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/radix-tree.h>
14 #include <linux/rbtree.h>
15 #include <linux/init.h>
16 #include <linux/pid.h>
17 #include <linux/bug.h>
18 #include <linux/mutex.h>
19 #include <linux/capability.h>
20 #include <linux/semaphore.h>
21 #include <linux/fiemap.h>
22 #include <linux/rculist_bl.h>
23 #include <linux/atomic.h>
24 #include <linux/shrinker.h>
25 #include <linux/migrate_mode.h>
26 #include <linux/uidgid.h>
27 #include <linux/lockdep.h>
28 #include <linux/percpu-rwsem.h>
29 #include <linux/blk_types.h>
31 #include <asm/byteorder.h>
32 #include <uapi/linux/fs.h>
34 struct export_operations
;
40 struct pipe_inode_info
;
41 struct poll_table_struct
;
43 struct vm_area_struct
;
46 struct swap_info_struct
;
48 extern void __init
inode_init(void);
49 extern void __init
inode_init_early(void);
50 extern void __init
files_init(unsigned long);
52 extern struct files_stat_struct files_stat
;
53 extern unsigned long get_max_files(void);
54 extern int sysctl_nr_open
;
55 extern struct inodes_stat_t inodes_stat
;
56 extern int leases_enable
, lease_break_time
;
57 extern int sysctl_protected_symlinks
;
58 extern int sysctl_protected_hardlinks
;
61 typedef int (get_block_t
)(struct inode
*inode
, sector_t iblock
,
62 struct buffer_head
*bh_result
, int create
);
63 typedef void (dio_iodone_t
)(struct kiocb
*iocb
, loff_t offset
,
64 ssize_t bytes
, void *private, int ret
,
68 * The below are the various read and write types that we support. Some of
69 * them include behavioral modifiers that send information down to the
70 * block layer and IO scheduler. Terminology:
72 * The block layer uses device plugging to defer IO a little bit, in
73 * the hope that we will see more IO very shortly. This increases
74 * coalescing of adjacent IO and thus reduces the number of IOs we
75 * have to send to the device. It also allows for better queuing,
76 * if the IO isn't mergeable. If the caller is going to be waiting
77 * for the IO, then he must ensure that the device is unplugged so
78 * that the IO is dispatched to the driver.
80 * All IO is handled async in Linux. This is fine for background
81 * writes, but for reads or writes that someone waits for completion
82 * on, we want to notify the block layer and IO scheduler so that they
83 * know about it. That allows them to make better scheduling
84 * decisions. So when the below references 'sync' and 'async', it
85 * is referencing this priority hint.
87 * With that in mind, the available types are:
89 * READ A normal read operation. Device will be plugged.
90 * READ_SYNC A synchronous read. Device is not plugged, caller can
91 * immediately wait on this read without caring about
93 * READA Used for read-ahead operations. Lower priority, and the
94 * block layer could (in theory) choose to ignore this
95 * request if it runs into resource problems.
96 * WRITE A normal async write. Device will be plugged.
97 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
98 * the hint that someone will be waiting on this IO
99 * shortly. The write equivalent of READ_SYNC.
100 * WRITE_ODIRECT Special case write for O_DIRECT only.
101 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
102 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
103 * non-volatile media on completion.
104 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
105 * by a cache flush and data is guaranteed to be on
106 * non-volatile media on completion.
109 #define RW_MASK REQ_WRITE
110 #define RWA_MASK REQ_RAHEAD
113 #define WRITE RW_MASK
114 #define READA RWA_MASK
115 #define KERNEL_READ (READ|REQ_KERNEL)
116 #define KERNEL_WRITE (WRITE|REQ_KERNEL)
118 #define READ_SYNC (READ | REQ_SYNC)
119 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
120 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
121 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
122 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
123 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
126 * Attribute flags. These should be or-ed together to figure out what
129 #define ATTR_MODE (1 << 0)
130 #define ATTR_UID (1 << 1)
131 #define ATTR_GID (1 << 2)
132 #define ATTR_SIZE (1 << 3)
133 #define ATTR_ATIME (1 << 4)
134 #define ATTR_MTIME (1 << 5)
135 #define ATTR_CTIME (1 << 6)
136 #define ATTR_ATIME_SET (1 << 7)
137 #define ATTR_MTIME_SET (1 << 8)
138 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
139 #define ATTR_ATTR_FLAG (1 << 10)
140 #define ATTR_KILL_SUID (1 << 11)
141 #define ATTR_KILL_SGID (1 << 12)
142 #define ATTR_FILE (1 << 13)
143 #define ATTR_KILL_PRIV (1 << 14)
144 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
145 #define ATTR_TIMES_SET (1 << 16)
148 * This is the Inode Attributes structure, used for notify_change(). It
149 * uses the above definitions as flags, to know which values have changed.
150 * Also, in this manner, a Filesystem can look at only the values it cares
151 * about. Basically, these are the attributes that the VFS layer can
152 * request to change from the FS layer.
154 * Derek Atkins <warlord@MIT.EDU> 94-10-20
157 unsigned int ia_valid
;
162 struct timespec ia_atime
;
163 struct timespec ia_mtime
;
164 struct timespec ia_ctime
;
167 * Not an attribute, but an auxiliary info for filesystems wanting to
168 * implement an ftruncate() like method. NOTE: filesystem should
169 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
171 struct file
*ia_file
;
175 * Includes for diskquotas.
177 #include <linux/quota.h>
180 * enum positive_aop_returns - aop return codes with specific semantics
182 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
183 * completed, that the page is still locked, and
184 * should be considered active. The VM uses this hint
185 * to return the page to the active list -- it won't
186 * be a candidate for writeback again in the near
187 * future. Other callers must be careful to unlock
188 * the page if they get this return. Returned by
191 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
192 * unlocked it and the page might have been truncated.
193 * The caller should back up to acquiring a new page and
194 * trying again. The aop will be taking reasonable
195 * precautions not to livelock. If the caller held a page
196 * reference, it should drop it before retrying. Returned
199 * address_space_operation functions return these large constants to indicate
200 * special semantics to the caller. These are much larger than the bytes in a
201 * page to allow for functions that return the number of bytes operated on in a
205 enum positive_aop_returns
{
206 AOP_WRITEPAGE_ACTIVATE
= 0x80000,
207 AOP_TRUNCATED_PAGE
= 0x80001,
210 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
211 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
212 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
213 * helper code (eg buffer layer)
214 * to clear GFP_FS from alloc */
217 * oh the beauties of C type declarations.
220 struct address_space
;
221 struct writeback_control
;
224 const struct iovec
*iov
;
225 unsigned long nr_segs
;
230 size_t iov_iter_copy_from_user_atomic(struct page
*page
,
231 struct iov_iter
*i
, unsigned long offset
, size_t bytes
);
232 size_t iov_iter_copy_from_user(struct page
*page
,
233 struct iov_iter
*i
, unsigned long offset
, size_t bytes
);
234 void iov_iter_advance(struct iov_iter
*i
, size_t bytes
);
235 int iov_iter_fault_in_readable(struct iov_iter
*i
, size_t bytes
);
236 size_t iov_iter_single_seg_count(struct iov_iter
*i
);
238 static inline void iov_iter_init(struct iov_iter
*i
,
239 const struct iovec
*iov
, unsigned long nr_segs
,
240 size_t count
, size_t written
)
243 i
->nr_segs
= nr_segs
;
245 i
->count
= count
+ written
;
247 iov_iter_advance(i
, written
);
250 static inline size_t iov_iter_count(struct iov_iter
*i
)
256 * "descriptor" for what we're up to with a read.
257 * This allows us to use the same read code yet
258 * have multiple different users of the data that
259 * we read from a file.
261 * The simplest case just copies the data to user
274 typedef int (*read_actor_t
)(read_descriptor_t
*, struct page
*,
275 unsigned long, unsigned long);
277 struct address_space_operations
{
278 int (*writepage
)(struct page
*page
, struct writeback_control
*wbc
);
279 int (*readpage
)(struct file
*, struct page
*);
281 /* Write back some dirty pages from this mapping. */
282 int (*writepages
)(struct address_space
*, struct writeback_control
*);
284 /* Set a page dirty. Return true if this dirtied it */
285 int (*set_page_dirty
)(struct page
*page
);
287 int (*readpages
)(struct file
*filp
, struct address_space
*mapping
,
288 struct list_head
*pages
, unsigned nr_pages
);
290 int (*write_begin
)(struct file
*, struct address_space
*mapping
,
291 loff_t pos
, unsigned len
, unsigned flags
,
292 struct page
**pagep
, void **fsdata
);
293 int (*write_end
)(struct file
*, struct address_space
*mapping
,
294 loff_t pos
, unsigned len
, unsigned copied
,
295 struct page
*page
, void *fsdata
);
297 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
298 sector_t (*bmap
)(struct address_space
*, sector_t
);
299 void (*invalidatepage
) (struct page
*, unsigned long);
300 int (*releasepage
) (struct page
*, gfp_t
);
301 void (*freepage
)(struct page
*);
302 ssize_t (*direct_IO
)(int, struct kiocb
*, const struct iovec
*iov
,
303 loff_t offset
, unsigned long nr_segs
);
304 int (*get_xip_mem
)(struct address_space
*, pgoff_t
, int,
305 void **, unsigned long *);
307 * migrate the contents of a page to the specified target. If sync
308 * is false, it must not block.
310 int (*migratepage
) (struct address_space
*,
311 struct page
*, struct page
*, enum migrate_mode
);
312 int (*launder_page
) (struct page
*);
313 int (*is_partially_uptodate
) (struct page
*, read_descriptor_t
*,
315 int (*error_remove_page
)(struct address_space
*, struct page
*);
317 /* swapfile support */
318 int (*swap_activate
)(struct swap_info_struct
*sis
, struct file
*file
,
320 void (*swap_deactivate
)(struct file
*file
);
323 extern const struct address_space_operations empty_aops
;
326 * pagecache_write_begin/pagecache_write_end must be used by general code
327 * to write into the pagecache.
