern int fcntl_setlk(unsigned int, struct file
*, unsigned int,
1060 struct flock __user
*);
1062 #if BITS_PER_LONG == 32
1063 extern int fcntl_getlk64(struct file
*, unsigned int, struct flock64 __user
*);
1064 extern int fcntl_setlk64(unsigned int, struct file
*, unsigned int,
1065 struct flock64 __user
*);
1068 extern int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
);
1069 extern int fcntl_getlease(struct file
*filp
);
1072 void locks_free_lock_context(struct inode
*inode
);
1073 void locks_free_lock(struct file_lock
*fl
);
1074 extern void locks_init_lock(struct file_lock
*);
1075 extern struct file_lock
* locks_alloc_lock(void);
1076 extern void locks_copy_lock(struct file_lock
*, struct file_lock
*);
1077 extern void locks_copy_conflock(struct file_lock
*, struct file_lock
*);
1078 extern void locks_remove_posix(struct file
*, fl_owner_t
);
1079 extern void locks_remove_file(struct file
*);
1080 extern void locks_release_private(struct file_lock
*);
1081 extern void posix_test_lock(struct file
*, struct file_lock
*);
1082 extern int posix_lock_file(struct file
*, struct file_lock
*, struct file_lock
*);
1083 extern int posix_unblock_lock(struct file_lock
*);
1084 extern int vfs_test_lock(struct file
*, struct file_lock
*);
1085 extern int vfs_lock_file(struct file
*, unsigned int, struct file_lock
*, struct file_lock
*);
1086 extern int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
);
1087 extern int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
);
1088 extern int __break_lease(struct inode
*inode
, unsigned int flags
, unsigned int type
);
1089 extern void lease_get_mtime(struct inode
*, struct timespec
*time
);
1090 extern int generic_setlease(struct file
*, long, struct file_lock
**, void **priv
);
1091 extern int vfs_setlease(struct file
*, long, struct file_lock
**, void **);
1092 extern int lease_modify(struct file_lock
*, int, struct list_head
*);
1093 struct files_struct
;
1094 extern void show_fd_locks(struct seq_file
*f
,
1095 struct file
*filp
, struct files_struct
*files
);
1096 #else /* !CONFIG_FILE_LOCKING */
1097 static inline int fcntl_getlk(struct file
*file
, unsigned int cmd
,
1098 struct flock __user
*user
)
1103 static inline int fcntl_setlk(unsigned int fd
, struct file
*file
,
1104 unsigned int cmd
, struct flock __user
*user
)
1109 #if BITS_PER_LONG == 32
1110 static inline int fcntl_getlk64(struct file
*file
, unsigned int cmd
,
1111 struct flock64 __user
*user
)
1116 static inline int fcntl_setlk64(unsigned int fd
, struct file
*file
,
1117 unsigned int cmd
, struct flock64 __user
*user
)
1122 static inline int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
)
1127 static inline int fcntl_getlease(struct file
*filp
)
1133 locks_free_lock_context(struct inode
*inode
)
1137 static inline void locks_init_lock(struct file_lock
*fl
)
1142 static inline void locks_copy_conflock(struct file_lock
*new, struct file_lock
*fl
)
1147 static inline void locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
1152 static inline void locks_remove_posix(struct file
*filp
, fl_owner_t owner
)
1157 static inline void locks_remove_file(struct file
*filp
)
1162 static inline void posix_test_lock(struct file
*filp
, struct file_lock
*fl
)
1167 static inline int posix_lock_file(struct file
*filp
, struct file_lock
*fl
,
1168 struct file_lock
*conflock
)
1173 static inline int posix_unblock_lock(struct file_lock
*waiter
)
1178 static inline int vfs_test_lock(struct file
*filp
, struct file_lock
*fl
)
1183 static inline int vfs_lock_file(struct file
*filp
, unsigned int cmd
,
1184 struct file_lock
*fl
, struct file_lock
*conf
)
1189 static inline int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
)
1194 static inline int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
)
1199 static inline int __break_lease(struct inode
*inode
, unsigned int mode
, unsigned int type
)
1204 static inline void lease_get_mtime(struct inode
*inode
, struct timespec
*time
)
1209 static inline int generic_setlease(struct file
*filp
, long arg
,
1210 struct file_lock
**flp
, void **priv
)
1215 static inline int vfs_setlease(struct file
*filp
, long arg
,
1216 struct file_lock
**lease
, void **priv
)
1221 static inline int lease_modify(struct file_lock
*fl
, int arg
,
1222 struct list_head
*dispose
)
1227 struct files_struct
;
1228 static inline void show_fd_locks(struct seq_file
*f
,
1229 struct file
*filp
, struct files_struct
*files
) {}
1230 #endif /* !CONFIG_FILE_LOCKING */
1232 static inline struct inode
*file_inode(const struct file
*f
)
1237 static inline int locks_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1239 return locks_lock_inode_wait(file_inode(filp
), fl
);
1242 struct fasync_struct
{
1246 struct fasync_struct
*fa_next
; /* singly linked list */
1247 struct file
*fa_file
;
1248 struct rcu_head fa_rcu
;
1251 #define FASYNC_MAGIC 0x4601
1253 /* SMP safe fasync helpers: */
1254 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1255 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1256 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1257 extern struct fasync_struct
*fasync_alloc(void);
1258 extern void fasync_free(struct fasync_struct
*);
1260 /* can be called from interrupts */
1261 extern void kill_fasync(struct fasync_struct
**, int, int);
1263 extern void __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1264 extern void f_setown(struct file
*filp
, unsigned long arg
, int force
);
1265 extern void f_delown(struct file
*filp
);
1266 extern pid_t
f_getown(struct file
*filp
);
1267 extern int send_sigurg(struct fown_struct
*fown
);
1275 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1276 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1277 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1278 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1279 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1282 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1283 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1285 /* Possible states of 'frozen' field */
1287 SB_UNFROZEN
= 0, /* FS is unfrozen */
1288 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1289 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1290 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1291 * internal threads if needed) */
1292 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1295 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1298 int frozen
; /* Is sb frozen? */
1299 wait_queue_head_t wait_unfrozen
; /* for get_super_thawed() */
1300 struct percpu_rw_semaphore rw_sem
[SB_FREEZE_LEVELS
];
1303 struct super_block
{
1304 struct list_head s_list
; /* Keep this first */
1305 dev_t s_dev
; /* search index; _not_ kdev_t */
1306 unsigned char s_blocksize_bits
;
1307 unsigned long s_blocksize
;
1308 loff_t s_maxbytes
; /* Max file size */
1309 struct file_system_type
*s_type
;
1310 const struct super_operations
*s_op
;
1311 const struct dquot_operations
*dq_op
;
1312 const struct quotactl_ops
*s_qcop
;
1313 const struct export_operations
*s_export_op
;
1314 unsigned long s_flags
;
1315 unsigned long s_iflags
; /* internal SB_I_* flags */
1316 unsigned long s_magic
;
1317 struct dentry
*s_root
;
1318 struct rw_semaphore s_umount
;
1321 #ifdef CONFIG_SECURITY
1324 const struct xattr_handler
**s_xattr
;
1326 struct hlist_bl_head s_anon
; /* anonymous dentries for (nfs) exporting */
1327 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1328 struct block_device
*s_bdev
;
1329 struct backing_dev_info
*s_bdi
;
1330 struct mtd_info
*s_mtd
;
1331 struct hlist_node s_instances
;
1332 unsigned int s_quota_types
; /* Bitmask of supported quota types */
1333 struct quota_info s_dquot
; /* Diskquota specific options */
1335 struct sb_writers s_writers
;
1337 char s_id
[32]; /* Informational name */
1338 u8 s_uuid
[16]; /* UUID */
1340 void *s_fs_info
; /* Filesystem private info */
1341 unsigned int s_max_links
;
1344 /* Granularity of c/m/atime in ns.
1345 Cannot be worse than a second */
1349 * The next field is for VFS *only*. No filesystems have any business
1350 * even looking at it. You had been warned.
