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 flags that we support. Some of
156 * them include behavioral modifiers that send information down to the
157 * block layer and IO scheduler. They should be used along with a req_op.
160 * The block layer uses device plugging to defer IO a little bit, in
161 * the hope that we will see more IO very shortly. This increases
162 * coalescing of adjacent IO and thus reduces the number of IOs we
163 * have to send to the device. It also allows for better queuing,
164 * if the IO isn't mergeable. If the caller is going to be waiting
165 * for the IO, then he must ensure that the device is unplugged so
166 * that the IO is dispatched to the driver.
168 * All IO is handled async in Linux. This is fine for background
169 * writes, but for reads or writes that someone waits for completion
170 * on, we want to notify the block layer and IO scheduler so that they
171 * know about it. That allows them to make better scheduling
172 * decisions. So when the below references 'sync' and 'async', it
173 * is referencing this priority hint.
175 * With that in mind, the available types are:
177 * READ A normal read operation. Device will be plugged.
178 * READ_SYNC A synchronous read. Device is not plugged, caller can
179 * immediately wait on this read without caring about
181 * WRITE A normal async write. Device will be plugged.
182 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
183 * the hint that someone will be waiting on this IO
184 * shortly. The write equivalent of READ_SYNC.
185 * WRITE_ODIRECT Special case write for O_DIRECT only.
186 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
187 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
188 * non-volatile media on completion.
189 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
190 * by a cache flush and data is guaranteed to be on
191 * non-volatile media on completion.
194 #define RW_MASK REQ_OP_WRITE
196 #define READ REQ_OP_READ
197 #define WRITE REQ_OP_WRITE
199 #define READ_SYNC REQ_SYNC
200 #define WRITE_SYNC (REQ_SYNC | REQ_NOIDLE)
201 #define WRITE_ODIRECT REQ_SYNC
202 #define WRITE_FLUSH (REQ_SYNC | REQ_NOIDLE | REQ_PREFLUSH)
203 #define WRITE_FUA (REQ_SYNC | REQ_NOIDLE | REQ_FUA)
204 #define WRITE_FLUSH_FUA (REQ_SYNC | REQ_NOIDLE | REQ_PREFLUSH | REQ_FUA)
207 * Attribute flags. These should be or-ed together to figure out what
210 #define ATTR_MODE (1 << 0)
211 #define ATTR_UID (1 << 1)
212 #define ATTR_GID (1 << 2)
213 #define ATTR_SIZE (1 << 3)
214 #define ATTR_ATIME (1 << 4)
215 #define ATTR_MTIME (1 << 5)
216 #define ATTR_CTIME (1 << 6)
217 #define ATTR_ATIME_SET (1 << 7)
218 #define ATTR_MTIME_SET (1 << 8)
219 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
220 #define ATTR_ATTR_FLAG (1 << 10)
221 #define ATTR_KILL_SUID (1 << 11)
222 #define ATTR_KILL_SGID (1 << 12)
223 #define ATTR_FILE (1 << 13)
224 #define ATTR_KILL_PRIV (1 << 14)
225 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
226 #define ATTR_TIMES_SET (1 << 16)
229 * Whiteout is represented by a char device. The following constants define the
230 * mode and device number to use.
232 #define WHITEOUT_MODE 0
233 #define WHITEOUT_DEV 0
236 * This is the Inode Attributes structure, used for notify_change(). It
237 * uses the above definitions as flags, to know which values have changed.
238 * Also, in this manner, a Filesystem can look at only the values it cares
239 * about. Basically, these are the attributes that the VFS layer can
240 * request to change from the FS layer.
242 * Derek Atkins <warlord@MIT.EDU> 94-10-20
245 unsigned int ia_valid
;
250 struct timespec ia_atime
;
251 struct timespec ia_mtime
;
252 struct timespec ia_ctime
;
255 * Not an attribute, but an auxiliary info for filesystems wanting to
256 * implement an ftruncate() like method. NOTE: filesystem should
257 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
259 struct file
*ia_file
;
263 * Includes for diskquotas.
265 #include <linux/quota.h>
268 * Maximum number of layers of fs stack. Needs to be limited to
269 * prevent kernel stack overflow
271 #define FILESYSTEM_MAX_STACK_DEPTH 2
274 * enum positive_aop_returns - aop return codes with specific semantics
276 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
277 * completed, that the page is still locked, and
278 * should be considered active. The VM uses this hint
279 * to return the page to the active list -- it won't
280 * be a candidate for writeback again in the near
281 * future. Other callers must be careful to unlock
282 * the page if they get this return. Returned by
285 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
286 * unlocked it and the page might have been truncated.
287 * The caller should back up to acquiring a new page and
288 * trying again. The aop will be taking reasonable
289 * precautions not to livelock. If the caller held a page
290 * reference, it should drop it before retrying. Returned
293 * address_space_operation functions return these large constants to indicate
294 * special semantics to the caller. These are much larger than the bytes in a
295 * page to allow for functions that return the number of bytes operated on in a
299 enum positive_aop_returns
{
300 AOP_WRITEPAGE_ACTIVATE
= 0x80000,
301 AOP_TRUNCATED_PAGE
= 0x80001,
304 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
305 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
306 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
307 * helper code (eg buffer layer)
308 * to clear GFP_FS from alloc */
311 * oh the beauties of C type declarations.
314 struct address_space
;
315 struct writeback_control
;
317 #define IOCB_EVENTFD (1 << 0)
318 #define IOCB_APPEND (1 << 1)
319 #define IOCB_DIRECT (1 << 2)
320 #define IOCB_HIPRI (1 << 3)
321 #define IOCB_DSYNC (1 << 4)
322 #define IOCB_SYNC (1 << 5)
325 struct file
*ki_filp
;
327 void (*ki_complete
)(struct kiocb
*iocb
, long ret
, long ret2
);
332 static inline bool is_sync_kiocb(struct kiocb
*kiocb
)
334 return kiocb
->ki_complete
== NULL
;
337 static inline int iocb_flags(struct file
*file
);
339 static inline void init_sync_kiocb(struct kiocb
*kiocb
, struct file
*filp
)
341 *kiocb
= (struct kiocb
) {
343 .ki_flags
= iocb_flags(filp
),
348 * "descriptor" for what we're up to with a read.
349 * This allows us to use the same read code yet
350 * have multiple different users of the data that
351 * we read from a file.