329 int pagecache_write_begin(struct file
*, struct address_space
*mapping
,
330 loff_t pos
, unsigned len
, unsigned flags
,
331 struct page
**pagep
, void **fsdata
);
333 int pagecache_write_end(struct file
*, struct address_space
*mapping
,
334 loff_t pos
, unsigned len
, unsigned copied
,
335 struct page
*page
, void *fsdata
);
337 struct backing_dev_info
;
338 struct address_space
{
339 struct inode
*host
; /* owner: inode, block_device */
340 struct radix_tree_root page_tree
; /* radix tree of all pages */
341 spinlock_t tree_lock
; /* and lock protecting it */
342 unsigned int i_mmap_writable
;/* count VM_SHARED mappings */
343 struct rb_root i_mmap
; /* tree of private and shared mappings */
344 struct list_head i_mmap_nonlinear
;/*list VM_NONLINEAR mappings */
345 struct mutex i_mmap_mutex
; /* protect tree, count, list */
346 /* Protected by tree_lock together with the radix tree */
347 unsigned long nrpages
; /* number of total pages */
348 pgoff_t writeback_index
;/* writeback starts here */
349 const struct address_space_operations
*a_ops
; /* methods */
350 unsigned long flags
; /* error bits/gfp mask */
351 struct backing_dev_info
*backing_dev_info
; /* device readahead, etc */
352 spinlock_t private_lock
; /* for use by the address_space */
353 struct list_head private_list
; /* ditto */
354 struct address_space
*assoc_mapping
; /* ditto */
355 } __attribute__((aligned(sizeof(long))));
357 * On most architectures that alignment is already the case; but
358 * must be enforced here for CRIS, to let the least significant bit
359 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
361 struct request_queue
;
363 struct block_device
{
364 dev_t bd_dev
; /* not a kdev_t - it's a search key */
366 struct inode
* bd_inode
; /* will die */
367 struct super_block
* bd_super
;
368 struct mutex bd_mutex
; /* open/close mutex */
369 struct list_head bd_inodes
;
373 bool bd_write_holder
;
375 struct list_head bd_holder_disks
;
377 struct block_device
* bd_contains
;
378 unsigned bd_block_size
;
379 struct hd_struct
* bd_part
;
380 /* number of times partitions within this device have been opened. */
381 unsigned bd_part_count
;
383 struct gendisk
* bd_disk
;
384 struct request_queue
* bd_queue
;
385 struct list_head bd_list
;
387 * Private data. You must have bd_claim'ed the block_device
388 * to use this. NOTE: bd_claim allows an owner to claim
389 * the same device multiple times, the owner must take special
390 * care to not mess up bd_private for that case.
392 unsigned long bd_private
;
394 /* The counter of freeze processes */
395 int bd_fsfreeze_count
;
396 /* Mutex for freeze */
397 struct mutex bd_fsfreeze_mutex
;
398 /* A semaphore that prevents I/O while block size is being changed */
399 struct percpu_rw_semaphore bd_block_size_semaphore
;
403 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
406 #define PAGECACHE_TAG_DIRTY 0
407 #define PAGECACHE_TAG_WRITEBACK 1
408 #define PAGECACHE_TAG_TOWRITE 2
410 int mapping_tagged(struct address_space
*mapping
, int tag
);
413 * Might pages of this file be mapped into userspace?
415 static inline int mapping_mapped(struct address_space
*mapping
)
417 return !RB_EMPTY_ROOT(&mapping
->i_mmap
) ||
418 !list_empty(&mapping
->i_mmap_nonlinear
);
422 * Might pages of this file have been modified in userspace?
423 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
424 * marks vma as VM_SHARED if it is shared, and the file was opened for
425 * writing i.e. vma may be mprotected writable even if now readonly.
427 static inline int mapping_writably_mapped(struct address_space
*mapping
)
429 return mapping
->i_mmap_writable
!= 0;
433 * Use sequence counter to get consistent i_size on 32-bit processors.
435 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
436 #include <linux/seqlock.h>
437 #define __NEED_I_SIZE_ORDERED
438 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
440 #define i_size_ordered_init(inode) do { } while (0)
444 #define ACL_NOT_CACHED ((void *)(-1))
446 #define IOP_FASTPERM 0x0001
447 #define IOP_LOOKUP 0x0002
448 #define IOP_NOFOLLOW 0x0004
451 * Keep mostly read-only and often accessed (especially for
452 * the RCU path lookup and 'stat' data) fields at the beginning
453 * of the 'struct inode'
457 unsigned short i_opflags
;
460 unsigned int i_flags
;
462 #ifdef CONFIG_FS_POSIX_ACL
463 struct posix_acl
*i_acl
;
464 struct posix_acl
*i_default_acl
;
467 const struct inode_operations
*i_op
;
468 struct super_block
*i_sb
;
469 struct address_space
*i_mapping
;
471 #ifdef CONFIG_SECURITY
475 /* Stat data, not accessed from path walking */
478 * Filesystems may only read i_nlink directly. They shall use the
479 * following functions for modification:
481 * (set|clear|inc|drop)_nlink
482 * inode_(inc|dec)_link_count
485 const unsigned int i_nlink
;
486 unsigned int __i_nlink
;
490 struct timespec i_atime
;
491 struct timespec i_mtime
;
492 struct timespec i_ctime
;
493 spinlock_t i_lock
; /* i_blocks, i_bytes, maybe i_size */
494 unsigned short i_bytes
;
495 unsigned int i_blkbits
;
498 #ifdef __NEED_I_SIZE_ORDERED
499 seqcount_t i_size_seqcount
;
503 unsigned long i_state
;
504 struct mutex i_mutex
;
506 unsigned long dirtied_when
; /* jiffies of first dirtying */
508 struct hlist_node i_hash
;
509 struct list_head i_wb_list
; /* backing dev IO list */
510 struct list_head i_lru
; /* inode LRU list */
511 struct list_head i_sb_list
;
513 struct hlist_head i_dentry
;
514 struct rcu_head i_rcu
;
518 atomic_t i_dio_count
;
519 atomic_t i_writecount
;
520 const struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
521 struct file_lock
*i_flock
;
522 struct address_space i_data
;
524 struct dquot
*i_dquot
[MAXQUOTAS
];
526 struct list_head i_devices
;
528 struct pipe_inode_info
*i_pipe
;
529 struct block_device
*i_bdev
;
535 #ifdef CONFIG_FSNOTIFY
536 __u32 i_fsnotify_mask
; /* all events this inode cares about */
537 struct hlist_head i_fsnotify_marks
;
541 atomic_t i_readcount
; /* struct files open RO */
543 void *i_private
; /* fs or device private pointer */
546 static inline int inode_unhashed(struct inode
*inode
)
548 return hlist_unhashed(&inode
->i_hash
);
552 * inode->i_mutex nesting subclasses for the lock validator:
554 * 0: the object of the current VFS operation
559 * The locking order between these classes is
560 * parent -> child -> normal -> xattr -> quota
562 enum inode_i_mutex_lock_class
572 * NOTE: in a 32bit arch with a preemptable kernel and
573 * an UP compile the i_size_read/write must be atomic
574 * with respect to the local cpu (unlike with preempt disabled),
575 * but they don't need to be atomic with respect to other cpus like in
576 * true SMP (so they need either to either locally disable irq around
577 * the read or for example on x86 they can be still implemented as a
578 * cmpxchg8b without the need of the lock prefix). For SMP compiles
579 * and 64bit archs it makes no difference if preempt is enabled or not.
581 static inline loff_t
i_size_read(const struct inode
*inode
)
583 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
588 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
589 i_size
= inode
->i_size
;
590 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
592 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
596 i_size
= inode
->i_size
;
600 return inode
->i_size
;
605 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
606 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
607 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
609 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
611 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
612 write_seqcount_begin(&inode
->i_size_seqcount
);
613 inode
->i_size
= i_size
;
614 write_seqcount_end(&inode
->i_size_seqcount
);
615 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
617 inode
->i_size
= i_size
;
620 inode
->i_size
= i_size
;
624 /* Helper functions so that in most cases filesystems will
625 * not need to deal directly with kuid_t and kgid_t and can
626 * instead deal with the raw numeric values that are stored
629 static inline uid_t
i_uid_read(const struct inode
*inode
)
631 return from_kuid(&init_user_ns
, inode
->i_uid
);
634 static inline gid_t
i_gid_read(const struct inode
*inode
)
636 return from_kgid(&init_user_ns
, inode
->i_gid
);
639 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
641 inode
->i_uid
= make_kuid(&init_user_ns
, uid
);
644 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
646 inode
->i_gid
= make_kgid(&init_user_ns
, gid
);
649 static inline unsigned iminor(const struct inode
*inode
)
651 return MINOR(inode
->i_rdev
);
654 static inline unsigned imajor(const struct inode
*inode
)
656 return MAJOR(inode
->i_rdev
);
659 extern struct block_device
*I_BDEV(struct inode
*inode
);
662 rwlock_t lock
; /* protects pid, uid, euid fields */
663 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
664 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
665 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
666 int signum
; /* posix.1b rt signal to be delivered on IO */
670 * Track a single file's readahead state
672 struct file_ra_state
{
673 pgoff_t start
; /* where readahead started */
674 unsigned int size
; /* # of readahead pages */
675 unsigned int async_size
; /* do asynchronous readahead when
676 there are only # of pages ahead */
678 unsigned int ra_pages
; /* Maximum readahead window */
679 unsigned int mmap_miss
; /* Cache miss stat for mmap accesses */
680 loff_t prev_pos
; /* Cache last read() position */
684 * Check if @index falls in the readahead windows.
686 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
688 return (index
>= ra
->start
&&
689 index
< ra
->start
+ ra
->size
);
692 #define FILE_MNT_WRITE_TAKEN 1
693 #define FILE_MNT_WRITE_RELEASED 2
697 * fu_list becomes invalid after file_free is called and queued via
698 * fu_rcuhead for RCU freeing
701 struct list_head fu_list
;
702 struct rcu_head fu_rcuhead
;
705 #define f_dentry f_path.dentry
706 #define f_vfsmnt f_path.mnt
707 const struct file_operations
*f_op
;
710 * Protects f_ep_links, f_flags, f_pos vs i_size in lseek SEEK_CUR.
711 * Must not be taken from IRQ context.
