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/capability.h>
22 #include <linux/semaphore.h>
23 #include <linux/fiemap.h>
24 #include <linux/rculist_bl.h>
25 #include <linux/atomic.h>
26 #include <linux/shrinker.h>
27 #include <linux/migrate_mode.h>
28 #include <linux/uidgid.h>
29 #include <linux/lockdep.h>
30 #include <linux/percpu-rwsem.h>
31 #include <linux/blk_types.h>
33 #include <asm/byteorder.h>
34 #include <uapi/linux/fs.h>
36 struct export_operations
;
42 struct pipe_inode_info
;
43 struct poll_table_struct
;
45 struct vm_area_struct
;
48 struct swap_info_struct
;
50 struct workqueue_struct
;
52 extern void __init
inode_init(void);
53 extern void __init
inode_init_early(void);
54 extern void __init
files_init(unsigned long);
56 extern struct files_stat_struct files_stat
;
57 extern unsigned long get_max_files(void);
58 extern int sysctl_nr_open
;
59 extern struct inodes_stat_t inodes_stat
;
60 extern int leases_enable
, lease_break_time
;
61 extern int sysctl_protected_symlinks
;
62 extern int sysctl_protected_hardlinks
;
65 typedef int (get_block_t
)(struct inode
*inode
, sector_t iblock
,
66 struct buffer_head
*bh_result
, int create
);
67 typedef void (dio_iodone_t
)(struct kiocb
*iocb
, loff_t offset
,
68 ssize_t bytes
, void *private);
70 #define MAY_EXEC 0x00000001
71 #define MAY_WRITE 0x00000002
72 #define MAY_READ 0x00000004
73 #define MAY_APPEND 0x00000008
74 #define MAY_ACCESS 0x00000010
75 #define MAY_OPEN 0x00000020
76 #define MAY_CHDIR 0x00000040
77 /* called from RCU mode, don't block */
78 #define MAY_NOT_BLOCK 0x00000080
81 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
82 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
85 /* file is open for reading */
86 #define FMODE_READ ((__force fmode_t)0x1)
87 /* file is open for writing */
88 #define FMODE_WRITE ((__force fmode_t)0x2)
89 /* file is seekable */
90 #define FMODE_LSEEK ((__force fmode_t)0x4)
91 /* file can be accessed using pread */
92 #define FMODE_PREAD ((__force fmode_t)0x8)
93 /* file can be accessed using pwrite */
94 #define FMODE_PWRITE ((__force fmode_t)0x10)
95 /* File is opened for execution with sys_execve / sys_uselib */
96 #define FMODE_EXEC ((__force fmode_t)0x20)
97 /* File is opened with O_NDELAY (only set for block devices) */
98 #define FMODE_NDELAY ((__force fmode_t)0x40)
99 /* File is opened with O_EXCL (only set for block devices) */
100 #define FMODE_EXCL ((__force fmode_t)0x80)
101 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
102 (specialy hack for floppy.c) */
103 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
104 /* 32bit hashes as llseek() offset (for directories) */
105 #define FMODE_32BITHASH ((__force fmode_t)0x200)
106 /* 64bit hashes as llseek() offset (for directories) */
107 #define FMODE_64BITHASH ((__force fmode_t)0x400)
110 * Don't update ctime and mtime.
112 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
113 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
115 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
117 /* Expect random access pattern */
118 #define FMODE_RANDOM ((__force fmode_t)0x1000)
120 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
121 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
123 /* File is opened with O_PATH; almost nothing can be done with it */
124 #define FMODE_PATH ((__force fmode_t)0x4000)
126 /* File needs atomic accesses to f_pos */
127 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
129 /* File was opened by fanotify and shouldn't generate fanotify events */
130 #define FMODE_NONOTIFY ((__force fmode_t)0x1000000)
133 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
134 * that indicates that they should check the contents of the iovec are
135 * valid, but not check the memory that the iovec elements
138 #define CHECK_IOVEC_ONLY -1
141 * The below are the various read and write types that we support. Some of
142 * them include behavioral modifiers that send information down to the
143 * block layer and IO scheduler. Terminology:
145 * The block layer uses device plugging to defer IO a little bit, in
146 * the hope that we will see more IO very shortly. This increases
147 * coalescing of adjacent IO and thus reduces the number of IOs we
148 * have to send to the device. It also allows for better queuing,
149 * if the IO isn't mergeable. If the caller is going to be waiting
150 * for the IO, then he must ensure that the device is unplugged so
151 * that the IO is dispatched to the driver.
153 * All IO is handled async in Linux. This is fine for background
154 * writes, but for reads or writes that someone waits for completion
155 * on, we want to notify the block layer and IO scheduler so that they
156 * know about it. That allows them to make better scheduling
157 * decisions. So when the below references 'sync' and 'async', it
158 * is referencing this priority hint.
160 * With that in mind, the available types are:
162 * READ A normal read operation. Device will be plugged.
163 * READ_SYNC A synchronous read. Device is not plugged, caller can
164 * immediately wait on this read without caring about
166 * READA Used for read-ahead operations. Lower priority, and the
167 * block layer could (in theory) choose to ignore this
168 * request if it runs into resource problems.
169 * WRITE A normal async write. Device will be plugged.
170 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
171 * the hint that someone will be waiting on this IO
172 * shortly. The write equivalent of READ_SYNC.
173 * WRITE_ODIRECT Special case write for O_DIRECT only.
174 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
175 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
176 * non-volatile media on completion.
177 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
178 * by a cache flush and data is guaranteed to be on
179 * non-volatile media on completion.
182 #define RW_MASK REQ_WRITE
183 #define RWA_MASK REQ_RAHEAD
186 #define WRITE RW_MASK
187 #define READA RWA_MASK
188 #define KERNEL_READ (READ|REQ_KERNEL)
189 #define KERNEL_WRITE (WRITE|REQ_KERNEL)
191 #define READ_SYNC (READ | REQ_SYNC)
192 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
193 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
194 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
195 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
196 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
199 * Attribute flags. These should be or-ed together to figure out what
202 #define ATTR_MODE (1 << 0)
203 #define ATTR_UID (1 << 1)
204 #define ATTR_GID (1 << 2)
205 #define ATTR_SIZE (1 << 3)
206 #define ATTR_ATIME (1 << 4)
207 #define ATTR_MTIME (1 << 5)
208 #define ATTR_CTIME (1 << 6)
209 #define ATTR_ATIME_SET (1 << 7)
210 #define ATTR_MTIME_SET (1 << 8)
211 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
212 #define ATTR_ATTR_FLAG (1 << 10)
213 #define ATTR_KILL_SUID (1 << 11)
214 #define ATTR_KILL_SGID (1 << 12)
215 #define ATTR_FILE (1 << 13)
216 #define ATTR_KILL_PRIV (1 << 14)
217 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
218 #define ATTR_TIMES_SET (1 << 16)
221 * This is the Inode Attributes structure, used for notify_change(). It
222 * uses the above definitions as flags, to know which values have changed.
223 * Also, in this manner, a Filesystem can look at only the values it cares
224 * about. Basically, these are the attributes that the VFS layer can
225 * request to change from the FS layer.
227 * Derek Atkins <warlord@MIT.EDU> 94-10-20
230 unsigned int ia_valid
;
235 struct timespec ia_atime
;
236 struct timespec ia_mtime
;
237 struct timespec ia_ctime
;
240 * Not an attribute, but an auxiliary info for filesystems wanting to
241 * implement an ftruncate() like method. NOTE: filesystem should
242 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
244 struct file
*ia_file
;
248 * Includes for diskquotas.
250 #include <linux/quota.h>
253 * enum positive_aop_returns - aop return codes with specific semantics
255 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
256 * completed, that the page is still locked, and
257 * should be considered active. The VM uses this hint
258 * to return the page to the active list -- it won't
259 * be a candidate for writeback again in the near
260 * future. Other callers must be careful to unlock
261 * the page if they get this return. Returned by
264 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
265 * unlocked it and the page might have been truncated.
266 * The caller should back up to acquiring a new page and
267 * trying again. The aop will be taking reasonable
268 * precautions not to livelock. If the caller held a page
269 * reference, it should drop it before retrying. Returned
272 * address_space_operation functions return these large constants to indicate
273 * special semantics to the caller. These are much larger than the bytes in a
274 * page to allow for functions that return the number of bytes operated on in a
278 enum positive_aop_returns
{
279 AOP_WRITEPAGE_ACTIVATE
= 0x80000,
280 AOP_TRUNCATED_PAGE
= 0x80001,
283 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
284 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
285 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
286 * helper code (eg buffer layer)
287 * to clear GFP_FS from alloc */
290 * oh the beauties of C type declarations.
293 struct address_space
;
294 struct writeback_control
;
297 const struct iovec
*iov
;
298 unsigned long nr_segs
;
303 size_t iov_iter_copy_from_user_atomic(struct page
*page
,
304 struct iov_iter
*i
, unsigned long offset
, size_t bytes
);
305 size_t iov_iter_copy_from_user(struct page
*page
,
306 struct iov_iter
*i
, unsigned long offset
, size_t bytes
);
307 void iov_iter_advance(struct iov_iter
*i
, size_t bytes
);
308 int iov_iter_fault_in_readable(struct iov_iter
*i
, size_t bytes
);
309 size_t iov_iter_single_seg_count(const struct iov_iter
*i
);
311 static inline void iov_iter_init(struct iov_iter
*i
,
312 const struct iovec
*iov
, unsigned long nr_segs
,
313 size_t count
, size_t written
)
316 i
->nr_segs
= nr_segs
;
318 i
->count
= count
+ written
;
320 iov_iter_advance(i
, written
);
323 static inline size_t iov_iter_count(struct iov_iter
*i
)
329 * "descriptor" for what we're up to with a read.
330 * This allows us to use the same read code yet
331 * have multiple different users of the data that
332 * we read from a file.
334 * The simplest case just copies the data to user
347 typedef int (*read_actor_t
)(read_descriptor_t
*, struct page
*,
348 unsigned long, unsigned long);
350 struct address_space_operations
{
351 int (*writepage
)(struct page
*page
, struct writeback_control
*wbc
);
352 int (*readpage
)(struct file
*, struct page
*);
354 /* Write back some dirty pages from this mapping. */
355 int (*writepages
)(struct address_space
*, struct writeback_control
*);
357 /* Set a page dirty. Return true if this dirtied it */
358 int (*set_page_dirty
)(struct page
*page
);
360 int (*readpages
)(struct file
*filp
, struct address_space
*mapping
,
361 struct list_head
*pages
, unsigned nr_pages
);
363 int (*write_begin
)(struct file
*, struct address_space
*mapping
,
364 loff_t pos
, unsigned len
, unsigned flags
,
365 struct page
**pagep
, void **fsdata
);
366 int (*write_end
)(struct file
*, struct address_space
*mapping
,
367 loff_t pos
, unsigned len
, unsigned copied
,
368 struct page
*page
, void *fsdata
);
370 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
371 sector_t (*bmap
)(struct address_space
*, sector_t
);
372 void (*invalidatepage
) (struct page
*, unsigned int, unsigned int);
373 int (*releasepage
) (struct page
*, gfp_t
);
374 void (*freepage
)(struct page
*);
375 ssize_t (*direct_IO
)(int, struct kiocb
*, const struct iovec
*iov
,
376 loff_t offset
, unsigned long nr_segs
);
377 int (*get_xip_mem
)(struct address_space
*, pgoff_t
, int,
378 void **, unsigned long *);
380 * migrate the contents of a page to the specified target. If
381 * migrate_mode is MIGRATE_ASYNC, it must not block.
