5 #include <linux/linkage.h>
6 #include <linux/wait.h>
7 #include <linux/kdev_t.h>
8 #include <linux/dcache.h>
9 #include <linux/path.h>
10 #include <linux/stat.h>
11 #include <linux/cache.h>
12 #include <linux/list.h>
13 #include <linux/list_lru.h>
14 #include <linux/llist.h>
15 #include <linux/radix-tree.h>
16 #include <linux/rbtree.h>
17 #include <linux/init.h>
18 #include <linux/pid.h>
19 #include <linux/bug.h>
20 #include <linux/mutex.h>
21 #include <linux/rwsem.h>
22 #include <linux/capability.h>
23 #include <linux/semaphore.h>
24 #include <linux/fiemap.h>
25 #include <linux/rculist_bl.h>
26 #include <linux/atomic.h>
27 #include <linux/shrinker.h>
28 #include <linux/migrate_mode.h>
29 #include <linux/uidgid.h>
30 #include <linux/lockdep.h>
31 #include <linux/percpu-rwsem.h>
32 #include <linux/blk_types.h>
34 #include <asm/byteorder.h>
35 #include <uapi/linux/fs.h>
37 struct backing_dev_info
;
38 struct export_operations
;
44 struct pipe_inode_info
;
45 struct poll_table_struct
;
47 struct vm_area_struct
;
50 struct swap_info_struct
;
52 struct workqueue_struct
;
56 extern void __init
inode_init(void);
57 extern void __init
inode_init_early(void);
58 extern void __init
files_init(unsigned long);
60 extern struct files_stat_struct files_stat
;
61 extern unsigned long get_max_files(void);
62 extern int sysctl_nr_open
;
63 extern struct inodes_stat_t inodes_stat
;
64 extern int leases_enable
, lease_break_time
;
65 extern int sysctl_protected_symlinks
;
66 extern int sysctl_protected_hardlinks
;
69 typedef int (get_block_t
)(struct inode
*inode
, sector_t iblock
,
70 struct buffer_head
*bh_result
, int create
);
71 typedef void (dio_iodone_t
)(struct kiocb
*iocb
, loff_t offset
,
72 ssize_t bytes
, void *private);
74 #define MAY_EXEC 0x00000001
75 #define MAY_WRITE 0x00000002
76 #define MAY_READ 0x00000004
77 #define MAY_APPEND 0x00000008
78 #define MAY_ACCESS 0x00000010
79 #define MAY_OPEN 0x00000020
80 #define MAY_CHDIR 0x00000040
81 /* called from RCU mode, don't block */
82 #define MAY_NOT_BLOCK 0x00000080
85 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
86 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
89 /* file is open for reading */
90 #define FMODE_READ ((__force fmode_t)0x1)
91 /* file is open for writing */
92 #define FMODE_WRITE ((__force fmode_t)0x2)
93 /* file is seekable */
94 #define FMODE_LSEEK ((__force fmode_t)0x4)
95 /* file can be accessed using pread */
96 #define FMODE_PREAD ((__force fmode_t)0x8)
97 /* file can be accessed using pwrite */
98 #define FMODE_PWRITE ((__force fmode_t)0x10)
99 /* File is opened for execution with sys_execve / sys_uselib */
100 #define FMODE_EXEC ((__force fmode_t)0x20)
101 /* File is opened with O_NDELAY (only set for block devices) */
102 #define FMODE_NDELAY ((__force fmode_t)0x40)
103 /* File is opened with O_EXCL (only set for block devices) */
104 #define FMODE_EXCL ((__force fmode_t)0x80)
105 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
106 (specialy hack for floppy.c) */
107 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
108 /* 32bit hashes as llseek() offset (for directories) */
109 #define FMODE_32BITHASH ((__force fmode_t)0x200)
110 /* 64bit hashes as llseek() offset (for directories) */
111 #define FMODE_64BITHASH ((__force fmode_t)0x400)
114 * Don't update ctime and mtime.
116 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
117 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
119 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
121 /* Expect random access pattern */
122 #define FMODE_RANDOM ((__force fmode_t)0x1000)
124 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
125 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
127 /* File is opened with O_PATH; almost nothing can be done with it */
128 #define FMODE_PATH ((__force fmode_t)0x4000)
130 /* File needs atomic accesses to f_pos */
131 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
132 /* Write access to underlying fs */
133 #define FMODE_WRITER ((__force fmode_t)0x10000)
134 /* Has read method(s) */
135 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
136 /* Has write method(s) */
137 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
139 /* File was opened by fanotify and shouldn't generate fanotify events */
140 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
143 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
144 * that indicates that they should check the contents of the iovec are
145 * valid, but not check the memory that the iovec elements
148 #define CHECK_IOVEC_ONLY -1
151 * The below are the various read and write types that we support. Some of
152 * them include behavioral modifiers that send information down to the
153 * block layer and IO scheduler. Terminology:
155 * The block layer uses device plugging to defer IO a little bit, in
156 * the hope that we will see more IO very shortly. This increases
157 * coalescing of adjacent IO and thus reduces the number of IOs we
158 * have to send to the device. It also allows for better queuing,
159 * if the IO isn't mergeable. If the caller is going to be waiting
160 * for the IO, then he must ensure that the device is unplugged so
161 * that the IO is dispatched to the driver.
163 * All IO is handled async in Linux. This is fine for background
164 * writes, but for reads or writes that someone waits for completion
165 * on, we want to notify the block layer and IO scheduler so that they
166 * know about it. That allows them to make better scheduling
167 * decisions. So when the below references 'sync' and 'async', it
168 * is referencing this priority hint.
170 * With that in mind, the available types are:
172 * READ A normal read operation. Device will be plugged.
173 * READ_SYNC A synchronous read. Device is not plugged, caller can
174 * immediately wait on this read without caring about
176 * READA Used for read-ahead operations. Lower priority, and the
177 * block layer could (in theory) choose to ignore this
178 * request if it runs into resource problems.
179 * WRITE A normal async write. Device will be plugged.
180 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
181 * the hint that someone will be waiting on this IO
182 * shortly. The write equivalent of READ_SYNC.
183 * WRITE_ODIRECT Special case write for O_DIRECT only.
184 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
185 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
186 * non-volatile media on completion.
187 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
188 * by a cache flush and data is guaranteed to be on
189 * non-volatile media on completion.
192 #define RW_MASK REQ_WRITE
193 #define RWA_MASK REQ_RAHEAD
196 #define WRITE RW_MASK
197 #define READA RWA_MASK
199 #define READ_SYNC (READ | REQ_SYNC)
200 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
201 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
202 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
203 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
204 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
207 * Attribute flags. These should be or-ed together to figure out what
210 #define ATTR_MODE (1 << 0)
211 #define ATTR_UID (1 << 1)
212 #define ATTR_GID (1 << 2)
213 #define ATTR_SIZE (1 << 3)
214 #define ATTR_ATIME (1 << 4)
215 #define ATTR_MTIME (1 << 5)
216 #define ATTR_CTIME (1 << 6)
217 #define ATTR_ATIME_SET (1 << 7)
218 #define ATTR_MTIME_SET (1 << 8)
219 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
220 #define ATTR_ATTR_FLAG (1 << 10)
221 #define ATTR_KILL_SUID (1 << 11)
222 #define ATTR_KILL_SGID (1 << 12)
223 #define ATTR_FILE (1 << 13)
224 #define ATTR_KILL_PRIV (1 << 14)
225 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
226 #define ATTR_TIMES_SET (1 << 16)
229 * Whiteout is represented by a char device. The following constants define the
230 * mode and device number to use.
232 #define WHITEOUT_MODE 0
233 #define WHITEOUT_DEV 0
236 * This is the Inode Attributes structure, used for notify_change(). It
237 * uses the above definitions as flags, to know which values have changed.
238 * Also, in this manner, a Filesystem can look at only the values it cares
239 * about. Basically, these are the attributes that the VFS layer can
240 * request to change from the FS layer.
242 * Derek Atkins <warlord@MIT.EDU> 94-10-20
245 unsigned int ia_valid
;
250 struct timespec ia_atime
;
251 struct timespec ia_mtime
;
252 struct timespec ia_ctime
;
255 * Not an attribute, but an auxiliary info for filesystems wanting to
256 * implement an ftruncate() like method. NOTE: filesystem should
257 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
259 struct file
*ia_file
;
263 * Includes for diskquotas.
265 #include <linux/quota.h>
268 * Maximum number of layers of fs stack. Needs to be limited to
269 * prevent kernel stack overflow
271 #define FILESYSTEM_MAX_STACK_DEPTH 2
274 * enum positive_aop_returns - aop return codes with specific semantics
276 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
277 * completed, that the page is still locked, and
278 * should be considered active. The VM uses this hint
279 * to return the page to the active list -- it won't
280 * be a candidate for writeback again in the near
281 * future. Other callers must be careful to unlock
282 * the page if they get this return. Returned by
285 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
286 * unlocked it and the page might have been truncated.
287 * The caller should back up to acquiring a new page and
288 * trying again. The aop will be taking reasonable
289 * precautions not to livelock. If the caller held a page
290 * reference, it should drop it before retrying. Returned
293 * address_space_operation functions return these large constants to indicate
294 * special semantics to the caller. These are much larger than the bytes in a
295 * page to allow for functions that return the number of bytes operated on in a
299 enum positive_aop_returns
{
300 AOP_WRITEPAGE_ACTIVATE
= 0x80000,
301 AOP_TRUNCATED_PAGE
= 0x80001,
304 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
305 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
306 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
307 * helper code (eg buffer layer)
308 * to clear GFP_FS from alloc */
311 * oh the beauties of C type declarations.
314 struct address_space
;
315 struct writeback_control
;
317 #define IOCB_EVENTFD (1 << 0)
318 #define IOCB_APPEND (1 << 1)
319 #define IOCB_DIRECT (1 << 2)
322 struct file
*ki_filp
;
324 void (*ki_complete
)(struct kiocb
*iocb
, long ret
, long ret2
);
329 static inline bool is_sync_kiocb(struct kiocb
*kiocb
)
331 return kiocb
->ki_complete
== NULL
;
334 static inline int iocb_flags(struct file
*file
);
336 static inline void init_sync_kiocb(struct kiocb
*kiocb
, struct file
*filp
)
338 *kiocb
= (struct kiocb
) {
340 .ki_flags
= iocb_flags(filp
),
345 * "descriptor" for what we're up to with a read.
346 * This allows us to use the same read code yet
347 * have multiple different users of the data that
348 * we read from a file.
350 * The simplest case just copies the data to user
363 typedef int (*read_actor_t
)(read_descriptor_t
*, struct page
*,
364 unsigned long, unsigned long);
366 struct address_space_operations
{
367 int (*writepage
)(struct page
*page
, struct writeback_control
*wbc
);
368 int (*readpage
)(struct file
*, struct page
*);
370 /* Write back some dirty pages from this mapping. */
371 int (*writepages
)(struct address_space
*, struct writeback_control
*);
373 /* Set a page dirty. Return true if this dirtied it */
374 int (*set_page_dirty
)(struct page
*page
);
376 int (*readpages
)(struct file
*filp
, struct address_space
*mapping
,
377 struct list_head
*pages
, unsigned nr_pages
);
379 int (*write_begin
)(struct file
*, struct address_space
*mapping
,
380 loff_t pos
, unsigned len
, unsigned flags
,
381 struct page
**pagep
, void **fsdata
);
382 int (*write_end
)(struct file
*, struct address_space
*mapping
,
383 loff_t pos
, unsigned len
, unsigned copied
,
384 struct page
*page
, void *fsdata
);
386 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
387 sector_t (*bmap
)(struct address_space
*, sector_t
);
388 void (*invalidatepage
) (struct page
*, unsigned int, unsigned int);
389 int (*releasepage
) (struct page
*, gfp_t
);
390 void (*freepage
)(struct page
*);
391 ssize_t (*direct_IO
)(struct kiocb
*, struct iov_iter
*iter
, loff_t offset
);
393 * migrate the contents of a page to the specified target. If
394 * migrate_mode is MIGRATE_ASYNC, it must not block.