717 atomic_long_t f_count
;
718 unsigned int f_flags
;
721 struct fown_struct f_owner
;
722 const struct cred
*f_cred
;
723 struct file_ra_state f_ra
;
726 #ifdef CONFIG_SECURITY
729 /* needed for tty driver, and maybe others */
733 /* Used by fs/eventpoll.c to link all the hooks to this file */
734 struct list_head f_ep_links
;
735 struct list_head f_tfile_llink
;
736 #endif /* #ifdef CONFIG_EPOLL */
737 struct address_space
*f_mapping
;
738 #ifdef CONFIG_DEBUG_WRITECOUNT
739 unsigned long f_mnt_write_state
;
746 /* file identifier */
747 unsigned char f_handle
[0];
750 static inline struct file
*get_file(struct file
*f
)
752 atomic_long_inc(&f
->f_count
);
755 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
756 #define file_count(x) atomic_long_read(&(x)->f_count)
758 #ifdef CONFIG_DEBUG_WRITECOUNT
759 static inline void file_take_write(struct file
*f
)
761 WARN_ON(f
->f_mnt_write_state
!= 0);
762 f
->f_mnt_write_state
= FILE_MNT_WRITE_TAKEN
;
764 static inline void file_release_write(struct file
*f
)
766 f
->f_mnt_write_state
|= FILE_MNT_WRITE_RELEASED
;
768 static inline void file_reset_write(struct file
*f
)
770 f
->f_mnt_write_state
= 0;
772 static inline void file_check_state(struct file
*f
)
775 * At this point, either both or neither of these bits
778 WARN_ON(f
->f_mnt_write_state
== FILE_MNT_WRITE_TAKEN
);
779 WARN_ON(f
->f_mnt_write_state
== FILE_MNT_WRITE_RELEASED
);
781 static inline int file_check_writeable(struct file
*f
)
783 if (f
->f_mnt_write_state
== FILE_MNT_WRITE_TAKEN
)
785 printk(KERN_WARNING
"writeable file with no "
786 "mnt_want_write()\n");
790 #else /* !CONFIG_DEBUG_WRITECOUNT */
791 static inline void file_take_write(struct file
*filp
) {}
792 static inline void file_release_write(struct file
*filp
) {}
793 static inline void file_reset_write(struct file
*filp
) {}
794 static inline void file_check_state(struct file
*filp
) {}
795 static inline int file_check_writeable(struct file
*filp
)
799 #endif /* CONFIG_DEBUG_WRITECOUNT */
801 #define MAX_NON_LFS ((1UL<<31) - 1)
803 /* Page cache limit. The filesystems should put that into their s_maxbytes
804 limits, otherwise bad things can happen in VM. */
805 #if BITS_PER_LONG==32
806 #define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
807 #elif BITS_PER_LONG==64
808 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
813 #define FL_ACCESS 8 /* not trying to lock, just looking */
814 #define FL_EXISTS 16 /* when unlocking, test for existence */
815 #define FL_LEASE 32 /* lease held on this file */
816 #define FL_CLOSE 64 /* unlock on close */
817 #define FL_SLEEP 128 /* A blocking lock */
818 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
819 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
822 * Special return value from posix_lock_file() and vfs_lock_file() for
823 * asynchronous locking.
825 #define FILE_LOCK_DEFERRED 1
828 * The POSIX file lock owner is determined by
829 * the "struct files_struct" in the thread group
830 * (or NULL for no owner - BSD locks).
832 * Lockd stuffs a "host" pointer into this.
834 typedef struct files_struct
*fl_owner_t
;
836 struct file_lock_operations
{
837 void (*fl_copy_lock
)(struct file_lock
*, struct file_lock
*);
838 void (*fl_release_private
)(struct file_lock
*);
841 struct lock_manager_operations
{
842 int (*lm_compare_owner
)(struct file_lock
*, struct file_lock
*);
843 void (*lm_notify
)(struct file_lock
*); /* unblock callback */
844 int (*lm_grant
)(struct file_lock
*, struct file_lock
*, int);
845 void (*lm_break
)(struct file_lock
*);
846 int (*lm_change
)(struct file_lock
**, int);
849 struct lock_manager
{
850 struct list_head list
;
854 void locks_start_grace(struct net
*, struct lock_manager
*);
855 void locks_end_grace(struct lock_manager
*);
856 int locks_in_grace(struct net
*);
858 /* that will die - we need it for nfs_lock_info */
859 #include <linux/nfs_fs_i.h>
862 struct file_lock
*fl_next
; /* singly linked list for this inode */
863 struct list_head fl_link
; /* doubly linked list of all locks */
864 struct list_head fl_block
; /* circular list of blocked processes */
866 unsigned int fl_flags
;
867 unsigned char fl_type
;
869 struct pid
*fl_nspid
;
870 wait_queue_head_t fl_wait
;
871 struct file
*fl_file
;
875 struct fasync_struct
* fl_fasync
; /* for lease break notifications */
876 /* for lease breaks: */
877 unsigned long fl_break_time
;
878 unsigned long fl_downgrade_time
;
880 const struct file_lock_operations
*fl_ops
; /* Callbacks for filesystems */
881 const struct lock_manager_operations
*fl_lmops
; /* Callbacks for lockmanagers */
883 struct nfs_lock_info nfs_fl
;
884 struct nfs4_lock_info nfs4_fl
;
886 struct list_head link
; /* link in AFS vnode's pending_locks list */
887 int state
; /* state of grant or error if -ve */
892 /* The following constant reflects the upper bound of the file/locking space */
894 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
895 #define OFFSET_MAX INT_LIMIT(loff_t)
896 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
899 #include <linux/fcntl.h>
901 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
903 #ifdef CONFIG_FILE_LOCKING
904 extern int fcntl_getlk(struct file
*, struct flock __user
*);
905 extern int fcntl_setlk(unsigned int, struct file
*, unsigned int,
906 struct flock __user
*);
908 #if BITS_PER_LONG == 32
909 extern int fcntl_getlk64(struct file
*, struct flock64 __user
*);
910 extern int fcntl_setlk64(unsigned int, struct file
*, unsigned int,
911 struct flock64 __user
*);
914 extern int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
);
915 extern int fcntl_getlease(struct file
*filp
);
918 void locks_free_lock(struct file_lock
*fl
);
919 extern void locks_init_lock(struct file_lock
*);
920 extern struct file_lock
* locks_alloc_lock(void);
921 extern void locks_copy_lock(struct file_lock
*, struct file_lock
*);
922 extern void __locks_copy_lock(struct file_lock
*, const struct file_lock
*);
923 extern void locks_remove_posix(struct file
*, fl_owner_t
);
924 extern void locks_remove_flock(struct file
*);
925 extern void locks_release_private(struct file_lock
*);
926 extern void posix_test_lock(struct file
*, struct file_lock
*);
927 extern int posix_lock_file(struct file
*, struct file_lock
*, struct file_lock
*);
928 extern int posix_lock_file_wait(struct file
*, struct file_lock
*);
929 extern int posix_unblock_lock(struct file
*, struct file_lock
*);
930 extern int vfs_test_lock(struct file
*, struct file_lock
*);
931 extern int vfs_lock_file(struct file
*, unsigned int, struct file_lock
*, struct file_lock
*);
932 extern int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
);
933 extern int flock_lock_file_wait(struct file
*filp
, struct file_lock
*fl
);
934 extern int __break_lease(struct inode
*inode
, unsigned int flags
);
935 extern void lease_get_mtime(struct inode
*, struct timespec
*time
);
936 extern int generic_setlease(struct file
*, long, struct file_lock
**);
937 extern int vfs_setlease(struct file
*, long, struct file_lock
**);
938 extern int lease_modify(struct file_lock
**, int);
939 extern int lock_may_read(struct inode
*, loff_t start
, unsigned long count
);
940 extern int lock_may_write(struct inode
*, loff_t start
, unsigned long count
);
941 extern void locks_delete_block(struct file_lock
*waiter
);
942 extern void lock_flocks(void);
943 extern void unlock_flocks(void);
944 #else /* !CONFIG_FILE_LOCKING */
945 static inline int fcntl_getlk(struct file
*file
, struct flock __user
*user
)
950 static inline int fcntl_setlk(unsigned int fd
, struct file
*file
,
951 unsigned int cmd
, struct flock __user
*user
)
956 #if BITS_PER_LONG == 32
957 static inline int fcntl_getlk64(struct file
*file
, struct flock64 __user
*user
)
962 static inline int fcntl_setlk64(unsigned int fd
, struct file
*file
,
963 unsigned int cmd
, struct flock64 __user
*user
)
968 static inline int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
)
973 static inline int fcntl_getlease(struct file
*filp
)
978 static inline void locks_init_lock(struct file_lock
*fl
)
983 static inline void __locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
988 static inline void locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
993 static inline void locks_remove_posix(struct file
*filp
, fl_owner_t owner
)
998 static inline void locks_remove_flock(struct file
*filp
)
1003 static inline void posix_test_lock(struct file
*filp
, struct file_lock
*fl
)
1008 static inline int posix_lock_file(struct file
*filp
, struct file_lock
*fl
,
1009 struct file_lock
*conflock
)
1014 static inline int posix_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1019 static inline int posix_unblock_lock(struct file
*filp
,
1020 struct file_lock
*waiter
)
1025 static inline int vfs_test_lock(struct file
*filp
, struct file_lock
*fl
)
1030 static inline int vfs_lock_file(struct file
*filp
, unsigned int cmd
,
1031 struct file_lock
*fl
, struct file_lock
*conf
)
1036 static inline int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
)
1041 static inline int flock_lock_file_wait(struct file
*filp
,
1042 struct file_lock
*request
)
1047 static inline int __break_lease(struct inode
*inode
, unsigned int mode
)
1052 static inline void lease_get_mtime(struct inode
*inode
, struct timespec
*time
)
1057 static inline int generic_setlease(struct file
*filp
, long arg
,
1058 struct file_lock
**flp
)
1063 static inline int vfs_setlease(struct file
*filp
, long arg
,
1064 struct file_lock
**lease
)
1069 static inline int lease_modify(struct file_lock
**before
, int arg
)
1074 static inline int lock_may_read(struct inode
*inode
, loff_t start
,
1080 static inline int lock_may_write(struct inode
*inode
, loff_t start
,
1086 static inline void locks_delete_block(struct file_lock
*waiter
)
1090 static inline void lock_flocks(void)
1094 static inline void unlock_flocks(void)
1098 #endif /* !CONFIG_FILE_LOCKING */
1101 struct fasync_struct
{
1105 struct fasync_struct
*fa_next
; /* singly linked list */
1106 struct file
*fa_file
;
1107 struct rcu_head fa_rcu
;
1110 #define FASYNC_MAGIC 0x4601
1112 /* SMP safe fasync helpers: */
1113 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1114 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1115 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1116 extern struct fasync_struct
*fasync_alloc(void);
1117 extern void fasync_free(struct fasync_struct