353 * The simplest case just copies the data to user
366 typedef int (*read_actor_t
)(read_descriptor_t
*, struct page
*,
367 unsigned long, unsigned long);
369 struct address_space_operations
{
370 int (*writepage
)(struct page
*page
, struct writeback_control
*wbc
);
371 int (*readpage
)(struct file
*, struct page
*);
373 /* Write back some dirty pages from this mapping. */
374 int (*writepages
)(struct address_space
*, struct writeback_control
*);
376 /* Set a page dirty. Return true if this dirtied it */
377 int (*set_page_dirty
)(struct page
*page
);
379 int (*readpages
)(struct file
*filp
, struct address_space
*mapping
,
380 struct list_head
*pages
, unsigned nr_pages
);
382 int (*write_begin
)(struct file
*, struct address_space
*mapping
,
383 loff_t pos
, unsigned len
, unsigned flags
,
384 struct page
**pagep
, void **fsdata
);
385 int (*write_end
)(struct file
*, struct address_space
*mapping
,
386 loff_t pos
, unsigned len
, unsigned copied
,
387 struct page
*page
, void *fsdata
);
389 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
390 sector_t (*bmap
)(struct address_space
*, sector_t
);
391 void (*invalidatepage
) (struct page
*, unsigned int, unsigned int);
392 int (*releasepage
) (struct page
*, gfp_t
);
393 void (*freepage
)(struct page
*);
394 ssize_t (*direct_IO
)(struct kiocb
*, struct iov_iter
*iter
);
396 * migrate the contents of a page to the specified target. If
397 * migrate_mode is MIGRATE_ASYNC, it must not block.
399 int (*migratepage
) (struct address_space
*,
400 struct page
*, struct page
*, enum migrate_mode
);
401 int (*launder_page
) (struct page
*);
402 int (*is_partially_uptodate
) (struct page
*, unsigned long,
404 void (*is_dirty_writeback
) (struct page
*, bool *, bool *);
405 int (*error_remove_page
)(struct address_space
*, struct page
*);
407 /* swapfile support */
408 int (*swap_activate
)(struct swap_info_struct
*sis
, struct file
*file
,
410 void (*swap_deactivate
)(struct file
*file
);
413 extern const struct address_space_operations empty_aops
;
416 * pagecache_write_begin/pagecache_write_end must be used by general code
417 * to write into the pagecache.
419 int pagecache_write_begin(struct file
*, struct address_space
*mapping
,
420 loff_t pos
, unsigned len
, unsigned flags
,
421 struct page
**pagep
, void **fsdata
);
423 int pagecache_write_end(struct file
*, struct address_space
*mapping
,
424 loff_t pos
, unsigned len
, unsigned copied
,
425 struct page
*page
, void *fsdata
);
427 struct address_space
{
428 struct inode
*host
; /* owner: inode, block_device */
429 struct radix_tree_root page_tree
; /* radix tree of all pages */
430 spinlock_t tree_lock
; /* and lock protecting it */
431 atomic_t i_mmap_writable
;/* count VM_SHARED mappings */
432 struct rb_root i_mmap
; /* tree of private and shared mappings */
433 struct rw_semaphore i_mmap_rwsem
; /* protect tree, count, list */
434 /* Protected by tree_lock together with the radix tree */
435 unsigned long nrpages
; /* number of total pages */
436 /* number of shadow or DAX exceptional entries */
437 unsigned long nrexceptional
;
438 pgoff_t writeback_index
;/* writeback starts here */
439 const struct address_space_operations
*a_ops
; /* methods */
440 unsigned long flags
; /* error bits/gfp mask */
441 spinlock_t private_lock
; /* for use by the address_space */
442 struct list_head private_list
; /* ditto */
443 void *private_data
; /* ditto */
444 } __attribute__((aligned(sizeof(long))));
446 * On most architectures that alignment is already the case; but
447 * must be enforced here for CRIS, to let the least significant bit
448 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
450 struct request_queue
;
452 struct block_device
{
453 dev_t bd_dev
; /* not a kdev_t - it's a search key */
455 struct inode
* bd_inode
; /* will die */
456 struct super_block
* bd_super
;
457 struct mutex bd_mutex
; /* open/close mutex */
458 struct list_head bd_inodes
;
462 bool bd_write_holder
;
464 struct list_head bd_holder_disks
;
466 struct block_device
* bd_contains
;
467 unsigned bd_block_size
;
468 struct hd_struct
* bd_part
;
469 /* number of times partitions within this device have been opened. */
470 unsigned bd_part_count
;
472 struct gendisk
* bd_disk
;
473 struct request_queue
* bd_queue
;
474 struct list_head bd_list
;
476 * Private data. You must have bd_claim'ed the block_device
477 * to use this. NOTE: bd_claim allows an owner to claim
478 * the same device multiple times, the owner must take special
479 * care to not mess up bd_private for that case.
481 unsigned long bd_private
;
483 /* The counter of freeze processes */
484 int bd_fsfreeze_count
;
485 /* Mutex for freeze */
486 struct mutex bd_fsfreeze_mutex
;
490 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
493 #define PAGECACHE_TAG_DIRTY 0
494 #define PAGECACHE_TAG_WRITEBACK 1
495 #define PAGECACHE_TAG_TOWRITE 2
497 int mapping_tagged(struct address_space
*mapping
, int tag
);
499 static inline void i_mmap_lock_write(struct address_space
*mapping
)
501 down_write(&mapping
->i_mmap_rwsem
);
504 static inline void i_mmap_unlock_write(struct address_space
*mapping
)
506 up_write(&mapping
->i_mmap_rwsem
);
509 static inline void i_mmap_lock_read(struct address_space
*mapping
)
511 down_read(&mapping
->i_mmap_rwsem
);
514 static inline void i_mmap_unlock_read(struct address_space
*mapping
)
516 up_read(&mapping
->i_mmap_rwsem
);
520 * Might pages of this file be mapped into userspace?
522 static inline int mapping_mapped(struct address_space
*mapping
)
524 return !RB_EMPTY_ROOT(&mapping
->i_mmap
);
528 * Might pages of this file have been modified in userspace?
529 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
530 * marks vma as VM_SHARED if it is shared, and the file was opened for
531 * writing i.e. vma may be mprotected writable even if now readonly.
533 * If i_mmap_writable is negative, no new writable mappings are allowed. You
534 * can only deny writable mappings, if none exists right now.
536 static inline int mapping_writably_mapped(struct address_space
*mapping
)
538 return atomic_read(&mapping
->i_mmap_writable
) > 0;
541 static inline int mapping_map_writable(struct address_space
*mapping
)
543 return atomic_inc_unless_negative(&mapping
->i_mmap_writable
) ?
