5 #include <linux/linkage.h>
6 #include <linux/wait.h>
7 #include <linux/kdev_t.h>
8 #include <linux/dcache.h>
9 #include <linux/path.h>
10 #include <linux/stat.h>
11 #include <linux/cache.h>
12 #include <linux/list.h>
13 #include <linux/list_lru.h>
14 #include <linux/llist.h>
15 #include <linux/radix-tree.h>
16 #include <linux/rbtree.h>
17 #include <linux/init.h>
18 #include <linux/pid.h>
19 #include <linux/bug.h>
20 #include <linux/mutex.h>
21 #include <linux/rwsem.h>
22 #include <linux/capability.h>
23 #include <linux/semaphore.h>
24 #include <linux/fiemap.h>
25 #include <linux/rculist_bl.h>
26 #include <linux/atomic.h>
27 #include <linux/shrinker.h>
28 #include <linux/migrate_mode.h>
29 #include <linux/uidgid.h>
30 #include <linux/lockdep.h>
31 #include <linux/percpu-rwsem.h>
32 #include <linux/blk_types.h>
34 #include <asm/byteorder.h>
35 #include <uapi/linux/fs.h>
37 struct backing_dev_info
;
38 struct export_operations
;
44 struct pipe_inode_info
;
45 struct poll_table_struct
;
47 struct vm_area_struct
;
50 struct swap_info_struct
;
52 struct workqueue_struct
;
55 extern void __init
inode_init(void);
56 extern void __init
inode_init_early(void);
57 extern void __init
files_init(unsigned long);
59 extern struct files_stat_struct files_stat
;
60 extern unsigned long get_max_files(void);
61 extern int sysctl_nr_open
;
62 extern struct inodes_stat_t inodes_stat
;
63 extern int leases_enable
, lease_break_time
;
64 extern int sysctl_protected_symlinks
;
65 extern int sysctl_protected_hardlinks
;
68 typedef int (get_block_t
)(struct inode
*inode
, sector_t iblock
,
69 struct buffer_head
*bh_result
, int create
);
70 typedef void (dio_iodone_t
)(struct kiocb
*iocb
, loff_t offset
,
71 ssize_t bytes
, void *private);
73 #define MAY_EXEC 0x00000001
74 #define MAY_WRITE 0x00000002
75 #define MAY_READ 0x00000004
76 #define MAY_APPEND 0x00000008
77 #define MAY_ACCESS 0x00000010
78 #define MAY_OPEN 0x00000020
79 #define MAY_CHDIR 0x00000040
80 /* called from RCU mode, don't block */
81 #define MAY_NOT_BLOCK 0x00000080
84 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
85 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
88 /* file is open for reading */
89 #define FMODE_READ ((__force fmode_t)0x1)
90 /* file is open for writing */
91 #define FMODE_WRITE ((__force fmode_t)0x2)
92 /* file is seekable */
93 #define FMODE_LSEEK ((__force fmode_t)0x4)
94 /* file can be accessed using pread */
95 #define FMODE_PREAD ((__force fmode_t)0x8)
96 /* file can be accessed using pwrite */
97 #define FMODE_PWRITE ((__force fmode_t)0x10)
98 /* File is opened for execution with sys_execve / sys_uselib */
99 #define FMODE_EXEC ((__force fmode_t)0x20)
100 /* File is opened with O_NDELAY (only set for block devices) */
101 #define FMODE_NDELAY ((__force fmode_t)0x40)
102 /* File is opened with O_EXCL (only set for block devices) */
103 #define FMODE_EXCL ((__force fmode_t)0x80)
104 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
105 (specialy hack for floppy.c) */
106 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
107 /* 32bit hashes as llseek() offset (for directories) */
108 #define FMODE_32BITHASH ((__force fmode_t)0x200)
109 /* 64bit hashes as llseek() offset (for directories) */
110 #define FMODE_64BITHASH ((__force fmode_t)0x400)
113 * Don't update ctime and mtime.
115 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
116 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
118 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
120 /* Expect random access pattern */
121 #define FMODE_RANDOM ((__force fmode_t)0x1000)
123 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
124 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
126 /* File is opened with O_PATH; almost nothing can be done with it */
127 #define FMODE_PATH ((__force fmode_t)0x4000)
129 /* File needs atomic accesses to f_pos */
130 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
131 /* Write access to underlying fs */
132 #define FMODE_WRITER ((__force fmode_t)0x10000)
133 /* Has read method(s) */
134 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
135 /* Has write method(s) */
136 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
138 /* File was opened by fanotify and shouldn't generate fanotify events */
139 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
142 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
143 * that indicates that they should check the contents of the iovec are
144 * valid, but not check the memory that the iovec elements
147 #define CHECK_IOVEC_ONLY -1
150 * The below are the various read and write types that we support. Some of
151 * them include behavioral modifiers that send information down to the
152 * block layer and IO scheduler. Terminology:
154 * The block layer uses device plugging to defer IO a little bit, in
155 * the hope that we will see more IO very shortly. This increases
156 * coalescing of adjacent IO and thus reduces the number of IOs we
157 * have to send to the device. It also allows for better queuing,
158 * if the IO isn't mergeable. If the caller is going to be waiting
159 * for the IO, then he must ensure that the device is unplugged so
160 * that the IO is dispatched to the driver.
162 * All IO is handled async in Linux. This is fine for background
163 * writes, but for reads or writes that someone waits for completion
164 * on, we want to notify the block layer and IO scheduler so that they
165 * know about it. That allows them to make better scheduling
166 * decisions. So when the below references 'sync' and 'async', it
167 * is referencing this priority hint.
169 * With that in mind, the available types are:
171 * READ A normal read operation. Device will be plugged.
172 * READ_SYNC A synchronous read. Device is not plugged, caller can
173 * immediately wait on this read without caring about
175 * READA Used for read-ahead operations. Lower priority, and the
176 * block layer could (in theory) choose to ignore this
177 * request if it runs into resource problems.
178 * WRITE A normal async write. Device will be plugged.
179 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
180 * the hint that someone will be waiting on this IO
181 * shortly. The write equivalent of READ_SYNC.
182 * WRITE_ODIRECT Special case write for O_DIRECT only.
183 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
184 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
185 * non-volatile media on completion.
186 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
187 * by a cache flush and data is guaranteed to be on
188 * non-volatile media on completion.
191 #define RW_MASK REQ_WRITE
192 #define RWA_MASK REQ_RAHEAD
195 #define WRITE RW_MASK
196 #define READA RWA_MASK
198 #define READ_SYNC (READ | REQ_SYNC)
199 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
200 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
201 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
202 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
203 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
206 * Attribute flags. These should be or-ed together to figure out what
209 #define ATTR_MODE (1 << 0)
210 #define ATTR_UID (1 << 1)
211 #define ATTR_GID (1 << 2)
212 #define ATTR_SIZE (1 << 3)
213 #define ATTR_ATIME (1 << 4)
214 #define ATTR_MTIME (1 << 5)
215 #define ATTR_CTIME (1 << 6)
216 #define ATTR_ATIME_SET (1 << 7)
217 #define ATTR_MTIME_SET (1 << 8)
218 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
219 #define ATTR_ATTR_FLAG (1 << 10)
220 #define ATTR_KILL_SUID (1 << 11)
221 #define ATTR_KILL_SGID (1 << 12)
222 #define ATTR_FILE (1 << 13)
223 #define ATTR_KILL_PRIV (1 << 14)
224 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
225 #define ATTR_TIMES_SET (1 << 16)
228 * Whiteout is represented by a char device. The following constants define the
229 * mode and device number to use.
231 #define WHITEOUT_MODE 0
232 #define WHITEOUT_DEV 0
235 * This is the Inode Attributes structure, used for notify_change(). It
236 * uses the above definitions as flags, to know which values have changed.
237 * Also, in this manner, a Filesystem can look at only the values it cares
238 * about. Basically, these are the attributes that the VFS layer can
239 * request to change from the FS layer.
241 * Derek Atkins <warlord@MIT.EDU> 94-10-20
244 unsigned int ia_valid
;
249 struct timespec ia_atime
;
250 struct timespec ia_mtime
;
251 struct timespec ia_ctime
;
254 * Not an attribute, but an auxiliary info for filesystems wanting to
255 * implement an ftruncate() like method. NOTE: filesystem should
256 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
258 struct file
*ia_file
;
262 * Includes for diskquotas.
264 #include <linux/quota.h>
267 * Maximum number of layers of fs stack. Needs to be limited to
268 * prevent kernel stack overflow
270 #define FILESYSTEM_MAX_STACK_DEPTH 2
273 * enum positive_aop_returns - aop return codes with specific semantics
275 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
276 * completed, that the page is still locked, and
277 * should be considered active. The VM uses this hint
278 * to return the page to the active list -- it won't
279 * be a candidate for writeback again in the near
280 * future. Other callers must be careful to unlock
281 * the page if they get this return. Returned by
284 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
285 * unlocked it and the page might have been truncated.
286 * The caller should back up to acquiring a new page and
287 * trying again. The aop will be taking reasonable
288 * precautions not to livelock. If the caller held a page
289 * reference, it should drop it before retrying. Returned
292 * address_space_operation functions return these large constants to indicate
293 * special semantics to the caller. These are much larger than the bytes in a
294 * page to allow for functions that return the number of bytes operated on in a
298 enum positive_aop_returns
{
299 AOP_WRITEPAGE_ACTIVATE
= 0x80000,
300 AOP_TRUNCATED_PAGE
= 0x80001,
303 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
304 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
305 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
306 * helper code (eg buffer layer)
307 * to clear GFP_FS from alloc */
310 * oh the beauties of C type declarations.
313 struct address_space
;
314 struct writeback_control
;
317 * "descriptor" for what we're up to with a read.
318 * This allows us to use the same read code yet
319 * have multiple different users of the data that
320 * we read from a file.
322 * The simplest case just copies the data to user
335 typedef int (*read_actor_t
)(read_descriptor_t
*, struct page
*,
336 unsigned long, unsigned long);
338 struct address_space_operations
{
339 int (*writepage
)(struct page
*page
, struct writeback_control
*wbc
);
340 int (*readpage
)(struct file
*, struct page
*);
342 /* Write back some dirty pages from this mapping. */
343 int (*writepages
)(struct address_space
*, struct writeback_control
*);
345 /* Set a page dirty. Return true if this dirtied it */
346 int (*set_page_dirty
)(struct page
*page
);
348 int (*readpages
)(struct file
*filp
, struct address_space
*mapping
,
349 struct list_head
*pages
, unsigned nr_pages
);
351 int (*write_begin
)(struct file
*, struct address_space
*mapping
,
352 loff_t pos
, unsigned len
, unsigned flags
,
353 struct page
**pagep
, void **fsdata
);
354 int (*write_end
)(struct file
*, struct address_space
*mapping
,
355 loff_t pos
, unsigned len
, unsigned copied
,
356 struct page
*page
, void *fsdata
);
358 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
359 sector_t (*bmap
)(struct address_space
*, sector_t
);
360 void (*invalidatepage
) (struct page
*, unsigned int, unsigned int);
361 int (*releasepage
) (struct page
*, gfp_t
);
362 void (*freepage
)(struct page
*);
363 ssize_t (*direct_IO
)(int, struct kiocb
*, struct iov_iter
*iter
, loff_t offset
);
364 int (*get_xip_mem
)(struct address_space
*, pgoff_t
, int,
365 void **, unsigned long *);
367 * migrate the contents of a page to the specified target. If
368 * migrate_mode is MIGRATE_ASYNC, it must not block.