396 int (*migratepage
) (struct address_space
*,
397 struct page
*, struct page
*, enum migrate_mode
);
398 int (*launder_page
) (struct page
*);
399 int (*is_partially_uptodate
) (struct page
*, unsigned long,
401 void (*is_dirty_writeback
) (struct page
*, bool *, bool *);
402 int (*error_remove_page
)(struct address_space
*, struct page
*);
404 /* swapfile support */
405 int (*swap_activate
)(struct swap_info_struct
*sis
, struct file
*file
,
407 void (*swap_deactivate
)(struct file
*file
);
410 extern const struct address_space_operations empty_aops
;
413 * pagecache_write_begin/pagecache_write_end must be used by general code
414 * to write into the pagecache.
416 int pagecache_write_begin(struct file
*, struct address_space
*mapping
,
417 loff_t pos
, unsigned len
, unsigned flags
,
418 struct page
**pagep
, void **fsdata
);
420 int pagecache_write_end(struct file
*, struct address_space
*mapping
,
421 loff_t pos
, unsigned len
, unsigned copied
,
422 struct page
*page
, void *fsdata
);
424 struct address_space
{
425 struct inode
*host
; /* owner: inode, block_device */
426 struct radix_tree_root page_tree
; /* radix tree of all pages */
427 spinlock_t tree_lock
; /* and lock protecting it */
428 atomic_t i_mmap_writable
;/* count VM_SHARED mappings */
429 struct rb_root i_mmap
; /* tree of private and shared mappings */
430 struct rw_semaphore i_mmap_rwsem
; /* protect tree, count, list */
431 /* Protected by tree_lock together with the radix tree */
432 unsigned long nrpages
; /* number of total pages */
433 unsigned long nrshadows
; /* number of shadow entries */
434 pgoff_t writeback_index
;/* writeback starts here */
435 const struct address_space_operations
*a_ops
; /* methods */
436 unsigned long flags
; /* error bits/gfp mask */
437 spinlock_t private_lock
; /* for use by the address_space */
438 struct list_head private_list
; /* ditto */
439 void *private_data
; /* ditto */
440 } __attribute__((aligned(sizeof(long))));
442 * On most architectures that alignment is already the case; but
443 * must be enforced here for CRIS, to let the least significant bit
444 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
446 struct request_queue
;
448 struct block_device
{
449 dev_t bd_dev
; /* not a kdev_t - it's a search key */
451 struct inode
* bd_inode
; /* will die */
452 struct super_block
* bd_super
;
453 struct mutex bd_mutex
; /* open/close mutex */
454 struct list_head bd_inodes
;
458 bool bd_write_holder
;
460 struct list_head bd_holder_disks
;
462 struct block_device
* bd_contains
;
463 unsigned bd_block_size
;
464 struct hd_struct
* bd_part
;
465 /* number of times partitions within this device have been opened. */
466 unsigned bd_part_count
;
468 struct gendisk
* bd_disk
;
469 struct request_queue
* bd_queue
;
470 struct list_head bd_list
;
472 * Private data. You must have bd_claim'ed the block_device
473 * to use this. NOTE: bd_claim allows an owner to claim
474 * the same device multiple times, the owner must take special
475 * care to not mess up bd_private for that case.
477 unsigned long bd_private
;
479 /* The counter of freeze processes */
480 int bd_fsfreeze_count
;
481 /* Mutex for freeze */
482 struct mutex bd_fsfreeze_mutex
;
486 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
489 #define PAGECACHE_TAG_DIRTY 0
490 #define PAGECACHE_TAG_WRITEBACK 1
491 #define PAGECACHE_TAG_TOWRITE 2
493 int mapping_tagged(struct address_space
*mapping
, int tag
);
495 static inline void i_mmap_lock_write(struct address_space
*mapping
)
497 down_write(&mapping
->i_mmap_rwsem
);
500 static inline void i_mmap_unlock_write(struct address_space
*mapping
)
502 up_write(&mapping
->i_mmap_rwsem
);
505 static inline void i_mmap_lock_read(struct address_space
*mapping
)
507 down_read(&mapping
->i_mmap_rwsem
);
510 static inline void i_mmap_unlock_read(struct address_space
*mapping
)
512 up_read(&mapping
->i_mmap_rwsem
);
516 * Might pages of this file be mapped into userspace?
518 static inline int mapping_mapped(struct address_space
*mapping
)
520 return !RB_EMPTY_ROOT(&mapping
->i_mmap
);
524 * Might pages of this file have been modified in userspace?
525 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
526 * marks vma as VM_SHARED if it is shared, and the file was opened for
527 * writing i.e. vma may be mprotected writable even if now readonly.
529 * If i_mmap_writable is negative, no new writable mappings are allowed. You
530 * can only deny writable mappings, if none exists right now.
532 static inline int mapping_writably_mapped(struct address_space
*mapping
)
534 return atomic_read(&mapping
->i_mmap_writable
) > 0;
537 static inline int mapping_map_writable(struct address_space
*mapping
)
539 return atomic_inc_unless_negative(&mapping
->i_mmap_writable
) ?
543 static inline void mapping_unmap_writable(struct address_space
*mapping
)
545 atomic_dec(&mapping
->i_mmap_writable
);
548 static inline int mapping_deny_writable(struct address_space
*mapping
)
550 return atomic_dec_unless_positive(&mapping
->i_mmap_writable
) ?
554 static inline void mapping_allow_writable(struct address_space
*mapping
)
556 atomic_inc(&mapping
->i_mmap_writable
);
560 * Use sequence counter to get consistent i_size on 32-bit processors.
562 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
563 #include <linux/seqlock.h>
564 #define __NEED_I_SIZE_ORDERED
565 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
567 #define i_size_ordered_init(inode) do { } while (0)
571 #define ACL_NOT_CACHED ((void *)(-1))
573 #define IOP_FASTPERM 0x0001
574 #define IOP_LOOKUP 0x0002
575 #define IOP_NOFOLLOW 0x0004
578 * Keep mostly read-only and often accessed (especially for
579 * the RCU path lookup and 'stat' data) fields at the beginning
580 * of the 'struct inode'
584 unsigned short i_opflags
;
587 unsigned int i_flags
;
589 #ifdef CONFIG_FS_POSIX_ACL
590 struct posix_acl
*i_acl
;
591 struct posix_acl
*i_default_acl
;
594 const struct inode_operations
*i_op
;
595 struct super_block
*i_sb
;
596 struct address_space
*i_mapping
;
598 #ifdef CONFIG_SECURITY
602 /* Stat data, not accessed from path walking */
605 * Filesystems may only read i_nlink directly. They shall use the
606 * following functions for modification:
608 * (set|clear|inc|drop)_nlink
609 * inode_(inc|dec)_link_count
612 const unsigned int i_nlink
;
613 unsigned int __i_nlink
;
617 struct timespec i_atime
;
618 struct timespec i_mtime
;
619 struct timespec i_ctime
;
620 spinlock_t i_lock
; /* i_blocks, i_bytes, maybe i_size */
621 unsigned short i_bytes
;
622 unsigned int i_blkbits
;
625 #ifdef __NEED_I_SIZE_ORDERED
626 seqcount_t i_size_seqcount
;
630 unsigned long i_state
;
631 struct mutex i_mutex
;
633 unsigned long dirtied_when
; /* jiffies of first dirtying */
634 unsigned long dirtied_time_when
;
636 struct hlist_node i_hash
;
637 struct list_head i_wb_list
; /* backing dev IO list */
638 struct list_head i_lru
; /* inode LRU list */
639 struct list_head i_sb_list
;
641 struct hlist_head i_dentry
;
642 struct rcu_head i_rcu
;
646 atomic_t i_dio_count
;
647 atomic_t i_writecount
;
649 atomic_t i_readcount
; /* struct files open RO */
651 const struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
652 struct file_lock_context
*i_flctx
;
653 struct address_space i_data
;
654 struct list_head i_devices
;
656 struct pipe_inode_info
*i_pipe
;
657 struct block_device
*i_bdev
;
664 #ifdef CONFIG_FSNOTIFY
665 __u32 i_fsnotify_mask
; /* all events this inode cares about */
666 struct hlist_head i_fsnotify_marks
;
669 void *i_private
; /* fs or device private pointer */
672 static inline int inode_unhashed(struct inode
*inode
)
674 return hlist_unhashed(&inode
->i_hash
);
678 * inode->i_mutex nesting subclasses for the lock validator:
680 * 0: the object of the current VFS operation
684 * 4: second non-directory
685 * 5: second parent (when locking independent directories in rename)
687 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
688 * non-directories at once.
690 * The locking order between these classes is
691 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
693 enum inode_i_mutex_lock_class
703 void lock_two_nondirectories(struct inode
*, struct inode
*);
704 void unlock_two_nondirectories(struct inode
*, struct inode
*);
707 * NOTE: in a 32bit arch with a preemptable kernel and
708 * an UP compile the i_size_read/write must be atomic
709 * with respect to the local cpu (unlike with preempt disabled),
710 * but they don't need to be atomic with respect to other cpus like in
711 * true SMP (so they need either to either locally disable irq around
712 * the read or for example on x86 they can be still implemented as a
713 * cmpxchg8b without the need of the lock prefix). For SMP compiles
714 * and 64bit archs it makes no difference if preempt is enabled or not.
716 static inline loff_t
i_size_read(const struct inode
*inode
)
718 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
723 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
724 i_size
= inode
->i_size
;
725 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
727 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
731 i_size
= inode
->i_size
;
735 return inode
->i_size
;
740 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
741 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
742 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
744 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
746 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
748 write_seqcount_begin(&inode
->i_size_seqcount
);
749 inode
->i_size
= i_size
;
750 write_seqcount_end(&inode
->i_size_seqcount
);
752 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
754 inode
->i_size
= i_size
;
757 inode
->i_size
= i_size
;
761 /* Helper functions so that in most cases filesystems will
762 * not need to deal directly with kuid_t and kgid_t and can
763 * instead deal with the raw numeric values that are stored
766 static inline uid_t
i_uid_read(const struct inode
*inode
)
768 return from_kuid(&init_user_ns
, inode
->i_uid
);
771 static inline gid_t
i_gid_read(const struct inode
*inode
)
773 return from_kgid(&init_user_ns
, inode
->i_gid
);
776 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
778 inode
->i_uid
= make_kuid(&init_user_ns
, uid
);
781 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
783 inode
->i_gid
= make_kgid(&init_user_ns
, gid
);
786 static inline unsigned iminor(const struct inode
*inode
)
788 return MINOR(inode
->i_rdev
);
791 static inline unsigned imajor(const struct inode
*inode
)
793 return MAJOR(inode
->i_rdev
);
796 extern struct block_device
*I_BDEV(struct inode
*inode
);
799 rwlock_t lock
; /* protects pid, uid, euid fields */
800 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
801 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
802 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
803 int signum
; /* posix.1b rt signal to be delivered on IO */
807 * Track a single file's readahead state
809 struct file_ra_state
{
810 pgoff_t start
; /* where readahead started */
811 unsigned int size
; /* # of readahead pages */
812 unsigned int async_size
; /* do asynchronous readahead when
813 there are only # of pages ahead */
815 unsigned int ra_pages
; /* Maximum readahead window */
816 unsigned int mmap_miss
; /* Cache miss stat for mmap accesses */
817 loff_t prev_pos
; /* Cache last read() position */
821 * Check if @index falls in the readahead windows.