383 int (*migratepage
) (struct address_space
*,
384 struct page
*, struct page
*, enum migrate_mode
);
385 int (*launder_page
) (struct page
*);
386 int (*is_partially_uptodate
) (struct page
*, read_descriptor_t
*,
388 void (*is_dirty_writeback
) (struct page
*, bool *, bool *);
389 int (*error_remove_page
)(struct address_space
*, struct page
*);
391 /* swapfile support */
392 int (*swap_activate
)(struct swap_info_struct
*sis
, struct file
*file
,
394 void (*swap_deactivate
)(struct file
*file
);
397 extern const struct address_space_operations empty_aops
;
400 * pagecache_write_begin/pagecache_write_end must be used by general code
401 * to write into the pagecache.
403 int pagecache_write_begin(struct file
*, struct address_space
*mapping
,
404 loff_t pos
, unsigned len
, unsigned flags
,
405 struct page
**pagep
, void **fsdata
);
407 int pagecache_write_end(struct file
*, struct address_space
*mapping
,
408 loff_t pos
, unsigned len
, unsigned copied
,
409 struct page
*page
, void *fsdata
);
411 struct backing_dev_info
;
412 struct address_space
{
413 struct inode
*host
; /* owner: inode, block_device */
414 struct radix_tree_root page_tree
; /* radix tree of all pages */
415 spinlock_t tree_lock
; /* and lock protecting it */
416 unsigned int i_mmap_writable
;/* count VM_SHARED mappings */
417 struct rb_root i_mmap
; /* tree of private and shared mappings */
418 struct list_head i_mmap_nonlinear
;/*list VM_NONLINEAR mappings */
419 struct mutex i_mmap_mutex
; /* protect tree, count, list */
420 /* Protected by tree_lock together with the radix tree */
421 unsigned long nrpages
; /* number of total pages */
422 unsigned long nrshadows
; /* number of shadow entries */
423 pgoff_t writeback_index
;/* writeback starts here */
424 const struct address_space_operations
*a_ops
; /* methods */
425 unsigned long flags
; /* error bits/gfp mask */
426 struct backing_dev_info
*backing_dev_info
; /* device readahead, etc */
427 spinlock_t private_lock
; /* for use by the address_space */
428 struct list_head private_list
; /* ditto */
429 void *private_data
; /* ditto */
430 } __attribute__((aligned(sizeof(long))));
432 * On most architectures that alignment is already the case; but
433 * must be enforced here for CRIS, to let the least significant bit
434 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
436 struct request_queue
;
438 struct block_device
{
439 dev_t bd_dev
; /* not a kdev_t - it's a search key */
441 struct inode
* bd_inode
; /* will die */
442 struct super_block
* bd_super
;
443 struct mutex bd_mutex
; /* open/close mutex */
444 struct list_head bd_inodes
;
448 bool bd_write_holder
;
450 struct list_head bd_holder_disks
;
452 struct block_device
* bd_contains
;
453 unsigned bd_block_size
;
454 struct hd_struct
* bd_part
;
455 /* number of times partitions within this device have been opened. */
456 unsigned bd_part_count
;
458 struct gendisk
* bd_disk
;
459 struct request_queue
* bd_queue
;
460 struct list_head bd_list
;
462 * Private data. You must have bd_claim'ed the block_device
463 * to use this. NOTE: bd_claim allows an owner to claim
464 * the same device multiple times, the owner must take special
465 * care to not mess up bd_private for that case.
467 unsigned long bd_private
;
469 /* The counter of freeze processes */
470 int bd_fsfreeze_count
;
471 /* Mutex for freeze */
472 struct mutex bd_fsfreeze_mutex
;
476 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
479 #define PAGECACHE_TAG_DIRTY 0
480 #define PAGECACHE_TAG_WRITEBACK 1
481 #define PAGECACHE_TAG_TOWRITE 2
483 int mapping_tagged(struct address_space
*mapping
, int tag
);
486 * Might pages of this file be mapped into userspace?
488 static inline int mapping_mapped(struct address_space
*mapping
)
490 return !RB_EMPTY_ROOT(&mapping
->i_mmap
) ||
491 !list_empty(&mapping
->i_mmap_nonlinear
);
495 * Might pages of this file have been modified in userspace?
496 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
497 * marks vma as VM_SHARED if it is shared, and the file was opened for
498 * writing i.e. vma may be mprotected writable even if now readonly.
500 static inline int mapping_writably_mapped(struct address_space
*mapping
)
502 return mapping
->i_mmap_writable
!= 0;
506 * Use sequence counter to get consistent i_size on 32-bit processors.
508 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
509 #include <linux/seqlock.h>
510 #define __NEED_I_SIZE_ORDERED
511 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
513 #define i_size_ordered_init(inode) do { } while (0)
517 #define ACL_NOT_CACHED ((void *)(-1))
519 #define IOP_FASTPERM 0x0001
520 #define IOP_LOOKUP 0x0002
521 #define IOP_NOFOLLOW 0x0004
524 * Keep mostly read-only and often accessed (especially for
525 * the RCU path lookup and 'stat' data) fields at the beginning
526 * of the 'struct inode'
530 unsigned short i_opflags
;
533 unsigned int i_flags
;
535 #ifdef CONFIG_FS_POSIX_ACL
536 struct posix_acl
*i_acl
;
537 struct posix_acl
*i_default_acl
;
540 const struct inode_operations
*i_op
;
541 struct super_block
*i_sb
;
542 struct address_space
*i_mapping
;
544 #ifdef CONFIG_SECURITY
548 /* Stat data, not accessed from path walking */
551 * Filesystems may only read i_nlink directly. They shall use the
552 * following functions for modification:
554 * (set|clear|inc|drop)_nlink
555 * inode_(inc|dec)_link_count
558 const unsigned int i_nlink
;
559 unsigned int __i_nlink
;
563 struct timespec i_atime
;
564 struct timespec i_mtime
;
565 struct timespec i_ctime
;
566 spinlock_t i_lock
; /* i_blocks, i_bytes, maybe i_size */
567 unsigned short i_bytes
;
568 unsigned int i_blkbits
;
571 #ifdef __NEED_I_SIZE_ORDERED
572 seqcount_t i_size_seqcount
;
576 unsigned long i_state
;
577 struct mutex i_mutex
;
579 unsigned long dirtied_when
; /* jiffies of first dirtying */
581 struct hlist_node i_hash
;
582 struct list_head i_wb_list
; /* backing dev IO list */
583 struct list_head i_lru
; /* inode LRU list */
584 struct list_head i_sb_list
;
586 struct hlist_head i_dentry
;
587 struct rcu_head i_rcu
;
591 atomic_t i_dio_count
;
592 atomic_t i_writecount
;
594 atomic_t i_readcount
; /* struct files open RO */
596 const struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
597 struct file_lock
*i_flock
;
598 struct address_space i_data
;
600 struct dquot
*i_dquot
[MAXQUOTAS
];
602 struct list_head i_devices
;
604 struct pipe_inode_info
*i_pipe
;
605 struct block_device
*i_bdev
;
611 #ifdef CONFIG_FSNOTIFY
612 __u32 i_fsnotify_mask
; /* all events this inode cares about */
613 struct hlist_head i_fsnotify_marks
;
616 void *i_private
; /* fs or device private pointer */
619 static inline int inode_unhashed(struct inode
*inode
)
621 return hlist_unhashed(&inode
->i_hash
);
625 * inode->i_mutex nesting subclasses for the lock validator:
627 * 0: the object of the current VFS operation
631 * 4: second non-directory
632 * The last is for certain operations (such as rename) which lock two
633 * non-directories at once.
635 * The locking order between these classes is
636 * parent -> child -> normal -> xattr -> second non-directory
638 enum inode_i_mutex_lock_class
647 void lock_two_nondirectories(struct inode
*, struct inode
*);
648 void unlock_two_nondirectories(struct inode
*, struct inode
*);
651 * NOTE: in a 32bit arch with a preemptable kernel and
652 * an UP compile the i_size_read/write must be atomic
653 * with respect to the local cpu (unlike with preempt disabled),
654 * but they don't need to be atomic with respect to other cpus like in
655 * true SMP (so they need either to either locally disable irq around
656 * the read or for example on x86 they can be still implemented as a
657 * cmpxchg8b without the need of the lock prefix). For SMP compiles
658 * and 64bit archs it makes no difference if preempt is enabled or not.
660 static inline loff_t
i_size_read(const struct inode
*inode
)
662 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
667 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
668 i_size
= inode
->i_size
;
669 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
671 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
675 i_size
= inode
->i_size
;
679 return inode
->i_size
;
684 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
685 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
686 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
688 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
690 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
692 write_seqcount_begin(&inode
->i_size_seqcount
);
693 inode
->i_size
= i_size
;
694 write_seqcount_end(&inode
->i_size_seqcount
);
696 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
698 inode
->i_size
= i_size
;
701 inode
->i_size
= i_size
;
705 /* Helper functions so that in most cases filesystems will
706 * not need to deal directly with kuid_t and kgid_t and can
707 * instead deal with the raw numeric values that are stored
710 static inline uid_t
i_uid_read(const struct inode
*inode
)
712 return from_kuid(&init_user_ns
, inode
->i_uid
);
715 static inline gid_t
i_gid_read(const struct inode
*inode
)
717 return from_kgid(&init_user_ns
, inode
->i_gid
);
720 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
722 inode
->i_uid
= make_kuid(&init_user_ns
, uid
);
725 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
727 inode
->i_gid
= make_kgid(&init_user_ns
, gid
);
730 static inline unsigned iminor(const struct inode
*inode
)
732 return MINOR(inode
->i_rdev
);
735 static inline unsigned imajor(const struct inode
*inode
)
737 return MAJOR(inode
->i_rdev
);
740 extern struct block_device
*I_BDEV(struct inode
*inode
);
743 rwlock_t lock
; /* protects pid, uid, euid fields */
744 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
745 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
746 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
747 int signum
; /* posix.1b rt signal to be delivered on IO */
751 * Track a single file's readahead state
753 struct file_ra_state
{
754 pgoff_t start
; /* where readahead started */
755 unsigned int size
; /* # of readahead pages */
756 unsigned int async_size
; /* do asynchronous readahead when
757 there are only # of pages ahead */
759 unsigned int ra_pages
; /* Maximum readahead window */
760 unsigned int mmap_miss
; /* Cache miss stat for mmap accesses */
761 loff_t prev_pos
; /* Cache last read() position */
765 * Check if @index falls in the readahead windows.