370 int (*migratepage
) (struct address_space
*,
371 struct page
*, struct page
*, enum migrate_mode
);
372 int (*launder_page
) (struct page
*);
373 int (*is_partially_uptodate
) (struct page
*, unsigned long,
375 void (*is_dirty_writeback
) (struct page
*, bool *, bool *);
376 int (*error_remove_page
)(struct address_space
*, struct page
*);
378 /* swapfile support */
379 int (*swap_activate
)(struct swap_info_struct
*sis
, struct file
*file
,
381 void (*swap_deactivate
)(struct file
*file
);
384 extern const struct address_space_operations empty_aops
;
387 * pagecache_write_begin/pagecache_write_end must be used by general code
388 * to write into the pagecache.
390 int pagecache_write_begin(struct file
*, struct address_space
*mapping
,
391 loff_t pos
, unsigned len
, unsigned flags
,
392 struct page
**pagep
, void **fsdata
);
394 int pagecache_write_end(struct file
*, struct address_space
*mapping
,
395 loff_t pos
, unsigned len
, unsigned copied
,
396 struct page
*page
, void *fsdata
);
398 struct address_space
{
399 struct inode
*host
; /* owner: inode, block_device */
400 struct radix_tree_root page_tree
; /* radix tree of all pages */
401 spinlock_t tree_lock
; /* and lock protecting it */
402 atomic_t i_mmap_writable
;/* count VM_SHARED mappings */
403 struct rb_root i_mmap
; /* tree of private and shared mappings */
404 struct rw_semaphore i_mmap_rwsem
; /* protect tree, count, list */
405 /* Protected by tree_lock together with the radix tree */
406 unsigned long nrpages
; /* number of total pages */
407 unsigned long nrshadows
; /* number of shadow entries */
408 pgoff_t writeback_index
;/* writeback starts here */
409 const struct address_space_operations
*a_ops
; /* methods */
410 unsigned long flags
; /* error bits/gfp mask */
411 spinlock_t private_lock
; /* for use by the address_space */
412 struct list_head private_list
; /* ditto */
413 void *private_data
; /* ditto */
414 } __attribute__((aligned(sizeof(long))));
416 * On most architectures that alignment is already the case; but
417 * must be enforced here for CRIS, to let the least significant bit
418 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
420 struct request_queue
;
422 struct block_device
{
423 dev_t bd_dev
; /* not a kdev_t - it's a search key */
425 struct inode
* bd_inode
; /* will die */
426 struct super_block
* bd_super
;
427 struct mutex bd_mutex
; /* open/close mutex */
428 struct list_head bd_inodes
;
432 bool bd_write_holder
;
434 struct list_head bd_holder_disks
;
436 struct block_device
* bd_contains
;
437 unsigned bd_block_size
;
438 struct hd_struct
* bd_part
;
439 /* number of times partitions within this device have been opened. */
440 unsigned bd_part_count
;
442 struct gendisk
* bd_disk
;
443 struct request_queue
* bd_queue
;
444 struct list_head bd_list
;
446 * Private data. You must have bd_claim'ed the block_device
447 * to use this. NOTE: bd_claim allows an owner to claim
448 * the same device multiple times, the owner must take special
449 * care to not mess up bd_private for that case.
451 unsigned long bd_private
;
453 /* The counter of freeze processes */
454 int bd_fsfreeze_count
;
455 /* Mutex for freeze */
456 struct mutex bd_fsfreeze_mutex
;
460 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
463 #define PAGECACHE_TAG_DIRTY 0
464 #define PAGECACHE_TAG_WRITEBACK 1
465 #define PAGECACHE_TAG_TOWRITE 2
467 int mapping_tagged(struct address_space
*mapping
, int tag
);
469 static inline void i_mmap_lock_write(struct address_space
*mapping
)
471 down_write(&mapping
->i_mmap_rwsem
);
474 static inline void i_mmap_unlock_write(struct address_space
*mapping
)
476 up_write(&mapping
->i_mmap_rwsem
);
479 static inline void i_mmap_lock_read(struct address_space
*mapping
)
481 down_read(&mapping
->i_mmap_rwsem
);
484 static inline void i_mmap_unlock_read(struct address_space
*mapping
)
486 up_read(&mapping
->i_mmap_rwsem
);
490 * Might pages of this file be mapped into userspace?
492 static inline int mapping_mapped(struct address_space
*mapping
)
494 return !RB_EMPTY_ROOT(&mapping
->i_mmap
);
498 * Might pages of this file have been modified in userspace?
499 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
500 * marks vma as VM_SHARED if it is shared, and the file was opened for
501 * writing i.e. vma may be mprotected writable even if now readonly.
503 * If i_mmap_writable is negative, no new writable mappings are allowed. You
504 * can only deny writable mappings, if none exists right now.
506 static inline int mapping_writably_mapped(struct address_space
*mapping
)
508 return atomic_read(&mapping
->i_mmap_writable
) > 0;
511 static inline int mapping_map_writable(struct address_space
*mapping
)
513 return atomic_inc_unless_negative(&mapping
->i_mmap_writable
) ?
517 static inline void mapping_unmap_writable(struct address_space
*mapping
)
519 atomic_dec(&mapping
->i_mmap_writable
);
522 static inline int mapping_deny_writable(struct address_space
*mapping
)
524 return atomic_dec_unless_positive(&mapping
->i_mmap_writable
) ?
528 static inline void mapping_allow_writable(struct address_space
*mapping
)
530 atomic_inc(&mapping
->i_mmap_writable
);
534 * Use sequence counter to get consistent i_size on 32-bit processors.
536 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
537 #include <linux/seqlock.h>
538 #define __NEED_I_SIZE_ORDERED
539 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
541 #define i_size_ordered_init(inode) do { } while (0)
545 #define ACL_NOT_CACHED ((void *)(-1))
547 #define IOP_FASTPERM 0x0001
548 #define IOP_LOOKUP 0x0002
549 #define IOP_NOFOLLOW 0x0004
552 * Keep mostly read-only and often accessed (especially for
553 * the RCU path lookup and 'stat' data) fields at the beginning
554 * of the 'struct inode'
558 unsigned short i_opflags
;
561 unsigned int i_flags
;
563 #ifdef CONFIG_FS_POSIX_ACL
564 struct posix_acl
*i_acl
;
565 struct posix_acl
*i_default_acl
;
568 const struct inode_operations
*i_op
;
569 struct super_block
*i_sb
;
570 struct address_space
*i_mapping
;
572 #ifdef CONFIG_SECURITY
576 /* Stat data, not accessed from path walking */
579 * Filesystems may only read i_nlink directly. They shall use the
580 * following functions for modification:
582 * (set|clear|inc|drop)_nlink
583 * inode_(inc|dec)_link_count
586 const unsigned int i_nlink
;
587 unsigned int __i_nlink
;
591 struct timespec i_atime
;
592 struct timespec i_mtime
;
593 struct timespec i_ctime
;
594 spinlock_t i_lock
; /* i_blocks, i_bytes, maybe i_size */
595 unsigned short i_bytes
;
596 unsigned int i_blkbits
;
599 #ifdef __NEED_I_SIZE_ORDERED
600 seqcount_t i_size_seqcount
;
604 unsigned long i_state
;
605 struct mutex i_mutex
;
607 unsigned long dirtied_when
; /* jiffies of first dirtying */
609 struct hlist_node i_hash
;
610 struct list_head i_wb_list
; /* backing dev IO list */
611 struct list_head i_lru
; /* inode LRU list */
612 struct list_head i_sb_list
;
614 struct hlist_head i_dentry
;
615 struct rcu_head i_rcu
;
619 atomic_t i_dio_count
;
620 atomic_t i_writecount
;
622 atomic_t i_readcount
; /* struct files open RO */
624 const struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
625 struct file_lock_context
*i_flctx
;
626 struct address_space i_data
;
627 struct list_head i_devices
;
629 struct pipe_inode_info
*i_pipe
;
630 struct block_device
*i_bdev
;
636 #ifdef CONFIG_FSNOTIFY
637 __u32 i_fsnotify_mask
; /* all events this inode cares about */
638 struct hlist_head i_fsnotify_marks
;
641 void *i_private
; /* fs or device private pointer */
644 static inline int inode_unhashed(struct inode
*inode
)
646 return hlist_unhashed(&inode
->i_hash
);
650 * inode->i_mutex nesting subclasses for the lock validator:
652 * 0: the object of the current VFS operation
656 * 4: second non-directory
657 * 5: second parent (when locking independent directories in rename)
659 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
660 * non-directories at once.
662 * The locking order between these classes is
663 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
665 enum inode_i_mutex_lock_class
675 void lock_two_nondirectories(struct inode
*, struct inode
*);
676 void unlock_two_nondirectories(struct inode
*, struct inode
*);
679 * NOTE: in a 32bit arch with a preemptable kernel and
680 * an UP compile the i_size_read/write must be atomic
681 * with respect to the local cpu (unlike with preempt disabled),
682 * but they don't need to be atomic with respect to other cpus like in
683 * true SMP (so they need either to either locally disable irq around
684 * the read or for example on x86 they can be still implemented as a
685 * cmpxchg8b without the need of the lock prefix). For SMP compiles
686 * and 64bit archs it makes no difference if preempt is enabled or not.
688 static inline loff_t
i_size_read(const struct inode
*inode
)
690 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
695 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
696 i_size
= inode
->i_size
;
697 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
699 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
703 i_size
= inode
->i_size
;
707 return inode
->i_size
;
712 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
713 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
714 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
716 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
718 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
720 write_seqcount_begin(&inode
->i_size_seqcount
);
721 inode
->i_size
= i_size
;
722 write_seqcount_end(&inode
->i_size_seqcount
);
724 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
726 inode
->i_size
= i_size
;
729 inode
->i_size
= i_size
;
733 /* Helper functions so that in most cases filesystems will
734 * not need to deal directly with kuid_t and kgid_t and can
735 * instead deal with the raw numeric values that are stored
738 static inline uid_t
i_uid_read(const struct inode
*inode
)
740 return from_kuid(&init_user_ns
, inode
->i_uid
);
743 static inline gid_t
i_gid_read(const struct inode
*inode
)
745 return from_kgid(&init_user_ns
, inode
->i_gid
);
748 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
750 inode
->i_uid
= make_kuid(&init_user_ns
, uid
);
753 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
755 inode
->i_gid
= make_kgid(&init_user_ns
, gid
);
758 static inline unsigned iminor(const struct inode
*inode
)
760 return MINOR(inode
->i_rdev
);
763 static inline unsigned imajor(const struct inode
*inode
)
765 return MAJOR(inode
->i_rdev
);
768 extern struct block_device
*I_BDEV(struct inode
*inode
);
771 rwlock_t lock
; /* protects pid, uid, euid fields */
772 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
773 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
774 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
775 int signum
; /* posix.1b rt signal to be delivered on IO */
779 * Track a single file's readahead state
781 struct file_ra_state
{
782 pgoff_t start
; /* where readahead started */
783 unsigned int size
; /* # of readahead pages */
784 unsigned int async_size
; /* do asynchronous readahead when
785 there are only # of pages ahead */
787 unsigned int ra_pages
; /* Maximum readahead window */
788 unsigned int mmap_miss
; /* Cache miss stat for mmap accesses */
789 loff_t prev_pos
; /* Cache last read() position */
793 * Check if @index falls in the readahead windows.
795 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
797 return (index
>= ra
->start
&&
798 index
< ra
->start
+ ra
->size
);
803 struct llist_node fu_llist
;
804 struct rcu_head fu_rcuhead
;
807 struct inode
*f_inode
; /* cached value */
808 const struct file_operations
*f_op
;
811 * Protects f_ep_links, f_flags.
