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
;
663 #ifdef CONFIG_FSNOTIFY
664 __u32 i_fsnotify_mask
; /* all events this inode cares about */
665 struct hlist_head i_fsnotify_marks
;
668 void *i_private
; /* fs or device private pointer */
671 static inline int inode_unhashed(struct inode
*inode
)
673 return hlist_unhashed(&inode
->i_hash
);
677 * inode->i_mutex nesting subclasses for the lock validator:
679 * 0: the object of the current VFS operation
683 * 4: second non-directory
684 * 5: second parent (when locking independent directories in rename)
686 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
687 * non-directories at once.
689 * The locking order between these classes is
690 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
692 enum inode_i_mutex_lock_class
702 void lock_two_nondirectories(struct inode
*, struct inode
*);
703 void unlock_two_nondirectories(struct inode
*, struct inode
*);
706 * NOTE: in a 32bit arch with a preemptable kernel and
707 * an UP compile the i_size_read/write must be atomic
708 * with respect to the local cpu (unlike with preempt disabled),
709 * but they don't need to be atomic with respect to other cpus like in
710 * true SMP (so they need either to either locally disable irq around
711 * the read or for example on x86 they can be still implemented as a
712 * cmpxchg8b without the need of the lock prefix). For SMP compiles
713 * and 64bit archs it makes no difference if preempt is enabled or not.
715 static inline loff_t
i_size_read(const struct inode
*inode
)
717 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
722 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
723 i_size
= inode
->i_size
;
724 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
726 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
730 i_size
= inode
->i_size
;
734 return inode
->i_size
;
739 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
740 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
741 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
743 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
745 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
747 write_seqcount_begin(&inode
->i_size_seqcount
);
748 inode
->i_size
= i_size
;
749 write_seqcount_end(&inode
->i_size_seqcount
);
751 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
753 inode
->i_size
= i_size
;
756 inode
->i_size
= i_size
;
760 /* Helper functions so that in most cases filesystems will
761 * not need to deal directly with kuid_t and kgid_t and can
762 * instead deal with the raw numeric values that are stored
765 static inline uid_t
i_uid_read(const struct inode
*inode
)
767 return from_kuid(&init_user_ns
, inode
->i_uid
);
770 static inline gid_t
i_gid_read(const struct inode
*inode
)
772 return from_kgid(&init_user_ns
, inode
->i_gid
);
775 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
777 inode
->i_uid
= make_kuid(&init_user_ns
, uid
);
780 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
782 inode
->i_gid
= make_kgid(&init_user_ns
, gid
);
785 static inline unsigned iminor(const struct inode
*inode
)
787 return MINOR(inode
->i_rdev
);
790 static inline unsigned imajor(const struct inode
*inode
)
792 return MAJOR(inode
->i_rdev
);
795 extern struct block_device
*I_BDEV(struct inode
*inode
);
798 rwlock_t lock
; /* protects pid, uid, euid fields */
799 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
800 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
801 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
802 int signum
; /* posix.1b rt signal to be delivered on IO */
806 * Track a single file's readahead state
808 struct file_ra_state
{
809 pgoff_t start
; /* where readahead started */
810 unsigned int size
; /* # of readahead pages */
811 unsigned int async_size
; /* do asynchronous readahead when
812 there are only # of pages ahead */
814 unsigned int ra_pages
; /* Maximum readahead window */
815 unsigned int mmap_miss
; /* Cache miss stat for mmap accesses */
816 loff_t prev_pos
; /* Cache last read() position */
820 * Check if @index falls in the readahead windows.
822 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
824 return (index
>= ra
->start
&&
825 index
< ra
->start
+ ra
->size
);
830 struct llist_node fu_llist
;
831 struct rcu_head fu_rcuhead
;
834 struct inode
*f_inode
; /* cached value */
835 const struct file_operations
*f_op
;
838 * Protects f_ep_links, f_flags.
839 * Must not be taken from IRQ context.
842 atomic_long_t f_count
;
843 unsigned int f_flags
;
845 struct mutex f_pos_lock
;
847 struct fown_struct f_owner
;
848 const struct cred
*f_cred
;
849 struct file_ra_state f_ra
;
852 #ifdef CONFIG_SECURITY
855 /* needed for tty driver, and maybe others */
859 /* Used by fs/eventpoll.c to link all the hooks to this file */
860 struct list_head f_ep_links
;
861 struct list_head f_tfile_llink
;
862 #endif /* #ifdef CONFIG_EPOLL */
863 struct address_space
*f_mapping
;
864 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
869 /* file identifier */
870 unsigned char f_handle
[0];
873 static inline struct file
*get_file(struct file
*f
)
875 atomic_long_inc(&f
->f_count
);
878 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
879 #define file_count(x) atomic_long_read(&(x)->f_count)
881 #define MAX_NON_LFS ((1UL<<31) - 1)
883 /* Page cache limit. The filesystems should put that into their s_maxbytes
884 limits, otherwise bad things can happen in VM. */
885 #if BITS_PER_LONG==32
886 #define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
887 #elif BITS_PER_LONG==64
888 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
893 #define FL_DELEG 4 /* NFSv4 delegation */
894 #define FL_ACCESS 8 /* not trying to lock, just looking */
895 #define FL_EXISTS 16 /* when unlocking, test for existence */
896 #define FL_LEASE 32 /* lease held on this file */
897 #define FL_CLOSE 64 /* unlock on close */
898 #define FL_SLEEP 128 /* A blocking lock */
899 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
900 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
901 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
902 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
905 * Special return value from posix_lock_file() and vfs_lock_file() for
906 * asynchronous locking.
908 #define FILE_LOCK_DEFERRED 1
910 /* legacy typedef, should eventually be removed */
911 typedef void *fl_owner_t
;
915 struct file_lock_operations
{
916 void (*fl_copy_lock
)(struct file_lock
*, struct file_lock
*);
917 void (*fl_release_private
)(struct file_lock
*);
920 struct lock_manager_operations
{
921 int (*lm_compare_owner
)(struct file_lock
*, struct file_lock
*);
922 unsigned long (*lm_owner_key
)(struct file_lock
*);
923 void (*lm_get_owner
)(struct file_lock
*, struct file_lock
*);
924 void (*lm_put_owner
)(struct file_lock
*);
925 void (*lm_notify
)(struct file_lock
*); /* unblock callback */
926 int (*lm_grant
)(struct file_lock
*, int);
927 bool (*lm_break
)(struct file_lock
*);
928 int (*lm_change
)(struct file_lock
*, int, struct list_head
*);
929 void (*lm_setup
)(struct file_lock
*, void **);
932 struct lock_manager
{
933 struct list_head list
;
937 void locks_start_grace(struct net
*, struct lock_manager
*);
938 void locks_end_grace(struct lock_manager
*);
939 int locks_in_grace(struct net
*);
941 /* that will die - we need it for nfs_lock_info */
942 #include <linux/nfs_fs_i.h>
945 * struct file_lock represents a generic "file lock". It's used to represent
946 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
947 * note that the same struct is used to represent both a request for a lock and
948 * the lock itself, but the same object is never used for both.
950 * FIXME: should we create a separate "struct lock_request" to help distinguish
953 * The varous i_flctx lists are ordered by:
956 * 2) lock range start
959 * Obviously, the last two criteria only matter for POSIX locks.