547 static inline void mapping_unmap_writable(struct address_space
*mapping
)
549 atomic_dec(&mapping
->i_mmap_writable
);
552 static inline int mapping_deny_writable(struct address_space
*mapping
)
554 return atomic_dec_unless_positive(&mapping
->i_mmap_writable
) ?
558 static inline void mapping_allow_writable(struct address_space
*mapping
)
560 atomic_inc(&mapping
->i_mmap_writable
);
564 * Use sequence counter to get consistent i_size on 32-bit processors.
566 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
567 #include <linux/seqlock.h>
568 #define __NEED_I_SIZE_ORDERED
569 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
571 #define i_size_ordered_init(inode) do { } while (0)
575 #define ACL_NOT_CACHED ((void *)(-1))
577 static inline struct posix_acl
*
578 uncached_acl_sentinel(struct task_struct
*task
)
580 return (void *)task
+ 1;
584 is_uncached_acl(struct posix_acl
*acl
)
586 return (long)acl
& 1;
589 #define IOP_FASTPERM 0x0001
590 #define IOP_LOOKUP 0x0002
591 #define IOP_NOFOLLOW 0x0004
594 * Keep mostly read-only and often accessed (especially for
595 * the RCU path lookup and 'stat' data) fields at the beginning
596 * of the 'struct inode'
600 unsigned short i_opflags
;
603 unsigned int i_flags
;
605 #ifdef CONFIG_FS_POSIX_ACL
606 struct posix_acl
*i_acl
;
607 struct posix_acl
*i_default_acl
;
610 const struct inode_operations
*i_op
;
611 struct super_block
*i_sb
;
612 struct address_space
*i_mapping
;
614 #ifdef CONFIG_SECURITY
618 /* Stat data, not accessed from path walking */
621 * Filesystems may only read i_nlink directly. They shall use the
622 * following functions for modification:
624 * (set|clear|inc|drop)_nlink
625 * inode_(inc|dec)_link_count
628 const unsigned int i_nlink
;
629 unsigned int __i_nlink
;
633 struct timespec i_atime
;
634 struct timespec i_mtime
;
635 struct timespec i_ctime
;
636 spinlock_t i_lock
; /* i_blocks, i_bytes, maybe i_size */
637 unsigned short i_bytes
;
638 unsigned int i_blkbits
;
641 #ifdef __NEED_I_SIZE_ORDERED
642 seqcount_t i_size_seqcount
;
646 unsigned long i_state
;
647 struct rw_semaphore i_rwsem
;
649 unsigned long dirtied_when
; /* jiffies of first dirtying */
650 unsigned long dirtied_time_when
;
652 struct hlist_node i_hash
;
653 struct list_head i_io_list
; /* backing dev IO list */
654 #ifdef CONFIG_CGROUP_WRITEBACK
655 struct bdi_writeback
*i_wb
; /* the associated cgroup wb */
657 /* foreign inode detection, see wbc_detach_inode() */
659 u16 i_wb_frn_avg_time
;
660 u16 i_wb_frn_history
;
662 struct list_head i_lru
; /* inode LRU list */
663 struct list_head i_sb_list
;
665 struct hlist_head i_dentry
;
666 struct rcu_head i_rcu
;
670 atomic_t i_dio_count
;
671 atomic_t i_writecount
;
673 atomic_t i_readcount
; /* struct files open RO */
675 const struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
676 struct file_lock_context
*i_flctx
;
677 struct address_space i_data
;
678 struct list_head i_devices
;
680 struct pipe_inode_info
*i_pipe
;
681 struct block_device
*i_bdev
;
689 #ifdef CONFIG_FSNOTIFY
690 __u32 i_fsnotify_mask
; /* all events this inode cares about */
691 struct hlist_head i_fsnotify_marks
;
694 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
695 struct fscrypt_info
*i_crypt_info
;
698 void *i_private
; /* fs or device private pointer */
701 static inline int inode_unhashed(struct inode
*inode
)
703 return hlist_unhashed(&inode
->i_hash
);
707 * inode->i_mutex nesting subclasses for the lock validator:
709 * 0: the object of the current VFS operation
713 * 4: second non-directory
714 * 5: second parent (when locking independent directories in rename)
716 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
717 * non-directories at once.
719 * The locking order between these classes is
720 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
722 enum inode_i_mutex_lock_class
732 static inline void inode_lock(struct inode
*inode
)
734 down_write(&inode
->i_rwsem
);
737 static inline void inode_unlock(struct inode
*inode
)
739 up_write(&inode
->i_rwsem
);
742 static inline void inode_lock_shared(struct inode
*inode
)
744 down_read(&inode
->i_rwsem
);
747 static inline void inode_unlock_shared(struct inode
*inode
)
749 up_read(&inode
->i_rwsem
);
752 static inline int inode_trylock(struct inode
*inode
)
754 return down_write_trylock(&inode
->i_rwsem
);
757 static inline int inode_trylock_shared(struct inode
*inode
)
759 return down_read_trylock(&inode
->i_rwsem
);
762 static inline int inode_is_locked(struct inode
*inode
)
764 return rwsem_is_locked(&inode
->i_rwsem
);
767 static inline void inode_lock_nested(struct inode
*inode
, unsigned subclass
)
769 down_write_nested(&inode
->i_rwsem
, subclass
);
772 void lock_two_nondirectories(struct inode
*, struct inode
*);
773 void unlock_two_nondirectories(struct inode
*, struct inode
*);
776 * NOTE: in a 32bit arch with a preemptable kernel and
777 * an UP compile the i_size_read/write must be atomic
778 * with respect to the local cpu (unlike with preempt disabled),
779 * but they don't need to be atomic with respect to other cpus like in
780 * true SMP (so they need either to either locally disable irq around
781 * the read or for example on x86 they can be still implemented as a
782 * cmpxchg8b without the need of the lock prefix). For SMP compiles
783 * and 64bit archs it makes no difference if preempt is enabled or not.
