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 export_operations
;
43 struct pipe_inode_info
;
44 struct poll_table_struct
;
46 struct vm_area_struct
;
49 struct swap_info_struct
;
51 struct workqueue_struct
;
54 extern void __init
inode_init(void);
55 extern void __init
inode_init_early(void);
56 extern void __init
files_init(unsigned long);
58 extern struct files_stat_struct files_stat
;
59 extern unsigned long get_max_files(void);
60 extern int sysctl_nr_open
;
61 extern struct inodes_stat_t inodes_stat
;
62 extern int leases_enable
, lease_break_time
;
63 extern int sysctl_protected_symlinks
;
64 extern int sysctl_protected_hardlinks
;
67 typedef int (get_block_t
)(struct inode
*inode
, sector_t iblock
,
68 struct buffer_head
*bh_result
, int create
);
69 typedef void (dio_iodone_t
)(struct kiocb
*iocb
, loff_t offset
,
70 ssize_t bytes
, void *private);
72 #define MAY_EXEC 0x00000001
73 #define MAY_WRITE 0x00000002
74 #define MAY_READ 0x00000004
75 #define MAY_APPEND 0x00000008
76 #define MAY_ACCESS 0x00000010
77 #define MAY_OPEN 0x00000020
78 #define MAY_CHDIR 0x00000040
79 /* called from RCU mode, don't block */
80 #define MAY_NOT_BLOCK 0x00000080
83 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
84 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
87 /* file is open for reading */
88 #define FMODE_READ ((__force fmode_t)0x1)
89 /* file is open for writing */
90 #define FMODE_WRITE ((__force fmode_t)0x2)
91 /* file is seekable */
92 #define FMODE_LSEEK ((__force fmode_t)0x4)
93 /* file can be accessed using pread */
94 #define FMODE_PREAD ((__force fmode_t)0x8)
95 /* file can be accessed using pwrite */
96 #define FMODE_PWRITE ((__force fmode_t)0x10)
97 /* File is opened for execution with sys_execve / sys_uselib */
98 #define FMODE_EXEC ((__force fmode_t)0x20)
99 /* File is opened with O_NDELAY (only set for block devices) */
100 #define FMODE_NDELAY ((__force fmode_t)0x40)
101 /* File is opened with O_EXCL (only set for block devices) */
102 #define FMODE_EXCL ((__force fmode_t)0x80)
103 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
104 (specialy hack for floppy.c) */
105 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
106 /* 32bit hashes as llseek() offset (for directories) */
107 #define FMODE_32BITHASH ((__force fmode_t)0x200)
108 /* 64bit hashes as llseek() offset (for directories) */
109 #define FMODE_64BITHASH ((__force fmode_t)0x400)
112 * Don't update ctime and mtime.
114 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
115 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
117 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
119 /* Expect random access pattern */
120 #define FMODE_RANDOM ((__force fmode_t)0x1000)
122 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
123 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
125 /* File is opened with O_PATH; almost nothing can be done with it */
126 #define FMODE_PATH ((__force fmode_t)0x4000)
128 /* File needs atomic accesses to f_pos */
129 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
130 /* Write access to underlying fs */
131 #define FMODE_WRITER ((__force fmode_t)0x10000)
132 /* Has read method(s) */
133 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
134 /* Has write method(s) */
135 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
137 /* File was opened by fanotify and shouldn't generate fanotify events */
138 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
141 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
142 * that indicates that they should check the contents of the iovec are
143 * valid, but not check the memory that the iovec elements
146 #define CHECK_IOVEC_ONLY -1
149 * The below are the various read and write types that we support. Some of
150 * them include behavioral modifiers that send information down to the
151 * block layer and IO scheduler. Terminology:
153 * The block layer uses device plugging to defer IO a little bit, in
154 * the hope that we will see more IO very shortly. This increases
155 * coalescing of adjacent IO and thus reduces the number of IOs we
156 * have to send to the device. It also allows for better queuing,
157 * if the IO isn't mergeable. If the caller is going to be waiting
158 * for the IO, then he must ensure that the device is unplugged so
159 * that the IO is dispatched to the driver.
161 * All IO is handled async in Linux. This is fine for background
162 * writes, but for reads or writes that someone waits for completion
163 * on, we want to notify the block layer and IO scheduler so that they
164 * know about it. That allows them to make better scheduling
165 * decisions. So when the below references 'sync' and 'async', it
166 * is referencing this priority hint.
168 * With that in mind, the available types are:
170 * READ A normal read operation. Device will be plugged.
171 * READ_SYNC A synchronous read. Device is not plugged, caller can
172 * immediately wait on this read without caring about
174 * READA Used for read-ahead operations. Lower priority, and the
175 * block layer could (in theory) choose to ignore this
176 * request if it runs into resource problems.
177 * WRITE A normal async write. Device will be plugged.
178 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
179 * the hint that someone will be waiting on this IO
180 * shortly. The write equivalent of READ_SYNC.
181 * WRITE_ODIRECT Special case write for O_DIRECT only.
182 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
183 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
184 * non-volatile media on completion.
185 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
186 * by a cache flush and data is guaranteed to be on
187 * non-volatile media on completion.
190 #define RW_MASK REQ_WRITE
191 #define RWA_MASK REQ_RAHEAD
194 #define WRITE RW_MASK
195 #define READA RWA_MASK
197 #define READ_SYNC (READ | REQ_SYNC)
198 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
199 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
200 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
201 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
202 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
205 * Attribute flags. These should be or-ed together to figure out what
208 #define ATTR_MODE (1 << 0)
209 #define ATTR_UID (1 << 1)
210 #define ATTR_GID (1 << 2)
211 #define ATTR_SIZE (1 << 3)
212 #define ATTR_ATIME (1 << 4)
213 #define ATTR_MTIME (1 << 5)
214 #define ATTR_CTIME (1 << 6)
215 #define ATTR_ATIME_SET (1 << 7)
216 #define ATTR_MTIME_SET (1 << 8)
217 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
218 #define ATTR_ATTR_FLAG (1 << 10)
219 #define ATTR_KILL_SUID (1 << 11)
220 #define ATTR_KILL_SGID (1 << 12)
221 #define ATTR_FILE (1 << 13)
222 #define ATTR_KILL_PRIV (1 << 14)
223 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
224 #define ATTR_TIMES_SET (1 << 16)
227 * Whiteout is represented by a char device. The following constants define the
228 * mode and device number to use.
230 #define WHITEOUT_MODE 0
231 #define WHITEOUT_DEV 0
234 * This is the Inode Attributes structure, used for notify_change(). It
235 * uses the above definitions as flags, to know which values have changed.
236 * Also, in this manner, a Filesystem can look at only the values it cares
237 * about. Basically, these are the attributes that the VFS layer can
238 * request to change from the FS layer.
240 * Derek Atkins <warlord@MIT.EDU> 94-10-20
243 unsigned int ia_valid
;
248 struct timespec ia_atime
;
249 struct timespec ia_mtime
;
250 struct timespec ia_ctime
;
253 * Not an attribute, but an auxiliary info for filesystems wanting to
254 * implement an ftruncate() like method. NOTE: filesystem should
255 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
257 struct file
*ia_file
;
261 * Includes for diskquotas.
263 #include <linux/quota.h>
266 * Maximum number of layers of fs stack. Needs to be limited to
267 * prevent kernel stack overflow
269 #define FILESYSTEM_MAX_STACK_DEPTH 2
272 * enum positive_aop_returns - aop return codes with specific semantics
274 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
275 * completed, that the page is still locked, and
276 * should be considered active. The VM uses this hint
277 * to return the page to the active list -- it won't
278 * be a candidate for writeback again in the near
279 * future. Other callers must be careful to unlock
280 * the page if they get this return. Returned by
283 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
284 * unlocked it and the page might have been truncated.
285 * The caller should back up to acquiring a new page and
286 * trying again. The aop will be taking reasonable
287 * precautions not to livelock. If the caller held a page
288 * reference, it should drop it before retrying. Returned
291 * address_space_operation functions return these large constants to indicate
292 * special semantics to the caller. These are much larger than the bytes in a
293 * page to allow for functions that return the number of bytes operated on in a
297 enum positive_aop_returns
{
298 AOP_WRITEPAGE_ACTIVATE
= 0x80000,
299 AOP_TRUNCATED_PAGE
= 0x80001,
302 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
303 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
304 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
305 * helper code (eg buffer layer)
306 * to clear GFP_FS from alloc */
309 * oh the beauties of C type declarations.
312 struct address_space
;
313 struct writeback_control
;
316 * "descriptor" for what we're up to with a read.
317 * This allows us to use the same read code yet
318 * have multiple different users of the data that
319 * we read from a file.
321 * The simplest case just copies the data to user
334 typedef int (*read_actor_t
)(read_descriptor_t
*, struct page
*,
335 unsigned long, unsigned long);
337 struct address_space_operations
{
338 int (*writepage
)(struct page
*page
, struct writeback_control
*wbc
);
339 int (*readpage
)(struct file
*, struct page
*);
341 /* Write back some dirty pages from this mapping. */
342 int (*writepages
)(struct address_space
*, struct writeback_control
*);
344 /* Set a page dirty. Return true if this dirtied it */
345 int (*set_page_dirty
)(struct page
*page
);
347 int (*readpages
)(struct file
*filp
, struct address_space
*mapping
,
348 struct list_head
*pages
, unsigned nr_pages
);
350 int (*write_begin
)(struct file
*, struct address_space
*mapping
,
351 loff_t pos
, unsigned len
, unsigned flags
,
352 struct page
**pagep
, void **fsdata
);
353 int (*write_end
)(struct file
*, struct address_space
*mapping
,
354 loff_t pos
, unsigned len
, unsigned copied
,
355 struct page
*page
, void *fsdata
);
357 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
358 sector_t (*bmap
)(struct address_space
*, sector_t
);
359 void (*invalidatepage
) (struct page
*, unsigned int, unsigned int);
360 int (*releasepage
) (struct page
*, gfp_t
);
361 void (*freepage
)(struct page
*);
362 ssize_t (*direct_IO
)(int, struct kiocb
*, struct iov_iter
*iter
, loff_t offset
);
363 int (*get_xip_mem
)(struct address_space
*, pgoff_t
, int,
364 void **, unsigned long *);
366 * migrate the contents of a page to the specified target. If
367 * migrate_mode is MIGRATE_ASYNC, it must not block.
369 int (*migratepage
) (struct address_space
*,
370 struct page
*, struct page
*, enum migrate_mode
);
371 int (*launder_page
) (struct page
*);
372 int (*is_partially_uptodate
) (struct page
*, unsigned long,
374 void (*is_dirty_writeback
) (struct page
*, bool *, bool *);
375 int (*error_remove_page
)(struct address_space
*, struct page
*);
377 /* swapfile support */
378 int (*swap_activate
)(struct swap_info_struct
*sis
, struct file
*file
,
380 void (*swap_deactivate
)(struct file
*file
);
383 extern const struct address_space_operations empty_aops
;
386 * pagecache_write_begin/pagecache_write_end must be used by general code
387 * to write into the pagecache.