962 struct file_lock
*fl_next
; /* singly linked list for this inode */
963 struct list_head fl_list
; /* link into file_lock_context */
964 struct hlist_node fl_link
; /* node in global lists */
965 struct list_head fl_block
; /* circular list of blocked processes */
967 unsigned int fl_flags
;
968 unsigned char fl_type
;
970 int fl_link_cpu
; /* what cpu's list is this on? */
971 struct pid
*fl_nspid
;
972 wait_queue_head_t fl_wait
;
973 struct file
*fl_file
;
977 struct fasync_struct
* fl_fasync
; /* for lease break notifications */
978 /* for lease breaks: */
979 unsigned long fl_break_time
;
980 unsigned long fl_downgrade_time
;
982 const struct file_lock_operations
*fl_ops
; /* Callbacks for filesystems */
983 const struct lock_manager_operations
*fl_lmops
; /* Callbacks for lockmanagers */
985 struct nfs_lock_info nfs_fl
;
986 struct nfs4_lock_info nfs4_fl
;
988 struct list_head link
; /* link in AFS vnode's pending_locks list */
989 int state
; /* state of grant or error if -ve */
994 struct file_lock_context
{
996 struct list_head flc_flock
;
997 struct list_head flc_posix
;
998 struct list_head flc_lease
;
1001 /* The following constant reflects the upper bound of the file/locking space */
1003 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1004 #define OFFSET_MAX INT_LIMIT(loff_t)
1005 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1008 #include <linux/fcntl.h>
1010 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
1012 #ifdef CONFIG_FILE_LOCKING
1013 extern int fcntl_getlk(struct file
*, unsigned int, struct flock __user
*);
1014 extern int fcntl_setlk(unsigned int, struct file
*, unsigned int,
1015 struct flock __user
*);
1017 #if BITS_PER_LONG == 32
1018 extern int fcntl_getlk64(struct file
*, unsigned int, struct flock64 __user
*);
1019 extern int fcntl_setlk64(unsigned int, struct file
*, unsigned int,
1020 struct flock64 __user
*);
1023 extern int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
);
1024 extern int fcntl_getlease(struct file
*filp
);
1027 void locks_free_lock_context(struct file_lock_context
*ctx
);
1028 void locks_free_lock(struct file_lock
*fl
);
1029 extern void locks_init_lock(struct file_lock
*);
1030 extern struct file_lock
* locks_alloc_lock(void);
1031 extern void locks_copy_lock(struct file_lock
*, struct file_lock
*);
1032 extern void locks_copy_conflock(struct file_lock
*, struct file_lock
*);
1033 extern void locks_remove_posix(struct file
*, fl_owner_t
);
1034 extern void locks_remove_file(struct file
*);
1035 extern void locks_release_private(struct file_lock
*);
1036 extern void posix_test_lock(struct file
*, struct file_lock
*);
1037 extern int posix_lock_file(struct file
*, struct file_lock
*, struct file_lock
*);
1038 extern int posix_lock_file_wait(struct file
*, struct file_lock
*);
1039 extern int posix_unblock_lock(struct file_lock
*);
1040 extern int vfs_test_lock(struct file
*, struct file_lock
*);
1041 extern int vfs_lock_file(struct file
*, unsigned int, struct file_lock
*, struct file_lock
*);
1042 extern int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
);
1043 extern int flock_lock_file_wait(struct file
*filp
, struct file_lock
*fl
);
1044 extern int __break_lease(struct inode
*inode
, unsigned int flags
, unsigned int type
);
1045 extern void lease_get_mtime(struct inode
*, struct timespec
*time
);
1046 extern int generic_setlease(struct file
*, long, struct file_lock
**, void **priv
);
1047 extern int vfs_setlease(struct file
*, long, struct file_lock
**, void **);
1048 extern int lease_modify(struct file_lock
*, int, struct list_head
*);
1049 #else /* !CONFIG_FILE_LOCKING */
1050 static inline int fcntl_getlk(struct file
*file
, unsigned int cmd
,
1051 struct flock __user
*user
)
1056 static inline int fcntl_setlk(unsigned int fd
, struct file
*file
,
1057 unsigned int cmd
, struct flock __user
*user
)
1062 #if BITS_PER_LONG == 32
1063 static inline int fcntl_getlk64(struct file
*file
, unsigned int cmd
,
1064 struct flock64 __user
*user
)
1069 static inline int fcntl_setlk64(unsigned int fd
, struct file
*file
,
1070 unsigned int cmd
, struct flock64 __user
*user
)
1075 static inline int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
)
1080 static inline int fcntl_getlease(struct file
*filp
)
1086 locks_free_lock_context(struct file_lock_context
*ctx
)
1090 static inline void locks_init_lock(struct file_lock
*fl
)
1095 static inline void locks_copy_conflock(struct file_lock
*new, struct file_lock
*fl
)
1100 static inline void locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
1105 static inline void locks_remove_posix(struct file
*filp
, fl_owner_t owner
)
1110 static inline void locks_remove_file(struct file
*filp
)
1115 static inline void posix_test_lock(struct file
*filp
, struct file_lock
*fl
)
1120 static inline int posix_lock_file(struct file
*filp
, struct file_lock
*fl
,
1121 struct file_lock
*conflock
)
1126 static inline int posix_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1131 static inline int posix_unblock_lock(struct file_lock
*waiter
)
1136 static inline int vfs_test_lock(struct file
*filp
, struct file_lock
*fl
)
1141 static inline int vfs_lock_file(struct file
*filp
, unsigned int cmd
,
1142 struct file_lock
*fl
, struct file_lock
*conf
)
1147 static inline int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
)
1152 static inline int flock_lock_file_wait(struct file
*filp
,
1153 struct file_lock
*request
)
1158 static inline int __break_lease(struct inode
*inode
, unsigned int mode
, unsigned int type
)
1163 static inline void lease_get_mtime(struct inode
*inode
, struct timespec
*time
)
1168 static inline int generic_setlease(struct file
*filp
, long arg
,
1169 struct file_lock
**flp
, void **priv
)
1174 static inline int vfs_setlease(struct file
*filp
, long arg
,
1175 struct file_lock
**lease
, void **priv
)
1180 static inline int lease_modify(struct file_lock
*fl
, int arg
,
1181 struct list_head
*dispose
)
1185 #endif /* !CONFIG_FILE_LOCKING */
1188 struct fasync_struct
{
1192 struct fasync_struct
*fa_next
; /* singly linked list */
1193 struct file
*fa_file
;
1194 struct rcu_head fa_rcu
;
1197 #define FASYNC_MAGIC 0x4601
1199 /* SMP safe fasync helpers: */
1200 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1201 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1202 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1203 extern struct fasync_struct
*fasync_alloc(void);
1204 extern void fasync_free(struct fasync_struct
*);
1206 /* can be called from interrupts */
1207 extern void kill_fasync(struct fasync_struct
**, int, int);
1209 extern void __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1210 extern void f_setown(struct file
*filp
, unsigned long arg
, int force
);
1211 extern void f_delown(struct file
*filp
);
1212 extern pid_t
f_getown(struct file
*filp
);
1213 extern int send_sigurg(struct fown_struct
*fown
);
1221 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1222 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1223 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1224 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1225 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1228 /* Possible states of 'frozen' field */
1230 SB_UNFROZEN
= 0, /* FS is unfrozen */
1231 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1232 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1233 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1234 * internal threads if needed) */
1235 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1238 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1241 /* Counters for counting writers at each level */
1242 struct percpu_counter counter
[SB_FREEZE_LEVELS
];
1243 wait_queue_head_t wait
; /* queue for waiting for
1244 writers / faults to finish */
1245 int frozen
; /* Is sb frozen? */
1246 wait_queue_head_t wait_unfrozen
; /* queue for waiting for
1248 #ifdef CONFIG_DEBUG_LOCK_ALLOC
1249 struct lockdep_map lock_map
[SB_FREEZE_LEVELS
];
1253 struct super_block
{
1254 struct list_head s_list
; /* Keep this first */
1255 dev_t s_dev
; /* search index; _not_ kdev_t */
1256 unsigned char s_blocksize_bits
;
1257 unsigned long s_blocksize
;
1258 loff_t s_maxbytes
; /* Max file size */
1259 struct file_system_type
*s_type
;
1260 const struct super_operations
*s_op
;
1261 const struct dquot_operations
*dq_op
;
1262 const struct quotactl_ops
*s_qcop
;
1263 const struct export_operations
*s_export_op
;
1264 unsigned long s_flags
;
1265 unsigned long s_magic
;
1266 struct dentry
*s_root
;
1267 struct rw_semaphore s_umount
;
1270 #ifdef CONFIG_SECURITY
1273 const struct xattr_handler
**s_xattr
;
1275 struct list_head s_inodes
; /* all inodes */
1276 struct hlist_bl_head s_anon
; /* anonymous dentries for (nfs) exporting */
1277 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1278 struct block_device
*s_bdev
;
1279 struct backing_dev_info
*s_bdi
;
1280 struct mtd_info
*s_mtd
;
1281 struct hlist_node s_instances
;
1282 unsigned int s_quota_types
; /* Bitmask of supported quota types */
1283 struct quota_info s_dquot
; /* Diskquota specific options */
1285 struct sb_writers s_writers
;
1287 char s_id
[32]; /* Informational name */
1288 u8 s_uuid
[16]; /* UUID */
1290 void *s_fs_info
; /* Filesystem private info */
1291 unsigned int s_max_links
;
1294 /* Granularity of c/m/atime in ns.
1295 Cannot be worse than a second */
1299 * The next field is for VFS *only*. No filesystems have any business
1300 * even looking at it. You had been warned.
1302 struct mutex s_vfs_rename_mutex
; /* Kludge */
1305 * Filesystem subtype. If non-empty the filesystem type field
1306 * in /proc/mounts will be "type.subtype"
1311 * Saved mount options for lazy filesystems using
1312 * generic_show_options()
1314 char __rcu
*s_options
;
1315 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1318 * Saved pool identifier for cleancache (-1 means none)
1320 int cleancache_poolid
;
1322 struct shrinker s_shrink
; /* per-sb shrinker handle */
1324 /* Number of inodes with nlink == 0 but still referenced */
1325 atomic_long_t s_remove_count
;
1327 /* Being remounted read-only */
1328 int s_readonly_remount
;
1330 /* AIO completions deferred from interrupt context */
1331 struct workqueue_struct
*s_dio_done_wq
;
1332 struct hlist_head s_pins
;
1335 * Keep the lru lists last in the structure so they always sit on their
1336 * own individual cachelines.
1338 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp
;
1339 struct list_lru s_inode_lru ____cacheline_aligned_in_smp
;
1340 struct rcu_head rcu
;
1343 * Indicates how deep in a filesystem stack this SB is
1348 extern struct timespec
current_fs_time(struct super_block
*sb
);
1351 * Snapshotting support.
1354 void __sb_end_write(struct super_block
*sb
, int level
);
1355 int __sb_start_write(struct super_block
*sb
, int level
, bool wait
);
1358 * sb_end_write - drop write access to a superblock
1359 * @sb: the super we wrote to
1361 * Decrement number of writers to the filesystem. Wake up possible waiters
1362 * wanting to freeze the filesystem.