785 static inline loff_t
i_size_read(const struct inode
*inode
)
787 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
792 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
793 i_size
= inode
->i_size
;
794 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
796 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
800 i_size
= inode
->i_size
;
804 return inode
->i_size
;
809 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
810 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
811 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
813 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
815 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
817 write_seqcount_begin(&inode
->i_size_seqcount
);
818 inode
->i_size
= i_size
;
819 write_seqcount_end(&inode
->i_size_seqcount
);
821 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
823 inode
->i_size
= i_size
;
826 inode
->i_size
= i_size
;
830 /* Helper functions so that in most cases filesystems will
831 * not need to deal directly with kuid_t and kgid_t and can
832 * instead deal with the raw numeric values that are stored
835 static inline uid_t
i_uid_read(const struct inode
*inode
)
837 return from_kuid(&init_user_ns
, inode
->i_uid
);
840 static inline gid_t
i_gid_read(const struct inode
*inode
)
842 return from_kgid(&init_user_ns
, inode
->i_gid
);
845 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
847 inode
->i_uid
= make_kuid(&init_user_ns
, uid
);
850 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
852 inode
->i_gid
= make_kgid(&init_user_ns
, gid
);
855 static inline unsigned iminor(const struct inode
*inode
)
857 return MINOR(inode
->i_rdev
);
860 static inline unsigned imajor(const struct inode
*inode
)
862 return MAJOR(inode
->i_rdev
);
865 extern struct block_device
*I_BDEV(struct inode
*inode
);
868 rwlock_t lock
; /* protects pid, uid, euid fields */
869 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
870 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
871 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
872 int signum
; /* posix.1b rt signal to be delivered on IO */
876 * Track a single file's readahead state
878 struct file_ra_state
{
879 pgoff_t start
; /* where readahead started */
880 unsigned int size
; /* # of readahead pages */
881 unsigned int async_size
; /* do asynchronous readahead when
882 there are only # of pages ahead */
884 unsigned int ra_pages
; /* Maximum readahead window */
885 unsigned int mmap_miss
; /* Cache miss stat for mmap accesses */
886 loff_t prev_pos
; /* Cache last read() position */
890 * Check if @index falls in the readahead windows.
892 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
894 return (index
>= ra
->start
&&
895 index
< ra
->start
+ ra
->size
);
900 struct llist_node fu_llist
;
901 struct rcu_head fu_rcuhead
;
904 struct inode
*f_inode
; /* cached value */
905 const struct file_operations
*f_op
;
908 * Protects f_ep_links, f_flags.
909 * Must not be taken from IRQ context.
912 atomic_long_t f_count
;
913 unsigned int f_flags
;
915 struct mutex f_pos_lock
;
917 struct fown_struct f_owner
;
918 const struct cred
*f_cred
;
919 struct file_ra_state f_ra
;
922 #ifdef CONFIG_SECURITY
925 /* needed for tty driver, and maybe others */
929 /* Used by fs/eventpoll.c to link all the hooks to this file */
930 struct list_head f_ep_links
;
931 struct list_head f_tfile_llink
;
932 #endif /* #ifdef CONFIG_EPOLL */
933 struct address_space
*f_mapping
;
934 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
939 /* file identifier */
940 unsigned char f_handle
[0];
943 static inline struct file
*get_file(struct file
*f
)
945 atomic_long_inc(&f
->f_count
);
948 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
949 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
950 #define file_count(x) atomic_long_read(&(x)->f_count)
952 #define MAX_NON_LFS ((1UL<<31) - 1)
954 /* Page cache limit. The filesystems should put that into their s_maxbytes
955 limits, otherwise bad things can happen in VM. */
956 #if BITS_PER_LONG==32
957 #define MAX_LFS_FILESIZE (((loff_t)PAGE_SIZE << (BITS_PER_LONG-1))-1)
958 #elif BITS_PER_LONG==64
959 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
964 #define FL_DELEG 4 /* NFSv4 delegation */
965 #define FL_ACCESS 8 /* not trying to lock, just looking */
966 #define FL_EXISTS 16 /* when unlocking, test for existence */
967 #define FL_LEASE 32 /* lease held on this file */
968 #define FL_CLOSE 64 /* unlock on close */
969 #define FL_SLEEP 128 /* A blocking lock */
970 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
971 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
972 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
973 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
976 * Special return value from posix_lock_file() and vfs_lock_file() for
977 * asynchronous locking.
979 #define FILE_LOCK_DEFERRED 1
981 /* legacy typedef, should eventually be removed */
982 typedef void *fl_owner_t
;
986 struct file_lock_operations
{
987 void (*fl_copy_lock
)(struct file_lock
*, struct file_lock
*);
988 void (*fl_release_private
)(struct file_lock
*);
991 struct lock_manager_operations
{
992 int (*lm_compare_owner
)(struct file_lock
*, struct file_lock
*);
993 unsigned long (*lm_owner_key
)(struct file_lock
*);
994 fl_owner_t (*lm_get_owner
)(fl_owner_t
);
995 void (*lm_put_owner
)(fl_owner_t
);
996 void (*lm_notify
)(struct file_lock
*); /* unblock callback */
997 int (*lm_grant
)(struct file_lock
*, int);
998 bool (*lm_break
)(struct file_lock
*);
999 int (*lm_change
)(struct file_lock
*, int, struct list_head
*);
1000 void (*lm_setup
)(struct file_lock
*, void **);
1003 struct lock_manager
{
1004 struct list_head list
;
1006 * NFSv4 and up also want opens blocked during the grace period;
1013 void locks_start_grace(struct net
*, struct lock_manager
*);
1014 void locks_end_grace(struct lock_manager
*);
1015 int locks_in_grace(struct net
*);
1016 int opens_in_grace(struct net
*);
1018 /* that will die - we need it for nfs_lock_info */
1019 #include <linux/nfs_fs_i.h>
1022 * struct file_lock represents a generic "file lock". It's used to represent
1023 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1024 * note that the same struct is used to represent both a request for a lock and
1025 * the lock itself, but the same object is never used for both.
1027 * FIXME: should we create a separate "struct lock_request" to help distinguish
1030 * The varous i_flctx lists are ordered by:
1033 * 2) lock range start
1036 * Obviously, the last two criteria only matter for POSIX locks.
