4 #include <linux/linkage.h>
5 #include <linux/wait.h>
6 #include <linux/kdev_t.h>
7 #include <linux/dcache.h>
8 #include <linux/path.h>
9 #include <linux/stat.h>
10 #include <linux/cache.h>
11 #include <linux/list.h>
12 #include <linux/list_lru.h>
13 #include <linux/llist.h>
14 #include <linux/radix-tree.h>
15 #include <linux/rbtree.h>
16 #include <linux/init.h>
17 #include <linux/pid.h>
18 #include <linux/bug.h>
19 #include <linux/mutex.h>
20 #include <linux/rwsem.h>
21 #include <linux/capability.h>
22 #include <linux/semaphore.h>
23 #include <linux/fiemap.h>
24 #include <linux/rculist_bl.h>
25 #include <linux/atomic.h>
26 #include <linux/shrinker.h>
27 #include <linux/migrate_mode.h>
28 #include <linux/uidgid.h>
29 #include <linux/lockdep.h>
30 #include <linux/percpu-rwsem.h>
31 #include <linux/blk_types.h>
32 #include <linux/workqueue.h>
33 #include <linux/percpu-rwsem.h>
34 #include <linux/delayed_call.h>
36 #include <asm/byteorder.h>
37 #include <uapi/linux/fs.h>
39 struct backing_dev_info
;
41 struct export_operations
;
46 struct pipe_inode_info
;
47 struct poll_table_struct
;
49 struct vm_area_struct
;
52 struct swap_info_struct
;
54 struct workqueue_struct
;
57 struct fscrypt_operations
;
59 extern void __init
inode_init(void);
60 extern void __init
inode_init_early(void);
61 extern void __init
files_init(void);
62 extern void __init
files_maxfiles_init(void);
64 extern struct files_stat_struct files_stat
;
65 extern unsigned long get_max_files(void);
66 extern int sysctl_nr_open
;
67 extern struct inodes_stat_t inodes_stat
;
68 extern int leases_enable
, lease_break_time
;
69 extern int sysctl_protected_symlinks
;
70 extern int sysctl_protected_hardlinks
;
73 typedef int (get_block_t
)(struct inode
*inode
, sector_t iblock
,
74 struct buffer_head
*bh_result
, int create
);
75 typedef int (dio_iodone_t
)(struct kiocb
*iocb
, loff_t offset
,
76 ssize_t bytes
, void *private);
78 #define MAY_EXEC 0x00000001
79 #define MAY_WRITE 0x00000002
80 #define MAY_READ 0x00000004
81 #define MAY_APPEND 0x00000008
82 #define MAY_ACCESS 0x00000010
83 #define MAY_OPEN 0x00000020
84 #define MAY_CHDIR 0x00000040
85 /* called from RCU mode, don't block */
86 #define MAY_NOT_BLOCK 0x00000080
89 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
90 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
93 /* file is open for reading */
94 #define FMODE_READ ((__force fmode_t)0x1)
95 /* file is open for writing */
96 #define FMODE_WRITE ((__force fmode_t)0x2)
97 /* file is seekable */
98 #define FMODE_LSEEK ((__force fmode_t)0x4)
99 /* file can be accessed using pread */
100 #define FMODE_PREAD ((__force fmode_t)0x8)
101 /* file can be accessed using pwrite */
102 #define FMODE_PWRITE ((__force fmode_t)0x10)
103 /* File is opened for execution with sys_execve / sys_uselib */
104 #define FMODE_EXEC ((__force fmode_t)0x20)
105 /* File is opened with O_NDELAY (only set for block devices) */
106 #define FMODE_NDELAY ((__force fmode_t)0x40)
107 /* File is opened with O_EXCL (only set for block devices) */
108 #define FMODE_EXCL ((__force fmode_t)0x80)
109 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
110 (specialy hack for floppy.c) */
111 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
112 /* 32bit hashes as llseek() offset (for directories) */
113 #define FMODE_32BITHASH ((__force fmode_t)0x200)
114 /* 64bit hashes as llseek() offset (for directories) */
115 #define FMODE_64BITHASH ((__force fmode_t)0x400)
118 * Don't update ctime and mtime.
120 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
121 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
123 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
125 /* Expect random access pattern */
126 #define FMODE_RANDOM ((__force fmode_t)0x1000)
128 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
129 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
131 /* File is opened with O_PATH; almost nothing can be done with it */
132 #define FMODE_PATH ((__force fmode_t)0x4000)
134 /* File needs atomic accesses to f_pos */
135 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
136 /* Write access to underlying fs */
137 #define FMODE_WRITER ((__force fmode_t)0x10000)
138 /* Has read method(s) */
139 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
140 /* Has write method(s) */
141 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
143 /* File was opened by fanotify and shouldn't generate fanotify events */
144 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
147 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
148 * that indicates that they should check the contents of the iovec are
149 * valid, but not check the memory that the iovec elements
152 #define CHECK_IOVEC_ONLY -1
155 * The below are the various read and write types that we support. Some of
156 * them include behavioral modifiers that send information down to the
157 * block layer and IO scheduler. Terminology:
159 * The block layer uses device plugging to defer IO a little bit, in
160 * the hope that we will see more IO very shortly. This increases
161 * coalescing of adjacent IO and thus reduces the number of IOs we
162 * have to send to the device. It also allows for better queuing,
163 * if the IO isn't mergeable. If the caller is going to be waiting
164 * for the IO, then he must ensure that the device is unplugged so
165 * that the IO is dispatched to the driver.
167 * All IO is handled async in Linux. This is fine for background
168 * writes, but for reads or writes that someone waits for completion
169 * on, we want to notify the block layer and IO scheduler so that they
170 * know about it. That allows them to make better scheduling
171 * decisions. So when the below references 'sync' and 'async', it
172 * is referencing this priority hint.
174 * With that in mind, the available types are:
176 * READ A normal read operation. Device will be plugged.
177 * READ_SYNC A synchronous read. Device is not plugged, caller can
178 * immediately wait on this read without caring about
180 * READA Used for read-ahead operations. Lower priority, and the
181 * block layer could (in theory) choose to ignore this
182 * request if it runs into resource problems.
183 * WRITE A normal async write. Device will be plugged.
184 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
185 * the hint that someone will be waiting on this IO
186 * shortly. The write equivalent of READ_SYNC.
187 * WRITE_ODIRECT Special case write for O_DIRECT only.
188 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
189 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
190 * non-volatile media on completion.
191 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
192 * by a cache flush and data is guaranteed to be on
193 * non-volatile media on completion.
196 #define RW_MASK REQ_WRITE
197 #define RWA_MASK REQ_RAHEAD
200 #define WRITE RW_MASK
201 #define READA RWA_MASK
203 #define READ_SYNC (READ | REQ_SYNC)
204 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
205 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
206 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
207 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
208 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
211 * Attribute flags. These should be or-ed together to figure out what
214 #define ATTR_MODE (1 << 0)
215 #define ATTR_UID (1 << 1)
216 #define ATTR_GID (1 << 2)
217 #define ATTR_SIZE (1 << 3)
218 #define ATTR_ATIME (1 << 4)
219 #define ATTR_MTIME (1 << 5)
220 #define ATTR_CTIME (1 << 6)
221 #define ATTR_ATIME_SET (1 << 7)
222 #define ATTR_MTIME_SET (1 << 8)
223 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
224 #define ATTR_ATTR_FLAG (1 << 10)
225 #define ATTR_KILL_SUID (1 << 11)
226 #define ATTR_KILL_SGID (1 << 12)
227 #define ATTR_FILE (1 << 13)
228 #define ATTR_KILL_PRIV (1 << 14)
229 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
230 #define ATTR_TIMES_SET (1 << 16)
233 * Whiteout is represented by a char device. The following constants define the
234 * mode and device number to use.
236 #define WHITEOUT_MODE 0
237 #define WHITEOUT_DEV 0
240 * This is the Inode Attributes structure, used for notify_change(). It
241 * uses the above definitions as flags, to know which values have changed.
242 * Also, in this manner, a Filesystem can look at only the values it cares
243 * about. Basically, these are the attributes that the VFS layer can
244 * request to change from the FS layer.
246 * Derek Atkins <warlord@MIT.EDU> 94-10-20
249 unsigned int ia_valid
;
254 struct timespec ia_atime
;
255 struct timespec ia_mtime
;
256 struct timespec ia_ctime
;
259 * Not an attribute, but an auxiliary info for filesystems wanting to
260 * implement an ftruncate() like method. NOTE: filesystem should
261 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
263 struct file
*ia_file
;
267 * Includes for diskquotas.
269 #include <linux/quota.h>
272 * Maximum number of layers of fs stack. Needs to be limited to
273 * prevent kernel stack overflow
275 #define FILESYSTEM_MAX_STACK_DEPTH 2
278 * enum positive_aop_returns - aop return codes with specific semantics
280 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
281 * completed, that the page is still locked, and
282 * should be considered active. The VM uses this hint
283 * to return the page to the active list -- it won't
284 * be a candidate for writeback again in the near
285 * future. Other callers must be careful to unlock
286 * the page if they get this return. Returned by
289 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
290 * unlocked it and the page might have been truncated.
291 * The caller should back up to acquiring a new page and
292 * trying again. The aop will be taking reasonable
293 * precautions not to livelock. If the caller held a page
294 * reference, it should drop it before retrying. Returned
297 * address_space_operation functions return these large constants to indicate
298 * special semantics to the caller. These are much larger than the bytes in a
299 * page to allow for functions that return the number of bytes operated on in a
303 enum positive_aop_returns
{
304 AOP_WRITEPAGE_ACTIVATE
= 0x80000,
305 AOP_TRUNCATED_PAGE
= 0x80001,
308 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
309 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
310 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
311 * helper code (eg buffer layer)
312 * to clear GFP_FS from alloc */
315 * oh the beauties of C type declarations.
318 struct address_space
;
319 struct writeback_control
;
321 #define IOCB_EVENTFD (1 << 0)
322 #define IOCB_APPEND (1 << 1)
323 #define IOCB_DIRECT (1 << 2)
324 #define IOCB_HIPRI (1 << 3)
325 #define IOCB_DSYNC (1 << 4)
326 #define IOCB_SYNC (1 << 5)
329 struct file
*ki_filp
;
331 void (*ki_complete
)(struct kiocb
*iocb
, long ret
, long ret2
);
336 static inline bool is_sync_kiocb(struct kiocb
*kiocb
)
338 return kiocb
->ki_complete
== NULL
;
341 static inline int iocb_flags(struct file
*file
);
343 static inline void init_sync_kiocb(struct kiocb
*kiocb
, struct file
*filp
)
345 *kiocb
= (struct kiocb
) {
347 .ki_flags
= iocb_flags(filp
),
352 * "descriptor" for what we're up to with a read.
353 * This allows us to use the same read code yet
354 * have multiple different users of the data that
355 * we read from a file.
357 * The simplest case just copies the data to user
370 typedef int (*read_actor_t
)(read_descriptor_t
*, struct page
*,
371 unsigned long, unsigned long);
373 struct address_space_operations
{
374 int (*writepage
)(struct page
*page
, struct writeback_control
*wbc
);
375 int (*readpage
)(struct file
*, struct page
*);
377 /* Write back some dirty pages from this mapping. */
378 int (*writepages
)(struct address_space
*, struct writeback_control
*);
380 /* Set a page dirty. Return true if this dirtied it */
381 int (*set_page_dirty
)(struct page
*page
);
383 int (*readpages
)(struct file
*filp
, struct address_space
*mapping
,
384 struct list_head
*pages
, unsigned nr_pages
);
386 int (*write_begin
)(struct file
*, struct address_space
*mapping
,
387 loff_t pos
, unsigned len
, unsigned flags
,
388 struct page
**pagep
, void **fsdata
);
389 int (*write_end
)(struct file
*, struct address_space
*mapping
,
390 loff_t pos
, unsigned len
, unsigned copied
,
391 struct page
*page
, void *fsdata
);
393 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
394 sector_t (*bmap
)(struct address_space
*, sector_t
);
395 void (*invalidatepage
) (struct page
*, unsigned int, unsigned int);
396 int (*releasepage
) (struct page
*, gfp_t
);
397 void (*freepage
)(struct page
*);
398 ssize_t (*direct_IO
)(struct kiocb
*, struct iov_iter
*iter
);
400 * migrate the contents of a page to the specified target. If
401 * migrate_mode is MIGRATE_ASYNC, it must not block.