812 * Must not be taken from IRQ context.
815 atomic_long_t f_count
;
816 unsigned int f_flags
;
818 struct mutex f_pos_lock
;
820 struct fown_struct f_owner
;
821 const struct cred
*f_cred
;
822 struct file_ra_state f_ra
;
825 #ifdef CONFIG_SECURITY
828 /* needed for tty driver, and maybe others */
832 /* Used by fs/eventpoll.c to link all the hooks to this file */
833 struct list_head f_ep_links
;
834 struct list_head f_tfile_llink
;
835 #endif /* #ifdef CONFIG_EPOLL */
836 struct address_space
*f_mapping
;
837 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
842 /* file identifier */
843 unsigned char f_handle
[0];
846 static inline struct file
*get_file(struct file
*f
)
848 atomic_long_inc(&f
->f_count
);
851 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
852 #define file_count(x) atomic_long_read(&(x)->f_count)
854 #define MAX_NON_LFS ((1UL<<31) - 1)
856 /* Page cache limit. The filesystems should put that into their s_maxbytes
857 limits, otherwise bad things can happen in VM. */
858 #if BITS_PER_LONG==32
859 #define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
860 #elif BITS_PER_LONG==64
861 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
866 #define FL_DELEG 4 /* NFSv4 delegation */
867 #define FL_ACCESS 8 /* not trying to lock, just looking */
868 #define FL_EXISTS 16 /* when unlocking, test for existence */
869 #define FL_LEASE 32 /* lease held on this file */
870 #define FL_CLOSE 64 /* unlock on close */
871 #define FL_SLEEP 128 /* A blocking lock */
872 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
873 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
874 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
875 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
878 * Special return value from posix_lock_file() and vfs_lock_file() for
879 * asynchronous locking.
881 #define FILE_LOCK_DEFERRED 1
883 /* legacy typedef, should eventually be removed */
884 typedef void *fl_owner_t
;
888 struct file_lock_operations
{
889 void (*fl_copy_lock
)(struct file_lock
*, struct file_lock
*);
890 void (*fl_release_private
)(struct file_lock
*);
893 struct lock_manager_operations
{
894 int (*lm_compare_owner
)(struct file_lock
*, struct file_lock
*);
895 unsigned long (*lm_owner_key
)(struct file_lock
*);
896 void (*lm_get_owner
)(struct file_lock
*, struct file_lock
*);
897 void (*lm_put_owner
)(struct file_lock
*);
898 void (*lm_notify
)(struct file_lock
*); /* unblock callback */
899 int (*lm_grant
)(struct file_lock
*, int);
900 bool (*lm_break
)(struct file_lock
*);
901 int (*lm_change
)(struct file_lock
*, int, struct list_head
*);
902 void (*lm_setup
)(struct file_lock
*, void **);
905 struct lock_manager
{
906 struct list_head list
;
910 void locks_start_grace(struct net
*, struct lock_manager
*);
911 void locks_end_grace(struct lock_manager
*);
912 int locks_in_grace(struct net
*);
914 /* that will die - we need it for nfs_lock_info */
915 #include <linux/nfs_fs_i.h>
918 * struct file_lock represents a generic "file lock". It's used to represent
919 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
920 * note that the same struct is used to represent both a request for a lock and
921 * the lock itself, but the same object is never used for both.
923 * FIXME: should we create a separate "struct lock_request" to help distinguish
926 * The varous i_flctx lists are ordered by:
929 * 2) lock range start
932 * Obviously, the last two criteria only matter for POSIX locks.
935 struct file_lock
*fl_next
; /* singly linked list for this inode */
936 struct list_head fl_list
; /* link into file_lock_context */
937 struct hlist_node fl_link
; /* node in global lists */
938 struct list_head fl_block
; /* circular list of blocked processes */
940 unsigned int fl_flags
;
941 unsigned char fl_type
;
943 int fl_link_cpu
; /* what cpu's list is this on? */
944 struct pid
*fl_nspid
;
945 wait_queue_head_t fl_wait
;
946 struct file
*fl_file
;
950 struct fasync_struct
* fl_fasync
; /* for lease break notifications */
951 /* for lease breaks: */
952 unsigned long fl_break_time
;
953 unsigned long fl_downgrade_time
;
955 const struct file_lock_operations
*fl_ops
; /* Callbacks for filesystems */
956 const struct lock_manager_operations
*fl_lmops
; /* Callbacks for lockmanagers */
958 struct nfs_lock_info nfs_fl
;
959 struct nfs4_lock_info nfs4_fl
;
961 struct list_head link
; /* link in AFS vnode's pending_locks list */
962 int state
; /* state of grant or error if -ve */
967 struct file_lock_context
{
969 struct list_head flc_flock
;
970 struct list_head flc_posix
;
971 struct list_head flc_lease
;
977 /* The following constant reflects the upper bound of the file/locking space */
979 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
980 #define OFFSET_MAX INT_LIMIT(loff_t)
981 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
984 #include <linux/fcntl.h>
986 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
988 #ifdef CONFIG_FILE_LOCKING
989 extern int fcntl_getlk(struct file
*, unsigned int, struct flock __user
*);
990 extern int fcntl_setlk(unsigned int, struct file
*, unsigned int,
991 struct flock __user
*);
993 #if BITS_PER_LONG == 32
994 extern int fcntl_getlk64(struct file
*, unsigned int, struct flock64 __user
*);
995 extern int fcntl_setlk64(unsigned int, struct file
*, unsigned int,
996 struct flock64 __user
*);
999 extern int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
);
1000 extern int fcntl_getlease(struct file
*filp
);
1003 void locks_free_lock_context(struct file_lock_context
*ctx
);
1004 void locks_free_lock(struct file_lock
*fl
);
1005 extern void locks_init_lock(struct file_lock
*);
1006 extern struct file_lock
* locks_alloc_lock(void);
1007 extern void locks_copy_lock(struct file_lock
*, struct file_lock
*);
1008 extern void locks_copy_conflock(struct file_lock
*, struct file_lock
*);
1009 extern void locks_remove_posix(struct file
*, fl_owner_t
);
1010 extern void locks_remove_file(struct file
*);
1011 extern void locks_release_private(struct file_lock
*);
1012 extern void posix_test_lock(struct file
*, struct file_lock
*);
1013 extern int posix_lock_file(struct file
*, struct file_lock
*, struct file_lock
*);
1014 extern int posix_lock_file_wait(struct file
*, struct file_lock
*);
1015 extern int posix_unblock_lock(struct file_lock
*);
1016 extern int vfs_test_lock(struct file
*, struct file_lock
*);
1017 extern int vfs_lock_file(struct file
*, unsigned int, struct file_lock
*, struct file_lock
*);
1018 extern int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
);
1019 extern int flock_lock_file_wait(struct file
*filp
, struct file_lock
*fl
);
1020 extern int __break_lease(struct inode
*inode
, unsigned int flags
, unsigned int type
);
1021 extern void lease_get_mtime(struct inode
*, struct timespec
*time
);
1022 extern int generic_setlease(struct file
*, long, struct file_lock
**, void **priv
);
1023 extern int vfs_setlease(struct file
*, long, struct file_lock
**, void **);
1024 extern int lease_modify(struct file_lock
*, int, struct list_head
*);
1025 #else /* !CONFIG_FILE_LOCKING */
1026 static inline int fcntl_getlk(struct file
*file
, unsigned int cmd
,
1027 struct flock __user
*user
)
1032 static inline int fcntl_setlk(unsigned int fd
, struct file
*file
,
1033 unsigned int cmd
, struct flock __user
*user
)
1038 #if BITS_PER_LONG == 32
1039 static inline int fcntl_getlk64(struct file
*file
, unsigned int cmd
,
1040 struct flock64 __user
*user
)
1045 static inline int fcntl_setlk64(unsigned int fd
, struct file
*file
,
1046 unsigned int cmd
, struct flock64 __user
*user
)
1051 static inline int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
)
1056 static inline int fcntl_getlease(struct file
*filp
)
1062 locks_free_lock_context(struct file_lock_context
*ctx
)
1066 static inline void locks_init_lock(struct file_lock
*fl
)
1071 static inline void locks_copy_conflock(struct file_lock
*new, struct file_lock
*fl
)
1076 static inline void locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
1081 static inline void locks_remove_posix(struct file
*filp
, fl_owner_t owner
)
1086 static inline void locks_remove_file(struct file
*filp
)
1091 static inline void posix_test_lock(struct file
*filp
, struct file_lock
*fl
)
1096 static inline int posix_lock_file(struct file
*filp
, struct file_lock
*fl
,
1097 struct file_lock
*conflock
)
1102 static inline int posix_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1107 static inline int posix_unblock_lock(struct file_lock
*waiter
)
1112 static inline int vfs_test_lock(struct file
*filp
, struct file_lock
*fl
)
1117 static inline int vfs_lock_file(struct file
*filp
, unsigned int cmd
,
1118 struct file_lock
*fl
, struct file_lock
*conf
)
1123 static inline int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
)
1128 static inline int flock_lock_file_wait(struct file
*filp
,
1129 struct file_lock
*request
)
1134 static inline int __break_lease(struct inode
*inode
, unsigned int mode
, unsigned int type
)
1139 static inline void lease_get_mtime(struct inode
*inode
, struct timespec
*time
)
1144 static inline int generic_setlease(struct file
*filp
, long arg
,
1145 struct file_lock
**flp
, void **priv
)
1150 static inline int vfs_setlease(struct file
*filp
, long arg
,
1151 struct file_lock
**lease
, void **priv
)
1156 static inline int lease_modify(struct file_lock
*fl
, int arg
,
1157 struct list_head
*dispose
)
1161 #endif /* !CONFIG_FILE_LOCKING */
1164 struct fasync_struct
{
1168 struct fasync_struct
*fa_next
; /* singly linked list */
1169 struct file
*fa_file
;
1170 struct rcu_head fa_rcu
;
1173 #define FASYNC_MAGIC 0x4601
1175 /* SMP safe fasync helpers: */
1176 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1177 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1178 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1179 extern struct fasync_struct
*fasync_alloc(void);
1180 extern void fasync_free(struct fasync_struct
*);
1182 /* can be called from interrupts */
1183 extern void kill_fasync(struct fasync_struct
**, int, int);
1185 extern void __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1186 extern void f_setown(struct file
*filp
, unsigned long arg
, int force
);
1187 extern void f_delown(struct file
*filp
);
1188 extern pid_t
f_getown(struct file
*filp
);
1189 extern int send_sigurg(struct fown_struct
*fown
);
1197 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1198 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1199 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1200 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1201 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1204 /* Possible states of 'frozen' field */
1206 SB_UNFROZEN
= 0, /* FS is unfrozen */
1207 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1208 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1209 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1210 * internal threads if needed) */
1211 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1214 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1217 /* Counters for counting writers at each level */
1218 struct percpu_counter counter
[SB_FREEZE_LEVELS
];
1219 wait_queue_head_t wait
; /* queue for waiting for
1220 writers / faults to finish */
1221 int frozen
; /* Is sb frozen? */
1222 wait_queue_head_t wait_unfrozen
; /* queue for waiting for
1224 #ifdef CONFIG_DEBUG_LOCK_ALLOC
1225 struct lockdep_map lock_map
[SB_FREEZE_LEVELS
];
1229 struct super_block
{
1230 struct list_head s_list
; /* Keep this first */
1231 dev_t s_dev
; /* search index; _not_ kdev_t */
1232 unsigned char s_blocksize_bits
;
1233 unsigned long s_blocksize
;
1234 loff_t s_maxbytes
; /* Max file size */
1235 struct file_system_type
*s_type
;
1236 const struct super_operations
*s_op
;
1237 const struct dquot_operations
*dq_op
;
1238 const struct quotactl_ops
*s_qcop
;
1239 const struct export_operations
*s_export_op
;
1240 unsigned long s_flags
;
1241 unsigned long s_magic
;
1242 struct dentry
*s_root
;
1243 struct rw_semaphore s_umount
;
1246 #ifdef CONFIG_SECURITY
1249 const struct xattr_handler
**s_xattr
;
1251 struct list_head s_inodes
; /* all inodes */
1252 struct hlist_bl_head s_anon
; /* anonymous dentries for (nfs) exporting */
1253 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1254 struct block_device
*s_bdev
;
1255 struct backing_dev_info
*s_bdi
;
1256 struct mtd_info
*s_mtd
;
1257 struct hlist_node s_instances
;
1258 unsigned int s_quota_types
; /* Bitmask of supported quota types */
1259 struct quota_info s_dquot
; /* Diskquota specific options */
1261 struct sb_writers s_writers
;
1263 char s_id
[32]; /* Informational name */
1264 u8 s_uuid
[16]; /* UUID */
1266 void *s_fs_info
; /* Filesystem private info */
1267 unsigned int s_max_links
;
1270 /* Granularity of c/m/atime in ns.