767 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
769 return (index
>= ra
->start
&&
770 index
< ra
->start
+ ra
->size
);
773 #define FILE_MNT_WRITE_TAKEN 1
774 #define FILE_MNT_WRITE_RELEASED 2
778 struct llist_node fu_llist
;
779 struct rcu_head fu_rcuhead
;
782 #define f_dentry f_path.dentry
783 struct inode
*f_inode
; /* cached value */
784 const struct file_operations
*f_op
;
787 * Protects f_ep_links, f_flags.
788 * Must not be taken from IRQ context.
791 atomic_long_t f_count
;
792 unsigned int f_flags
;
794 struct mutex f_pos_lock
;
796 struct fown_struct f_owner
;
797 const struct cred
*f_cred
;
798 struct file_ra_state f_ra
;
801 #ifdef CONFIG_SECURITY
804 /* needed for tty driver, and maybe others */
808 /* Used by fs/eventpoll.c to link all the hooks to this file */
809 struct list_head f_ep_links
;
810 struct list_head f_tfile_llink
;
811 #endif /* #ifdef CONFIG_EPOLL */
812 struct address_space
*f_mapping
;
813 #ifdef CONFIG_DEBUG_WRITECOUNT
814 unsigned long f_mnt_write_state
;
816 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
821 /* file identifier */
822 unsigned char f_handle
[0];
825 static inline struct file
*get_file(struct file
*f
)
827 atomic_long_inc(&f
->f_count
);
830 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
831 #define file_count(x) atomic_long_read(&(x)->f_count)
833 #ifdef CONFIG_DEBUG_WRITECOUNT
834 static inline void file_take_write(struct file
*f
)
836 WARN_ON(f
->f_mnt_write_state
!= 0);
837 f
->f_mnt_write_state
= FILE_MNT_WRITE_TAKEN
;
839 static inline void file_release_write(struct file
*f
)
841 f
->f_mnt_write_state
|= FILE_MNT_WRITE_RELEASED
;
843 static inline void file_reset_write(struct file
*f
)
845 f
->f_mnt_write_state
= 0;
847 static inline void file_check_state(struct file
*f
)
850 * At this point, either both or neither of these bits
853 WARN_ON(f
->f_mnt_write_state
== FILE_MNT_WRITE_TAKEN
);
854 WARN_ON(f
->f_mnt_write_state
== FILE_MNT_WRITE_RELEASED
);
856 static inline int file_check_writeable(struct file
*f
)
858 if (f
->f_mnt_write_state
== FILE_MNT_WRITE_TAKEN
)
860 printk(KERN_WARNING
"writeable file with no "
861 "mnt_want_write()\n");
865 #else /* !CONFIG_DEBUG_WRITECOUNT */
866 static inline void file_take_write(struct file
*filp
) {}
867 static inline void file_release_write(struct file
*filp
) {}
868 static inline void file_reset_write(struct file
*filp
) {}
869 static inline void file_check_state(struct file
*filp
) {}
870 static inline int file_check_writeable(struct file
*filp
)
874 #endif /* CONFIG_DEBUG_WRITECOUNT */
876 #define MAX_NON_LFS ((1UL<<31) - 1)
878 /* Page cache limit. The filesystems should put that into their s_maxbytes
879 limits, otherwise bad things can happen in VM. */
880 #if BITS_PER_LONG==32
881 #define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
882 #elif BITS_PER_LONG==64
883 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
888 #define FL_DELEG 4 /* NFSv4 delegation */
889 #define FL_ACCESS 8 /* not trying to lock, just looking */
890 #define FL_EXISTS 16 /* when unlocking, test for existence */
891 #define FL_LEASE 32 /* lease held on this file */
892 #define FL_CLOSE 64 /* unlock on close */
893 #define FL_SLEEP 128 /* A blocking lock */
894 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
895 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
896 #define FL_FILE_PVT 1024 /* lock is private to the file */
899 * Special return value from posix_lock_file() and vfs_lock_file() for
900 * asynchronous locking.
902 #define FILE_LOCK_DEFERRED 1
905 * The POSIX file lock owner is determined by
906 * the "struct files_struct" in the thread group
907 * (or NULL for no owner - BSD locks).
909 * Lockd stuffs a "host" pointer into this.
911 typedef struct files_struct
*fl_owner_t
;
913 struct file_lock_operations
{
914 void (*fl_copy_lock
)(struct file_lock
*, struct file_lock
*);
915 void (*fl_release_private
)(struct file_lock
*);
918 struct lock_manager_operations
{
919 int (*lm_compare_owner
)(struct file_lock
*, struct file_lock
*);
920 unsigned long (*lm_owner_key
)(struct file_lock
*);
921 void (*lm_notify
)(struct file_lock
*); /* unblock callback */
922 int (*lm_grant
)(struct file_lock
*, struct file_lock
*, int);
923 void (*lm_break
)(struct file_lock
*);
924 int (*lm_change
)(struct file_lock
**, int);
927 struct lock_manager
{
928 struct list_head list
;
932 void locks_start_grace(struct net
*, struct lock_manager
*);
933 void locks_end_grace(struct lock_manager
*);
934 int locks_in_grace(struct net
*);
936 /* that will die - we need it for nfs_lock_info */
937 #include <linux/nfs_fs_i.h>
940 * struct file_lock represents a generic "file lock". It's used to represent
941 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
942 * note that the same struct is used to represent both a request for a lock and
943 * the lock itself, but the same object is never used for both.
945 * FIXME: should we create a separate "struct lock_request" to help distinguish
948 * The i_flock list is ordered by:
950 * 1) lock type -- FL_LEASEs first, then FL_FLOCK, and finally FL_POSIX
952 * 3) lock range start
955 * Obviously, the last two criteria only matter for POSIX locks.
958 struct file_lock
*fl_next
; /* singly linked list for this inode */
959 struct hlist_node fl_link
; /* node in global lists */
960 struct list_head fl_block
; /* circular list of blocked processes */
962 unsigned int fl_flags
;
963 unsigned char fl_type
;
965 int fl_link_cpu
; /* what cpu's list is this on? */
966 struct pid
*fl_nspid
;
967 wait_queue_head_t fl_wait
;
968 struct file
*fl_file
;
972 struct fasync_struct
* fl_fasync
; /* for lease break notifications */
973 /* for lease breaks: */
974 unsigned long fl_break_time
;
975 unsigned long fl_downgrade_time
;
977 const struct file_lock_operations
*fl_ops
; /* Callbacks for filesystems */
978 const struct lock_manager_operations
*fl_lmops
; /* Callbacks for lockmanagers */
980 struct nfs_lock_info nfs_fl
;
981 struct nfs4_lock_info nfs4_fl
;
983 struct list_head link
; /* link in AFS vnode's pending_locks list */
984 int state
; /* state of grant or error if -ve */
989 /* The following constant reflects the upper bound of the file/locking space */
991 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
992 #define OFFSET_MAX INT_LIMIT(loff_t)
993 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
996 #include <linux/fcntl.h>
998 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
1000 #ifdef CONFIG_FILE_LOCKING
1001 extern int fcntl_getlk(struct file
*, unsigned int, struct flock __user
*);
1002 extern int fcntl_setlk(unsigned int, struct file
*, unsigned int,
1003 struct flock __user
*);
1005 #if BITS_PER_LONG == 32
1006 extern int fcntl_getlk64(struct file
*, unsigned int, struct flock64 __user
*);
1007 extern int fcntl_setlk64(unsigned int, struct file
*, unsigned int,
1008 struct flock64 __user
*);
1011 extern int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
);
1012 extern int fcntl_getlease(struct file
*filp
);
1015 void locks_free_lock(struct file_lock
*fl
);
1016 extern void locks_init_lock(struct file_lock
*);
1017 extern struct file_lock
* locks_alloc_lock(void);
1018 extern void locks_copy_lock(struct file_lock
*, struct file_lock
*);
1019 extern void __locks_copy_lock(struct file_lock
*, const struct file_lock
*);
1020 extern void locks_remove_posix(struct file
*, fl_owner_t
);
1021 extern void locks_remove_file(struct file
*);
1022 extern void locks_release_private(struct file_lock
*);
1023 extern void posix_test_lock(struct file
*, struct file_lock
*);
1024 extern int posix_lock_file(struct file
*, struct file_lock
*, struct file_lock
*);
1025 extern int posix_lock_file_wait(struct file
*, struct file_lock
*);
1026 extern int posix_unblock_lock(struct file_lock
*);
1027 extern int vfs_test_lock(struct file
*, struct file_lock
*);
1028 extern int vfs_lock_file(struct file
*, unsigned int, struct file_lock
*, struct file_lock
*);
1029 extern int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
);
1030 extern int flock_lock_file_wait(struct file
*filp
, struct file_lock
*fl
);
1031 extern int __break_lease(struct inode
*inode
, unsigned int flags
, unsigned int type
);
1032 extern void lease_get_mtime(struct inode
*, struct timespec
*time
);
1033 extern int generic_setlease(struct file
*, long, struct file_lock
**);
1034 extern int vfs_setlease(struct file
*, long, struct file_lock
**);
1035 extern int lease_modify(struct file_lock
**, int);
1036 extern int lock_may_read(struct inode
*, loff_t start
, unsigned long count
);
1037 extern int lock_may_write(struct inode
*, loff_t start
, unsigned long count
);
1038 #else /* !CONFIG_FILE_LOCKING */
1039 static inline int fcntl_getlk(struct file
*file
, unsigned int cmd
,
1040 struct flock __user
*user
)
1045 static inline int fcntl_setlk(unsigned int fd
, struct file
*file
,
1046 unsigned int cmd
, struct flock __user
*user
)
1051 #if BITS_PER_LONG == 32
1052 static inline int fcntl_getlk64(struct file
*file
, unsigned int cmd
,
1053 struct flock64 __user
*user
)
1058 static inline int fcntl_setlk64(unsigned int fd
, struct file
*file
,
1059 unsigned int cmd
, struct flock64 __user
*user
)
1064 static inline int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
)
1069 static inline int fcntl_getlease(struct file
*filp
)
1074 static inline void locks_init_lock(struct file_lock
*fl
)
1079 static inline void __locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
1084 static inline void locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
1089 static inline void locks_remove_posix(struct file
*filp
, fl_owner_t owner
)
1094 static inline void locks_remove_file(struct file
*filp
)
1099 static inline void posix_test_lock(struct file
*filp
, struct file_lock
*fl
)
1104 static inline int posix_lock_file(struct file
*filp
, struct file_lock
*fl
,
1105 struct file_lock
*conflock
)
1110 static inline int posix_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1115 static inline int posix_unblock_lock(struct file_lock
*waiter
)
1120 static inline int vfs_test_lock(struct file
*filp
, struct file_lock
*fl
)
1125 static inline int vfs_lock_file(struct file
*filp
, unsigned int cmd
,
1126 struct file_lock
*fl
, struct file_lock
*conf
)
1131 static inline int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
)
1136 static inline int flock_lock_file_wait(struct file
*filp
,
1137 struct file_lock
*request
)
1142 static inline int __break_lease(struct inode
*inode
, unsigned int mode
, unsigned int type
)
1147 static inline void lease_get_mtime(struct inode
*inode
, struct timespec
*time
)
1152 static inline int generic_setlease(struct file
*filp
, long arg
,
1153 struct file_lock
**flp
)
1158 static inline int vfs_setlease(struct file