403 int (*migratepage
) (struct address_space
*,
404 struct page
*, struct page
*, enum migrate_mode
);
405 int (*launder_page
) (struct page
*);
406 int (*is_partially_uptodate
) (struct page
*, unsigned long,
408 void (*is_dirty_writeback
) (struct page
*, bool *, bool *);
409 int (*error_remove_page
)(struct address_space
*, struct page
*);
411 /* swapfile support */
412 int (*swap_activate
)(struct swap_info_struct
*sis
, struct file
*file
,
414 void (*swap_deactivate
)(struct file
*file
);
417 extern const struct address_space_operations empty_aops
;
420 * pagecache_write_begin/pagecache_write_end must be used by general code
421 * to write into the pagecache.
423 int pagecache_write_begin(struct file
*, struct address_space
*mapping
,
424 loff_t pos
, unsigned len
, unsigned flags
,
425 struct page
**pagep
, void **fsdata
);
427 int pagecache_write_end(struct file
*, struct address_space
*mapping
,
428 loff_t pos
, unsigned len
, unsigned copied
,
429 struct page
*page
, void *fsdata
);
431 struct address_space
{
432 struct inode
*host
; /* owner: inode, block_device */
433 struct radix_tree_root page_tree
; /* radix tree of all pages */
434 spinlock_t tree_lock
; /* and lock protecting it */
435 atomic_t i_mmap_writable
;/* count VM_SHARED mappings */
436 struct rb_root i_mmap
; /* tree of private and shared mappings */
437 struct rw_semaphore i_mmap_rwsem
; /* protect tree, count, list */
438 /* Protected by tree_lock together with the radix tree */
439 unsigned long nrpages
; /* number of total pages */
440 /* number of shadow or DAX exceptional entries */
441 unsigned long nrexceptional
;
442 pgoff_t writeback_index
;/* writeback starts here */
443 const struct address_space_operations
*a_ops
; /* methods */
444 unsigned long flags
; /* error bits/gfp mask */
445 spinlock_t private_lock
; /* for use by the address_space */
446 struct list_head private_list
; /* ditto */
447 void *private_data
; /* ditto */
448 } __attribute__((aligned(sizeof(long))));
450 * On most architectures that alignment is already the case; but
451 * must be enforced here for CRIS, to let the least significant bit
452 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
454 struct request_queue
;
456 struct block_device
{
457 dev_t bd_dev
; /* not a kdev_t - it's a search key */
459 struct inode
* bd_inode
; /* will die */
460 struct super_block
* bd_super
;
461 struct mutex bd_mutex
; /* open/close mutex */
462 struct list_head bd_inodes
;
466 bool bd_write_holder
;
468 struct list_head bd_holder_disks
;
470 struct block_device
* bd_contains
;
471 unsigned bd_block_size
;
472 struct hd_struct
* bd_part
;
473 /* number of times partitions within this device have been opened. */
474 unsigned bd_part_count
;
476 struct gendisk
* bd_disk
;
477 struct request_queue
* bd_queue
;
478 struct list_head bd_list
;
480 * Private data. You must have bd_claim'ed the block_device
481 * to use this. NOTE: bd_claim allows an owner to claim
482 * the same device multiple times, the owner must take special
483 * care to not mess up bd_private for that case.
485 unsigned long bd_private
;
487 /* The counter of freeze processes */
488 int bd_fsfreeze_count
;
489 /* Mutex for freeze */
490 struct mutex bd_fsfreeze_mutex
;
494 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
497 #define PAGECACHE_TAG_DIRTY 0
498 #define PAGECACHE_TAG_WRITEBACK 1
499 #define PAGECACHE_TAG_TOWRITE 2
501 int mapping_tagged(struct address_space
*mapping
, int tag
);
503 static inline void i_mmap_lock_write(struct address_space
*mapping
)
505 down_write(&mapping
->i_mmap_rwsem
);
508 static inline void i_mmap_unlock_write(struct address_space
*mapping
)
510 up_write(&mapping
->i_mmap_rwsem
);
513 static inline void i_mmap_lock_read(struct address_space
*mapping
)
515 down_read(&mapping
->i_mmap_rwsem
);
518 static inline void i_mmap_unlock_read(struct address_space
*mapping
)
520 up_read(&mapping
->i_mmap_rwsem
);
524 * Might pages of this file be mapped into userspace?
526 static inline int mapping_mapped(struct address_space
*mapping
)
528 return !RB_EMPTY_ROOT(&mapping
->i_mmap
);
532 * Might pages of this file have been modified in userspace?
533 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
534 * marks vma as VM_SHARED if it is shared, and the file was opened for
535 * writing i.e. vma may be mprotected writable even if now readonly.
537 * If i_mmap_writable is negative, no new writable mappings are allowed. You
538 * can only deny writable mappings, if none exists right now.
540 static inline int mapping_writably_mapped(struct address_space
*mapping
)
542 return atomic_read(&mapping
->i_mmap_writable
) > 0;
545 static inline int mapping_map_writable(struct address_space
*mapping
)
547 return atomic_inc_unless_negative(&mapping
->i_mmap_writable
) ?
551 static inline void mapping_unmap_writable(struct address_space
*mapping
)
553 atomic_dec(&mapping
->i_mmap_writable
);
556 static inline int mapping_deny_writable(struct address_space
*mapping
)
558 return atomic_dec_unless_positive(&mapping
->i_mmap_writable
) ?
562 static inline void mapping_allow_writable(struct address_space
*mapping
)
564 atomic_inc(&mapping
->i_mmap_writable
);
568 * Use sequence counter to get consistent i_size on 32-bit processors.
570 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
571 #include <linux/seqlock.h>
572 #define __NEED_I_SIZE_ORDERED
573 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
575 #define i_size_ordered_init(inode) do { } while (0)
579 #define ACL_NOT_CACHED ((void *)(-1))
581 static inline struct posix_acl
*
582 uncached_acl_sentinel(struct task_struct
*task
)
584 return (void *)task
+ 1;
588 is_uncached_acl(struct posix_acl
*acl
)
590 return (long)acl
& 1;
593 #define IOP_FASTPERM 0x0001
594 #define IOP_LOOKUP 0x0002
595 #define IOP_NOFOLLOW 0x0004
598 * Keep mostly read-only and often accessed (especially for
599 * the RCU path lookup and 'stat' data) fields at the beginning
600 * of the 'struct inode'
604 unsigned short i_opflags
;
607 unsigned int i_flags
;
609 #ifdef CONFIG_FS_POSIX_ACL
610 struct posix_acl
*i_acl
;
611 struct posix_acl
*i_default_acl
;
614 const struct inode_operations
*i_op
;
615 struct super_block
*i_sb
;
616 struct address_space
*i_mapping
;
618 #ifdef CONFIG_SECURITY
622 /* Stat data, not accessed from path walking */
625 * Filesystems may only read i_nlink directly. They shall use the
626 * following functions for modification:
628 * (set|clear|inc|drop)_nlink
629 * inode_(inc|dec)_link_count
632 const unsigned int i_nlink
;
633 unsigned int __i_nlink
;
637 struct timespec i_atime
;
638 struct timespec i_mtime
;
639 struct timespec i_ctime
;
640 spinlock_t i_lock
; /* i_blocks, i_bytes, maybe i_size */
641 unsigned short i_bytes
;
642 unsigned int i_blkbits
;
645 #ifdef __NEED_I_SIZE_ORDERED
646 seqcount_t i_size_seqcount
;
650 unsigned long i_state
;
651 struct rw_semaphore i_rwsem
;
653 unsigned long dirtied_when
; /* jiffies of first dirtying */
654 unsigned long dirtied_time_when
;
656 struct hlist_node i_hash
;
657 struct list_head i_io_list
; /* backing dev IO list */
658 #ifdef CONFIG_CGROUP_WRITEBACK
659 struct bdi_writeback
*i_wb
; /* the associated cgroup wb */
661 /* foreign inode detection, see wbc_detach_inode() */
663 u16 i_wb_frn_avg_time
;
664 u16 i_wb_frn_history
;
666 struct list_head i_lru
; /* inode LRU list */
667 struct list_head i_sb_list
;
669 struct hlist_head i_dentry
;
670 struct rcu_head i_rcu
;
674 atomic_t i_dio_count
;
675 atomic_t i_writecount
;
677 atomic_t i_readcount
; /* struct files open RO */
679 const struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
680 struct file_lock_context
*i_flctx
;
681 struct address_space i_data
;
682 struct list_head i_devices
;
684 struct pipe_inode_info
*i_pipe
;
685 struct block_device
*i_bdev
;
693 #ifdef CONFIG_FSNOTIFY
694 __u32 i_fsnotify_mask
; /* all events this inode cares about */
695 struct hlist_head i_fsnotify_marks
;
698 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
699 struct fscrypt_info
*i_crypt_info
;
702 void *i_private
; /* fs or device private pointer */
705 static inline int inode_unhashed(struct inode
*inode
)
707 return hlist_unhashed(&inode
->i_hash
);
711 * inode->i_mutex nesting subclasses for the lock validator:
713 * 0: the object of the current VFS operation
717 * 4: second non-directory
718 * 5: second parent (when locking independent directories in rename)
720 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
721 * non-directories at once.
723 * The locking order between these classes is
724 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
726 enum inode_i_mutex_lock_class
736 static inline void inode_lock(struct inode
*inode
)
738 down_write(&inode
->i_rwsem
);
741 static inline void inode_unlock(struct inode
*inode
)
743 up_write(&inode
->i_rwsem
);
746 static inline void inode_lock_shared(struct inode
*inode
)
748 down_read(&inode
->i_rwsem
);
751 static inline void inode_unlock_shared(struct inode
*inode
)
753 up_read(&inode
->i_rwsem
);
756 static inline int inode_trylock(struct inode
*inode
)
758 return down_write_trylock(&inode
->i_rwsem
);
761 static inline int inode_trylock_shared(struct inode
*inode
)
763 return down_read_trylock(&inode
->i_rwsem
);
766 static inline int inode_is_locked(struct inode
*inode
)
768 return rwsem_is_locked(&inode
->i_rwsem
);
771 static inline void inode_lock_nested(struct inode
*inode
, unsigned subclass
)
773 down_write_nested(&inode
->i_rwsem
, subclass
);
776 void lock_two_nondirectories(struct inode
*, struct inode
*);
777 void unlock_two_nondirectories(struct inode
*, struct inode
*);
780 * NOTE: in a 32bit arch with a preemptable kernel and
781 * an UP compile the i_size_read/write must be atomic
782 * with respect to the local cpu (unlike with preempt disabled),
783 * but they don't need to be atomic with respect to other cpus like in
784 * true SMP (so they need either to either locally disable irq around
785 * the read or for example on x86 they can be still implemented as a
786 * cmpxchg8b without the need of the lock prefix). For SMP compiles
787 * and 64bit archs it makes no difference if preempt is enabled or not.
789 static inline loff_t
i_size_read(const struct inode
*inode
)
791 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
796 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
797 i_size
= inode
->i_size
;
798 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
800 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
804 i_size
= inode
->i_size
;
808 return inode
->i_size
;
813 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
814 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
815 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
817 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
819 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
821 write_seqcount_begin(&inode
->i_size_seqcount
);
822 inode
->i_size
= i_size
;
823 write_seqcount_end(&inode
->i_size_seqcount
);
825 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
827 inode
->i_size
= i_size
;
830 inode
->i_size
= i_size
;
834 /* Helper functions so that in most cases filesystems will
835 * not need to deal directly with kuid_t and kgid_t and can
836 * instead deal with the raw numeric values that are stored
839 static inline uid_t
i_uid_read(const struct inode
*inode
)
841 return from_kuid(&init_user_ns
, inode
->i_uid
);
844 static inline gid_t
i_gid_read(const struct inode
*inode
)
846 return from_kgid(&init_user_ns
, inode
->i_gid
);
849 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
851 inode
->i_uid
= make_kuid(&init_user_ns
, uid
);
854 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
856 inode
->i_gid
= make_kgid(&init_user_ns
, gid
);
859 static inline unsigned iminor(const struct inode
*inode
)
861 return MINOR(inode
->i_rdev
);
864 static inline unsigned imajor(const struct inode
*inode
)
866 return MAJOR(inode
->i_rdev
);
869 extern struct block_device
*I_BDEV(struct inode
*inode
);
872 rwlock_t lock
; /* protects pid, uid, euid fields */
873 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
874 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
875 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
876 int signum
; /* posix.1b rt signal to be delivered on IO */
880 * Track a single file's readahead state
882 struct file_ra_state
{
883 pgoff_t start
; /* where readahead started */
884 unsigned int size
; /* # of readahead pages */
885 unsigned int async_size
; /* do asynchronous readahead when
886 there are only # of pages ahead */
888 unsigned int ra_pages
; /* Maximum readahead window */
889 unsigned int mmap_miss
; /* Cache miss stat for mmap accesses */
890 loff_t prev_pos
; /* Cache last read() position */
894 * Check if @index falls in the readahead windows.