1039 struct file_lock
*fl_next
; /* singly linked list for this inode */
1040 struct list_head fl_list
; /* link into file_lock_context */
1041 struct hlist_node fl_link
; /* node in global lists */
1042 struct list_head fl_block
; /* circular list of blocked processes */
1043 fl_owner_t fl_owner
;
1044 unsigned int fl_flags
;
1045 unsigned char fl_type
;
1046 unsigned int fl_pid
;
1047 int fl_link_cpu
; /* what cpu's list is this on? */
1048 struct pid
*fl_nspid
;
1049 wait_queue_head_t fl_wait
;
1050 struct file
*fl_file
;
1054 struct fasync_struct
* fl_fasync
; /* for lease break notifications */
1055 /* for lease breaks: */
1056 unsigned long fl_break_time
;
1057 unsigned long fl_downgrade_time
;
1059 const struct file_lock_operations
*fl_ops
; /* Callbacks for filesystems */
1060 const struct lock_manager_operations
*fl_lmops
; /* Callbacks for lockmanagers */
1062 struct nfs_lock_info nfs_fl
;
1063 struct nfs4_lock_info nfs4_fl
;
1065 struct list_head link
; /* link in AFS vnode's pending_locks list */
1066 int state
; /* state of grant or error if -ve */
1071 struct file_lock_context
{
1072 spinlock_t flc_lock
;
1073 struct list_head flc_flock
;
1074 struct list_head flc_posix
;
1075 struct list_head flc_lease
;
1078 /* The following constant reflects the upper bound of the file/locking space */
1080 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1081 #define OFFSET_MAX INT_LIMIT(loff_t)
1082 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1085 #include <linux/fcntl.h>
1087 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
1089 #ifdef CONFIG_FILE_LOCKING
1090 extern int fcntl_getlk(struct file
*, unsigned int, struct flock __user
*);
1091 extern int fcntl_setlk(unsigned int, struct file
*, unsigned int,
1092 struct flock __user
*);
1094 #if BITS_PER_LONG == 32
1095 extern int fcntl_getlk64(struct file
*, unsigned int, struct flock64 __user
*);
1096 extern int fcntl_setlk64(unsigned int, struct file
*, unsigned int,
1097 struct flock64 __user
*);
1100 extern int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
);
1101 extern int fcntl_getlease(struct file
*filp
);
1104 void locks_free_lock_context(struct inode
*inode
);
1105 void locks_free_lock(struct file_lock
*fl
);
1106 extern void locks_init_lock(struct file_lock
*);
1107 extern struct file_lock
* locks_alloc_lock(void);
1108 extern void locks_copy_lock(struct file_lock
*, struct file_lock
*);
1109 extern void locks_copy_conflock(struct file_lock
*, struct file_lock
*);
1110 extern void locks_remove_posix(struct file
*, fl_owner_t
);
1111 extern void locks_remove_file(struct file
*);
1112 extern void locks_release_private(struct file_lock
*);
1113 extern void posix_test_lock(struct file
*, struct file_lock
*);
1114 extern int posix_lock_file(struct file
*, struct file_lock
*, struct file_lock
*);
1115 extern int posix_unblock_lock(struct file_lock
*);
1116 extern int vfs_test_lock(struct file
*, struct file_lock
*);
1117 extern int vfs_lock_file(struct file
*, unsigned int, struct file_lock
*, struct file_lock
*);
1118 extern int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
);
1119 extern int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
);
1120 extern int __break_lease(struct inode
*inode
, unsigned int flags
, unsigned int type
);
1121 extern void lease_get_mtime(struct inode
*, struct timespec
*time
);
1122 extern int generic_setlease(struct file
*, long, struct file_lock
**, void **priv
);
1123 extern int vfs_setlease(struct file
*, long, struct file_lock
**, void **);
1124 extern int lease_modify(struct file_lock
*, int, struct list_head
*);
1125 struct files_struct
;
1126 extern void show_fd_locks(struct seq_file
*f
,
1127 struct file
*filp
, struct files_struct
*files
);
1128 #else /* !CONFIG_FILE_LOCKING */
1129 static inline int fcntl_getlk(struct file
*file
, unsigned int cmd
,
1130 struct flock __user
*user
)
1135 static inline int fcntl_setlk(unsigned int fd
, struct file
*file
,
1136 unsigned int cmd
, struct flock __user
*user
)
1141 #if BITS_PER_LONG == 32
1142 static inline int fcntl_getlk64(struct file
*file
, unsigned int cmd
,
1143 struct flock64 __user
*user
)
1148 static inline int fcntl_setlk64(unsigned int fd
, struct file
*file
,
1149 unsigned int cmd
, struct flock64 __user
*user
)
1154 static inline int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
)
1159 static inline int fcntl_getlease(struct file
*filp
)
1165 locks_free_lock_context(struct inode
*inode
)
1169 static inline void locks_init_lock(struct file_lock
*fl
)
1174 static inline void locks_copy_conflock(struct file_lock
*new, struct file_lock
*fl
)
1179 static inline void locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
1184 static inline void locks_remove_posix(struct file
*filp
, fl_owner_t owner
)
1189 static inline void locks_remove_file(struct file
*filp
)
1194 static inline void posix_test_lock(struct file
*filp
, struct file_lock
*fl
)
1199 static inline int posix_lock_file(struct file
*filp
, struct file_lock
*fl
,
1200 struct file_lock
*conflock
)
1205 static inline int posix_unblock_lock(struct file_lock
*waiter
)
1210 static inline int vfs_test_lock(struct file
*filp
, struct file_lock
*fl
)
1215 static inline int vfs_lock_file(struct file
*filp
, unsigned int cmd
,
1216 struct file_lock
*fl
, struct file_lock
*conf
)
1221 static inline int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
)
1226 static inline int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
)
1231 static inline int __break_lease(struct inode
*inode
, unsigned int mode
, unsigned int type
)
1236 static inline void lease_get_mtime(struct inode
*inode
, struct timespec
*time
)
1241 static inline int generic_setlease(struct file
*filp
, long arg
,
1242 struct file_lock
**flp
, void **priv
)
1247 static inline int vfs_setlease(struct file
*filp
, long arg
,
1248 struct file_lock
**lease
, void **priv
)
1253 static inline int lease_modify(struct file_lock
*fl
, int arg
,
1254 struct list_head
*dispose
)
1259 struct files_struct
;
1260 static inline void show_fd_locks(struct seq_file
*f
,
1261 struct file
*filp
, struct files_struct
*files
) {}
1262 #endif /* !CONFIG_FILE_LOCKING */
1264 static inline struct inode
*file_inode(const struct file
*f
)
1269 static inline struct dentry
*file_dentry(const struct file
*file
)
1271 struct dentry
*dentry
= file
->f_path
.dentry
;
1273 if (unlikely(dentry
->d_flags
& DCACHE_OP_REAL
))
1274 return dentry
->d_op
->d_real(dentry
, file_inode(file
));
1279 static inline int locks_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1281 return locks_lock_inode_wait(file_inode(filp
), fl
);
1284 struct fasync_struct
{
1288 struct fasync_struct
*fa_next
; /* singly linked list */
1289 struct file
*fa_file
;
1290 struct rcu_head fa_rcu
;
1293 #define FASYNC_MAGIC 0x4601
1295 /* SMP safe fasync helpers: */
1296 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1297 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1298 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1299 extern struct fasync_struct
*fasync_alloc(void);
1300 extern void fasync_free(struct fasync_struct
*);
1302 /* can be called from interrupts */
1303 extern void kill_fasync(struct fasync_struct