*filp
, long arg
,
1159 struct file_lock
**lease
)
1164 static inline int lease_modify(struct file_lock
**before
, int arg
)
1169 static inline int lock_may_read(struct inode
*inode
, loff_t start
,
1175 static inline int lock_may_write(struct inode
*inode
, loff_t start
,
1180 #endif /* !CONFIG_FILE_LOCKING */
1183 struct fasync_struct
{
1187 struct fasync_struct
*fa_next
; /* singly linked list */
1188 struct file
*fa_file
;
1189 struct rcu_head fa_rcu
;
1192 #define FASYNC_MAGIC 0x4601
1194 /* SMP safe fasync helpers: */
1195 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1196 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1197 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1198 extern struct fasync_struct
*fasync_alloc(void);
1199 extern void fasync_free(struct fasync_struct
*);
1201 /* can be called from interrupts */
1202 extern void kill_fasync(struct fasync_struct
**, int, int);
1204 extern int __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1205 extern int f_setown(struct file
*filp
, unsigned long arg
, int force
);
1206 extern void f_delown(struct file
*filp
);
1207 extern pid_t
f_getown(struct file
*filp
);
1208 extern int send_sigurg(struct fown_struct
*fown
);
1216 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1217 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1218 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1219 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1220 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1222 extern struct list_head super_blocks
;
1223 extern spinlock_t sb_lock
;
1225 /* Possible states of 'frozen' field */
1227 SB_UNFROZEN
= 0, /* FS is unfrozen */
1228 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1229 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1230 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1231 * internal threads if needed) */
1232 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1235 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1238 /* Counters for counting writers at each level */
1239 struct percpu_counter counter
[SB_FREEZE_LEVELS
];
1240 wait_queue_head_t wait
; /* queue for waiting for
1241 writers / faults to finish */
1242 int frozen
; /* Is sb frozen? */
1243 wait_queue_head_t wait_unfrozen
; /* queue for waiting for
1245 #ifdef CONFIG_DEBUG_LOCK_ALLOC
1246 struct lockdep_map lock_map
[SB_FREEZE_LEVELS
];
1250 struct super_block
{
1251 struct list_head s_list
; /* Keep this first */
1252 dev_t s_dev
; /* search index; _not_ kdev_t */
1253 unsigned char s_blocksize_bits
;
1254 unsigned long s_blocksize
;
1255 loff_t s_maxbytes
; /* Max file size */
1256 struct file_system_type
*s_type
;
1257 const struct super_operations
*s_op
;
1258 const struct dquot_operations
*dq_op
;
1259 const struct quotactl_ops
*s_qcop
;
1260 const struct export_operations
*s_export_op
;
1261 unsigned long s_flags
;
1262 unsigned long s_magic
;
1263 struct dentry
*s_root
;
1264 struct rw_semaphore s_umount
;
1267 #ifdef CONFIG_SECURITY
1270 const struct xattr_handler
**s_xattr
;
1272 struct list_head s_inodes
; /* all inodes */
1273 struct hlist_bl_head s_anon
; /* anonymous dentries for (nfs) exporting */
1274 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1275 struct block_device
*s_bdev
;
1276 struct backing_dev_info
*s_bdi
;
1277 struct mtd_info
*s_mtd
;
1278 struct hlist_node s_instances
;
1279 struct quota_info s_dquot
; /* Diskquota specific options */
1281 struct sb_writers s_writers
;
1283 char s_id
[32]; /* Informational name */
1284 u8 s_uuid
[16]; /* UUID */
1286 void *s_fs_info
; /* Filesystem private info */
1287 unsigned int s_max_links
;
1290 /* Granularity of c/m/atime in ns.
1291 Cannot be worse than a second */
1295 * The next field is for VFS *only*. No filesystems have any business
1296 * even looking at it. You had been warned.
1298 struct mutex s_vfs_rename_mutex
; /* Kludge */
1301 * Filesystem subtype. If non-empty the filesystem type field
1302 * in /proc/mounts will be "type.subtype"
1307 * Saved mount options for lazy filesystems using
1308 * generic_show_options()
1310 char __rcu
*s_options
;
1311 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1314 * Saved pool identifier for cleancache (-1 means none)
1316 int cleancache_poolid
;
1318 struct shrinker s_shrink
; /* per-sb shrinker handle */
1320 /* Number of inodes with nlink == 0 but still referenced */
1321 atomic_long_t s_remove_count
;
1323 /* Being remounted read-only */
1324 int s_readonly_remount
;
1326 /* AIO completions deferred from interrupt context */
1327 struct workqueue_struct
*s_dio_done_wq
;
1330 * Keep the lru lists last in the structure so they always sit on their
1331 * own individual cachelines.
1333 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp
;
1334 struct list_lru s_inode_lru ____cacheline_aligned_in_smp
;
1335 struct rcu_head rcu
;
1338 extern struct timespec
current_fs_time(struct super_block
*sb
);
1341 * Snapshotting support.
1344 void __sb_end_write(struct super_block
*sb
, int level
);
1345 int __sb_start_write(struct super_block
*sb
, int level
, bool wait
);
1348 * sb_end_write - drop write access to a superblock
1349 * @sb: the super we wrote to
1351 * Decrement number of writers to the filesystem. Wake up possible waiters
1352 * wanting to freeze the filesystem.
1354 static inline void sb_end_write(struct super_block
*sb
)
1356 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1360 * sb_end_pagefault - drop write access to a superblock from a page fault
1361 * @sb: the super we wrote to
1363 * Decrement number of processes handling write page fault to the filesystem.
1364 * Wake up possible waiters wanting to freeze the filesystem.
1366 static inline void sb_end_pagefault(struct super_block
*sb
)
1368 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1372 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1373 * @sb: the super we wrote to
1375 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1376 * waiters wanting to freeze the filesystem.
1378 static inline void sb_end_intwrite(struct super_block
*sb
)
1380 __sb_end_write(sb
, SB_FREEZE_FS
);
1384 * sb_start_write - get write access to a superblock
1385 * @sb: the super we write to
1387 * When a process wants to write data or metadata to a file system (i.e. dirty
1388 * a page or an inode), it should embed the operation in a sb_start_write() -
1389 * sb_end_write() pair to get exclusion against file system freezing. This
1390 * function increments number of writers preventing freezing. If the file
1391 * system is already frozen, the function waits until the file system is
1394 * Since freeze protection behaves as a lock, users have to preserve
1395 * ordering of freeze protection and other filesystem locks. Generally,
1396 * freeze protection should be the outermost lock. In particular, we have:
1399 * -> i_mutex (write path, truncate, directory ops, ...)
1400 * -> s_umount (freeze_super, thaw_super)
1402 static inline void sb_start_write(struct super_block
*sb
)
1404 __sb_start_write(sb
, SB_FREEZE_WRITE
, true);
1407 static inline int sb_start_write_trylock(struct super_block
*sb
)
1409 return __sb_start_write(sb
, SB_FREEZE_WRITE
, false);
1413 * sb_start_pagefault - get write access to a superblock from a page fault
1414 * @sb: the super we write to
1416 * When a process starts handling write page fault, it should embed the
1417 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1418 * exclusion against file system freezing. This is needed since the page fault
1419 * is going to dirty a page. This function increments number of running page
1420 * faults preventing freezing. If the file system is already frozen, the
1421 * function waits until the file system is thawed.
1423 * Since page fault freeze protection behaves as a lock, users have to preserve
1424 * ordering of freeze protection and other filesystem locks. It is advised to
1425 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1426 * handling code implies lock dependency:
1429 * -> sb_start_pagefault
1431 static inline void sb_start_pagefault(struct super_block
*sb
)
1433 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
, true);
1437 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1438 * @sb: the super we write to
1440 * This is the third level of protection against filesystem freezing. It is
1441 * free for use by a filesystem. The only requirement is that it must rank
1442 * below sb_start_pagefault.
1444 * For example filesystem can call sb_start_intwrite() when starting a
1445 * transaction which somewhat eases handling of freezing for internal sources
1446 * of filesystem changes (internal fs threads, discarding preallocation on file
1449 static inline void sb_start_intwrite(struct super_block
*sb
)
1451 __sb_start_write(sb
, SB_FREEZE_FS
, true);
1455 extern bool inode_owner_or_capable(const struct inode
*inode
);
1458 * VFS helper functions..
1460 extern int vfs_create(struct inode
*, struct dentry
*, umode_t
, bool);
1461 extern int vfs_mkdir(struct inode
*, struct dentry
*, umode_t
);
1462 extern int vfs_mknod(struct inode
*, struct dentry
*, umode_t
, dev_t
);
1463 extern int vfs_symlink(struct inode
*, struct dentry
*, const char *);
1464 extern int vfs_link(struct dentry
*, struct inode
*, struct dentry
*, struct inode
**);
1465 extern int vfs_rmdir(struct inode
*, struct dentry
*);
1466 extern int vfs_unlink(struct inode
*, struct dentry
*, struct inode
**);
1467 extern int vfs_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*, struct inode
**, unsigned int);
1470 * VFS dentry helper functions.
1472 extern void dentry_unhash(struct dentry
*dentry
);
1475 * VFS file helper functions.
1477 extern void inode_init_owner(struct inode
*inode
, const struct inode
*dir
,
1480 * VFS FS_IOC_FIEMAP helper definitions.
1482 struct fiemap_extent_info
{
1483 unsigned int fi_flags
; /* Flags as passed from user */
1484 unsigned int fi_extents_mapped
; /* Number of mapped extents */
1485 unsigned int fi_extents_max
; /* Size of fiemap_extent array */
1486 struct fiemap_extent __user
*fi_extents_start
; /* Start of
1487 fiemap_extent array */
1489 int fiemap_fill_next_extent(struct fiemap_extent_info
*info
, u64 logical
,
1490 u64 phys
, u64 len
, u32 flags
);
1491 int fiemap_check_flags(struct fiemap_extent_info
*fieinfo
, u32 fs_flags
);
1496 * NOTE! These match bits 12..15 of stat.st_mode
1497 * (ie "(i_mode >> 12) & 15").
1499 #define DT_UNKNOWN 0
1510 * This is the "filldir" function type, used by readdir() to let
1511 * the kernel specify what kind of dirent layout it wants to have.
1512 * This allows the kernel to read directories into kernel space or
1513 * to have different dirent layouts depending on the binary type.