1352 struct mutex s_vfs_rename_mutex
; /* Kludge */
1355 * Filesystem subtype. If non-empty the filesystem type field
1356 * in /proc/mounts will be "type.subtype"
1361 * Saved mount options for lazy filesystems using
1362 * generic_show_options()
1364 char __rcu
*s_options
;
1365 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1368 * Saved pool identifier for cleancache (-1 means none)
1370 int cleancache_poolid
;
1372 struct shrinker s_shrink
; /* per-sb shrinker handle */
1374 /* Number of inodes with nlink == 0 but still referenced */
1375 atomic_long_t s_remove_count
;
1377 /* Being remounted read-only */
1378 int s_readonly_remount
;
1380 /* AIO completions deferred from interrupt context */
1381 struct workqueue_struct
*s_dio_done_wq
;
1382 struct hlist_head s_pins
;
1385 * Keep the lru lists last in the structure so they always sit on their
1386 * own individual cachelines.
1388 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp
;
1389 struct list_lru s_inode_lru ____cacheline_aligned_in_smp
;
1390 struct rcu_head rcu
;
1391 struct work_struct destroy_work
;
1393 struct mutex s_sync_lock
; /* sync serialisation lock */
1396 * Indicates how deep in a filesystem stack this SB is
1400 /* s_inode_list_lock protects s_inodes */
1401 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp
;
1402 struct list_head s_inodes
; /* all inodes */
1405 extern struct timespec
current_fs_time(struct super_block
*sb
);
1408 * Snapshotting support.
1411 void __sb_end_write(struct super_block
*sb
, int level
);
1412 int __sb_start_write(struct super_block
*sb
, int level
, bool wait
);
1414 #define __sb_writers_acquired(sb, lev) \
1415 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1416 #define __sb_writers_release(sb, lev) \
1417 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1420 * sb_end_write - drop write access to a superblock
1421 * @sb: the super we wrote to
1423 * Decrement number of writers to the filesystem. Wake up possible waiters
1424 * wanting to freeze the filesystem.
1426 static inline void sb_end_write(struct super_block
*sb
)
1428 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1432 * sb_end_pagefault - drop write access to a superblock from a page fault
1433 * @sb: the super we wrote to
1435 * Decrement number of processes handling write page fault to the filesystem.
1436 * Wake up possible waiters wanting to freeze the filesystem.
1438 static inline void sb_end_pagefault(struct super_block
*sb
)
1440 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1444 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1445 * @sb: the super we wrote to
1447 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1448 * waiters wanting to freeze the filesystem.
1450 static inline void sb_end_intwrite(struct super_block
*sb
)
1452 __sb_end_write(sb
, SB_FREEZE_FS
);
1456 * sb_start_write - get write access to a superblock
1457 * @sb: the super we write to
1459 * When a process wants to write data or metadata to a file system (i.e. dirty
1460 * a page or an inode), it should embed the operation in a sb_start_write() -
1461 * sb_end_write() pair to get exclusion against file system freezing. This
1462 * function increments number of writers preventing freezing. If the file
1463 * system is already frozen, the function waits until the file system is
1466 * Since freeze protection behaves as a lock, users have to preserve
1467 * ordering of freeze protection and other filesystem locks. Generally,
1468 * freeze protection should be the outermost lock. In particular, we have:
1471 * -> i_mutex (write path, truncate, directory ops, ...)
1472 * -> s_umount (freeze_super, thaw_super)
1474 static inline void sb_start_write(struct super_block
*sb
)
1476 __sb_start_write(sb
, SB_FREEZE_WRITE
, true);
1479 static inline int sb_start_write_trylock(struct super_block
*sb
)
1481 return __sb_start_write(sb
, SB_FREEZE_WRITE
, false);
1485 * sb_start_pagefault - get write access to a superblock from a page fault
1486 * @sb: the super we write to
1488 * When a process starts handling write page fault, it should embed the
1489 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1490 * exclusion against file system freezing. This is needed since the page fault
1491 * is going to dirty a page. This function increments number of running page
1492 * faults preventing freezing. If the file system is already frozen, the
1493 * function waits until the file system is thawed.
1495 * Since page fault freeze protection behaves as a lock, users have to preserve
1496 * ordering of freeze protection and other filesystem locks. It is advised to
1497 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1498 * handling code implies lock dependency:
1501 * -> sb_start_pagefault
1503 static inline void sb_start_pagefault(struct super_block
*sb
)
1505 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
, true);
1509 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1510 * @sb: the super we write to
1512 * This is the third level of protection against filesystem freezing. It is
1513 * free for use by a filesystem. The only requirement is that it must rank
1514 * below sb_start_pagefault.
1516 * For example filesystem can call sb_start_intwrite() when starting a
1517 * transaction which somewhat eases handling of freezing for internal sources
1518 * of filesystem changes (internal fs threads, discarding preallocation on file
1521 static inline void sb_start_intwrite(struct super_block
*sb
)
1523 __sb_start_write(sb
, SB_FREEZE_FS
, true);
1527 extern bool inode_owner_or_capable(const struct inode
*inode
);
1530 * VFS helper functions..
1532 extern int vfs_create(struct inode
*, struct dentry
*, umode_t
, bool);
1533 extern int vfs_mkdir(struct inode
*, struct dentry
*, umode_t
);
1534 extern int vfs_mknod(struct inode
*, struct dentry
*, umode_t
, dev_t
);
1535 extern int vfs_symlink(struct inode
*, struct dentry
*, const char *);
1536 extern int vfs_link(struct dentry
*, struct inode
*, struct dentry
*, struct inode
**);
1537 extern int vfs_rmdir(struct inode
*, struct dentry
*);
1538 extern int vfs_unlink(struct inode
*, struct dentry
*, struct inode
**);
1539 extern int vfs_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*, struct inode
**, unsigned int);
1540 extern int vfs_whiteout(struct inode
*, struct dentry
*);
1543 * VFS dentry helper functions.
1545 extern void dentry_unhash(struct dentry
*dentry
);
1548 * VFS file helper functions.
1550 extern void inode_init_owner(struct inode
*inode
, const struct inode
*dir
,
1553 * VFS FS_IOC_FIEMAP helper definitions.
1555 struct fiemap_extent_info
{
1556 unsigned int fi_flags
; /* Flags as passed from user */
1557 unsigned int fi_extents_mapped
; /* Number of mapped extents */
1558 unsigned int fi_extents_max
; /* Size of fiemap_extent array */
1559 struct fiemap_extent __user
*fi_extents_start
; /* Start of
1560 fiemap_extent array */
1562 int fiemap_fill_next_extent(struct fiemap_extent_info
*info
, u64 logical
,
1563 u64 phys
, u64 len
, u32 flags
);
1564 int fiemap_check_flags(struct fiemap_extent_info
*fieinfo
, u32 fs_flags
);
1569 * NOTE! These match bits 12..15 of stat.st_mode
1570 * (ie "(i_mode >> 12) & 15").
1572 #define DT_UNKNOWN 0
1583 * This is the "filldir" function type, used by readdir() to let
1584 * the kernel specify what kind of dirent layout it wants to have.
1585 * This allows the kernel to read directories into kernel space or
1586 * to have different dirent layouts depending on the binary type.