823 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
825 return (index
>= ra
->start
&&
826 index
< ra
->start
+ ra
->size
);
831 struct llist_node fu_llist
;
832 struct rcu_head fu_rcuhead
;
835 struct inode
*f_inode
; /* cached value */
836 const struct file_operations
*f_op
;
839 * Protects f_ep_links, f_flags.
840 * Must not be taken from IRQ context.
843 atomic_long_t f_count
;
844 unsigned int f_flags
;
846 struct mutex f_pos_lock
;
848 struct fown_struct f_owner
;
849 const struct cred
*f_cred
;
850 struct file_ra_state f_ra
;
853 #ifdef CONFIG_SECURITY
856 /* needed for tty driver, and maybe others */
860 /* Used by fs/eventpoll.c to link all the hooks to this file */
861 struct list_head f_ep_links
;
862 struct list_head f_tfile_llink
;
863 #endif /* #ifdef CONFIG_EPOLL */
864 struct address_space
*f_mapping
;
865 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
870 /* file identifier */
871 unsigned char f_handle
[0];
874 static inline struct file
*get_file(struct file
*f
)
876 atomic_long_inc(&f
->f_count
);
879 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
880 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
881 #define file_count(x) atomic_long_read(&(x)->f_count)
883 #define MAX_NON_LFS ((1UL<<31) - 1)
885 /* Page cache limit. The filesystems should put that into their s_maxbytes
886 limits, otherwise bad things can happen in VM. */
887 #if BITS_PER_LONG==32
888 #define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
889 #elif BITS_PER_LONG==64
890 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
895 #define FL_DELEG 4 /* NFSv4 delegation */
896 #define FL_ACCESS 8 /* not trying to lock, just looking */
897 #define FL_EXISTS 16 /* when unlocking, test for existence */
898 #define FL_LEASE 32 /* lease held on this file */
899 #define FL_CLOSE 64 /* unlock on close */
900 #define FL_SLEEP 128 /* A blocking lock */
901 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
902 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
903 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
904 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
907 * Special return value from posix_lock_file() and vfs_lock_file() for
908 * asynchronous locking.
910 #define FILE_LOCK_DEFERRED 1
912 /* legacy typedef, should eventually be removed */
913 typedef void *fl_owner_t
;
917 struct file_lock_operations
{
918 void (*fl_copy_lock
)(struct file_lock
*, struct file_lock
*);
919 void (*fl_release_private
)(struct file_lock
*);
922 struct lock_manager_operations
{
923 int (*lm_compare_owner
)(struct file_lock
*, struct file_lock
*);
924 unsigned long (*lm_owner_key
)(struct file_lock
*);
925 fl_owner_t (*lm_get_owner
)(fl_owner_t
);
926 void (*lm_put_owner
)(fl_owner_t
);
927 void (*lm_notify
)(struct file_lock
*); /* unblock callback */
928 int (*lm_grant
)(struct file_lock
*, int);
929 bool (*lm_break
)(struct file_lock
*);
930 int (*lm_change
)(struct file_lock
*, int, struct list_head
*);
931 void (*lm_setup
)(struct file_lock
*, void **);
934 struct lock_manager
{
935 struct list_head list
;
939 void locks_start_grace(struct net
*, struct lock_manager
*);
940 void locks_end_grace(struct lock_manager
*);
941 int locks_in_grace(struct net
*);
943 /* that will die - we need it for nfs_lock_info */
944 #include <linux/nfs_fs_i.h>
947 * struct file_lock represents a generic "file lock". It's used to represent
948 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
949 * note that the same struct is used to represent both a request for a lock and
950 * the lock itself, but the same object is never used for both.
952 * FIXME: should we create a separate "struct lock_request" to help distinguish
955 * The varous i_flctx lists are ordered by:
958 * 2) lock range start
961 * Obviously, the last two criteria only matter for POSIX locks.
964 struct file_lock
*fl_next
; /* singly linked list for this inode */
965 struct list_head fl_list
; /* link into file_lock_context */
966 struct hlist_node fl_link
; /* node in global lists */
967 struct list_head fl_block
; /* circular list of blocked processes */
969 unsigned int fl_flags
;
970 unsigned char fl_type
;
972 int fl_link_cpu
; /* what cpu's list is this on? */
973 struct pid
*fl_nspid
;
974 wait_queue_head_t fl_wait
;
975 struct file
*fl_file
;
979 struct fasync_struct
* fl_fasync
; /* for lease break notifications */
980 /* for lease breaks: */
981 unsigned long fl_break_time
;
982 unsigned long fl_downgrade_time
;
984 const struct file_lock_operations
*fl_ops
; /* Callbacks for filesystems */
985 const struct lock_manager_operations
*fl_lmops
; /* Callbacks for lockmanagers */
987 struct nfs_lock_info nfs_fl
;
988 struct nfs4_lock_info nfs4_fl
;
990 struct list_head link
; /* link in AFS vnode's pending_locks list */
991 int state
; /* state of grant or error if -ve */
996 struct file_lock_context
{
998 struct list_head flc_flock
;
999 struct list_head flc_posix
;
1000 struct list_head flc_lease
;
1003 /* The following constant reflects the upper bound of the file/locking space */
1005 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1006 #define OFFSET_MAX INT_LIMIT(loff_t)
1007 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1010 #include <linux/fcntl.h>
1012 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
1014 #ifdef CONFIG_FILE_LOCKING
1015 extern int fcntl_getlk(struct file
*, unsigned int, struct flock __user
*);
1016 extern int fcntl_setlk(unsigned int, struct file
*, unsigned int,
1017 struct flock __user
*);
1019 #if BITS_PER_LONG == 32
1020 extern int fcntl_getlk64(struct file
*, unsigned int, struct flock64 __user
*);
1021 extern int fcntl_setlk64(unsigned int, struct file
*, unsigned int,
1022 struct flock64 __user
*);
1025 extern int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
);
1026 extern int fcntl_getlease(struct file
*filp
);
1029 void locks_free_lock_context(struct file_lock_context
*ctx
);
1030 void locks_free_lock(struct file_lock
*fl
);
1031 extern void locks_init_lock(struct file_lock
*);
1032 extern struct file_lock
* locks_alloc_lock(void);
1033 extern void locks_copy_lock(struct file_lock
*, struct file_lock
*);
1034 extern void locks_copy_conflock(struct file_lock
*, struct file_lock
*);
1035 extern void locks_remove_posix(struct file
*, fl_owner_t
);
1036 extern void locks_remove_file(struct file
*);
1037 extern void locks_release_private(struct file_lock
*);
1038 extern void posix_test_lock(struct file
*, struct file_lock
*);
1039 extern int posix_lock_file(struct file
*, struct file_lock
*, struct file_lock
*);
1040 extern int posix_lock_file_wait(struct file
*, struct file_lock
*);
1041 extern int posix_unblock_lock(struct file_lock
*);
1042 extern int vfs_test_lock(struct file
*, struct file_lock
*);
1043 extern int vfs_lock_file(struct file
*, unsigned int, struct file_lock
*, struct file_lock
*);
1044 extern int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
);
1045 extern int flock_lock_file_wait(struct file
*filp
, struct file_lock
*fl
);
1046 extern int __break_lease(struct inode
*inode
, unsigned int flags
, unsigned int type
);
1047 extern void lease_get_mtime(struct inode
*, struct timespec
*time
);
1048 extern int generic_setlease(struct file
*, long, struct file_lock
**, void **priv
);
1049 extern int vfs_setlease(struct file
*, long, struct file_lock
**, void **);
1050 extern int lease_modify(struct file_lock
*, int, struct list_head
*);
1051 struct files_struct
;
1052 extern void show_fd_locks(struct seq_file
*f
,
1053 struct file
*filp
, struct files_struct
*files
);
1054 #else /* !CONFIG_FILE_LOCKING */
1055 static inline int fcntl_getlk(struct file
*file
, unsigned int cmd
,
1056 struct flock __user
*user
)
1061 static inline int fcntl_setlk(unsigned int fd
, struct file
*file
,
1062 unsigned int cmd
, struct flock __user
*user
)
1067 #if BITS_PER_LONG == 32
1068 static inline int fcntl_getlk64(struct file
*file
, unsigned int cmd
,
1069 struct flock64 __user
*user
)
1074 static inline int fcntl_setlk64(unsigned int fd
, struct file
*file
,
1075 unsigned int cmd
, struct flock64 __user
*user
)
1080 static inline int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
)
1085 static inline int fcntl_getlease(struct file
*filp
)
1091 locks_free_lock_context(struct file_lock_context
*ctx
)
1095 static inline void locks_init_lock(struct file_lock
*fl
)
1100 static inline void locks_copy_conflock(struct file_lock
*new, struct file_lock
*fl
)
1105 static inline void locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
1110 static inline void locks_remove_posix(struct file
*filp
, fl_owner_t owner
)
1115 static inline void locks_remove_file(struct file
*filp
)
1120 static inline void posix_test_lock(struct file
*filp
, struct file_lock
*fl
)
1125 static inline int posix_lock_file(struct file
*filp
, struct file_lock
*fl
,
1126 struct file_lock
*conflock
)
1131 static inline int posix_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1136 static inline int posix_unblock_lock(struct file_lock
*waiter
)
1141 static inline int vfs_test_lock(struct file
*filp
, struct file_lock
*fl
)
1146 static inline int vfs_lock_file(struct file
*filp
, unsigned int cmd
,
1147 struct file_lock
*fl
, struct file_lock
*conf
)
1152 static inline int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
)
1157 static inline int flock_lock_file_wait(struct file
*filp
,
1158 struct file_lock
*request
)
1163 static inline int __break_lease(struct inode
*inode
, unsigned int mode
, unsigned int type
)
1168 static inline void lease_get_mtime(struct inode
*inode
, struct timespec
*time
)
1173 static inline int generic_setlease(struct file
*filp
, long arg
,
1174 struct file_lock
**flp
, void **priv
)
1179 static inline int vfs_setlease(struct file
*filp
, long arg
,
1180 struct file_lock
**lease
, void **priv
)
1185 static inline int lease_modify(struct file_lock
*fl
, int arg
,
1186 struct list_head
*dispose
)
1191 struct files_struct
;
1192 static inline void show_fd_locks(struct seq_file
*f
,
1193 struct file
*filp
, struct files_struct
*files
) {}
1194 #endif /* !CONFIG_FILE_LOCKING */
1197 struct fasync_struct
{
1201 struct fasync_struct
*fa_next
; /* singly linked list */
1202 struct file
*fa_file
;
1203 struct rcu_head fa_rcu
;
1206 #define FASYNC_MAGIC 0x4601
1208 /* SMP safe fasync helpers: */
1209 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1210 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1211 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1212 extern struct fasync_struct
*fasync_alloc(void);
1213 extern void fasync_free(struct fasync_struct
*);
1215 /* can be called from interrupts */
1216 extern void kill_fasync(struct fasync_struct
**, int, int);
1218 extern void __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1219 extern void f_setown(struct file
*filp
, unsigned long arg
, int force
);
1220 extern void f_delown(struct file
*filp
);
1221 extern pid_t
f_getown(struct file
*filp
);
1222 extern int send_sigurg(struct fown_struct
*fown
);
1230 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1231 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1232 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1233 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1234 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1237 /* Possible states of 'frozen' field */
1239 SB_UNFROZEN
= 0, /* FS is unfrozen */
1240 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1241 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1242 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1243 * internal threads if needed) */
1244 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1247 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1250 /* Counters for counting writers at each level */
1251 struct percpu_counter counter
[SB_FREEZE_LEVELS
];
1252 wait_queue_head_t wait
; /* queue for waiting for
1253 writers / faults to finish */
1254 int frozen
; /* Is sb frozen? */
1255 wait_queue_head_t wait_unfrozen
; /* queue for waiting for
1257 #ifdef CONFIG_DEBUG_LOCK_ALLOC
1258 struct lockdep_map lock_map
[SB_FREEZE_LEVELS
];
1262 struct super_block
{
1263 struct list_head s_list
; /* Keep this first */
1264 dev_t s_dev
; /* search index; _not_ kdev_t */
1265 unsigned char s_blocksize_bits
;
1266 unsigned long s_blocksize
;
1267 loff_t s_maxbytes
; /* Max file size */
1268 struct file_system_type
*s_type
;
1269 const struct super_operations
*s_op
;
1270 const struct dquot_operations
*dq_op
;
1271 const struct quotactl_ops
*s_qcop
;
1272 const struct export_operations
*s_export_op
;
1273 unsigned long s_flags
;
1274 unsigned long s_magic
;
1275 struct dentry
*s_root
;
1276 struct rw_semaphore s_umount
;
1279 #ifdef CONFIG_SECURITY
1282 const struct xattr_handler
**s_xattr
;
1284 struct list_head s_inodes
; /* all inodes */
1285 struct hlist_bl_head s_anon
; /* anonymous dentries for (nfs) exporting */
1286 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1287 struct block_device
*s_bdev
;
1288 struct backing_dev_info
*s_bdi
;
1289 struct mtd_info
*s_mtd
;
1290 struct hlist_node s_instances
;
1291 unsigned int s_quota_types
; /* Bitmask of supported quota types */
1292 struct quota_info s_dquot
; /* Diskquota specific options */
1294 struct sb_writers s_writers
;
1296 char s_id
[32]; /* Informational name */
1297 u8 s_uuid
[16]; /* UUID */
1299 void *s_fs_info
; /* Filesystem private info */
1300 unsigned int s_max_links
;
1303 /* Granularity of c/m/atime in ns.