1515 typedef int (*filldir_t
)(void *, const char *, int, loff_t
, u64
, unsigned);
1516 struct dir_context
{
1517 const filldir_t actor
;
1521 struct block_device_operations
;
1523 /* These macros are for out of kernel modules to test that
1524 * the kernel supports the unlocked_ioctl and compat_ioctl
1525 * fields in struct file_operations. */
1526 #define HAVE_COMPAT_IOCTL 1
1527 #define HAVE_UNLOCKED_IOCTL 1
1529 struct file_operations
{
1530 struct module
*owner
;
1531 loff_t (*llseek
) (struct file
*, loff_t
, int);
1532 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
1533 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
1534 ssize_t (*aio_read
) (struct kiocb
*, const struct iovec
*, unsigned long, loff_t
);
1535 ssize_t (*aio_write
) (struct kiocb
*, const struct iovec
*, unsigned long, loff_t
);
1536 int (*iterate
) (struct file
*, struct dir_context
*);
1537 unsigned int (*poll
) (struct file
*, struct poll_table_struct
*);
1538 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
1539 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
1540 int (*mmap
) (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
**);
1555 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
1557 int (*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);
1594 } ____cacheline_aligned
;
1596 ssize_t
rw_copy_check_uvector(int type
, const struct iovec __user
* uvector
,
1597 unsigned long nr_segs
, unsigned long fast_segs
,
1598 struct iovec
*fast_pointer
,
1599 struct iovec
**ret_pointer
);
1601 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1602 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1603 extern ssize_t
vfs_readv(struct file
*, const struct iovec __user
*,
1604 unsigned long, loff_t
*);
1605 extern ssize_t
vfs_writev(struct file
*, const struct iovec __user
*,
1606 unsigned long, loff_t
*);
1608 struct super_operations
{
1609 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
1610 void (*destroy_inode
)(struct inode
*);
1612 void (*dirty_inode
) (struct inode
*, int flags
);
1613 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
1614 int (*drop_inode
) (struct inode
*);
1615 void (*evict_inode
) (struct inode
*);
1616 void (*put_super
) (struct super_block
*);
1617 int (*sync_fs
)(struct super_block
*sb
, int wait
);
1618 int (*freeze_fs
) (struct super_block
*);
1619 int (*unfreeze_fs
) (struct super_block
*);
1620 int (*statfs
) (struct dentry
*, struct kstatfs
*);
1621 int (*remount_fs
) (struct super_block
*, int *, char *);
1622 void (*umount_begin
) (struct super_block
*);
1624 int (*show_options
)(struct seq_file
*, struct dentry
*);
1625 int (*show_devname
)(struct seq_file
*, struct dentry
*);
1626 int (*show_path
)(struct seq_file
*, struct dentry
*);
1627 int (*show_stats
)(struct seq_file
*, struct dentry
*);
1629 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
1630 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
1632 int (*bdev_try_to_free_page
)(struct super_block
*, struct page
*, gfp_t
);
1633 long (*nr_cached_objects
)(struct super_block
*, int);
1634 long (*free_cached_objects
)(struct super_block
*, long, int);
1638 * Inode flags - they have no relation to superblock flags now
1640 #define S_SYNC 1 /* Writes are synced at once */
1641 #define S_NOATIME 2 /* Do not update access times */
1642 #define S_APPEND 4 /* Append-only file */
1643 #define S_IMMUTABLE 8 /* Immutable file */
1644 #define S_DEAD 16 /* removed, but still open directory */
1645 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1646 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1647 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1648 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1649 #define S_PRIVATE 512 /* Inode is fs-internal */
1650 #define S_IMA 1024 /* Inode has an associated IMA struct */
1651 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1652 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1655 * Note that nosuid etc flags are inode-specific: setting some file-system
1656 * flags just means all the inodes inherit those flags by default. It might be
1657 * possible to override it selectively if you really wanted to with some
1658 * ioctl() that is not currently implemented.
1660 * Exception: MS_RDONLY is always applied to the entire file system.
1662 * Unfortunately, it is possible to change a filesystems flags with it mounted
1663 * with files in use. This means that all of the inodes will not have their
1664 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1665 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1667 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1669 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1670 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1671 ((inode)->i_flags & S_SYNC))
1672 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1673 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1674 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1675 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1676 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1678 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1679 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1680 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1681 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1683 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1684 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1685 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1686 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1687 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1688 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1689 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1692 * Inode state bits. Protected by inode->i_lock
1694 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1695 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1697 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1698 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1699 * various stages of removing an inode.
1701 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1703 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1704 * fdatasync(). i_atime is the usual cause.
1705 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1706 * these changes separately from I_DIRTY_SYNC so that we
1707 * don't have to write inode on fdatasync() when only
1708 * mtime has changed in it.
1709 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1710 * I_NEW Serves as both a mutex and completion notification.
1711 * New inodes set I_NEW. If two processes both create
1712 * the same inode, one of them will release its inode and
1713 * wait for I_NEW to be released before returning.
1714 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1715 * also cause waiting on I_NEW, without I_NEW actually
1716 * being set. find_inode() uses this to prevent returning
1717 * nearly-dead inodes.
1718 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1719 * is zero. I_FREEING must be set when I_WILL_FREE is
1721 * I_FREEING Set when inode is about to be freed but still has dirty
1722 * pages or buffers attached or the inode itself is still
1724 * I_CLEAR Added by clear_inode(). In this state the inode is
1725 * clean and can be destroyed. Inode keeps I_FREEING.
1727 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1728 * prohibited for many purposes. iget() must wait for
1729 * the inode to be completely released, then create it
1730 * anew. Other functions will just ignore such inodes,
1731 * if appropriate. I_NEW is used for waiting.
1733 * I_SYNC Writeback of inode is running. The bit is set during
1734 * data writeback, and cleared with a wakeup on the bit
1735 * address once it is done. The bit is also used to pin
1736 * the inode in memory for flusher thread.
1738 * I_REFERENCED Marks the inode as recently references on the LRU list.
1740 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1742 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1744 #define I_DIRTY_SYNC (1 << 0)
1745 #define I_DIRTY_DATASYNC (1 << 1)
1746 #define I_DIRTY_PAGES (1 << 2)
1748 #define I_NEW (1 << __I_NEW)
1749 #define I_WILL_FREE (1 << 4)
1750 #define I_FREEING (1 << 5)
1751 #define I_CLEAR (1 << 6)
1753 #define I_SYNC (1 << __I_SYNC)
1754 #define I_REFERENCED (1 << 8)
1755 #define __I_DIO_WAKEUP 9
1756 #define I_DIO_WAKEUP (1 << I_DIO_WAKEUP)
1757 #define I_LINKABLE (1 << 10)
1759 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1761 extern void __mark_inode_dirty(struct inode
*, int);
1762 static inline void mark_inode_dirty(struct inode
*inode
)
1764 __mark_inode_dirty(inode
, I_DIRTY
);
1767 static inline void mark_inode_dirty_sync(struct inode
*inode
)
1769 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
1772 extern void inc_nlink(struct inode
*inode
);
1773 extern void drop_nlink(struct inode
*inode
);
1774 extern void clear_nlink(struct inode
*inode
);
1775 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
1777 static inline void inode_inc_link_count(struct inode
*inode
)
1780 mark_inode_dirty(inode
);
1783 static inline void inode_dec_link_count(struct inode
*inode
)
1786 mark_inode_dirty(inode
);
1790 * inode_inc_iversion - increments i_version
1791 * @inode: inode that need to be updated
1793 * Every time the inode is modified, the i_version field will be incremented.
1794 * The filesystem has to be mounted with i_version flag
1797 static inline void inode_inc_iversion(struct inode
*inode
)
1799 spin_lock(&inode
->i_lock
);
1801 spin_unlock(&inode
->i_lock
);
1804 enum file_time_flags
{
1811 extern void touch_atime(const struct path
*);
1812 static inline void file_accessed(struct file
*file
)
1814 if (!(file
->f_flags
& O_NOATIME
))
1815 touch_atime(&file
->f_path
);
1818 int sync_inode(struct inode
*inode
, struct writeback_control
*wbc
);
1819 int sync_inode_metadata(struct inode
*inode
, int wait
);
1821 struct file_system_type
{
1824 #define FS_REQUIRES_DEV 1
1825 #define FS_BINARY_MOUNTDATA 2
1826 #define FS_HAS_SUBTYPE 4
1827 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
1828 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
1829 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
1830 struct dentry
*(*mount
) (struct file_system_type
*, int,
1831 const char *, void *);
1832 void (*kill_sb
) (struct super_block
*);
1833 struct module
*owner
;
1834 struct file_system_type
* next
;
1835 struct hlist_head fs_supers
;
1837 struct lock_class_key s_lock_key
;
1838 struct lock_class_key s_umount_key
;
1839 struct lock_class_key s_vfs_rename_key
;
1840 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
1842 struct lock_class_key i_lock_key
;
1843 struct lock_class_key i_mutex_key
;
1844 struct lock_class_key i_mutex_dir_key
;
1847 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1849 extern struct dentry
*mount_ns(struct file_system_type
*fs_type
, int flags
,
1850 void *data
, int (*fill_super
)(struct super_block
*, void *, int));
1851 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
1852 int flags
, const char *dev_name
, void *data
,
1853 int (*fill_super
)(struct super_block
*, void *, int));
1854 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
1855 int flags
, void *data
,
1856 int (*fill_super
)(struct super_block
*, void *, int));
1857 extern struct dentry
*mount_nodev(struct file_system_type
*fs_type
,
1858 int flags
, void *data
,
1859 int (*fill_super
)(struct super_block
*, void *, int));
1860 extern struct dentry
*mount_subtree(struct vfsmount
*mnt
, const char *path
);
1861 void generic_shutdown_super(struct super_block
*sb
);
1862 void kill_block_super(struct super_block
*sb
);
1863 void kill_anon_super(struct super_block
*sb
);
1864 void kill_litter_super(struct super_block
*sb
);
1865 void deactivate_super(struct super_block
*sb
);
1866 void deactivate_locked_super(struct super_block
*sb
);
1867 int set_anon_super(struct super_block
*s
, void *data
);
1868 int get_anon_bdev(dev_t
*);
1869 void free_anon_bdev(dev_t
);
1870 struct super_block
*sget(struct file_system_type
*type
,
1871 int (*test
)(struct super_block
*,void *),
1872 int (*set
)(struct super_block
*,void *),
1873 int flags
, void *data
);
1874 extern struct dentry
*mount_pseudo(struct file_system_type
*, char *,
1875 const struct super_operations
*ops
,
1876 const struct dentry_operations
*dops
,
1879 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1880 #define fops_get(fops) \
1881 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1882 #define fops_put(fops) \
1883 do { if (fops) module_put((fops)->owner); } while(0)
1885 * This one is to be used *ONLY* from ->open() instances.
1886 * fops must be non-NULL, pinned down *and* module dependencies
1887 * should be sufficient to pin the caller down as well.