1589 typedef int (*filldir_t
)(struct dir_context
*, const char *, int, loff_t
, u64
,
1592 struct dir_context
{
1593 const filldir_t actor
;
1597 struct block_device_operations
;
1599 /* These macros are for out of kernel modules to test that
1600 * the kernel supports the unlocked_ioctl and compat_ioctl
1601 * fields in struct file_operations. */
1602 #define HAVE_COMPAT_IOCTL 1
1603 #define HAVE_UNLOCKED_IOCTL 1
1606 * These flags let !MMU mmap() govern direct device mapping vs immediate
1607 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1609 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1610 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1611 * NOMMU_MAP_READ: Can be mapped for reading
1612 * NOMMU_MAP_WRITE: Can be mapped for writing
1613 * NOMMU_MAP_EXEC: Can be mapped for execution
1615 #define NOMMU_MAP_COPY 0x00000001
1616 #define NOMMU_MAP_DIRECT 0x00000008
1617 #define NOMMU_MAP_READ VM_MAYREAD
1618 #define NOMMU_MAP_WRITE VM_MAYWRITE
1619 #define NOMMU_MAP_EXEC VM_MAYEXEC
1621 #define NOMMU_VMFLAGS \
1622 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1627 struct file_operations
{
1628 struct module
*owner
;
1629 loff_t (*llseek
) (struct file
*, loff_t
, int);
1630 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
1631 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
1632 ssize_t (*read_iter
) (struct kiocb
*, struct iov_iter
*);
1633 ssize_t (*write_iter
) (struct kiocb
*, struct iov_iter
*);
1634 int (*iterate
) (struct file
*, struct dir_context
*);
1635 unsigned int (*poll
) (struct file
*, struct poll_table_struct
*);
1636 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
1637 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
1638 int (*mmap
) (struct file
*, struct vm_area_struct
*);
1639 int (*open
) (struct inode
*, struct file
*);
1640 int (*flush
) (struct file
*, fl_owner_t id
);
1641 int (*release
) (struct inode
*, struct file
*);
1642 int (*fsync
) (struct file
*, loff_t
, loff_t
, int datasync
);
1643 int (*aio_fsync
) (struct kiocb
*, int datasync
);
1644 int (*fasync
) (int, struct file
*, int);
1645 int (*lock
) (struct file
*, int, struct file_lock
*);
1646 ssize_t (*sendpage
) (struct file
*, struct page
*, int, size_t, loff_t
*, int);
1647 unsigned long (*get_unmapped_area
)(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1648 int (*check_flags
)(int);
1649 int (*flock
) (struct file
*, int, struct file_lock
*);
1650 ssize_t (*splice_write
)(struct pipe_inode_info
*, struct file
*, loff_t
*, size_t, unsigned int);
1651 ssize_t (*splice_read
)(struct file
*, loff_t
*, struct pipe_inode_info
*, size_t, unsigned int);
1652 int (*setlease
)(struct file
*, long, struct file_lock
**, void **);
1653 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
1655 void (*show_fdinfo
)(struct seq_file
*m
, struct file
*f
);
1657 unsigned (*mmap_capabilities
)(struct file
*);
1659 ssize_t (*copy_file_range
)(struct file
*, loff_t
, struct file
*,
1660 loff_t
, size_t, unsigned int);
1661 int (*clone_file_range
)(struct file
*, loff_t
, struct file
*, loff_t
,
1663 ssize_t (*dedupe_file_range
)(struct file
*, u64
, u64
, struct file
*,
1667 struct inode_operations
{
1668 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*, unsigned int);
1669 const char * (*get_link
) (struct dentry
*, struct inode
*, struct delayed_call
*);
1670 int (*permission
) (struct inode
*, int);
1671 struct posix_acl
* (*get_acl
)(struct inode
*, int);
1673 int (*readlink
) (struct dentry
*, char __user
*,int);
1675 int (*create
) (struct inode
*,struct dentry
*, umode_t
, bool);
1676 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
1677 int (*unlink
) (struct inode
*,struct dentry
*);
1678 int (*symlink
) (struct inode
*,struct dentry
*,const char *);
1679 int (*mkdir
) (struct inode
*,struct dentry
*,umode_t
);
1680 int (*rmdir
) (struct inode
*,struct dentry
*);
1681 int (*mknod
) (struct inode
*,struct dentry
*,umode_t
,dev_t
);
1682 int (*rename
) (struct inode
*, struct dentry
*,
1683 struct inode
*, struct dentry
*);
1684 int (*rename2
) (struct inode
*, struct dentry
*,
1685 struct inode
*, struct dentry
*, unsigned int);
1686 int (*setattr
) (struct dentry
*, struct iattr
*);
1687 int (*getattr
) (struct vfsmount
*mnt
, struct dentry
*, struct kstat
*);
1688 int (*setxattr
) (struct dentry
*, const char *,const void *,size_t,int);
1689 ssize_t (*getxattr
) (struct dentry
*, const char *, void *, size_t);
1690 ssize_t (*listxattr
) (struct dentry
*, char *, size_t);
1691 int (*removexattr
) (struct dentry
*, const char *);
1692 int (*fiemap
)(struct inode
*, struct fiemap_extent_info
*, u64 start
,
1694 int (*update_time
)(struct inode
*, struct timespec
*, int);
1695 int (*atomic_open
)(struct inode
*, struct dentry
*,
1696 struct file
*, unsigned open_flag
,
1697 umode_t create_mode
, int *opened
);
1698 int (*tmpfile
) (struct inode
*, struct dentry
*, umode_t
);
1699 int (*set_acl
)(struct inode
*, struct posix_acl
*, int);
1700 } ____cacheline_aligned
;
1702 ssize_t
rw_copy_check_uvector(int type
, const struct iovec __user
* uvector
,
1703 unsigned long nr_segs
, unsigned long fast_segs
,
1704 struct iovec
*fast_pointer
,
1705 struct iovec
**ret_pointer
);
1707 extern ssize_t
__vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1708 extern ssize_t
__vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1709 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1710 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1711 extern ssize_t
vfs_readv(struct file
*, const struct iovec __user
*,
1712 unsigned long, loff_t
*);
1713 extern ssize_t
vfs_writev(struct file
*, const struct iovec __user
*,
1714 unsigned long, loff_t
*);
1715 extern ssize_t
vfs_copy_file_range(struct file
*, loff_t
, struct file
*,
1716 loff_t
, size_t, unsigned int);
1717 extern int vfs_clone_file_range(struct file
*file_in
, loff_t pos_in
,
1718 struct file
*file_out
, loff_t pos_out
, u64 len
);
1719 extern int vfs_dedupe_file_range(struct file
*file
,
1720 struct file_dedupe_range
*same
);
1722 struct super_operations
{
1723 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
1724 void (*destroy_inode
)(struct inode
*);
1726 void (*dirty_inode
) (struct inode
*, int flags
);
1727 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
1728 int (*drop_inode
) (struct inode
*);
1729 void (*evict_inode
) (struct inode
*);
1730 void (*put_super
) (struct super_block
*);
1731 int (*sync_fs
)(struct super_block
*sb
, int wait
);
1732 int (*freeze_super
) (struct super_block
*);
1733 int (*freeze_fs
) (struct super_block
*);
1734 int (*thaw_super
) (struct super_block
*);
1735 int (*unfreeze_fs
) (struct super_block
*);
1736 int (*statfs
) (struct dentry
*, struct kstatfs
*);
1737 int (*remount_fs
) (struct super_block
*, int *, char *);
1738 void (*umount_begin
) (struct super_block
*);
1740 int (*show_options
)(struct seq_file
*, struct dentry
*);
1741 int (*show_devname
)(struct seq_file
*, struct dentry
*);
1742 int (*show_path
)(struct seq_file
*, struct dentry
*);
1743 int (*show_stats
)(struct seq_file
*, struct dentry
*);
1745 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
1746 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
1747 struct dquot
**(*get_dquots
)(struct inode
*);
1749 int (*bdev_try_to_free_page
)(struct super_block
*, struct page
*, gfp_t
);
1750 long (*nr_cached_objects
)(struct super_block
*,
1751 struct shrink_control
*);
1752 long (*free_cached_objects
)(struct super_block
*,
1753 struct shrink_control
*);
1757 * Inode flags - they have no relation to superblock flags now
1759 #define S_SYNC 1 /* Writes are synced at once */
1760 #define S_NOATIME 2 /* Do not update access times */
1761 #define S_APPEND 4 /* Append-only file */
1762 #define S_IMMUTABLE 8 /* Immutable file */
1763 #define S_DEAD 16 /* removed, but still open directory */
1764 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1765 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1766 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1767 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1768 #define S_PRIVATE 512 /* Inode is fs-internal */
1769 #define S_IMA 1024 /* Inode has an associated IMA struct */
1770 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1771 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1772 #ifdef CONFIG_FS_DAX
1773 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1775 #define S_DAX 0 /* Make all the DAX code disappear */
1779 * Note that nosuid etc flags are inode-specific: setting some file-system
1780 * flags just means all the inodes inherit those flags by default. It might be
1781 * possible to override it selectively if you really wanted to with some
1782 * ioctl() that is not currently implemented.
1784 * Exception: MS_RDONLY is always applied to the entire file system.
1786 * Unfortunately, it is possible to change a filesystems flags with it mounted
1787 * with files in use. This means that all of the inodes will not have their
1788 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1789 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1791 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1793 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1794 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1795 ((inode)->i_flags & S_SYNC))
1796 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1797 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1798 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1799 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1800 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1802 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1803 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1804 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1805 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1807 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1808 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1809 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1810 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1811 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1812 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1813 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1814 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1816 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1817 (inode)->i_rdev == WHITEOUT_DEV)
1820 * Inode state bits. Protected by inode->i_lock
1822 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1823 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1825 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1826 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1827 * various stages of removing an inode.
1829 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1831 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1832 * fdatasync(). i_atime is the usual cause.
1833 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1834 * these changes separately from I_DIRTY_SYNC so that we
1835 * don't have to write inode on fdatasync() when only
1836 * mtime has changed in it.