389 int pagecache_write_begin(struct file
*, struct address_space
*mapping
,
390 loff_t pos
, unsigned len
, unsigned flags
,
391 struct page
**pagep
, void **fsdata
);
393 int pagecache_write_end(struct file
*, struct address_space
*mapping
,
394 loff_t pos
, unsigned len
, unsigned copied
,
395 struct page
*page
, void *fsdata
);
397 struct backing_dev_info
;
398 struct address_space
{
399 struct inode
*host
; /* owner: inode, block_device */
400 struct radix_tree_root page_tree
; /* radix tree of all pages */
401 spinlock_t tree_lock
; /* and lock protecting it */
402 atomic_t i_mmap_writable
;/* count VM_SHARED mappings */
403 struct rb_root i_mmap
; /* tree of private and shared mappings */
404 struct list_head i_mmap_nonlinear
;/*list VM_NONLINEAR mappings */
405 struct rw_semaphore i_mmap_rwsem
; /* protect tree, count, list */
406 /* Protected by tree_lock together with the radix tree */
407 unsigned long nrpages
; /* number of total pages */
408 unsigned long nrshadows
; /* number of shadow entries */
409 pgoff_t writeback_index
;/* writeback starts here */
410 const struct address_space_operations
*a_ops
; /* methods */
411 unsigned long flags
; /* error bits/gfp mask */
412 struct backing_dev_info
*backing_dev_info
; /* device readahead, etc */
413 spinlock_t private_lock
; /* for use by the address_space */
414 struct list_head private_list
; /* ditto */
415 void *private_data
; /* ditto */
416 } __attribute__((aligned(sizeof(long))));
418 * On most architectures that alignment is already the case; but
419 * must be enforced here for CRIS, to let the least significant bit
420 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
422 struct request_queue
;
424 struct block_device
{
425 dev_t bd_dev
; /* not a kdev_t - it's a search key */
427 struct inode
* bd_inode
; /* will die */
428 struct super_block
* bd_super
;
429 struct mutex bd_mutex
; /* open/close mutex */
430 struct list_head bd_inodes
;
434 bool bd_write_holder
;
436 struct list_head bd_holder_disks
;
438 struct block_device
* bd_contains
;
439 unsigned bd_block_size
;
440 struct hd_struct
* bd_part
;
441 /* number of times partitions within this device have been opened. */
442 unsigned bd_part_count
;
444 struct gendisk
* bd_disk
;
445 struct request_queue
* bd_queue
;
446 struct list_head bd_list
;
448 * Private data. You must have bd_claim'ed the block_device
449 * to use this. NOTE: bd_claim allows an owner to claim
450 * the same device multiple times, the owner must take special
451 * care to not mess up bd_private for that case.
453 unsigned long bd_private
;
455 /* The counter of freeze processes */
456 int bd_fsfreeze_count
;
457 /* Mutex for freeze */
458 struct mutex bd_fsfreeze_mutex
;
462 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
465 #define PAGECACHE_TAG_DIRTY 0
466 #define PAGECACHE_TAG_WRITEBACK 1
467 #define PAGECACHE_TAG_TOWRITE 2
469 int mapping_tagged(struct address_space
*mapping
, int tag
);
471 static inline void i_mmap_lock_write(struct address_space
*mapping
)
473 down_write(&mapping
->i_mmap_rwsem
);
476 static inline void i_mmap_unlock_write(struct address_space
*mapping
)
478 up_write(&mapping
->i_mmap_rwsem
);
481 static inline void i_mmap_lock_read(struct address_space
*mapping
)
483 down_read(&mapping
->i_mmap_rwsem
);
486 static inline void i_mmap_unlock_read(struct address_space
*mapping
)
488 up_read(&mapping
->i_mmap_rwsem
);
492 * Might pages of this file be mapped into userspace?
494 static inline int mapping_mapped(struct address_space
*mapping
)
496 return !RB_EMPTY_ROOT(&mapping
->i_mmap
) ||
497 !list_empty(&mapping
->i_mmap_nonlinear
);
501 * Might pages of this file have been modified in userspace?
502 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
503 * marks vma as VM_SHARED if it is shared, and the file was opened for
504 * writing i.e. vma may be mprotected writable even if now readonly.
506 * If i_mmap_writable is negative, no new writable mappings are allowed. You
507 * can only deny writable mappings, if none exists right now.
509 static inline int mapping_writably_mapped(struct address_space
*mapping
)
511 return atomic_read(&mapping
->i_mmap_writable
) > 0;
514 static inline int mapping_map_writable(struct address_space
*mapping
)
516 return atomic_inc_unless_negative(&mapping
->i_mmap_writable
) ?
520 static inline void mapping_unmap_writable(struct address_space
*mapping
)
522 atomic_dec(&mapping
->i_mmap_writable
);
525 static inline int mapping_deny_writable(struct address_space
*mapping
)
527 return atomic_dec_unless_positive(&mapping
->i_mmap_writable
) ?
531 static inline void mapping_allow_writable(struct address_space
*mapping
)
533 atomic_inc(&mapping
->i_mmap_writable
);
537 * Use sequence counter to get consistent i_size on 32-bit processors.
539 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
540 #include <linux/seqlock.h>
541 #define __NEED_I_SIZE_ORDERED
542 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
544 #define i_size_ordered_init(inode) do { } while (0)
548 #define ACL_NOT_CACHED ((void *)(-1))
550 #define IOP_FASTPERM 0x0001
551 #define IOP_LOOKUP 0x0002
552 #define IOP_NOFOLLOW 0x0004
555 * Keep mostly read-only and often accessed (especially for
556 * the RCU path lookup and 'stat' data) fields at the beginning
557 * of the 'struct inode'
561 unsigned short i_opflags
;
564 unsigned int i_flags
;
566 #ifdef CONFIG_FS_POSIX_ACL
567 struct posix_acl
*i_acl
;
568 struct posix_acl
*i_default_acl
;
571 const struct inode_operations
*i_op
;
572 struct super_block
*i_sb
;
573 struct address_space
*i_mapping
;
575 #ifdef CONFIG_SECURITY
579 /* Stat data, not accessed from path walking */
582 * Filesystems may only read i_nlink directly. They shall use the
583 * following functions for modification:
585 * (set|clear|inc|drop)_nlink
586 * inode_(inc|dec)_link_count
589 const unsigned int i_nlink
;
590 unsigned int __i_nlink
;
594 struct timespec i_atime
;
595 struct timespec i_mtime
;
596 struct timespec i_ctime
;
597 spinlock_t i_lock
; /* i_blocks, i_bytes, maybe i_size */
598 unsigned short i_bytes
;
599 unsigned int i_blkbits
;
602 #ifdef __NEED_I_SIZE_ORDERED
603 seqcount_t i_size_seqcount
;
607 unsigned long i_state
;
608 struct mutex i_mutex
;
610 unsigned long dirtied_when
; /* jiffies of first dirtying */
612 struct hlist_node i_hash
;
613 struct list_head i_wb_list
; /* backing dev IO list */
614 struct list_head i_lru
; /* inode LRU list */
615 struct list_head i_sb_list
;
617 struct hlist_head i_dentry
;
618 struct rcu_head i_rcu
;
622 atomic_t i_dio_count
;
623 atomic_t i_writecount
;
625 atomic_t i_readcount
; /* struct files open RO */
627 const struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
628 struct file_lock_context
*i_flctx
;
629 struct address_space i_data
;
630 struct list_head i_devices
;
632 struct pipe_inode_info
*i_pipe
;
633 struct block_device
*i_bdev
;
639 #ifdef CONFIG_FSNOTIFY
640 __u32 i_fsnotify_mask
; /* all events this inode cares about */
641 struct hlist_head i_fsnotify_marks
;
644 void *i_private
; /* fs or device private pointer */
647 static inline int inode_unhashed(struct inode
*inode
)
649 return hlist_unhashed(&inode
->i_hash
);
653 * inode->i_mutex nesting subclasses for the lock validator:
655 * 0: the object of the current VFS operation
659 * 4: second non-directory
660 * 5: second parent (when locking independent directories in rename)
662 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
663 * non-directories at once.
665 * The locking order between these classes is
666 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
668 enum inode_i_mutex_lock_class
678 void lock_two_nondirectories(struct inode
*, struct inode
*);
679 void unlock_two_nondirectories(struct inode
*, struct inode
*);
682 * NOTE: in a 32bit arch with a preemptable kernel and
683 * an UP compile the i_size_read/write must be atomic
684 * with respect to the local cpu (unlike with preempt disabled),
685 * but they don't need to be atomic with respect to other cpus like in
686 * true SMP (so they need either to either locally disable irq around
687 * the read or for example on x86 they can be still implemented as a
688 * cmpxchg8b without the need of the lock prefix). For SMP compiles
689 * and 64bit archs it makes no difference if preempt is enabled or not.
691 static inline loff_t
i_size_read(const struct inode
*inode
)
693 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
698 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
699 i_size
= inode
->i_size
;
700 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
702 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
706 i_size
= inode
->i_size
;
710 return inode
->i_size
;
715 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
716 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
717 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
719 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
721 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
723 write_seqcount_begin(&inode
->i_size_seqcount
);
724 inode
->i_size
= i_size
;
725 write_seqcount_end(&inode
->i_size_seqcount
);
727 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
729 inode
->i_size
= i_size
;
732 inode
->i_size
= i_size
;
736 /* Helper functions so that in most cases filesystems will
737 * not need to deal directly with kuid_t and kgid_t and can
738 * instead deal with the raw numeric values that are stored
741 static inline uid_t
i_uid_read(const struct inode
*inode
)
743 return from_kuid(&init_user_ns
, inode
->i_uid
);
746 static inline gid_t
i_gid_read(const struct inode
*inode
)
748 return from_kgid(&init_user_ns
, inode
->i_gid
);
751 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
753 inode
->i_uid
= make_kuid(&init_user_ns
, uid
);
756 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
758 inode
->i_gid
= make_kgid(&init_user_ns
, gid
);
761 static inline unsigned iminor(const struct inode
*inode
)
763 return MINOR(inode
->i_rdev
);
766 static inline unsigned imajor(const struct inode
*inode
)
768 return MAJOR(inode
->i_rdev
);
771 extern struct block_device
*I_BDEV(struct inode
*inode
);
774 rwlock_t lock
; /* protects pid, uid, euid fields */
775 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
776 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
777 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
778 int signum
; /* posix.1b rt signal to be delivered on IO */
782 * Track a single file's readahead state
784 struct file_ra_state
{
785 pgoff_t start
; /* where readahead started */
786 unsigned int size
; /* # of readahead pages */
787 unsigned int async_size
; /* do asynchronous readahead when
788 there are only # of pages ahead */
790 unsigned int ra_pages
; /* Maximum readahead window */
791 unsigned int mmap_miss
; /* Cache miss stat for mmap accesses */
792 loff_t prev_pos
; /* Cache last read() position */
796 * Check if @index falls in the readahead windows.
798 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
800 return (index
>= ra
->start
&&
801 index
< ra
->start
+ ra
->size
);
806 struct llist_node fu_llist
;
807 struct rcu_head fu_rcuhead
;
810 struct inode
*f_inode
; /* cached value */
811 const struct file_operations
*f_op
;
814 * Protects f_ep_links, f_flags.
815 * Must not be taken from IRQ context.