1889 #define replace_fops(f, fops) \
1891 struct file *__file = (f); \
1892 fops_put(__file->f_op); \
1893 BUG_ON(!(__file->f_op = (fops))); \
1896 extern int register_filesystem(struct file_system_type
*);
1897 extern int unregister_filesystem(struct file_system_type
*);
1898 extern struct vfsmount
*kern_mount_data(struct file_system_type
*, void *data
);
1899 #define kern_mount(type) kern_mount_data(type, NULL)
1900 extern void kern_unmount(struct vfsmount
*mnt
);
1901 extern int may_umount_tree(struct vfsmount
*);
1902 extern int may_umount(struct vfsmount
*);
1903 extern long do_mount(const char *, const char *, const char *, unsigned long, void *);
1904 extern struct vfsmount
*collect_mounts(struct path
*);
1905 extern void drop_collected_mounts(struct vfsmount
*);
1906 extern int iterate_mounts(int (*)(struct vfsmount
*, void *), void *,
1908 extern int vfs_statfs(struct path
*, struct kstatfs
*);
1909 extern int user_statfs(const char __user
*, struct kstatfs
*);
1910 extern int fd_statfs(int, struct kstatfs
*);
1911 extern int vfs_ustat(dev_t
, struct kstatfs
*);
1912 extern int freeze_super(struct super_block
*super
);
1913 extern int thaw_super(struct super_block
*super
);
1914 extern bool our_mnt(struct vfsmount
*mnt
);
1915 extern bool fs_fully_visible(struct file_system_type
*);
1917 extern int current_umask(void);
1919 extern void ihold(struct inode
* inode
);
1920 extern void iput(struct inode
*);
1922 static inline struct inode
*file_inode(struct file
*f
)
1928 extern struct kobject
*fs_kobj
;
1930 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1932 #define FLOCK_VERIFY_READ 1
1933 #define FLOCK_VERIFY_WRITE 2
1935 #ifdef CONFIG_FILE_LOCKING
1936 extern int locks_mandatory_locked(struct file
*);
1937 extern int locks_mandatory_area(int, struct inode
*, struct file
*, loff_t
, size_t);
1940 * Candidates for mandatory locking have the setgid bit set
1941 * but no group execute bit - an otherwise meaningless combination.
1944 static inline int __mandatory_lock(struct inode
*ino
)
1946 return (ino
->i_mode
& (S_ISGID
| S_IXGRP
)) == S_ISGID
;
1950 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1951 * otherwise these will be advisory locks
1954 static inline int mandatory_lock(struct inode
*ino
)
1956 return IS_MANDLOCK(ino
) && __mandatory_lock(ino
);
1959 static inline int locks_verify_locked(struct file
*file
)
1961 if (mandatory_lock(file_inode(file
)))
1962 return locks_mandatory_locked(file
);
1966 static inline int locks_verify_truncate(struct inode
*inode
,
1970 if (inode
->i_flock
&& mandatory_lock(inode
))
1971 return locks_mandatory_area(
1972 FLOCK_VERIFY_WRITE
, inode
, filp
,
1973 size
< inode
->i_size
? size
: inode
->i_size
,
1974 (size
< inode
->i_size
? inode
->i_size
- size
1975 : size
- inode
->i_size
)
1980 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
1983 * Since this check is lockless, we must ensure that any refcounts
1984 * taken are done before checking inode->i_flock. Otherwise, we could
1985 * end up racing with tasks trying to set a new lease on this file.
1989 return __break_lease(inode
, mode
, FL_LEASE
);
1993 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
1996 return __break_lease(inode
, mode
, FL_DELEG
);
2000 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2004 ret
= break_deleg(inode
, O_WRONLY
|O_NONBLOCK
);
2005 if (ret
== -EWOULDBLOCK
&& delegated_inode
) {
2006 *delegated_inode
= inode
;
2012 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2016 ret
= break_deleg(*delegated_inode
, O_WRONLY
);
2017 iput(*delegated_inode
);
2018 *delegated_inode
= NULL
;
2022 #else /* !CONFIG_FILE_LOCKING */
2023 static inline int locks_mandatory_locked(struct file
*file
)
2028 static inline int locks_mandatory_area(int rw
, struct inode
*inode
,
2029 struct file
*filp
, loff_t offset
,
2035 static inline int __mandatory_lock(struct inode
*inode
)
2040 static inline int mandatory_lock(struct inode
*inode
)
2045 static inline int locks_verify_locked(struct file
*file
)
2050 static inline int locks_verify_truncate(struct inode
*inode
, struct file
*filp
,
2056 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2061 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2066 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2071 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2077 #endif /* CONFIG_FILE_LOCKING */
2082 const char *name
; /* pointer to actual string */
2083 const __user
char *uptr
; /* original userland pointer */
2084 struct audit_names
*aname
;
2085 bool separate
; /* should "name" be freed? */
2088 extern long vfs_truncate(struct path
*, loff_t
);
2089 extern int do_truncate(struct dentry
*, loff_t start
, unsigned int time_attrs
,
2091 extern int do_fallocate(struct file
*file
, int mode
, loff_t offset
,
2093 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
2095 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
2096 extern struct file
*filp_open(const char *, int, umode_t
);
2097 extern struct file
*file_open_root(struct dentry
*, struct vfsmount
*,
2099 extern struct file
* dentry_open(const struct path
*, int, const struct cred
*);
2100 extern int filp_close(struct file
*, fl_owner_t id
);
2102 extern struct filename
*getname(const char __user
*);
2103 extern struct filename
*getname_kernel(const char *);
2109 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
2110 int (*open
)(struct inode
*, struct file
*),
2112 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
2116 extern int ioctl_preallocate(struct file
*filp
, void __user
*argp
);
2119 extern void __init
vfs_caches_init_early(void);
2120 extern void __init
vfs_caches_init(unsigned long);
2122 extern struct kmem_cache
*names_cachep
;
2124 extern void final_putname(struct filename
*name
);
2126 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2127 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2128 #ifndef CONFIG_AUDITSYSCALL
2129 #define putname(name) final_putname(name)
2131 extern void putname(struct filename
*name
);
2135 extern int register_blkdev(unsigned int, const char *);
2136 extern void unregister_blkdev(unsigned int, const char *);
2137 extern struct block_device
*bdget(dev_t
);
2138 extern struct block_device
*bdgrab(struct block_device
*bdev
);
2139 extern void bd_set_size(struct block_device
*, loff_t size
);
2140 extern void bd_forget(struct inode
*inode
);
2141 extern void bdput(struct block_device
*);
2142 extern void invalidate_bdev(struct block_device
*);
2143 extern void iterate_bdevs(void (*)(struct block_device
*, void *), void *);
2144 extern int sync_blockdev(struct block_device
*bdev
);
2145 extern void kill_bdev(struct block_device
*);
2146 extern struct super_block
*freeze_bdev(struct block_device
*);
2147 extern void emergency_thaw_all(void);
2148 extern int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
);
2149 extern int fsync_bdev(struct block_device
*);
2150 extern int sb_is_blkdev_sb(struct super_block
*sb
);
2152 static inline void bd_forget(struct inode
*inode
) {}
2153 static inline int sync_blockdev(struct block_device
*bdev
) { return 0; }
2154 static inline void kill_bdev(struct block_device
*bdev
) {}
2155 static inline void invalidate_bdev(struct block_device
*bdev
) {}
2157 static inline struct super_block
*freeze_bdev(struct block_device
*sb
)
2162 static inline int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
)
2167 static inline void iterate_bdevs(void (*f
)(struct block_device
*, void *), void *arg
)
2171 static inline int sb_is_blkdev_sb(struct super_block
*sb
)
2176 extern int sync_filesystem(struct super_block
*);
2177 extern const struct file_operations def_blk_fops
;
2178 extern const struct file_operations def_chr_fops
;
2179 extern const struct file_operations bad_sock_fops
;
2181 extern int ioctl_by_bdev(struct block_device
*, unsigned, unsigned long);
2182 extern int blkdev_ioctl(struct block_device
*, fmode_t
, unsigned, unsigned long);
2183 extern long compat_blkdev_ioctl(struct file
*, unsigned, unsigned long);
2184 extern int blkdev_get(struct block_device
*bdev
, fmode_t mode
, void *holder
);
2185 extern struct block_device
*blkdev_get_by_path(const char *path
, fmode_t mode
,
2187 extern struct block_device
*blkdev_get_by_dev(dev_t dev
, fmode_t mode
,
2189 extern void blkdev_put(struct block_device
*bdev
, fmode_t mode
);
2191 extern int bd_link_disk_holder(struct block_device
*bdev
, struct gendisk
*disk
);
2192 extern void bd_unlink_disk_holder(struct block_device
*bdev
,
2193 struct gendisk
*disk
);
2195 static inline int bd_link_disk_holder(struct block_device
*bdev
,
2196 struct gendisk
*disk
)
2200 static inline void bd_unlink_disk_holder(struct block_device
*bdev
,
2201 struct gendisk
*disk
)
2208 #define CHRDEV_MAJOR_HASH_SIZE 255
2209 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
2210 extern int register_chrdev_region(dev_t
, unsigned, const char *);
2211 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
2212 unsigned int count
, const char *name
,
2213 const struct file_operations
*fops
);
2214 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
2215 unsigned int count
, const char *name
);
2216 extern void unregister_chrdev_region(dev_t
, unsigned);
2217 extern void chrdev_show(struct seq_file
*,off_t
);
2219 static inline int register_chrdev(unsigned int major
, const char *name
,
2220 const struct file_operations
*fops
)
2222 return __register_chrdev(major
, 0, 256, name
, fops
);
2225 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2227 __unregister_chrdev(major
, 0, 256, name
);
2230 /* fs/block_dev.c */
2231 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2232 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2235 #define BLKDEV_MAJOR_HASH_SIZE 255
2236 extern const char *__bdevname(dev_t
, char *buffer
);
2237 extern const char *bdevname(struct block_device
*bdev
, char *buffer
);
2238 extern struct block_device
*lookup_bdev(const char *);
2239 extern void blkdev_show(struct seq_file
*,off_t
);
2242 #define BLKDEV_MAJOR_HASH_SIZE 0
2245 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2247 /* Invalid inode operations -- fs/bad_inode.c */
2248 extern void make_bad_inode(struct inode
*);
2249 extern int is_bad_inode(struct inode
*);
2253 * return READ, READA, or WRITE
2255 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2258 * return data direction, READ or WRITE
2260 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2262 extern void check_disk_size_change(struct gendisk
*disk
,
2263 struct block_device
*bdev
);
2264 extern int revalidate_disk(struct gendisk
*);
2265 extern int check_disk_change(struct block_device
*);
2266 extern int __invalidate_device(struct block_device
*, bool);
2267 extern int invalidate_partition(struct gendisk
*, int);
2269 unsigned long invalidate_mapping_pages(struct address_space
*mapping
,
2270 pgoff_t start
, pgoff_t end
);
2272 static inline void invalidate_remote_inode(struct inode
*inode
)
2274 if (S_ISREG(inode
->i_mode
) || S_ISDIR(inode
->i_mode
) ||
2275 S_ISLNK(inode
->i_mode
))
2276 invalidate_mapping_pages(inode
->i_mapping
, 0, -1);
2278 extern int invalidate_inode_pages2(struct address_space
*mapping
);
2279 extern int invalidate_inode_pages2_range(struct address_space
*mapping
,
2280 pgoff_t start
, pgoff_t end
);
2281 extern int write_inode_now(struct inode
*, int);
2282 extern int filemap_fdatawrite(struct address_space
*);
2283 extern int filemap_flush(struct address_space
*);
2284 extern int filemap_fdatawait(struct address_space
*);
2285 extern int filemap_fdatawait_range(struct address_space
*, loff_t lstart
,
2287 extern int filemap_write_and_wait(struct address_space
*mapping
);
2288 extern int filemap_write_and_wait_range(struct address_space
*mapping
,
2289 loff_t lstart
, loff_t lend
);
2290 extern int __filemap_fdatawrite_range(struct address_space
*mapping
,
2291 loff_t start
, loff_t end
, int sync_mode
);
2292 extern int filemap_fdatawrite_range(struct address_space
*mapping
,
2293 loff_t start
, loff_t end
);
2295 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
2297 extern int vfs_fsync(struct file
*file
, int datasync
);
2298 static inline int generic_write_sync(struct file
*file
, loff_t pos
, loff_t count
)
2300 if (!(file
->f_flags
& O_DSYNC
) && !IS_SYNC(file
->f_mapping
->host
))
2302 return vfs_fsync_range(file
, pos
, pos
+ count
- 1,
2303 (file
->f_flags
& __O_SYNC
) ? 0 : 1);
2305 extern void emergency_sync(void);
2306 extern void emergency_remount(void);
2308 extern sector_t
bmap(struct inode
*, sector_t
);
2310 extern int notify_change(struct dentry
*, struct iattr
*, struct inode
**);
2311 extern int inode_permission(struct inode
*, int);
2312 extern int generic_permission(struct inode
*, int);
2314 static inline bool execute_ok(struct inode
*inode
)
2316 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
2319 static inline void file_start_write(struct file
*file
)
2321 if (!S_ISREG(file_inode(file
)->i_mode
))
2323 __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, true);
2326 static inline bool file_start_write_trylock(struct file
*file
)
2328 if (!S_ISREG(file_inode(file
)->i_mode
))
2330 return __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, false);
2333 static inline void file_end_write(struct file
*file
)
2335 if (!S_ISREG(file_inode(file
)->i_mode
))
2337 __sb_end_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
);
2341 * get_write_access() gets write permission for a file.