**, int, int);
1305 extern void __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1306 extern void f_setown(struct file
*filp
, unsigned long arg
, int force
);
1307 extern void f_delown(struct file
*filp
);
1308 extern pid_t
f_getown(struct file
*filp
);
1309 extern int send_sigurg(struct fown_struct
*fown
);
1317 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1318 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1319 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1320 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1321 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1324 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1325 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1327 /* Possible states of 'frozen' field */
1329 SB_UNFROZEN
= 0, /* FS is unfrozen */
1330 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1331 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1332 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1333 * internal threads if needed) */
1334 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1337 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1340 int frozen
; /* Is sb frozen? */
1341 wait_queue_head_t wait_unfrozen
; /* for get_super_thawed() */
1342 struct percpu_rw_semaphore rw_sem
[SB_FREEZE_LEVELS
];
1345 struct super_block
{
1346 struct list_head s_list
; /* Keep this first */
1347 dev_t s_dev
; /* search index; _not_ kdev_t */
1348 unsigned char s_blocksize_bits
;
1349 unsigned long s_blocksize
;
1350 loff_t s_maxbytes
; /* Max file size */
1351 struct file_system_type
*s_type
;
1352 const struct super_operations
*s_op
;
1353 const struct dquot_operations
*dq_op
;
1354 const struct quotactl_ops
*s_qcop
;
1355 const struct export_operations
*s_export_op
;
1356 unsigned long s_flags
;
1357 unsigned long s_iflags
; /* internal SB_I_* flags */
1358 unsigned long s_magic
;
1359 struct dentry
*s_root
;
1360 struct rw_semaphore s_umount
;
1363 #ifdef CONFIG_SECURITY
1366 const struct xattr_handler
**s_xattr
;
1368 const struct fscrypt_operations
*s_cop
;
1370 struct hlist_bl_head s_anon
; /* anonymous dentries for (nfs) exporting */
1371 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1372 struct block_device
*s_bdev
;
1373 struct backing_dev_info
*s_bdi
;
1374 struct mtd_info
*s_mtd
;
1375 struct hlist_node s_instances
;
1376 unsigned int s_quota_types
; /* Bitmask of supported quota types */
1377 struct quota_info s_dquot
; /* Diskquota specific options */
1379 struct sb_writers s_writers
;
1381 char s_id
[32]; /* Informational name */
1382 u8 s_uuid
[16]; /* UUID */
1384 void *s_fs_info
; /* Filesystem private info */
1385 unsigned int s_max_links
;
1388 /* Granularity of c/m/atime in ns.
1389 Cannot be worse than a second */
1393 * The next field is for VFS *only*. No filesystems have any business
1394 * even looking at it. You had been warned.
1396 struct mutex s_vfs_rename_mutex
; /* Kludge */
1399 * Filesystem subtype. If non-empty the filesystem type field
1400 * in /proc/mounts will be "type.subtype"
1405 * Saved mount options for lazy filesystems using
1406 * generic_show_options()
1408 char __rcu
*s_options
;
1409 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1412 * Saved pool identifier for cleancache (-1 means none)
1414 int cleancache_poolid
;
1416 struct shrinker s_shrink
; /* per-sb shrinker handle */
1418 /* Number of inodes with nlink == 0 but still referenced */
1419 atomic_long_t s_remove_count
;
1421 /* Being remounted read-only */
1422 int s_readonly_remount
;
1424 /* AIO completions deferred from interrupt context */
1425 struct workqueue_struct
*s_dio_done_wq
;
1426 struct hlist_head s_pins
;
1429 * Keep the lru lists last in the structure so they always sit on their
1430 * own individual cachelines.
1432 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp
;
1433 struct list_lru s_inode_lru ____cacheline_aligned_in_smp
;
1434 struct rcu_head rcu
;
1435 struct work_struct destroy_work
;
1437 struct mutex s_sync_lock
; /* sync serialisation lock */
1440 * Indicates how deep in a filesystem stack this SB is
1444 /* s_inode_list_lock protects s_inodes */
1445 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp
;
1446 struct list_head s_inodes
; /* all inodes */
1449 extern struct timespec
current_fs_time(struct super_block
*sb
);
1452 * Snapshotting support.
1455 void __sb_end_write(struct super_block
*sb
, int level
);
1456 int __sb_start_write(struct super_block
*sb
, int level
, bool wait
);
1458 #define __sb_writers_acquired(sb, lev) \
1459 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1460 #define __sb_writers_release(sb, lev) \
1461 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1464 * sb_end_write - drop write access to a superblock
1465 * @sb: the super we wrote to
1467 * Decrement number of writers to the filesystem. Wake up possible waiters
1468 * wanting to freeze the filesystem.
1470 static inline void sb_end_write(struct super_block
*sb
)
1472 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1476 * sb_end_pagefault - drop write access to a superblock from a page fault
1477 * @sb: the super we wrote to
1479 * Decrement number of processes handling write page fault to the filesystem.
1480 * Wake up possible waiters wanting to freeze the filesystem.
1482 static inline void sb_end_pagefault(struct super_block
*sb
)
1484 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1488 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1489 * @sb: the super we wrote to
1491 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1492 * waiters wanting to freeze the filesystem.
1494 static inline void sb_end_intwrite(struct super_block
*sb
)
1496 __sb_end_write(sb
, SB_FREEZE_FS
);
1500 * sb_start_write - get write access to a superblock
1501 * @sb: the super we write to
1503 * When a process wants to write data or metadata to a file system (i.e. dirty
1504 * a page or an inode), it should embed the operation in a sb_start_write() -
1505 * sb_end_write() pair to get exclusion against file system freezing. This
1506 * function increments number of writers preventing freezing. If the file
1507 * system is already frozen, the function waits until the file system is
1510 * Since freeze protection behaves as a lock, users have to preserve
1511 * ordering of freeze protection and other filesystem locks. Generally,
1512 * freeze protection should be the outermost lock. In particular, we have:
1515 * -> i_mutex (write path, truncate, directory ops, ...)
1516 * -> s_umount (freeze_super, thaw_super)
1518 static inline void sb_start_write(struct super_block
*sb
)
1520 __sb_start_write(sb
, SB_FREEZE_WRITE
, true);
1523 static inline int sb_start_write_trylock(struct super_block
*sb
)
1525 return __sb_start_write(sb
, SB_FREEZE_WRITE
, false);
1529 * sb_start_pagefault - get write access to a superblock from a page fault
1530 * @sb: the super we write to
1532 * When a process starts handling write page fault, it should embed the
1533 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1534 * exclusion against file system freezing. This is needed since the page fault
1535 * is going to dirty a page. This function increments number of running page
1536 * faults preventing freezing. If the file system is already frozen, the
1537 * function waits until the file system is thawed.