818 atomic_long_t f_count
;
819 unsigned int f_flags
;
821 struct mutex f_pos_lock
;
823 struct fown_struct f_owner
;
824 const struct cred
*f_cred
;
825 struct file_ra_state f_ra
;
828 #ifdef CONFIG_SECURITY
831 /* needed for tty driver, and maybe others */
835 /* Used by fs/eventpoll.c to link all the hooks to this file */
836 struct list_head f_ep_links
;
837 struct list_head f_tfile_llink
;
838 #endif /* #ifdef CONFIG_EPOLL */
839 struct address_space
*f_mapping
;
840 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
845 /* file identifier */
846 unsigned char f_handle
[0];
849 static inline struct file
*get_file(struct file
*f
)
851 atomic_long_inc(&f
->f_count
);
854 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
855 #define file_count(x) atomic_long_read(&(x)->f_count)
857 #define MAX_NON_LFS ((1UL<<31) - 1)
859 /* Page cache limit. The filesystems should put that into their s_maxbytes
860 limits, otherwise bad things can happen in VM. */
861 #if BITS_PER_LONG==32
862 #define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
863 #elif BITS_PER_LONG==64
864 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
869 #define FL_DELEG 4 /* NFSv4 delegation */
870 #define FL_ACCESS 8 /* not trying to lock, just looking */
871 #define FL_EXISTS 16 /* when unlocking, test for existence */
872 #define FL_LEASE 32 /* lease held on this file */
873 #define FL_CLOSE 64 /* unlock on close */
874 #define FL_SLEEP 128 /* A blocking lock */
875 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
876 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
877 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
878 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
881 * Special return value from posix_lock_file() and vfs_lock_file() for
882 * asynchronous locking.
884 #define FILE_LOCK_DEFERRED 1
886 /* legacy typedef, should eventually be removed */
887 typedef void *fl_owner_t
;
891 struct file_lock_operations
{
892 void (*fl_copy_lock
)(struct file_lock
*, struct file_lock
*);
893 void (*fl_release_private
)(struct file_lock
*);
896 struct lock_manager_operations
{
897 int (*lm_compare_owner
)(struct file_lock
*, struct file_lock
*);
898 unsigned long (*lm_owner_key
)(struct file_lock
*);
899 void (*lm_get_owner
)(struct file_lock
*, struct file_lock
*);
900 void (*lm_put_owner
)(struct file_lock
*);
901 void (*lm_notify
)(struct file_lock
*); /* unblock callback */
902 int (*lm_grant
)(struct file_lock
*, int);
903 bool (*lm_break
)(struct file_lock
*);
904 int (*lm_change
)(struct file_lock
*, int, struct list_head
*);
905 void (*lm_setup
)(struct file_lock
*, void **);
908 struct lock_manager
{
909 struct list_head list
;
913 void locks_start_grace(struct net
*, struct lock_manager
*);
914 void locks_end_grace(struct lock_manager
*);
915 int locks_in_grace(struct net
*);
917 /* that will die - we need it for nfs_lock_info */
918 #include <linux/nfs_fs_i.h>
921 * struct file_lock represents a generic "file lock". It's used to represent
922 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
923 * note that the same struct is used to represent both a request for a lock and
924 * the lock itself, but the same object is never used for both.
926 * FIXME: should we create a separate "struct lock_request" to help distinguish
929 * The varous i_flctx lists are ordered by:
932 * 2) lock range start
935 * Obviously, the last two criteria only matter for POSIX locks.
938 struct file_lock
*fl_next
; /* singly linked list for this inode */
939 struct list_head fl_list
; /* link into file_lock_context */
940 struct hlist_node fl_link
; /* node in global lists */
941 struct list_head fl_block
; /* circular list of blocked processes */
943 unsigned int fl_flags
;
944 unsigned char fl_type
;
946 int fl_link_cpu
; /* what cpu's list is this on? */
947 struct pid
*fl_nspid
;
948 wait_queue_head_t fl_wait
;
949 struct file
*fl_file
;
953 struct fasync_struct
* fl_fasync
; /* for lease break notifications */
954 /* for lease breaks: */
955 unsigned long fl_break_time
;
956 unsigned long fl_downgrade_time
;
958 const struct file_lock_operations
*fl_ops
; /* Callbacks for filesystems */
959 const struct lock_manager_operations
*fl_lmops
; /* Callbacks for lockmanagers */
961 struct nfs_lock_info nfs_fl
;
962 struct nfs4_lock_info nfs4_fl
;
964 struct list_head link
; /* link in AFS vnode's pending_locks list */
965 int state
; /* state of grant or error if -ve */
970 struct file_lock_context
{
972 struct list_head flc_flock
;
973 struct list_head flc_posix
;
974 struct list_head flc_lease
;
980 /* The following constant reflects the upper bound of the file/locking space */
982 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
983 #define OFFSET_MAX INT_LIMIT(loff_t)
984 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
987 #include <linux/fcntl.h>
989 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
991 #ifdef CONFIG_FILE_LOCKING
992 extern int fcntl_getlk(struct file
*, unsigned int, struct flock __user
*);
993 extern int fcntl_setlk(unsigned int, struct file
*, unsigned int,
994 struct flock __user
*);
996 #if BITS_PER_LONG == 32
997 extern int fcntl_getlk64(struct file
*, unsigned int, struct flock64 __user
*);
998 extern int fcntl_setlk64(unsigned int, struct file
*, unsigned int,
999 struct flock64 __user
*);
1002 extern int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
);
1003 extern int fcntl_getlease(struct file
*filp
);
1006 void locks_free_lock_context(struct file_lock_context
*ctx
);
1007 void locks_free_lock(struct file_lock
*fl
);
1008 extern void locks_init_lock(struct file_lock
*);
1009 extern struct file_lock
* locks_alloc_lock(void);
1010 extern void locks_copy_lock(struct file_lock
*, struct file_lock
*);
1011 extern void locks_copy_conflock(struct file_lock
*, struct file_lock
*);
1012 extern void locks_remove_posix(struct file
*, fl_owner_t
);
1013 extern void locks_remove_file(struct file
*);
1014 extern void locks_release_private(struct file_lock
*);
1015 extern void posix_test_lock(struct file
*, struct file_lock
*);
1016 extern int posix_lock_file(struct file
*, struct file_lock
*, struct file_lock
*);
1017 extern int posix_lock_file_wait(struct file
*, struct file_lock
*);
1018 extern int posix_unblock_lock(struct file_lock
*);
1019 extern int vfs_test_lock(struct file
*, struct file_lock
*);
1020 extern int vfs_lock_file(struct file
*, unsigned int, struct file_lock
*, struct file_lock
*);
1021 extern int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
);
1022 extern int flock_lock_file_wait(struct file
*filp
, struct file_lock
*fl
);
1023 extern int __break_lease(struct inode
*inode
, unsigned int flags
, unsigned int type
);
1024 extern void lease_get_mtime(struct inode
*, struct timespec
*time
);
1025 extern int generic_setlease(struct file
*, long, struct file_lock
**, void **priv
);
1026 extern int vfs_setlease(struct file
*, long, struct file_lock
**, void **);
1027 extern int lease_modify(struct file_lock
*, int, struct list_head
*);
1028 #else /* !CONFIG_FILE_LOCKING */
1029 static inline int fcntl_getlk(struct file
*file
, unsigned int cmd
,
1030 struct flock __user
*user
)
1035 static inline int fcntl_setlk(unsigned int fd
, struct file
*file
,
1036 unsigned int cmd
, struct flock __user
*user
)
1041 #if BITS_PER_LONG == 32
1042 static inline int fcntl_getlk64(struct file
*file
, unsigned int cmd
,
1043 struct flock64 __user
*user
)
1048 static inline int fcntl_setlk64(unsigned int fd
, struct file
*file
,
1049 unsigned int cmd
, struct flock64 __user
*user
)
1054 static inline int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
)
1059 static inline int fcntl_getlease(struct file
*filp
)
1065 locks_free_lock_context(struct file_lock_context
*ctx
)
1069 static inline void locks_init_lock(struct file_lock
*fl
)
1074 static inline void locks_copy_conflock(struct file_lock
*new, struct file_lock
*fl
)
1079 static inline void locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
1084 static inline void locks_remove_posix(struct file
*filp
, fl_owner_t owner
)
1089 static inline void locks_remove_file(struct file
*filp
)
1094 static inline void posix_test_lock(struct file
*filp
, struct file_lock
*fl
)
1099 static inline int posix_lock_file(struct file
*filp
, struct file_lock
*fl
,
1100 struct file_lock
*conflock
)
1105 static inline int posix_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1110 static inline int posix_unblock_lock(struct file_lock
*waiter
)
1115 static inline int vfs_test_lock(struct file
*filp
, struct file_lock
*fl
)
1120 static inline int vfs_lock_file(struct file
*filp
, unsigned int cmd
,
1121 struct file_lock
*fl
, struct file_lock
*conf
)
1126 static inline int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
)
1131 static inline int flock_lock_file_wait(struct file
*filp
,
1132 struct file_lock
*request
)
1137 static inline int __break_lease(struct inode
*inode
, unsigned int mode
, unsigned int type
)
1142 static inline void lease_get_mtime(struct inode
*inode
, struct timespec
*time
)
1147 static inline int generic_setlease(struct file
*filp
, long arg
,
1148 struct file_lock
**flp
, void **priv
)
1153 static inline int vfs_setlease(struct file
*filp
, long arg
,
1154 struct file_lock
**lease
, void **priv
)
1159 static inline int lease_modify(struct file_lock
*fl
, int arg
,
1160 struct list_head
*dispose
)
1164 #endif /* !CONFIG_FILE_LOCKING */
1167 struct fasync_struct
{
1171 struct fasync_struct
*fa_next
; /* singly linked list */
1172 struct file
*fa_file
;
1173 struct rcu_head fa_rcu
;
1176 #define FASYNC_MAGIC 0x4601
1178 /* SMP safe fasync helpers: */
1179 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1180 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1181 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1182 extern struct fasync_struct
*fasync_alloc(void);
1183 extern void fasync_free(struct fasync_struct
*);
1185 /* can be called from interrupts */
1186 extern void kill_fasync(struct fasync_struct
**, int, int);
1188 extern void __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1189 extern void f_setown(struct file
*filp
, unsigned long arg
, int force
);
1190 extern void f_delown(struct file
*filp
);
1191 extern pid_t
f_getown(struct file
*filp
);
1192 extern int send_sigurg(struct fown_struct
*fown
);
1200 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1201 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1202 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1203 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1204 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1206 extern struct list_head super_blocks
;
1207 extern spinlock_t sb_lock
;
1209 /* Possible states of 'frozen' field */
1211 SB_UNFROZEN
= 0, /* FS is unfrozen */
1212 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1213 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1214 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1215 * internal threads if needed) */
1216 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1219 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1222 /* Counters for counting writers at each level */
1223 struct percpu_counter counter
[SB_FREEZE_LEVELS
];
1224 wait_queue_head_t wait
; /* queue for waiting for
1225 writers / faults to finish */
1226 int frozen
; /* Is sb frozen? */
1227 wait_queue_head_t wait_unfrozen
; /* queue for waiting for
1229 #ifdef CONFIG_DEBUG_LOCK_ALLOC
1230 struct lockdep_map lock_map
[SB_FREEZE_LEVELS
];
1234 struct super_block
{
1235 struct list_head s_list
; /* Keep this first */
1236 dev_t s_dev
; /* search index; _not_ kdev_t */
1237 unsigned char s_blocksize_bits
;
1238 unsigned long s_blocksize
;
1239 loff_t s_maxbytes
; /* Max file size */
1240 struct file_system_type
*s_type
;
1241 const struct super_operations
*s_op
;
1242 const struct dquot_operations
*dq_op
;
1243 const struct quotactl_ops
*s_qcop
;
1244 const struct export_operations
*s_export_op
;
1245 unsigned long s_flags
;
1246 unsigned long s_magic
;
1247 struct dentry
*s_root
;
1248 struct rw_semaphore s_umount
;
1251 #ifdef CONFIG_SECURITY
1254 const struct xattr_handler
**s_xattr
;
1256 struct list_head s_inodes
; /* all inodes */
1257 struct hlist_bl_head s_anon
; /* anonymous dentries for (nfs) exporting */
1258 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1259 struct block_device
*s_bdev
;
1260 struct backing_dev_info
*s_bdi
;
1261 struct mtd_info
*s_mtd
;
1262 struct hlist_node s_instances
;
1263 unsigned int s_quota_types
; /* Bitmask of supported quota types */
1264 struct quota_info s_dquot
; /* Diskquota specific options */
1266 struct sb_writers s_writers
;
1268 char s_id
[32]; /* Informational name */
1269 u8 s_uuid
[16]; /* UUID */
1271 void *s_fs_info
; /* Filesystem private info */
1272 unsigned int s_max_links
;
1275 /* Granularity of c/m/atime in ns.