896 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
898 return (index
>= ra
->start
&&
899 index
< ra
->start
+ ra
->size
);
904 struct llist_node fu_llist
;
905 struct rcu_head fu_rcuhead
;
908 struct inode
*f_inode
; /* cached value */
909 const struct file_operations
*f_op
;
912 * Protects f_ep_links, f_flags.
913 * Must not be taken from IRQ context.
916 atomic_long_t f_count
;
917 unsigned int f_flags
;
919 struct mutex f_pos_lock
;
921 struct fown_struct f_owner
;
922 const struct cred
*f_cred
;
923 struct file_ra_state f_ra
;
926 #ifdef CONFIG_SECURITY
929 /* needed for tty driver, and maybe others */
933 /* Used by fs/eventpoll.c to link all the hooks to this file */
934 struct list_head f_ep_links
;
935 struct list_head f_tfile_llink
;
936 #endif /* #ifdef CONFIG_EPOLL */
937 struct address_space
*f_mapping
;
938 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
943 /* file identifier */
944 unsigned char f_handle
[0];
947 static inline struct file
*get_file(struct file
*f
)
949 atomic_long_inc(&f
->f_count
);
952 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
953 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
954 #define file_count(x) atomic_long_read(&(x)->f_count)
956 #define MAX_NON_LFS ((1UL<<31) - 1)
958 /* Page cache limit. The filesystems should put that into their s_maxbytes
959 limits, otherwise bad things can happen in VM. */
960 #if BITS_PER_LONG==32
961 #define MAX_LFS_FILESIZE (((loff_t)PAGE_SIZE << (BITS_PER_LONG-1))-1)
962 #elif BITS_PER_LONG==64
963 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
968 #define FL_DELEG 4 /* NFSv4 delegation */
969 #define FL_ACCESS 8 /* not trying to lock, just looking */
970 #define FL_EXISTS 16 /* when unlocking, test for existence */
971 #define FL_LEASE 32 /* lease held on this file */
972 #define FL_CLOSE 64 /* unlock on close */
973 #define FL_SLEEP 128 /* A blocking lock */
974 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
975 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
976 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
977 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
980 * Special return value from posix_lock_file() and vfs_lock_file() for
981 * asynchronous locking.
983 #define FILE_LOCK_DEFERRED 1
985 /* legacy typedef, should eventually be removed */
986 typedef void *fl_owner_t
;
990 struct file_lock_operations
{
991 void (*fl_copy_lock
)(struct file_lock
*, struct file_lock
*);
992 void (*fl_release_private
)(struct file_lock
*);
995 struct lock_manager_operations
{
996 int (*lm_compare_owner
)(struct file_lock
*, struct file_lock
*);
997 unsigned long (*lm_owner_key
)(struct file_lock
*);
998 fl_owner_t (*lm_get_owner
)(fl_owner_t
);
999 void (*lm_put_owner
)(fl_owner_t
);
1000 void (*lm_notify
)(struct file_lock
*); /* unblock callback */
1001 int (*lm_grant
)(struct file_lock
*, int);
1002 bool (*lm_break
)(struct file_lock
*);
1003 int (*lm_change
)(struct file_lock
*, int, struct list_head
*);
1004 void (*lm_setup
)(struct file_lock
*, void **);
1007 struct lock_manager
{
1008 struct list_head list
;
1010 * NFSv4 and up also want opens blocked during the grace period;
1017 void locks_start_grace(struct net
*, struct lock_manager
*);
1018 void locks_end_grace(struct lock_manager
*);
1019 int locks_in_grace(struct net
*);
1020 int opens_in_grace(struct net
*);
1022 /* that will die - we need it for nfs_lock_info */
1023 #include <linux/nfs_fs_i.h>
1026 * struct file_lock represents a generic "file lock". It's used to represent
1027 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1028 * note that the same struct is used to represent both a request for a lock and
1029 * the lock itself, but the same object is never used for both.
1031 * FIXME: should we create a separate "struct lock_request" to help distinguish
1034 * The varous i_flctx lists are ordered by:
1037 * 2) lock range start
1040 * Obviously, the last two criteria only matter for POSIX locks.
1043 struct file_lock
*fl_next
; /* singly linked list for this inode */
1044 struct list_head fl_list
; /* link into file_lock_context */
1045 struct hlist_node fl_link
; /* node in global lists */
1046 struct list_head fl_block
; /* circular list of blocked processes */
1047 fl_owner_t fl_owner
;
1048 unsigned int fl_flags
;
1049 unsigned char fl_type
;
1050 unsigned int fl_pid
;
1051 int fl_link_cpu
; /* what cpu's list is this on? */
1052 struct pid
*fl_nspid
;
1053 wait_queue_head_t fl_wait
;
1054 struct file
*fl_file
;
1058 struct fasync_struct
* fl_fasync
; /* for lease break notifications */
1059 /* for lease breaks: */
1060 unsigned long fl_break_time
;
1061 unsigned long fl_downgrade_time
;
1063 const struct file_lock_operations
*fl_ops
; /* Callbacks for filesystems */
1064 const struct lock_manager_operations
*fl_lmops
; /* Callbacks for lockmanagers */
1066 struct nfs_lock_info nfs_fl
;
1067 struct nfs4_lock_info nfs4_fl
;
1069 struct list_head link
; /* link in AFS vnode's pending_locks list */
1070 int state
; /* state of grant or error if -ve */
1075 struct file_lock_context
{
1076 spinlock_t flc_lock
;
1077 struct list_head flc_flock
;
1078 struct list_head flc_posix
;
1079 struct list_head flc_lease
;
1082 /* The following constant reflects the upper bound of the file/locking space */
1084 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1085 #define OFFSET_MAX INT_LIMIT(loff_t)
1086 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1089 #include <linux/fcntl.h>
1091 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
1093 #ifdef CONFIG_FILE_LOCKING
1094 extern int fcntl_getlk(struct file
*, unsigned int, struct flock __user
*);
1095 extern int fcntl_setlk(unsigned int, struct file
*, unsigned int,
1096 struct flock __user
*);
1098 #if BITS_PER_LONG == 32
1099 extern int fcntl_getlk64(struct file
*, unsigned int, struct flock64 __user
*);
1100 extern int fcntl_setlk64(unsigned int, struct file
*, unsigned int,
1101 struct flock64 __user
*);
1104 extern int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
);
1105 extern int fcntl_getlease(struct file
*filp
);
1108 void locks_free_lock_context(struct inode
*inode
);
1109 void locks_free_lock(struct file_lock
*fl
);
1110 extern void locks_init_lock(struct file_lock
*);
1111 extern struct file_lock
* locks_alloc_lock(void);
1112 extern void locks_copy_lock(struct file_lock
*, struct file_lock
*);
1113 extern void locks_copy_conflock(struct file_lock
*, struct file_lock
*);
1114 extern void locks_remove_posix(struct file
*, fl_owner_t
);
1115 extern void locks_remove_file(struct file
*);
1116 extern void locks_release_private(struct file_lock
*);
1117 extern void posix_test_lock(struct file
*, struct file_lock
*);
1118 extern int posix_lock_file(struct file
*, struct file_lock
*, struct file_lock
*);
1119 extern int posix_unblock_lock(struct file_lock
*);
1120 extern int vfs_test_lock(struct file
*, struct file_lock
*);
1121 extern int vfs_lock_file(struct file
*, unsigned int, struct file_lock
*, struct file_lock
*);
1122 extern int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
);
1123 extern int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
);
1124 extern int __break_lease(struct inode
*inode
, unsigned int flags
, unsigned int type
);
1125 extern void lease_get_mtime(struct inode
*, struct timespec
*time
);
1126 extern int generic_setlease(struct file
*, long, struct file_lock
**, void **priv
);
1127 extern int vfs_setlease(struct file
*, long, struct file_lock
**, void **);
1128 extern int lease_modify(struct file_lock
*, int, struct list_head
*);
1129 struct files_struct
;
1130 extern void show_fd_locks(struct seq_file
*f
,
1131 struct file
*filp
, struct files_struct
*files
);
1132 #else /* !CONFIG_FILE_LOCKING */
1133 static inline int fcntl_getlk(struct file
*file
, unsigned int cmd
,
1134 struct flock __user
*user
)
1139 static inline int fcntl_setlk(unsigned int fd
, struct file
*file
,
1140 unsigned int cmd
, struct flock __user
*user
)
1145 #if BITS_PER_LONG == 32
1146 static inline int fcntl_getlk64(struct file
*file
, unsigned int cmd
,
1147 struct flock64 __user
*user
)
1152 static inline int fcntl_setlk64(unsigned int fd
, struct file
*file
,
1153 unsigned int cmd
, struct flock64 __user
*user
)
1158 static inline int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
)
1163 static inline int fcntl_getlease(struct file
*filp
)
1169 locks_free_lock_context(struct inode
*inode
)
1173 static inline void locks_init_lock(struct file_lock
*fl
)
1178 static inline void locks_copy_conflock(struct file_lock
*new, struct file_lock
*fl
)
1183 static inline void locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
1188 static inline void locks_remove_posix(struct file
*filp
, fl_owner_t owner
)
1193 static inline void locks_remove_file(struct file
*filp
)
1198 static inline void posix_test_lock(struct file
*filp
, struct file_lock
*fl
)
1203 static inline int posix_lock_file(struct file
*filp
, struct file_lock
*fl
,
1204 struct file_lock
*conflock
)
1209 static inline int posix_unblock_lock(struct file_lock
*waiter
)
1214 static inline int vfs_test_lock(struct file
*filp
, struct file_lock
*fl
)
1219 static inline int vfs_lock_file(struct file
*filp
, unsigned int cmd
,
1220 struct file_lock
*fl
, struct file_lock
*conf
)
1225 static inline int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
)
1230 static inline int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
)
1235 static inline int __break_lease(struct inode
*inode
, unsigned int mode
, unsigned int type
)
1240 static inline void lease_get_mtime(struct inode
*inode
, struct timespec
*time
)
1245 static inline int generic_setlease(struct file
*filp
, long arg
,
1246 struct file_lock
**flp
, void **priv
)
1251 static inline int vfs_setlease(struct file
*filp
, long arg
,
1252 struct file_lock
**lease
, void **priv
)
1257 static inline int lease_modify(struct file_lock
*fl
, int arg
,
1258 struct list_head
*dispose
)
1263 struct files_struct
;
1264 static inline void show_fd_locks(struct seq_file
*f
,
1265 struct file
*filp
, struct files_struct
*files
) {}
1266 #endif /* !CONFIG_FILE_LOCKING */
1268 static inline struct inode
*file_inode(const struct file
*f
)
1273 static inline struct dentry
*file_dentry(const struct file
*file
)
1275 struct dentry
*dentry
= file
->f_path
.dentry
;
1277 if (unlikely(dentry
->d_flags
& DCACHE_OP_REAL
))
1278 return dentry
->d_op
->d_real(dentry
, file_inode(file
));
1283 static inline int locks_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1285 return locks_lock_inode_wait(file_inode(filp
), fl
);
1288 struct fasync_struct
{
1292 struct fasync_struct
*fa_next
; /* singly linked list */
1293 struct file
*fa_file
;
1294 struct rcu_head fa_rcu
;
1297 #define FASYNC_MAGIC 0x4601
1299 /* SMP safe fasync helpers: */
1300 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1301 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1302 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1303 extern struct fasync_struct
*fasync_alloc(void);
1304 extern void fasync_free(struct fasync_struct
*);
1306 /* can be called from interrupts */
1307 extern void kill_fasync(struct fasync_struct
**, int, int);
1309 extern void __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1310 extern void f_setown(struct file
*filp
, unsigned long arg
, int force
);
1311 extern void f_delown(struct file
*filp
);
1312 extern pid_t
f_getown(struct file
*filp
);
1313 extern int send_sigurg(struct fown_struct
*fown
);
1321 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1322 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1323 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1324 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1325 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1328 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1329 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1331 /* Possible states of 'frozen' field */
1333 SB_UNFROZEN
= 0, /* FS is unfrozen */
1334 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1335 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1336 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1337 * internal threads if needed) */
1338 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1341 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1344 int frozen
; /* Is sb frozen? */
1345 wait_queue_head_t wait_unfrozen
; /* for get_super_thawed() */
1346 struct percpu_rw_semaphore rw_sem
[SB_FREEZE_LEVELS
];
1349 struct super_block
{
1350 struct list_head s_list
; /* Keep this first */
1351 dev_t s_dev
; /* search index; _not_ kdev_t */
1352 unsigned char s_blocksize_bits
;
1353 unsigned long s_blocksize
;
1354 loff_t s_maxbytes
; /* Max file size */
1355 struct file_system_type
*s_type
;
1356 const struct super_operations
*s_op
;
1357 const struct dquot_operations
*dq_op
;
1358 const struct quotactl_ops
*s_qcop
;
1359 const struct export_operations
*s_export_op
;
1360 unsigned long s_flags
;
1361 unsigned long s_iflags
; /* internal SB_I_* flags */
1362 unsigned long s_magic
;
1363 struct dentry
*s_root
;
1364 struct rw_semaphore s_umount
;
1367 #ifdef CONFIG_SECURITY
1370 const struct xattr_handler
**s_xattr
;
1372 const struct fscrypt_operations
*s_cop
;
1374 struct hlist_bl_head s_anon
; /* anonymous dentries for (nfs) exporting */
1375 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1376 struct block_device
*s_bdev
;
1377 struct backing_dev_info
*s_bdi
;
1378 struct mtd_info
*s_mtd
;
1379 struct hlist_node s_instances
;
1380 unsigned int s_quota_types
; /* Bitmask of supported quota types */
1381 struct quota_info s_dquot
; /* Diskquota specific options */
1383 struct sb_writers s_writers
;
1385 char s_id
[32]; /* Informational name */
1386 u8 s_uuid
[16]; /* UUID */
1388 void *s_fs_info
; /* Filesystem private info */
1389 unsigned int s_max_links
;
1392 /* Granularity of c/m/atime in ns.