1837 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1838 * I_NEW Serves as both a mutex and completion notification.
1839 * New inodes set I_NEW. If two processes both create
1840 * the same inode, one of them will release its inode and
1841 * wait for I_NEW to be released before returning.
1842 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1843 * also cause waiting on I_NEW, without I_NEW actually
1844 * being set. find_inode() uses this to prevent returning
1845 * nearly-dead inodes.
1846 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1847 * is zero. I_FREEING must be set when I_WILL_FREE is
1849 * I_FREEING Set when inode is about to be freed but still has dirty
1850 * pages or buffers attached or the inode itself is still
1852 * I_CLEAR Added by clear_inode(). In this state the inode is
1853 * clean and can be destroyed. Inode keeps I_FREEING.
1855 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1856 * prohibited for many purposes. iget() must wait for
1857 * the inode to be completely released, then create it
1858 * anew. Other functions will just ignore such inodes,
1859 * if appropriate. I_NEW is used for waiting.
1861 * I_SYNC Writeback of inode is running. The bit is set during
1862 * data writeback, and cleared with a wakeup on the bit
1863 * address once it is done. The bit is also used to pin
1864 * the inode in memory for flusher thread.
1866 * I_REFERENCED Marks the inode as recently references on the LRU list.
1868 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1870 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1871 * synchronize competing switching instances and to tell
1872 * wb stat updates to grab mapping->tree_lock. See
1873 * inode_switch_wb_work_fn() for details.
1875 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1877 #define I_DIRTY_SYNC (1 << 0)
1878 #define I_DIRTY_DATASYNC (1 << 1)
1879 #define I_DIRTY_PAGES (1 << 2)
1881 #define I_NEW (1 << __I_NEW)
1882 #define I_WILL_FREE (1 << 4)
1883 #define I_FREEING (1 << 5)
1884 #define I_CLEAR (1 << 6)
1886 #define I_SYNC (1 << __I_SYNC)
1887 #define I_REFERENCED (1 << 8)
1888 #define __I_DIO_WAKEUP 9
1889 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1890 #define I_LINKABLE (1 << 10)
1891 #define I_DIRTY_TIME (1 << 11)
1892 #define __I_DIRTY_TIME_EXPIRED 12
1893 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1894 #define I_WB_SWITCH (1 << 13)
1896 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1897 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1899 extern void __mark_inode_dirty(struct inode
*, int);
1900 static inline void mark_inode_dirty(struct inode
*inode
)
1902 __mark_inode_dirty(inode
, I_DIRTY
);
1905 static inline void mark_inode_dirty_sync(struct inode
*inode
)
1907 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
1910 extern void inc_nlink(struct inode
*inode
);
1911 extern void drop_nlink(struct inode
*inode
);
1912 extern void clear_nlink(struct inode
*inode
);
1913 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
1915 static inline void inode_inc_link_count(struct inode
*inode
)
1918 mark_inode_dirty(inode
);
1921 static inline void inode_dec_link_count(struct inode
*inode
)
1924 mark_inode_dirty(inode
);
1928 * inode_inc_iversion - increments i_version
1929 * @inode: inode that need to be updated
1931 * Every time the inode is modified, the i_version field will be incremented.
1932 * The filesystem has to be mounted with i_version flag
1935 static inline void inode_inc_iversion(struct inode
*inode
)
1937 spin_lock(&inode
->i_lock
);
1939 spin_unlock(&inode
->i_lock
);
1942 enum file_time_flags
{
1949 extern bool atime_needs_update(const struct path
*, struct inode
*);
1950 extern void touch_atime(const struct path
*);
1951 static inline void file_accessed(struct file
*file
)
1953 if (!(file
->f_flags
& O_NOATIME
))
1954 touch_atime(&file
->f_path
);
1957 int sync_inode(struct inode
*inode
, struct writeback_control
*wbc
);
1958 int sync_inode_metadata(struct inode
*inode
, int wait
);
1960 struct file_system_type
{
1963 #define FS_REQUIRES_DEV 1
1964 #define FS_BINARY_MOUNTDATA 2
1965 #define FS_HAS_SUBTYPE 4
1966 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
1967 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
1968 #define FS_USERNS_VISIBLE 32 /* FS must already be visible */
1969 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
1970 struct dentry
*(*mount
) (struct file_system_type
*, int,
1971 const char *, void *);
1972 void (*kill_sb
) (struct super_block
*);
1973 struct module
*owner
;
1974 struct file_system_type
* next
;
1975 struct hlist_head fs_supers
;
1977 struct lock_class_key s_lock_key
;
1978 struct lock_class_key s_umount_key
;
1979 struct lock_class_key s_vfs_rename_key
;
1980 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
1982 struct lock_class_key i_lock_key
;
1983 struct lock_class_key i_mutex_key
;
1984 struct lock_class_key i_mutex_dir_key
;
1987 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1989 extern struct dentry
*mount_ns(struct file_system_type
*fs_type
, int flags
,
1990 void *data
, int (*fill_super
)(struct super_block
*, void *, int));
1991 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
1992 int flags
, const char *dev_name
, void *data
,
1993 int (*fill_super
)(struct super_block
*, void *, int));
1994 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
1995 int flags
, void *data
,
1996 int (*fill_super
)(struct super_block
*, void *, int));
1997 extern struct dentry
*mount_nodev(struct file_system_type
*fs_type
,
1998 int flags
, void *data
,
1999 int (*fill_super
)(struct super_block
*, void *, int));
2000 extern struct dentry
*mount_subtree(struct vfsmount
*mnt
, const char *path
);
2001 void generic_shutdown_super(struct super_block
*sb
);
2002 void kill_block_super(struct super_block
*sb
);
2003 void kill_anon_super(struct super_block
*sb
);
2004 void kill_litter_super(struct super_block
*sb
);
2005 void deactivate_super(struct super_block
*sb
);
2006 void deactivate_locked_super(struct super_block
*sb
);
2007 int set_anon_super(struct super_block
*s
, void *data
);
2008 int get_anon_bdev(dev_t
*);
2009 void free_anon_bdev(dev_t
);
2010 struct super_block
*sget(struct file_system_type
*type
,
2011 int (*test
)(struct super_block
*,void *),
2012 int (*set
)(struct super_block
*,void *),
2013 int flags
, void *data
);
2014 extern struct dentry
*mount_pseudo(struct file_system_type
*, char *,
2015 const struct super_operations
*ops
,
2016 const struct dentry_operations
*dops
,
2019 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2020 #define fops_get(fops) \
2021 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2022 #define fops_put(fops) \
2023 do { if (fops) module_put((fops)->owner); } while(0)
2025 * This one is to be used *ONLY* from ->open() instances.
2026 * fops must be non-NULL, pinned down *and* module dependencies
2027 * should be sufficient to pin the caller down as well.
2029 #define replace_fops(f, fops) \
2031 struct file *__file = (f); \
2032 fops_put(__file->f_op); \
2033 BUG_ON(!(__file->f_op = (fops))); \
2036 extern int register_filesystem(struct file_system_type
*);
2037 extern int unregister_filesystem(struct file_system_type
*);
2038 extern struct vfsmount
*kern_mount_data(struct file_system_type
*, void *data
);
2039 #define kern_mount(type) kern_mount_data(type, NULL)
2040 extern void kern_unmount(struct vfsmount
*mnt
);
2041 extern int may_umount_tree(struct vfsmount
*);
2042 extern int may_umount(struct vfsmount
*);
2043 extern long do_mount(const char *, const char __user
*,
2044 const char *, unsigned long, void *);
2045 extern struct vfsmount
*collect_mounts(struct path
*);
2046 extern void drop_collected_mounts(struct vfsmount
*);
2047 extern int iterate_mounts(int (*)(struct vfsmount
*, void *), void *,
2049 extern int vfs_statfs(struct path
*, struct kstatfs
*);
2050 extern int user_statfs(const char __user
*, struct kstatfs
*);
2051 extern int fd_statfs(int, struct kstatfs
*);
2052 extern int vfs_ustat(dev_t
, struct kstatfs
*);
2053 extern int freeze_super(struct super_block
*super
);
2054 extern int thaw_super(struct super_block
*super
);
2055 extern bool our_mnt(struct vfsmount
*mnt
);
2057 extern int current_umask(void);
2059 extern void ihold(struct inode
* inode
);
2060 extern void iput(struct inode
*);
2061 extern int generic_update_time(struct inode
*, struct timespec
*, int);
2064 extern struct kobject
*fs_kobj
;
2066 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
2068 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2069 extern int locks_mandatory_locked(struct file
*);
2070 extern int locks_mandatory_area(struct inode
*, struct file
*, loff_t
, loff_t
, unsigned char);
2073 * Candidates for mandatory locking have the setgid bit set
2074 * but no group execute bit - an otherwise meaningless combination.