1271 Cannot be worse than a second */
1275 * The next field is for VFS *only*. No filesystems have any business
1276 * even looking at it. You had been warned.
1278 struct mutex s_vfs_rename_mutex
; /* Kludge */
1281 * Filesystem subtype. If non-empty the filesystem type field
1282 * in /proc/mounts will be "type.subtype"
1287 * Saved mount options for lazy filesystems using
1288 * generic_show_options()
1290 char __rcu
*s_options
;
1291 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1294 * Saved pool identifier for cleancache (-1 means none)
1296 int cleancache_poolid
;
1298 struct shrinker s_shrink
; /* per-sb shrinker handle */
1300 /* Number of inodes with nlink == 0 but still referenced */
1301 atomic_long_t s_remove_count
;
1303 /* Being remounted read-only */
1304 int s_readonly_remount
;
1306 /* AIO completions deferred from interrupt context */
1307 struct workqueue_struct
*s_dio_done_wq
;
1308 struct hlist_head s_pins
;
1311 * Keep the lru lists last in the structure so they always sit on their
1312 * own individual cachelines.
1314 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp
;
1315 struct list_lru s_inode_lru ____cacheline_aligned_in_smp
;
1316 struct rcu_head rcu
;
1319 * Indicates how deep in a filesystem stack this SB is
1324 extern struct timespec
current_fs_time(struct super_block
*sb
);
1327 * Snapshotting support.
1330 void __sb_end_write(struct super_block
*sb
, int level
);
1331 int __sb_start_write(struct super_block
*sb
, int level
, bool wait
);
1334 * sb_end_write - drop write access to a superblock
1335 * @sb: the super we wrote to
1337 * Decrement number of writers to the filesystem. Wake up possible waiters
1338 * wanting to freeze the filesystem.
1340 static inline void sb_end_write(struct super_block
*sb
)
1342 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1346 * sb_end_pagefault - drop write access to a superblock from a page fault
1347 * @sb: the super we wrote to
1349 * Decrement number of processes handling write page fault to the filesystem.
1350 * Wake up possible waiters wanting to freeze the filesystem.
1352 static inline void sb_end_pagefault(struct super_block
*sb
)
1354 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1358 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1359 * @sb: the super we wrote to
1361 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1362 * waiters wanting to freeze the filesystem.
1364 static inline void sb_end_intwrite(struct super_block
*sb
)
1366 __sb_end_write(sb
, SB_FREEZE_FS
);
1370 * sb_start_write - get write access to a superblock
1371 * @sb: the super we write to
1373 * When a process wants to write data or metadata to a file system (i.e. dirty
1374 * a page or an inode), it should embed the operation in a sb_start_write() -
1375 * sb_end_write() pair to get exclusion against file system freezing. This
1376 * function increments number of writers preventing freezing. If the file
1377 * system is already frozen, the function waits until the file system is
1380 * Since freeze protection behaves as a lock, users have to preserve
1381 * ordering of freeze protection and other filesystem locks. Generally,
1382 * freeze protection should be the outermost lock. In particular, we have:
1385 * -> i_mutex (write path, truncate, directory ops, ...)
1386 * -> s_umount (freeze_super, thaw_super)
1388 static inline void sb_start_write(struct super_block
*sb
)
1390 __sb_start_write(sb
, SB_FREEZE_WRITE
, true);
1393 static inline int sb_start_write_trylock(struct super_block
*sb
)
1395 return __sb_start_write(sb
, SB_FREEZE_WRITE
, false);
1399 * sb_start_pagefault - get write access to a superblock from a page fault
1400 * @sb: the super we write to
1402 * When a process starts handling write page fault, it should embed the
1403 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1404 * exclusion against file system freezing. This is needed since the page fault
1405 * is going to dirty a page. This function increments number of running page
1406 * faults preventing freezing. If the file system is already frozen, the
1407 * function waits until the file system is thawed.
1409 * Since page fault freeze protection behaves as a lock, users have to preserve
1410 * ordering of freeze protection and other filesystem locks. It is advised to
1411 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1412 * handling code implies lock dependency:
1415 * -> sb_start_pagefault
1417 static inline void sb_start_pagefault(struct super_block
*sb
)
1419 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
, true);
1423 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1424 * @sb: the super we write to
1426 * This is the third level of protection against filesystem freezing. It is
1427 * free for use by a filesystem. The only requirement is that it must rank
1428 * below sb_start_pagefault.
1430 * For example filesystem can call sb_start_intwrite() when starting a
1431 * transaction which somewhat eases handling of freezing for internal sources
1432 * of filesystem changes (internal fs threads, discarding preallocation on file
1435 static inline void sb_start_intwrite(struct super_block
*sb
)
1437 __sb_start_write(sb
, SB_FREEZE_FS
, true);
1441 extern bool inode_owner_or_capable(const struct inode
*inode
);
1444 * VFS helper functions..
1446 extern int vfs_create(struct inode
*, struct dentry
*, umode_t
, bool);
1447 extern int vfs_mkdir(struct inode
*, struct dentry
*, umode_t
);
1448 extern int vfs_mknod(struct inode
*, struct dentry
*, umode_t
, dev_t
);
1449 extern int vfs_symlink(struct inode
*, struct dentry
*, const char *);
1450 extern int vfs_link(struct dentry
*, struct inode
*, struct dentry
*, struct inode
**);
1451 extern int vfs_rmdir(struct inode
*, struct dentry
*);
1452 extern int vfs_unlink(struct inode
*, struct dentry
*, struct inode
**);
1453 extern int vfs_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*, struct inode
**, unsigned int);
1454 extern int vfs_whiteout(struct inode
*, struct dentry
*);
1457 * VFS dentry helper functions.
1459 extern void dentry_unhash(struct dentry
*dentry
);
1462 * VFS file helper functions.
1464 extern void inode_init_owner(struct inode
*inode
, const struct inode
*dir
,
1467 * VFS FS_IOC_FIEMAP helper definitions.
1469 struct fiemap_extent_info
{
1470 unsigned int fi_flags
; /* Flags as passed from user */
1471 unsigned int fi_extents_mapped
; /* Number of mapped extents */
1472 unsigned int fi_extents_max
; /* Size of fiemap_extent array */
1473 struct fiemap_extent __user
*fi_extents_start
; /* Start of
1474 fiemap_extent array */
1476 int fiemap_fill_next_extent(struct fiemap_extent_info
*info
, u64 logical
,
1477 u64 phys
, u64 len
, u32 flags
);
1478 int fiemap_check_flags(struct fiemap_extent_info
*fieinfo
, u32 fs_flags
);
1483 * NOTE! These match bits 12..15 of stat.st_mode
1484 * (ie "(i_mode >> 12) & 15").
1486 #define DT_UNKNOWN 0
1497 * This is the "filldir" function type, used by readdir() to let
1498 * the kernel specify what kind of dirent layout it wants to have.
1499 * This allows the kernel to read directories into kernel space or
1500 * to have different dirent layouts depending on the binary type.