1364 static inline void sb_end_write(struct super_block
*sb
)
1366 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1370 * sb_end_pagefault - drop write access to a superblock from a page fault
1371 * @sb: the super we wrote to
1373 * Decrement number of processes handling write page fault to the filesystem.
1374 * Wake up possible waiters wanting to freeze the filesystem.
1376 static inline void sb_end_pagefault(struct super_block
*sb
)
1378 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1382 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1383 * @sb: the super we wrote to
1385 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1386 * waiters wanting to freeze the filesystem.
1388 static inline void sb_end_intwrite(struct super_block
*sb
)
1390 __sb_end_write(sb
, SB_FREEZE_FS
);
1394 * sb_start_write - get write access to a superblock
1395 * @sb: the super we write to
1397 * When a process wants to write data or metadata to a file system (i.e. dirty
1398 * a page or an inode), it should embed the operation in a sb_start_write() -
1399 * sb_end_write() pair to get exclusion against file system freezing. This
1400 * function increments number of writers preventing freezing. If the file
1401 * system is already frozen, the function waits until the file system is
1404 * Since freeze protection behaves as a lock, users have to preserve
1405 * ordering of freeze protection and other filesystem locks. Generally,
1406 * freeze protection should be the outermost lock. In particular, we have:
1409 * -> i_mutex (write path, truncate, directory ops, ...)
1410 * -> s_umount (freeze_super, thaw_super)
1412 static inline void sb_start_write(struct super_block
*sb
)
1414 __sb_start_write(sb
, SB_FREEZE_WRITE
, true);
1417 static inline int sb_start_write_trylock(struct super_block
*sb
)
1419 return __sb_start_write(sb
, SB_FREEZE_WRITE
, false);
1423 * sb_start_pagefault - get write access to a superblock from a page fault
1424 * @sb: the super we write to
1426 * When a process starts handling write page fault, it should embed the
1427 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1428 * exclusion against file system freezing. This is needed since the page fault
1429 * is going to dirty a page. This function increments number of running page
1430 * faults preventing freezing. If the file system is already frozen, the
1431 * function waits until the file system is thawed.
1433 * Since page fault freeze protection behaves as a lock, users have to preserve
1434 * ordering of freeze protection and other filesystem locks. It is advised to
1435 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1436 * handling code implies lock dependency:
1439 * -> sb_start_pagefault
1441 static inline void sb_start_pagefault(struct super_block
*sb
)
1443 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
, true);
1447 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1448 * @sb: the super we write to
1450 * This is the third level of protection against filesystem freezing. It is
1451 * free for use by a filesystem. The only requirement is that it must rank
1452 * below sb_start_pagefault.
1454 * For example filesystem can call sb_start_intwrite() when starting a
1455 * transaction which somewhat eases handling of freezing for internal sources
1456 * of filesystem changes (internal fs threads, discarding preallocation on file
1459 static inline void sb_start_intwrite(struct super_block
*sb
)
1461 __sb_start_write(sb
, SB_FREEZE_FS
, true);
1465 extern bool inode_owner_or_capable(const struct inode
*inode
);
1468 * VFS helper functions..
1470 extern int vfs_create(struct inode
*, struct dentry
*, umode_t
, bool);
1471 extern int vfs_mkdir(struct inode
*, struct dentry
*, umode_t
);
1472 extern int vfs_mknod(struct inode
*, struct dentry
*, umode_t
, dev_t
);
1473 extern int vfs_symlink(struct inode
*, struct dentry
*, const char *);
1474 extern int vfs_link(struct dentry
*, struct inode
*, struct dentry
*, struct inode
**);
1475 extern int vfs_rmdir(struct inode
*, struct dentry
*);
1476 extern int vfs_unlink(struct inode
*, struct dentry
*, struct inode
**);
1477 extern int vfs_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*, struct inode
**, unsigned int);
1478 extern int vfs_whiteout(struct inode
*, struct dentry
*);
1481 * VFS dentry helper functions.
1483 extern void dentry_unhash(struct dentry
*dentry
);
1486 * VFS file helper functions.
1488 extern void inode_init_owner(struct inode
*inode
, const struct inode
*dir
,
1491 * VFS FS_IOC_FIEMAP helper definitions.
1493 struct fiemap_extent_info
{
1494 unsigned int fi_flags
; /* Flags as passed from user */
1495 unsigned int fi_extents_mapped
; /* Number of mapped extents */
1496 unsigned int fi_extents_max
; /* Size of fiemap_extent array */
1497 struct fiemap_extent __user
*fi_extents_start
; /* Start of
1498 fiemap_extent array */
1500 int fiemap_fill_next_extent(struct fiemap_extent_info
*info
, u64 logical
,
1501 u64 phys
, u64 len
, u32 flags
);
1502 int fiemap_check_flags(struct fiemap_extent_info
*fieinfo
, u32 fs_flags
);
1507 * NOTE! These match bits 12..15 of stat.st_mode
1508 * (ie "(i_mode >> 12) & 15").
1510 #define DT_UNKNOWN 0
1521 * This is the "filldir" function type, used by readdir() to let
1522 * the kernel specify what kind of dirent layout it wants to have.
1523 * This allows the kernel to read directories into kernel space or
1524 * to have different dirent layouts depending on the binary type.
1527 typedef int (*filldir_t
)(struct dir_context
*, const char *, int, loff_t
, u64
,
1530 struct dir_context
{
1531 const filldir_t actor
;
1535 struct block_device_operations
;
1537 /* These macros are for out of kernel modules to test that
1538 * the kernel supports the unlocked_ioctl and compat_ioctl
1539 * fields in struct file_operations. */
1540 #define HAVE_COMPAT_IOCTL 1
1541 #define HAVE_UNLOCKED_IOCTL 1
1544 * These flags let !MMU mmap() govern direct device mapping vs immediate
1545 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1547 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1548 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1549 * NOMMU_MAP_READ: Can be mapped for reading
1550 * NOMMU_MAP_WRITE: Can be mapped for writing
1551 * NOMMU_MAP_EXEC: Can be mapped for execution
1553 #define NOMMU_MAP_COPY 0x00000001
1554 #define NOMMU_MAP_DIRECT 0x00000008
1555 #define NOMMU_MAP_READ VM_MAYREAD
1556 #define NOMMU_MAP_WRITE VM_MAYWRITE
1557 #define NOMMU_MAP_EXEC VM_MAYEXEC
1559 #define NOMMU_VMFLAGS \
1560 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1565 struct file_operations
{
1566 struct module
*owner
;
1567 loff_t (*llseek
) (struct file
*, loff_t
, int);
1568 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
1569 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
1570 ssize_t (*read_iter
) (struct kiocb
*, struct iov_iter
*);
1571 ssize_t (*write_iter
) (struct kiocb
*, struct iov_iter
*);
1572 int (*iterate
) (struct file
*, struct dir_context
*);
1573 unsigned int (*poll
) (struct file
*, struct poll_table_struct
*);
1574 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
1575 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
1576 int (*mmap
) (struct file
*, struct vm_area_struct
*);
1577 int (*mremap
)(struct file
*, struct vm_area_struct
*);
1578 int (*open
) (struct inode
*, struct file
*);
1579 int (*flush
) (struct file
*, fl_owner_t id
);
1580 int (*release
) (struct inode
*, struct file
*);
1581 int (*fsync
) (struct file
*, loff_t
, loff_t
, int datasync
);
1582 int (*aio_fsync
) (struct kiocb
*, int datasync
);
1583 int (*fasync
) (int, struct file
*, int);
1584 int (*lock
) (struct file
*, int, struct file_lock
*);
1585 ssize_t (*sendpage
) (struct file
*, struct page
*, int, size_t, loff_t
*, int);
1586 unsigned long (*get_unmapped_area
)(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1587 int (*check_flags
)(int);
1588 int (*flock
) (struct file
*, int, struct file_lock
*);
1589 ssize_t (*splice_write
)(struct pipe_inode_info
*, struct file
*, loff_t
*, size_t, unsigned int);
1590 ssize_t (*splice_read
)(struct file
*, loff_t
*, struct pipe_inode_info
*, size_t, unsigned int);
1591 int (*setlease
)(struct file
*, long, struct file_lock
**, void **);
1592 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
1594 void (*show_fdinfo
)(struct seq_file
*m
, struct file
*f
);
1596 unsigned (*mmap_capabilities
)(struct file
*);
1600 struct inode_operations
{
1601 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*, unsigned int);
1602 void * (*follow_link
) (struct dentry
*, struct nameidata
*);
1603 int (*permission
) (struct inode
*, int);
1604 struct posix_acl
* (*get_acl
)(struct inode
*, int);
1606 int (*readlink
) (struct dentry
*, char __user
*,int);
1607 void (*put_link
) (struct dentry
*, struct nameidata
*, void *);
1609 int (*create
) (struct inode
*,struct dentry
*, umode_t
, bool);
1610 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
1611 int (*unlink
) (struct inode
*,struct dentry
*);
1612 int (*symlink
) (struct inode
*,struct dentry
*,const char *);
1613 int (*mkdir
) (struct inode
*,struct dentry
*,umode_t
);
1614 int (*rmdir
) (struct inode
*,struct dentry
*);
1615 int (*mknod
) (struct inode
*,struct dentry
*,umode_t
,dev_t
);
1616 int (*rename
) (struct inode
*, struct dentry
*,
1617 struct inode
*, struct dentry
*);
1618 int (*rename2
) (struct inode
*, struct dentry
*,
1619 struct inode
*, struct dentry
*, unsigned int);
1620 int (*setattr
) (struct dentry
*, struct iattr
*);
1621 int (*getattr
) (struct vfsmount
*mnt
, struct dentry
*, struct kstat
*);
1622 int (*setxattr
) (struct dentry
*, const char *,const void *,size_t,int);