1304 Cannot be worse than a second */
1308 * The next field is for VFS *only*. No filesystems have any business
1309 * even looking at it. You had been warned.
1311 struct mutex s_vfs_rename_mutex
; /* Kludge */
1314 * Filesystem subtype. If non-empty the filesystem type field
1315 * in /proc/mounts will be "type.subtype"
1320 * Saved mount options for lazy filesystems using
1321 * generic_show_options()
1323 char __rcu
*s_options
;
1324 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1327 * Saved pool identifier for cleancache (-1 means none)
1329 int cleancache_poolid
;
1331 struct shrinker s_shrink
; /* per-sb shrinker handle */
1333 /* Number of inodes with nlink == 0 but still referenced */
1334 atomic_long_t s_remove_count
;
1336 /* Being remounted read-only */
1337 int s_readonly_remount
;
1339 /* AIO completions deferred from interrupt context */
1340 struct workqueue_struct
*s_dio_done_wq
;
1341 struct hlist_head s_pins
;
1344 * Keep the lru lists last in the structure so they always sit on their
1345 * own individual cachelines.
1347 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp
;
1348 struct list_lru s_inode_lru ____cacheline_aligned_in_smp
;
1349 struct rcu_head rcu
;
1352 * Indicates how deep in a filesystem stack this SB is
1357 extern struct timespec
current_fs_time(struct super_block
*sb
);
1360 * Snapshotting support.
1363 void __sb_end_write(struct super_block
*sb
, int level
);
1364 int __sb_start_write(struct super_block
*sb
, int level
, bool wait
);
1367 * sb_end_write - drop write access to a superblock
1368 * @sb: the super we wrote to
1370 * Decrement number of writers to the filesystem. Wake up possible waiters
1371 * wanting to freeze the filesystem.
1373 static inline void sb_end_write(struct super_block
*sb
)
1375 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1379 * sb_end_pagefault - drop write access to a superblock from a page fault
1380 * @sb: the super we wrote to
1382 * Decrement number of processes handling write page fault to the filesystem.
1383 * Wake up possible waiters wanting to freeze the filesystem.
1385 static inline void sb_end_pagefault(struct super_block
*sb
)
1387 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1391 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1392 * @sb: the super we wrote to
1394 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1395 * waiters wanting to freeze the filesystem.
1397 static inline void sb_end_intwrite(struct super_block
*sb
)
1399 __sb_end_write(sb
, SB_FREEZE_FS
);
1403 * sb_start_write - get write access to a superblock
1404 * @sb: the super we write to
1406 * When a process wants to write data or metadata to a file system (i.e. dirty
1407 * a page or an inode), it should embed the operation in a sb_start_write() -
1408 * sb_end_write() pair to get exclusion against file system freezing. This
1409 * function increments number of writers preventing freezing. If the file
1410 * system is already frozen, the function waits until the file system is
1413 * Since freeze protection behaves as a lock, users have to preserve
1414 * ordering of freeze protection and other filesystem locks. Generally,
1415 * freeze protection should be the outermost lock. In particular, we have:
1418 * -> i_mutex (write path, truncate, directory ops, ...)
1419 * -> s_umount (freeze_super, thaw_super)
1421 static inline void sb_start_write(struct super_block
*sb
)
1423 __sb_start_write(sb
, SB_FREEZE_WRITE
, true);
1426 static inline int sb_start_write_trylock(struct super_block
*sb
)
1428 return __sb_start_write(sb
, SB_FREEZE_WRITE
, false);
1432 * sb_start_pagefault - get write access to a superblock from a page fault
1433 * @sb: the super we write to
1435 * When a process starts handling write page fault, it should embed the
1436 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1437 * exclusion against file system freezing. This is needed since the page fault
1438 * is going to dirty a page. This function increments number of running page
1439 * faults preventing freezing. If the file system is already frozen, the
1440 * function waits until the file system is thawed.
1442 * Since page fault freeze protection behaves as a lock, users have to preserve
1443 * ordering of freeze protection and other filesystem locks. It is advised to
1444 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1445 * handling code implies lock dependency:
1448 * -> sb_start_pagefault
1450 static inline void sb_start_pagefault(struct super_block
*sb
)
1452 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
, true);
1456 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1457 * @sb: the super we write to
1459 * This is the third level of protection against filesystem freezing. It is
1460 * free for use by a filesystem. The only requirement is that it must rank
1461 * below sb_start_pagefault.
1463 * For example filesystem can call sb_start_intwrite() when starting a
1464 * transaction which somewhat eases handling of freezing for internal sources
1465 * of filesystem changes (internal fs threads, discarding preallocation on file
1468 static inline void sb_start_intwrite(struct super_block
*sb
)
1470 __sb_start_write(sb
, SB_FREEZE_FS
, true);
1474 extern bool inode_owner_or_capable(const struct inode
*inode
);
1477 * VFS helper functions..
1479 extern int vfs_create(struct inode
*, struct dentry
*, umode_t
, bool);
1480 extern int vfs_mkdir(struct inode
*, struct dentry
*, umode_t
);
1481 extern int vfs_mknod(struct inode
*, struct dentry
*, umode_t
, dev_t
);
1482 extern int vfs_symlink(struct inode
*, struct dentry
*, const char *);
1483 extern int vfs_link(struct dentry
*, struct inode
*, struct dentry
*, struct inode
**);
1484 extern int vfs_rmdir(struct inode
*, struct dentry
*);
1485 extern int vfs_unlink(struct inode
*, struct dentry
*, struct inode
**);
1486 extern int vfs_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*, struct inode
**, unsigned int);
1487 extern int vfs_whiteout(struct inode
*, struct dentry
*);
1490 * VFS dentry helper functions.
1492 extern void dentry_unhash(struct dentry
*dentry
);
1495 * VFS file helper functions.
1497 extern void inode_init_owner(struct inode
*inode
, const struct inode
*dir
,
1500 * VFS FS_IOC_FIEMAP helper definitions.
1502 struct fiemap_extent_info
{
1503 unsigned int fi_flags
; /* Flags as passed from user */
1504 unsigned int fi_extents_mapped
; /* Number of mapped extents */
1505 unsigned int fi_extents_max
; /* Size of fiemap_extent array */
1506 struct fiemap_extent __user
*fi_extents_start
; /* Start of
1507 fiemap_extent array */
1509 int fiemap_fill_next_extent(struct fiemap_extent_info
*info
, u64 logical
,
1510 u64 phys
, u64 len
, u32 flags
);
1511 int fiemap_check_flags(struct fiemap_extent_info
*fieinfo
, u32 fs_flags
);
1516 * NOTE! These match bits 12..15 of stat.st_mode
1517 * (ie "(i_mode >> 12) & 15").
1519 #define DT_UNKNOWN 0
1530 * This is the "filldir" function type, used by readdir() to let
1531 * the kernel specify what kind of dirent layout it wants to have.
1532 * This allows the kernel to read directories into kernel space or
1533 * to have different dirent layouts depending on the binary type.