1393 Cannot be worse than a second */
1397 * The next field is for VFS *only*. No filesystems have any business
1398 * even looking at it. You had been warned.
1400 struct mutex s_vfs_rename_mutex
; /* Kludge */
1403 * Filesystem subtype. If non-empty the filesystem type field
1404 * in /proc/mounts will be "type.subtype"
1409 * Saved mount options for lazy filesystems using
1410 * generic_show_options()
1412 char __rcu
*s_options
;
1413 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1416 * Saved pool identifier for cleancache (-1 means none)
1418 int cleancache_poolid
;
1420 struct shrinker s_shrink
; /* per-sb shrinker handle */
1422 /* Number of inodes with nlink == 0 but still referenced */
1423 atomic_long_t s_remove_count
;
1425 /* Being remounted read-only */
1426 int s_readonly_remount
;
1428 /* AIO completions deferred from interrupt context */
1429 struct workqueue_struct
*s_dio_done_wq
;
1430 struct hlist_head s_pins
;
1433 * Keep the lru lists last in the structure so they always sit on their
1434 * own individual cachelines.
1436 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp
;
1437 struct list_lru s_inode_lru ____cacheline_aligned_in_smp
;
1438 struct rcu_head rcu
;
1439 struct work_struct destroy_work
;
1441 struct mutex s_sync_lock
; /* sync serialisation lock */
1444 * Indicates how deep in a filesystem stack this SB is
1448 /* s_inode_list_lock protects s_inodes */
1449 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp
;
1450 struct list_head s_inodes
; /* all inodes */
1453 extern struct timespec
current_fs_time(struct super_block
*sb
);
1456 * Snapshotting support.
1459 void __sb_end_write(struct super_block
*sb
, int level
);
1460 int __sb_start_write(struct super_block
*sb
, int level
, bool wait
);
1462 #define __sb_writers_acquired(sb, lev) \
1463 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1464 #define __sb_writers_release(sb, lev) \
1465 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1468 * sb_end_write - drop write access to a superblock
1469 * @sb: the super we wrote to
1471 * Decrement number of writers to the filesystem. Wake up possible waiters
1472 * wanting to freeze the filesystem.
1474 static inline void sb_end_write(struct super_block
*sb
)
1476 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1480 * sb_end_pagefault - drop write access to a superblock from a page fault
1481 * @sb: the super we wrote to
1483 * Decrement number of processes handling write page fault to the filesystem.
1484 * Wake up possible waiters wanting to freeze the filesystem.
1486 static inline void sb_end_pagefault(struct super_block
*sb
)
1488 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1492 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1493 * @sb: the super we wrote to
1495 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1496 * waiters wanting to freeze the filesystem.
1498 static inline void sb_end_intwrite(struct super_block
*sb
)
1500 __sb_end_write(sb
, SB_FREEZE_FS
);
1504 * sb_start_write - get write access to a superblock
1505 * @sb: the super we write to
1507 * When a process wants to write data or metadata to a file system (i.e. dirty
1508 * a page or an inode), it should embed the operation in a sb_start_write() -
1509 * sb_end_write() pair to get exclusion against file system freezing. This
1510 * function increments number of writers preventing freezing. If the file
1511 * system is already frozen, the function waits until the file system is
1514 * Since freeze protection behaves as a lock, users have to preserve
1515 * ordering of freeze protection and other filesystem locks. Generally,
1516 * freeze protection should be the outermost lock. In particular, we have:
1519 * -> i_mutex (write path, truncate, directory ops, ...)
1520 * -> s_umount (freeze_super, thaw_super)
1522 static inline void sb_start_write(struct super_block
*sb
)
1524 __sb_start_write(sb
, SB_FREEZE_WRITE
, true);
1527 static inline int sb_start_write_trylock(struct super_block
*sb
)
1529 return __sb_start_write(sb
, SB_FREEZE_WRITE
, false);
1533 * sb_start_pagefault - get write access to a superblock from a page fault
1534 * @sb: the super we write to
1536 * When a process starts handling write page fault, it should embed the
1537 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1538 * exclusion against file system freezing. This is needed since the page fault
1539 * is going to dirty a page. This function increments number of running page
1540 * faults preventing freezing. If the file system is already frozen, the
1541 * function waits until the file system is thawed.
1543 * Since page fault freeze protection behaves as a lock, users have to preserve
1544 * ordering of freeze protection and other filesystem locks. It is advised to
1545 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1546 * handling code implies lock dependency:
1549 * -> sb_start_pagefault
1551 static inline void sb_start_pagefault(struct super_block
*sb
)
1553 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
, true);
1557 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1558 * @sb: the super we write to
1560 * This is the third level of protection against filesystem freezing. It is
1561 * free for use by a filesystem. The only requirement is that it must rank
1562 * below sb_start_pagefault.
1564 * For example filesystem can call sb_start_intwrite() when starting a
1565 * transaction which somewhat eases handling of freezing for internal sources
1566 * of filesystem changes (internal fs threads, discarding preallocation on file
1569 static inline void sb_start_intwrite(struct super_block
*sb
)
1571 __sb_start_write(sb
, SB_FREEZE_FS
, true);
1575 extern bool inode_owner_or_capable(const struct inode
*inode
);
1578 * VFS helper functions..
1580 extern int vfs_create(struct inode
*, struct dentry
*, umode_t
, bool);
1581 extern int vfs_mkdir(struct inode
*, struct dentry
*, umode_t
);
1582 extern int vfs_mknod(struct inode
*, struct dentry
*, umode_t
, dev_t
);
1583 extern int vfs_symlink(struct inode
*, struct dentry
*, const char *);
1584 extern int vfs_link(struct dentry
*, struct inode
*, struct dentry
*, struct inode
**);
1585 extern int vfs_rmdir(struct inode
*, struct dentry
*);
1586 extern int vfs_unlink(struct inode
*, struct dentry
*, struct inode
**);
1587 extern int vfs_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*, struct inode
**, unsigned int);
1588 extern int vfs_whiteout(struct inode
*, struct dentry
*);
1591 * VFS file helper functions.
1593 extern void inode_init_owner(struct inode
*inode
, const struct inode
*dir
,
1596 * VFS FS_IOC_FIEMAP helper definitions.
1598 struct fiemap_extent_info
{
1599 unsigned int fi_flags
; /* Flags as passed from user */
1600 unsigned int fi_extents_mapped
; /* Number of mapped extents */
1601 unsigned int fi_extents_max
; /* Size of fiemap_extent array */
1602 struct fiemap_extent __user
*fi_extents_start
; /* Start of
1603 fiemap_extent array */
1605 int fiemap_fill_next_extent(struct fiemap_extent_info
*info
, u64 logical
,
1606 u64 phys
, u64 len
, u32 flags
);
1607 int fiemap_check_flags(struct fiemap_extent_info
*fieinfo
, u32 fs_flags
);
1612 * NOTE! These match bits 12..15 of stat.st_mode
1613 * (ie "(i_mode >> 12) & 15").
1615 #define DT_UNKNOWN 0
1626 * This is the "filldir" function type, used by readdir() to let
1627 * the kernel specify what kind of dirent layout it wants to have.
1628 * This allows the kernel to read directories into kernel space or
1629 * to have different dirent layouts depending on the binary type.