2077 static inline int __mandatory_lock(struct inode
*ino
)
2079 return (ino
->i_mode
& (S_ISGID
| S_IXGRP
)) == S_ISGID
;
2083 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2084 * otherwise these will be advisory locks
2087 static inline int mandatory_lock(struct inode
*ino
)
2089 return IS_MANDLOCK(ino
) && __mandatory_lock(ino
);
2092 static inline int locks_verify_locked(struct file
*file
)
2094 if (mandatory_lock(file_inode(file
)))
2095 return locks_mandatory_locked(file
);
2099 static inline int locks_verify_truncate(struct inode
*inode
,
2103 if (!inode
->i_flctx
|| !mandatory_lock(inode
))
2106 if (size
< inode
->i_size
) {
2107 return locks_mandatory_area(inode
, f
, size
, inode
->i_size
- 1,
2110 return locks_mandatory_area(inode
, f
, inode
->i_size
, size
- 1,
2115 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2117 static inline int locks_mandatory_locked(struct file
*file
)
2122 static inline int locks_mandatory_area(struct inode
*inode
, struct file
*filp
,
2123 loff_t start
, loff_t end
, unsigned char type
)
2128 static inline int __mandatory_lock(struct inode
*inode
)
2133 static inline int mandatory_lock(struct inode
*inode
)
2138 static inline int locks_verify_locked(struct file
*file
)
2143 static inline int locks_verify_truncate(struct inode
*inode
, struct file
*filp
,
2149 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2152 #ifdef CONFIG_FILE_LOCKING
2153 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2156 * Since this check is lockless, we must ensure that any refcounts
2157 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2158 * could end up racing with tasks trying to set a new lease on this
2162 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2163 return __break_lease(inode
, mode
, FL_LEASE
);
2167 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2170 * Since this check is lockless, we must ensure that any refcounts
2171 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2172 * could end up racing with tasks trying to set a new lease on this
2176 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2177 return __break_lease(inode
, mode
, FL_DELEG
);
2181 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2185 ret
= break_deleg(inode
, O_WRONLY
|O_NONBLOCK
);
2186 if (ret
== -EWOULDBLOCK
&& delegated_inode
) {
2187 *delegated_inode
= inode
;
2193 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2197 ret
= break_deleg(*delegated_inode
, O_WRONLY
);
2198 iput(*delegated_inode
);
2199 *delegated_inode
= NULL
;
2203 static inline int break_layout(struct inode
*inode
, bool wait
)
2206 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2207 return __break_lease(inode
,
2208 wait
? O_WRONLY
: O_WRONLY
| O_NONBLOCK
,
2213 #else /* !CONFIG_FILE_LOCKING */
2214 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2219 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2224 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2229 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2235 static inline int break_layout(struct inode
*inode
, bool wait
)
2240 #endif /* CONFIG_FILE_LOCKING */
2245 const char *name
; /* pointer to actual string */
2246 const __user
char *uptr
; /* original userland pointer */
2247 struct audit_names
*aname
;
2252 extern long vfs_truncate(struct path
*, loff_t
);
2253 extern int do_truncate(struct dentry
*, loff_t start
, unsigned int time_attrs
,
2255 extern int vfs_fallocate(struct file
*file
, int mode
, loff_t offset
,
2257 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
2259 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
2260 extern struct file
*filp_open(const char *, int, umode_t
);
2261 extern struct file
*file_open_root(struct dentry
*, struct vfsmount
*,
2263 extern struct file
* dentry_open(const struct path
*, int, const struct cred
*);
2264 extern int filp_close(struct file
*, fl_owner_t id
);
2266 extern struct filename
*getname_flags(const char __user
*, int, int *);
2267 extern struct filename
*getname(const char __user
*);
2268 extern struct filename
*getname_kernel(const char *);
2269 extern void putname(struct filename
*name
);
2275 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
2276 int (*open
)(struct inode
*, struct file
*),
2278 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
2282 extern int ioctl_preallocate(struct file
*filp
, void __user
*argp
);
2285 extern void __init
vfs_caches_init_early(void);
2286 extern void __init
vfs_caches_init(void);
2288 extern struct kmem_cache
*names_cachep
;
2290 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2291 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2294 extern int register_blkdev(unsigned int, const char *);
2295 extern void unregister_blkdev(unsigned int, const char *);
2296 extern struct block_device
*bdget(dev_t
);
2297 extern struct block_device
*bdgrab(struct block_device
*bdev
);
2298 extern void bd_set_size(struct block_device
*, loff_t size
);
2299 extern void bd_forget(struct inode
*inode
);
2300 extern void bdput(struct block_device
*);
2301 extern void invalidate_bdev(struct block_device
*);
2302 extern void iterate_bdevs(void (*)(struct block_device
*, void *), void *);
2303 extern int sync_blockdev(struct block_device
*bdev
);
2304 extern void kill_bdev(struct block_device
*);
2305 extern struct super_block
*freeze_bdev(struct block_device
*);
2306 extern void emergency_thaw_all(void);
2307 extern int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
);
2308 extern int fsync_bdev(struct block_device
*);
2309 #ifdef CONFIG_FS_DAX
2310 extern bool blkdev_dax_capable(struct block_device
*bdev
);
2312 static inline bool blkdev_dax_capable(struct block_device
*bdev
)
2318 extern struct super_block
*blockdev_superblock
;
2320 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2322 return sb
== blockdev_superblock
;
2325 static inline void bd_forget(struct inode
*inode
) {}
2326 static inline int sync_blockdev(struct block_device
*bdev
) { return 0; }
2327 static inline void kill_bdev(struct block_device
*bdev
) {}
2328 static inline void invalidate_bdev(struct block_device
*bdev
) {}
2330 static inline struct super_block
*freeze_bdev(struct block_device
*sb
)
2335 static inline int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
)
2340 static inline void iterate_bdevs(void (*f
)(struct block_device
*, void *), void *arg
)
2344 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2349 extern int sync_filesystem(struct super_block
*);
2350 extern const struct file_operations def_blk_fops
;
2351 extern const struct file_operations def_chr_fops
;
2353 extern int ioctl_by_bdev(struct block_device
*, unsigned, unsigned long);
2354 extern int blkdev_ioctl(struct block_device
*, fmode_t
, unsigned, unsigned long);
2355 extern long compat_blkdev_ioctl(struct file
*, unsigned, unsigned long);
2356 extern int blkdev_get(struct block_device
*bdev
, fmode_t mode
, void *holder
);
2357 extern struct block_device
*blkdev_get_by_path(const char *path
, fmode_t mode
,
2359 extern struct block_device
*blkdev_get_by_dev(dev_t dev
, fmode_t mode
,
2361 extern void blkdev_put(struct block_device
*bdev
, fmode_t mode
);
2362 extern int __blkdev_reread_part(struct block_device
*bdev
);
2363 extern int blkdev_reread_part(struct block_device
*bdev
);
2366 extern int bd_link_disk_holder(struct block_device
*bdev
, struct gendisk
*disk
);
2367 extern void bd_unlink_disk_holder(struct block_device
*bdev
,
2368 struct gendisk
*disk
);
2370 static inline int bd_link_disk_holder(struct block_device
*bdev
,
2371 struct gendisk