1536 typedef int (*filldir_t
)(struct dir_context
*, const char *, int, loff_t
, u64
,
1539 struct dir_context
{
1540 const filldir_t actor
;
1544 struct block_device_operations
;
1546 /* These macros are for out of kernel modules to test that
1547 * the kernel supports the unlocked_ioctl and compat_ioctl
1548 * fields in struct file_operations. */
1549 #define HAVE_COMPAT_IOCTL 1
1550 #define HAVE_UNLOCKED_IOCTL 1
1553 * These flags let !MMU mmap() govern direct device mapping vs immediate
1554 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1556 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1557 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1558 * NOMMU_MAP_READ: Can be mapped for reading
1559 * NOMMU_MAP_WRITE: Can be mapped for writing
1560 * NOMMU_MAP_EXEC: Can be mapped for execution
1562 #define NOMMU_MAP_COPY 0x00000001
1563 #define NOMMU_MAP_DIRECT 0x00000008
1564 #define NOMMU_MAP_READ VM_MAYREAD
1565 #define NOMMU_MAP_WRITE VM_MAYWRITE
1566 #define NOMMU_MAP_EXEC VM_MAYEXEC
1568 #define NOMMU_VMFLAGS \
1569 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1574 struct file_operations
{
1575 struct module
*owner
;
1576 loff_t (*llseek
) (struct file
*, loff_t
, int);
1577 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
1578 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
1579 ssize_t (*read_iter
) (struct kiocb
*, struct iov_iter
*);
1580 ssize_t (*write_iter
) (struct kiocb
*, struct iov_iter
*);
1581 int (*iterate
) (struct file
*, struct dir_context
*);
1582 unsigned int (*poll
) (struct file
*, struct poll_table_struct
*);
1583 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
1584 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
1585 int (*mmap
) (struct file
*, struct vm_area_struct
*);
1586 int (*mremap
)(struct file
*, struct vm_area_struct
*);
1587 int (*open
) (struct inode
*, struct file
*);
1588 int (*flush
) (struct file
*, fl_owner_t id
);
1589 int (*release
) (struct inode
*, struct file
*);
1590 int (*fsync
) (struct file
*, loff_t
, loff_t
, int datasync
);
1591 int (*aio_fsync
) (struct kiocb
*, int datasync
);
1592 int (*fasync
) (int, struct file
*, int);
1593 int (*lock
) (struct file
*, int, struct file_lock
*);
1594 ssize_t (*sendpage
) (struct file
*, struct page
*, int, size_t, loff_t
*, int);
1595 unsigned long (*get_unmapped_area
)(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1596 int (*check_flags
)(int);
1597 int (*flock
) (struct file
*, int, struct file_lock
*);
1598 ssize_t (*splice_write
)(struct pipe_inode_info
*, struct file
*, loff_t
*, size_t, unsigned int);
1599 ssize_t (*splice_read
)(struct file
*, loff_t
*, struct pipe_inode_info
*, size_t, unsigned int);
1600 int (*setlease
)(struct file
*, long, struct file_lock
**, void **);
1601 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
1603 void (*show_fdinfo
)(struct seq_file
*m
, struct file
*f
);
1605 unsigned (*mmap_capabilities
)(struct file
*);
1609 struct inode_operations
{
1610 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*, unsigned int);
1611 const char * (*follow_link
) (struct dentry
*, void **, struct nameidata
*);
1612 int (*permission
) (struct inode
*, int);
1613 struct posix_acl
* (*get_acl
)(struct inode
*, int);
1615 int (*readlink
) (struct dentry
*, char __user
*,int);
1616 void (*put_link
) (struct dentry
*, void *);
1618 int (*create
) (struct inode
*,struct dentry
*, umode_t
, bool);
1619 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
1620 int (*unlink
) (struct inode
*,struct dentry
*);
1621 int (*symlink
) (struct inode
*,struct dentry
*,const char *);
1622 int (*mkdir
) (struct inode
*,struct dentry
*,umode_t
);
1623 int (*rmdir
) (struct inode
*,struct dentry
*);
1624 int (*mknod
) (struct inode
*,struct dentry
*,umode_t
,dev_t
);
1625 int (*rename
) (struct inode
*, struct dentry
*,
1626 struct inode
*, struct dentry
*);
1627 int (*rename2
) (struct inode
*, struct dentry
*,
1628 struct inode
*, struct dentry
*, unsigned int);
1629 int (*setattr
) (struct dentry
*, struct iattr
*);
1630 int (*getattr
) (struct vfsmount
*mnt
, struct dentry
*, struct kstat
*);
1631 int (*setxattr
) (struct dentry
*, const char *,const void *,size_t,int);
1632 ssize_t (*getxattr
) (struct dentry
*, const char *, void *, size_t);
1633 ssize_t (*listxattr
) (struct dentry
*, char *, size_t);
1634 int (*removexattr
) (struct dentry
*, const char *);
1635 int (*fiemap
)(struct inode
*, struct fiemap_extent_info
*, u64 start
,
1637 int (*update_time
)(struct inode
*, struct timespec
*, int);
1638 int (*atomic_open
)(struct inode
*, struct dentry
*,
1639 struct file
*, unsigned open_flag
,
1640 umode_t create_mode
, int *opened
);
1641 int (*tmpfile
) (struct inode
*, struct dentry
*, umode_t
);
1642 int (*set_acl
)(struct inode
*, struct posix_acl
*, int);
1644 /* WARNING: probably going away soon, do not use! */
1645 int (*dentry_open
)(struct dentry
*, struct file
*, const struct cred
*);
1646 } ____cacheline_aligned
;
1648 ssize_t
rw_copy_check_uvector(int type
, const struct iovec __user
* uvector
,
1649 unsigned long nr_segs
, unsigned long fast_segs
,
1650 struct iovec
*fast_pointer
,
1651 struct iovec
**ret_pointer
);
1653 extern ssize_t
__vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1654 extern ssize_t
__vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1655 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1656 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1657 extern ssize_t
vfs_readv(struct file
*, const struct iovec __user
*,
1658 unsigned long, loff_t
*);
1659 extern ssize_t
vfs_writev(struct file
*, const struct iovec __user
*,
1660 unsigned long, loff_t
*);
1662 struct super_operations
{
1663 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
1664 void (*destroy_inode
)(struct inode
*);
1666 void (*dirty_inode
) (struct inode
*, int flags
);
1667 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
1668 int (*drop_inode
) (struct inode
*);
1669 void (*evict_inode
) (struct inode
*);
1670 void (*put_super
) (struct super_block
*);
1671 int (*sync_fs
)(struct super_block
*sb
, int wait
);
1672 int (*freeze_super
) (struct super_block
*);
1673 int (*freeze_fs
) (struct super_block
*);
1674 int (*thaw_super
) (struct super_block
*);
1675 int (*unfreeze_fs
) (struct super_block
*);
1676 int (*statfs
) (struct dentry
*, struct kstatfs
*);
1677 int (*remount_fs
) (struct super_block
*, int *, char *);
1678 void (*umount_begin
) (struct super_block
*);
1680 int (*show_options
)(struct seq_file
*, struct dentry
*);
1681 int (*show_devname
)(struct seq_file
*, struct dentry
*);
1682 int (*show_path
)(struct seq_file
*, struct dentry
*);
1683 int (*show_stats
)(struct seq_file
*, struct dentry
*);
1685 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
1686 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
1687 struct dquot
**(*get_dquots
)(struct inode
*);
1689 int (*bdev_try_to_free_page
)(struct super_block
*, struct page
*, gfp_t
);
1690 long (*nr_cached_objects
)(struct super_block
*,
1691 struct shrink_control
*);
1692 long (*free_cached_objects
)(struct super_block
*,
1693 struct shrink_control
*);
1697 * Inode flags - they have no relation to superblock flags now
1699 #define S_SYNC 1 /* Writes are synced at once */
1700 #define S_NOATIME 2 /* Do not update access times */
1701 #define S_APPEND 4 /* Append-only file */
1702 #define S_IMMUTABLE 8 /* Immutable file */
1703 #define S_DEAD 16 /* removed, but still open directory */
1704 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1705 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1706 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1707 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1708 #define S_PRIVATE 512 /* Inode is fs-internal */
1709 #define S_IMA 1024 /* Inode has an associated IMA struct */
1710 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1711 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1712 #ifdef CONFIG_FS_DAX
1713 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1715 #define S_DAX 0 /* Make all the DAX code disappear */
1719 * Note that nosuid etc flags are inode-specific: setting some file-system
1720 * flags just means all the inodes inherit those flags by default. It might be
1721 * possible to override it selectively if you really wanted to with some
1722 * ioctl() that is not currently implemented.
1724 * Exception: MS_RDONLY is always applied to the entire file system.
1726 * Unfortunately, it is possible to change a filesystems flags with it mounted
1727 * with files in use. This means that all of the inodes will not have their
1728 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1729 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1731 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1733 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1734 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1735 ((inode)->i_flags & S_SYNC))
1736 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1737 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1738 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1739 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1740 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1742 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1743 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1744 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1745 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1747 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1748 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1749 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1750 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1751 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1752 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1753 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1754 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1756 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1757 (inode)->i_rdev == WHITEOUT_DEV)
1760 * Inode state bits. Protected by inode->i_lock
1762 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1763 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1765 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1766 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1767 * various stages of removing an inode.
1769 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1771 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1772 * fdatasync(). i_atime is the usual cause.
1773 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1774 * these changes separately from I_DIRTY_SYNC so that we
1775 * don't have to write inode on fdatasync() when only
1776 * mtime has changed in it.
1777 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1778 * I_NEW Serves as both a mutex and completion notification.
1779 * New inodes set I_NEW. If two processes both create
1780 * the same inode, one of them will release its inode and
1781 * wait for I_NEW to be released before returning.
1782 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1783 * also cause waiting on I_NEW, without I_NEW actually
1784 * being set. find_inode() uses this to prevent returning
1785 * nearly-dead inodes.
1786 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1787 * is zero. I_FREEING must be set when I_WILL_FREE is
1789 * I_FREEING Set when inode is about to be freed but still has dirty
1790 * pages or buffers attached or the inode itself is still
1792 * I_CLEAR Added by clear_inode(). In this state the inode is
1793 * clean and can be destroyed. Inode keeps I_FREEING.
1795 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1796 * prohibited for many purposes. iget() must wait for
1797 * the inode to be completely released, then create it
1798 * anew. Other functions will just ignore such inodes,
1799 * if appropriate. I_NEW is used for waiting.
1801 * I_SYNC Writeback of inode is running. The bit is set during
1802 * data writeback, and cleared with a wakeup on the bit
1803 * address once it is done. The bit is also used to pin
1804 * the inode in memory for flusher thread.
1806 * I_REFERENCED Marks the inode as recently references on the LRU list.
1808 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1810 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1812 #define I_DIRTY_SYNC (1 << 0)
1813 #define I_DIRTY_DATASYNC (1 << 1)
1814 #define I_DIRTY_PAGES (1 << 2)
1816 #define I_NEW (1 << __I_NEW)
1817 #define I_WILL_FREE (1 << 4)
1818 #define I_FREEING (1 << 5)
1819 #define I_CLEAR (1 << 6)
1821 #define I_SYNC (1 << __I_SYNC)
1822 #define I_REFERENCED (1 << 8)
1823 #define __I_DIO_WAKEUP 9
1824 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1825 #define I_LINKABLE (1 << 10)
1826 #define I_DIRTY_TIME (1 << 11)
1827 #define __I_DIRTY_TIME_EXPIRED 12
1828 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1830 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1831 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1833 extern void __mark_inode_dirty(struct inode
*, int);
1834 static inline void mark_inode_dirty(struct inode
*inode
)
1836 __mark_inode_dirty(inode
, I_DIRTY
);
1839 static inline void mark_inode_dirty_sync(struct inode
*inode
)
1841 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
1844 extern void inc_nlink(struct inode
*inode
);
1845 extern void drop_nlink(struct inode
*inode
);
1846 extern void clear_nlink(struct inode
*inode
);
1847 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
1849 static inline void inode_inc_link_count(struct inode
*inode
)
1852 mark_inode_dirty(inode
);
1855 static inline void inode_dec_link_count(struct inode
*inode
)
1858 mark_inode_dirty(inode
);
1862 * inode_inc_iversion - increments i_version
1863 * @inode: inode that need to be updated
1865 * Every time the inode is modified, the i_version field will be incremented.
1866 * The filesystem has to be mounted with i_version flag
1869 static inline void inode_inc_iversion(struct inode
*inode
)
1871 spin_lock(&inode
->i_lock
);
1873 spin_unlock(&inode
->i_lock
);
1876 enum file_time_flags
{
1883 extern void touch_atime(const struct path
*);
1884 static inline void file_accessed(struct file
*file
)
1886 if (!(file
->f_flags
& O_NOATIME
))
1887 touch_atime(&file
->f_path
);
1890 int sync_inode(struct inode
*inode
, struct writeback_control
*wbc
);
1891 int sync_inode_metadata(struct inode
*inode
, int wait
);
1893 struct file_system_type
{
1896 #define FS_REQUIRES_DEV 1
1897 #define FS_BINARY_MOUNTDATA 2
1898 #define FS_HAS_SUBTYPE 4
1899 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
1900 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
1901 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
1902 struct dentry
*(*mount
) (struct file_system_type
*, int,
1903 const char *, void *);
1904 void (*kill_sb
) (struct super_block
*);
1905 struct module
*owner
;
1906 struct file_system_type
* next
;
1907 struct hlist_head fs_supers
;
1909 struct lock_class_key s_lock_key
;
1910 struct lock_class_key s_umount_key
;
1911 struct lock_class_key s_vfs_rename_key
;
1912 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
1914 struct lock_class_key i_lock_key
;
1915 struct lock_class_key i_mutex_key
;
1916 struct lock_class_key i_mutex_dir_key
;
1919 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1921 extern struct dentry
*mount_ns(struct file_system_type
*fs_type
, int flags
,
1922 void *data
, int (*fill_super
)(struct super_block
*, void *, int));
1923 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
1924 int flags
, const char *dev_name
, void *data
,
1925 int (*fill_super
)(struct super_block
*, void *, int));
1926 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
1927 int flags
, void *data
,
1928 int (*fill_super
)(struct super_block
*, void *, int));
1929 extern struct dentry
*mount_nodev(struct file_system_type
*fs_type
,
1930 int flags
, void *data
,
1931 int (*fill_super
)(struct super_block
*, void *, int));
1932 extern struct dentry
*mount_subtree(struct vfsmount
*mnt
, const char *path
);
1933 void generic_shutdown_super(struct super_block
*sb
);
1934 void kill_block_super(struct super_block
*sb
);
1935 void kill_anon_super(struct super_block
*sb
);
1936 void kill_litter_super(struct super_block
*sb
);
1937 void deactivate_super(struct super_block
*sb
);
1938 void deactivate_locked_super(struct super_block
*sb
);
1939 int set_anon_super(struct super_block
*s
, void *data
);
1940 int get_anon_bdev(dev_t
*);
1941 void free_anon_bdev(dev_t
);
1942 struct super_block
*sget(struct file_system_type
*type
,
1943 int (*test
)(struct super_block
*,void *),
1944 int (*set
)(struct super_block
*,void *),
1945 int flags
, void *data
);
1946 extern struct dentry
*mount_pseudo(struct file_system_type
*, char *,
1947 const struct super_operations
*ops
,
1948 const struct dentry_operations
*dops
,
1951 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1952 #define fops_get(fops) \
1953 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1954 #define fops_put(fops) \
1955 do { if (fops) module_put((fops)->owner); } while(0)
1957 * This one is to be used *ONLY* from ->open() instances.