1503 typedef int (*filldir_t
)(struct dir_context
*, const char *, int, loff_t
, u64
,
1506 struct dir_context
{
1507 const filldir_t actor
;
1511 struct block_device_operations
;
1513 /* These macros are for out of kernel modules to test that
1514 * the kernel supports the unlocked_ioctl and compat_ioctl
1515 * fields in struct file_operations. */
1516 #define HAVE_COMPAT_IOCTL 1
1517 #define HAVE_UNLOCKED_IOCTL 1
1520 * These flags let !MMU mmap() govern direct device mapping vs immediate
1521 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1523 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1524 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1525 * NOMMU_MAP_READ: Can be mapped for reading
1526 * NOMMU_MAP_WRITE: Can be mapped for writing
1527 * NOMMU_MAP_EXEC: Can be mapped for execution
1529 #define NOMMU_MAP_COPY 0x00000001
1530 #define NOMMU_MAP_DIRECT 0x00000008
1531 #define NOMMU_MAP_READ VM_MAYREAD
1532 #define NOMMU_MAP_WRITE VM_MAYWRITE
1533 #define NOMMU_MAP_EXEC VM_MAYEXEC
1535 #define NOMMU_VMFLAGS \
1536 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1541 struct file_operations
{
1542 struct module
*owner
;
1543 loff_t (*llseek
) (struct file
*, loff_t
, int);
1544 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
1545 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
1546 ssize_t (*aio_read
) (struct kiocb
*, const struct iovec
*, unsigned long, loff_t
);
1547 ssize_t (*aio_write
) (struct kiocb
*, const struct iovec
*, unsigned long, loff_t
);
1548 ssize_t (*read_iter
) (struct kiocb
*, struct iov_iter
*);
1549 ssize_t (*write_iter
) (struct kiocb
*, struct iov_iter
*);
1550 int (*iterate
) (struct file
*, struct dir_context
*);
1551 unsigned int (*poll
) (struct file
*, struct poll_table_struct
*);
1552 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
1553 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
1554 int (*mmap
) (struct file
*, struct vm_area_struct
*);
1555 void (*mremap
)(struct file
*, struct vm_area_struct
*);
1556 int (*open
) (struct inode
*, struct file
*);
1557 int (*flush
) (struct file
*, fl_owner_t id
);
1558 int (*release
) (struct inode
*, struct file
*);
1559 int (*fsync
) (struct file
*, loff_t
, loff_t
, int datasync
);
1560 int (*aio_fsync
) (struct kiocb
*, int datasync
);
1561 int (*fasync
) (int, struct file
*, int);
1562 int (*lock
) (struct file
*, int, struct file_lock
*);
1563 ssize_t (*sendpage
) (struct file
*, struct page
*, int, size_t, loff_t
*, int);
1564 unsigned long (*get_unmapped_area
)(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1565 int (*check_flags
)(int);
1566 int (*flock
) (struct file
*, int, struct file_lock
*);
1567 ssize_t (*splice_write
)(struct pipe_inode_info
*, struct file
*, loff_t
*, size_t, unsigned int);
1568 ssize_t (*splice_read
)(struct file
*, loff_t
*, struct pipe_inode_info
*, size_t, unsigned int);
1569 int (*setlease
)(struct file
*, long, struct file_lock
**, void **);
1570 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
1572 void (*show_fdinfo
)(struct seq_file
*m
, struct file
*f
);
1574 unsigned (*mmap_capabilities
)(struct file
*);
1578 struct inode_operations
{
1579 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*, unsigned int);
1580 void * (*follow_link
) (struct dentry
*, struct nameidata
*);
1581 int (*permission
) (struct inode
*, int);
1582 struct posix_acl
* (*get_acl
)(struct inode
*, int);
1584 int (*readlink
) (struct dentry
*, char __user
*,int);
1585 void (*put_link
) (struct dentry
*, struct nameidata
*, void *);
1587 int (*create
) (struct inode
*,struct dentry
*, umode_t
, bool);
1588 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
1589 int (*unlink
) (struct inode
*,struct dentry
*);
1590 int (*symlink
) (struct inode
*,struct dentry
*,const char *);
1591 int (*mkdir
) (struct inode
*,struct dentry
*,umode_t
);
1592 int (*rmdir
) (struct inode
*,struct dentry
*);
1593 int (*mknod
) (struct inode
*,struct dentry
*,umode_t
,dev_t
);
1594 int (*rename
) (struct inode
*, struct dentry
*,
1595 struct inode
*, struct dentry
*);
1596 int (*rename2
) (struct inode
*, struct dentry
*,
1597 struct inode
*, struct dentry
*, unsigned int);
1598 int (*setattr
) (struct dentry
*, struct iattr
*);
1599 int (*getattr
) (struct vfsmount
*mnt
, struct dentry
*, struct kstat
*);
1600 int (*setxattr
) (struct dentry
*, const char *,const void *,size_t,int);
1601 ssize_t (*getxattr
) (struct dentry
*, const char *, void *, size_t);
1602 ssize_t (*listxattr
) (struct dentry
*, char *, size_t);
1603 int (*removexattr
) (struct dentry
*, const char *);
1604 int (*fiemap
)(struct inode
*, struct fiemap_extent_info
*, u64 start
,
1606 int (*update_time
)(struct inode
*, struct timespec
*, int);
1607 int (*atomic_open
)(struct inode
*, struct dentry
*,
1608 struct file
*, unsigned open_flag
,
1609 umode_t create_mode
, int *opened
);
1610 int (*tmpfile
) (struct inode
*, struct dentry
*, umode_t
);
1611 int (*set_acl
)(struct inode
*, struct posix_acl
*, int);
1613 /* WARNING: probably going away soon, do not use! */
1614 int (*dentry_open
)(struct dentry
*, struct file
*, const struct cred
*);
1615 } ____cacheline_aligned
;
1617 ssize_t
rw_copy_check_uvector(int type
, const struct iovec __user
* uvector
,
1618 unsigned long nr_segs
, unsigned long fast_segs
,
1619 struct iovec
*fast_pointer
,
1620 struct iovec
**ret_pointer
);
1622 extern ssize_t
__vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1623 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1624 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1625 extern ssize_t
vfs_readv(struct file
*, const struct iovec __user
*,
1626 unsigned long, loff_t
*);
1627 extern ssize_t
vfs_writev(struct file
*, const struct iovec __user
*,
1628 unsigned long, loff_t
*);
1630 struct super_operations
{
1631 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
1632 void (*destroy_inode
)(struct inode
*);
1634 void (*dirty_inode
) (struct inode
*, int flags
);
1635 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
1636 int (*drop_inode
) (struct inode
*);
1637 void (*evict_inode
) (struct inode
*);
1638 void (*put_super
) (struct super_block
*);
1639 int (*sync_fs
)(struct super_block
*sb
, int wait
);
1640 int (*freeze_super
) (struct super_block
*);
1641 int (*freeze_fs
) (struct super_block
*);
1642 int (*thaw_super
) (struct super_block
*);
1643 int (*unfreeze_fs
) (struct super_block
*);
1644 int (*statfs
) (struct dentry
*, struct kstatfs
*);
1645 int (*remount_fs
) (struct super_block
*, int *, char *);
1646 void (*umount_begin
) (struct super_block
*);
1648 int (*show_options
)(struct seq_file
*, struct dentry
*);
1649 int (*show_devname
)(struct seq_file
*, struct dentry
*);
1650 int (*show_path
)(struct seq_file
*, struct dentry
*);
1651 int (*show_stats
)(struct seq_file
*, struct dentry
*);
1653 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
1654 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
1655 struct dquot
**(*get_dquots
)(struct inode
*);
1657 int (*bdev_try_to_free_page
)(struct super_block
*, struct page
*, gfp_t
);
1658 long (*nr_cached_objects
)(struct super_block
*,
1659 struct shrink_control
*);
1660 long (*free_cached_objects
)(struct super_block
*,
1661 struct shrink_control
*);
1665 * Inode flags - they have no relation to superblock flags now
1667 #define S_SYNC 1 /* Writes are synced at once */
1668 #define S_NOATIME 2 /* Do not update access times */
1669 #define S_APPEND 4 /* Append-only file */
1670 #define S_IMMUTABLE 8 /* Immutable file */
1671 #define S_DEAD 16 /* removed, but still open directory */
1672 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1673 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1674 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1675 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1676 #define S_PRIVATE 512 /* Inode is fs-internal */
1677 #define S_IMA 1024 /* Inode has an associated IMA struct */
1678 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1679 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1682 * Note that nosuid etc flags are inode-specific: setting some file-system
1683 * flags just means all the inodes inherit those flags by default. It might be
1684 * possible to override it selectively if you really wanted to with some
1685 * ioctl() that is not currently implemented.
1687 * Exception: MS_RDONLY is always applied to the entire file system.
1689 * Unfortunately, it is possible to change a filesystems flags with it mounted
1690 * with files in use. This means that all of the inodes will not have their
1691 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1692 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1694 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1696 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1697 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1698 ((inode)->i_flags & S_SYNC))
1699 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1700 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1701 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1702 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1703 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1705 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1706 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1707 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1708 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1710 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1711 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1712 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1713 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1714 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1715 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1716 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1718 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1719 (inode)->i_rdev == WHITEOUT_DEV)
1722 * Inode state bits. Protected by inode->i_lock
1724 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1725 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1727 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1728 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1729 * various stages of removing an inode.
1731 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1733 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1734 * fdatasync(). i_atime is the usual cause.
1735 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1736 * these changes separately from I_DIRTY_SYNC so that we
1737 * don't have to write inode on fdatasync() when only
1738 * mtime has changed in it.
1739 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1740 * I_NEW Serves as both a mutex and completion notification.
1741 * New inodes set I_NEW. If two processes both create
1742 * the same inode, one of them will release its inode and
1743 * wait for I_NEW to be released before returning.
1744 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1745 * also cause waiting on I_NEW, without I_NEW actually
1746 * being set. find_inode() uses this to prevent returning
1747 * nearly-dead inodes.
1748 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1749 * is zero. I_FREEING must be set when I_WILL_FREE is
1751 * I_FREEING Set when inode is about to be freed but still has dirty
1752 * pages or buffers attached or the inode itself is still
1754 * I_CLEAR Added by clear_inode(). In this state the inode is
1755 * clean and can be destroyed. Inode keeps I_FREEING.
1757 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1758 * prohibited for many purposes. iget() must wait for
1759 * the inode to be completely released, then create it
1760 * anew. Other functions will just ignore such inodes,
1761 * if appropriate. I_NEW is used for waiting.
1763 * I_SYNC Writeback of inode is running. The bit is set during
1764 * data writeback, and cleared with a wakeup on the bit
1765 * address once it is done. The bit is also used to pin
1766 * the inode in memory for flusher thread.
1768 * I_REFERENCED Marks the inode as recently references on the LRU list.
1770 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1772 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1774 #define I_DIRTY_SYNC (1 << 0)
1775 #define I_DIRTY_DATASYNC (1 << 1)
1776 #define I_DIRTY_PAGES (1 << 2)
1778 #define I_NEW (1 << __I_NEW)
1779 #define I_WILL_FREE (1 << 4)
1780 #define I_FREEING (1 << 5)
1781 #define I_CLEAR (1 << 6)
1783 #define I_SYNC (1 << __I_SYNC)
1784 #define I_REFERENCED (1 << 8)
1785 #define __I_DIO_WAKEUP 9
1786 #define I_DIO_WAKEUP (1 << I_DIO_WAKEUP)
1787 #define I_LINKABLE (1 << 10)
1789 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1791 extern void __mark_inode_dirty(struct inode
*, int);
1792 static inline void mark_inode_dirty(struct inode
*inode
)
1794 __mark_inode_dirty(inode
, I_DIRTY
);
1797 static inline void mark_inode_dirty_sync(struct inode
*inode
)
1799 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
1802 extern void inc_nlink(struct inode
*inode
);
1803 extern void drop_nlink(struct inode
*inode
);
1804 extern void clear_nlink(struct inode
*inode
);
1805 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
1807 static inline void inode_inc_link_count(struct inode
*inode
)
1810 mark_inode_dirty(inode
);
1813 static inline void inode_dec_link_count(struct inode
*inode
)
1816 mark_inode_dirty(inode
);
1820 * inode_inc_iversion - increments i_version
1821 * @inode: inode that need to be updated
1823 * Every time the inode is modified, the i_version field will be incremented.
1824 * The filesystem has to be mounted with i_version flag
1827 static inline void inode_inc_iversion(struct inode
*inode
)
1829 spin_lock(&inode
->i_lock
);
1831 spin_unlock(&inode
->i_lock
);
1834 enum file_time_flags
{
1841 extern void touch_atime(const struct path
*);
1842 static inline void file_accessed(struct file
*file
)
1844 if (!(file
->f_flags
& O_NOATIME
))
1845 touch_atime(&file
->f_path
);
1848 int sync_inode(struct inode
*inode
, struct writeback_control
*wbc
);
1849 int sync_inode_metadata(struct inode
*inode
, int wait
);
1851 struct file_system_type
{
1854 #define FS_REQUIRES_DEV 1
1855 #define FS_BINARY_MOUNTDATA 2
1856 #define FS_HAS_SUBTYPE 4
1857 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
1858 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
1859 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
1860 struct dentry
*(*mount
) (struct file_system_type
*, int,
1861 const char *, void *);
1862 void (*kill_sb
) (struct super_block
*);
1863 struct module
*owner
;
1864 struct file_system_type
* next
;
1865 struct hlist_head fs_supers
;
1867 struct lock_class_key s_lock_key
;
1868 struct lock_class_key s_umount_key
;
1869 struct lock_class_key s_vfs_rename_key
;
1870 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
1872 struct lock_class_key i_lock_key
;
1873 struct lock_class_key i_mutex_key
;
1874 struct lock_class_key i_mutex_dir_key
;
1877 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1879 extern struct dentry
*mount_ns(struct file_system_type
*fs_type
, int flags
,
1880 void *data
, int (*fill_super
)(struct super_block
*, void *, int));
1881 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
1882 int flags
, const char *dev_name
, void *data
,
1883 int (*fill_super
)(struct super_block
*, void *, int));
1884 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
1885 int flags
, void *data
,
1886 int (*fill_super
)(struct super_block
*, void *, int));
1887 extern struct dentry
*mount_nodev(struct file_system_type
*fs_type
,
1888 int flags
, void *data
,
1889 int (*fill_super
)(struct super_block
*, void *, int));
1890 extern struct dentry
*mount_subtree(struct vfsmount
*mnt
, const char *path
);
1891 void generic_shutdown_super(struct super_block
*sb
);
1892 void kill_block_super(struct super_block
*sb
);
1893 void kill_anon_super(struct super_block
*sb
);
1894 void kill_litter_super(struct super_block
*sb
);
1895 void deactivate_super(struct super_block
*sb
);
1896 void deactivate_locked_super(struct super_block
*sb
);
1897 int set_anon_super(struct super_block
*s
, void *data
);
1898 int get_anon_bdev(dev_t
*);
1899 void free_anon_bdev(dev_t
);
1900 struct super_block
*sget(struct file_system_type
*type
,
1901 int (*test
)(struct super_block
*,void *),
1902 int (*set
)(struct super_block
*,void *),
1903 int flags
, void *data
);
1904 extern struct dentry
*mount_pseudo(struct file_system_type
*, char *,
1905 const struct super_operations
*ops
,
1906 const struct dentry_operations
*dops
,
1909 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1910 #define fops_get(fops) \
1911 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1912 #define fops_put(fops) \
1913 do { if (fops) module_put((fops)->owner); } while(0)
1915 * This one is to be used *ONLY* from ->open() instances.