1276 Cannot be worse than a second */
1280 * The next field is for VFS *only*. No filesystems have any business
1281 * even looking at it. You had been warned.
1283 struct mutex s_vfs_rename_mutex
; /* Kludge */
1286 * Filesystem subtype. If non-empty the filesystem type field
1287 * in /proc/mounts will be "type.subtype"
1292 * Saved mount options for lazy filesystems using
1293 * generic_show_options()
1295 char __rcu
*s_options
;
1296 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1299 * Saved pool identifier for cleancache (-1 means none)
1301 int cleancache_poolid
;
1303 struct shrinker s_shrink
; /* per-sb shrinker handle */
1305 /* Number of inodes with nlink == 0 but still referenced */
1306 atomic_long_t s_remove_count
;
1308 /* Being remounted read-only */
1309 int s_readonly_remount
;
1311 /* AIO completions deferred from interrupt context */
1312 struct workqueue_struct
*s_dio_done_wq
;
1313 struct hlist_head s_pins
;
1316 * Keep the lru lists last in the structure so they always sit on their
1317 * own individual cachelines.
1319 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp
;
1320 struct list_lru s_inode_lru ____cacheline_aligned_in_smp
;
1321 struct rcu_head rcu
;
1324 * Indicates how deep in a filesystem stack this SB is
1329 extern struct timespec
current_fs_time(struct super_block
*sb
);
1332 * Snapshotting support.
1335 void __sb_end_write(struct super_block
*sb
, int level
);
1336 int __sb_start_write(struct super_block
*sb
, int level
, bool wait
);
1339 * sb_end_write - drop write access to a superblock
1340 * @sb: the super we wrote to
1342 * Decrement number of writers to the filesystem. Wake up possible waiters
1343 * wanting to freeze the filesystem.
1345 static inline void sb_end_write(struct super_block
*sb
)
1347 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1351 * sb_end_pagefault - drop write access to a superblock from a page fault
1352 * @sb: the super we wrote to
1354 * Decrement number of processes handling write page fault to the filesystem.
1355 * Wake up possible waiters wanting to freeze the filesystem.
1357 static inline void sb_end_pagefault(struct super_block
*sb
)
1359 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1363 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1364 * @sb: the super we wrote to
1366 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1367 * waiters wanting to freeze the filesystem.
1369 static inline void sb_end_intwrite(struct super_block
*sb
)
1371 __sb_end_write(sb
, SB_FREEZE_FS
);
1375 * sb_start_write - get write access to a superblock
1376 * @sb: the super we write to
1378 * When a process wants to write data or metadata to a file system (i.e. dirty
1379 * a page or an inode), it should embed the operation in a sb_start_write() -
1380 * sb_end_write() pair to get exclusion against file system freezing. This
1381 * function increments number of writers preventing freezing. If the file
1382 * system is already frozen, the function waits until the file system is
1385 * Since freeze protection behaves as a lock, users have to preserve
1386 * ordering of freeze protection and other filesystem locks. Generally,
1387 * freeze protection should be the outermost lock. In particular, we have:
1390 * -> i_mutex (write path, truncate, directory ops, ...)
1391 * -> s_umount (freeze_super, thaw_super)
1393 static inline void sb_start_write(struct super_block
*sb
)
1395 __sb_start_write(sb
, SB_FREEZE_WRITE
, true);
1398 static inline int sb_start_write_trylock(struct super_block
*sb
)
1400 return __sb_start_write(sb
, SB_FREEZE_WRITE
, false);
1404 * sb_start_pagefault - get write access to a superblock from a page fault
1405 * @sb: the super we write to
1407 * When a process starts handling write page fault, it should embed the
1408 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1409 * exclusion against file system freezing. This is needed since the page fault
1410 * is going to dirty a page. This function increments number of running page
1411 * faults preventing freezing. If the file system is already frozen, the
1412 * function waits until the file system is thawed.
1414 * Since page fault freeze protection behaves as a lock, users have to preserve
1415 * ordering of freeze protection and other filesystem locks. It is advised to
1416 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1417 * handling code implies lock dependency:
1420 * -> sb_start_pagefault
1422 static inline void sb_start_pagefault(struct super_block
*sb
)
1424 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
, true);
1428 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1429 * @sb: the super we write to
1431 * This is the third level of protection against filesystem freezing. It is
1432 * free for use by a filesystem. The only requirement is that it must rank
1433 * below sb_start_pagefault.
1435 * For example filesystem can call sb_start_intwrite() when starting a
1436 * transaction which somewhat eases handling of freezing for internal sources
1437 * of filesystem changes (internal fs threads, discarding preallocation on file
1440 static inline void sb_start_intwrite(struct super_block
*sb
)
1442 __sb_start_write(sb
, SB_FREEZE_FS
, true);
1446 extern bool inode_owner_or_capable(const struct inode
*inode
);
1449 * VFS helper functions..
1451 extern int vfs_create(struct inode
*, struct dentry
*, umode_t
, bool);
1452 extern int vfs_mkdir(struct inode
*, struct dentry
*, umode_t
);
1453 extern int vfs_mknod(struct inode
*, struct dentry
*, umode_t
, dev_t
);
1454 extern int vfs_symlink(struct inode
*, struct dentry
*, const char *);
1455 extern int vfs_link(struct dentry
*, struct inode
*, struct dentry
*, struct inode
**);
1456 extern int vfs_rmdir(struct inode
*, struct dentry
*);
1457 extern int vfs_unlink(struct inode
*, struct dentry
*, struct inode
**);
1458 extern int vfs_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*, struct inode
**, unsigned int);
1459 extern int vfs_whiteout(struct inode
*, struct dentry
*);
1462 * VFS dentry helper functions.
1464 extern void dentry_unhash(struct dentry
*dentry
);
1467 * VFS file helper functions.
1469 extern void inode_init_owner(struct inode
*inode
, const struct inode
*dir
,
1472 * VFS FS_IOC_FIEMAP helper definitions.
1474 struct fiemap_extent_info
{
1475 unsigned int fi_flags
; /* Flags as passed from user */
1476 unsigned int fi_extents_mapped
; /* Number of mapped extents */
1477 unsigned int fi_extents_max
; /* Size of fiemap_extent array */
1478 struct fiemap_extent __user
*fi_extents_start
; /* Start of
1479 fiemap_extent array */
1481 int fiemap_fill_next_extent(struct fiemap_extent_info
*info
, u64 logical
,
1482 u64 phys
, u64 len
, u32 flags
);
1483 int fiemap_check_flags(struct fiemap_extent_info
*fieinfo
, u32 fs_flags
);
1488 * NOTE! These match bits 12..15 of stat.st_mode
1489 * (ie "(i_mode >> 12) & 15").
1491 #define DT_UNKNOWN 0
1502 * This is the "filldir" function type, used by readdir() to let
1503 * the kernel specify what kind of dirent layout it wants to have.
1504 * This allows the kernel to read directories into kernel space or
1505 * to have different dirent layouts depending on the binary type.