*);
1119 /* can be called from interrupts */
1120 extern void kill_fasync(struct fasync_struct
**, int, int);
1122 extern int __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1123 extern int f_setown(struct file
*filp
, unsigned long arg
, int force
);
1124 extern void f_delown(struct file
*filp
);
1125 extern pid_t
f_getown(struct file
*filp
);
1126 extern int send_sigurg(struct fown_struct
*fown
);
1134 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1135 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1136 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1137 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1138 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1140 extern struct list_head super_blocks
;
1141 extern spinlock_t sb_lock
;
1143 /* Possible states of 'frozen' field */
1145 SB_UNFROZEN
= 0, /* FS is unfrozen */
1146 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1147 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1148 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1149 * internal threads if needed) */
1150 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1153 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1156 /* Counters for counting writers at each level */
1157 struct percpu_counter counter
[SB_FREEZE_LEVELS
];
1158 wait_queue_head_t wait
; /* queue for waiting for
1159 writers / faults to finish */
1160 int frozen
; /* Is sb frozen? */
1161 wait_queue_head_t wait_unfrozen
; /* queue for waiting for
1163 #ifdef CONFIG_DEBUG_LOCK_ALLOC
1164 struct lockdep_map lock_map
[SB_FREEZE_LEVELS
];
1168 struct super_block
{
1169 struct list_head s_list
; /* Keep this first */
1170 dev_t s_dev
; /* search index; _not_ kdev_t */
1171 unsigned char s_blocksize_bits
;
1172 unsigned long s_blocksize
;
1173 loff_t s_maxbytes
; /* Max file size */
1174 struct file_system_type
*s_type
;
1175 const struct super_operations
*s_op
;
1176 const struct dquot_operations
*dq_op
;
1177 const struct quotactl_ops
*s_qcop
;
1178 const struct export_operations
*s_export_op
;
1179 unsigned long s_flags
;
1180 unsigned long s_magic
;
1181 struct dentry
*s_root
;
1182 struct rw_semaphore s_umount
;
1185 #ifdef CONFIG_SECURITY
1188 const struct xattr_handler
**s_xattr
;
1190 struct list_head s_inodes
; /* all inodes */
1191 struct hlist_bl_head s_anon
; /* anonymous dentries for (nfs) exporting */
1193 struct list_head __percpu
*s_files
;
1195 struct list_head s_files
;
1197 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1198 /* s_dentry_lru, s_nr_dentry_unused protected by dcache.c lru locks */
1199 struct list_head s_dentry_lru
; /* unused dentry lru */
1200 int s_nr_dentry_unused
; /* # of dentry on lru */
1202 /* s_inode_lru_lock protects s_inode_lru and s_nr_inodes_unused */
1203 spinlock_t s_inode_lru_lock ____cacheline_aligned_in_smp
;
1204 struct list_head s_inode_lru
; /* unused inode lru */
1205 int s_nr_inodes_unused
; /* # of inodes on lru */
1207 struct block_device
*s_bdev
;
1208 struct backing_dev_info
*s_bdi
;
1209 struct mtd_info
*s_mtd
;
1210 struct hlist_node s_instances
;
1211 struct quota_info s_dquot
; /* Diskquota specific options */
1213 struct sb_writers s_writers
;
1215 char s_id
[32]; /* Informational name */
1216 u8 s_uuid
[16]; /* UUID */
1218 void *s_fs_info
; /* Filesystem private info */
1219 unsigned int s_max_links
;
1222 /* Granularity of c/m/atime in ns.
1223 Cannot be worse than a second */
1227 * The next field is for VFS *only*. No filesystems have any business
1228 * even looking at it. You had been warned.
1230 struct mutex s_vfs_rename_mutex
; /* Kludge */
1233 * Filesystem subtype. If non-empty the filesystem type field
1234 * in /proc/mounts will be "type.subtype"
1239 * Saved mount options for lazy filesystems using
1240 * generic_show_options()
1242 char __rcu
*s_options
;
1243 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1246 * Saved pool identifier for cleancache (-1 means none)
1248 int cleancache_poolid
;
1250 struct shrinker s_shrink
; /* per-sb shrinker handle */
1252 /* Number of inodes with nlink == 0 but still referenced */
1253 atomic_long_t s_remove_count
;
1255 /* Being remounted read-only */
1256 int s_readonly_remount
;
1259 /* superblock cache pruning functions */
1260 extern void prune_icache_sb(struct super_block
*sb
, int nr_to_scan
);
1261 extern void prune_dcache_sb(struct super_block
*sb
, int nr_to_scan
);
1263 extern struct timespec
current_fs_time(struct super_block
*sb
);
1266 * Snapshotting support.
1269 void __sb_end_write(struct super_block
*sb
, int level
);
1270 int __sb_start_write(struct super_block
*sb
, int level
, bool wait
);
1273 * sb_end_write - drop write access to a superblock
1274 * @sb: the super we wrote to
1276 * Decrement number of writers to the filesystem. Wake up possible waiters
1277 * wanting to freeze the filesystem.
1279 static inline void sb_end_write(struct super_block
*sb
)
1281 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1285 * sb_end_pagefault - drop write access to a superblock from a page fault
1286 * @sb: the super we wrote to
1288 * Decrement number of processes handling write page fault to the filesystem.
1289 * Wake up possible waiters wanting to freeze the filesystem.
1291 static inline void sb_end_pagefault(struct super_block
*sb
)
1293 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1297 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1298 * @sb: the super we wrote to
1300 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1301 * waiters wanting to freeze the filesystem.
1303 static inline void sb_end_intwrite(struct super_block
*sb
)
1305 __sb_end_write(sb
, SB_FREEZE_FS
);
1309 * sb_start_write - get write access to a superblock
1310 * @sb: the super we write to
1312 * When a process wants to write data or metadata to a file system (i.e. dirty
1313 * a page or an inode), it should embed the operation in a sb_start_write() -
1314 * sb_end_write() pair to get exclusion against file system freezing. This
1315 * function increments number of writers preventing freezing. If the file
1316 * system is already frozen, the function waits until the file system is
1319 * Since freeze protection behaves as a lock, users have to preserve
1320 * ordering of freeze protection and other filesystem locks. Generally,
1321 * freeze protection should be the outermost lock. In particular, we have:
1324 * -> i_mutex (write path, truncate, directory ops, ...)
1325 * -> s_umount (freeze_super, thaw_super)
1327 static inline void sb_start_write(struct super_block
*sb
)
1329 __sb_start_write(sb
, SB_FREEZE_WRITE
, true);
1332 static inline int sb_start_write_trylock(struct super_block
*sb
)
1334 return __sb_start_write(sb
, SB_FREEZE_WRITE
, false);
1338 * sb_start_pagefault - get write access to a superblock from a page fault
1339 * @sb: the super we write to
1341 * When a process starts handling write page fault, it should embed the
1342 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1343 * exclusion against file system freezing. This is needed since the page fault
1344 * is going to dirty a page. This function increments number of running page
1345 * faults preventing freezing. If the file system is already frozen, the
1346 * function waits until the file system is thawed.
1348 * Since page fault freeze protection behaves as a lock, users have to preserve
1349 * ordering of freeze protection and other filesystem locks. It is advised to
1350 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1351 * handling code implies lock dependency:
1354 * -> sb_start_pagefault
1356 static inline void sb_start_pagefault(struct super_block
*sb
)
1358 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
, true);
1362 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1363 * @sb: the super we write to
1365 * This is the third level of protection against filesystem freezing. It is
1366 * free for use by a filesystem. The only requirement is that it must rank
1367 * below sb_start_pagefault.
1369 * For example filesystem can call sb_start_intwrite() when starting a
1370 * transaction which somewhat eases handling of freezing for internal sources
1371 * of filesystem changes (internal fs threads, discarding preallocation on file
1374 static inline void sb_start_intwrite(struct super_block
*sb
)
1376 __sb_start_write(sb
, SB_FREEZE_FS
, true);
1380 extern bool inode_owner_or_capable(const struct inode
*inode
);
1382 /* not quite ready to be deprecated, but... */
1383 extern void lock_super(struct super_block
*);
1384 extern void unlock_super(struct super_block
*);
1387 * VFS helper functions..
1389 extern int vfs_create(struct inode
*, struct dentry
*, umode_t
, bool);
1390 extern int vfs_mkdir(struct inode
*, struct dentry
*, umode_t
);
1391 extern int vfs_mknod(struct inode
*, struct dentry
*, umode_t
, dev_t
);
1392 extern int vfs_symlink(struct inode
*, struct dentry
*, const char *);
1393 extern int vfs_link(struct dentry
*, struct inode
*, struct dentry
*);
1394 extern int vfs_rmdir(struct inode
*, struct dentry
*);
1395 extern int vfs_unlink(struct inode
*, struct dentry
*);
1396 extern int vfs_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*);
1399 * VFS dentry helper functions.
1401 extern void dentry_unhash(struct dentry
*dentry
);
1404 * VFS file helper functions.
1406 extern void inode_init_owner(struct inode
*inode
, const struct inode
*dir
,
1409 * VFS FS_IOC_FIEMAP helper definitions.
1411 struct fiemap_extent_info
{
1412 unsigned int fi_flags
; /* Flags as passed from user */
1413 unsigned int fi_extents_mapped
; /* Number of mapped extents */
1414 unsigned int fi_extents_max
; /* Size of fiemap_extent array */
1415 struct fiemap_extent __user
*fi_extents_start
; /* Start of
1416 fiemap_extent array */
1418 int fiemap_fill_next_extent(struct fiemap_extent_info
*info
, u64 logical
,
1419 u64 phys
, u64 len
, u32 flags
);
1420 int fiemap_check_flags(struct fiemap_extent_info
*fieinfo
, u32 fs_flags
);
1425 * NOTE! These match bits 12..15 of stat.st_mode
1426 * (ie "(i_mode >> 12) & 15").
1428 #define DT_UNKNOWN 0
1439 * This is the "filldir" function type, used by readdir() to let
1440 * the kernel specify what kind of dirent layout it wants to have.
1441 * This allows the kernel to read directories into kernel space or
1442 * to have different dirent layouts depending on the binary type.