1916 * fops must be non-NULL, pinned down *and* module dependencies
1917 * should be sufficient to pin the caller down as well.
1919 #define replace_fops(f, fops) \
1921 struct file *__file = (f); \
1922 fops_put(__file->f_op); \
1923 BUG_ON(!(__file->f_op = (fops))); \
1926 extern int register_filesystem(struct file_system_type
*);
1927 extern int unregister_filesystem(struct file_system_type
*);
1928 extern struct vfsmount
*kern_mount_data(struct file_system_type
*, void *data
);
1929 #define kern_mount(type) kern_mount_data(type, NULL)
1930 extern void kern_unmount(struct vfsmount
*mnt
);
1931 extern int may_umount_tree(struct vfsmount
*);
1932 extern int may_umount(struct vfsmount
*);
1933 extern long do_mount(const char *, const char __user
*,
1934 const char *, unsigned long, void *);
1935 extern struct vfsmount
*collect_mounts(struct path
*);
1936 extern void drop_collected_mounts(struct vfsmount
*);
1937 extern int iterate_mounts(int (*)(struct vfsmount
*, void *), void *,
1939 extern int vfs_statfs(struct path
*, struct kstatfs
*);
1940 extern int user_statfs(const char __user
*, struct kstatfs
*);
1941 extern int fd_statfs(int, struct kstatfs
*);
1942 extern int vfs_ustat(dev_t
, struct kstatfs
*);
1943 extern int freeze_super(struct super_block
*super
);
1944 extern int thaw_super(struct super_block
*super
);
1945 extern bool our_mnt(struct vfsmount
*mnt
);
1946 extern bool fs_fully_visible(struct file_system_type
*);
1948 extern int current_umask(void);
1950 extern void ihold(struct inode
* inode
);
1951 extern void iput(struct inode
*);
1953 static inline struct inode
*file_inode(const struct file
*f
)
1959 extern struct kobject
*fs_kobj
;
1961 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1963 #define FLOCK_VERIFY_READ 1
1964 #define FLOCK_VERIFY_WRITE 2
1966 #ifdef CONFIG_FILE_LOCKING
1967 extern int locks_mandatory_locked(struct file
*);
1968 extern int locks_mandatory_area(int, struct inode
*, struct file
*, loff_t
, size_t);
1971 * Candidates for mandatory locking have the setgid bit set
1972 * but no group execute bit - an otherwise meaningless combination.
1975 static inline int __mandatory_lock(struct inode
*ino
)
1977 return (ino
->i_mode
& (S_ISGID
| S_IXGRP
)) == S_ISGID
;
1981 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1982 * otherwise these will be advisory locks
1985 static inline int mandatory_lock(struct inode
*ino
)
1987 return IS_MANDLOCK(ino
) && __mandatory_lock(ino
);
1990 static inline int locks_verify_locked(struct file
*file
)
1992 if (mandatory_lock(file_inode(file
)))
1993 return locks_mandatory_locked(file
);
1997 static inline int locks_verify_truncate(struct inode
*inode
,
2001 if (inode
->i_flctx
&& mandatory_lock(inode
))
2002 return locks_mandatory_area(
2003 FLOCK_VERIFY_WRITE
, inode
, filp
,
2004 size
< inode
->i_size
? size
: inode
->i_size
,
2005 (size
< inode
->i_size
? inode
->i_size
- size
2006 : size
- inode
->i_size
)
2011 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2014 * Since this check is lockless, we must ensure that any refcounts
2015 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2016 * could end up racing with tasks trying to set a new lease on this
2020 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2021 return __break_lease(inode
, mode
, FL_LEASE
);
2025 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2028 * Since this check is lockless, we must ensure that any refcounts
2029 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2030 * could end up racing with tasks trying to set a new lease on this
2034 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2035 return __break_lease(inode
, mode
, FL_DELEG
);
2039 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2043 ret
= break_deleg(inode
, O_WRONLY
|O_NONBLOCK
);
2044 if (ret
== -EWOULDBLOCK
&& delegated_inode
) {
2045 *delegated_inode
= inode
;
2051 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2055 ret
= break_deleg(*delegated_inode
, O_WRONLY
);
2056 iput(*delegated_inode
);
2057 *delegated_inode
= NULL
;
2061 static inline int break_layout(struct inode
*inode
, bool wait
)
2064 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2065 return __break_lease(inode
,
2066 wait
? O_WRONLY
: O_WRONLY
| O_NONBLOCK
,
2071 #else /* !CONFIG_FILE_LOCKING */
2072 static inline int locks_mandatory_locked(struct file
*file
)
2077 static inline int locks_mandatory_area(int rw
, struct inode
*inode
,
2078 struct file
*filp
, loff_t offset
,
2084 static inline int __mandatory_lock(struct inode
*inode
)
2089 static inline int mandatory_lock(struct inode
*inode
)
2094 static inline int locks_verify_locked(struct file
*file
)
2099 static inline int locks_verify_truncate(struct inode
*inode
, struct file
*filp
,
2105 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2110 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2115 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2120 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2126 static inline int break_layout(struct inode
*inode
, bool wait
)
2131 #endif /* CONFIG_FILE_LOCKING */
2136 const char *name
; /* pointer to actual string */
2137 const __user
char *uptr
; /* original userland pointer */
2138 struct audit_names
*aname
;
2139 bool separate
; /* should "name" be freed? */
2142 extern long vfs_truncate(struct path
*, loff_t
);
2143 extern int do_truncate(struct dentry
*, loff_t start
, unsigned int time_attrs
,
2145 extern int vfs_fallocate(struct file
*file
, int mode
, loff_t offset
,
2147 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
2149 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
2150 extern struct file
*filp_open(const char *, int, umode_t
);
2151 extern struct file
*file_open_root(struct dentry
*, struct vfsmount
*,
2153 extern int vfs_open(const struct path
*, struct file
*, const struct cred
*);
2154 extern struct file
* dentry_open(const struct path
*, int, const struct cred
*);
2155 extern int filp_close(struct file
*, fl_owner_t id
);
2157 extern struct filename
*getname_flags(const char __user
*, int, int *);
2158 extern struct filename
*getname(const char __user
*);
2159 extern struct filename
*getname_kernel(const char *);
2165 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
2166 int (*open
)(struct inode
*, struct file
*),
2168 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
2172 extern int ioctl_preallocate(struct file
*filp
, void __user
*argp
);
2175 extern void __init
vfs_caches_init_early(void);
2176 extern void __init
vfs_caches_init(unsigned long);
2178 extern struct kmem_cache
*names_cachep
;
2180 extern void final_putname(struct filename
*name
);
2182 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2183 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2184 #ifndef CONFIG_AUDITSYSCALL
2185 #define putname(name) final_putname(name)
2187 extern void putname(struct filename
*name
);
2191 extern int register_blkdev(unsigned int, const char *);
2192 extern void unregister_blkdev(unsigned int, const char *);
2193 extern struct block_device
*bdget(dev_t
);
2194 extern struct block_device
*bdgrab(struct block_device
*bdev
);
2195 extern void bd_set_size(struct block_device
*, loff_t size
);
2196 extern void bd_forget(struct inode
*inode
);
2197 extern void bdput(struct block_device
*);
2198 extern void invalidate_bdev(struct block_device
*);
2199 extern void iterate_bdevs(void (*)(struct block_device
*, void *), void *);
2200 extern int sync_blockdev(struct block_device
*bdev
);
2201 extern void kill_bdev(struct block_device
*);
2202 extern struct super_block
*freeze_bdev(struct block_device
*);
2203 extern void emergency_thaw_all(void);
2204 extern int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
);
2205 extern int fsync_bdev(struct block_device
*);
2206 extern int sb_is_blkdev_sb(struct super_block
*sb
);
2208 static inline void bd_forget(struct inode
*inode
) {}
2209 static inline int sync_blockdev(struct block_device
*bdev
) { return 0; }
2210 static inline void kill_bdev(struct block_device
*bdev
) {}
2211 static inline void invalidate_bdev(struct block_device
*bdev
) {}
2213 static inline struct super_block
*freeze_bdev(struct block_device
*sb
)
2218 static inline int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
)
2223 static inline void iterate_bdevs(void (*f
)(struct block_device
*, void *), void *arg
)
2227 static inline int sb_is_blkdev_sb(struct super_block
*sb
)
2232 extern int sync_filesystem(struct super_block
*);
2233 extern const struct file_operations def_blk_fops
;
2234 extern const struct file_operations def_chr_fops
;
2236 extern int ioctl_by_bdev(struct block_device
*, unsigned, unsigned long);
2237 extern int blkdev_ioctl(struct block_device
*, fmode_t
, unsigned, unsigned long);
2238 extern long compat_blkdev_ioctl(struct file
*, unsigned, unsigned long);
2239 extern int blkdev_get(struct block_device
*bdev
, fmode_t mode
, void *holder
);
2240 extern struct block_device
*blkdev_get_by_path(const char *path
, fmode_t mode
,
2242 extern struct block_device
*blkdev_get_by_dev(dev_t dev
, fmode_t mode
,
2244 extern void blkdev_put(struct block_device
*bdev
, fmode_t mode
);
2246 extern int bd_link_disk_holder(struct block_device
*bdev
, struct gendisk
*disk
);
2247 extern void bd_unlink_disk_holder(struct block_device
*bdev
,
2248 struct gendisk
*disk
);
2250 static inline int bd_link_disk_holder(struct block_device
*bdev
,
2251 struct gendisk
*disk
)
2255 static inline void bd_unlink_disk_holder(struct block_device
*bdev
,
2256 struct gendisk
*disk
)
2263 #define CHRDEV_MAJOR_HASH_SIZE 255
2264 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
2265 extern int register_chrdev_region(dev_t
, unsigned, const char *);
2266 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
2267 unsigned int count
, const char *name
,
2268 const struct file_operations
*fops
);
2269 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
2270 unsigned int count
, const char *name
);
2271 extern void unregister_chrdev_region(dev_t
, unsigned);
2272 extern void chrdev_show(struct seq_file
*,off_t
);
2274 static inline int register_chrdev(unsigned int major
, const char *name
,
2275 const struct file_operations
*fops
)
2277 return __register_chrdev(major
, 0, 256, name
, fops
);
2280 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2282 __unregister_chrdev(major
, 0, 256, name
);
2285 /* fs/block_dev.c */
2286 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2287 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2290 #define BLKDEV_MAJOR_HASH_SIZE 255
2291 extern const char *__bdevname(dev_t
, char *buffer
);
2292 extern const char *bdevname(struct block_device
*bdev
, char *buffer
);
2293 extern struct block_device
*lookup_bdev(const char *);
2294 extern void blkdev_show(struct seq_file
*,off_t
);
2297 #define BLKDEV_MAJOR_HASH_SIZE 0
2300 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2302 /* Invalid inode operations -- fs/bad_inode.c */
2303 extern void make_bad_inode(struct inode
*);
2304 extern int is_bad_inode(struct inode
*);
2308 * return READ, READA, or WRITE
2310 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2313 * return data direction, READ or WRITE
2315 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2317 extern void check_disk_size_change(struct gendisk
*disk
,
2318 struct block_device
*bdev
);
2319 extern int revalidate_disk(struct gendisk
*);
2320 extern int check_disk_change(struct block_device
*);
2321 extern int __invalidate_device(struct block_device