1539 * Since page fault freeze protection behaves as a lock, users have to preserve
1540 * ordering of freeze protection and other filesystem locks. It is advised to
1541 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1542 * handling code implies lock dependency:
1545 * -> sb_start_pagefault
1547 static inline void sb_start_pagefault(struct super_block
*sb
)
1549 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
, true);
1553 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1554 * @sb: the super we write to
1556 * This is the third level of protection against filesystem freezing. It is
1557 * free for use by a filesystem. The only requirement is that it must rank
1558 * below sb_start_pagefault.
1560 * For example filesystem can call sb_start_intwrite() when starting a
1561 * transaction which somewhat eases handling of freezing for internal sources
1562 * of filesystem changes (internal fs threads, discarding preallocation on file
1565 static inline void sb_start_intwrite(struct super_block
*sb
)
1567 __sb_start_write(sb
, SB_FREEZE_FS
, true);
1571 extern bool inode_owner_or_capable(const struct inode
*inode
);
1574 * VFS helper functions..
1576 extern int vfs_create(struct inode
*, struct dentry
*, umode_t
, bool);
1577 extern int vfs_mkdir(struct inode
*, struct dentry
*, umode_t
);
1578 extern int vfs_mknod(struct inode
*, struct dentry
*, umode_t
, dev_t
);
1579 extern int vfs_symlink(struct inode
*, struct dentry
*, const char *);
1580 extern int vfs_link(struct dentry
*, struct inode
*, struct dentry
*, struct inode
**);
1581 extern int vfs_rmdir(struct inode
*, struct dentry
*);
1582 extern int vfs_unlink(struct inode
*, struct dentry
*, struct inode
**);
1583 extern int vfs_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*, struct inode
**, unsigned int);
1584 extern int vfs_whiteout(struct inode
*, struct dentry
*);
1587 * VFS file helper functions.
1589 extern void inode_init_owner(struct inode
*inode
, const struct inode
*dir
,
1592 * VFS FS_IOC_FIEMAP helper definitions.
1594 struct fiemap_extent_info
{
1595 unsigned int fi_flags
; /* Flags as passed from user */
1596 unsigned int fi_extents_mapped
; /* Number of mapped extents */
1597 unsigned int fi_extents_max
; /* Size of fiemap_extent array */
1598 struct fiemap_extent __user
*fi_extents_start
; /* Start of
1599 fiemap_extent array */
1601 int fiemap_fill_next_extent(struct fiemap_extent_info
*info
, u64 logical
,
1602 u64 phys
, u64 len
, u32 flags
);
1603 int fiemap_check_flags(struct fiemap_extent_info
*fieinfo
, u32 fs_flags
);
1608 * NOTE! These match bits 12..15 of stat.st_mode
1609 * (ie "(i_mode >> 12) & 15").
1611 #define DT_UNKNOWN 0
1622 * This is the "filldir" function type, used by readdir() to let
1623 * the kernel specify what kind of dirent layout it wants to have.
1624 * This allows the kernel to read directories into kernel space or
1625 * to have different dirent layouts depending on the binary type.
1628 typedef int (*filldir_t
)(struct dir_context
*, const char *, int, loff_t
, u64
,
1631 struct dir_context
{
1632 const filldir_t actor
;
1636 struct block_device_operations
;
1638 /* These macros are for out of kernel modules to test that
1639 * the kernel supports the unlocked_ioctl and compat_ioctl
1640 * fields in struct file_operations. */
1641 #define HAVE_COMPAT_IOCTL 1
1642 #define HAVE_UNLOCKED_IOCTL 1
1645 * These flags let !MMU mmap() govern direct device mapping vs immediate
1646 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1648 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1649 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1650 * NOMMU_MAP_READ: Can be mapped for reading
1651 * NOMMU_MAP_WRITE: Can be mapped for writing
1652 * NOMMU_MAP_EXEC: Can be mapped for execution
1654 #define NOMMU_MAP_COPY 0x00000001
1655 #define NOMMU_MAP_DIRECT 0x00000008
1656 #define NOMMU_MAP_READ VM_MAYREAD
1657 #define NOMMU_MAP_WRITE VM_MAYWRITE
1658 #define NOMMU_MAP_EXEC VM_MAYEXEC
1660 #define NOMMU_VMFLAGS \
1661 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1666 struct file_operations
{
1667 struct module
*owner
;
1668 loff_t (*llseek
) (struct file
*, loff_t
, int);
1669 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
1670 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
1671 ssize_t (*read_iter
) (struct kiocb
*, struct iov_iter
*);
1672 ssize_t (*write_iter
) (struct kiocb
*, struct iov_iter
*);
1673 int (*iterate
) (struct file
*, struct dir_context
*);
1674 int (*iterate_shared
) (struct file
*, struct dir_context
*);
1675 unsigned int (*poll
) (struct file
*, struct poll_table_struct
*);
1676 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
1677 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
1678 int (*mmap
) (struct file
*, struct vm_area_struct
*);
1679 int (*open
) (struct inode
*, struct file
*);
1680 int (*flush
) (struct file
*, fl_owner_t id
);
1681 int (*release
) (struct inode
*, struct file
*);
1682 int (*fsync
) (struct file
*, loff_t
, loff_t
, int datasync
);
1683 int (*aio_fsync
) (struct kiocb
*, int datasync
);
1684 int (*fasync
) (int, struct file
*, int);
1685 int (*lock
) (struct file
*, int, struct file_lock
*);
1686 ssize_t (*sendpage
) (struct file
*, struct page
*, int, size_t, loff_t
*, int);
1687 unsigned long (*get_unmapped_area
)(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1688 int (*check_flags
)(int);
1689 int (*flock
) (struct file
*, int, struct file_lock
*);
1690 ssize_t (*splice_write
)(struct pipe_inode_info
*, struct file
*, loff_t
*, size_t, unsigned int);
1691 ssize_t (*splice_read
)(struct file
*, loff_t
*, struct pipe_inode_info
*, size_t, unsigned int);
1692 int (*setlease
)(struct file
*, long, struct file_lock
**, void **);
1693 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
1695 void (*show_fdinfo
)(struct seq_file
*m
, struct file
*f
);
1697 unsigned (*mmap_capabilities
)(struct file
*);
1699 ssize_t (*copy_file_range
)(struct file
*, loff_t
, struct file
*,
1700 loff_t
, size_t, unsigned int);
1701 int (*clone_file_range
)(struct file
*, loff_t
, struct file
*, loff_t
,
1703 ssize_t (*dedupe_file_range
)(struct file
*, u64
, u64
, struct file
*,
1707 struct inode_operations
{
1708 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*, unsigned int);
1709 const char * (*get_link
) (struct dentry
*, struct inode
*, struct delayed_call