*disk
)
2375 static inline void bd_unlink_disk_holder(struct block_device
*bdev
,
2376 struct gendisk
*disk
)
2383 #define CHRDEV_MAJOR_HASH_SIZE 255
2384 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
2385 extern int register_chrdev_region(dev_t
, unsigned, const char *);
2386 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
2387 unsigned int count
, const char *name
,
2388 const struct file_operations
*fops
);
2389 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
2390 unsigned int count
, const char *name
);
2391 extern void unregister_chrdev_region(dev_t
, unsigned);
2392 extern void chrdev_show(struct seq_file
*,off_t
);
2394 static inline int register_chrdev(unsigned int major
, const char *name
,
2395 const struct file_operations
*fops
)
2397 return __register_chrdev(major
, 0, 256, name
, fops
);
2400 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2402 __unregister_chrdev(major
, 0, 256, name
);
2405 /* fs/block_dev.c */
2406 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2407 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2410 #define BLKDEV_MAJOR_HASH_SIZE 255
2411 extern const char *__bdevname(dev_t
, char *buffer
);
2412 extern const char *bdevname(struct block_device
*bdev
, char *buffer
);
2413 extern struct block_device
*lookup_bdev(const char *);
2414 extern void blkdev_show(struct seq_file
*,off_t
);
2417 #define BLKDEV_MAJOR_HASH_SIZE 0
2420 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2422 /* Invalid inode operations -- fs/bad_inode.c */
2423 extern void make_bad_inode(struct inode
*);
2424 extern bool is_bad_inode(struct inode
*);
2428 * return READ, READA, or WRITE
2430 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2433 * return data direction, READ or WRITE
2435 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2437 extern void check_disk_size_change(struct gendisk
*disk
,
2438 struct block_device
*bdev
);
2439 extern int revalidate_disk(struct gendisk
*);
2440 extern int check_disk_change(struct block_device
*);
2441 extern int __invalidate_device(struct block_device
*, bool);
2442 extern int invalidate_partition(struct gendisk
*, int);
2444 unsigned long invalidate_mapping_pages(struct address_space
*mapping
,
2445 pgoff_t start
, pgoff_t end
);
2447 static inline void invalidate_remote_inode(struct inode
*inode
)
2449 if (S_ISREG(inode
->i_mode
) || S_ISDIR(inode
->i_mode
) ||
2450 S_ISLNK(inode
->i_mode
))
2451 invalidate_mapping_pages(inode
->i_mapping
, 0, -1);
2453 extern int invalidate_inode_pages2(struct address_space
*mapping
);
2454 extern int invalidate_inode_pages2_range(struct address_space
*mapping
,
2455 pgoff_t start
, pgoff_t end
);
2456 extern int write_inode_now(struct inode
*, int);
2457 extern int filemap_fdatawrite(struct address_space
*);
2458 extern int filemap_flush(struct address_space
*);
2459 extern int filemap_fdatawait(struct address_space
*);
2460 extern void filemap_fdatawait_keep_errors(struct address_space
*);
2461 extern int filemap_fdatawait_range(struct address_space
*, loff_t lstart
,
2463 extern int filemap_write_and_wait(struct address_space
*mapping
);
2464 extern int filemap_write_and_wait_range(struct address_space
*mapping
,
2465 loff_t lstart
, loff_t lend
);
2466 extern int __filemap_fdatawrite_range(struct address_space
*mapping
,
2467 loff_t start
, loff_t end
, int sync_mode
);
2468 extern int filemap_fdatawrite_range(struct address_space
*mapping
,
2469 loff_t start
, loff_t end
);
2471 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
2473 extern int vfs_fsync(struct file
*file
, int datasync
);
2474 static inline int generic_write_sync(struct file
*file
, loff_t pos
, loff_t count
)
2476 if (!(file
->f_flags
& O_DSYNC
) && !IS_SYNC(file
->f_mapping
->host
))
2478 return vfs_fsync_range(file
, pos
, pos
+ count
- 1,
2479 (file
->f_flags
& __O_SYNC
) ? 0 : 1);
2481 extern void emergency_sync(void);
2482 extern void emergency_remount(void);
2484 extern sector_t
bmap(struct inode
*, sector_t
);
2486 extern int notify_change(struct dentry
*, struct iattr
*, struct inode
**);
2487 extern int inode_permission(struct inode
*, int);
2488 extern int __inode_permission(struct inode
*, int);
2489 extern int generic_permission(struct inode
*, int);
2490 extern int __check_sticky(struct inode
*dir
, struct inode
*inode
);
2492 static inline bool execute_ok(struct inode
*inode
)
2494 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
2497 static inline void file_start_write(struct file
*file
)
2499 if (!S_ISREG(file_inode(file
)->i_mode
))
2501 __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, true);
2504 static inline bool file_start_write_trylock(struct file
*file
)
2506 if (!S_ISREG(file_inode(file
)->i_mode
))
2508 return __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, false);
2511 static inline void file_end_write(struct file
*file
)
2513 if (!S_ISREG(file_inode(file
)->i_mode
))
2515 __sb_end_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
);
2519 * get_write_access() gets write permission for a file.
2520 * put_write_access() releases this write permission.
2521 * This is used for regular files.
2522 * We cannot support write (and maybe mmap read-write shared) accesses and
2523 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2524 * can have the following values:
2525 * 0: no writers, no VM_DENYWRITE mappings
2526 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2527 * > 0: (i_writecount) users are writing to the file.
2529 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2530 * except for the cases where we don't hold i_writecount yet. Then we need to
2531 * use {get,deny}_write_access() - these functions check the sign and refuse
2532 * to do the change if sign is wrong.
2534 static inline int get_write_access(struct inode
*inode
)
2536 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2538 static inline int deny_write_access(struct file
*file
)
2540 struct inode
*inode
= file_inode(file
);
2541 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2543 static inline void put_write_access(struct inode
* inode
)
2545 atomic_dec(&inode
->i_writecount
);
2547 static inline void allow_write_access(struct file
*file
)
2550 atomic_inc(&file_inode(file
)->i_writecount
);
2552 static inline bool inode_is_open_for_write(const struct inode
*inode
)
2554 return atomic_read(&inode
->i_writecount
) > 0;
2558 static inline void i_readcount_dec(struct inode
*inode
)
2560 BUG_ON(!atomic_read(&inode
->i_readcount
));
2561 atomic_dec(&inode
->i_readcount
);
2563 static inline void i_readcount_inc(struct inode
*inode
)
2565 atomic_inc(&inode
->i_readcount
);
2568 static inline void i_readcount_dec(struct inode
*inode
)
2572 static inline void i_readcount_inc(struct inode
*inode
)
2577 extern int do_pipe_flags(int *, int);
2579 enum kernel_read_file_id
{
2580 READING_FIRMWARE
= 1,
2584 extern int kernel_read(struct file
*, loff_t
, char *, unsigned long);
2585 extern int kernel_read_file(struct file
*, void **, loff_t
*, loff_t
,
2586 enum kernel_read_file_id
);
2587 extern int kernel_read_file_from_path(char *, void **, loff_t
*, loff_t
,
2588 enum kernel_read_file_id
);
2589 extern int kernel_read_file_from_fd(int, void **, loff_t
*, loff_t
,
2590 enum kernel_read_file_id
);
2591 extern ssize_t
kernel_write(struct file
*, const char *, size_t, loff_t
);
2592 extern ssize_t
__kernel_write(struct file
*, const char *, size_t, loff_t
*);
2593 extern struct file
* open_exec(const char *);
2595 /* fs/dcache.c -- generic fs support functions */
2596 extern bool is_subdir(struct dentry