1623 ssize_t (*getxattr
) (struct dentry
*, const char *, void *, size_t);
1624 ssize_t (*listxattr
) (struct dentry
*, char *, size_t);
1625 int (*removexattr
) (struct dentry
*, const char *);
1626 int (*fiemap
)(struct inode
*, struct fiemap_extent_info
*, u64 start
,
1628 int (*update_time
)(struct inode
*, struct timespec
*, int);
1629 int (*atomic_open
)(struct inode
*, struct dentry
*,
1630 struct file
*, unsigned open_flag
,
1631 umode_t create_mode
, int *opened
);
1632 int (*tmpfile
) (struct inode
*, struct dentry
*, umode_t
);
1633 int (*set_acl
)(struct inode
*, struct posix_acl
*, int);
1635 /* WARNING: probably going away soon, do not use! */
1636 int (*dentry_open
)(struct dentry
*, struct file
*, const struct cred
*);
1637 } ____cacheline_aligned
;
1639 ssize_t
rw_copy_check_uvector(int type
, const struct iovec __user
* uvector
,
1640 unsigned long nr_segs
, unsigned long fast_segs
,
1641 struct iovec
*fast_pointer
,
1642 struct iovec
**ret_pointer
);
1644 extern ssize_t
__vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1645 extern ssize_t
__vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1646 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1647 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1648 extern ssize_t
vfs_readv(struct file
*, const struct iovec __user
*,
1649 unsigned long, loff_t
*);
1650 extern ssize_t
vfs_writev(struct file
*, const struct iovec __user
*,
1651 unsigned long, loff_t
*);
1653 struct super_operations
{
1654 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
1655 void (*destroy_inode
)(struct inode
*);
1657 void (*dirty_inode
) (struct inode
*, int flags
);
1658 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
1659 int (*drop_inode
) (struct inode
*);
1660 void (*evict_inode
) (struct inode
*);
1661 void (*put_super
) (struct super_block
*);
1662 int (*sync_fs
)(struct super_block
*sb
, int wait
);
1663 int (*freeze_super
) (struct super_block
*);
1664 int (*freeze_fs
) (struct super_block
*);
1665 int (*thaw_super
) (struct super_block
*);
1666 int (*unfreeze_fs
) (struct super_block
*);
1667 int (*statfs
) (struct dentry
*, struct kstatfs
*);
1668 int (*remount_fs
) (struct super_block
*, int *, char *);
1669 void (*umount_begin
) (struct super_block
*);
1671 int (*show_options
)(struct seq_file
*, struct dentry
*);
1672 int (*show_devname
)(struct seq_file
*, struct dentry
*);
1673 int (*show_path
)(struct seq_file
*, struct dentry
*);
1674 int (*show_stats
)(struct seq_file
*, struct dentry
*);
1676 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
1677 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
1678 struct dquot
**(*get_dquots
)(struct inode
*);
1680 int (*bdev_try_to_free_page
)(struct super_block
*, struct page
*, gfp_t
);
1681 long (*nr_cached_objects
)(struct super_block
*,
1682 struct shrink_control
*);
1683 long (*free_cached_objects
)(struct super_block
*,
1684 struct shrink_control
*);
1688 * Inode flags - they have no relation to superblock flags now
1690 #define S_SYNC 1 /* Writes are synced at once */
1691 #define S_NOATIME 2 /* Do not update access times */
1692 #define S_APPEND 4 /* Append-only file */
1693 #define S_IMMUTABLE 8 /* Immutable file */
1694 #define S_DEAD 16 /* removed, but still open directory */
1695 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1696 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1697 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1698 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1699 #define S_PRIVATE 512 /* Inode is fs-internal */
1700 #define S_IMA 1024 /* Inode has an associated IMA struct */
1701 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1702 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1703 #ifdef CONFIG_FS_DAX
1704 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1706 #define S_DAX 0 /* Make all the DAX code disappear */
1710 * Note that nosuid etc flags are inode-specific: setting some file-system
1711 * flags just means all the inodes inherit those flags by default. It might be
1712 * possible to override it selectively if you really wanted to with some
1713 * ioctl() that is not currently implemented.
1715 * Exception: MS_RDONLY is always applied to the entire file system.
1717 * Unfortunately, it is possible to change a filesystems flags with it mounted
1718 * with files in use. This means that all of the inodes will not have their
1719 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1720 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1722 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1724 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1725 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1726 ((inode)->i_flags & S_SYNC))
1727 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1728 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1729 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1730 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1731 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1733 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1734 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1735 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1736 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1738 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1739 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1740 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1741 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1742 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1743 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1744 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1745 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1747 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1748 (inode)->i_rdev == WHITEOUT_DEV)
1751 * Inode state bits. Protected by inode->i_lock
1753 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1754 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1756 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1757 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1758 * various stages of removing an inode.
1760 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1762 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1763 * fdatasync(). i_atime is the usual cause.
1764 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1765 * these changes separately from I_DIRTY_SYNC so that we
1766 * don't have to write inode on fdatasync() when only
1767 * mtime has changed in it.
1768 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1769 * I_NEW Serves as both a mutex and completion notification.
1770 * New inodes set I_NEW. If two processes both create
1771 * the same inode, one of them will release its inode and
1772 * wait for I_NEW to be released before returning.
1773 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1774 * also cause waiting on I_NEW, without I_NEW actually
1775 * being set. find_inode() uses this to prevent returning
1776 * nearly-dead inodes.
1777 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1778 * is zero. I_FREEING must be set when I_WILL_FREE is
1780 * I_FREEING Set when inode is about to be freed but still has dirty
1781 * pages or buffers attached or the inode itself is still
1783 * I_CLEAR Added by clear_inode(). In this state the inode is
1784 * clean and can be destroyed. Inode keeps I_FREEING.
1786 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1787 * prohibited for many purposes. iget() must wait for
1788 * the inode to be completely released, then create it
1789 * anew. Other functions will just ignore such inodes,
1790 * if appropriate. I_NEW is used for waiting.
1792 * I_SYNC Writeback of inode is running. The bit is set during
1793 * data writeback, and cleared with a wakeup on the bit
1794 * address once it is done. The bit is also used to pin
1795 * the inode in memory for flusher thread.
1797 * I_REFERENCED Marks the inode as recently references on the LRU list.
1799 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1801 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1803 #define I_DIRTY_SYNC (1 << 0)
1804 #define I_DIRTY_DATASYNC (1 << 1)
1805 #define I_DIRTY_PAGES (1 << 2)
1807 #define I_NEW (1 << __I_NEW)
1808 #define I_WILL_FREE (1 << 4)
1809 #define I_FREEING (1 << 5)
1810 #define I_CLEAR (1 << 6)
1812 #define I_SYNC (1 << __I_SYNC)
1813 #define I_REFERENCED (1 << 8)
1814 #define __I_DIO_WAKEUP 9
1815 #define I_DIO_WAKEUP (1 << I_DIO_WAKEUP)
1816 #define I_LINKABLE (1 << 10)
1817 #define I_DIRTY_TIME (1 << 11)
1818 #define __I_DIRTY_TIME_EXPIRED 12
1819 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1821 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1822 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1824 extern void __mark_inode_dirty(struct inode
*, int);
1825 static inline void mark_inode_dirty(struct inode
*inode
)
1827 __mark_inode_dirty(inode
, I_DIRTY
);
1830 static inline void mark_inode_dirty_sync(struct inode
*inode
)
1832 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
1835 extern void inc_nlink(struct inode
*inode
);
1836 extern void drop_nlink(struct inode
*inode
);
1837 extern void clear_nlink(struct inode
*inode
);
1838 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
1840 static inline void inode_inc_link_count(struct inode
*inode
)
1843 mark_inode_dirty(inode
);
1846 static inline void inode_dec_link_count(struct inode
*inode
)
1849 mark_inode_dirty(inode
);
1853 * inode_inc_iversion - increments i_version
1854 * @inode: inode that need to be updated
1856 * Every time the inode is modified, the i_version field will be incremented.