1632 typedef int (*filldir_t
)(struct dir_context
*, const char *, int, loff_t
, u64
,
1635 struct dir_context
{
1636 const filldir_t actor
;
1640 struct block_device_operations
;
1642 /* These macros are for out of kernel modules to test that
1643 * the kernel supports the unlocked_ioctl and compat_ioctl
1644 * fields in struct file_operations. */
1645 #define HAVE_COMPAT_IOCTL 1
1646 #define HAVE_UNLOCKED_IOCTL 1
1649 * These flags let !MMU mmap() govern direct device mapping vs immediate
1650 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1652 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1653 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1654 * NOMMU_MAP_READ: Can be mapped for reading
1655 * NOMMU_MAP_WRITE: Can be mapped for writing
1656 * NOMMU_MAP_EXEC: Can be mapped for execution
1658 #define NOMMU_MAP_COPY 0x00000001
1659 #define NOMMU_MAP_DIRECT 0x00000008
1660 #define NOMMU_MAP_READ VM_MAYREAD
1661 #define NOMMU_MAP_WRITE VM_MAYWRITE
1662 #define NOMMU_MAP_EXEC VM_MAYEXEC
1664 #define NOMMU_VMFLAGS \
1665 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1670 struct file_operations
{
1671 struct module
*owner
;
1672 loff_t (*llseek
) (struct file
*, loff_t
, int);
1673 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
1674 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
1675 ssize_t (*read_iter
) (struct kiocb
*, struct iov_iter
*);
1676 ssize_t (*write_iter
) (struct kiocb
*, struct iov_iter
*);
1677 int (*iterate
) (struct file
*, struct dir_context
*);
1678 int (*iterate_shared
) (struct file
*, struct dir_context
*);
1679 unsigned int (*poll
) (struct file
*, struct poll_table_struct
*);
1680 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
1681 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
1682 int (*mmap
) (struct file
*, struct vm_area_struct
*);
1683 int (*open
) (struct inode
*, struct file
*);
1684 int (*flush
) (struct file
*, fl_owner_t id
);
1685 int (*release
) (struct inode
*, struct file
*);
1686 int (*fsync
) (struct file
*, loff_t
, loff_t
, int datasync
);
1687 int (*aio_fsync
) (struct kiocb
*, int datasync
);
1688 int (*fasync
) (int, struct file
*, int);
1689 int (*lock
) (struct file
*, int, struct file_lock
*);
1690 ssize_t (*sendpage
) (struct file
*, struct page
*, int, size_t, loff_t
*, int);
1691 unsigned long (*get_unmapped_area
)(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1692 int (*check_flags
)(int);
1693 int (*flock
) (struct file
*, int, struct file_lock
*);
1694 ssize_t (*splice_write
)(struct pipe_inode_info
*, struct file
*, loff_t
*, size_t, unsigned int);
1695 ssize_t (*splice_read
)(struct file
*, loff_t
*, struct pipe_inode_info
*, size_t, unsigned int);
1696 int (*setlease
)(struct file
*, long, struct file_lock
**, void **);
1697 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
1699 void (*show_fdinfo
)(struct seq_file
*m
, struct file
*f
);
1701 unsigned (*mmap_capabilities
)(struct file
*);
1703 ssize_t (*copy_file_range
)(struct file
*, loff_t
, struct file
*,
1704 loff_t
, size_t, unsigned int);
1705 int (*clone_file_range
)(struct file
*, loff_t
, struct file
*, loff_t
,
1707 ssize_t (*dedupe_file_range
)(struct file
*, u64
, u64
, struct file
*,
1711 struct inode_operations
{
1712 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*, unsigned int);
1713 const char * (*get_link
) (struct dentry
*, struct inode
*, struct delayed_call
*);
1714 int (*permission
) (struct inode
*, int);
1715 struct posix_acl
* (*get_acl
)(struct inode
*, int);
1717 int (*readlink
) (struct dentry
*, char __user
*,int);
1719 int (*create
) (struct inode
*,struct dentry
*, umode_t
, bool);
1720 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
1721 int (*unlink
) (struct inode
*,struct dentry
*);
1722 int (*symlink
) (struct inode
*,struct dentry
*,const char *);
1723 int (*mkdir
) (struct inode
*,struct dentry
*,umode_t
);
1724 int (*rmdir
) (struct inode
*,struct dentry
*);
1725 int (*mknod
) (struct inode
*,struct dentry
*,umode_t
,dev_t
);
1726 int (*rename
) (struct inode
*, struct dentry
*,
1727 struct inode
*, struct dentry
*);
1728 int (*rename2
) (struct inode
*, struct dentry
*,
1729 struct inode
*, struct dentry
*, unsigned int);
1730 int (*setattr
) (struct dentry
*, struct iattr
*);
1731 int (*getattr
) (struct vfsmount
*mnt
, struct dentry
*, struct kstat
*);
1732 int (*setxattr
) (struct dentry
*, struct inode
*,
1733 const char *, const void *, size_t, int);
1734 ssize_t (*getxattr
) (struct dentry
*, struct inode
*,
1735 const char *, void *, size_t);
1736 ssize_t (*listxattr
) (struct dentry
*, char *, size_t);
1737 int (*removexattr
) (struct dentry
*, const char *);
1738 int (*fiemap
)(struct inode
*, struct fiemap_extent_info
*, u64 start
,
1740 int (*update_time
)(struct inode
*, struct timespec
*, int);
1741 int (*atomic_open
)(struct inode
*, struct dentry
*,
1742 struct file
*, unsigned open_flag
,
1743 umode_t create_mode
, int *opened
);
1744 int (*tmpfile
) (struct inode
*, struct dentry
*, umode_t
);
1745 int (*set_acl
)(struct inode
*, struct posix_acl
*, int);
1746 } ____cacheline_aligned
;
1748 ssize_t
rw_copy_check_uvector(int type
, const struct iovec __user
* uvector
,
1749 unsigned long nr_segs
, unsigned long fast_segs
,
1750 struct iovec
*fast_pointer
,
1751 struct iovec
**ret_pointer
);
1753 extern ssize_t
__vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1754 extern ssize_t
__vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1755 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1756 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1757 extern ssize_t
vfs_readv(struct file
*, const struct iovec __user
*,
1758 unsigned long, loff_t
*, int);
1759 extern ssize_t
vfs_writev(struct file
*, const struct iovec __user
*,
1760 unsigned long, loff_t
*, int);
1761 extern ssize_t
vfs_copy_file_range(struct file
*, loff_t
, struct file
*,
1762 loff_t
, size_t, unsigned int);
1763 extern int vfs_clone_file_range(struct file
*file_in
, loff_t pos_in
,
1764 struct file
*file_out
, loff_t pos_out
, u64 len
);
1765 extern int vfs_dedupe_file_range(struct file
*file
,
1766 struct file_dedupe_range
*same
);
1768 struct super_operations
{
1769 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
1770 void (*destroy_inode
)(struct inode
*);
1772 void (*dirty_inode
) (struct inode
*, int flags
);
1773 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
1774 int (*drop_inode
) (struct inode
*);
1775 void (*evict_inode
) (struct inode
*);
1776 void (*put_super
) (struct super_block
*);
1777 int (*sync_fs
)(struct super_block
*sb
, int wait
);
1778 int (*freeze_super
) (struct super_block
*);
1779 int (*freeze_fs
) (struct super_block
*);
1780 int (*thaw_super
) (struct super_block
*);
1781 int (*unfreeze_fs
) (struct super_block
*);
1782 int (*statfs
) (struct dentry
*, struct kstatfs
*);
1783 int (*remount_fs
) (struct super_block
*, int *, char *);
1784 void (*umount_begin
) (struct super_block
*);
1786 int (*show_options
)(struct seq_file
*, struct dentry
*);
1787 int (*show_devname
)(struct seq_file
*, struct dentry
*);
1788 int (*show_path
)(struct seq_file
*, struct dentry
*);
1789 int (*show_stats
)(struct seq_file
*, struct dentry
*);
1791 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
1792 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
1793 struct dquot
**(*get_dquots
)(struct inode
*);
1795 int (*bdev_try_to_free_page
)(struct super_block
*, struct page
*, gfp_t
);
1796 long (*nr_cached_objects
)(struct super_block
*,
1797 struct shrink_control
*);
1798 long (*free_cached_objects
)(struct super_block
*,
1799 struct shrink_control
*);
1803 * Inode flags - they have no relation to superblock flags now
1805 #define S_SYNC 1 /* Writes are synced at once */
1806 #define S_NOATIME 2 /* Do not update access times */
1807 #define S_APPEND 4 /* Append-only file */
1808 #define S_IMMUTABLE 8 /* Immutable file */
1809 #define S_DEAD 16 /* removed, but still open directory */
1810 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1811 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1812 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1813 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1814 #define S_PRIVATE 512 /* Inode is fs-internal */
1815 #define S_IMA 1024 /* Inode has an associated IMA struct */
1816 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1817 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1818 #ifdef CONFIG_FS_DAX
1819 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1821 #define S_DAX 0 /* Make all the DAX code disappear */
1825 * Note that nosuid etc flags are inode-specific: setting some file-system
1826 * flags just means all the inodes inherit those flags by default. It might be
1827 * possible to override it selectively if you really wanted to with some
1828 * ioctl() that is not currently implemented.
1830 * Exception: MS_RDONLY is always applied to the entire file system.
1832 * Unfortunately, it is possible to change a filesystems flags with it mounted
1833 * with files in use. This means that all of the inodes will not have their
1834 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1835 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1837 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1839 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1840 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1841 ((inode)->i_flags & S_SYNC))
1842 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1843 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1844 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1845 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1846 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1848 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1849 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1850 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1851 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1853 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1854 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1855 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1856 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1857 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1858 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1859 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1860 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1862 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1863 (inode)->i_rdev == WHITEOUT_DEV)
1866 * Inode state bits. Protected by inode->i_lock
1868 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1869 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1871 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1872 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1873 * various stages of removing an inode.
1875 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1877 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1878 * fdatasync(). i_atime is the usual cause.
1879 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1880 * these changes separately from I_DIRTY_SYNC so that we
1881 * don't have to write inode on fdatasync() when only
1882 * mtime has changed in it.
1883 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1884 * I_NEW Serves as both a mutex and completion notification.
1885 * New inodes set I_NEW. If two processes both create
1886 * the same inode, one of them will release its inode and
1887 * wait for I_NEW to be released before returning.
1888 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1889 * also cause waiting on I_NEW, without I_NEW actually
1890 * being set. find_inode() uses this to prevent returning
1891 * nearly-dead inodes.
1892 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1893 * is zero. I_FREEING must be set when I_WILL_FREE is
1895 * I_FREEING Set when inode is about to be freed but still has dirty
1896 * pages or buffers attached or the inode itself is still
1898 * I_CLEAR Added by clear_inode(). In this state the inode is
1899 * clean and can be destroyed. Inode keeps I_FREEING.
1901 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1902 * prohibited for many purposes. iget() must wait for
1903 * the inode to be completely released, then create it
1904 * anew. Other functions will just ignore such inodes,
1905 * if appropriate. I_NEW is used for waiting.
1907 * I_SYNC Writeback of inode is running. The bit is set during
1908 * data writeback, and cleared with a wakeup on the bit
1909 * address once it is done. The bit is also used to pin
1910 * the inode in memory for flusher thread.
1912 * I_REFERENCED Marks the inode as recently references on the LRU list.
1914 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1916 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1917 * synchronize competing switching instances and to tell
1918 * wb stat updates to grab mapping->tree_lock. See
1919 * inode_switch_wb_work_fn() for details.
1921 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1923 #define I_DIRTY_SYNC (1 << 0)
1924 #define I_DIRTY_DATASYNC (1 << 1)
1925 #define I_DIRTY_PAGES (1 << 2)
1927 #define I_NEW (1 << __I_NEW)
1928 #define I_WILL_FREE (1 << 4)
1929 #define I_FREEING (1 << 5)
1930 #define I_CLEAR (1 << 6)
1932 #define I_SYNC (1 << __I_SYNC)
1933 #define I_REFERENCED (1 << 8)
1934 #define __I_DIO_WAKEUP 9
1935 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1936 #define I_LINKABLE (1 << 10)
1937 #define I_DIRTY_TIME (1 << 11)
1938 #define __I_DIRTY_TIME_EXPIRED 12
1939 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1940 #define I_WB_SWITCH (1 << 13)
1942 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1943 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1945 extern void __mark_inode_dirty(struct inode
*, int);
1946 static inline void mark_inode_dirty(struct inode
*inode
)
1948 __mark_inode_dirty(inode
, I_DIRTY
);
1951 static inline void mark_inode_dirty_sync(struct inode
*inode
)
1953 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
1956 extern void inc_nlink(struct inode
*inode
);
1957 extern void drop_nlink(struct inode
*inode
);
1958 extern void clear_nlink(struct inode
*inode
);
1959 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
1961 static inline void inode_inc_link_count(struct inode
*inode
)
1964 mark_inode_dirty(inode
);
1967 static inline void inode_dec_link_count(struct inode
*inode
)
1970 mark_inode_dirty(inode
);
1974 * inode_inc_iversion - increments i_version
1975 * @inode: inode that need to be updated
1977 * Every time the inode is modified, the i_version field will be incremented.
1978 * The filesystem has to be mounted with i_version flag
1981 static inline void inode_inc_iversion(struct inode
*inode
)
1983 spin_lock(&inode
->i_lock
);
1985 spin_unlock(&inode
->i_lock
);
1988 enum file_time_flags
{
1995 extern bool atime_needs_update(const struct path
*, struct inode
*);
1996 extern void touch_atime(const struct path
*);
1997 static inline void file_accessed(struct file
*file
)
1999 if (!(file
->f_flags
& O_NOATIME
))
2000 touch_atime(&file
->f_path
);
2003 int sync_inode(struct inode
*inode
, struct writeback_control
*wbc
);
2004 int sync_inode_metadata(struct inode
*inode
, int wait
);
2006 struct file_system_type
{
2009 #define FS_REQUIRES_DEV 1
2010 #define FS_BINARY_MOUNTDATA 2
2011 #define FS_HAS_SUBTYPE 4
2012 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2013 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
2014 #define FS_USERNS_VISIBLE 32 /* FS must already be visible */
2015 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2016 struct dentry
*(*mount
) (struct file_system_type
*, int,
2017 const char *, void *);
2018 void (*kill_sb
) (struct super_block
*);
2019 struct module
*owner
;
2020 struct file_system_type
* next
;
2021 struct hlist_head fs_supers
;
2023 struct lock_class_key s_lock_key
;
2024 struct lock_class_key s_umount_key
;
2025 struct lock_class_key s_vfs_rename_key
;
2026 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
2028 struct lock_class_key i_lock_key
;
2029 struct lock_class_key i_mutex_key
;
2030 struct lock_class_key i_mutex_dir_key
;
2033 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2035 extern struct dentry
*mount_ns(struct file_system_type
*fs_type
, int flags
,
2036 void *data
, int (*fill_super
)(struct super_block
*, void *, int));
2037 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
2038 int flags
, const char *dev_name
, void *data
,
2039 int (*fill_super
)(struct super_block
*, void *, int));
2040 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
2041 int flags
, void *data
,
2042 int (*fill_super
)(struct super_block
*, void *, int));
2043 extern struct dentry
*mount_nodev(struct file_system_type
*fs_type
,
2044 int flags
, void *data
,
2045 int (*fill_super
)(struct super_block
*, void *, int));
2046 extern struct dentry
*mount_subtree(struct vfsmount
*mnt
, const char *path
);
2047 void generic_shutdown_super(struct super_block
*sb
);
2048 void kill_block_super(struct super_block
*sb
);
2049 void kill_anon_super(struct super_block
*sb
);
2050 void kill_litter_super(struct super_block
*sb
);
2051 void deactivate_super(struct super_block
*sb
);
2052 void deactivate_locked_super(struct super_block
*sb
);
2053 int set_anon_super(struct super_block
*s
, void *data
);
2054 int get_anon_bdev(dev_t
*);
2055 void free_anon_bdev(dev_t
);
2056 struct super_block
*sget(struct file_system_type
*type
,
2057 int (*test
)(struct super_block
*,void *),
2058 int (*set
)(struct super_block
*,void *),
2059 int flags
, void *data
);
2060 extern struct dentry
*mount_pseudo(struct file_system_type
*, char *,
2061 const struct super_operations
*ops
,
2062 const struct dentry_operations
*dops
,
2065 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2066 #define fops_get(fops) \
2067 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2068 #define fops_put(fops) \
2069 do { if (fops) module_put((fops)->owner); } while(0)
2071 * This one is to be used *ONLY* from ->open() instances.