*, bool);
2322 extern int invalidate_partition(struct gendisk
*, int);
2324 unsigned long invalidate_mapping_pages(struct address_space
*mapping
,
2325 pgoff_t start
, pgoff_t end
);
2327 static inline void invalidate_remote_inode(struct inode
*inode
)
2329 if (S_ISREG(inode
->i_mode
) || S_ISDIR(inode
->i_mode
) ||
2330 S_ISLNK(inode
->i_mode
))
2331 invalidate_mapping_pages(inode
->i_mapping
, 0, -1);
2333 extern int invalidate_inode_pages2(struct address_space
*mapping
);
2334 extern int invalidate_inode_pages2_range(struct address_space
*mapping
,
2335 pgoff_t start
, pgoff_t end
);
2336 extern int write_inode_now(struct inode
*, int);
2337 extern int filemap_fdatawrite(struct address_space
*);
2338 extern int filemap_flush(struct address_space
*);
2339 extern int filemap_fdatawait(struct address_space
*);
2340 extern int filemap_fdatawait_range(struct address_space
*, loff_t lstart
,
2342 extern int filemap_write_and_wait(struct address_space
*mapping
);
2343 extern int filemap_write_and_wait_range(struct address_space
*mapping
,
2344 loff_t lstart
, loff_t lend
);
2345 extern int __filemap_fdatawrite_range(struct address_space
*mapping
,
2346 loff_t start
, loff_t end
, int sync_mode
);
2347 extern int filemap_fdatawrite_range(struct address_space
*mapping
,
2348 loff_t start
, loff_t end
);
2350 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
2352 extern int vfs_fsync(struct file
*file
, int datasync
);
2353 static inline int generic_write_sync(struct file
*file
, loff_t pos
, loff_t count
)
2355 if (!(file
->f_flags
& O_DSYNC
) && !IS_SYNC(file
->f_mapping
->host
))
2357 return vfs_fsync_range(file
, pos
, pos
+ count
- 1,
2358 (file
->f_flags
& __O_SYNC
) ? 0 : 1);
2360 extern void emergency_sync(void);
2361 extern void emergency_remount(void);
2363 extern sector_t
bmap(struct inode
*, sector_t
);
2365 extern int notify_change(struct dentry
*, struct iattr
*, struct inode
**);
2366 extern int inode_permission(struct inode
*, int);
2367 extern int __inode_permission(struct inode
*, int);
2368 extern int generic_permission(struct inode
*, int);
2369 extern int __check_sticky(struct inode
*dir
, struct inode
*inode
);
2371 static inline bool execute_ok(struct inode
*inode
)
2373 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
2376 static inline void file_start_write(struct file
*file
)
2378 if (!S_ISREG(file_inode(file
)->i_mode
))
2380 __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, true);
2383 static inline bool file_start_write_trylock(struct file
*file
)
2385 if (!S_ISREG(file_inode(file
)->i_mode
))
2387 return __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, false);
2390 static inline void file_end_write(struct file
*file
)
2392 if (!S_ISREG(file_inode(file
)->i_mode
))
2394 __sb_end_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
);
2398 * get_write_access() gets write permission for a file.
2399 * put_write_access() releases this write permission.
2400 * This is used for regular files.
2401 * We cannot support write (and maybe mmap read-write shared) accesses and
2402 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2403 * can have the following values:
2404 * 0: no writers, no VM_DENYWRITE mappings
2405 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2406 * > 0: (i_writecount) users are writing to the file.
2408 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2409 * except for the cases where we don't hold i_writecount yet. Then we need to
2410 * use {get,deny}_write_access() - these functions check the sign and refuse
2411 * to do the change if sign is wrong.
2413 static inline int get_write_access(struct inode
*inode
)
2415 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2417 static inline int deny_write_access(struct file
*file
)
2419 struct inode
*inode
= file_inode(file
);
2420 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2422 static inline void put_write_access(struct inode
* inode
)
2424 atomic_dec(&inode
->i_writecount
);
2426 static inline void allow_write_access(struct file
*file
)
2429 atomic_inc(&file_inode(file
)->i_writecount
);
2431 static inline bool inode_is_open_for_write(const struct inode
*inode
)
2433 return atomic_read(&inode
->i_writecount
) > 0;
2437 static inline void i_readcount_dec(struct inode
*inode
)
2439 BUG_ON(!atomic_read(&inode
->i_readcount
));
2440 atomic_dec(&inode
->i_readcount
);
2442 static inline void i_readcount_inc(struct inode
*inode
)
2444 atomic_inc(&inode
->i_readcount
);
2447 static inline void i_readcount_dec(struct inode
*inode
)
2451 static inline void i_readcount_inc(struct inode
*inode
)
2456 extern int do_pipe_flags(int *, int);
2458 extern int kernel_read(struct file
*, loff_t
, char *, unsigned long);
2459 extern ssize_t
kernel_write(struct file
*, const char *, size_t, loff_t
);
2460 extern ssize_t
__kernel_write(struct file
*, const char *, size_t, loff_t
*);
2461 extern struct file
* open_exec(const char *);
2463 /* fs/dcache.c -- generic fs support functions */
2464 extern int is_subdir(struct dentry
*, struct dentry
*);
2465 extern int path_is_under(struct path
*, struct path
*);
2467 #include <linux/err.h>
2469 /* needed for stackable file system support */
2470 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int whence
);
2472 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int whence
);
2474 extern int inode_init_always(struct super_block
*, struct inode
*);
2475 extern void inode_init_once(struct inode
*);
2476 extern void address_space_init_once(struct address_space
*mapping
);
2477 extern struct inode
* igrab(struct inode
*);
2478 extern ino_t
iunique(struct super_block
*, ino_t
);
2479 extern int inode_needs_sync(struct inode
*inode
);
2480 extern int generic_delete_inode(struct inode
*inode
);
2481 static inline int generic_drop_inode(struct inode
*inode
)
2483 return !inode
->i_nlink
|| inode_unhashed(inode
);
2486 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
2487 unsigned long hashval
, int (*test
)(struct inode
*, void *),
2489 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
2490 int (*test
)(struct inode
*, void *), void *data
);
2491 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
2493 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
2494 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
2495 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
2496 extern int insert_inode_locked(struct inode
*);
2497 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2498 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
2500 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
2502 extern void unlock_new_inode(struct inode
*);
2503 extern unsigned int get_next_ino(void);
2505 extern void __iget(struct inode
* inode
);
2506 extern void iget_failed(struct inode
*);
2507 extern void clear_inode(struct inode
*);
2508 extern void __destroy_inode(struct inode
*);
2509 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
2510 extern struct inode
*new_inode(struct super_block
*sb
);
2511 extern void free_inode_nonrcu(struct inode
*inode
);
2512 extern int should_remove_suid(struct dentry
*);
2513 extern int file_remove_suid(struct file
*);
2515 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
2516 static inline void insert_inode_hash(struct inode
*inode
)
2518 __insert_inode_hash(inode
, inode
->i_ino
);
2521 extern void __remove_inode_hash(struct inode
*);
2522 static inline void remove_inode_hash(struct inode
*inode
)
2524 if (!inode_unhashed(inode
))
2525 __remove_inode_hash(inode
);
2528 extern void inode_sb_list_add(struct inode
*inode
);
2531 extern void submit_bio(int, struct bio
*);
2532 extern int bdev_read_only(struct block_device
*);
2534 extern int set_blocksize(struct block_device
*, int);
2535 extern int sb_set_blocksize(struct super_block
*, int);
2536 extern int sb_min_blocksize(struct super_block
*, int);
2538 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
2539 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
2540 int generic_write_checks(struct file
*file
, loff_t
*pos
, size_t *count
, int isblk
);
2541 extern ssize_t
generic_file_read_iter(struct kiocb
*, struct iov_iter
*);
2542 extern ssize_t
__generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2543 extern ssize_t
generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2544 extern ssize_t
generic_file_direct_write(struct kiocb
*, struct iov_iter
*, loff_t
);
2545 extern ssize_t
generic_perform_write(struct file
*, struct iov_iter
*, loff_t
);
2546 extern ssize_t
do_sync_read(struct file
*filp
, char __user
*buf
, size_t len
, loff_t
*ppos
);
2547 extern ssize_t
do_sync_write(struct file
*filp
, const char __user
*buf
, size_t len
, loff_t
*ppos
);
2548 extern ssize_t
new_sync_read(struct file
*filp
, char __user
*buf
, size_t len
, loff_t
*ppos
);
2549 extern ssize_t
new_sync_write(struct file
*filp
, const char __user
*buf
, size_t len
, loff_t
*ppos
);
2551 /* fs/block_dev.c */
2552 extern ssize_t
blkdev_read_iter(struct kiocb
*iocb
, struct iov_iter
*to
);
2553 extern ssize_t
blkdev_write_iter(struct kiocb
*iocb
, struct iov_iter
*from
);
2554 extern int blkdev_fsync(struct file
*filp
, loff_t start
, loff_t end
,
2556 extern void block_sync_page(struct page
*page
);
2559 extern ssize_t
generic_file_splice_read(struct file
*, loff_t
*,
2560 struct pipe_inode_info
*, size_t, unsigned int);
2561 extern ssize_t
default_file_splice_read(struct file
*, loff_t
*,
2562 struct pipe_inode_info
*, size_t, unsigned int);
2563 extern ssize_t
iter_file_splice_write(struct pipe_inode_info
*,
2564 struct file
*, loff_t
*, size_t, unsigned int);
2565 extern ssize_t
generic_splice_sendpage(struct pipe_inode_info
*pipe
,
2566 struct file
*out
, loff_t
*, size_t len
, unsigned int flags
);
2567 extern long do_splice_direct(struct file
*in
, loff_t
*ppos
, struct file
*out
,
2568 loff_t
*opos
, size_t len
, unsigned int flags
);
2572 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
2573 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int whence
);
2574 extern loff_t
no_llseek(struct file
*file
, loff_t offset
, int whence
);
2575 extern loff_t
vfs_setpos(struct file
*file
, loff_t offset
, loff_t maxsize
);
2576 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int whence
);
2577 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
2578 int whence
, loff_t maxsize
, loff_t eof
);
2579 extern loff_t
fixed_size_llseek(struct file
*file
, loff_t offset
,
2580 int whence
, loff_t size
);
2581 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
2582 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
2584 #ifdef CONFIG_FS_XIP
2585 extern ssize_t
xip_file_read(struct file
*filp
, char __user
*buf
, size_t len
,
2587 extern int xip_file_mmap(struct file
* file
, struct vm_area_struct
* vma
);
2588 extern ssize_t
xip_file_write(struct file
*filp
, const char __user
*buf
,
2589 size_t len
, loff_t
*ppos
);
2590 extern int xip_truncate_page(struct address_space
*mapping
, loff_t from
);
2592 static inline int xip_truncate_page(struct address_space
*mapping
, loff_t from
)
2599 typedef void (dio_submit_t
)(int rw
, struct bio
*bio
, struct inode
*inode
,
2600 loff_t file_offset
);
2603 /* need locking between buffered and direct access */
2606 /* filesystem does not support filling holes */
2607 DIO_SKIP_HOLES
= 0x02,
2609 /* filesystem can handle aio writes beyond i_size */
2610 DIO_ASYNC_EXTEND
= 0x04,
2613 void dio_end_io(struct bio
*bio
, int error
);
2615 ssize_t
__blockdev_direct_IO(int rw
, struct kiocb