1958 * fops must be non-NULL, pinned down *and* module dependencies
1959 * should be sufficient to pin the caller down as well.
1961 #define replace_fops(f, fops) \
1963 struct file *__file = (f); \
1964 fops_put(__file->f_op); \
1965 BUG_ON(!(__file->f_op = (fops))); \
1968 extern int register_filesystem(struct file_system_type
*);
1969 extern int unregister_filesystem(struct file_system_type
*);
1970 extern struct vfsmount
*kern_mount_data(struct file_system_type
*, void *data
);
1971 #define kern_mount(type) kern_mount_data(type, NULL)
1972 extern void kern_unmount(struct vfsmount
*mnt
);
1973 extern int may_umount_tree(struct vfsmount
*);
1974 extern int may_umount(struct vfsmount
*);
1975 extern long do_mount(const char *, const char __user
*,
1976 const char *, unsigned long, void *);
1977 extern struct vfsmount
*collect_mounts(struct path
*);
1978 extern void drop_collected_mounts(struct vfsmount
*);
1979 extern int iterate_mounts(int (*)(struct vfsmount
*, void *), void *,
1981 extern int vfs_statfs(struct path
*, struct kstatfs
*);
1982 extern int user_statfs(const char __user
*, struct kstatfs
*);
1983 extern int fd_statfs(int, struct kstatfs
*);
1984 extern int vfs_ustat(dev_t
, struct kstatfs
*);
1985 extern int freeze_super(struct super_block
*super
);
1986 extern int thaw_super(struct super_block
*super
);
1987 extern bool our_mnt(struct vfsmount
*mnt
);
1988 extern bool fs_fully_visible(struct file_system_type
*);
1990 extern int current_umask(void);
1992 extern void ihold(struct inode
* inode
);
1993 extern void iput(struct inode
*);
1994 extern int generic_update_time(struct inode
*, struct timespec
*, int);
1996 static inline struct inode
*file_inode(const struct file
*f
)
2002 extern struct kobject
*fs_kobj
;
2004 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
2006 #define FLOCK_VERIFY_READ 1
2007 #define FLOCK_VERIFY_WRITE 2
2009 #ifdef CONFIG_FILE_LOCKING
2010 extern int locks_mandatory_locked(struct file
*);
2011 extern int locks_mandatory_area(int, struct inode
*, struct file
*, loff_t
, size_t);
2014 * Candidates for mandatory locking have the setgid bit set
2015 * but no group execute bit - an otherwise meaningless combination.
2018 static inline int __mandatory_lock(struct inode
*ino
)
2020 return (ino
->i_mode
& (S_ISGID
| S_IXGRP
)) == S_ISGID
;
2024 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2025 * otherwise these will be advisory locks
2028 static inline int mandatory_lock(struct inode
*ino
)
2030 return IS_MANDLOCK(ino
) && __mandatory_lock(ino
);
2033 static inline int locks_verify_locked(struct file
*file
)
2035 if (mandatory_lock(file_inode(file
)))
2036 return locks_mandatory_locked(file
);
2040 static inline int locks_verify_truncate(struct inode
*inode
,
2044 if (inode
->i_flctx
&& mandatory_lock(inode
))
2045 return locks_mandatory_area(
2046 FLOCK_VERIFY_WRITE
, inode
, filp
,
2047 size
< inode
->i_size
? size
: inode
->i_size
,
2048 (size
< inode
->i_size
? inode
->i_size
- size
2049 : size
- inode
->i_size
)
2054 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2057 * Since this check is lockless, we must ensure that any refcounts
2058 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2059 * could end up racing with tasks trying to set a new lease on this
2063 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2064 return __break_lease(inode
, mode
, FL_LEASE
);
2068 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2071 * Since this check is lockless, we must ensure that any refcounts
2072 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2073 * could end up racing with tasks trying to set a new lease on this
2077 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2078 return __break_lease(inode
, mode
, FL_DELEG
);
2082 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2086 ret
= break_deleg(inode
, O_WRONLY
|O_NONBLOCK
);
2087 if (ret
== -EWOULDBLOCK
&& delegated_inode
) {
2088 *delegated_inode
= inode
;
2094 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2098 ret
= break_deleg(*delegated_inode
, O_WRONLY
);
2099 iput(*delegated_inode
);
2100 *delegated_inode
= NULL
;
2104 static inline int break_layout(struct inode
*inode
, bool wait
)
2107 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2108 return __break_lease(inode
,
2109 wait
? O_WRONLY
: O_WRONLY
| O_NONBLOCK
,
2114 #else /* !CONFIG_FILE_LOCKING */
2115 static inline int locks_mandatory_locked(struct file
*file
)
2120 static inline int locks_mandatory_area(int rw
, struct inode
*inode
,
2121 struct file
*filp
, loff_t offset
,
2127 static inline int __mandatory_lock(struct inode
*inode
)
2132 static inline int mandatory_lock(struct inode
*inode
)
2137 static inline int locks_verify_locked(struct file
*file
)
2142 static inline int locks_verify_truncate(struct inode
*inode
, struct file
*filp
,
2148 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2153 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2158 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2163 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2169 static inline int break_layout(struct inode
*inode
, bool wait
)
2174 #endif /* CONFIG_FILE_LOCKING */
2179 const char *name
; /* pointer to actual string */
2180 const __user
char *uptr
; /* original userland pointer */
2181 struct audit_names
*aname
;
2186 extern long vfs_truncate(struct path
*, loff_t
);
2187 extern int do_truncate(struct dentry
*, loff_t start
, unsigned int time_attrs
,
2189 extern int vfs_fallocate(struct file
*file
, int mode
, loff_t offset
,
2191 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
2193 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
2194 extern struct file
*filp_open(const char *, int, umode_t
);
2195 extern struct file
*file_open_root(struct dentry
*, struct vfsmount
*,
2197 extern int vfs_open(const struct path
*, struct file
*, const struct cred
*);
2198 extern struct file
* dentry_open(const struct path
*, int, const struct cred
*);
2199 extern int filp_close(struct file
*, fl_owner_t id
);
2201 extern struct filename
*getname_flags(const char __user
*, int, int *);
2202 extern struct filename
*getname(const char __user
*);
2203 extern struct filename
*getname_kernel(const char *);
2204 extern void putname(struct filename
*name
);
2210 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
2211 int (*open
)(struct inode
*, struct file
*),
2213 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
2217 extern int ioctl_preallocate(struct file
*filp
, void __user
*argp
);
2220 extern void __init
vfs_caches_init_early(void);
2221 extern void __init
vfs_caches_init(unsigned long);
2223 extern struct kmem_cache
*names_cachep
;
2225 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2226 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2229 extern int register_blkdev(unsigned int, const char *);
2230 extern void unregister_blkdev(unsigned int, const char *);
2231 extern struct block_device
*bdget(dev_t
);
2232 extern struct block_device
*bdgrab(struct block_device
*bdev
);
2233 extern void bd_set_size(struct block_device
*, loff_t size
);
2234 extern void bd_forget(struct inode
*inode
);
2235 extern void bdput(struct block_device
*);
2236 extern void invalidate_bdev(struct block_device
*);
2237 extern void iterate_bdevs(void (*)(struct block_device
*, void *), void *);
2238 extern int sync_blockdev(struct block_device
*bdev
);
2239 extern void kill_bdev(struct block_device
*);
2240 extern struct super_block
*freeze_bdev(struct block_device
*);
2241 extern void emergency_thaw_all(void);
2242 extern int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
);
2243 extern int fsync_bdev(struct block_device
*);
2244 extern int sb_is_blkdev_sb(struct super_block
*sb
);
2246 static inline void bd_forget(struct inode
*inode
) {}
2247 static inline int sync_blockdev(struct block_device
*bdev
) { return 0; }
2248 static inline void kill_bdev(struct block_device
*bdev
) {}
2249 static inline void invalidate_bdev(struct block_device
*bdev
) {}
2251 static inline struct super_block
*freeze_bdev(struct block_device
*sb
)
2256 static inline int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
)
2261 static inline void iterate_bdevs(void (*f
)(struct block_device
*, void *), void *arg
)
2265 static inline int sb_is_blkdev_sb(struct super_block
*sb
)
2270 extern int sync_filesystem(struct super_block
*);
2271 extern const struct file_operations def_blk_fops
;
2272 extern const struct file_operations def_chr_fops
;
2274 extern int ioctl_by_bdev(struct block_device
*, unsigned, unsigned long);
2275 extern int blkdev_ioctl(struct block_device
*, fmode_t
, unsigned, unsigned long);
2276 extern long compat_blkdev_ioctl(struct file
*, unsigned, unsigned long);
2277 extern int blkdev_get(struct block_device
*bdev
, fmode_t mode
, void *holder
);
2278 extern struct block_device
*blkdev_get_by_path(const char *path
, fmode_t mode
,
2280 extern struct block_device
*blkdev_get_by_dev(dev_t dev
, fmode_t mode
,
2282 extern void blkdev_put(struct block_device
*bdev
, fmode_t mode
);
2284 extern int bd_link_disk_holder(struct block_device
*bdev
, struct gendisk
*disk
);
2285 extern void bd_unlink_disk_holder(struct block_device
*bdev
,
2286 struct gendisk
*disk
);
2288 static inline int bd_link_disk_holder(struct block_device
*bdev
,
2289 struct gendisk
*disk
)
2293 static inline void bd_unlink_disk_holder(struct block_device
*bdev
,
2294 struct gendisk
*disk
)
2301 #define CHRDEV_MAJOR_HASH_SIZE 255
2302 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
2303 extern int register_chrdev_region(dev_t
, unsigned, const char *);
2304 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
2305 unsigned int count
, const char *name
,
2306 const struct file_operations
*fops
);
2307 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
2308 unsigned int count
, const char *name
);
2309 extern void unregister_chrdev_region(dev_t
, unsigned);
2310 extern void chrdev_show(struct seq_file
*,off_t
);
2312 static inline int register_chrdev(unsigned int major
, const char *name
,
2313 const struct file_operations
*fops
)
2315 return __register_chrdev(major
, 0, 256, name
, fops
);
2318 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2320 __unregister_chrdev(major
, 0, 256, name
);
2323 /* fs/block_dev.c */
2324 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2325 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2328 #define BLKDEV_MAJOR_HASH_SIZE 255
2329 extern const char *__bdevname(dev_t
, char *buffer
);
2330 extern const char *bdevname(struct block_device
*bdev
, char *buffer
);
2331 extern struct block_device
*lookup_bdev(const char *);
2332 extern void blkdev_show(struct seq_file
*,off_t
);
2335 #define BLKDEV_MAJOR_HASH_SIZE 0
2338 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2340 /* Invalid inode operations -- fs/bad_inode.c */
2341 extern void make_bad_inode(struct inode
*);
2342 extern int is_bad_inode(struct inode
*);
2346 * return READ, READA, or WRITE
2348 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2351 * return data direction, READ or WRITE
2353 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2355 extern void check_disk_size_change(struct gendisk
*disk
,
2356 struct block_device
*bdev
);
2357 extern int revalidate_disk(struct gendisk
*);
2358 extern int check_disk_change(struct block_device
*);
2359 extern int __invalidate_device(struct block_device
*, bool);
2360 extern int invalidate_partition(struct gendisk
*, int);
2362 unsigned long invalidate_mapping_pages(struct address_space
*mapping
,
2363 pgoff_t start
, pgoff_t end
);
2365 static inline void invalidate_remote_inode(struct inode
*inode
)
2367 if (S_ISREG(inode
->i_mode
) || S_ISDIR(inode
->i_mode
) ||
2368 S_ISLNK(inode
->i_mode
))
2369 invalidate_mapping_pages(inode
->i_mapping
, 0, -1);
2371 extern int invalidate_inode_pages2(struct address_space
*mapping
);
2372 extern int invalidate_inode_pages2_range(struct address_space
*mapping
,
2373 pgoff_t start
, pgoff_t end
);
2374 extern int write_inode_now(struct inode
*, int);
2375 extern int filemap_fdatawrite(struct address_space
*);
2376 extern int filemap_flush(struct address_space
*);
2377 extern int filemap_fdatawait(struct address_space
*);
2378 extern int filemap_fdatawait_range(struct address_space
*, loff_t lstart
,
2380 extern int filemap_write_and_wait(struct address_space
*mapping
);
2381 extern int filemap_write_and_wait_range(struct address_space
*mapping
,
2382 loff_t lstart
, loff_t lend
);
2383 extern int __filemap_fdatawrite_range(struct address_space
*mapping
,
2384 loff_t start
, loff_t end
, int sync_mode
);
2385 extern int filemap_fdatawrite_range(struct address_space
*mapping
,
2386 loff_t start
, loff_t end
);
2388 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
2390 extern int vfs_fsync(struct file
*file
, int datasync
);
2391 static inline int generic_write_sync(struct file
*file
, loff_t pos
, loff_t count
)
2393 if (!(file
->f_flags
& O_DSYNC
) && !IS_SYNC(file
->f_mapping
->host
))
2395 return vfs_fsync_range(file
, pos
, pos
+ count
- 1,
2396 (file
->f_flags
& __O_SYNC
) ? 0 : 1);
2398 extern void emergency_sync(void);
2399 extern void emergency_remount(void);
2401 extern sector_t
bmap(struct inode
*, sector_t
);
2403 extern int notify_change(struct dentry
*, struct iattr
*, struct inode
**);
2404 extern int inode_permission(struct inode
*, int);
2405 extern int __inode_permission(struct inode
*, int);
2406 extern int generic_permission(struct inode
*, int);
2407 extern int __check_sticky(struct inode
*dir
, struct inode
*inode
);
2409 static inline bool execute_ok(struct inode
*inode
)
2411 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
2414 static inline void file_start_write(struct file
*file
)
2416 if (!S_ISREG(file_inode(file
)->i_mode
))
2418 __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, true);
2421 static inline bool file_start_write_trylock(struct file
*file
)
2423 if (!S_ISREG(file_inode(file
)->i_mode
))
2425 return __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, false);
2428 static inline void file_end_write(struct file
*file
)
2430 if (!S_ISREG(file_inode(file
)->i_mode
))
2432 __sb_end_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
);
2436 * get_write_access() gets write permission for a file.