1857 * The filesystem has to be mounted with i_version flag
1860 static inline void inode_inc_iversion(struct inode
*inode
)
1862 spin_lock(&inode
->i_lock
);
1864 spin_unlock(&inode
->i_lock
);
1867 enum file_time_flags
{
1874 extern void touch_atime(const struct path
*);
1875 static inline void file_accessed(struct file
*file
)
1877 if (!(file
->f_flags
& O_NOATIME
))
1878 touch_atime(&file
->f_path
);
1881 int sync_inode(struct inode
*inode
, struct writeback_control
*wbc
);
1882 int sync_inode_metadata(struct inode
*inode
, int wait
);
1884 struct file_system_type
{
1887 #define FS_REQUIRES_DEV 1
1888 #define FS_BINARY_MOUNTDATA 2
1889 #define FS_HAS_SUBTYPE 4
1890 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
1891 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
1892 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
1893 struct dentry
*(*mount
) (struct file_system_type
*, int,
1894 const char *, void *);
1895 void (*kill_sb
) (struct super_block
*);
1896 struct module
*owner
;
1897 struct file_system_type
* next
;
1898 struct hlist_head fs_supers
;
1900 struct lock_class_key s_lock_key
;
1901 struct lock_class_key s_umount_key
;
1902 struct lock_class_key s_vfs_rename_key
;
1903 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
1905 struct lock_class_key i_lock_key
;
1906 struct lock_class_key i_mutex_key
;
1907 struct lock_class_key i_mutex_dir_key
;
1910 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1912 extern struct dentry
*mount_ns(struct file_system_type
*fs_type
, int flags
,
1913 void *data
, int (*fill_super
)(struct super_block
*, void *, int));
1914 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
1915 int flags
, const char *dev_name
, void *data
,
1916 int (*fill_super
)(struct super_block
*, void *, int));
1917 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
1918 int flags
, void *data
,
1919 int (*fill_super
)(struct super_block
*, void *, int));
1920 extern struct dentry
*mount_nodev(struct file_system_type
*fs_type
,
1921 int flags
, void *data
,
1922 int (*fill_super
)(struct super_block
*, void *, int));
1923 extern struct dentry
*mount_subtree(struct vfsmount
*mnt
, const char *path
);
1924 void generic_shutdown_super(struct super_block
*sb
);
1925 void kill_block_super(struct super_block
*sb
);
1926 void kill_anon_super(struct super_block
*sb
);
1927 void kill_litter_super(struct super_block
*sb
);
1928 void deactivate_super(struct super_block
*sb
);
1929 void deactivate_locked_super(struct super_block
*sb
);
1930 int set_anon_super(struct super_block
*s
, void *data
);
1931 int get_anon_bdev(dev_t
*);
1932 void free_anon_bdev(dev_t
);
1933 struct super_block
*sget(struct file_system_type
*type
,
1934 int (*test
)(struct super_block
*,void *),
1935 int (*set
)(struct super_block
*,void *),
1936 int flags
, void *data
);
1937 extern struct dentry
*mount_pseudo(struct file_system_type
*, char *,
1938 const struct super_operations
*ops
,
1939 const struct dentry_operations
*dops
,
1942 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1943 #define fops_get(fops) \
1944 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1945 #define fops_put(fops) \
1946 do { if (fops) module_put((fops)->owner); } while(0)
1948 * This one is to be used *ONLY* from ->open() instances.
1949 * fops must be non-NULL, pinned down *and* module dependencies
1950 * should be sufficient to pin the caller down as well.
1952 #define replace_fops(f, fops) \
1954 struct file *__file = (f); \
1955 fops_put(__file->f_op); \
1956 BUG_ON(!(__file->f_op = (fops))); \
1959 extern int register_filesystem(struct file_system_type
*);
1960 extern int unregister_filesystem(struct file_system_type
*);
1961 extern struct vfsmount
*kern_mount_data(struct file_system_type
*, void *data
);
1962 #define kern_mount(type) kern_mount_data(type, NULL)
1963 extern void kern_unmount(struct vfsmount
*mnt
);
1964 extern int may_umount_tree(struct vfsmount
*);
1965 extern int may_umount(struct vfsmount
*);
1966 extern long do_mount(const char *, const char __user
*,
1967 const char *, unsigned long, void *);
1968 extern struct vfsmount
*collect_mounts(struct path
*);
1969 extern void drop_collected_mounts(struct vfsmount
*);
1970 extern int iterate_mounts(int (*)(struct vfsmount
*, void *), void *,
1972 extern int vfs_statfs(struct path
*, struct kstatfs
*);
1973 extern int user_statfs(const char __user
*, struct kstatfs
*);
1974 extern int fd_statfs(int, struct kstatfs
*);
1975 extern int vfs_ustat(dev_t
, struct kstatfs
*);
1976 extern int freeze_super(struct super_block
*super
);
1977 extern int thaw_super(struct super_block
*super
);
1978 extern bool our_mnt(struct vfsmount
*mnt
);
1979 extern bool fs_fully_visible(struct file_system_type
*);
1981 extern int current_umask(void);
1983 extern void ihold(struct inode
* inode
);
1984 extern void iput(struct inode
*);
1985 extern int generic_update_time(struct inode
*, struct timespec
*, int);
1987 static inline struct inode
*file_inode(const struct file
*f
)
1993 extern struct kobject
*fs_kobj
;
1995 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1997 #define FLOCK_VERIFY_READ 1
1998 #define FLOCK_VERIFY_WRITE 2
2000 #ifdef CONFIG_FILE_LOCKING
2001 extern int locks_mandatory_locked(struct file
*);
2002 extern int locks_mandatory_area(int, struct inode
*, struct file
*, loff_t
, size_t);
2005 * Candidates for mandatory locking have the setgid bit set
2006 * but no group execute bit - an otherwise meaningless combination.
2009 static inline int __mandatory_lock(struct inode
*ino
)
2011 return (ino
->i_mode
& (S_ISGID
| S_IXGRP
)) == S_ISGID
;
2015 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2016 * otherwise these will be advisory locks
2019 static inline int mandatory_lock(struct inode
*ino
)
2021 return IS_MANDLOCK(ino
) && __mandatory_lock(ino
);
2024 static inline int locks_verify_locked(struct file
*file
)
2026 if (mandatory_lock(file_inode(file
)))
2027 return locks_mandatory_locked(file
);
2031 static inline int locks_verify_truncate(struct inode
*inode
,
2035 if (inode
->i_flctx
&& mandatory_lock(inode
))
2036 return locks_mandatory_area(
2037 FLOCK_VERIFY_WRITE
, inode
, filp
,
2038 size
< inode
->i_size
? size
: inode
->i_size
,
2039 (size
< inode
->i_size
? inode
->i_size
- size
2040 : size
- inode
->i_size
)
2045 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2048 * Since this check is lockless, we must ensure that any refcounts
2049 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2050 * could end up racing with tasks trying to set a new lease on this
2054 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2055 return __break_lease(inode
, mode
, FL_LEASE
);
2059 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2062 * Since this check is lockless, we must ensure that any refcounts
2063 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2064 * could end up racing with tasks trying to set a new lease on this
2068 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2069 return __break_lease(inode
, mode
, FL_DELEG
);
2073 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2077 ret
= break_deleg(inode
, O_WRONLY
|O_NONBLOCK
);
2078 if (ret
== -EWOULDBLOCK
&& delegated_inode
) {
2079 *delegated_inode
= inode
;
2085 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2089 ret
= break_deleg(*delegated_inode
, O_WRONLY
);
2090 iput(*delegated_inode
);
2091 *delegated_inode
= NULL
;
2095 static inline int break_layout(struct inode
*inode
, bool wait
)
2098 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2099 return __break_lease(inode
,
2100 wait
? O_WRONLY
: O_WRONLY
| O_NONBLOCK
,
2105 #else /* !CONFIG_FILE_LOCKING */
2106 static inline int locks_mandatory_locked(struct file
*file
)
2111 static inline int locks_mandatory_area(int rw
, struct inode
*inode
,
2112 struct file
*filp
, loff_t offset
,
2118 static inline int __mandatory_lock(struct inode
*inode
)
2123 static inline int mandatory_lock(struct inode
*inode
)
2128 static inline int locks_verify_locked(struct file
*file
)
2133 static inline int locks_verify_truncate(struct inode
*inode
, struct file
*filp
,
2139 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2144 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2149 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2154 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2160 static inline int break_layout(struct inode
*inode
, bool wait
)
2165 #endif /* CONFIG_FILE_LOCKING */
2170 const char *name
; /* pointer to actual string */
2171 const __user
char *uptr
; /* original userland pointer */
2172 struct audit_names
*aname
;
2177 extern long vfs_truncate(struct path
*, loff_t
);
2178 extern int do_truncate(struct dentry
*, loff_t start
, unsigned int time_attrs
,
2180 extern int vfs_fallocate(struct file
*file
, int mode
, loff_t offset
,
2182 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
2184 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
2185 extern struct file
*filp_open(const char *, int, umode_t
);
2186 extern struct file
*file_open_root(struct dentry
*, struct vfsmount
*,
2188 extern int vfs_open(const struct path
*, struct file