1508 typedef int (*filldir_t
)(struct dir_context
*, const char *, int, loff_t
, u64
,
1511 struct dir_context
{
1512 const filldir_t actor
;
1516 struct block_device_operations
;
1518 /* These macros are for out of kernel modules to test that
1519 * the kernel supports the unlocked_ioctl and compat_ioctl
1520 * fields in struct file_operations. */
1521 #define HAVE_COMPAT_IOCTL 1
1522 #define HAVE_UNLOCKED_IOCTL 1
1526 struct file_operations
{
1527 struct module
*owner
;
1528 loff_t (*llseek
) (struct file
*, loff_t
, int);
1529 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
1530 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
1531 ssize_t (*aio_read
) (struct kiocb
*, const struct iovec
*, unsigned long, loff_t
);
1532 ssize_t (*aio_write
) (struct kiocb
*, const struct iovec
*, unsigned long, loff_t
);
1533 ssize_t (*read_iter
) (struct kiocb
*, struct iov_iter
*);
1534 ssize_t (*write_iter
) (struct kiocb
*, struct iov_iter
*);
1535 int (*iterate
) (struct file
*, struct dir_context
*);
1536 unsigned int (*poll
) (struct file
*, struct poll_table_struct
*);
1537 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
1538 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
1539 int (*mmap
) (struct file
*, struct vm_area_struct
*);
1540 void (*mremap
)(struct file
*, struct vm_area_struct
*);
1541 int (*open
) (struct inode
*, struct file
*);
1542 int (*flush
) (struct file
*, fl_owner_t id
);
1543 int (*release
) (struct inode
*, struct file
*);
1544 int (*fsync
) (struct file
*, loff_t
, loff_t
, int datasync
);
1545 int (*aio_fsync
) (struct kiocb
*, int datasync
);
1546 int (*fasync
) (int, struct file
*, int);
1547 int (*lock
) (struct file
*, int, struct file_lock
*);
1548 ssize_t (*sendpage
) (struct file
*, struct page
*, int, size_t, loff_t
*, int);
1549 unsigned long (*get_unmapped_area
)(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1550 int (*check_flags
)(int);
1551 int (*flock
) (struct file
*, int, struct file_lock
*);
1552 ssize_t (*splice_write
)(struct pipe_inode_info
*, struct file
*, loff_t
*, size_t, unsigned int);
1553 ssize_t (*splice_read
)(struct file
*, loff_t
*, struct pipe_inode_info
*, size_t, unsigned int);
1554 int (*setlease
)(struct file
*, long, struct file_lock
**, void **);
1555 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
1557 void (*show_fdinfo
)(struct seq_file
*m
, struct file
*f
);
1560 struct inode_operations
{
1561 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*, unsigned int);
1562 void * (*follow_link
) (struct dentry
*, struct nameidata
*);
1563 int (*permission
) (struct inode
*, int);
1564 struct posix_acl
* (*get_acl
)(struct inode
*, int);
1566 int (*readlink
) (struct dentry
*, char __user
*,int);
1567 void (*put_link
) (struct dentry
*, struct nameidata
*, void *);
1569 int (*create
) (struct inode
*,struct dentry
*, umode_t
, bool);
1570 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
1571 int (*unlink
) (struct inode
*,struct dentry
*);
1572 int (*symlink
) (struct inode
*,struct dentry
*,const char *);
1573 int (*mkdir
) (struct inode
*,struct dentry
*,umode_t
);
1574 int (*rmdir
) (struct inode
*,struct dentry
*);
1575 int (*mknod
) (struct inode
*,struct dentry
*,umode_t
,dev_t
);
1576 int (*rename
) (struct inode
*, struct dentry
*,
1577 struct inode
*, struct dentry
*);
1578 int (*rename2
) (struct inode
*, struct dentry
*,
1579 struct inode
*, struct dentry
*, unsigned int);
1580 int (*setattr
) (struct dentry
*, struct iattr
*);
1581 int (*getattr
) (struct vfsmount
*mnt
, struct dentry
*, struct kstat
*);
1582 int (*setxattr
) (struct dentry
*, const char *,const void *,size_t,int);
1583 ssize_t (*getxattr
) (struct dentry
*, const char *, void *, size_t);
1584 ssize_t (*listxattr
) (struct dentry
*, char *, size_t);
1585 int (*removexattr
) (struct dentry
*, const char *);
1586 int (*fiemap
)(struct inode
*, struct fiemap_extent_info
*, u64 start
,
1588 int (*update_time
)(struct inode
*, struct timespec
*, int);
1589 int (*atomic_open
)(struct inode
*, struct dentry
*,
1590 struct file
*, unsigned open_flag
,
1591 umode_t create_mode
, int *opened
);
1592 int (*tmpfile
) (struct inode
*, struct dentry
*, umode_t
);
1593 int (*set_acl
)(struct inode
*, struct posix_acl
*, int);
1595 /* WARNING: probably going away soon, do not use! */
1596 int (*dentry_open
)(struct dentry
*, struct file
*, const struct cred
*);
1597 } ____cacheline_aligned
;
1599 ssize_t
rw_copy_check_uvector(int type
, const struct iovec __user
* uvector
,
1600 unsigned long nr_segs
, unsigned long fast_segs
,
1601 struct iovec
*fast_pointer
,
1602 struct iovec
**ret_pointer
);
1604 extern ssize_t
__vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1605 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1606 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1607 extern ssize_t
vfs_readv(struct file
*, const struct iovec __user
*,
1608 unsigned long, loff_t
*);
1609 extern ssize_t
vfs_writev(struct file
*, const struct iovec __user
*,
1610 unsigned long, loff_t
*);
1612 struct super_operations
{
1613 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
1614 void (*destroy_inode
)(struct inode
*);
1616 void (*dirty_inode
) (struct inode
*, int flags
);
1617 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
1618 int (*drop_inode
) (struct inode
*);
1619 void (*evict_inode
) (struct inode
*);
1620 void (*put_super
) (struct super_block
*);
1621 int (*sync_fs
)(struct super_block
*sb
, int wait
);
1622 int (*freeze_super
) (struct super_block
*);
1623 int (*freeze_fs
) (struct super_block
*);
1624 int (*thaw_super
) (struct super_block
*);
1625 int (*unfreeze_fs
) (struct super_block
*);
1626 int (*statfs
) (struct dentry
*, struct kstatfs
*);
1627 int (*remount_fs
) (struct super_block
*, int *, char *);
1628 void (*umount_begin
) (struct super_block
*);
1630 int (*show_options
)(struct seq_file
*, struct dentry
*);
1631 int (*show_devname
)(struct seq_file
*, struct dentry
*);
1632 int (*show_path
)(struct seq_file
*, struct dentry
*);
1633 int (*show_stats
)(struct seq_file
*, struct dentry
*);
1635 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
1636 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
1637 struct dquot
**(*get_dquots
)(struct inode
*);
1639 int (*bdev_try_to_free_page
)(struct super_block
*, struct page
*, gfp_t
);
1640 long (*nr_cached_objects
)(struct super_block
*, int);
1641 long (*free_cached_objects
)(struct super_block
*, long, int);
1645 * Inode flags - they have no relation to superblock flags now
1647 #define S_SYNC 1 /* Writes are synced at once */
1648 #define S_NOATIME 2 /* Do not update access times */
1649 #define S_APPEND 4 /* Append-only file */
1650 #define S_IMMUTABLE 8 /* Immutable file */
1651 #define S_DEAD 16 /* removed, but still open directory */
1652 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1653 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1654 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1655 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1656 #define S_PRIVATE 512 /* Inode is fs-internal */
1657 #define S_IMA 1024 /* Inode has an associated IMA struct */
1658 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1659 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1662 * Note that nosuid etc flags are inode-specific: setting some file-system
1663 * flags just means all the inodes inherit those flags by default. It might be
1664 * possible to override it selectively if you really wanted to with some
1665 * ioctl() that is not currently implemented.
1667 * Exception: MS_RDONLY is always applied to the entire file system.
1669 * Unfortunately, it is possible to change a filesystems flags with it mounted
1670 * with files in use. This means that all of the inodes will not have their
1671 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1672 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1674 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1676 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1677 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1678 ((inode)->i_flags & S_SYNC))
1679 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1680 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1681 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1682 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1683 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1685 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1686 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1687 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1688 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1690 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1691 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1692 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1693 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1694 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1695 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1696 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1698 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1699 (inode)->i_rdev == WHITEOUT_DEV)
1702 * Inode state bits. Protected by inode->i_lock
1704 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1705 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1707 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1708 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1709 * various stages of removing an inode.
1711 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1713 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1714 * fdatasync(). i_atime is the usual cause.
1715 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1716 * these changes separately from I_DIRTY_SYNC so that we
1717 * don't have to write inode on fdatasync() when only
1718 * mtime has changed in it.
1719 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1720 * I_NEW Serves as both a mutex and completion notification.
1721 * New inodes set I_NEW. If two processes both create
1722 * the same inode, one of them will release its inode and
1723 * wait for I_NEW to be released before returning.
1724 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1725 * also cause waiting on I_NEW, without I_NEW actually
1726 * being set. find_inode() uses this to prevent returning
1727 * nearly-dead inodes.
1728 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1729 * is zero. I_FREEING must be set when I_WILL_FREE is
1731 * I_FREEING Set when inode is about to be freed but still has dirty
1732 * pages or buffers attached or the inode itself is still
1734 * I_CLEAR Added by clear_inode(). In this state the inode is
1735 * clean and can be destroyed. Inode keeps I_FREEING.
1737 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1738 * prohibited for many purposes. iget() must wait for
1739 * the inode to be completely released, then create it
1740 * anew. Other functions will just ignore such inodes,
1741 * if appropriate. I_NEW is used for waiting.
1743 * I_SYNC Writeback of inode is running. The bit is set during
1744 * data writeback, and cleared with a wakeup on the bit
1745 * address once it is done. The bit is also used to pin
1746 * the inode in memory for flusher thread.
1748 * I_REFERENCED Marks the inode as recently references on the LRU list.
1750 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1752 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1754 #define I_DIRTY_SYNC (1 << 0)
1755 #define I_DIRTY_DATASYNC (1 << 1)
1756 #define I_DIRTY_PAGES (1 << 2)
1758 #define I_NEW (1 << __I_NEW)
1759 #define I_WILL_FREE (1 << 4)
1760 #define I_FREEING (1 << 5)
1761 #define I_CLEAR (1 << 6)
1763 #define I_SYNC (1 << __I_SYNC)
1764 #define I_REFERENCED (1 << 8)
1765 #define __I_DIO_WAKEUP 9
1766 #define I_DIO_WAKEUP (1 << I_DIO_WAKEUP)
1767 #define I_LINKABLE (1 << 10)
1769 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1771 extern void __mark_inode_dirty(struct inode
*, int);
1772 static inline void mark_inode_dirty(struct inode
*inode
)
1774 __mark_inode_dirty(inode
, I_DIRTY
);
1777 static inline void mark_inode_dirty_sync(struct inode
*inode
)
1779 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
1782 extern void inc_nlink(struct inode
*inode
);
1783 extern void drop_nlink(struct inode
*inode
);
1784 extern void clear_nlink(struct inode
*inode
);
1785 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
1787 static inline void inode_inc_link_count(struct inode
*inode
)
1790 mark_inode_dirty(inode
);
1793 static inline void inode_dec_link_count(struct inode
*inode
)
1796 mark_inode_dirty(inode
);
1800 * inode_inc_iversion - increments i_version
1801 * @inode: inode that need to be updated
1803 * Every time the inode is modified, the i_version field will be incremented.
1804 * The filesystem has to be mounted with i_version flag
1807 static inline void inode_inc_iversion(struct inode
*inode
)
1809 spin_lock(&inode
->i_lock
);
1811 spin_unlock(&inode
->i_lock
);
1814 enum file_time_flags
{
1821 extern void touch_atime(const struct path
*);
1822 static inline void file_accessed(struct file
*file
)
1824 if (!(file
->f_flags
& O_NOATIME
))
1825 touch_atime(&file
->f_path
);
1828 int sync_inode(struct inode
*inode
, struct writeback_control
*wbc
);
1829 int sync_inode_metadata(struct inode
*inode
, int wait
);
1831 struct file_system_type
{
1834 #define FS_REQUIRES_DEV 1
1835 #define FS_BINARY_MOUNTDATA 2
1836 #define FS_HAS_SUBTYPE 4
1837 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
1838 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
1839 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
1840 struct dentry
*(*mount
) (struct file_system_type
*, int,
1841 const char *, void *);
1842 void (*kill_sb
) (struct super_block
*);
1843 struct module
*owner
;
1844 struct file_system_type
* next
;
1845 struct hlist_head fs_supers
;
1847 struct lock_class_key s_lock_key
;
1848 struct lock_class_key s_umount_key
;
1849 struct lock_class_key s_vfs_rename_key
;
1850 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
1852 struct lock_class_key i_lock_key
;
1853 struct lock_class_key i_mutex_key
;
1854 struct lock_class_key i_mutex_dir_key
;
1857 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1859 extern struct dentry
*mount_ns(struct file_system_type
*fs_type
, int flags
,
1860 void *data
, int (*fill_super
)(struct super_block
*, void *, int));
1861 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
1862 int flags
, const char *dev_name
, void *data
,
1863 int (*fill_super
)(struct super_block
*, void *, int));
1864 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
1865 int flags
, void *data
,
1866 int (*fill_super
)(struct super_block
*, void *, int));
1867 extern struct dentry
*mount_nodev(struct file_system_type
*fs_type
,
1868 int flags
, void *data
,
1869 int (*fill_super
)(struct super_block
*, void *, int));
1870 extern struct dentry
*mount_subtree(struct vfsmount
*mnt
, const char *path
);
1871 void generic_shutdown_super(struct super_block
*sb
);
1872 void kill_block_super(struct super_block
*sb
);
1873 void kill_anon_super(struct super_block
*sb
);
1874 void kill_litter_super(struct super_block
*sb
);
1875 void deactivate_super(struct super_block
*sb
);
1876 void deactivate_locked_super(struct super_block
*sb
);
1877 int set_anon_super(struct super_block
*s
, void *data
);
1878 int get_anon_bdev(dev_t
*);
1879 void free_anon_bdev(dev_t
);
1880 struct super_block
*sget(struct file_system_type
*type
,
1881 int (*test
)(struct super_block
*,void *),
1882 int (*set
)(struct super_block
*,void *),
1883 int flags
, void *data
);
1884 extern struct dentry
*mount_pseudo(struct file_system_type
*, char *,
1885 const struct super_operations
*ops
,
1886 const struct dentry_operations
*dops
,
1889 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1890 #define fops_get(fops) \
1891 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1892 #define fops_put(fops) \
1893 do { if (fops) module_put((fops)->owner); } while(0)
1895 * This one is to be used *ONLY* from ->open() instances.