2437 * put_write_access() releases this write permission.
2438 * This is used for regular files.
2439 * We cannot support write (and maybe mmap read-write shared) accesses and
2440 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2441 * can have the following values:
2442 * 0: no writers, no VM_DENYWRITE mappings
2443 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2444 * > 0: (i_writecount) users are writing to the file.
2446 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2447 * except for the cases where we don't hold i_writecount yet. Then we need to
2448 * use {get,deny}_write_access() - these functions check the sign and refuse
2449 * to do the change if sign is wrong.
2451 static inline int get_write_access(struct inode
*inode
)
2453 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2455 static inline int deny_write_access(struct file
*file
)
2457 struct inode
*inode
= file_inode(file
);
2458 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2460 static inline void put_write_access(struct inode
* inode
)
2462 atomic_dec(&inode
->i_writecount
);
2464 static inline void allow_write_access(struct file
*file
)
2467 atomic_inc(&file_inode(file
)->i_writecount
);
2469 static inline bool inode_is_open_for_write(const struct inode
*inode
)
2471 return atomic_read(&inode
->i_writecount
) > 0;
2475 static inline void i_readcount_dec(struct inode
*inode
)
2477 BUG_ON(!atomic_read(&inode
->i_readcount
));
2478 atomic_dec(&inode
->i_readcount
);
2480 static inline void i_readcount_inc(struct inode
*inode
)
2482 atomic_inc(&inode
->i_readcount
);
2485 static inline void i_readcount_dec(struct inode
*inode
)
2489 static inline void i_readcount_inc(struct inode
*inode
)
2494 extern int do_pipe_flags(int *, int);
2496 extern int kernel_read(struct file
*, loff_t
, char *, unsigned long);
2497 extern ssize_t
kernel_write(struct file
*, const char *, size_t, loff_t
);
2498 extern ssize_t
__kernel_write(struct file
*, const char *, size_t, loff_t
*);
2499 extern struct file
* open_exec(const char *);
2501 /* fs/dcache.c -- generic fs support functions */
2502 extern int is_subdir(struct dentry
*, struct dentry
*);
2503 extern int path_is_under(struct path
*, struct path
*);
2505 #include <linux/err.h>
2507 /* needed for stackable file system support */
2508 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int whence
);
2510 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int whence
);
2512 extern int inode_init_always(struct super_block
*, struct inode
*);
2513 extern void inode_init_once(struct inode
*);
2514 extern void address_space_init_once(struct address_space
*mapping
);
2515 extern struct inode
* igrab(struct inode
*);
2516 extern ino_t
iunique(struct super_block
*, ino_t
);
2517 extern int inode_needs_sync(struct inode
*inode
);
2518 extern int generic_delete_inode(struct inode
*inode
);
2519 static inline int generic_drop_inode(struct inode
*inode
)
2521 return !inode
->i_nlink
|| inode_unhashed(inode
);
2524 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
2525 unsigned long hashval
, int (*test
)(struct inode
*, void *),
2527 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
2528 int (*test
)(struct inode
*, void *), void *data
);
2529 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
2531 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
2532 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
2533 extern struct inode
*find_inode_nowait(struct super_block
*,
2535 int (*match
)(struct inode
*,
2536 unsigned long, void *),
2538 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
2539 extern int insert_inode_locked(struct inode
*);
2540 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2541 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
2543 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
2545 extern void unlock_new_inode(struct inode
*);
2546 extern unsigned int get_next_ino(void);
2548 extern void __iget(struct inode
* inode
);
2549 extern void iget_failed(struct inode
*);
2550 extern void clear_inode(struct inode
*);
2551 extern void __destroy_inode(struct inode
*);
2552 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
2553 extern struct inode
*new_inode(struct super_block
*sb
);
2554 extern void free_inode_nonrcu(struct inode
*inode
);
2555 extern int should_remove_suid(struct dentry
*);
2556 extern int file_remove_suid(struct file
*);
2558 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
2559 static inline void insert_inode_hash(struct inode
*inode
)
2561 __insert_inode_hash(inode
, inode
->i_ino
);
2564 extern void __remove_inode_hash(struct inode
*);
2565 static inline void remove_inode_hash(struct inode
*inode
)
2567 if (!inode_unhashed(inode
))
2568 __remove_inode_hash(inode
);
2571 extern void inode_sb_list_add(struct inode
*inode
);
2574 extern void submit_bio(int, struct bio
*);
2575 extern int bdev_read_only(struct block_device
*);
2577 extern int set_blocksize(struct block_device
*, int);
2578 extern int sb_set_blocksize(struct super_block
*, int);
2579 extern int sb_min_blocksize(struct super_block
*, int);
2581 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
2582 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
2583 extern ssize_t
generic_write_checks(struct kiocb
*, struct iov_iter
*);
2584 extern ssize_t
generic_file_read_iter(struct kiocb
*, struct iov_iter
*);
2585 extern ssize_t
__generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2586 extern ssize_t
generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2587 extern ssize_t
generic_file_direct_write(struct kiocb
*, struct iov_iter
*, loff_t
);
2588 extern ssize_t
generic_perform_write(struct file
*, struct iov_iter
*, loff_t
);
2590 ssize_t
vfs_iter_read(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
);
2591 ssize_t
vfs_iter_write(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
);
2593 /* fs/block_dev.c */
2594 extern ssize_t
blkdev_read_iter(struct kiocb
*iocb
, struct iov_iter
*to
);
2595 extern ssize_t
blkdev_write_iter(struct kiocb
*iocb
, struct iov_iter
*from
);
2596 extern int blkdev_fsync(struct file
*filp
, loff_t start
, loff_t end
,
2598 extern void block_sync_page(struct page
*page
);
2601 extern ssize_t
generic_file_splice_read(struct file
*, loff_t
*,
2602 struct pipe_inode_info
*, size_t, unsigned int);
2603 extern ssize_t
default_file_splice_read(struct file
*, loff_t
*,
2604 struct pipe_inode_info
*, size_t, unsigned int);
2605 extern ssize_t
iter_file_splice_write(struct pipe_inode_info
*,
2606 struct file
*, loff_t
*, size_t, unsigned int);
2607 extern ssize_t
generic_splice_sendpage(struct pipe_inode_info
*pipe
,
2608 struct file
*out
, loff_t
*, size_t len
, unsigned int flags
);
2609 extern long do_splice_direct(struct file
*in
, loff_t
*ppos
, struct file
*out
,
2610 loff_t
*opos
, size_t len
, unsigned int flags
);
2614 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
2615 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int whence
);
2616 extern loff_t
no_llseek(struct file
*file
, loff_t offset
, int whence
);
2617 extern loff_t
vfs_setpos(struct file
*file
, loff_t offset
, loff_t maxsize
);
2618 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int whence
);
2619 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
2620 int whence
, loff_t maxsize
, loff_t eof
);
2621 extern loff_t
fixed_size_llseek(struct file
*file
, loff_t offset
,
2622 int whence
, loff_t size
);
2623 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
2624 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
2626 ssize_t
dax_do_io(struct kiocb
*, struct inode
*, struct iov_iter
*, loff_t
,
2627 get_block_t
, dio_iodone_t
, int flags
);
2628 int dax_clear_blocks(struct inode
*, sector_t block
, long size
);
2629 int dax_zero_page_range(struct inode
*, loff_t from
, unsigned len
, get_block_t
);
2630 int dax_truncate_page(struct inode
*, loff_t from
, get_block_t
);
2631 int dax_fault(struct vm_area_struct
*, struct vm_fault
*, get_block_t
);
2632 int dax_pfn_mkwrite(struct vm_area_struct
*, struct vm_fault
*);
2633 #define dax_mkwrite(vma, vmf, gb) dax_fault(vma, vmf, gb)
2636 typedef void (dio_submit_t
)(int rw
, struct bio
*bio
, struct inode
*inode
,
2637 loff_t file_offset
);
2640 /* need locking between buffered and direct access */
2643 /* filesystem does not support filling holes */
2644 DIO_SKIP_HOLES
= 0x02,
2646 /* filesystem can handle aio writes beyond i_size */
2647 DIO_ASYNC_EXTEND
= 0x04,
2649 /* inode/fs/bdev does not need truncate protection */
2650 DIO_SKIP_DIO_COUNT
= 0x08,
2653 void dio_end_io(struct bio
*bio
, int error
);
2655 ssize_t
__blockdev_direct_IO(struct kiocb
*iocb
, struct inode
*inode
,
2656 struct block_device
*bdev
, struct iov_iter
*iter
,
2657 loff_t offset
, get_block_t get_block
,
2658 dio_iodone_t end_io
, dio_submit_t submit_io
,
2661 static inline ssize_t
blockdev_direct_IO(struct kiocb
*iocb
,
2662 struct inode
*inode
,
2663 struct iov_iter
*iter
, loff_t offset
,
2664 get_block_t get_block
)
2666 return __blockdev_direct_IO(iocb
, inode
, inode
->i_sb
->s_bdev
, iter
,
2667 offset
, get_block
, NULL
, NULL
,
2668 DIO_LOCKING
| DIO_SKIP_HOLES
);
2672 void inode_dio_wait(struct inode
*inode
);
2675 * inode_dio_begin - signal start of a direct I/O requests
2676 * @inode: inode the direct I/O happens on
2678 * This is called once we've finished processing a direct I/O request,
2679 * and is used to wake up callers waiting for direct I/O to be quiesced.