*, const struct cred
*);
2189 extern struct file
* dentry_open(const struct path
*, int, const struct cred
*);
2190 extern int filp_close(struct file
*, fl_owner_t id
);
2192 extern struct filename
*getname_flags(const char __user
*, int, int *);
2193 extern struct filename
*getname(const char __user
*);
2194 extern struct filename
*getname_kernel(const char *);
2195 extern void putname(struct filename
*name
);
2201 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
2202 int (*open
)(struct inode
*, struct file
*),
2204 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
2208 extern int ioctl_preallocate(struct file
*filp
, void __user
*argp
);
2211 extern void __init
vfs_caches_init_early(void);
2212 extern void __init
vfs_caches_init(unsigned long);
2214 extern struct kmem_cache
*names_cachep
;
2216 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2217 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2220 extern int register_blkdev(unsigned int, const char *);
2221 extern void unregister_blkdev(unsigned int, const char *);
2222 extern struct block_device
*bdget(dev_t
);
2223 extern struct block_device
*bdgrab(struct block_device
*bdev
);
2224 extern void bd_set_size(struct block_device
*, loff_t size
);
2225 extern void bd_forget(struct inode
*inode
);
2226 extern void bdput(struct block_device
*);
2227 extern void invalidate_bdev(struct block_device
*);
2228 extern void iterate_bdevs(void (*)(struct block_device
*, void *), void *);
2229 extern int sync_blockdev(struct block_device
*bdev
);
2230 extern void kill_bdev(struct block_device
*);
2231 extern struct super_block
*freeze_bdev(struct block_device
*);
2232 extern void emergency_thaw_all(void);
2233 extern int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
);
2234 extern int fsync_bdev(struct block_device
*);
2235 extern int sb_is_blkdev_sb(struct super_block
*sb
);
2237 static inline void bd_forget(struct inode
*inode
) {}
2238 static inline int sync_blockdev(struct block_device
*bdev
) { return 0; }
2239 static inline void kill_bdev(struct block_device
*bdev
) {}
2240 static inline void invalidate_bdev(struct block_device
*bdev
) {}
2242 static inline struct super_block
*freeze_bdev(struct block_device
*sb
)
2247 static inline int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
)
2252 static inline void iterate_bdevs(void (*f
)(struct block_device
*, void *), void *arg
)
2256 static inline int sb_is_blkdev_sb(struct super_block
*sb
)
2261 extern int sync_filesystem(struct super_block
*);
2262 extern const struct file_operations def_blk_fops
;
2263 extern const struct file_operations def_chr_fops
;
2265 extern int ioctl_by_bdev(struct block_device
*, unsigned, unsigned long);
2266 extern int blkdev_ioctl(struct block_device
*, fmode_t
, unsigned, unsigned long);
2267 extern long compat_blkdev_ioctl(struct file
*, unsigned, unsigned long);
2268 extern int blkdev_get(struct block_device
*bdev
, fmode_t mode
, void *holder
);
2269 extern struct block_device
*blkdev_get_by_path(const char *path
, fmode_t mode
,
2271 extern struct block_device
*blkdev_get_by_dev(dev_t dev
, fmode_t mode
,
2273 extern void blkdev_put(struct block_device
*bdev
, fmode_t mode
);
2275 extern int bd_link_disk_holder(struct block_device
*bdev
, struct gendisk
*disk
);
2276 extern void bd_unlink_disk_holder(struct block_device
*bdev
,
2277 struct gendisk
*disk
);
2279 static inline int bd_link_disk_holder(struct block_device
*bdev
,
2280 struct gendisk
*disk
)
2284 static inline void bd_unlink_disk_holder(struct block_device
*bdev
,
2285 struct gendisk
*disk
)
2292 #define CHRDEV_MAJOR_HASH_SIZE 255
2293 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
2294 extern int register_chrdev_region(dev_t
, unsigned, const char *);
2295 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
2296 unsigned int count
, const char *name
,
2297 const struct file_operations
*fops
);
2298 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
2299 unsigned int count
, const char *name
);
2300 extern void unregister_chrdev_region(dev_t
, unsigned);
2301 extern void chrdev_show(struct seq_file
*,off_t
);
2303 static inline int register_chrdev(unsigned int major
, const char *name
,
2304 const struct file_operations
*fops
)
2306 return __register_chrdev(major
, 0, 256, name
, fops
);
2309 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2311 __unregister_chrdev(major
, 0, 256, name
);
2314 /* fs/block_dev.c */
2315 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2316 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2319 #define BLKDEV_MAJOR_HASH_SIZE 255
2320 extern const char *__bdevname(dev_t
, char *buffer
);
2321 extern const char *bdevname(struct block_device
*bdev
, char *buffer
);
2322 extern struct block_device
*lookup_bdev(const char *);
2323 extern void blkdev_show(struct seq_file
*,off_t
);
2326 #define BLKDEV_MAJOR_HASH_SIZE 0
2329 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2331 /* Invalid inode operations -- fs/bad_inode.c */
2332 extern void make_bad_inode(struct inode
*);
2333 extern int is_bad_inode(struct inode
*);
2337 * return READ, READA, or WRITE
2339 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2342 * return data direction, READ or WRITE
2344 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2346 extern void check_disk_size_change(struct gendisk
*disk
,
2347 struct block_device
*bdev
);
2348 extern int revalidate_disk(struct gendisk
*);
2349 extern int check_disk_change(struct block_device
*);
2350 extern int __invalidate_device(struct block_device
*, bool);
2351 extern int invalidate_partition(struct gendisk
*, int);
2353 unsigned long invalidate_mapping_pages(struct address_space
*mapping
,
2354 pgoff_t start
, pgoff_t end
);
2356 static inline void invalidate_remote_inode(struct inode
*inode
)
2358 if (S_ISREG(inode
->i_mode
) || S_ISDIR(inode
->i_mode
) ||
2359 S_ISLNK(inode
->i_mode
))
2360 invalidate_mapping_pages(inode
->i_mapping
, 0, -1);
2362 extern int invalidate_inode_pages2(struct address_space
*mapping
);
2363 extern int invalidate_inode_pages2_range(struct address_space
*mapping
,
2364 pgoff_t start
, pgoff_t end
);
2365 extern int write_inode_now(struct inode
*, int);
2366 extern int filemap_fdatawrite(struct address_space
*);
2367 extern int filemap_flush(struct address_space
*);
2368 extern int filemap_fdatawait(struct address_space
*);
2369 extern int filemap_fdatawait_range(struct address_space
*, loff_t lstart
,
2371 extern int filemap_write_and_wait(struct address_space
*mapping
);
2372 extern int filemap_write_and_wait_range(struct address_space
*mapping
,
2373 loff_t lstart
, loff_t lend
);
2374 extern int __filemap_fdatawrite_range(struct address_space
*mapping
,
2375 loff_t start
, loff_t end
, int sync_mode
);
2376 extern int filemap_fdatawrite_range(struct address_space
*mapping
,
2377 loff_t start
, loff_t end
);
2379 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
2381 extern int vfs_fsync(struct file
*file
, int datasync
);
2382 static inline int generic_write_sync(struct file
*file
, loff_t pos
, loff_t count
)
2384 if (!(file
->f_flags
& O_DSYNC
) && !IS_SYNC(file
->f_mapping
->host
))
2386 return vfs_fsync_range(file
, pos
, pos
+ count
- 1,
2387 (file
->f_flags
& __O_SYNC
) ? 0 : 1);
2389 extern void emergency_sync(void);
2390 extern void emergency_remount(void);
2392 extern sector_t
bmap(struct inode
*, sector_t
);
2394 extern int notify_change(struct dentry
*, struct iattr
*, struct inode
**);
2395 extern int inode_permission(struct inode
*, int);
2396 extern int __inode_permission(struct inode
*, int);
2397 extern int generic_permission(struct inode
*, int);
2398 extern int __check_sticky(struct inode
*dir
, struct inode
*inode
);
2400 static inline bool execute_ok(struct inode
*inode
)
2402 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
2405 static inline void file_start_write(struct file
*file
)
2407 if (!S_ISREG(file_inode(file
)->i_mode
))
2409 __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, true);
2412 static inline bool file_start_write_trylock(struct file
*file
)
2414 if (!S_ISREG(file_inode(file
)->i_mode
))
2416 return __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, false);
2419 static inline void file_end_write(struct file
*file
)
2421 if (!S_ISREG(file_inode(file
)->i_mode
))
2423 __sb_end_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
);
2427 * get_write_access() gets write permission for a file.
2428 * put_write_access() releases this write permission.
2429 * This is used for regular files.
2430 * We cannot support write (and maybe mmap read-write shared) accesses and
2431 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2432 * can have the following values:
2433 * 0: no writers, no VM_DENYWRITE mappings
2434 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2435 * > 0: (i_writecount) users are writing to the file.
2437 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2438 * except for the cases where we don't hold i_writecount yet. Then we need to
2439 * use {get,deny}_write_access() - these functions check the sign and refuse
2440 * to do the change if sign is wrong.