1896 * fops must be non-NULL, pinned down *and* module dependencies
1897 * should be sufficient to pin the caller down as well.
1899 #define replace_fops(f, fops) \
1901 struct file *__file = (f); \
1902 fops_put(__file->f_op); \
1903 BUG_ON(!(__file->f_op = (fops))); \
1906 extern int register_filesystem(struct file_system_type
*);
1907 extern int unregister_filesystem(struct file_system_type
*);
1908 extern struct vfsmount
*kern_mount_data(struct file_system_type
*, void *data
);
1909 #define kern_mount(type) kern_mount_data(type, NULL)
1910 extern void kern_unmount(struct vfsmount
*mnt
);
1911 extern int may_umount_tree(struct vfsmount
*);
1912 extern int may_umount(struct vfsmount
*);
1913 extern long do_mount(const char *, const char __user
*,
1914 const char *, unsigned long, void *);
1915 extern struct vfsmount
*collect_mounts(struct path
*);
1916 extern void drop_collected_mounts(struct vfsmount
*);
1917 extern int iterate_mounts(int (*)(struct vfsmount
*, void *), void *,
1919 extern int vfs_statfs(struct path
*, struct kstatfs
*);
1920 extern int user_statfs(const char __user
*, struct kstatfs
*);
1921 extern int fd_statfs(int, struct kstatfs
*);
1922 extern int vfs_ustat(dev_t
, struct kstatfs
*);
1923 extern int freeze_super(struct super_block
*super
);
1924 extern int thaw_super(struct super_block
*super
);
1925 extern bool our_mnt(struct vfsmount
*mnt
);
1926 extern bool fs_fully_visible(struct file_system_type
*);
1928 extern int current_umask(void);
1930 extern void ihold(struct inode
* inode
);
1931 extern void iput(struct inode
*);
1933 static inline struct inode
*file_inode(const struct file
*f
)
1939 extern struct kobject
*fs_kobj
;
1941 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1943 #define FLOCK_VERIFY_READ 1
1944 #define FLOCK_VERIFY_WRITE 2
1946 #ifdef CONFIG_FILE_LOCKING
1947 extern int locks_mandatory_locked(struct file
*);
1948 extern int locks_mandatory_area(int, struct inode
*, struct file
*, loff_t
, size_t);
1951 * Candidates for mandatory locking have the setgid bit set
1952 * but no group execute bit - an otherwise meaningless combination.
1955 static inline int __mandatory_lock(struct inode
*ino
)
1957 return (ino
->i_mode
& (S_ISGID
| S_IXGRP
)) == S_ISGID
;
1961 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1962 * otherwise these will be advisory locks
1965 static inline int mandatory_lock(struct inode
*ino
)
1967 return IS_MANDLOCK(ino
) && __mandatory_lock(ino
);
1970 static inline int locks_verify_locked(struct file
*file
)
1972 if (mandatory_lock(file_inode(file
)))
1973 return locks_mandatory_locked(file
);
1977 static inline int locks_verify_truncate(struct inode
*inode
,
1981 if (inode
->i_flctx
&& mandatory_lock(inode
))
1982 return locks_mandatory_area(
1983 FLOCK_VERIFY_WRITE
, inode
, filp
,
1984 size
< inode
->i_size
? size
: inode
->i_size
,
1985 (size
< inode
->i_size
? inode
->i_size
- size
1986 : size
- inode
->i_size
)
1991 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
1994 * Since this check is lockless, we must ensure that any refcounts
1995 * taken are done before checking i_flctx->flc_lease. Otherwise, we
1996 * could end up racing with tasks trying to set a new lease on this
2000 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2001 return __break_lease(inode
, mode
, FL_LEASE
);
2005 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2008 * Since this check is lockless, we must ensure that any refcounts
2009 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2010 * could end up racing with tasks trying to set a new lease on this
2014 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2015 return __break_lease(inode
, mode
, FL_DELEG
);
2019 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2023 ret
= break_deleg(inode
, O_WRONLY
|O_NONBLOCK
);
2024 if (ret
== -EWOULDBLOCK
&& delegated_inode
) {
2025 *delegated_inode
= inode
;
2031 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2035 ret
= break_deleg(*delegated_inode
, O_WRONLY
);
2036 iput(*delegated_inode
);
2037 *delegated_inode
= NULL
;
2041 static inline int break_layout(struct inode
*inode
, bool wait
)
2044 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2045 return __break_lease(inode
,
2046 wait
? O_WRONLY
: O_WRONLY
| O_NONBLOCK
,
2051 #else /* !CONFIG_FILE_LOCKING */
2052 static inline int locks_mandatory_locked(struct file
*file
)
2057 static inline int locks_mandatory_area(int rw
, struct inode
*inode
,
2058 struct file
*filp
, loff_t offset
,
2064 static inline int __mandatory_lock(struct inode
*inode
)
2069 static inline int mandatory_lock(struct inode
*inode
)
2074 static inline int locks_verify_locked(struct file
*file
)
2079 static inline int locks_verify_truncate(struct inode
*inode
, struct file
*filp
,
2085 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2090 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2095 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2100 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2106 static inline int break_layout(struct inode
*inode
, bool wait
)
2111 #endif /* CONFIG_FILE_LOCKING */
2116 const char *name
; /* pointer to actual string */
2117 const __user
char *uptr
; /* original userland pointer */
2118 struct audit_names
*aname
;
2119 bool separate
; /* should "name" be freed? */
2122 extern long vfs_truncate(struct path
*, loff_t
);
2123 extern int do_truncate(struct dentry
*, loff_t start
, unsigned int time_attrs
,
2125 extern int vfs_fallocate(struct file
*file
, int mode
, loff_t offset
,
2127 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
2129 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
2130 extern struct file
*filp_open(const char *, int, umode_t
);
2131 extern struct file
*file_open_root(struct dentry
*, struct vfsmount
*,
2133 extern int vfs_open(const struct path
*, struct file
*, const struct cred
*);
2134 extern struct file
* dentry_open(const struct path
*, int, const struct cred
*);
2135 extern int filp_close(struct file
*, fl_owner_t id
);
2137 extern struct filename
*getname_flags(const char __user
*, int, int *);
2138 extern struct filename
*getname(const char __user
*);
2139 extern struct filename
*getname_kernel(const char *);
2145 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
2146 int (*open
)(struct inode
*, struct file
*),
2148 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
2152 extern int ioctl_preallocate(struct file
*filp
, void __user
*argp
);
2155 extern void __init
vfs_caches_init_early(void);
2156 extern void __init
vfs_caches_init(unsigned long);
2158 extern struct kmem_cache
*names_cachep
;
2160 extern void final_putname(struct filename
*name
);
2162 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2163 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2164 #ifndef CONFIG_AUDITSYSCALL
2165 #define putname(name) final_putname(name)
2167 extern void putname(struct filename
*name
);
2171 extern int register_blkdev(unsigned int, const char *);
2172 extern void unregister_blkdev(unsigned int, const char *);
2173 extern struct block_device
*bdget(dev_t
);
2174 extern struct block_device
*bdgrab(struct block_device
*bdev
);
2175 extern void bd_set_size(struct block_device
*, loff_t size
);
2176 extern void bd_forget(struct inode
*inode
);
2177 extern void bdput(struct block_device
*);
2178 extern void invalidate_bdev(struct block_device
*);
2179 extern void iterate_bdevs(void (*)(struct block_device
*, void *), void *);
2180 extern int sync_blockdev(struct block_device
*bdev
);
2181 extern void kill_bdev(struct block_device
*);
2182 extern struct super_block
*freeze_bdev(struct block_device
*);
2183 extern void emergency_thaw_all(void);
2184 extern int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
);
2185 extern int fsync_bdev(struct block_device
*);
2186 extern int sb_is_blkdev_sb(struct super_block
*sb
);
2188 static inline void bd_forget(struct inode
*inode
) {}
2189 static inline int sync_blockdev(struct block_device
*bdev
) { return 0; }
2190 static inline void kill_bdev(struct block_device
*bdev
) {}
2191 static inline void invalidate_bdev(struct block_device
*bdev
) {}
2193 static inline struct super_block
*freeze_bdev(struct block_device
*sb
)
2198 static inline int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
)
2203 static inline void iterate_bdevs(void (*f
)(struct block_device
*, void *), void *arg
)
2207 static inline int sb_is_blkdev_sb(struct super_block
*sb
)
2212 extern int sync_filesystem(struct super_block
*);
2213 extern const struct file_operations def_blk_fops
;
2214 extern const struct file_operations def_chr_fops
;
2216 extern int ioctl_by_bdev(struct block_device
*, unsigned, unsigned long);
2217 extern int blkdev_ioctl(struct block_device
*, fmode_t
, unsigned, unsigned long);
2218 extern long compat_blkdev_ioctl(struct file
*, unsigned, unsigned long);
2219 extern int blkdev_get(struct block_device
*bdev
, fmode_t mode
, void *holder
);
2220 extern struct block_device
*blkdev_get_by_path(const char *path
, fmode_t mode
,
2222 extern struct block_device
*blkdev_get_by_dev(dev_t dev
, fmode_t mode
,
2224 extern void blkdev_put(struct block_device
*bdev
, fmode_t mode
);
2226 extern int bd_link_disk_holder(struct block_device
*bdev
, struct gendisk
*disk
);
2227 extern void bd_unlink_disk_holder(struct block_device
*bdev
,
2228 struct gendisk
*disk
);
2230 static inline int bd_link_disk_holder(struct block_device
*bdev
,
2231 struct gendisk
*disk
)
2235 static inline void bd_unlink_disk_holder(struct block_device
*bdev
,
2236 struct gendisk
*disk
)
2243 #define CHRDEV_MAJOR_HASH_SIZE 255
2244 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
2245 extern int register_chrdev_region(dev_t
, unsigned, const char *);
2246 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
2247 unsigned int count
, const char *name
,
2248 const struct file_operations
*fops
);
2249 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
2250 unsigned int count
, const char *name
);
2251 extern void unregister_chrdev_region(dev_t
, unsigned);
2252 extern void chrdev_show(struct seq_file
*,off_t
);
2254 static inline int register_chrdev(unsigned int major
, const char *name
,
2255 const struct file_operations
*fops
)
2257 return __register_chrdev(major
, 0, 256, name
, fops
);
2260 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2262 __unregister_chrdev(major
, 0, 256, name
);
2265 /* fs/block_dev.c */
2266 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2267 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2270 #define BLKDEV_MAJOR_HASH_SIZE 255
2271 extern const char *__bdevname(dev_t
, char *buffer
);
2272 extern const char *bdevname(struct block_device
*bdev
, char *buffer
);
2273 extern struct block_device
*lookup_bdev(const char *);
2274 extern void blkdev_show(struct seq_file
*,off_t
);
2277 #define BLKDEV_MAJOR_HASH_SIZE 0
2280 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2282 /* Invalid inode operations -- fs/bad_inode.c */
2283 extern void make_bad_inode(struct inode
*);
2284 extern int is_bad_inode(struct inode
*);
2288 * return READ, READA, or WRITE
2290 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2293 * return data direction, READ or WRITE
2295 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2297 extern void check_disk_size_change(struct gendisk
*disk
,
2298 struct block_device
*bdev
);
2299 extern int revalidate_disk(struct gendisk
*);
2300 extern int check_disk_change(struct block_device
*);
2301 extern int __invalidate_device(struct block_device
*, bool);
2302 extern int invalidate_partition(struct gendisk
*, int);
2304 unsigned long invalidate_mapping_pages(struct address_space
*mapping
,
2305 pgoff_t start
, pgoff_t end
);
2307 static inline void invalidate_remote_inode(struct inode
*inode
)
2309 if (S_ISREG(inode
->i_mode
) || S_ISDIR(inode
->i_mode
) ||
2310 S_ISLNK(inode
->i_mode
))
2311 invalidate_mapping_pages(inode
->i_mapping
, 0, -1);
2313 extern int invalidate_inode_pages2(struct address_space
*mapping
);
2314 extern int invalidate_inode_pages2_range(struct address_space
*mapping
,
2315 pgoff_t start
, pgoff_t end
);
2316 extern int write_inode_now(struct inode
*, int);
2317 extern int filemap_fdatawrite(struct address_space
*);
2318 extern int filemap_flush(struct address_space
*);
2319 extern int filemap_fdatawait(struct address_space
*);
2320 extern int filemap_fdatawait_range(struct address_space
*, loff_t lstart
,
2322 extern int filemap_write_and_wait(struct address_space
*mapping
);
2323 extern int filemap_write_and_wait_range(struct address_space
*mapping
,
2324 loff_t lstart
, loff_t lend
);
2325 extern int __filemap_fdatawrite_range(struct address_space
*mapping
,
2326 loff_t start
, loff_t end
, int sync_mode
);
2327 extern int filemap_fdatawrite_range(struct address_space
*mapping
,
2328 loff_t start
, loff_t end
);
2330 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
2332 extern int vfs_fsync(struct file
*file
, int datasync
);
2333 static inline int generic_write_sync(struct file
*file
, loff_t pos
, loff_t count
)
2335 if (!(file
->f_flags
& O_DSYNC
) && !IS_SYNC(file
->f_mapping
->host
))
2337 return vfs_fsync_range(file
, pos
, pos
+ count
- 1,
2338 (file
->f_flags
& __O_SYNC
) ? 0 : 1);
2340 extern void emergency_sync(void);
2341 extern void emergency_remount(void);
2343 extern sector_t
bmap(struct inode
*, sector_t
);
2345 extern int notify_change(struct dentry
*, struct iattr
*, struct inode
**);
2346 extern int inode_permission(struct inode
*, int);
2347 extern int __inode_permission(struct inode
*, int);
2348 extern int generic_permission(struct inode
*, int);
2349 extern int __check_sticky(struct inode
*dir
, struct inode
*inode
);
2351 static inline bool execute_ok(struct inode
*inode
)
2353 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
2356 static inline void file_start_write(struct file
*file
)
2358 if (!S_ISREG(file_inode(file
)->i_mode
))
2360 __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, true);
2363 static inline bool file_start_write_trylock(struct file
*file
)
2365 if (!S_ISREG(file_inode(file
)->i_mode
))
2367 return __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, false);
2370 static inline void file_end_write(struct file
*file
)
2372 if (!S_ISREG(file_inode(file
)->i_mode
))
2374 __sb_end_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
);
2378 * get_write_access() gets write permission for a file.