2681 static inline void inode_dio_begin(struct inode
*inode
)
2683 atomic_inc(&inode
->i_dio_count
);
2687 * inode_dio_end - signal finish of a direct I/O requests
2688 * @inode: inode the direct I/O happens on
2690 * This is called once we've finished processing a direct I/O request,
2691 * and is used to wake up callers waiting for direct I/O to be quiesced.
2693 static inline void inode_dio_end(struct inode
*inode
)
2695 if (atomic_dec_and_test(&inode
->i_dio_count
))
2696 wake_up_bit(&inode
->i_state
, __I_DIO_WAKEUP
);
2699 extern void inode_set_flags(struct inode
*inode
, unsigned int flags
,
2702 extern const struct file_operations generic_ro_fops
;
2704 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2706 extern int readlink_copy(char __user
*, int, const char *);
2707 extern int page_readlink(struct dentry
*, char __user
*, int);
2708 extern const char *page_follow_link_light(struct dentry
*, void **, struct nameidata
*);
2709 extern void page_put_link(struct dentry
*, void *);
2710 extern int __page_symlink(struct inode
*inode
, const char *symname
, int len
,
2712 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
2713 extern const struct inode_operations page_symlink_inode_operations
;
2714 extern void kfree_put_link(struct dentry
*, void *);
2715 extern int generic_readlink(struct dentry
*, char __user
*, int);
2716 extern void generic_fillattr(struct inode
*, struct kstat
*);
2717 int vfs_getattr_nosec(struct path
*path
, struct kstat
*stat
);
2718 extern int vfs_getattr(struct path
*, struct kstat
*);
2719 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2720 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2721 void __inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2722 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2723 loff_t
inode_get_bytes(struct inode
*inode
);
2724 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
2725 const char *simple_follow_link(struct dentry
*, void **, struct nameidata
*);
2726 extern const struct inode_operations simple_symlink_inode_operations
;
2728 extern int iterate_dir(struct file
*, struct dir_context
*);
2730 extern int vfs_stat(const char __user
*, struct kstat
*);
2731 extern int vfs_lstat(const char __user
*, struct kstat
*);
2732 extern int vfs_fstat(unsigned int, struct kstat
*);
2733 extern int vfs_fstatat(int , const char __user
*, struct kstat
*, int);
2735 extern int do_vfs_ioctl(struct file
*filp
, unsigned int fd
, unsigned int cmd
,
2737 extern int __generic_block_fiemap(struct inode
*inode
,
2738 struct fiemap_extent_info
*fieinfo
,
2739 loff_t start
, loff_t len
,
2740 get_block_t
*get_block
);
2741 extern int generic_block_fiemap(struct inode
*inode
,
2742 struct fiemap_extent_info
*fieinfo
, u64 start
,
2743 u64 len
, get_block_t
*get_block
);
2745 extern void get_filesystem(struct file_system_type
*fs
);
2746 extern void put_filesystem(struct file_system_type
*fs
);
2747 extern struct file_system_type
*get_fs_type(const char *name
);
2748 extern struct super_block
*get_super(struct block_device
*);
2749 extern struct super_block
*get_super_thawed(struct block_device
*);
2750 extern struct super_block
*get_active_super(struct block_device
*bdev
);
2751 extern void drop_super(struct super_block
*sb
);
2752 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
2753 extern void iterate_supers_type(struct file_system_type
*,
2754 void (*)(struct super_block
*, void *), void *);
2756 extern int dcache_dir_open(struct inode
*, struct file
*);
2757 extern int dcache_dir_close(struct inode
*, struct file
*);
2758 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
2759 extern int dcache_readdir(struct file
*, struct dir_context
*);
2760 extern int simple_setattr(struct dentry
*, struct iattr
*);
2761 extern int simple_getattr(struct vfsmount
*, struct dentry
*, struct kstat
*);
2762 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
2763 extern int simple_open(struct inode
*inode
, struct file
*file
);
2764 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
2765 extern int simple_unlink(struct inode
*, struct dentry
*);
2766 extern int simple_rmdir(struct inode
*, struct dentry
*);
2767 extern int simple_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*);
2768 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
2769 extern int simple_empty(struct dentry
*);
2770 extern int simple_readpage(struct file
*file
, struct page
*page
);
2771 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
2772 loff_t pos
, unsigned len
, unsigned flags
,
2773 struct page
**pagep
, void **fsdata
);
2774 extern int simple_write_end(struct file
*file
, struct address_space
*mapping
,
2775 loff_t pos
, unsigned len
, unsigned copied
,
2776 struct page
*page
, void *fsdata
);
2777 extern int always_delete_dentry(const struct dentry
*);
2778 extern struct inode
*alloc_anon_inode(struct super_block
*);
2779 extern int simple_nosetlease(struct file
*, long, struct file_lock
**, void **);
2780 extern const struct dentry_operations simple_dentry_operations
;
2782 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
2783 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
2784 extern const struct file_operations simple_dir_operations
;
2785 extern const struct inode_operations simple_dir_inode_operations
;
2786 struct tree_descr
{ char *name
; const struct file_operations
*ops
; int mode
; };
2787 struct dentry
*d_alloc_name(struct dentry
*, const char *);
2788 extern int simple_fill_super(struct super_block
*, unsigned long, struct tree_descr
*);
2789 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
2790 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
2792 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
2793 loff_t
*ppos
, const void *from
, size_t available
);
2794 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
2795 const void __user
*from
, size_t count
);
2797 extern int __generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
2798 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
2800 extern int generic_check_addressable(unsigned, u64
);
2802 #ifdef CONFIG_MIGRATION
2803 extern int buffer_migrate_page(struct address_space
*,
2804 struct page
*, struct page
*,
2807 #define buffer_migrate_page NULL
2810 extern int inode_change_ok(const struct inode
*, struct iattr
*);
2811 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
2812 extern void setattr_copy(struct inode
*inode
, const struct iattr
*attr
);
2814 extern int file_update_time(struct file
*file
);
2816 extern int generic_show_options(struct seq_file
*m
, struct dentry
*root
);
2817 extern void save_mount_options(struct super_block
*sb
, char *options
);
2818 extern void replace_mount_options(struct super_block
*sb
, char *options
);
2820 static inline bool io_is_direct(struct file
*filp
)
2822 return (filp
->f_flags
& O_DIRECT
) || IS_DAX(file_inode(filp
));
2825 static inline int iocb_flags(struct file
*file
)
2828 if (file
->f_flags
& O_APPEND
)
2830 if (io_is_direct(file
))
2835 static inline ino_t
parent_ino(struct dentry
*dentry
)
2840 * Don't strictly need d_lock here? If the parent ino could change
2841 * then surely we'd have a deeper race in the caller?
2843 spin_lock(&dentry
->d_lock
);
2844 res
= dentry
->d_parent
->d_inode
->i_ino
;
2845 spin_unlock(&dentry
->d_lock
);
2849 /* Transaction based IO helpers */
2852 * An argresp is stored in an allocated page and holds the
2853 * size of the argument or response, along with its content
2855 struct simple_transaction_argresp
{
2860 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2862 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
2864 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
2865 size_t size
, loff_t
*pos
);
2866 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
2868 void simple_transaction_set(struct file
*file
, size_t n
);
2871 * simple attribute files
2873 * These attributes behave similar to those in sysfs:
2875 * Writing to an attribute immediately sets a value, an open file can be
2876 * written to multiple times.
2878 * Reading from an attribute creates a buffer from the value that might get
2879 * read with multiple read calls. When the attribute has been read
2880 * completely, no further read calls are possible until the file is opened
2883 * All attributes contain a text representation of a numeric value
2884 * that are accessed with the get() and set() functions.
2886 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2887 static int __fops ## _open(struct inode *inode, struct file *file) \
2889 __simple_attr_check_format(__fmt, 0ull); \
2890 return simple_attr_open(inode, file, __get, __set, __fmt); \
2892 static const struct file_operations __fops = { \
2893 .owner = THIS_MODULE, \
2894 .open = __fops ## _open, \
2895 .release = simple_attr_release, \
2896 .read = simple_attr_read, \
2897 .write = simple_attr_write, \
2898 .llseek = generic_file_llseek, \
2901 static inline __printf(1, 2)
2902 void __simple_attr_check_format(const char *fmt
, ...)
2904 /* don't do anything, just let the compiler check the arguments; */
2907 int simple_attr_open(struct inode
*inode
, struct file
*file
,
2908 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
2910 int simple_attr_release(struct inode
*inode
, struct file
*file
);
2911 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
2912 size_t len
, loff_t
*ppos
);
2913 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
2914 size_t len
, loff_t
*ppos
);
2917 int proc_nr_files(struct ctl_table
*table
, int write
,
2918 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
2919 int proc_nr_dentry(struct ctl_table
*table
, int write
,
2920 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
2921 int proc_nr_inodes(struct ctl_table
*table
, int write
,
2922 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
2923 int __init
get_filesystem_list(char *buf
);
2925 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2926 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2928 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2929 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2930 (flag & __FMODE_NONOTIFY)))
2932 static inline int is_sxid(umode_t mode
)
2934 return (mode
& S_ISUID
) || ((mode
& S_ISGID
) && (mode
& S_IXGRP
));
2937 static inline int check_sticky(struct inode
*dir
, struct inode
*inode
)
2939 if (!(dir
->i_mode
& S_ISVTX
))
2942 return __check_sticky(dir
, inode
);
2945 static inline void inode_has_no_xattr(struct inode
*inode
)
2947 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& MS_NOSEC
))
2948 inode
->i_flags
|= S_NOSEC
;
2951 static inline bool is_root_inode(struct inode
*inode
)
2953 return inode
== inode
->i_sb
->s_root
->d_inode
;
2956 static inline bool dir_emit(struct dir_context
*ctx
,
2957 const char *name
, int namelen
,
2958 u64 ino
, unsigned type
)
2960 return ctx
->actor(ctx
, name
, namelen
, ctx
->pos
, ino
, type
) == 0;
2962 static inline bool dir_emit_dot(struct file
*file
, struct dir_context
*ctx
)
2964 return ctx
->actor(ctx
, ".", 1, ctx
->pos
,
2965 file
->f_path
.dentry
->d_inode
->i_ino
, DT_DIR
) == 0;
2967 static inline bool dir_emit_dotdot(struct file
*file
, struct dir_context
*ctx
)
2969 return ctx
->actor(ctx
, "..", 2, ctx
->pos
,
2970 parent_ino(file
->f_path
.dentry
), DT_DIR
) == 0;
2972 static inline bool dir_emit_dots(struct file
*file
, struct dir_context
*ctx
)
2974 if (ctx
->pos
== 0) {
2975 if (!dir_emit_dot(file
, ctx
))
2979 if (ctx
->pos
== 1) {
2980 if (!dir_emit_dotdot(file
, ctx
))
2986 static inline bool dir_relax(struct inode
*inode
)
2988 mutex_unlock(&inode
->i_mutex
);
2989 mutex_lock(&inode
->i_mutex
);
2990 return !IS_DEADDIR(inode
);
2993 #endif /* _LINUX_FS_H */