1444 typedef int (*filldir_t
)(void *, const char *, int, loff_t
, u64
, unsigned);
1445 struct block_device_operations
;
1447 /* These macros are for out of kernel modules to test that
1448 * the kernel supports the unlocked_ioctl and compat_ioctl
1449 * fields in struct file_operations. */
1450 #define HAVE_COMPAT_IOCTL 1
1451 #define HAVE_UNLOCKED_IOCTL 1
1453 struct file_operations
{
1454 struct module
*owner
;
1455 loff_t (*llseek
) (struct file
*, loff_t
, int);
1456 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
1457 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
1458 ssize_t (*aio_read
) (struct kiocb
*, const struct iovec
*, unsigned long, loff_t
);
1459 ssize_t (*aio_write
) (struct kiocb
*, const struct iovec
*, unsigned long, loff_t
);
1460 int (*readdir
) (struct file
*, void *, filldir_t
);
1461 unsigned int (*poll
) (struct file
*, struct poll_table_struct
*);
1462 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
1463 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
1464 int (*mmap
) (struct file
*, struct vm_area_struct
*);
1465 int (*open
) (struct inode
*, struct file
*);
1466 int (*flush
) (struct file
*, fl_owner_t id
);
1467 int (*release
) (struct inode
*, struct file
*);
1468 int (*fsync
) (struct file
*, loff_t
, loff_t
, int datasync
);
1469 int (*aio_fsync
) (struct kiocb
*, int datasync
);
1470 int (*fasync
) (int, struct file
*, int);
1471 int (*lock
) (struct file
*, int, struct file_lock
*);
1472 ssize_t (*sendpage
) (struct file
*, struct page
*, int, size_t, loff_t
*, int);
1473 unsigned long (*get_unmapped_area
)(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1474 int (*check_flags
)(int);
1475 int (*flock
) (struct file
*, int, struct file_lock
*);
1476 ssize_t (*splice_write
)(struct pipe_inode_info
*, struct file
*, loff_t
*, size_t, unsigned int);
1477 ssize_t (*splice_read
)(struct file
*, loff_t
*, struct pipe_inode_info
*, size_t, unsigned int);
1478 int (*setlease
)(struct file
*, long, struct file_lock
**);
1479 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
1483 struct inode_operations
{
1484 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*, unsigned int);
1485 void * (*follow_link
) (struct dentry
*, struct nameidata
*);
1486 int (*permission
) (struct inode
*, int);
1487 struct posix_acl
* (*get_acl
)(struct inode
*, int);
1489 int (*readlink
) (struct dentry
*, char __user
*,int);
1490 void (*put_link
) (struct dentry
*, struct nameidata
*, void *);
1492 int (*create
) (struct inode
*,struct dentry
*, umode_t
, bool);
1493 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
1494 int (*unlink
) (struct inode
*,struct dentry
*);
1495 int (*symlink
) (struct inode
*,struct dentry
*,const char *);
1496 int (*mkdir
) (struct inode
*,struct dentry
*,umode_t
);
1497 int (*rmdir
) (struct inode
*,struct dentry
*);
1498 int (*mknod
) (struct inode
*,struct dentry
*,umode_t
,dev_t
);
1499 int (*rename
) (struct inode
*, struct dentry
*,
1500 struct inode
*, struct dentry
*);
1501 void (*truncate
) (struct inode
*);
1502 int (*setattr
) (struct dentry
*, struct iattr
*);
1503 int (*getattr
) (struct vfsmount
*mnt
, struct dentry
*, struct kstat
*);
1504 int (*setxattr
) (struct dentry
*, const char *,const void *,size_t,int);
1505 ssize_t (*getxattr
) (struct dentry
*, const char *, void *, size_t);
1506 ssize_t (*listxattr
) (struct dentry
*, char *, size_t);
1507 int (*removexattr
) (struct dentry
*, const char *);
1508 int (*fiemap
)(struct inode
*, struct fiemap_extent_info
*, u64 start
,
1510 int (*update_time
)(struct inode
*, struct timespec
*, int);
1511 int (*atomic_open
)(struct inode
*, struct dentry
*,
1512 struct file
*, unsigned open_flag
,
1513 umode_t create_mode
, int *opened
);
1514 } ____cacheline_aligned
;
1518 ssize_t
rw_copy_check_uvector(int type
, const struct iovec __user
* uvector
,
1519 unsigned long nr_segs
, unsigned long fast_segs
,
1520 struct iovec
*fast_pointer
,
1521 struct iovec
**ret_pointer
);
1523 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1524 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1525 extern ssize_t
vfs_readv(struct file
*, const struct iovec __user
*,
1526 unsigned long, loff_t
*);
1527 extern ssize_t
vfs_writev(struct file
*, const struct iovec __user
*,
1528 unsigned long, loff_t
*);
1530 struct super_operations
{
1531 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
1532 void (*destroy_inode
)(struct inode
*);
1534 void (*dirty_inode
) (struct inode
*, int flags
);
1535 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
1536 int (*drop_inode
) (struct inode
*);
1537 void (*evict_inode
) (struct inode
*);
1538 void (*put_super
) (struct super_block
*);
1539 int (*sync_fs
)(struct super_block
*sb
, int wait
);
1540 int (*freeze_fs
) (struct super_block
*);
1541 int (*unfreeze_fs
) (struct super_block
*);
1542 int (*statfs
) (struct dentry
*, struct kstatfs
*);
1543 int (*remount_fs
) (struct super_block
*, int *, char *);
1544 void (*umount_begin
) (struct super_block
*);
1546 int (*show_options
)(struct seq_file
*, struct dentry
*);
1547 int (*show_devname
)(struct seq_file
*, struct dentry
*);
1548 int (*show_path
)(struct seq_file
*, struct dentry
*);
1549 int (*show_stats
)(struct seq_file
*, struct dentry
*);
1551 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
1552 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
1554 int (*bdev_try_to_free_page
)(struct super_block
*, struct page
*, gfp_t
);
1555 int (*nr_cached_objects
)(struct super_block
*);
1556 void (*free_cached_objects
)(struct super_block
*, int);
1560 * Inode state bits. Protected by inode->i_lock
1562 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1563 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1565 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1566 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1567 * various stages of removing an inode.
1569 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1571 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1572 * fdatasync(). i_atime is the usual cause.
1573 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1574 * these changes separately from I_DIRTY_SYNC so that we
1575 * don't have to write inode on fdatasync() when only
1576 * mtime has changed in it.
1577 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1578 * I_NEW Serves as both a mutex and completion notification.
1579 * New inodes set I_NEW. If two processes both create
1580 * the same inode, one of them will release its inode and
1581 * wait for I_NEW to be released before returning.
1582 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1583 * also cause waiting on I_NEW, without I_NEW actually
1584 * being set. find_inode() uses this to prevent returning
1585 * nearly-dead inodes.
1586 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1587 * is zero. I_FREEING must be set when I_WILL_FREE is
1589 * I_FREEING Set when inode is about to be freed but still has dirty
1590 * pages or buffers attached or the inode itself is still
1592 * I_CLEAR Added by clear_inode(). In this state the inode is
1593 * clean and can be destroyed. Inode keeps I_FREEING.
1595 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1596 * prohibited for many purposes. iget() must wait for
1597 * the inode to be completely released, then create it
1598 * anew. Other functions will just ignore such inodes,
1599 * if appropriate. I_NEW is used for waiting.
1601 * I_SYNC Writeback of inode is running. The bit is set during
1602 * data writeback, and cleared with a wakeup on the bit
1603 * address once it is done. The bit is also used to pin
1604 * the inode in memory for flusher thread.
1606 * I_REFERENCED Marks the inode as recently references on the LRU list.
1608 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1610 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1612 #define I_DIRTY_SYNC (1 << 0)
1613 #define I_DIRTY_DATASYNC (1 << 1)
1614 #define I_DIRTY_PAGES (1 << 2)
1616 #define I_NEW (1 << __I_NEW)
1617 #define I_WILL_FREE (1 << 4)
1618 #define I_FREEING (1 << 5)
1619 #define I_CLEAR (1 << 6)
1621 #define I_SYNC (1 << __I_SYNC)
1622 #define I_REFERENCED (1 << 8)
1623 #define __I_DIO_WAKEUP 9
1624 #define I_DIO_WAKEUP (1 << I_DIO_WAKEUP)
1626 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1628 extern void __mark_inode_dirty(struct inode
*, int);
1629 static inline void mark_inode_dirty(struct inode
*inode
)
1631 __mark_inode_dirty(inode
, I_DIRTY
);
1634 static inline void mark_inode_dirty_sync(struct inode
*inode
)
1636 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
1639 extern void inc_nlink(struct inode
*inode
);
1640 extern void drop_nlink(struct inode
*inode
);
1641 extern void clear_nlink(struct inode
*inode
);
1642 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
1644 static inline void inode_inc_link_count(struct inode
*inode
)
1647 mark_inode_dirty(inode
);
1650 static inline void inode_dec_link_count(struct inode
*inode
)
1653 mark_inode_dirty(inode
);
1657 * inode_inc_iversion - increments i_version
1658 * @inode: inode that need to be updated
1660 * Every time the inode is modified, the i_version field will be incremented.