2342 * put_write_access() releases this write permission.
2343 * This is used for regular files.
2344 * We cannot support write (and maybe mmap read-write shared) accesses and
2345 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2346 * can have the following values:
2347 * 0: no writers, no VM_DENYWRITE mappings
2348 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2349 * > 0: (i_writecount) users are writing to the file.
2351 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2352 * except for the cases where we don't hold i_writecount yet. Then we need to
2353 * use {get,deny}_write_access() - these functions check the sign and refuse
2354 * to do the change if sign is wrong.
2356 static inline int get_write_access(struct inode
*inode
)
2358 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2360 static inline int deny_write_access(struct file
*file
)
2362 struct inode
*inode
= file_inode(file
);
2363 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2365 static inline void put_write_access(struct inode
* inode
)
2367 atomic_dec(&inode
->i_writecount
);
2369 static inline void allow_write_access(struct file
*file
)
2372 atomic_inc(&file_inode(file
)->i_writecount
);
2374 static inline bool inode_is_open_for_write(const struct inode
*inode
)
2376 return atomic_read(&inode
->i_writecount
) > 0;
2380 static inline void i_readcount_dec(struct inode
*inode
)
2382 BUG_ON(!atomic_read(&inode
->i_readcount
));
2383 atomic_dec(&inode
->i_readcount
);
2385 static inline void i_readcount_inc(struct inode
*inode
)
2387 atomic_inc(&inode
->i_readcount
);
2390 static inline void i_readcount_dec(struct inode
*inode
)
2394 static inline void i_readcount_inc(struct inode
*inode
)
2399 extern int do_pipe_flags(int *, int);
2401 extern int kernel_read(struct file
*, loff_t
, char *, unsigned long);
2402 extern ssize_t
kernel_write(struct file
*, const char *, size_t, loff_t
);
2403 extern struct file
* open_exec(const char *);
2405 /* fs/dcache.c -- generic fs support functions */
2406 extern int is_subdir(struct dentry
*, struct dentry
*);
2407 extern int path_is_under(struct path
*, struct path
*);
2409 #include <linux/err.h>
2411 /* needed for stackable file system support */
2412 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int whence
);
2414 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int whence
);
2416 extern int inode_init_always(struct super_block
*, struct inode
*);
2417 extern void inode_init_once(struct inode
*);
2418 extern void address_space_init_once(struct address_space
*mapping
);
2419 extern struct inode
* igrab(struct inode
*);
2420 extern ino_t
iunique(struct super_block
*, ino_t
);
2421 extern int inode_needs_sync(struct inode
*inode
);
2422 extern int generic_delete_inode(struct inode
*inode
);
2423 static inline int generic_drop_inode(struct inode
*inode
)
2425 return !inode
->i_nlink
|| inode_unhashed(inode
);
2428 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
2429 unsigned long hashval
, int (*test
)(struct inode
*, void *),
2431 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
2432 int (*test
)(struct inode
*, void *), void *data
);
2433 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
2435 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
2436 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
2437 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
2438 extern int insert_inode_locked(struct inode
*);
2439 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2440 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
2442 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
2444 extern void unlock_new_inode(struct inode
*);
2445 extern unsigned int get_next_ino(void);
2447 extern void __iget(struct inode
* inode
);
2448 extern void iget_failed(struct inode
*);
2449 extern void clear_inode(struct inode
*);
2450 extern void __destroy_inode(struct inode
*);
2451 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
2452 extern struct inode
*new_inode(struct super_block
*sb
);
2453 extern void free_inode_nonrcu(struct inode
*inode
);
2454 extern int should_remove_suid(struct dentry
*);
2455 extern int file_remove_suid(struct file
*);
2457 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
2458 static inline void insert_inode_hash(struct inode
*inode
)
2460 __insert_inode_hash(inode
, inode
->i_ino
);
2463 extern void __remove_inode_hash(struct inode
*);
2464 static inline void remove_inode_hash(struct inode
*inode
)
2466 if (!inode_unhashed(inode
))
2467 __remove_inode_hash(inode
);
2470 extern void inode_sb_list_add(struct inode
*inode
);
2473 extern void submit_bio(int, struct bio
*);
2474 extern int bdev_read_only(struct block_device
*);
2476 extern int set_blocksize(struct block_device
*, int);
2477 extern int sb_set_blocksize(struct super_block
*, int);
2478 extern int sb_min_blocksize(struct super_block
*, int);
2480 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
2481 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
2482 extern int generic_file_remap_pages(struct vm_area_struct
*, unsigned long addr
,
2483 unsigned long size
, pgoff_t pgoff
);
2484 extern int file_read_actor(read_descriptor_t
* desc
, struct page
*page
, unsigned long offset
, unsigned long size
);
2485 int generic_write_checks(struct file
*file
, loff_t
*pos
, size_t *count
, int isblk
);
2486 extern ssize_t
generic_file_aio_read(struct kiocb
*, const struct iovec
*, unsigned long, loff_t
);
2487 extern ssize_t
__generic_file_aio_write(struct kiocb
*, const struct iovec
*, unsigned long,
2489 extern ssize_t
generic_file_aio_write(struct kiocb
*, const struct iovec
*, unsigned long, loff_t
);
2490 extern ssize_t
generic_file_direct_write(struct kiocb
*, const struct iovec
*,
2491 unsigned long *, loff_t
, loff_t
*, size_t, size_t);
2492 extern ssize_t
generic_file_buffered_write(struct kiocb
*, const struct iovec
*,
2493 unsigned long, loff_t
, loff_t
*, size_t, ssize_t
);
2494 extern ssize_t
do_sync_read(struct file
*filp
, char __user
*buf
, size_t len
, loff_t
*ppos
);
2495 extern ssize_t
do_sync_write(struct file
*filp
, const char __user
*buf
, size_t len
, loff_t
*ppos
);
2496 extern int generic_segment_checks(const struct iovec
*iov
,
2497 unsigned long *nr_segs
, size_t *count
, int access_flags
);
2499 /* fs/block_dev.c */
2500 extern ssize_t
blkdev_aio_write(struct kiocb
*iocb
, const struct iovec
*iov
,
2501 unsigned long nr_segs
, loff_t pos
);
2502 extern int blkdev_fsync(struct file
*filp
, loff_t start
, loff_t end
,
2504 extern void block_sync_page(struct page
*page
);
2507 extern ssize_t
generic_file_splice_read(struct file
*, loff_t
*,
2508 struct pipe_inode_info
*, size_t, unsigned int);
2509 extern ssize_t
default_file_splice_read(struct file
*, loff_t
*,
2510 struct pipe_inode_info
*, size_t, unsigned int);
2511 extern ssize_t
generic_file_splice_write(struct pipe_inode_info
*,
2512 struct file
*, loff_t
*, size_t, unsigned int);
2513 extern ssize_t
generic_splice_sendpage(struct pipe_inode_info
*pipe
,
2514 struct file
*out
, loff_t
*, size_t len
, unsigned int flags
);
2517 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
2518 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int whence
);
2519 extern loff_t
no_llseek(struct file
*file
, loff_t offset
, int whence
);
2520 extern loff_t
vfs_setpos(struct file
*file
, loff_t offset
, loff_t maxsize
);
2521 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int whence
);
2522 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
2523 int whence
, loff_t maxsize
, loff_t eof
);
2524 extern loff_t
fixed_size_llseek(struct file
*file
, loff_t offset
,
2525 int whence
, loff_t size
);
2526 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
2527 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
2529 #ifdef CONFIG_FS_XIP
2530 extern ssize_t
xip_file_read(struct file
*filp
, char __user
*buf
, size_t len
,
2532 extern int xip_file_mmap(struct file
* file
, struct vm_area_struct
* vma
);
2533 extern ssize_t
xip_file_write(struct file
*filp
, const char __user
*buf
,
2534 size_t len
, loff_t
*ppos
);
2535 extern int xip_truncate_page(struct address_space
*mapping
, loff_t from
);
2537 static inline int xip_truncate_page(struct address_space
*mapping
, loff_t from
)
2544 typedef void (dio_submit_t
)(int rw
, struct bio
*bio
, struct inode
*inode
,
2545 loff_t file_offset
);
2548 /* need locking between buffered and direct access */
2551 /* filesystem does not support filling holes */
2552 DIO_SKIP_HOLES
= 0x02,
2554 /* filesystem can handle aio writes beyond i_size */
2555 DIO_ASYNC_EXTEND
= 0x04,
2558 void dio_end_io(struct bio
*bio
, int error
);
2560 ssize_t
__blockdev_direct_IO(int rw
, struct kiocb
*iocb
, struct inode
*inode
,
2561 struct block_device
*bdev
, const struct iovec
*iov
, loff_t offset
,
2562 unsigned long nr_segs
, get_block_t get_block
, dio_iodone_t end_io
,
2563 dio_submit_t submit_io
, int flags
);
2565 static inline ssize_t
blockdev_direct_IO(int rw
, struct kiocb
*iocb
,
2566 struct inode
*inode
, const struct iovec
*iov
, loff_t offset
,
2567 unsigned long nr_segs
, get_block_t get_block
)
2569 return __blockdev_direct_IO(rw
, iocb
, inode
, inode
->i_sb
->s_bdev
, iov
,
2570 offset
, nr_segs
, get_block
, NULL
, NULL
,
2571 DIO_LOCKING
| DIO_SKIP_HOLES
);
2575 void inode_dio_wait(struct inode
*inode
);
2576 void inode_dio_done(struct inode
*inode
);
2578 extern void inode_set_flags(struct inode
*inode
, unsigned int flags
,
2581 extern const struct file_operations generic_ro_fops
;
2583 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2585 extern int vfs_readlink(struct dentry
*, char __user
*, int, const char *);
2586 extern int page_readlink(struct dentry
*, char __user
*, int);
2587 extern void *page_follow_link_light(struct dentry
*, struct nameidata
*);
2588 extern void page_put_link(struct dentry
*, struct nameidata
*, void *);
2589 extern int __page_symlink(struct inode