2072 * fops must be non-NULL, pinned down *and* module dependencies
2073 * should be sufficient to pin the caller down as well.
2075 #define replace_fops(f, fops) \
2077 struct file *__file = (f); \
2078 fops_put(__file->f_op); \
2079 BUG_ON(!(__file->f_op = (fops))); \
2082 extern int register_filesystem(struct file_system_type
*);
2083 extern int unregister_filesystem(struct file_system_type
*);
2084 extern struct vfsmount
*kern_mount_data(struct file_system_type
*, void *data
);
2085 #define kern_mount(type) kern_mount_data(type, NULL)
2086 extern void kern_unmount(struct vfsmount
*mnt
);
2087 extern int may_umount_tree(struct vfsmount
*);
2088 extern int may_umount(struct vfsmount
*);
2089 extern long do_mount(const char *, const char __user
*,
2090 const char *, unsigned long, void *);
2091 extern struct vfsmount
*collect_mounts(struct path
*);
2092 extern void drop_collected_mounts(struct vfsmount
*);
2093 extern int iterate_mounts(int (*)(struct vfsmount
*, void *), void *,
2095 extern int vfs_statfs(struct path
*, struct kstatfs
*);
2096 extern int user_statfs(const char __user
*, struct kstatfs
*);
2097 extern int fd_statfs(int, struct kstatfs
*);
2098 extern int vfs_ustat(dev_t
, struct kstatfs
*);
2099 extern int freeze_super(struct super_block
*super
);
2100 extern int thaw_super(struct super_block
*super
);
2101 extern bool our_mnt(struct vfsmount
*mnt
);
2103 extern int current_umask(void);
2105 extern void ihold(struct inode
* inode
);
2106 extern void iput(struct inode
*);
2107 extern int generic_update_time(struct inode
*, struct timespec
*, int);
2110 extern struct kobject
*fs_kobj
;
2112 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2114 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2115 extern int locks_mandatory_locked(struct file
*);
2116 extern int locks_mandatory_area(struct inode
*, struct file
*, loff_t
, loff_t
, unsigned char);
2119 * Candidates for mandatory locking have the setgid bit set
2120 * but no group execute bit - an otherwise meaningless combination.
2123 static inline int __mandatory_lock(struct inode
*ino
)
2125 return (ino
->i_mode
& (S_ISGID
| S_IXGRP
)) == S_ISGID
;
2129 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2130 * otherwise these will be advisory locks
2133 static inline int mandatory_lock(struct inode
*ino
)
2135 return IS_MANDLOCK(ino
) && __mandatory_lock(ino
);
2138 static inline int locks_verify_locked(struct file
*file
)
2140 if (mandatory_lock(file_inode(file
)))
2141 return locks_mandatory_locked(file
);
2145 static inline int locks_verify_truncate(struct inode
*inode
,
2149 if (!inode
->i_flctx
|| !mandatory_lock(inode
))
2152 if (size
< inode
->i_size
) {
2153 return locks_mandatory_area(inode
, f
, size
, inode
->i_size
- 1,
2156 return locks_mandatory_area(inode
, f
, inode
->i_size
, size
- 1,
2161 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2163 static inline int locks_mandatory_locked(struct file
*file
)
2168 static inline int locks_mandatory_area(struct inode
*inode
, struct file
*filp
,
2169 loff_t start
, loff_t end
, unsigned char type
)
2174 static inline int __mandatory_lock(struct inode
*inode
)
2179 static inline int mandatory_lock(struct inode
*inode
)
2184 static inline int locks_verify_locked(struct file
*file
)
2189 static inline int locks_verify_truncate(struct inode
*inode
, struct file
*filp
,
2195 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2198 #ifdef CONFIG_FILE_LOCKING
2199 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2202 * Since this check is lockless, we must ensure that any refcounts
2203 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2204 * could end up racing with tasks trying to set a new lease on this
2208 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2209 return __break_lease(inode
, mode
, FL_LEASE
);
2213 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2216 * Since this check is lockless, we must ensure that any refcounts
2217 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2218 * could end up racing with tasks trying to set a new lease on this
2222 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2223 return __break_lease(inode
, mode
, FL_DELEG
);
2227 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2231 ret
= break_deleg(inode
, O_WRONLY
|O_NONBLOCK
);
2232 if (ret
== -EWOULDBLOCK
&& delegated_inode
) {
2233 *delegated_inode
= inode
;
2239 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2243 ret
= break_deleg(*delegated_inode
, O_WRONLY
);
2244 iput(*delegated_inode
);
2245 *delegated_inode
= NULL
;
2249 static inline int break_layout(struct inode
*inode
, bool wait
)
2252 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2253 return __break_lease(inode
,
2254 wait
? O_WRONLY
: O_WRONLY
| O_NONBLOCK
,
2259 #else /* !CONFIG_FILE_LOCKING */
2260 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2265 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2270 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2275 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2281 static inline int break_layout(struct inode
*inode
, bool wait
)
2286 #endif /* CONFIG_FILE_LOCKING */
2291 const char *name
; /* pointer to actual string */
2292 const __user
char *uptr
; /* original userland pointer */
2293 struct audit_names
*aname
;
2298 extern long vfs_truncate(const struct path
*, loff_t
);
2299 extern int do_truncate(struct dentry
*, loff_t start
, unsigned int time_attrs
,
2301 extern int vfs_fallocate(struct file
*file
, int mode
, loff_t offset
,
2303 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
2305 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
2306 extern struct file
*filp_open(const char *, int, umode_t
);
2307 extern struct file
*file_open_root(struct dentry
*, struct vfsmount
*,
2308 const char *, int, umode_t
);
2309 extern struct file
* dentry_open(const struct path
*, int, const struct cred
*);
2310 extern int filp_close(struct file
*, fl_owner_t id
);
2312 extern struct filename
*getname_flags(const char __user
*, int, int *);
2313 extern struct filename
*getname(const char __user
*);
2314 extern struct filename
*getname_kernel(const char *);
2315 extern void putname(struct filename
*name
);
2321 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
2322 int (*open
)(struct inode
*, struct file
*),
2324 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
2328 extern int ioctl_preallocate(struct file
*filp
, void __user
*argp
);
2331 extern void __init
vfs_caches_init_early(void);
2332 extern void __init
vfs_caches_init(void);
2334 extern struct kmem_cache
*names_cachep
;
2336 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2337 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2340 extern int register_blkdev(unsigned int, const char *);
2341 extern void unregister_blkdev(unsigned int, const char *);
2342 extern struct block_device
*bdget(dev_t
);
2343 extern struct block_device
*bdgrab(struct block_device
*bdev
);
2344 extern void bd_set_size(struct block_device
*, loff_t size
);
2345 extern void bd_forget(struct inode
*inode
);
2346 extern void bdput(struct block_device
*);
2347 extern void invalidate_bdev(struct block_device
*);
2348 extern void iterate_bdevs(void (*)(struct block_device
*, void *), void *);
2349 extern int sync_blockdev(struct block_device
*bdev
);
2350 extern void kill_bdev(struct block_device
*);
2351 extern struct super_block
*freeze_bdev(struct block_device
*);
2352 extern void emergency_thaw_all(void);
2353 extern int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
);
2354 extern int fsync_bdev(struct block_device
*);
2356 extern struct super_block
*blockdev_superblock
;
2358 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2360 return sb
== blockdev_superblock
;
2363 static inline void bd_forget(struct inode
*inode
) {}
2364 static inline int sync_blockdev(struct block_device
*bdev
) { return 0; }
2365 static inline void kill_bdev(struct block_device
*bdev
) {}
2366 static inline void invalidate_bdev(struct block_device
*bdev
) {}
2368 static inline struct super_block
*freeze_bdev(struct block_device
*sb
)
2373 static inline int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
)
2378 static inline void iterate_bdevs(void (*f
)(struct block_device
*, void *), void *arg
)
2382 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2387 extern int sync_filesystem(struct super_block
*);
2388 extern const struct file_operations def_blk_fops
;
2389 extern const struct file_operations def_chr_fops
;
2391 extern int ioctl_by_bdev(struct block_device
*, unsigned, unsigned long);
2392 extern int blkdev_ioctl(struct block_device
*, fmode_t
, unsigned, unsigned long);
2393 extern long compat_blkdev_ioctl(struct file
*, unsigned, unsigned long);
2394 extern int blkdev_get(struct block_device
*bdev
, fmode_t mode
, void *holder
);
2395 extern struct block_device
*blkdev_get_by_path(const char *path
, fmode_t mode
,
2397 extern struct block_device
*blkdev_get_by_dev(dev_t dev
, fmode_t mode
,
2399 extern void blkdev_put(struct block_device
*bdev
, fmode_t mode
);
2400 extern int __blkdev_reread_part(struct block_device
*bdev
);
2401 extern int blkdev_reread_part(struct block_device
*bdev
);
2404 extern int bd_link_disk_holder(struct block_device
*bdev
, struct gendisk
*disk
);
2405 extern void bd_unlink_disk_holder(struct block_device
*bdev
,
2406 struct gendisk
*disk
);
2408 static inline int bd_link_disk_holder(struct block_device
*bdev
,
2409 struct gendisk
*disk
)
2413 static inline void bd_unlink_disk_holder(struct block_device
*bdev
,
2414 struct gendisk
*disk
)
2421 #define CHRDEV_MAJOR_HASH_SIZE 255
2422 /* Marks the bottom of the first segment of free char majors */
2423 #define CHRDEV_MAJOR_DYN_END 234
2424 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
2425 extern int register_chrdev_region(dev_t
, unsigned, const char *);
2426 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
2427 unsigned int count
, const char *name
,
2428 const struct file_operations
*fops
);
2429 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
2430 unsigned int count
, const char *name
);
2431 extern void unregister_chrdev_region(dev_t
, unsigned);
2432 extern void chrdev_show(struct seq_file
*,off_t
);
2434 static inline int register_chrdev(unsigned int major
, const char *name
,
2435 const struct file_operations
*fops
)
2437 return __register_chrdev(major
, 0, 256, name
, fops
);
2440 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2442 __unregister_chrdev(major
, 0, 256, name
);
2445 /* fs/block_dev.c */
2446 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2447 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2450 #define BLKDEV_MAJOR_HASH_SIZE 255
2451 extern const char *__bdevname(dev_t
, char *buffer
);
2452 extern const char *bdevname(struct block_device
*bdev
, char *buffer
);
2453 extern struct block_device
*lookup_bdev(const char *);
2454 extern void blkdev_show(struct seq_file
*,off_t
);
2457 #define BLKDEV_MAJOR_HASH_SIZE 0
2460 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2462 /* Invalid inode operations -- fs/bad_inode.c */
2463 extern void make_bad_inode(struct inode
*);
2464 extern bool is_bad_inode(struct inode
*);
2468 * return READ, READA, or WRITE
2470 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2473 * return data direction, READ or WRITE
2475 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2477 extern void check_disk_size_change(struct gendisk
*disk
,
2478 struct block_device
*bdev
);
2479 extern int revalidate_disk(struct gendisk
*);
2480 extern int check_disk_change(struct block_device
*);
2481 extern int __invalidate_device(struct block_device
*, bool);
2482 extern int invalidate_partition(struct gendisk
*, int);
2484 unsigned long invalidate_mapping_pages(struct address_space
*mapping
,
2485 pgoff_t start
, pgoff_t end
);
2487 static inline void invalidate_remote_inode(struct inode
*inode
)
2489 if (S_ISREG(inode
->i_mode
) || S_ISDIR(inode
->i_mode
) ||
2490 S_ISLNK(inode
->i_mode
))
2491 invalidate_mapping_pages(inode
->i_mapping
, 0, -1);
2493 extern int invalidate_inode_pages2(struct address_space
*mapping
);
2494 extern int invalidate_inode_pages2_range(struct address_space
*mapping
,
2495 pgoff_t start
, pgoff_t end
);
2496 extern int write_inode_now(struct inode
*, int);
2497 extern int filemap_fdatawrite(struct address_space
*);
2498 extern int filemap_flush(struct address_space
*);
2499 extern int filemap_fdatawait(struct address_space
*);
2500 extern void filemap_fdatawait_keep_errors(struct address_space
*);
2501 extern int filemap_fdatawait_range(struct address_space
*, loff_t lstart
,
2503 extern int filemap_write_and_wait(struct address_space
*mapping
);
2504 extern int filemap_write_and_wait_range(struct address_space
*mapping
,
2505 loff_t lstart
, loff_t lend
);
2506 extern int __filemap_fdatawrite_range(struct address_space
*mapping
,
2507 loff_t start
, loff_t end
, int sync_mode
);
2508 extern int filemap_fdatawrite_range(struct address_space
*mapping
,
2509 loff_t start
, loff_t end
);
2511 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
2513 extern int vfs_fsync(struct file
*file
, int datasync
);
2516 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2517 * to already be updated for the write, and will return either the amount
2518 * of bytes passed in, or an error if syncing the file failed.