2379 * put_write_access() releases this write permission.
2380 * This is used for regular files.
2381 * We cannot support write (and maybe mmap read-write shared) accesses and
2382 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2383 * can have the following values:
2384 * 0: no writers, no VM_DENYWRITE mappings
2385 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2386 * > 0: (i_writecount) users are writing to the file.
2388 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2389 * except for the cases where we don't hold i_writecount yet. Then we need to
2390 * use {get,deny}_write_access() - these functions check the sign and refuse
2391 * to do the change if sign is wrong.
2393 static inline int get_write_access(struct inode
*inode
)
2395 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2397 static inline int deny_write_access(struct file
*file
)
2399 struct inode
*inode
= file_inode(file
);
2400 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2402 static inline void put_write_access(struct inode
* inode
)
2404 atomic_dec(&inode
->i_writecount
);
2406 static inline void allow_write_access(struct file
*file
)
2409 atomic_inc(&file_inode(file
)->i_writecount
);
2411 static inline bool inode_is_open_for_write(const struct inode
*inode
)
2413 return atomic_read(&inode
->i_writecount
) > 0;
2417 static inline void i_readcount_dec(struct inode
*inode
)
2419 BUG_ON(!atomic_read(&inode
->i_readcount
));
2420 atomic_dec(&inode
->i_readcount
);
2422 static inline void i_readcount_inc(struct inode
*inode
)
2424 atomic_inc(&inode
->i_readcount
);
2427 static inline void i_readcount_dec(struct inode
*inode
)
2431 static inline void i_readcount_inc(struct inode
*inode
)
2436 extern int do_pipe_flags(int *, int);
2438 extern int kernel_read(struct file
*, loff_t
, char *, unsigned long);
2439 extern ssize_t
kernel_write(struct file
*, const char *, size_t, loff_t
);
2440 extern ssize_t
__kernel_write(struct file
*, const char *, size_t, loff_t
*);
2441 extern struct file
* open_exec(const char *);
2443 /* fs/dcache.c -- generic fs support functions */
2444 extern int is_subdir(struct dentry
*, struct dentry
*);
2445 extern int path_is_under(struct path
*, struct path
*);
2447 #include <linux/err.h>
2449 /* needed for stackable file system support */
2450 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int whence
);
2452 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int whence
);
2454 extern int inode_init_always(struct super_block
*, struct inode
*);
2455 extern void inode_init_once(struct inode
*);
2456 extern void address_space_init_once(struct address_space
*mapping
);
2457 extern struct inode
* igrab(struct inode
*);
2458 extern ino_t
iunique(struct super_block
*, ino_t
);
2459 extern int inode_needs_sync(struct inode
*inode
);
2460 extern int generic_delete_inode(struct inode
*inode
);
2461 static inline int generic_drop_inode(struct inode
*inode
)
2463 return !inode
->i_nlink
|| inode_unhashed(inode
);
2466 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
2467 unsigned long hashval
, int (*test
)(struct inode
*, void *),
2469 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
2470 int (*test
)(struct inode
*, void *), void *data
);
2471 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
2473 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
2474 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
2475 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
2476 extern int insert_inode_locked(struct inode
*);
2477 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2478 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
2480 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
2482 extern void unlock_new_inode(struct inode
*);
2483 extern unsigned int get_next_ino(void);
2485 extern void __iget(struct inode
* inode
);
2486 extern void iget_failed(struct inode
*);
2487 extern void clear_inode(struct inode
*);
2488 extern void __destroy_inode(struct inode
*);
2489 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
2490 extern struct inode
*new_inode(struct super_block
*sb
);
2491 extern void free_inode_nonrcu(struct inode
*inode
);
2492 extern int should_remove_suid(struct dentry
*);
2493 extern int file_remove_suid(struct file
*);
2495 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
2496 static inline void insert_inode_hash(struct inode
*inode
)
2498 __insert_inode_hash(inode
, inode
->i_ino
);
2501 extern void __remove_inode_hash(struct inode
*);
2502 static inline void remove_inode_hash(struct inode
*inode
)
2504 if (!inode_unhashed(inode
))
2505 __remove_inode_hash(inode
);
2508 extern void inode_sb_list_add(struct inode
*inode
);
2511 extern void submit_bio(int, struct bio
*);
2512 extern int bdev_read_only(struct block_device
*);
2514 extern int set_blocksize(struct block_device
*, int);
2515 extern int sb_set_blocksize(struct super_block
*, int);
2516 extern int sb_min_blocksize(struct super_block
*, int);
2518 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
2519 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
2520 extern int generic_file_remap_pages(struct vm_area_struct
*, unsigned long addr
,
2521 unsigned long size
, pgoff_t pgoff
);
2522 int generic_write_checks(struct file
*file
, loff_t
*pos
, size_t *count
, int isblk
);
2523 extern ssize_t
generic_file_read_iter(struct kiocb
*, struct iov_iter
*);
2524 extern ssize_t
__generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2525 extern ssize_t
generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2526 extern ssize_t
generic_file_direct_write(struct kiocb
*, struct iov_iter
*, loff_t
);
2527 extern ssize_t
generic_perform_write(struct file
*, struct iov_iter
*, loff_t
);
2528 extern ssize_t
do_sync_read(struct file
*filp
, char __user
*buf
, size_t len
, loff_t
*ppos
);
2529 extern ssize_t
do_sync_write(struct file
*filp
, const char __user
*buf
, size_t len
, loff_t
*ppos
);
2530 extern ssize_t
new_sync_read(struct file
*filp
, char __user
*buf
, size_t len
, loff_t
*ppos
);
2531 extern ssize_t
new_sync_write(struct file
*filp
, const char __user
*buf
, size_t len
, loff_t
*ppos
);
2533 /* fs/block_dev.c */
2534 extern ssize_t
blkdev_read_iter(struct kiocb
*iocb
, struct iov_iter
*to
);
2535 extern ssize_t
blkdev_write_iter(struct kiocb
*iocb
, struct iov_iter
*from
);
2536 extern int blkdev_fsync(struct file
*filp
, loff_t start
, loff_t end
,
2538 extern void block_sync_page(struct page
*page
);
2541 extern ssize_t
generic_file_splice_read(struct file
*, loff_t
*,
2542 struct pipe_inode_info
*, size_t, unsigned int);
2543 extern ssize_t
default_file_splice_read(struct file
*, loff_t
*,
2544 struct pipe_inode_info
*, size_t, unsigned int);
2545 extern ssize_t
iter_file_splice_write(struct pipe_inode_info
*,
2546 struct file
*, loff_t
*, size_t, unsigned int);
2547 extern ssize_t
generic_splice_sendpage(struct pipe_inode_info
*pipe
,
2548 struct file
*out
, loff_t
*, size_t len
, unsigned int flags
);
2549 extern long do_splice_direct(struct file
*in
, loff_t
*ppos
, struct file
*out
,
2550 loff_t
*opos
, size_t len
, unsigned int flags
);
2554 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
2555 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int whence
);
2556 extern loff_t
no_llseek(struct file
*file
, loff_t offset
, int whence
);
2557 extern loff_t
vfs_setpos(struct file
*file
, loff_t offset
, loff_t maxsize
);
2558 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int whence
);
2559 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
2560 int whence
, loff_t maxsize
, loff_t eof
);
2561 extern loff_t
fixed_size_llseek(struct file
*file
, loff_t offset
,
2562 int whence
, loff_t size
);
2563 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
2564 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
2566 #ifdef CONFIG_FS_XIP
2567 extern ssize_t
xip_file_read(struct file
*filp
, char __user
*buf
, size_t len
,
2569 extern int xip_file_mmap(struct file
* file
, struct vm_area_struct
* vma
);
2570 extern ssize_t
xip_file_write(struct file
*filp
, const char __user
*buf
,
2571 size_t len
, loff_t
*ppos
);
2572 extern int xip_truncate_page(struct address_space
*mapping
, loff_t from
);
2574 static inline int xip_truncate_page(struct address_space
*mapping
, loff_t from
)
2581 typedef void (dio_submit_t
)(int rw
, struct bio
*bio
, struct inode
*inode
,
2582 loff_t file_offset
);
2585 /* need locking between buffered and direct access */
2588 /* filesystem does not support filling holes */
2589 DIO_SKIP_HOLES
= 0x02,
2591 /* filesystem can handle aio writes beyond i_size */
2592 DIO_ASYNC_EXTEND
= 0x04,
2595 void dio_end_io(struct bio
*bio
, int error
);
2597 ssize_t
__blockdev_direct_IO(int rw
, struct kiocb
*iocb
, struct inode
*inode
,
2598 struct block_device
*bdev
, struct iov_iter
*iter
, loff_t offset
,
2599 get_block_t get_block
, dio_iodone_t end_io
,
2600 dio_submit_t submit_io
, int flags
);
2602 static inline ssize_t
blockdev_direct_IO(int rw
, struct kiocb
*iocb
,
2603 struct inode
*inode
, struct iov_iter
*iter
, loff_t offset
,
2604 get_block_t get_block
)
2606 return __blockdev_direct_IO(rw
, iocb
, inode
, inode
->i_sb
->s_bdev
, iter
,
2607 offset
, get_block
, NULL
, NULL
,
2608 DIO_LOCKING
| DIO_SKIP_HOLES
);
2612 void inode_dio_wait(struct inode
*inode
);
2613 void inode_dio_done(struct inode
*inode
);
2615 extern void inode_set_flags(struct inode
*inode
, unsigned int flags
,
2618 extern const struct file_operations generic_ro_fops
;
2620 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2622 extern int readlink_copy(char __user
*, int, const char *);
2623 extern int page_readlink(struct dentry
*, char __user
*, int);
2624 extern void *page_follow_link_light(struct dentry
*, struct nameidata
*);
2625 extern void page_put_link(struct dentry
*, struct nameidata
*, void *);
2626 extern int __page_symlink(struct inode
*inode
, const char *symname
, int len
,
2628 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
2629 extern const struct inode_operations page_symlink_inode_operations
;
2630 extern void kfree_put_link(struct dentry