1661 * The filesystem has to be mounted with i_version flag
1664 static inline void inode_inc_iversion(struct inode
*inode
)
1666 spin_lock(&inode
->i_lock
);
1668 spin_unlock(&inode
->i_lock
);
1671 enum file_time_flags
{
1678 extern void touch_atime(struct path
*);
1679 static inline void file_accessed(struct file
*file
)
1681 if (!(file
->f_flags
& O_NOATIME
))
1682 touch_atime(&file
->f_path
);
1685 int sync_inode(struct inode
*inode
, struct writeback_control
*wbc
);
1686 int sync_inode_metadata(struct inode
*inode
, int wait
);
1688 struct file_system_type
{
1691 struct dentry
*(*mount
) (struct file_system_type
*, int,
1692 const char *, void *);
1693 void (*kill_sb
) (struct super_block
*);
1694 struct module
*owner
;
1695 struct file_system_type
* next
;
1696 struct hlist_head fs_supers
;
1698 struct lock_class_key s_lock_key
;
1699 struct lock_class_key s_umount_key
;
1700 struct lock_class_key s_vfs_rename_key
;
1701 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
1703 struct lock_class_key i_lock_key
;
1704 struct lock_class_key i_mutex_key
;
1705 struct lock_class_key i_mutex_dir_key
;
1708 extern struct dentry
*mount_ns(struct file_system_type
*fs_type
, int flags
,
1709 void *data
, int (*fill_super
)(struct super_block
*, void *, int));
1710 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
1711 int flags
, const char *dev_name
, void *data
,
1712 int (*fill_super
)(struct super_block
*, void *, int));
1713 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
1714 int flags
, void *data
,
1715 int (*fill_super
)(struct super_block
*, void *, int));
1716 extern struct dentry
*mount_nodev(struct file_system_type
*fs_type
,
1717 int flags
, void *data
,
1718 int (*fill_super
)(struct super_block
*, void *, int));
1719 extern struct dentry
*mount_subtree(struct vfsmount
*mnt
, const char *path
);
1720 void generic_shutdown_super(struct super_block
*sb
);
1721 void kill_block_super(struct super_block
*sb
);
1722 void kill_anon_super(struct super_block
*sb
);
1723 void kill_litter_super(struct super_block
*sb
);
1724 void deactivate_super(struct super_block
*sb
);
1725 void deactivate_locked_super(struct super_block
*sb
);
1726 int set_anon_super(struct super_block
*s
, void *data
);
1727 int get_anon_bdev(dev_t
*);
1728 void free_anon_bdev(dev_t
);
1729 struct super_block
*sget(struct file_system_type
*type
,
1730 int (*test
)(struct super_block
*,void *),
1731 int (*set
)(struct super_block
*,void *),
1732 int flags
, void *data
);
1733 extern struct dentry
*mount_pseudo(struct file_system_type
*, char *,
1734 const struct super_operations
*ops
,
1735 const struct dentry_operations
*dops
,
1738 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1739 #define fops_get(fops) \
1740 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1741 #define fops_put(fops) \
1742 do { if (fops) module_put((fops)->owner); } while(0)
1744 extern int register_filesystem(struct file_system_type
*);
1745 extern int unregister_filesystem(struct file_system_type
*);
1746 extern struct vfsmount
*kern_mount_data(struct file_system_type
*, void *data
);
1747 #define kern_mount(type) kern_mount_data(type, NULL)
1748 extern void kern_unmount(struct vfsmount
*mnt
);
1749 extern int may_umount_tree(struct vfsmount
*);
1750 extern int may_umount(struct vfsmount
*);
1751 extern long do_mount(const char *, const char *, const char *, unsigned long, void *);
1752 extern struct vfsmount
*collect_mounts(struct path
*);
1753 extern void drop_collected_mounts(struct vfsmount
*);
1754 extern int iterate_mounts(int (*)(struct vfsmount
*, void *), void *,
1756 extern int vfs_statfs(struct path
*, struct kstatfs
*);
1757 extern int user_statfs(const char __user
*, struct kstatfs
*);
1758 extern int fd_statfs(int, struct kstatfs
*);
1759 extern int vfs_ustat(dev_t
, struct kstatfs
*);
1760 extern int freeze_super(struct super_block
*super
);
1761 extern int thaw_super(struct super_block
*super
);
1762 extern bool our_mnt(struct vfsmount
*mnt
);
1764 extern int current_umask(void);
1767 extern struct kobject
*fs_kobj
;
1769 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1770 extern int rw_verify_area(int, struct file
*, loff_t
*, size_t);
1772 #define FLOCK_VERIFY_READ 1
1773 #define FLOCK_VERIFY_WRITE 2
1775 #ifdef CONFIG_FILE_LOCKING
1776 extern int locks_mandatory_locked(struct inode
*);
1777 extern int locks_mandatory_area(int, struct inode
*, struct file
*, loff_t
, size_t);
1780 * Candidates for mandatory locking have the setgid bit set
1781 * but no group execute bit - an otherwise meaningless combination.
1784 static inline int __mandatory_lock(struct inode
*ino
)
1786 return (ino
->i_mode
& (S_ISGID
| S_IXGRP
)) == S_ISGID
;
1790 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1791 * otherwise these will be advisory locks
1794 static inline int mandatory_lock(struct inode
*ino
)
1796 return IS_MANDLOCK(ino
) && __mandatory_lock(ino
);
1799 static inline int locks_verify_locked(struct inode
*inode
)
1801 if (mandatory_lock(inode
))
1802 return locks_mandatory_locked(inode
);
1806 static inline int locks_verify_truncate(struct inode
*inode
,
1810 if (inode
->i_flock
&& mandatory_lock(inode
))
1811 return locks_mandatory_area(
1812 FLOCK_VERIFY_WRITE
, inode
, filp
,
1813 size
< inode
->i_size
? size
: inode
->i_size
,
1814 (size
< inode
->i_size
? inode
->i_size
- size
1815 : size
- inode
->i_size
)
1820 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
1823 return __break_lease(inode
, mode
);
1826 #else /* !CONFIG_FILE_LOCKING */
1827 static inline int locks_mandatory_locked(struct inode
*inode
)
1832 static inline int locks_mandatory_area(int rw
, struct inode
*inode
,
1833 struct file
*filp
, loff_t offset
,
1839 static inline int __mandatory_lock(struct inode
*inode
)
1844 static inline int mandatory_lock(struct inode
*inode
)
1849 static inline int locks_verify_locked(struct inode
*inode
)
1854 static inline int locks_verify_truncate(struct inode
*inode
, struct file
*filp
,
1860 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
1865 #endif /* CONFIG_FILE_LOCKING */
1870 const char *name
; /* pointer to actual string */
1871 const __user
char *uptr
; /* original userland pointer */
1872 struct audit_names
*aname
;
1873 bool separate
; /* should "name" be freed? */
1876 extern int do_truncate(struct dentry
*, loff_t start
, unsigned int time_attrs
,
1878 extern int do_fallocate(struct file
*file
, int mode
, loff_t offset
,
1880 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
1882 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
1883 extern struct file
*filp_open(const char *, int, umode_t
);
1884 extern struct file
*file_open_root(struct dentry
*, struct vfsmount
*,
1886 extern struct file
* dentry_open(const struct path
*, int, const struct cred
*);
1887 extern int filp_close(struct file
*, fl_owner_t id
);
1889 extern struct filename
*getname(const char __user
*);
1895 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
1896 int (*open
)(struct inode
*, struct file
*),
1898 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
1902 extern int ioctl_preallocate(struct file
*filp
, void __user
*argp
);
1905 extern void __init
vfs_caches_init_early(void);
1906 extern void __init
vfs_caches_init(unsigned long);
1908 extern struct kmem_cache
*names_cachep
;
1910 extern void final_putname(struct filename
*name
);
1912 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
1913 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
1914 #ifndef CONFIG_AUDITSYSCALL
1915 #define putname(name) final_putname(name)
1917 extern void putname(struct filename
*name
);
1921 extern int register_blkdev(unsigned int, const char *);
1922 extern void unregister_blkdev(unsigned int, const char *);
1923 extern struct block_device
*bdget(dev_t
);
1924 extern struct block_device
*bdgrab(struct block_device
*bdev
);
1925 extern void bd_set_size(struct block_device
*, loff_t size
);
1926 extern sector_t
blkdev_max_block(struct block_device
*bdev
);
1927 extern void bd_forget(struct inode
*inode
);
1928 extern void bdput(struct block_device
*);
1929 extern void invalidate_bdev(struct block_device
*);
1930 extern void iterate_bdevs(void (*)(struct block_device
*, void *), void *);
1931 extern int sync_blockdev(struct block_device
*bdev
);
1932 extern void kill_bdev(struct block_device
*);
1933 extern struct super_block
*freeze_bdev(struct block_device
*);
1934 extern void emergency_thaw_all(void);
1935 extern int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
);
1936 extern int fsync_bdev(struct block_device
*);
1938 static inline void bd_forget(struct inode
*inode
) {}
1939 static inline int sync_blockdev(struct block_device
*bdev
) { return 0; }
1940 static inline void kill_bdev(struct block_device
*bdev
) {}
1941 static inline void invalidate_bdev(struct block_device
*bdev
) {}
1943 static inline struct super_block
*freeze_bdev(struct block_device
*sb
)
1948 static inline int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
)
1953 static inline void iterate_bdevs(void (*f
)(struct block_device
*, void *), void *arg
)
1957 extern int sync_filesystem(struct super_block
*);
1958 extern const struct file_operations def_blk_fops
;
1959 extern const struct file_operations def_chr_fops
;
1960 extern const struct file_operations bad_sock_fops
;
1961 extern const struct file_operations def_fifo_fops
;
1963 extern int ioctl_by_bdev(struct block_device
*, unsigned, unsigned long);
1964 extern int blkdev_ioctl(struct block_device
*, fmode_t
, unsigned, unsigned long);
1965 extern long compat_blkdev_ioctl(struct file
*, unsigned, unsigned long);
1966 extern int blkdev_get(struct block_device
*bdev
, fmode_t mode
, void *holder
);
1967 extern struct block_device
*blkdev_get_by_path(const char *path
, fmode_t mode
,
1969 extern struct block_device
*blkdev_get_by_dev(dev_t dev
, fmode_t mode
,
1971 extern int blkdev_put(struct block_device
*bdev
, fmode_t mode
);
1973 extern int bd_link_disk_holder(struct block_device
*bdev
, struct gendisk
*disk
);
1974 extern void bd_unlink_disk_holder(struct block_device
*bdev
,
1975 struct gendisk
*disk
);
1977 static inline int bd_link_disk_holder(struct block_device
*bdev
,
1978 struct gendisk
*disk
)
1982 static inline void bd_unlink_disk_holder(struct block_device
*bdev
,
1983 struct gendisk
*disk
)
1990 #define CHRDEV_MAJOR_HASH_SIZE 255
1991 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
1992 extern int register_chrdev_region(dev_t
, unsigned, const char *);
1993 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
1994 unsigned int count
, const char *name
,
1995 const struct file_operations
*fops
);
1996 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
1997 unsigned int count
, const char *name
);
1998 extern void unregister_chrdev_region(dev_t
, unsigned);
1999 extern void chrdev_show(struct seq_file
*,off_t
);
2001 static inline int register_chrdev(unsigned int major