*, struct dentry
*);
2597 extern bool path_is_under(struct path
*, struct path
*);
2599 extern char *file_path(struct file
*, char *, int);
2601 #include <linux/err.h>
2603 /* needed for stackable file system support */
2604 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int whence
);
2606 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int whence
);
2608 extern int inode_init_always(struct super_block
*, struct inode
*);
2609 extern void inode_init_once(struct inode
*);
2610 extern void address_space_init_once(struct address_space
*mapping
);
2611 extern struct inode
* igrab(struct inode
*);
2612 extern ino_t
iunique(struct super_block
*, ino_t
);
2613 extern int inode_needs_sync(struct inode
*inode
);
2614 extern int generic_delete_inode(struct inode
*inode
);
2615 static inline int generic_drop_inode(struct inode
*inode
)
2617 return !inode
->i_nlink
|| inode_unhashed(inode
);
2620 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
2621 unsigned long hashval
, int (*test
)(struct inode
*, void *),
2623 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
2624 int (*test
)(struct inode
*, void *), void *data
);
2625 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
2627 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
2628 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
2629 extern struct inode
*find_inode_nowait(struct super_block
*,
2631 int (*match
)(struct inode
*,
2632 unsigned long, void *),
2634 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
2635 extern int insert_inode_locked(struct inode
*);
2636 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2637 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
2639 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
2641 extern void unlock_new_inode(struct inode
*);
2642 extern unsigned int get_next_ino(void);
2644 extern void __iget(struct inode
* inode
);
2645 extern void iget_failed(struct inode
*);
2646 extern void clear_inode(struct inode
*);
2647 extern void __destroy_inode(struct inode
*);
2648 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
2649 extern struct inode
*new_inode(struct super_block
*sb
);
2650 extern void free_inode_nonrcu(struct inode
*inode
);
2651 extern int should_remove_suid(struct dentry
*);
2652 extern int file_remove_privs(struct file
*);
2653 extern int dentry_needs_remove_privs(struct dentry
*dentry
);
2654 static inline int file_needs_remove_privs(struct file
*file
)
2656 return dentry_needs_remove_privs(file
->f_path
.dentry
);
2659 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
2660 static inline void insert_inode_hash(struct inode
*inode
)
2662 __insert_inode_hash(inode
, inode
->i_ino
);
2665 extern void __remove_inode_hash(struct inode
*);
2666 static inline void remove_inode_hash(struct inode
*inode
)
2668 if (!inode_unhashed(inode
) && !hlist_fake(&inode
->i_hash
))
2669 __remove_inode_hash(inode
);
2672 extern void inode_sb_list_add(struct inode
*inode
);
2675 extern blk_qc_t
submit_bio(int, struct bio
*);
2676 extern int bdev_read_only(struct block_device
*);
2678 extern int set_blocksize(struct block_device
*, int);
2679 extern int sb_set_blocksize(struct super_block
*, int);
2680 extern int sb_min_blocksize(struct super_block
*, int);
2682 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
2683 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
2684 extern ssize_t
generic_write_checks(struct kiocb
*, struct iov_iter
*);
2685 extern ssize_t
generic_file_read_iter(struct kiocb
*, struct iov_iter
*);
2686 extern ssize_t
__generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2687 extern ssize_t
generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2688 extern ssize_t
generic_file_direct_write(struct kiocb
*, struct iov_iter
*, loff_t
);
2689 extern ssize_t
generic_perform_write(struct file
*, struct iov_iter
*, loff_t
);
2691 ssize_t
vfs_iter_read(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
);
2692 ssize_t
vfs_iter_write(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
);
2694 /* fs/block_dev.c */
2695 extern ssize_t
blkdev_read_iter(struct kiocb
*iocb
, struct iov_iter
*to
);
2696 extern ssize_t
blkdev_write_iter(struct kiocb
*iocb
, struct iov_iter
*from
);
2697 extern int blkdev_fsync(struct file
*filp
, loff_t start
, loff_t end
,
2699 extern void block_sync_page(struct page
*page
);
2702 extern ssize_t
generic_file_splice_read(struct file
*, loff_t
*,
2703 struct pipe_inode_info
*, size_t, unsigned int);
2704 extern ssize_t
default_file_splice_read(struct file
*, loff_t
*,
2705 struct pipe_inode_info
*, size_t, unsigned int);
2706 extern ssize_t
iter_file_splice_write(struct pipe_inode_info
*,
2707 struct file
*, loff_t
*, size_t, unsigned int);
2708 extern ssize_t
generic_splice_sendpage(struct pipe_inode_info
*pipe
,
2709 struct file
*out
, loff_t
*, size_t len
, unsigned int flags
);
2710 extern long do_splice_direct(struct file
*in
, loff_t
*ppos
, struct file
*out
,
2711 loff_t
*opos
, size_t len
, unsigned int flags
);
2715 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
2716 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int whence
);
2717 extern loff_t
no_llseek(struct file
*file
, loff_t offset
, int whence
);
2718 extern loff_t
vfs_setpos(struct file
*file
, loff_t offset
, loff_t maxsize
);
2719 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int whence
);
2720 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
2721 int whence
, loff_t maxsize
, loff_t eof
);
2722 extern loff_t
fixed_size_llseek(struct file
*file
, loff_t offset
,
2723 int whence
, loff_t size
);
2724 extern loff_t
no_seek_end_llseek_size(struct file
*, loff_t
, int, loff_t
);
2725 extern loff_t
no_seek_end_llseek(struct file
*, loff_t
, int);
2726 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
2727 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
2730 typedef void (dio_submit_t
)(int rw
, struct bio
*bio
, struct inode
*inode
,
2731 loff_t file_offset
);
2734 /* need locking between buffered and direct access */
2737 /* filesystem does not support filling holes */
2738 DIO_SKIP_HOLES
= 0x02,
2740 /* filesystem can handle aio writes beyond i_size */
2741 DIO_ASYNC_EXTEND
= 0x04,
2743 /* inode/fs/bdev does not need truncate protection */
2744 DIO_SKIP_DIO_COUNT
= 0x08,
2747 void dio_end_io(struct bio
*bio
, int error
);
2749 ssize_t
__blockdev_direct_IO(struct kiocb
*iocb
, struct inode
*inode
,
2750 struct block_device
*bdev
, struct iov_iter
*iter
,
2751 loff_t offset
, get_block_t get_block
,
2752 dio_iodone_t end_io
, dio_submit_t submit_io
,
2755 static inline ssize_t
blockdev_direct_IO(struct kiocb
*iocb
,
2756 struct inode
*inode
,
2757 struct iov_iter
*iter
, loff_t offset
,
2758 get_block_t get_block
)
2760 return __blockdev_direct_IO(iocb
, inode
, inode
->i_sb
->s_bdev
, iter
,
2761 offset
, get_block
, NULL
, NULL
,
2762 DIO_LOCKING
| DIO_SKIP_HOLES
);
2766 void inode_dio_wait(struct inode
*inode
);
2769 * inode_dio_begin - signal start of a direct I/O requests
2770 * @inode: inode the direct I/O happens on
2772 * This is called once we've finished processing a direct I/O request,
2773 * and is used to wake up callers waiting for direct I/O to be quiesced.
2775 static inline void inode_dio_begin(struct inode
*inode
)
2777 atomic_inc(&inode
->i_dio_count
);
2781 * inode_dio_end - signal finish of a direct I/O requests
2782 * @inode: inode the direct I/O happens on
2784 * This is called once we've finished processing a direct I/O request,
2785 * and is used to wake up callers waiting for direct I/O to be quiesced.