*, struct nameidata
*, void *);
2631 extern int generic_readlink(struct dentry
*, char __user
*, int);
2632 extern void generic_fillattr(struct inode
*, struct kstat
*);
2633 int vfs_getattr_nosec(struct path
*path
, struct kstat
*stat
);
2634 extern int vfs_getattr(struct path
*, struct kstat
*);
2635 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2636 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2637 void __inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2638 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2639 loff_t
inode_get_bytes(struct inode
*inode
);
2640 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
2642 extern int vfs_readdir(struct file
*, filldir_t
, void *);
2643 extern int iterate_dir(struct file
*, struct dir_context
*);
2645 extern int vfs_stat(const char __user
*, struct kstat
*);
2646 extern int vfs_lstat(const char __user
*, struct kstat
*);
2647 extern int vfs_fstat(unsigned int, struct kstat
*);
2648 extern int vfs_fstatat(int , const char __user
*, struct kstat
*, int);
2650 extern int do_vfs_ioctl(struct file
*filp
, unsigned int fd
, unsigned int cmd
,
2652 extern int __generic_block_fiemap(struct inode
*inode
,
2653 struct fiemap_extent_info
*fieinfo
,
2654 loff_t start
, loff_t len
,
2655 get_block_t
*get_block
);
2656 extern int generic_block_fiemap(struct inode
*inode
,
2657 struct fiemap_extent_info
*fieinfo
, u64 start
,
2658 u64 len
, get_block_t
*get_block
);
2660 extern void get_filesystem(struct file_system_type
*fs
);
2661 extern void put_filesystem(struct file_system_type
*fs
);
2662 extern struct file_system_type
*get_fs_type(const char *name
);
2663 extern struct super_block
*get_super(struct block_device
*);
2664 extern struct super_block
*get_super_thawed(struct block_device
*);
2665 extern struct super_block
*get_active_super(struct block_device
*bdev
);
2666 extern void drop_super(struct super_block
*sb
);
2667 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
2668 extern void iterate_supers_type(struct file_system_type
*,
2669 void (*)(struct super_block
*, void *), void *);
2671 extern int dcache_dir_open(struct inode
*, struct file
*);
2672 extern int dcache_dir_close(struct inode
*, struct file
*);
2673 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
2674 extern int dcache_readdir(struct file
*, struct dir_context
*);
2675 extern int simple_setattr(struct dentry
*, struct iattr
*);
2676 extern int simple_getattr(struct vfsmount
*, struct dentry
*, struct kstat
*);
2677 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
2678 extern int simple_open(struct inode
*inode
, struct file
*file
);
2679 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
2680 extern int simple_unlink(struct inode
*, struct dentry
*);
2681 extern int simple_rmdir(struct inode
*, struct dentry
*);
2682 extern int simple_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*);
2683 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
2684 extern int simple_empty(struct dentry
*);
2685 extern int simple_readpage(struct file
*file
, struct page
*page
);
2686 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
2687 loff_t pos
, unsigned len
, unsigned flags
,
2688 struct page
**pagep
, void **fsdata
);
2689 extern int simple_write_end(struct file
*file
, struct address_space
*mapping
,
2690 loff_t pos
, unsigned len
, unsigned copied
,
2691 struct page
*page
, void *fsdata
);
2692 extern int always_delete_dentry(const struct dentry
*);
2693 extern struct inode
*alloc_anon_inode(struct super_block
*);
2694 extern int simple_nosetlease(struct file
*, long, struct file_lock
**, void **);
2695 extern const struct dentry_operations simple_dentry_operations
;
2697 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
2698 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
2699 extern const struct file_operations simple_dir_operations
;
2700 extern const struct inode_operations simple_dir_inode_operations
;
2701 struct tree_descr
{ char *name
; const struct file_operations
*ops
; int mode
; };
2702 struct dentry
*d_alloc_name(struct dentry
*, const char *);
2703 extern int simple_fill_super(struct super_block
*, unsigned long, struct tree_descr
*);
2704 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
2705 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
2707 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
2708 loff_t
*ppos
, const void *from
, size_t available
);
2709 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
2710 const void __user
*from
, size_t count
);
2712 extern int __generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
2713 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
2715 extern int generic_check_addressable(unsigned, u64
);
2717 #ifdef CONFIG_MIGRATION
2718 extern int buffer_migrate_page(struct address_space
*,
2719 struct page
*, struct page
*,
2722 #define buffer_migrate_page NULL
2725 extern int inode_change_ok(const struct inode
*, struct iattr
*);
2726 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
2727 extern void setattr_copy(struct inode
*inode
, const struct iattr
*attr
);
2729 extern int file_update_time(struct file
*file
);
2731 extern int generic_show_options(struct seq_file
*m
, struct dentry
*root
);
2732 extern void save_mount_options(struct super_block
*sb
, char *options
);
2733 extern void replace_mount_options(struct super_block
*sb
, char *options
);
2735 static inline ino_t
parent_ino(struct dentry
*dentry
)
2740 * Don't strictly need d_lock here? If the parent ino could change
2741 * then surely we'd have a deeper race in the caller?
2743 spin_lock(&dentry
->d_lock
);
2744 res
= dentry
->d_parent
->d_inode
->i_ino
;
2745 spin_unlock(&dentry
->d_lock
);
2749 /* Transaction based IO helpers */
2752 * An argresp is stored in an allocated page and holds the
2753 * size of the argument or response, along with its content
2755 struct simple_transaction_argresp
{
2760 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2762 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
2764 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
2765 size_t size
, loff_t
*pos
);
2766 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
2768 void simple_transaction_set(struct file
*file
, size_t n
);
2771 * simple attribute files
2773 * These attributes behave similar to those in sysfs:
2775 * Writing to an attribute immediately sets a value, an open file can be
2776 * written to multiple times.
2778 * Reading from an attribute creates a buffer from the value that might get
2779 * read with multiple read calls. When the attribute has been read
2780 * completely, no further read calls are possible until the file is opened
2783 * All attributes contain a text representation of a numeric value
2784 * that are accessed with the get() and set() functions.
2786 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2787 static int __fops ## _open(struct inode *inode, struct file *file) \
2789 __simple_attr_check_format(__fmt, 0ull); \
2790 return simple_attr_open(inode, file, __get, __set, __fmt); \
2792 static const struct file_operations __fops = { \
2793 .owner = THIS_MODULE, \
2794 .open = __fops ## _open, \
2795 .release = simple_attr_release, \
2796 .read = simple_attr_read, \
2797 .write = simple_attr_write, \
2798 .llseek = generic_file_llseek, \
2801 static inline __printf(1, 2)
2802 void __simple_attr_check_format(const char *fmt
, ...)
2804 /* don't do anything, just let the compiler check the arguments; */
2807 int simple_attr_open(struct inode
*inode
, struct file
*file
,
2808 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
2810 int simple_attr_release(struct inode
*inode
, struct file
*file
);
2811 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
2812 size_t len
, loff_t
*ppos
);
2813 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
2814 size_t len
, loff_t
*ppos
);
2817 int proc_nr_files(struct ctl_table
*table
, int write
,
2818 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
2819 int proc_nr_dentry(struct ctl_table
*table
, int write
,
2820 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
2821 int proc_nr_inodes(struct ctl_table
*table
, int write
,
2822 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
2823 int __init
get_filesystem_list(char *buf
);
2825 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2826 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2828 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2829 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2830 (flag & __FMODE_NONOTIFY)))
2832 static inline int is_sxid(umode_t mode
)
2834 return (mode
& S_ISUID
) || ((mode
& S_ISGID
) && (mode
& S_IXGRP
));
2837 static inline int check_sticky(struct inode
*dir
, struct inode
*inode
)
2839 if (!(dir
->i_mode
& S_ISVTX
))
2842 return __check_sticky(dir
, inode
);
2845 static inline void inode_has_no_xattr(struct inode
*inode
)
2847 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& MS_NOSEC
))
2848 inode
->i_flags
|= S_NOSEC
;
2851 static inline bool is_root_inode(struct inode
*inode
)
2853 return inode
== inode
->i_sb
->s_root
->d_inode
;
2856 static inline bool dir_emit(struct dir_context
*ctx
,
2857 const char *name
, int namelen
,
2858 u64 ino
, unsigned type
)
2860 return ctx
->actor(ctx
, name
, namelen
, ctx
->pos
, ino
, type
) == 0;
2862 static inline bool dir_emit_dot(struct file
*file
, struct dir_context
*ctx
)
2864 return ctx
->actor(ctx
, ".", 1, ctx
->pos
,
2865 file
->f_path
.dentry
->d_inode
->i_ino
, DT_DIR
) == 0;
2867 static inline bool dir_emit_dotdot(struct file
*file
, struct dir_context
*ctx
)
2869 return ctx
->actor(ctx
, "..", 2, ctx
->pos
,
2870 parent_ino(file
->f_path
.dentry
), DT_DIR
) == 0;
2872 static inline bool dir_emit_dots(struct file
*file
, struct dir_context
*ctx
)
2874 if (ctx
->pos
== 0) {
2875 if (!dir_emit_dot(file
, ctx
))
2879 if (ctx
->pos
== 1) {
2880 if (!dir_emit_dotdot(file
, ctx
))
2886 static inline bool dir_relax(struct inode
*inode
)
2888 mutex_unlock(&inode
->i_mutex
);
2889 mutex_lock(&inode
->i_mutex
);
2890 return !IS_DEADDIR(inode
);
2893 #endif /* _LINUX_FS_H */