, const char *name
,
2002 const struct file_operations
*fops
)
2004 return __register_chrdev(major
, 0, 256, name
, fops
);
2007 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2009 __unregister_chrdev(major
, 0, 256, name
);
2012 /* fs/block_dev.c */
2013 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2014 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2017 #define BLKDEV_MAJOR_HASH_SIZE 255
2018 extern const char *__bdevname(dev_t
, char *buffer
);
2019 extern const char *bdevname(struct block_device
*bdev
, char *buffer
);
2020 extern struct block_device
*lookup_bdev(const char *);
2021 extern void blkdev_show(struct seq_file
*,off_t
);
2024 #define BLKDEV_MAJOR_HASH_SIZE 0
2027 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2029 /* Invalid inode operations -- fs/bad_inode.c */
2030 extern void make_bad_inode(struct inode
*);
2031 extern int is_bad_inode(struct inode
*);
2033 extern const struct file_operations read_pipefifo_fops
;
2034 extern const struct file_operations write_pipefifo_fops
;
2035 extern const struct file_operations rdwr_pipefifo_fops
;
2039 * return READ, READA, or WRITE
2041 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2044 * return data direction, READ or WRITE
2046 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2048 extern void check_disk_size_change(struct gendisk
*disk
,
2049 struct block_device
*bdev
);
2050 extern int revalidate_disk(struct gendisk
*);
2051 extern int check_disk_change(struct block_device
*);
2052 extern int __invalidate_device(struct block_device
*, bool);
2053 extern int invalidate_partition(struct gendisk
*, int);
2055 unsigned long invalidate_mapping_pages(struct address_space
*mapping
,
2056 pgoff_t start
, pgoff_t end
);
2058 static inline void invalidate_remote_inode(struct inode
*inode
)
2060 if (S_ISREG(inode
->i_mode
) || S_ISDIR(inode
->i_mode
) ||
2061 S_ISLNK(inode
->i_mode
))
2062 invalidate_mapping_pages(inode
->i_mapping
, 0, -1);
2064 extern int invalidate_inode_pages2(struct address_space
*mapping
);
2065 extern int invalidate_inode_pages2_range(struct address_space
*mapping
,
2066 pgoff_t start
, pgoff_t end
);
2067 extern int write_inode_now(struct inode
*, int);
2068 extern int filemap_fdatawrite(struct address_space
*);
2069 extern int filemap_flush(struct address_space
*);
2070 extern int filemap_fdatawait(struct address_space
*);
2071 extern int filemap_fdatawait_range(struct address_space
*, loff_t lstart
,
2073 extern int filemap_write_and_wait(struct address_space
*mapping
);
2074 extern int filemap_write_and_wait_range(struct address_space
*mapping
,
2075 loff_t lstart
, loff_t lend
);
2076 extern int __filemap_fdatawrite_range(struct address_space
*mapping
,
2077 loff_t start
, loff_t end
, int sync_mode
);
2078 extern int filemap_fdatawrite_range(struct address_space
*mapping
,
2079 loff_t start
, loff_t end
);
2081 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
2083 extern int vfs_fsync(struct file
*file
, int datasync
);
2084 extern int generic_write_sync(struct file
*file
, loff_t pos
, loff_t count
);
2085 extern void emergency_sync(void);
2086 extern void emergency_remount(void);
2088 extern sector_t
bmap(struct inode
*, sector_t
);
2090 extern int notify_change(struct dentry
*, struct iattr
*);
2091 extern int inode_permission(struct inode
*, int);
2092 extern int generic_permission(struct inode
*, int);
2094 static inline bool execute_ok(struct inode
*inode
)
2096 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
2100 * get_write_access() gets write permission for a file.
2101 * put_write_access() releases this write permission.
2102 * This is used for regular files.
2103 * We cannot support write (and maybe mmap read-write shared) accesses and
2104 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2105 * can have the following values:
2106 * 0: no writers, no VM_DENYWRITE mappings
2107 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2108 * > 0: (i_writecount) users are writing to the file.
2110 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2111 * except for the cases where we don't hold i_writecount yet. Then we need to
2112 * use {get,deny}_write_access() - these functions check the sign and refuse
2113 * to do the change if sign is wrong.
2115 static inline int get_write_access(struct inode
*inode
)
2117 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2119 static inline int deny_write_access(struct file
*file
)
2121 struct inode
*inode
= file
->f_path
.dentry
->d_inode
;
2122 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2124 static inline void put_write_access(struct inode
* inode
)
2126 atomic_dec(&inode
->i_writecount
);
2128 static inline void allow_write_access(struct file
*file
)
2131 atomic_inc(&file
->f_path
.dentry
->d_inode
->i_writecount
);
2134 static inline void i_readcount_dec(struct inode
*inode
)
2136 BUG_ON(!atomic_read(&inode
->i_readcount
));
2137 atomic_dec(&inode
->i_readcount
);
2139 static inline void i_readcount_inc(struct inode
*inode
)
2141 atomic_inc(&inode
->i_readcount
);
2144 static inline void i_readcount_dec(struct inode
*inode
)
2148 static inline void i_readcount_inc(struct inode
*inode
)
2153 extern int do_pipe_flags(int *, int);
2155 extern int kernel_read(struct file
*, loff_t
, char *, unsigned long);
2156 extern struct file
* open_exec(const char *);
2158 /* fs/dcache.c -- generic fs support functions */
2159 extern int is_subdir(struct dentry
*, struct dentry
*);
2160 extern int path_is_under(struct path
*, struct path
*);
2161 extern ino_t
find_inode_number(struct dentry
*, struct qstr
*);
2163 #include <linux/err.h>
2165 /* needed for stackable file system support */
2166 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int origin
);
2168 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int origin
);
2170 extern int inode_init_always(struct super_block
*, struct inode
*);
2171 extern void inode_init_once(struct inode
*);
2172 extern void address_space_init_once(struct address_space
*mapping
);
2173 extern void ihold(struct inode
* inode
);
2174 extern void iput(struct inode
*);
2175 extern struct inode
* igrab(struct inode
*);
2176 extern ino_t
iunique(struct super_block
*, ino_t
);
2177 extern int inode_needs_sync(struct inode
*inode
);
2178 extern int generic_delete_inode(struct inode
*inode
);
2179 static inline int generic_drop_inode(struct inode
*inode
)
2181 return !inode
->i_nlink
|| inode_unhashed(inode
);
2184 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
2185 unsigned long hashval
, int (*test
)(struct inode
*, void *),
2187 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
2188 int (*test
)(struct inode
*, void *), void *data
);
2189 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
2191 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
2192 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
2193 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
2194 extern int insert_inode_locked(struct inode
*);
2195 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2196 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
2198 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
2200 extern void unlock_new_inode(struct inode
*);
2201 extern unsigned int get_next_ino(void);
2203 extern void __iget(struct inode
* inode
);
2204 extern void iget_failed(struct inode
*);
2205 extern void clear_inode(struct inode
*);
2206 extern void __destroy_inode(struct inode
*);
2207 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
2208 extern struct inode
*new_inode(struct super_block
*sb
);
2209 extern void free_inode_nonrcu(struct inode
*inode
);
2210 extern int should_remove_suid(struct dentry
*);
2211 extern int file_remove_suid(struct file
*);
2213 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
2214 static inline void insert_inode_hash(struct inode
*inode
)
2216 __insert_inode_hash(inode
, inode
->i_ino
);
2219 extern void __remove_inode_hash(struct inode
*);
2220 static inline void remove_inode_hash(struct inode
*inode
)
2222 if (!inode_unhashed(inode
))
2223 __remove_inode_hash(inode
);
2226 extern void inode_sb_list_add(struct inode
*inode
);
2229 extern void submit_bio(int, struct bio
*);
2230 extern int bdev_read_only(struct block_device
*);
2232 extern int set_blocksize(struct block_device
*, int);
2233 extern int sb_set_blocksize(struct super_block
*, int);
2234 extern int sb_min_blocksize(struct super_block
*, int);
2236 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
2237 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
2238 extern int generic_file_remap_pages(struct vm_area_struct
*, unsigned long addr
,
2239 unsigned long size
, pgoff_t pgoff
);
2240 extern int file_read_actor(read_descriptor_t
* desc
, struct page
*page
, unsigned long offset
, unsigned long size
);
2241 int generic_write_checks(struct file
*file
, loff_t
*pos
, size_t *count
, int isblk
);
2242 extern ssize_t
generic_file_aio_read(struct kiocb
*, const struct iovec
*, unsigned long, loff_t
);
2243 extern ssize_t
__generic_file_aio_write(struct kiocb
*, const struct iovec
*, unsigned long,
2245 extern ssize_t
generic_file_aio_write(struct kiocb
*, const struct iovec
*, unsigned long, loff_t
);
2246 extern ssize_t
generic_file_direct_write(struct kiocb
*, const struct iovec
*,
2247 unsigned long *, loff_t
, loff_t
*, size_t, size_t);
2248 extern ssize_t
generic_file_buffered_write(struct kiocb
*, const struct iovec
*,
2249 unsigned long, loff_t
, loff_t
*, size_t, ssize_t
);
2250 extern ssize_t
do_sync_read(struct file
*filp
, char __user
*buf
, size_t len
, loff_t
*ppos
);
2251 extern ssize_t
do_sync_write(struct file
*filp
, const char __user
*buf
, size_t len
, loff_t
*ppos
);
2252 extern int generic_segment_checks(const struct iovec
*iov
,
2253 unsigned long *nr_segs
, size_t *count
, int access_flags
);
2255 /* fs/block_dev.c */
2256 extern ssize_t
blkdev_aio_read(struct kiocb
*iocb
, const struct iovec
*iov
,
2257 unsigned long nr_segs
, loff_t pos
);
2258 extern ssize_t
blkdev_aio_write(struct kiocb
*iocb
, const struct iovec
*iov
,
2259 unsigned long nr_segs
, loff_t pos
);
2260 extern int blkdev_fsync(struct file
*filp
, loff_t start
, loff_t end
,
2262 extern void block_sync_page(struct page
*page
);
2265 extern ssize_t
generic_file_splice_read(struct file
*, loff_t
*,
2266 struct pipe_inode_info
*, size_t, unsigned int);
2267 extern ssize_t
default_file_splice_read(struct file
*, loff_t
*,
2268 struct pipe_inode_info
*, size_t, unsigned int);
2269 extern ssize_t
generic_file_splice_write(struct pipe_inode_info
*,
2270 struct file
*, loff_t
*, size_t, unsigned int);
2271 extern ssize_t
generic_splice_sendpage(struct pipe_inode_info
*pipe
,
2272 struct file
*out
, loff_t
*, size_t len
, unsigned int flags
);
2273 extern long do_splice_direct(struct file
*in
, loff_t
*ppos
, struct file
*out
,
2274 size_t len
, unsigned int flags
);
2277 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