*inode
, const char *symname
, int len
,
2591 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
2592 extern const struct inode_operations page_symlink_inode_operations
;
2593 extern void kfree_put_link(struct dentry
*, struct nameidata
*, void *);
2594 extern int generic_readlink(struct dentry
*, char __user
*, int);
2595 extern void generic_fillattr(struct inode
*, struct kstat
*);
2596 int vfs_getattr_nosec(struct path
*path
, struct kstat
*stat
);
2597 extern int vfs_getattr(struct path
*, struct kstat
*);
2598 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2599 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2600 void __inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2601 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2602 loff_t
inode_get_bytes(struct inode
*inode
);
2603 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
2605 extern int vfs_readdir(struct file
*, filldir_t
, void *);
2606 extern int iterate_dir(struct file
*, struct dir_context
*);
2608 extern int vfs_stat(const char __user
*, struct kstat
*);
2609 extern int vfs_lstat(const char __user
*, struct kstat
*);
2610 extern int vfs_fstat(unsigned int, struct kstat
*);
2611 extern int vfs_fstatat(int , const char __user
*, struct kstat
*, int);
2613 extern int do_vfs_ioctl(struct file
*filp
, unsigned int fd
, unsigned int cmd
,
2615 extern int __generic_block_fiemap(struct inode
*inode
,
2616 struct fiemap_extent_info
*fieinfo
,
2617 loff_t start
, loff_t len
,
2618 get_block_t
*get_block
);
2619 extern int generic_block_fiemap(struct inode
*inode
,
2620 struct fiemap_extent_info
*fieinfo
, u64 start
,
2621 u64 len
, get_block_t
*get_block
);
2623 extern void get_filesystem(struct file_system_type
*fs
);
2624 extern void put_filesystem(struct file_system_type
*fs
);
2625 extern struct file_system_type
*get_fs_type(const char *name
);
2626 extern struct super_block
*get_super(struct block_device
*);
2627 extern struct super_block
*get_super_thawed(struct block_device
*);
2628 extern struct super_block
*get_active_super(struct block_device
*bdev
);
2629 extern void drop_super(struct super_block
*sb
);
2630 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
2631 extern void iterate_supers_type(struct file_system_type
*,
2632 void (*)(struct super_block
*, void *), void *);
2634 extern int dcache_dir_open(struct inode
*, struct file
*);
2635 extern int dcache_dir_close(struct inode
*, struct file
*);
2636 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
2637 extern int dcache_readdir(struct file
*, struct dir_context
*);
2638 extern int simple_setattr(struct dentry
*, struct iattr
*);
2639 extern int simple_getattr(struct vfsmount
*, struct dentry
*, struct kstat
*);
2640 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
2641 extern int simple_open(struct inode
*inode
, struct file
*file
);
2642 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
2643 extern int simple_unlink(struct inode
*, struct dentry
*);
2644 extern int simple_rmdir(struct inode
*, struct dentry
*);
2645 extern int simple_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*);
2646 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
2647 extern int simple_empty(struct dentry
*);
2648 extern int simple_readpage(struct file
*file
, struct page
*page
);
2649 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
2650 loff_t pos
, unsigned len
, unsigned flags
,
2651 struct page
**pagep
, void **fsdata
);
2652 extern int simple_write_end(struct file
*file
, struct address_space
*mapping
,
2653 loff_t pos
, unsigned len
, unsigned copied
,
2654 struct page
*page
, void *fsdata
);
2655 extern int always_delete_dentry(const struct dentry
*);
2656 extern struct inode
*alloc_anon_inode(struct super_block
*);
2657 extern const struct dentry_operations simple_dentry_operations
;
2659 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
2660 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
2661 extern const struct file_operations simple_dir_operations
;
2662 extern const struct inode_operations simple_dir_inode_operations
;
2663 struct tree_descr
{ char *name
; const struct file_operations
*ops
; int mode
; };
2664 struct dentry
*d_alloc_name(struct dentry
*, const char *);
2665 extern int simple_fill_super(struct super_block
*, unsigned long, struct tree_descr
*);
2666 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
2667 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
2669 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
2670 loff_t
*ppos
, const void *from
, size_t available
);
2671 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
2672 const void __user
*from
, size_t count
);
2674 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
2676 extern int generic_check_addressable(unsigned, u64
);
2678 #ifdef CONFIG_MIGRATION
2679 extern int buffer_migrate_page(struct address_space
*,
2680 struct page
*, struct page
*,
2683 #define buffer_migrate_page NULL
2686 extern int inode_change_ok(const struct inode
*, struct iattr
*);
2687 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
2688 extern void setattr_copy(struct inode
*inode
, const struct iattr
*attr
);
2690 extern int file_update_time(struct file
*file
);
2692 extern int generic_show_options(struct seq_file
*m
, struct dentry
*root
);
2693 extern void save_mount_options(struct super_block
*sb
, char *options
);
2694 extern void replace_mount_options(struct super_block
*sb
, char *options
);
2696 static inline ino_t
parent_ino(struct dentry
*dentry
)
2701 * Don't strictly need d_lock here? If the parent ino could change
2702 * then surely we'd have a deeper race in the caller?
2704 spin_lock(&dentry
->d_lock
);
2705 res
= dentry
->d_parent
->d_inode
->i_ino
;
2706 spin_unlock(&dentry
->d_lock
);
2710 /* Transaction based IO helpers */
2713 * An argresp is stored in an allocated page and holds the
2714 * size of the argument or response, along with its content
2716 struct simple_transaction_argresp
{
2721 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2723 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
2725 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
2726 size_t size
, loff_t
*pos
);
2727 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
2729 void simple_transaction_set(struct file
*file
, size_t n
);
2732 * simple attribute files
2734 * These attributes behave similar to those in sysfs:
2736 * Writing to an attribute immediately sets a value, an open file can be
2737 * written to multiple times.
2739 * Reading from an attribute creates a buffer from the value that might get
2740 * read with multiple read calls. When the attribute has been read
2741 * completely, no further read calls are possible until the file is opened
2744 * All attributes contain a text representation of a numeric value
2745 * that are accessed with the get() and set() functions.
2747 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2748 static int __fops ## _open(struct inode *inode, struct file *file) \
2750 __simple_attr_check_format(__fmt, 0ull); \
2751 return simple_attr_open(inode, file, __get, __set, __fmt); \
2753 static const struct file_operations __fops = { \
2754 .owner = THIS_MODULE, \
2755 .open = __fops ## _open, \
2756 .release = simple_attr_release, \
2757 .read = simple_attr_read, \
2758 .write = simple_attr_write, \
2759 .llseek = generic_file_llseek, \
2762 static inline __printf(1, 2)
2763 void __simple_attr_check_format(const char *fmt
, ...)
2765 /* don't do anything, just let the compiler check the arguments; */
2768 int simple_attr_open(struct inode
*inode
, struct file
*file
,
2769 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
2771 int simple_attr_release(struct inode
*inode
, struct file
*file
);
2772 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
2773 size_t len
, loff_t
*ppos
);
2774 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
2775 size_t len
, loff_t
*ppos
);
2778 int proc_nr_files(struct ctl_table
*table
, int write
,
2779 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
2780 int proc_nr_dentry(struct ctl_table
*table
, int write
,
2781 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
2782 int proc_nr_inodes(struct ctl_table
*table
, int write
,
2783 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
2784 int __init
get_filesystem_list(char *buf
);
2786 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2787 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2789 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2790 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2791 (flag & __FMODE_NONOTIFY)))
2793 static inline int is_sxid(umode_t mode
)
2795 return (mode
& S_ISUID
) || ((mode
& S_ISGID
) && (mode
& S_IXGRP
));
2798 static inline void inode_has_no_xattr(struct inode
*inode
)
2800 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& MS_NOSEC
))
2801 inode
->i_flags
|= S_NOSEC
;
2804 static inline bool dir_emit(struct dir_context
*ctx
,
2805 const char *name
, int namelen
,
2806 u64 ino
, unsigned type
)
2808 return ctx
->actor(ctx
, name
, namelen
, ctx
->pos
, ino
, type
) == 0;
2810 static inline bool dir_emit_dot(struct file
*file
, struct dir_context
*ctx
)
2812 return ctx
->actor(ctx
, ".", 1, ctx
->pos
,
2813 file
->f_path
.dentry
->d_inode
->i_ino
, DT_DIR
) == 0;
2815 static inline bool dir_emit_dotdot(struct file
*file
, struct dir_context
*ctx
)
2817 return ctx
->actor(ctx
, "..", 2, ctx
->pos
,
2818 parent_ino(file
->f_path
.dentry
), DT_DIR
) == 0;
2820 static inline bool dir_emit_dots(struct file
*file
, struct dir_context
*ctx
)
2822 if (ctx
->pos
== 0) {
2823 if (!dir_emit_dot(file
, ctx
))
2827 if (ctx
->pos
== 1) {
2828 if (!dir_emit_dotdot(file
, ctx
))
2834 static inline bool dir_relax(struct inode
*inode
)
2836 mutex_unlock(&inode
->i_mutex
);
2837 mutex_lock(&inode
->i_mutex
);
2838 return !IS_DEADDIR(inode
);
2841 #endif /* _LINUX_FS_H */