2442 static inline int get_write_access(struct inode
*inode
)
2444 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2446 static inline int deny_write_access(struct file
*file
)
2448 struct inode
*inode
= file_inode(file
);
2449 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2451 static inline void put_write_access(struct inode
* inode
)
2453 atomic_dec(&inode
->i_writecount
);
2455 static inline void allow_write_access(struct file
*file
)
2458 atomic_inc(&file_inode(file
)->i_writecount
);
2460 static inline bool inode_is_open_for_write(const struct inode
*inode
)
2462 return atomic_read(&inode
->i_writecount
) > 0;
2466 static inline void i_readcount_dec(struct inode
*inode
)
2468 BUG_ON(!atomic_read(&inode
->i_readcount
));
2469 atomic_dec(&inode
->i_readcount
);
2471 static inline void i_readcount_inc(struct inode
*inode
)
2473 atomic_inc(&inode
->i_readcount
);
2476 static inline void i_readcount_dec(struct inode
*inode
)
2480 static inline void i_readcount_inc(struct inode
*inode
)
2485 extern int do_pipe_flags(int *, int);
2487 extern int kernel_read(struct file
*, loff_t
, char *, unsigned long);
2488 extern ssize_t
kernel_write(struct file
*, const char *, size_t, loff_t
);
2489 extern ssize_t
__kernel_write(struct file
*, const char *, size_t, loff_t
*);
2490 extern struct file
* open_exec(const char *);
2492 /* fs/dcache.c -- generic fs support functions */
2493 extern int is_subdir(struct dentry
*, struct dentry
*);
2494 extern int path_is_under(struct path
*, struct path
*);
2496 #include <linux/err.h>
2498 /* needed for stackable file system support */
2499 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int whence
);
2501 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int whence
);
2503 extern int inode_init_always(struct super_block
*, struct inode
*);
2504 extern void inode_init_once(struct inode
*);
2505 extern void address_space_init_once(struct address_space
*mapping
);
2506 extern struct inode
* igrab(struct inode
*);
2507 extern ino_t
iunique(struct super_block
*, ino_t
);
2508 extern int inode_needs_sync(struct inode
*inode
);
2509 extern int generic_delete_inode(struct inode
*inode
);
2510 static inline int generic_drop_inode(struct inode
*inode
)
2512 return !inode
->i_nlink
|| inode_unhashed(inode
);
2515 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
2516 unsigned long hashval
, int (*test
)(struct inode
*, void *),
2518 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
2519 int (*test
)(struct inode
*, void *), void *data
);
2520 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
2522 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
2523 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
2524 extern struct inode
*find_inode_nowait(struct super_block
*,
2526 int (*match
)(struct inode
*,
2527 unsigned long, void *),
2529 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
2530 extern int insert_inode_locked(struct inode
*);
2531 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2532 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
2534 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
2536 extern void unlock_new_inode(struct inode
*);
2537 extern unsigned int get_next_ino(void);
2539 extern void __iget(struct inode
* inode
);
2540 extern void iget_failed(struct inode
*);
2541 extern void clear_inode(struct inode
*);
2542 extern void __destroy_inode(struct inode
*);
2543 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
2544 extern struct inode
*new_inode(struct super_block
*sb
);
2545 extern void free_inode_nonrcu(struct inode
*inode
);
2546 extern int should_remove_suid(struct dentry
*);
2547 extern int file_remove_suid(struct file
*);
2549 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
2550 static inline void insert_inode_hash(struct inode
*inode
)
2552 __insert_inode_hash(inode
, inode
->i_ino
);
2555 extern void __remove_inode_hash(struct inode
*);
2556 static inline void remove_inode_hash(struct inode
*inode
)
2558 if (!inode_unhashed(inode
))
2559 __remove_inode_hash(inode
);
2562 extern void inode_sb_list_add(struct inode
*inode
);
2565 extern void submit_bio(int, struct bio
*);
2566 extern int bdev_read_only(struct block_device
*);
2568 extern int set_blocksize(struct block_device
*, int);
2569 extern int sb_set_blocksize(struct super_block
*, int);
2570 extern int sb_min_blocksize(struct super_block
*, int);
2572 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
2573 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
2574 extern ssize_t
generic_write_checks(struct kiocb
*, struct iov_iter
*);
2575 extern ssize_t
generic_file_read_iter(struct kiocb
*, struct iov_iter
*);
2576 extern ssize_t
__generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2577 extern ssize_t
generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2578 extern ssize_t
generic_file_direct_write(struct kiocb
*, struct iov_iter
*, loff_t
);
2579 extern ssize_t
generic_perform_write(struct file
*, struct iov_iter
*, loff_t
);
2581 ssize_t
vfs_iter_read(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
);
2582 ssize_t
vfs_iter_write(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
);
2584 /* fs/block_dev.c */
2585 extern ssize_t
blkdev_read_iter(struct kiocb
*iocb
, struct iov_iter
*to
);
2586 extern ssize_t
blkdev_write_iter(struct kiocb
*iocb
, struct iov_iter
*from
);
2587 extern int blkdev_fsync(struct file
*filp
, loff_t start
, loff_t end
,
2589 extern void block_sync_page(struct page
*page
);
2592 extern ssize_t
generic_file_splice_read(struct file
*, loff_t
*,
2593 struct pipe_inode_info
*, size_t, unsigned int);
2594 extern ssize_t
default_file_splice_read(struct file
*, loff_t
*,
2595 struct pipe_inode_info
*, size_t, unsigned int);
2596 extern ssize_t
iter_file_splice_write(struct pipe_inode_info
*,
2597 struct file
*, loff_t
*, size_t, unsigned int);
2598 extern ssize_t
generic_splice_sendpage(struct pipe_inode_info
*pipe
,
2599 struct file
*out
, loff_t
*, size_t len
, unsigned int flags
);
2600 extern long do_splice_direct(struct file
*in
, loff_t
*ppos
, struct file
*out
,
2601 loff_t
*opos
, size_t len
, unsigned int flags
);
2605 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
2606 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int whence
);
2607 extern loff_t
no_llseek(struct file
*file
, loff_t offset
, int whence
);
2608 extern loff_t
vfs_setpos(struct file
*file
, loff_t offset
, loff_t maxsize
);
2609 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int whence
);
2610 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
2611 int whence
, loff_t maxsize
, loff_t eof
);
2612 extern loff_t
fixed_size_llseek(struct file
*file
, loff_t offset
,
2613 int whence
, loff_t size
);
2614 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
2615 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
2617 ssize_t
dax_do_io(struct kiocb
*, struct inode
*, struct iov_iter
*, loff_t
,
2618 get_block_t
, dio_iodone_t
, int flags
);
2619 int dax_clear_blocks(struct inode
*, sector_t block
, long size
);
2620 int dax_zero_page_range(struct inode
*, loff_t from
, unsigned len
, get_block_t
);
2621 int dax_truncate_page(struct inode
*, loff_t from
, get_block_t
);
2622 int dax_fault(struct vm_area_struct
*, struct vm_fault
*, get_block_t
);
2623 #define dax_mkwrite(vma, vmf, gb) dax_fault(vma, vmf, gb)
2626 typedef void (dio_submit_t
)(int rw
, struct bio
*bio
, struct inode
*inode
,
2627 loff_t file_offset
);
2630 /* need locking between buffered and direct access */
2633 /* filesystem does not support filling holes */
2634 DIO_SKIP_HOLES
= 0x02,
2636 /* filesystem can handle aio writes beyond i_size */
2637 DIO_ASYNC_EXTEND
= 0x04,
2640 void dio_end_io(struct bio
*bio
, int error
);
2642 ssize_t
__blockdev_direct_IO(struct kiocb
*iocb
, struct inode
*inode
,
2643 struct block_device
*bdev
, struct iov_iter
*iter
,
2644 loff_t offset
, get_block_t get_block
,
2645 dio_iodone_t end_io
, dio_submit_t submit_io
,
2648 static inline ssize_t
blockdev_direct_IO(struct kiocb
*iocb
,
2649 struct inode
*inode
,
2650 struct iov_iter
*iter
, loff_t offset
,
2651 get_block_t get_block
)
2653 return __blockdev_direct_IO(iocb
, inode
, inode
->i_sb
->s_bdev
, iter
,
2654 offset
, get_block
, NULL
, NULL
,
2655 DIO_LOCKING
| DIO_SKIP_HOLES
);
2659 void inode_dio_wait(struct inode
*inode
);
2660 void inode_dio_done(struct inode
*inode
);
2662 extern void inode_set_flags(struct inode
*inode
, unsigned int flags
,
2665 extern const struct file_operations generic_ro_fops
;
2667 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2669 extern int readlink_copy(char __user
*, int, const char *);
2670 extern int page_readlink(struct dentry
*, char __user
*, int);
2671 extern void *page_follow_link_light(struct dentry
*, struct nameidata
*);
2672 extern void page_put_link(struct dentry
*, struct nameidata
*, void *);
2673 extern int __page_symlink(struct inode
*inode
, const char *symname
, int len
,
2675 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
2676 extern const struct inode_operations page_symlink_inode_operations
;
2677 extern void kfree_put_link(struct dentry
*, struct nameidata
*, void *);
2678 extern int generic_readlink(struct dentry
*, char __user
*, int);
2679 extern void generic_fillattr(struct inode
*, struct kstat
*);
2680 int vfs_getattr_nosec(struct path
*path
, struct kstat
*stat
);
2681 extern int vfs_getattr(struct path
*, struct kstat
*);
2682 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2683 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2684 void __inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2685 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2686 loff_t
inode_get_bytes(struct inode
*inode
);