2787 static inline void inode_dio_end(struct inode
*inode
)
2789 if (atomic_dec_and_test(&inode
->i_dio_count
))
2790 wake_up_bit(&inode
->i_state
, __I_DIO_WAKEUP
);
2793 extern void inode_set_flags(struct inode
*inode
, unsigned int flags
,
2796 extern const struct file_operations generic_ro_fops
;
2798 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2800 extern int readlink_copy(char __user
*, int, const char *);
2801 extern int page_readlink(struct dentry
*, char __user
*, int);
2802 extern const char *page_get_link(struct dentry
*, struct inode
*,
2803 struct delayed_call
*);
2804 extern void page_put_link(void *);
2805 extern int __page_symlink(struct inode
*inode
, const char *symname
, int len
,
2807 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
2808 extern const struct inode_operations page_symlink_inode_operations
;
2809 extern void kfree_link(void *);
2810 extern int generic_readlink(struct dentry
*, char __user
*, int);
2811 extern void generic_fillattr(struct inode
*, struct kstat
*);
2812 int vfs_getattr_nosec(struct path
*path
, struct kstat
*stat
);
2813 extern int vfs_getattr(struct path
*, struct kstat
*);
2814 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2815 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2816 void __inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2817 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2818 loff_t
inode_get_bytes(struct inode
*inode
);
2819 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
2820 const char *simple_get_link(struct dentry
*, struct inode
*,
2821 struct delayed_call
*);
2822 extern const struct inode_operations simple_symlink_inode_operations
;
2824 extern int iterate_dir(struct file
*, struct dir_context
*);
2826 extern int vfs_stat(const char __user
*, struct kstat
*);
2827 extern int vfs_lstat(const char __user
*, struct kstat
*);
2828 extern int vfs_fstat(unsigned int, struct kstat
*);
2829 extern int vfs_fstatat(int , const char __user
*, struct kstat
*, int);
2831 extern int __generic_block_fiemap(struct inode
*inode
,
2832 struct fiemap_extent_info
*fieinfo
,
2833 loff_t start
, loff_t len
,
2834 get_block_t
*get_block
);
2835 extern int generic_block_fiemap(struct inode
*inode
,
2836 struct fiemap_extent_info
*fieinfo
, u64 start
,
2837 u64 len
, get_block_t
*get_block
);
2839 extern void get_filesystem(struct file_system_type
*fs
);
2840 extern void put_filesystem(struct file_system_type
*fs
);
2841 extern struct file_system_type
*get_fs_type(const char *name
);
2842 extern struct super_block
*get_super(struct block_device
*);
2843 extern struct super_block
*get_super_thawed(struct block_device
*);
2844 extern struct super_block
*get_active_super(struct block_device
*bdev
);
2845 extern void drop_super(struct super_block
*sb
);
2846 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
2847 extern void iterate_supers_type(struct file_system_type
*,
2848 void (*)(struct super_block
*, void *), void *);
2850 extern int dcache_dir_open(struct inode
*, struct file
*);
2851 extern int dcache_dir_close(struct inode
*, struct file
*);
2852 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
2853 extern int dcache_readdir(struct file
*, struct dir_context
*);
2854 extern int simple_setattr(struct dentry
*, struct iattr
*);
2855 extern int simple_getattr(struct vfsmount
*, struct dentry
*, struct kstat
*);
2856 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
2857 extern int simple_open(struct inode
*inode
, struct file
*file
);
2858 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
2859 extern int simple_unlink(struct inode
*, struct dentry
*);
2860 extern int simple_rmdir(struct inode
*, struct dentry
*);
2861 extern int simple_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*);
2862 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
2863 extern int simple_empty(struct dentry
*);
2864 extern int simple_readpage(struct file
*file
, struct page
*page
);
2865 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
2866 loff_t pos
, unsigned len
, unsigned flags
,
2867 struct page
**pagep
, void **fsdata
);
2868 extern int simple_write_end(struct file
*file
, struct address_space
*mapping
,
2869 loff_t pos
, unsigned len
, unsigned copied
,
2870 struct page
*page
, void *fsdata
);
2871 extern int always_delete_dentry(const struct dentry
*);
2872 extern struct inode
*alloc_anon_inode(struct super_block
*);
2873 extern int simple_nosetlease(struct file
*, long, struct file_lock
**, void **);
2874 extern const struct dentry_operations simple_dentry_operations
;
2876 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
2877 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
2878 extern const struct file_operations simple_dir_operations
;
2879 extern const struct inode_operations simple_dir_inode_operations
;
2880 extern void make_empty_dir_inode(struct inode
*inode
);
2881 extern bool is_empty_dir_inode(struct inode
*inode
);
2882 struct tree_descr
{ char *name
; const struct file_operations
*ops
; int mode
; };
2883 struct dentry
*d_alloc_name(struct dentry
*, const char *);
2884 extern int simple_fill_super(struct super_block
*, unsigned long, struct tree_descr
*);
2885 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
2886 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
2888 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
2889 loff_t
*ppos
, const void *from
, size_t available
);
2890 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
2891 const void __user
*from
, size_t count
);
2893 extern int __generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
2894 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
2896 extern int generic_check_addressable(unsigned, u64
);
2898 #ifdef CONFIG_MIGRATION
2899 extern int buffer_migrate_page(struct address_space
*,
2900 struct page
*, struct page
*,
2903 #define buffer_migrate_page NULL
2906 extern int inode_change_ok(const struct inode
*, struct iattr
*);
2907 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
2908 extern void setattr_copy(struct inode
*inode
, const struct iattr
*attr
);
2910 extern int file_update_time(struct file
*file
);
2912 extern int generic_show_options(struct seq_file
*m
, struct dentry
*root
);
2913 extern void save_mount_options(struct super_block
*sb
, char *options
);
2914 extern void replace_mount_options(struct super_block
*sb
, char *options
);
2916 static inline bool io_is_direct(struct file
*filp
)
2918 return (filp
->f_flags
& O_DIRECT
) || IS_DAX(filp
->f_mapping
->host
);
2921 static inline int iocb_flags(struct file
*file
)
2924 if (file
->f_flags
& O_APPEND
)
2926 if (io_is_direct(file
))
2931 static inline ino_t
parent_ino(struct dentry
*dentry
)
2936 * Don't strictly need d_lock here? If the parent ino could change
2937 * then surely we'd have a deeper race in the caller?
2939 spin_lock(&dentry
->d_lock
);
2940 res
= dentry
->d_parent
->d_inode
->i_ino
;
2941 spin_unlock(&dentry
->d_lock
);
2945 /* Transaction based IO helpers */
2948 * An argresp is stored in an allocated page and holds the
2949 * size of the argument or response, along with its content
2951 struct simple_transaction_argresp
{
2956 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2958 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
2960 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
2961 size_t size
, loff_t
*pos
);
2962 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
2964 void simple_transaction_set(struct file
*file
, size_t n
);
2967 * simple attribute files
2969 * These attributes behave similar to those in sysfs:
2971 * Writing to an attribute immediately sets a value, an open file can be
2972 * written to multiple times.
2974 * Reading from an attribute creates a buffer from the value that might get
2975 * read with multiple read calls. When the attribute has been read
2976 * completely, no further read calls are possible until the file is opened
2979 * All attributes contain a text representation of a numeric value
2980 * that are accessed with the get() and set() functions.
2982 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2983 static int __fops ## _open(struct inode *inode, struct file *file) \
2985 __simple_attr_check_format(__fmt, 0ull); \
2986 return simple_attr_open(inode, file, __get, __set, __fmt); \
2988 static const struct file_operations __fops = { \
2989 .owner = THIS_MODULE, \
2990 .open = __fops ## _open, \
2991 .release = simple_attr_release, \
2992 .read = simple_attr_read, \
2993 .write = simple_attr_write, \
2994 .llseek = generic_file_llseek, \
2997 static inline __printf(1, 2)
2998 void __simple_attr_check_format(const char *fmt
, ...)
3000 /* don't do anything, just let the compiler check the arguments; */
3003 int simple_attr_open(struct inode
*inode
, struct file
*file
,
3004 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
3006 int simple_attr_release(struct inode
*inode
, struct file
*file
);
3007 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
3008 size_t len
, loff_t
*ppos
);
3009 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
3010 size_t len
, loff_t
*ppos
);
3013 int proc_nr_files(struct ctl_table
*table
, int write
,
3014 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3015 int proc_nr_dentry(struct ctl_table
*table
, int write
,
3016 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3017 int proc_nr_inodes(struct ctl_table
*table
, int write
,
3018 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3019 int __init
get_filesystem_list(char *buf
);
3021 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3022 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3024 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3025 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3026 (flag & __FMODE_NONOTIFY)))
3028 static inline bool is_sxid(umode_t mode
)
3030 return (mode
& S_ISUID
) || ((mode
& S_ISGID
) && (mode
& S_IXGRP
));
3033 static inline int check_sticky(struct inode
*dir
, struct inode
*inode
)
3035 if (!(dir
->i_mode
& S_ISVTX
))
3038 return __check_sticky(dir
, inode
);
3041 static inline void inode_has_no_xattr(struct inode
*inode
)
3043 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& MS_NOSEC
))
3044 inode
->i_flags
|= S_NOSEC
;
3047 static inline bool is_root_inode(struct inode
*inode
)
3049 return inode
== inode
->i_sb
->s_root
->d_inode
;
3052 static inline bool dir_emit(struct dir_context
*ctx
,
3053 const char *name
, int namelen
,
3054 u64 ino
, unsigned type
)
3056 return ctx
->actor(ctx
, name
, namelen
, ctx
->pos
, ino
, type
) == 0;
3058 static inline bool dir_emit_dot(struct file
*file
, struct dir_context
*ctx
)
3060 return ctx
->actor(ctx
, ".", 1, ctx
->pos
,
3061 file
->f_path
.dentry
->d_inode
->i_ino
, DT_DIR
) == 0;
3063 static inline bool dir_emit_dotdot(struct file
*file
, struct dir_context
*ctx
)
3065 return ctx
->actor(ctx
, "..", 2, ctx
->pos
,
3066 parent_ino(file
->f_path
.dentry
), DT_DIR
) == 0;
3068 static inline bool dir_emit_dots(struct file
*file
, struct dir_context
*ctx
)
3070 if (ctx
->pos
== 0) {
3071 if (!dir_emit_dot(file
, ctx
))
3075 if (ctx
->pos
== 1) {
3076 if (!dir_emit_dotdot(file
, ctx
))
3082 static inline bool dir_relax(struct inode
*inode
)
3084 inode_unlock(inode
);
3086 return !IS_DEADDIR(inode
);
3089 extern bool path_noexec(const struct path
*path
);
3090 extern void inode_nohighmem(struct inode
*inode
);
3092 #endif /* _LINUX_FS_H */