2278 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int origin
);
2279 extern loff_t
no_llseek(struct file
*file
, loff_t offset
, int origin
);
2280 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int origin
);
2281 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
2282 int origin
, loff_t maxsize
, loff_t eof
);
2283 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
2284 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
2286 #ifdef CONFIG_FS_XIP
2287 extern ssize_t
xip_file_read(struct file
*filp
, char __user
*buf
, size_t len
,
2289 extern int xip_file_mmap(struct file
* file
, struct vm_area_struct
* vma
);
2290 extern ssize_t
xip_file_write(struct file
*filp
, const char __user
*buf
,
2291 size_t len
, loff_t
*ppos
);
2292 extern int xip_truncate_page(struct address_space
*mapping
, loff_t from
);
2294 static inline int xip_truncate_page(struct address_space
*mapping
, loff_t from
)
2301 typedef void (dio_submit_t
)(int rw
, struct bio
*bio
, struct inode
*inode
,
2302 loff_t file_offset
);
2305 /* need locking between buffered and direct access */
2308 /* filesystem does not support filling holes */
2309 DIO_SKIP_HOLES
= 0x02,
2312 void dio_end_io(struct bio
*bio
, int error
);
2314 ssize_t
__blockdev_direct_IO(int rw
, struct kiocb
*iocb
, struct inode
*inode
,
2315 struct block_device
*bdev
, const struct iovec
*iov
, loff_t offset
,
2316 unsigned long nr_segs
, get_block_t get_block
, dio_iodone_t end_io
,
2317 dio_submit_t submit_io
, int flags
);
2319 static inline ssize_t
blockdev_direct_IO(int rw
, struct kiocb
*iocb
,
2320 struct inode
*inode
, const struct iovec
*iov
, loff_t offset
,
2321 unsigned long nr_segs
, get_block_t get_block
)
2323 return __blockdev_direct_IO(rw
, iocb
, inode
, inode
->i_sb
->s_bdev
, iov
,
2324 offset
, nr_segs
, get_block
, NULL
, NULL
,
2325 DIO_LOCKING
| DIO_SKIP_HOLES
);
2329 void inode_dio_wait(struct inode
*inode
);
2330 void inode_dio_done(struct inode
*inode
);
2332 extern const struct file_operations generic_ro_fops
;
2334 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2336 extern int vfs_readlink(struct dentry
*, char __user
*, int, const char *);
2337 extern int vfs_follow_link(struct nameidata
*, const char *);
2338 extern int page_readlink(struct dentry
*, char __user
*, int);
2339 extern void *page_follow_link_light(struct dentry
*, struct nameidata
*);
2340 extern void page_put_link(struct dentry
*, struct nameidata
*, void *);
2341 extern int __page_symlink(struct inode
*inode
, const char *symname
, int len
,
2343 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
2344 extern const struct inode_operations page_symlink_inode_operations
;
2345 extern int generic_readlink(struct dentry
*, char __user
*, int);
2346 extern void generic_fillattr(struct inode
*, struct kstat
*);
2347 extern int vfs_getattr(struct vfsmount
*, struct dentry
*, struct kstat
*);
2348 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2349 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2350 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2351 loff_t
inode_get_bytes(struct inode
*inode
);
2352 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
2354 extern int vfs_readdir(struct file
*, filldir_t
, void *);
2356 extern int vfs_stat(const char __user
*, struct kstat
*);
2357 extern int vfs_lstat(const char __user
*, struct kstat
*);
2358 extern int vfs_fstat(unsigned int, struct kstat
*);
2359 extern int vfs_fstatat(int , const char __user
*, struct kstat
*, int);
2361 extern int do_vfs_ioctl(struct file
*filp
, unsigned int fd
, unsigned int cmd
,
2363 extern int __generic_block_fiemap(struct inode
*inode
,
2364 struct fiemap_extent_info
*fieinfo
,
2365 loff_t start
, loff_t len
,
2366 get_block_t
*get_block
);
2367 extern int generic_block_fiemap(struct inode
*inode
,
2368 struct fiemap_extent_info
*fieinfo
, u64 start
,
2369 u64 len
, get_block_t
*get_block
);
2371 extern void get_filesystem(struct file_system_type
*fs
);
2372 extern void put_filesystem(struct file_system_type
*fs
);
2373 extern struct file_system_type
*get_fs_type(const char *name
);
2374 extern struct super_block
*get_super(struct block_device
*);
2375 extern struct super_block
*get_super_thawed(struct block_device
*);
2376 extern struct super_block
*get_active_super(struct block_device
*bdev
);
2377 extern void drop_super(struct super_block
*sb
);
2378 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
2379 extern void iterate_supers_type(struct file_system_type
*,
2380 void (*)(struct super_block
*, void *), void *);
2382 extern int dcache_dir_open(struct inode
*, struct file
*);
2383 extern int dcache_dir_close(struct inode
*, struct file
*);
2384 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
2385 extern int dcache_readdir(struct file
*, void *, filldir_t
);
2386 extern int simple_setattr(struct dentry
*, struct iattr
*);
2387 extern int simple_getattr(struct vfsmount
*, struct dentry
*, struct kstat
*);
2388 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
2389 extern int simple_open(struct inode
*inode
, struct file
*file
);
2390 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
2391 extern int simple_unlink(struct inode
*, struct dentry
*);
2392 extern int simple_rmdir(struct inode
*, struct dentry
*);
2393 extern int simple_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*);
2394 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
2395 extern int simple_empty(struct dentry
*);
2396 extern int simple_readpage(struct file
*file
, struct page
*page
);
2397 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
2398 loff_t pos
, unsigned len
, unsigned flags
,
2399 struct page
**pagep
, void **fsdata
);
2400 extern int simple_write_end(struct file
*file
, struct address_space
*mapping
,
2401 loff_t pos
, unsigned len
, unsigned copied
,
2402 struct page
*page
, void *fsdata
);
2404 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
2405 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
2406 extern const struct file_operations simple_dir_operations
;
2407 extern const struct inode_operations simple_dir_inode_operations
;
2408 struct tree_descr
{ char *name
; const struct file_operations
*ops
; int mode
; };
2409 struct dentry
*d_alloc_name(struct dentry
*, const char *);
2410 extern int simple_fill_super(struct super_block
*, unsigned long, struct tree_descr
*);
2411 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
2412 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
2414 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
2415 loff_t
*ppos
, const void *from
, size_t available
);
2416 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
2417 const void __user
*from
, size_t count
);
2419 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
2421 extern int generic_check_addressable(unsigned, u64
);
2423 #ifdef CONFIG_MIGRATION
2424 extern int buffer_migrate_page(struct address_space
*,
2425 struct page
*, struct page
*,
2428 #define buffer_migrate_page NULL
2431 extern int inode_change_ok(const struct inode
*, struct iattr
*);
2432 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
2433 extern void setattr_copy(struct inode
*inode
, const struct iattr
*attr
);
2435 extern int file_update_time(struct file
*file
);
2437 extern int generic_show_options(struct seq_file
*m
, struct dentry
*root
);
2438 extern void save_mount_options(struct super_block
*sb
, char *options
);
2439 extern void replace_mount_options(struct super_block
*sb
, char *options
);
2441 static inline ino_t
parent_ino(struct dentry
*dentry
)
2446 * Don't strictly need d_lock here? If the parent ino could change
2447 * then surely we'd have a deeper race in the caller?
2449 spin_lock(&dentry
->d_lock
);
2450 res
= dentry
->d_parent
->d_inode
->i_ino
;
2451 spin_unlock(&dentry
->d_lock
);
2455 /* Transaction based IO helpers */
2458 * An argresp is stored in an allocated page and holds the
2459 * size of the argument or response, along with its content
2461 struct simple_transaction_argresp
{
2466 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2468 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
2470 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
2471 size_t size
, loff_t
*pos
);
2472 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
2474 void simple_transaction_set(struct file
*file
, size_t n
);
2477 * simple attribute files
2479 * These attributes behave similar to those in sysfs:
2481 * Writing to an attribute immediately sets a value, an open file can be
2482 * written to multiple times.
2484 * Reading from an attribute creates a buffer from the value that might get
2485 * read with multiple read calls. When the attribute has been read
2486 * completely, no further read calls are possible until the file is opened
2489 * All attributes contain a text representation of a numeric value
2490 * that are accessed with the get() and set() functions.
2492 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2493 static int __fops ## _open(struct inode *inode, struct file *file) \
2495 __simple_attr_check_format(__fmt, 0ull); \
2496 return simple_attr_open(inode, file, __get, __set, __fmt); \
2498 static const struct file_operations __fops = { \
2499 .owner = THIS_MODULE, \
2500 .open = __fops ## _open, \
2501 .release = simple_attr_release, \
2502 .read = simple_attr_read, \
2503 .write = simple_attr_write, \
2504 .llseek = generic_file_llseek, \
2507 static inline __printf(1, 2)
2508 void __simple_attr_check_format(const char *fmt
, ...)
2510 /* don't do anything, just let the compiler check the arguments; */
2513 int simple_attr_open(struct inode
*inode
, struct file
*file
,
2514 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
2516 int simple_attr_release(struct inode
*inode
, struct file
*file
);
2517 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
2518 size_t len
, loff_t
*ppos
);
2519 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
2520 size_t len
, loff_t
*ppos
);
2523 int proc_nr_files(struct ctl_table
*table
, int write
,
2524 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
2525 int proc_nr_dentry(struct ctl_table
*table
, int write
,
2526 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
2527 int proc_nr_inodes(struct ctl_table
*table
, int write
,
2528 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
2529 int __init
get_filesystem_list(char *buf
);
2531 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2532 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2534 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2535 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2536 (flag & __FMODE_NONOTIFY)))
2538 static inline int is_sxid(umode_t mode
)
2540 return (mode
& S_ISUID
) || ((mode
& S_ISGID
) && (mode
& S_IXGRP
));
2543 static inline void inode_has_no_xattr(struct inode
*inode
)
2545 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& MS_NOSEC
))
2546 inode
->i_flags
|= S_NOSEC
;
2549 #endif /* _LINUX_FS_H */