2687 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
2689 extern int vfs_readdir(struct file
*, filldir_t
, void *);
2690 extern int iterate_dir(struct file
*, struct dir_context
*);
2692 extern int vfs_stat(const char __user
*, struct kstat
*);
2693 extern int vfs_lstat(const char __user
*, struct kstat
*);
2694 extern int vfs_fstat(unsigned int, struct kstat
*);
2695 extern int vfs_fstatat(int , const char __user
*, struct kstat
*, int);
2697 extern int do_vfs_ioctl(struct file
*filp
, unsigned int fd
, unsigned int cmd
,
2699 extern int __generic_block_fiemap(struct inode
*inode
,
2700 struct fiemap_extent_info
*fieinfo
,
2701 loff_t start
, loff_t len
,
2702 get_block_t
*get_block
);
2703 extern int generic_block_fiemap(struct inode
*inode
,
2704 struct fiemap_extent_info
*fieinfo
, u64 start
,
2705 u64 len
, get_block_t
*get_block
);
2707 extern void get_filesystem(struct file_system_type
*fs
);
2708 extern void put_filesystem(struct file_system_type
*fs
);
2709 extern struct file_system_type
*get_fs_type(const char *name
);
2710 extern struct super_block
*get_super(struct block_device
*);
2711 extern struct super_block
*get_super_thawed(struct block_device
*);
2712 extern struct super_block
*get_active_super(struct block_device
*bdev
);
2713 extern void drop_super(struct super_block
*sb
);
2714 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
2715 extern void iterate_supers_type(struct file_system_type
*,
2716 void (*)(struct super_block
*, void *), void *);
2718 extern int dcache_dir_open(struct inode
*, struct file
*);
2719 extern int dcache_dir_close(struct inode
*, struct file
*);
2720 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
2721 extern int dcache_readdir(struct file
*, struct dir_context
*);
2722 extern int simple_setattr(struct dentry
*, struct iattr
*);
2723 extern int simple_getattr(struct vfsmount
*, struct dentry
*, struct kstat
*);
2724 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
2725 extern int simple_open(struct inode
*inode
, struct file
*file
);
2726 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
2727 extern int simple_unlink(struct inode
*, struct dentry
*);
2728 extern int simple_rmdir(struct inode
*, struct dentry
*);
2729 extern int simple_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*);
2730 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
2731 extern int simple_empty(struct dentry
*);
2732 extern int simple_readpage(struct file
*file
, struct page
*page
);
2733 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
2734 loff_t pos
, unsigned len
, unsigned flags
,
2735 struct page
**pagep
, void **fsdata
);
2736 extern int simple_write_end(struct file
*file
, struct address_space
*mapping
,
2737 loff_t pos
, unsigned len
, unsigned copied
,
2738 struct page
*page
, void *fsdata
);
2739 extern int always_delete_dentry(const struct dentry
*);
2740 extern struct inode
*alloc_anon_inode(struct super_block
*);
2741 extern int simple_nosetlease(struct file
*, long, struct file_lock
**, void **);
2742 extern const struct dentry_operations simple_dentry_operations
;
2744 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
2745 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
2746 extern const struct file_operations simple_dir_operations
;
2747 extern const struct inode_operations simple_dir_inode_operations
;
2748 struct tree_descr
{ char *name
; const struct file_operations
*ops
; int mode
; };
2749 struct dentry
*d_alloc_name(struct dentry
*, const char *);
2750 extern int simple_fill_super(struct super_block
*, unsigned long, struct tree_descr
*);
2751 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
2752 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
2754 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
2755 loff_t
*ppos
, const void *from
, size_t available
);
2756 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
2757 const void __user
*from
, size_t count
);
2759 extern int __generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
2760 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
2762 extern int generic_check_addressable(unsigned, u64
);
2764 #ifdef CONFIG_MIGRATION
2765 extern int buffer_migrate_page(struct address_space
*,
2766 struct page
*, struct page
*,
2769 #define buffer_migrate_page NULL
2772 extern int inode_change_ok(const struct inode
*, struct iattr
*);
2773 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
2774 extern void setattr_copy(struct inode
*inode
, const struct iattr
*attr
);
2776 extern int file_update_time(struct file
*file
);
2778 extern int generic_show_options(struct seq_file
*m
, struct dentry
*root
);
2779 extern void save_mount_options(struct super_block
*sb
, char *options
);
2780 extern void replace_mount_options(struct super_block
*sb
, char *options
);
2782 static inline bool io_is_direct(struct file
*filp
)
2784 return (filp
->f_flags
& O_DIRECT
) || IS_DAX(file_inode(filp
));
2787 static inline int iocb_flags(struct file
*file
)
2790 if (file
->f_flags
& O_APPEND
)
2792 if (io_is_direct(file
))
2797 static inline ino_t
parent_ino(struct dentry
*dentry
)
2802 * Don't strictly need d_lock here? If the parent ino could change
2803 * then surely we'd have a deeper race in the caller?
2805 spin_lock(&dentry
->d_lock
);
2806 res
= dentry
->d_parent
->d_inode
->i_ino
;
2807 spin_unlock(&dentry
->d_lock
);
2811 /* Transaction based IO helpers */
2814 * An argresp is stored in an allocated page and holds the
2815 * size of the argument or response, along with its content
2817 struct simple_transaction_argresp
{
2822 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2824 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
2826 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
2827 size_t size
, loff_t
*pos
);
2828 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
2830 void simple_transaction_set(struct file
*file
, size_t n
);
2833 * simple attribute files
2835 * These attributes behave similar to those in sysfs:
2837 * Writing to an attribute immediately sets a value, an open file can be
2838 * written to multiple times.
2840 * Reading from an attribute creates a buffer from the value that might get
2841 * read with multiple read calls. When the attribute has been read
2842 * completely, no further read calls are possible until the file is opened
2845 * All attributes contain a text representation of a numeric value
2846 * that are accessed with the get() and set() functions.
2848 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2849 static int __fops ## _open(struct inode *inode, struct file *file) \
2851 __simple_attr_check_format(__fmt, 0ull); \
2852 return simple_attr_open(inode, file, __get, __set, __fmt); \
2854 static const struct file_operations __fops = { \
2855 .owner = THIS_MODULE, \
2856 .open = __fops ## _open, \
2857 .release = simple_attr_release, \
2858 .read = simple_attr_read, \
2859 .write = simple_attr_write, \
2860 .llseek = generic_file_llseek, \
2863 static inline __printf(1, 2)
2864 void __simple_attr_check_format(const char *fmt
, ...)
2866 /* don't do anything, just let the compiler check the arguments; */
2869 int simple_attr_open(struct inode
*inode
, struct file
*file
,
2870 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
2872 int simple_attr_release(struct inode
*inode
, struct file
*file
);
2873 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
2874 size_t len
, loff_t
*ppos
);
2875 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
2876 size_t len
, loff_t
*ppos
);
2879 int proc_nr_files(struct ctl_table
*table
, int write
,
2880 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
2881 int proc_nr_dentry(struct ctl_table
*table
, int write
,
2882 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
2883 int proc_nr_inodes(struct ctl_table
*table
, int write
,
2884 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
2885 int __init
get_filesystem_list(char *buf
);
2887 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2888 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2890 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2891 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2892 (flag & __FMODE_NONOTIFY)))
2894 static inline int is_sxid(umode_t mode
)
2896 return (mode
& S_ISUID
) || ((mode
& S_ISGID
) && (mode
& S_IXGRP
));
2899 static inline int check_sticky(struct inode
*dir
, struct inode
*inode
)
2901 if (!(dir
->i_mode
& S_ISVTX
))
2904 return __check_sticky(dir
, inode
);
2907 static inline void inode_has_no_xattr(struct inode
*inode
)
2909 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& MS_NOSEC
))
2910 inode
->i_flags
|= S_NOSEC
;
2913 static inline bool is_root_inode(struct inode
*inode
)
2915 return inode
== inode
->i_sb
->s_root
->d_inode
;
2918 static inline bool dir_emit(struct dir_context
*ctx
,
2919 const char *name
, int namelen
,
2920 u64 ino
, unsigned type
)
2922 return ctx
->actor(ctx
, name
, namelen
, ctx
->pos
, ino
, type
) == 0;
2924 static inline bool dir_emit_dot(struct file
*file
, struct dir_context
*ctx
)
2926 return ctx
->actor(ctx
, ".", 1, ctx
->pos
,
2927 file
->f_path
.dentry
->d_inode
->i_ino
, DT_DIR
) == 0;
2929 static inline bool dir_emit_dotdot(struct file
*file
, struct dir_context
*ctx
)
2931 return ctx
->actor(ctx
, "..", 2, ctx
->pos
,
2932 parent_ino(file
->f_path
.dentry
), DT_DIR
) == 0;
2934 static inline bool dir_emit_dots(struct file
*file
, struct dir_context
*ctx
)
2936 if (ctx
->pos
== 0) {
2937 if (!dir_emit_dot(file
, ctx
))
2941 if (ctx
->pos
== 1) {
2942 if (!dir_emit_dotdot(file
, ctx
))
2948 static inline bool dir_relax(struct inode
*inode
)
2950 mutex_unlock(&inode
->i_mutex
);
2951 mutex_lock(&inode
->i_mutex
);
2952 return !IS_DEADDIR(inode
);
2955 #endif /* _LINUX_FS_H */