2520 static inline ssize_t
generic_write_sync(struct kiocb
*iocb
, ssize_t count
)
2522 if (iocb
->ki_flags
& IOCB_DSYNC
) {
2523 int ret
= vfs_fsync_range(iocb
->ki_filp
,
2524 iocb
->ki_pos
- count
, iocb
->ki_pos
- 1,
2525 (iocb
->ki_flags
& IOCB_SYNC
) ? 0 : 1);
2533 extern void emergency_sync(void);
2534 extern void emergency_remount(void);
2536 extern sector_t
bmap(struct inode
*, sector_t
);
2538 extern int notify_change(struct dentry
*, struct iattr
*, struct inode
**);
2539 extern int inode_permission(struct inode
*, int);
2540 extern int __inode_permission(struct inode
*, int);
2541 extern int generic_permission(struct inode
*, int);
2542 extern int __check_sticky(struct inode
*dir
, struct inode
*inode
);
2544 static inline bool execute_ok(struct inode
*inode
)
2546 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
2549 static inline void file_start_write(struct file
*file
)
2551 if (!S_ISREG(file_inode(file
)->i_mode
))
2553 __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, true);
2556 static inline bool file_start_write_trylock(struct file
*file
)
2558 if (!S_ISREG(file_inode(file
)->i_mode
))
2560 return __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, false);
2563 static inline void file_end_write(struct file
*file
)
2565 if (!S_ISREG(file_inode(file
)->i_mode
))
2567 __sb_end_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
);
2571 * get_write_access() gets write permission for a file.
2572 * put_write_access() releases this write permission.
2573 * This is used for regular files.
2574 * We cannot support write (and maybe mmap read-write shared) accesses and
2575 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2576 * can have the following values:
2577 * 0: no writers, no VM_DENYWRITE mappings
2578 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2579 * > 0: (i_writecount) users are writing to the file.
2581 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2582 * except for the cases where we don't hold i_writecount yet. Then we need to
2583 * use {get,deny}_write_access() - these functions check the sign and refuse
2584 * to do the change if sign is wrong.
2586 static inline int get_write_access(struct inode
*inode
)
2588 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2590 static inline int deny_write_access(struct file
*file
)
2592 struct inode
*inode
= file_inode(file
);
2593 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2595 static inline void put_write_access(struct inode
* inode
)
2597 atomic_dec(&inode
->i_writecount
);
2599 static inline void allow_write_access(struct file
*file
)
2602 atomic_inc(&file_inode(file
)->i_writecount
);
2604 static inline bool inode_is_open_for_write(const struct inode
*inode
)
2606 return atomic_read(&inode
->i_writecount
) > 0;
2610 static inline void i_readcount_dec(struct inode
*inode
)
2612 BUG_ON(!atomic_read(&inode
->i_readcount
));
2613 atomic_dec(&inode
->i_readcount
);
2615 static inline void i_readcount_inc(struct inode
*inode
)
2617 atomic_inc(&inode
->i_readcount
);
2620 static inline void i_readcount_dec(struct inode
*inode
)
2624 static inline void i_readcount_inc(struct inode
*inode
)
2629 extern int do_pipe_flags(int *, int);
2631 #define __kernel_read_file_id(id) \
2632 id(UNKNOWN, unknown) \
2633 id(FIRMWARE, firmware) \
2634 id(MODULE, kernel-module) \
2635 id(KEXEC_IMAGE, kexec-image) \
2636 id(KEXEC_INITRAMFS, kexec-initramfs) \
2637 id(POLICY, security-policy) \
2640 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2641 #define __fid_stringify(dummy, str) #str,
2643 enum kernel_read_file_id
{
2644 __kernel_read_file_id(__fid_enumify
)
2647 static const char * const kernel_read_file_str
[] = {
2648 __kernel_read_file_id(__fid_stringify
)
2651 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id
)
2653 if (id
< 0 || id
>= READING_MAX_ID
)
2654 return kernel_read_file_str
[READING_UNKNOWN
];
2656 return kernel_read_file_str
[id
];
2659 extern int kernel_read(struct file
*, loff_t
, char *, unsigned long);
2660 extern int kernel_read_file(struct file
*, void **, loff_t
*, loff_t
,
2661 enum kernel_read_file_id
);
2662 extern int kernel_read_file_from_path(char *, void **, loff_t
*, loff_t
,
2663 enum kernel_read_file_id
);
2664 extern int kernel_read_file_from_fd(int, void **, loff_t
*, loff_t
,
2665 enum kernel_read_file_id
);
2666 extern ssize_t
kernel_write(struct file
*, const char *, size_t, loff_t
);
2667 extern ssize_t
__kernel_write(struct file
*, const char *, size_t, loff_t
*);
2668 extern struct file
* open_exec(const char *);
2670 /* fs/dcache.c -- generic fs support functions */
2671 extern bool is_subdir(struct dentry
*, struct dentry
*);
2672 extern bool path_is_under(struct path
*, struct path
*);
2674 extern char *file_path(struct file
*, char *, int);
2676 #include <linux/err.h>
2678 /* needed for stackable file system support */
2679 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int whence
);
2681 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int whence
);
2683 extern int inode_init_always(struct super_block
*, struct inode
*);
2684 extern void inode_init_once(struct inode
*);
2685 extern void address_space_init_once(struct address_space
*mapping
);
2686 extern struct inode
* igrab(struct inode
*);
2687 extern ino_t
iunique(struct super_block
*, ino_t
);
2688 extern int inode_needs_sync(struct inode
*inode
);
2689 extern int generic_delete_inode(struct inode
*inode
);
2690 static inline int generic_drop_inode(struct inode
*inode
)
2692 return !inode
->i_nlink
|| inode_unhashed(inode
);
2695 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
2696 unsigned long hashval
, int (*test
)(struct inode
*, void *),
2698 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
2699 int (*test
)(struct inode
*, void *), void *data
);
2700 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
2702 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
2703 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
2704 extern struct inode
*find_inode_nowait(struct super_block
*,
2706 int (*match
)(struct inode
*,
2707 unsigned long, void *),
2709 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
2710 extern int insert_inode_locked(struct inode
*);
2711 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2712 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
2714 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
2716 extern void unlock_new_inode(struct inode
*);
2717 extern unsigned int get_next_ino(void);
2719 extern void __iget(struct inode
* inode
);
2720 extern void iget_failed(struct inode
*);
2721 extern void clear_inode(struct inode
*);
2722 extern void __destroy_inode(struct inode
*);
2723 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
2724 extern struct inode
*new_inode(struct super_block
*sb
);
2725 extern void free_inode_nonrcu(struct inode
*inode
);
2726 extern int should_remove_suid(struct dentry
*);
2727 extern int file_remove_privs(struct file
*);
2728 extern int dentry_needs_remove_privs(struct dentry
*dentry
);
2729 static inline int file_needs_remove_privs(struct file
*file
)
2731 return dentry_needs_remove_privs(file
->f_path
.dentry
);
2734 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
2735 static inline void insert_inode_hash(struct inode
*inode
)
2737 __insert_inode_hash(inode
, inode
->i_ino
);
2740 extern void __remove_inode_hash(struct inode
*);
2741 static inline void remove_inode_hash(struct inode
*inode
)
2743 if (!inode_unhashed(inode
) && !hlist_fake(&inode
->i_hash
))
2744 __remove_inode_hash(inode
);
2747 extern void inode_sb_list_add(struct inode
*inode
);
2750 extern blk_qc_t
submit_bio(int, struct bio
*);
2751 extern int bdev_read_only(struct block_device
*);
2753 extern int set_blocksize(struct block_device
*, int);
2754 extern int sb_set_blocksize(struct super_block
*, int);
2755 extern int sb_min_blocksize(struct super_block
*, int);
2757 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
2758 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
2759 extern ssize_t
generic_write_checks(struct kiocb
*, struct iov_iter
*);
2760 extern ssize_t
generic_file_read_iter(struct kiocb
*, struct iov_iter
*);
2761 extern ssize_t
__generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2762 extern ssize_t
generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2763 extern ssize_t
generic_file_direct_write(struct kiocb
*, struct iov_iter
*);
2764 extern ssize_t
generic_perform_write(struct file
*, struct iov_iter
*, loff_t
);
2766 ssize_t
vfs_iter_read(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
);
2767 ssize_t
vfs_iter_write(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
);
2769 /* fs/block_dev.c */
2770 extern ssize_t
blkdev_read_iter(struct kiocb
*iocb
, struct iov_iter
*to
);
2771 extern ssize_t
blkdev_write_iter(struct kiocb
*iocb
, struct iov_iter
*from
);
2772 extern int blkdev_fsync(struct file
*filp
, loff_t start
, loff_t end
,
2774 extern void block_sync_page(struct page
*page
);
2777 extern ssize_t
generic_file_splice_read(struct file
*, loff_t
*,
2778 struct pipe_inode_info
*, size_t, unsigned int);
2779 extern ssize_t
default_file_splice_read(struct file
*, loff_t
*,
2780 struct pipe_inode_info
*, size_t, unsigned int);
2781 extern ssize_t
iter_file_splice_write(struct pipe_inode_info
*,
2782 struct file
*, loff_t
*, size_t, unsigned int);
2783 extern ssize_t
generic_splice_sendpage(struct pipe_inode_info
*pipe
,
2784 struct file
*out
, loff_t
*, size_t len
, unsigned int flags
);
2785 extern long do_splice_direct(struct file
*in
, loff_t
*ppos
, struct file
*out
,
2786 loff_t
*opos
, size_t len
, unsigned int flags
);
2790 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
2791 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int whence
);
2792 extern loff_t
no_llseek(struct file
*file
, loff_t offset
, int whence
);
2793 extern loff_t
vfs_setpos(struct file
*file
, loff_t offset
, loff_t maxsize
);
2794 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int whence
);
2795 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
2796 int whence
, loff_t maxsize
, loff_t eof
);
2797 extern loff_t
fixed_size_llseek(struct file
*file
, loff_t offset
,
2798 int whence
, loff_t size
);
2799 extern loff_t
no_seek_end_llseek_size(struct file
*, loff_t
, int, loff_t
);
2800 extern loff_t
no_seek_end_llseek(struct file
*, loff_t
, int);
2801 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
2802 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
2805 typedef void (dio_submit_t
)(int rw
, struct bio
*bio
, struct inode
*inode
,
2806 loff_t file_offset
);
2809 /* need locking between buffered and direct access */
2812 /* filesystem does not support filling holes */
2813 DIO_SKIP_HOLES
= 0x02,
2815 /* filesystem can handle aio writes beyond i_size */
2816 DIO_ASYNC_EXTEND
= 0x04,
2818 /* inode/fs/bdev does not need truncate protection */
2819 DIO_SKIP_DIO_COUNT
= 0x08,
2822 void dio_end_io(struct bio
*bio
, int error
);
2824 ssize_t
__blockdev_direct_IO(struct kiocb
*iocb
, struct inode
*inode
,
2825 struct block_device
*bdev
, struct iov_iter
*iter
,
2826 get_block_t get_block
,
2827 dio_iodone_t end_io
, dio_submit_t submit_io
,
2830 static inline ssize_t
blockdev_direct_IO(struct kiocb
*iocb
,
2831 struct inode
*inode
,
2832 struct iov_iter
*iter
,
2833 get_block_t get_block
)
2835 return __blockdev_direct_IO(iocb
, inode
, inode
->i_sb
->s_bdev
, iter
,
2836 get_block
, NULL
, NULL
, DIO_LOCKING
| DIO_SKIP_HOLES
);
2840 void inode_dio_wait(struct inode
*inode
);
2843 * inode_dio_begin - signal start of a direct I/O requests
2844 * @inode: inode the direct I/O happens on
2846 * This is called once we've finished processing a direct I/O request,
2847 * and is used to wake up callers waiting for direct I/O to be quiesced.