*iocb
, struct inode
*inode
,
2616 struct block_device
*bdev
, struct iov_iter
*iter
, loff_t offset
,
2617 get_block_t get_block
, dio_iodone_t end_io
,
2618 dio_submit_t submit_io
, int flags
);
2620 static inline ssize_t
blockdev_direct_IO(int rw
, struct kiocb
*iocb
,
2621 struct inode
*inode
, struct iov_iter
*iter
, loff_t offset
,
2622 get_block_t get_block
)
2624 return __blockdev_direct_IO(rw
, iocb
, inode
, inode
->i_sb
->s_bdev
, iter
,
2625 offset
, get_block
, NULL
, NULL
,
2626 DIO_LOCKING
| DIO_SKIP_HOLES
);
2630 void inode_dio_wait(struct inode
*inode
);
2631 void inode_dio_done(struct inode
*inode
);
2633 extern void inode_set_flags(struct inode
*inode
, unsigned int flags
,
2636 extern const struct file_operations generic_ro_fops
;
2638 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2640 extern int readlink_copy(char __user
*, int, const char *);
2641 extern int page_readlink(struct dentry
*, char __user
*, int);
2642 extern void *page_follow_link_light(struct dentry
*, struct nameidata
*);
2643 extern void page_put_link(struct dentry
*, struct nameidata
*, void *);
2644 extern int __page_symlink(struct inode
*inode
, const char *symname
, int len
,
2646 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
2647 extern const struct inode_operations page_symlink_inode_operations
;
2648 extern void kfree_put_link(struct dentry
*, struct nameidata
*, void *);
2649 extern int generic_readlink(struct dentry
*, char __user
*, int);
2650 extern void generic_fillattr(struct inode
*, struct kstat
*);
2651 int vfs_getattr_nosec(struct path
*path
, struct kstat
*stat
);
2652 extern int vfs_getattr(struct path
*, struct kstat
*);
2653 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2654 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2655 void __inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2656 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2657 loff_t
inode_get_bytes(struct inode
*inode
);
2658 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
2660 extern int vfs_readdir(struct file
*, filldir_t
, void *);
2661 extern int iterate_dir(struct file
*, struct dir_context
*);
2663 extern int vfs_stat(const char __user
*, struct kstat
*);
2664 extern int vfs_lstat(const char __user
*, struct kstat
*);
2665 extern int vfs_fstat(unsigned int, struct kstat
*);
2666 extern int vfs_fstatat(int , const char __user
*, struct kstat
*, int);
2668 extern int do_vfs_ioctl(struct file
*filp
, unsigned int fd
, unsigned int cmd
,
2670 extern int __generic_block_fiemap(struct inode
*inode
,
2671 struct fiemap_extent_info
*fieinfo
,
2672 loff_t start
, loff_t len
,
2673 get_block_t
*get_block
);
2674 extern int generic_block_fiemap(struct inode
*inode
,
2675 struct fiemap_extent_info
*fieinfo
, u64 start
,
2676 u64 len
, get_block_t
*get_block
);
2678 extern void get_filesystem(struct file_system_type
*fs
);
2679 extern void put_filesystem(struct file_system_type
*fs
);
2680 extern struct file_system_type
*get_fs_type(const char *name
);
2681 extern struct super_block
*get_super(struct block_device
*);
2682 extern struct super_block
*get_super_thawed(struct block_device
*);
2683 extern struct super_block
*get_active_super(struct block_device
*bdev
);
2684 extern void drop_super(struct super_block
*sb
);
2685 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
2686 extern void iterate_supers_type(struct file_system_type
*,
2687 void (*)(struct super_block
*, void *), void *);
2689 extern int dcache_dir_open(struct inode
*, struct file
*);
2690 extern int dcache_dir_close(struct inode
*, struct file
*);
2691 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
2692 extern int dcache_readdir(struct file
*, struct dir_context
*);
2693 extern int simple_setattr(struct dentry
*, struct iattr
*);
2694 extern int simple_getattr(struct vfsmount
*, struct dentry
*, struct kstat
*);
2695 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
2696 extern int simple_open(struct inode
*inode
, struct file
*file
);
2697 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
2698 extern int simple_unlink(struct inode
*, struct dentry
*);
2699 extern int simple_rmdir(struct inode
*, struct dentry
*);
2700 extern int simple_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*);
2701 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
2702 extern int simple_empty(struct dentry
*);
2703 extern int simple_readpage(struct file
*file
, struct page
*page
);
2704 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
2705 loff_t pos
, unsigned len
, unsigned flags
,
2706 struct page
**pagep
, void **fsdata
);
2707 extern int simple_write_end(struct file
*file
, struct address_space
*mapping
,
2708 loff_t pos
, unsigned len
, unsigned copied
,
2709 struct page
*page
, void *fsdata
);
2710 extern int always_delete_dentry(const struct dentry
*);
2711 extern struct inode
*alloc_anon_inode(struct super_block
*);
2712 extern int simple_nosetlease(struct file
*, long, struct file_lock
**, void **);
2713 extern const struct dentry_operations simple_dentry_operations
;
2715 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
2716 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
2717 extern const struct file_operations simple_dir_operations
;
2718 extern const struct inode_operations simple_dir_inode_operations
;
2719 struct tree_descr
{ char *name
; const struct file_operations
*ops
; int mode
; };
2720 struct dentry
*d_alloc_name(struct dentry
*, const char *);
2721 extern int simple_fill_super(struct super_block
*, unsigned long, struct tree_descr
*);
2722 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
2723 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
2725 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
2726 loff_t
*ppos
, const void *from
, size_t available
);
2727 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
2728 const void __user
*from
, size_t count
);
2730 extern int __generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
2731 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
2733 extern int generic_check_addressable(unsigned, u64
);
2735 #ifdef CONFIG_MIGRATION
2736 extern int buffer_migrate_page(struct address_space
*,
2737 struct page
*, struct page
*,
2740 #define buffer_migrate_page NULL
2743 extern int inode_change_ok(const struct inode
*, struct iattr
*);
2744 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
2745 extern void setattr_copy(struct inode
*inode
, const struct iattr
*attr
);
2747 extern int file_update_time(struct file
*file
);
2749 extern int generic_show_options(struct seq_file
*m
, struct dentry
*root
);
2750 extern void save_mount_options(struct super_block
*sb
, char *options
);
2751 extern void replace_mount_options(struct super_block
*sb
, char *options
);
2753 static inline ino_t
parent_ino(struct dentry
*dentry
)
2758 * Don't strictly need d_lock here? If the parent ino could change
2759 * then surely we'd have a deeper race in the caller?
2761 spin_lock(&dentry
->d_lock
);
2762 res
= dentry
->d_parent
->d_inode
->i_ino
;
2763 spin_unlock(&dentry
->d_lock
);
2767 /* Transaction based IO helpers */
2770 * An argresp is stored in an allocated page and holds the
2771 * size of the argument or response, along with its content
2773 struct simple_transaction_argresp
{
2778 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2780 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
2782 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
2783 size_t size
, loff_t
*pos
);
2784 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
2786 void simple_transaction_set(struct file
*file
, size_t n
);
2789 * simple attribute files
2791 * These attributes behave similar to those in sysfs:
2793 * Writing to an attribute immediately sets a value, an open file can be
2794 * written to multiple times.
2796 * Reading from an attribute creates a buffer from the value that might get
2797 * read with multiple read calls. When the attribute has been read
2798 * completely, no further read calls are possible until the file is opened
2801 * All attributes contain a text representation of a numeric value
2802 * that are accessed with the get() and set() functions.
2804 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2805 static int __fops ## _open(struct inode *inode, struct file *file) \
2807 __simple_attr_check_format(__fmt, 0ull); \
2808 return simple_attr_open(inode, file, __get, __set, __fmt); \
2810 static const struct file_operations __fops = { \
2811 .owner = THIS_MODULE, \
2812 .open = __fops ## _open, \
2813 .release = simple_attr_release, \
2814 .read = simple_attr_read, \
2815 .write = simple_attr_write, \
2816 .llseek = generic_file_llseek, \
2819 static inline __printf(1, 2)
2820 void __simple_attr_check_format(const char *fmt
, ...)
2822 /* don't do anything, just let the compiler check the arguments; */
2825 int simple_attr_open(struct inode
*inode
, struct file
*file
,
2826 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
2828 int simple_attr_release(struct inode
*inode
, struct file
*file
);
2829 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
2830 size_t len
, loff_t
*ppos
);
2831 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
2832 size_t len
, loff_t
*ppos
);
2835 int proc_nr_files(struct ctl_table
*table
, int write
,
2836 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
2837 int proc_nr_dentry(struct ctl_table
*table
, int write
,
2838 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
2839 int proc_nr_inodes(struct ctl_table
*table
, int write
,
2840 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
2841 int __init
get_filesystem_list(char *buf
);
2843 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2844 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2846 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2847 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2848 (flag & __FMODE_NONOTIFY)))
2850 static inline int is_sxid(umode_t mode
)
2852 return (mode
& S_ISUID
) || ((mode
& S_ISGID
) && (mode
& S_IXGRP
));
2855 static inline int check_sticky(struct inode
*dir
, struct inode
*inode
)
2857 if (!(dir
->i_mode
& S_ISVTX
))
2860 return __check_sticky(dir
, inode
);
2863 static inline void inode_has_no_xattr(struct inode
*inode
)
2865 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& MS_NOSEC
))
2866 inode
->i_flags
|= S_NOSEC
;
2869 static inline bool is_root_inode(struct inode
*inode
)
2871 return inode
== inode
->i_sb
->s_root
->d_inode
;
2874 static inline bool dir_emit(struct dir_context
*ctx
,
2875 const char *name
, int namelen
,
2876 u64 ino
, unsigned type
)
2878 return ctx
->actor(ctx
, name
, namelen
, ctx
->pos
, ino
, type
) == 0;
2880 static inline bool dir_emit_dot(struct file
*file
, struct dir_context
*ctx
)
2882 return ctx
->actor(ctx
, ".", 1, ctx
->pos
,
2883 file
->f_path
.dentry
->d_inode
->i_ino
, DT_DIR
) == 0;
2885 static inline bool dir_emit_dotdot(struct file
*file
, struct dir_context
*ctx
)
2887 return ctx
->actor(ctx
, "..", 2, ctx
->pos
,
2888 parent_ino(file
->f_path
.dentry
), DT_DIR
) == 0;
2890 static inline bool dir_emit_dots(struct file
*file
, struct dir_context
*ctx
)
2892 if (ctx
->pos
== 0) {
2893 if (!dir_emit_dot(file
, ctx
))
2897 if (ctx
->pos
== 1) {
2898 if (!dir_emit_dotdot(file
, ctx
))
2904 static inline bool dir_relax(struct inode
*inode
)
2906 mutex_unlock(&inode
->i_mutex
);
2907 mutex_lock(&inode
->i_mutex
);
2908 return !IS_DEADDIR(inode
);
2911 #endif /* _LINUX_FS_H */