2849 static inline void inode_dio_begin(struct inode
*inode
)
2851 atomic_inc(&inode
->i_dio_count
);
2855 * inode_dio_end - signal finish of a direct I/O requests
2856 * @inode: inode the direct I/O happens on
2858 * This is called once we've finished processing a direct I/O request,
2859 * and is used to wake up callers waiting for direct I/O to be quiesced.
2861 static inline void inode_dio_end(struct inode
*inode
)
2863 if (atomic_dec_and_test(&inode
->i_dio_count
))
2864 wake_up_bit(&inode
->i_state
, __I_DIO_WAKEUP
);
2867 extern void inode_set_flags(struct inode
*inode
, unsigned int flags
,
2870 extern const struct file_operations generic_ro_fops
;
2872 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2874 extern int readlink_copy(char __user
*, int, const char *);
2875 extern int page_readlink(struct dentry
*, char __user
*, int);
2876 extern const char *page_get_link(struct dentry
*, struct inode
*,
2877 struct delayed_call
*);
2878 extern void page_put_link(void *);
2879 extern int __page_symlink(struct inode
*inode
, const char *symname
, int len
,
2881 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
2882 extern const struct inode_operations page_symlink_inode_operations
;
2883 extern void kfree_link(void *);
2884 extern int generic_readlink(struct dentry
*, char __user
*, int);
2885 extern void generic_fillattr(struct inode
*, struct kstat
*);
2886 int vfs_getattr_nosec(struct path
*path
, struct kstat
*stat
);
2887 extern int vfs_getattr(struct path
*, struct kstat
*);
2888 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2889 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2890 void __inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2891 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2892 loff_t
inode_get_bytes(struct inode
*inode
);
2893 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
2894 const char *simple_get_link(struct dentry
*, struct inode
*,
2895 struct delayed_call
*);
2896 extern const struct inode_operations simple_symlink_inode_operations
;
2898 extern int iterate_dir(struct file
*, struct dir_context
*);
2900 extern int vfs_stat(const char __user
*, struct kstat
*);
2901 extern int vfs_lstat(const char __user
*, struct kstat
*);
2902 extern int vfs_fstat(unsigned int, struct kstat
*);
2903 extern int vfs_fstatat(int , const char __user
*, struct kstat
*, int);
2905 extern int __generic_block_fiemap(struct inode
*inode
,
2906 struct fiemap_extent_info
*fieinfo
,
2907 loff_t start
, loff_t len
,
2908 get_block_t
*get_block
);
2909 extern int generic_block_fiemap(struct inode
*inode
,
2910 struct fiemap_extent_info
*fieinfo
, u64 start
,
2911 u64 len
, get_block_t
*get_block
);
2913 extern void get_filesystem(struct file_system_type
*fs
);
2914 extern void put_filesystem(struct file_system_type
*fs
);
2915 extern struct file_system_type
*get_fs_type(const char *name
);
2916 extern struct super_block
*get_super(struct block_device
*);
2917 extern struct super_block
*get_super_thawed(struct block_device
*);
2918 extern struct super_block
*get_active_super(struct block_device
*bdev
);
2919 extern void drop_super(struct super_block
*sb
);
2920 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
2921 extern void iterate_supers_type(struct file_system_type
*,
2922 void (*)(struct super_block
*, void *), void *);
2924 extern int dcache_dir_open(struct inode
*, struct file
*);
2925 extern int dcache_dir_close(struct inode
*, struct file
*);
2926 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
2927 extern int dcache_readdir(struct file
*, struct dir_context
*);
2928 extern int simple_setattr(struct dentry
*, struct iattr
*);
2929 extern int simple_getattr(struct vfsmount
*, struct dentry
*, struct kstat
*);
2930 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
2931 extern int simple_open(struct inode
*inode
, struct file
*file
);
2932 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
2933 extern int simple_unlink(struct inode
*, struct dentry
*);
2934 extern int simple_rmdir(struct inode
*, struct dentry
*);
2935 extern int simple_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*);
2936 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
2937 extern int simple_empty(struct dentry
*);
2938 extern int simple_readpage(struct file
*file
, struct page
*page
);
2939 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
2940 loff_t pos
, unsigned len
, unsigned flags
,
2941 struct page
**pagep
, void **fsdata
);
2942 extern int simple_write_end(struct file
*file
, struct address_space
*mapping
,
2943 loff_t pos
, unsigned len
, unsigned copied
,
2944 struct page
*page
, void *fsdata
);
2945 extern int always_delete_dentry(const struct dentry
*);
2946 extern struct inode
*alloc_anon_inode(struct super_block
*);
2947 extern int simple_nosetlease(struct file
*, long, struct file_lock
**, void **);
2948 extern const struct dentry_operations simple_dentry_operations
;
2950 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
2951 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
2952 extern const struct file_operations simple_dir_operations
;
2953 extern const struct inode_operations simple_dir_inode_operations
;
2954 extern void make_empty_dir_inode(struct inode
*inode
);
2955 extern bool is_empty_dir_inode(struct inode
*inode
);
2956 struct tree_descr
{ char *name
; const struct file_operations
*ops
; int mode
; };
2957 struct dentry
*d_alloc_name(struct dentry
*, const char *);
2958 extern int simple_fill_super(struct super_block
*, unsigned long, struct tree_descr
*);
2959 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
2960 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
2962 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
2963 loff_t
*ppos
, const void *from
, size_t available
);
2964 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
2965 const void __user
*from
, size_t count
);
2967 extern int __generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
2968 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
2970 extern int generic_check_addressable(unsigned, u64
);
2972 #ifdef CONFIG_MIGRATION
2973 extern int buffer_migrate_page(struct address_space
*,
2974 struct page
*, struct page
*,
2977 #define buffer_migrate_page NULL
2980 extern int inode_change_ok(const struct inode
*, struct iattr
*);
2981 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
2982 extern void setattr_copy(struct inode
*inode
, const struct iattr
*attr
);
2984 extern int file_update_time(struct file
*file
);
2986 extern int generic_show_options(struct seq_file
*m
, struct dentry
*root
);
2987 extern void save_mount_options(struct super_block
*sb
, char *options
);
2988 extern void replace_mount_options(struct super_block
*sb
, char *options
);
2990 static inline bool io_is_direct(struct file
*filp
)
2992 return (filp
->f_flags
& O_DIRECT
) || IS_DAX(filp
->f_mapping
->host
);
2995 static inline int iocb_flags(struct file
*file
)
2998 if (file
->f_flags
& O_APPEND
)
3000 if (io_is_direct(file
))
3002 if ((file
->f_flags
& O_DSYNC
) || IS_SYNC(file
->f_mapping
->host
))
3004 if (file
->f_flags
& __O_SYNC
)
3009 static inline ino_t
parent_ino(struct dentry
*dentry
)
3014 * Don't strictly need d_lock here? If the parent ino could change
3015 * then surely we'd have a deeper race in the caller?
3017 spin_lock(&dentry
->d_lock
);
3018 res
= dentry
->d_parent
->d_inode
->i_ino
;
3019 spin_unlock(&dentry
->d_lock
);
3023 /* Transaction based IO helpers */
3026 * An argresp is stored in an allocated page and holds the
3027 * size of the argument or response, along with its content
3029 struct simple_transaction_argresp
{
3034 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3036 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
3038 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
3039 size_t size
, loff_t
*pos
);
3040 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
3042 void simple_transaction_set(struct file
*file
, size_t n
);
3045 * simple attribute files
3047 * These attributes behave similar to those in sysfs:
3049 * Writing to an attribute immediately sets a value, an open file can be
3050 * written to multiple times.
3052 * Reading from an attribute creates a buffer from the value that might get
3053 * read with multiple read calls. When the attribute has been read
3054 * completely, no further read calls are possible until the file is opened
3057 * All attributes contain a text representation of a numeric value
3058 * that are accessed with the get() and set() functions.
3060 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3061 static int __fops ## _open(struct inode *inode, struct file *file) \
3063 __simple_attr_check_format(__fmt, 0ull); \
3064 return simple_attr_open(inode, file, __get, __set, __fmt); \
3066 static const struct file_operations __fops = { \
3067 .owner = THIS_MODULE, \
3068 .open = __fops ## _open, \
3069 .release = simple_attr_release, \
3070 .read = simple_attr_read, \
3071 .write = simple_attr_write, \
3072 .llseek = generic_file_llseek, \
3075 static inline __printf(1, 2)
3076 void __simple_attr_check_format(const char *fmt
, ...)
3078 /* don't do anything, just let the compiler check the arguments; */
3081 int simple_attr_open(struct inode
*inode
, struct file
*file
,
3082 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
3084 int simple_attr_release(struct inode
*inode
, struct file
*file
);
3085 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
3086 size_t len
, loff_t
*ppos
);
3087 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
3088 size_t len
, loff_t
*ppos
);
3091 int proc_nr_files(struct ctl_table
*table
, int write
,
3092 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3093 int proc_nr_dentry(struct ctl_table
*table
, int write
,
3094 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3095 int proc_nr_inodes(struct ctl_table
*table
, int write
,
3096 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3097 int __init
get_filesystem_list(char *buf
);
3099 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3100 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3102 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3103 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3104 (flag & __FMODE_NONOTIFY)))
3106 static inline bool is_sxid(umode_t mode
)
3108 return (mode
& S_ISUID
) || ((mode
& S_ISGID
) && (mode
& S_IXGRP
));
3111 static inline int check_sticky(struct inode
*dir
, struct inode
*inode
)
3113 if (!(dir
->i_mode
& S_ISVTX
))
3116 return __check_sticky(dir
, inode
);
3119 static inline void inode_has_no_xattr(struct inode
*inode
)
3121 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& MS_NOSEC
))
3122 inode
->i_flags
|= S_NOSEC
;
3125 static inline bool is_root_inode(struct inode
*inode
)
3127 return inode
== inode
->i_sb
->s_root
->d_inode
;
3130 static inline bool dir_emit(struct dir_context
*ctx
,
3131 const char *name
, int namelen
,
3132 u64 ino
, unsigned type
)
3134 return ctx
->actor(ctx
, name
, namelen
, ctx
->pos
, ino
, type
) == 0;
3136 static inline bool dir_emit_dot(struct file
*file
, struct dir_context
*ctx
)
3138 return ctx
->actor(ctx
, ".", 1, ctx
->pos
,
3139 file
->f_path
.dentry
->d_inode
->i_ino
, DT_DIR
) == 0;
3141 static inline bool dir_emit_dotdot(struct file
*file
, struct dir_context
*ctx
)
3143 return ctx
->actor(ctx
, "..", 2, ctx
->pos
,
3144 parent_ino(file
->f_path
.dentry
), DT_DIR
) == 0;
3146 static inline bool dir_emit_dots(struct file
*file
, struct dir_context
*ctx
)
3148 if (ctx
->pos
== 0) {
3149 if (!dir_emit_dot(file
, ctx
))
3153 if (ctx
->pos
== 1) {
3154 if (!dir_emit_dotdot(file
, ctx
))
3160 static inline bool dir_relax(struct inode
*inode
)
3162 inode_unlock(inode
);
3164 return !IS_DEADDIR(inode
);
3167 static inline bool dir_relax_shared(struct inode
*inode
)
3169 inode_unlock_shared(inode
);
3170 inode_lock_shared(inode
);
3171 return !IS_DEADDIR(inode
);
3174 extern bool path_noexec(const struct path
*path
);
3175 extern void inode_nohighmem(struct inode
*inode
);
3177 #endif /* _LINUX_FS_H */