*);
1710 int (*permission
) (struct inode
*, int);
1711 struct posix_acl
* (*get_acl
)(struct inode
*, int);
1713 int (*readlink
) (struct dentry
*, char __user
*,int);
1715 int (*create
) (struct inode
*,struct dentry
*, umode_t
, bool);
1716 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
1717 int (*unlink
) (struct inode
*,struct dentry
*);
1718 int (*symlink
) (struct inode
*,struct dentry
*,const char *);
1719 int (*mkdir
) (struct inode
*,struct dentry
*,umode_t
);
1720 int (*rmdir
) (struct inode
*,struct dentry
*);
1721 int (*mknod
) (struct inode
*,struct dentry
*,umode_t
,dev_t
);
1722 int (*rename
) (struct inode
*, struct dentry
*,
1723 struct inode
*, struct dentry
*);
1724 int (*rename2
) (struct inode
*, struct dentry
*,
1725 struct inode
*, struct dentry
*, unsigned int);
1726 int (*setattr
) (struct dentry
*, struct iattr
*);
1727 int (*getattr
) (struct vfsmount
*mnt
, struct dentry
*, struct kstat
*);
1728 int (*setxattr
) (struct dentry
*, struct inode
*,
1729 const char *, const void *, size_t, int);
1730 ssize_t (*getxattr
) (struct dentry
*, struct inode
*,
1731 const char *, void *, size_t);
1732 ssize_t (*listxattr
) (struct dentry
*, char *, size_t);
1733 int (*removexattr
) (struct dentry
*, const char *);
1734 int (*fiemap
)(struct inode
*, struct fiemap_extent_info
*, u64 start
,
1736 int (*update_time
)(struct inode
*, struct timespec
*, int);
1737 int (*atomic_open
)(struct inode
*, struct dentry
*,
1738 struct file
*, unsigned open_flag
,
1739 umode_t create_mode
, int *opened
);
1740 int (*tmpfile
) (struct inode
*, struct dentry
*, umode_t
);
1741 int (*set_acl
)(struct inode
*, struct posix_acl
*, int);
1742 } ____cacheline_aligned
;
1744 ssize_t
rw_copy_check_uvector(int type
, const struct iovec __user
* uvector
,
1745 unsigned long nr_segs
, unsigned long fast_segs
,
1746 struct iovec
*fast_pointer
,
1747 struct iovec
**ret_pointer
);
1749 extern ssize_t
__vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1750 extern ssize_t
__vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1751 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1752 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1753 extern ssize_t
vfs_readv(struct file
*, const struct iovec __user
*,
1754 unsigned long, loff_t
*, int);
1755 extern ssize_t
vfs_writev(struct file
*, const struct iovec __user
*,
1756 unsigned long, loff_t
*, int);
1757 extern ssize_t
vfs_copy_file_range(struct file
*, loff_t
, struct file
*,
1758 loff_t
, size_t, unsigned int);
1759 extern int vfs_clone_file_range(struct file
*file_in
, loff_t pos_in
,
1760 struct file
*file_out
, loff_t pos_out
, u64 len
);
1761 extern int vfs_dedupe_file_range(struct file
*file
,
1762 struct file_dedupe_range
*same
);
1764 struct super_operations
{
1765 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
1766 void (*destroy_inode
)(struct inode
*);
1768 void (*dirty_inode
) (struct inode
*, int flags
);
1769 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
1770 int (*drop_inode
) (struct inode
*);
1771 void (*evict_inode
) (struct inode
*);
1772 void (*put_super
) (struct super_block
*);
1773 int (*sync_fs
)(struct super_block
*sb
, int wait
);
1774 int (*freeze_super
) (struct super_block
*);
1775 int (*freeze_fs
) (struct super_block
*);
1776 int (*thaw_super
) (struct super_block
*);
1777 int (*unfreeze_fs
) (struct super_block
*);
1778 int (*statfs
) (struct dentry
*, struct kstatfs
*);
1779 int (*remount_fs
) (struct super_block
*, int *, char *);
1780 void (*umount_begin
) (struct super_block
*);
1782 int (*show_options
)(struct seq_file
*, struct dentry
*);
1783 int (*show_devname
)(struct seq_file
*, struct dentry
*);
1784 int (*show_path
)(struct seq_file
*, struct dentry
*);
1785 int (*show_stats
)(struct seq_file
*, struct dentry
*);
1787 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
1788 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
1789 struct dquot
**(*get_dquots
)(struct inode
*);
1791 int (*bdev_try_to_free_page
)(struct super_block
*, struct page
*, gfp_t
);
1792 long (*nr_cached_objects
)(struct super_block
*,
1793 struct shrink_control
*);
1794 long (*free_cached_objects
)(struct super_block
*,
1795 struct shrink_control
*);
1799 * Inode flags - they have no relation to superblock flags now
1801 #define S_SYNC 1 /* Writes are synced at once */
1802 #define S_NOATIME 2 /* Do not update access times */
1803 #define S_APPEND 4 /* Append-only file */
1804 #define S_IMMUTABLE 8 /* Immutable file */
1805 #define S_DEAD 16 /* removed, but still open directory */
1806 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1807 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1808 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1809 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1810 #define S_PRIVATE 512 /* Inode is fs-internal */
1811 #define S_IMA 1024 /* Inode has an associated IMA struct */
1812 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1813 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1814 #ifdef CONFIG_FS_DAX
1815 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1817 #define S_DAX 0 /* Make all the DAX code disappear */
1821 * Note that nosuid etc flags are inode-specific: setting some file-system
1822 * flags just means all the inodes inherit those flags by default. It might be
1823 * possible to override it selectively if you really wanted to with some
1824 * ioctl() that is not currently implemented.
1826 * Exception: MS_RDONLY is always applied to the entire file system.
1828 * Unfortunately, it is possible to change a filesystems flags with it mounted
1829 * with files in use. This means that all of the inodes will not have their
1830 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1831 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1833 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1835 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1836 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1837 ((inode)->i_flags & S_SYNC))
1838 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1839 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1840 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1841 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1842 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1844 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1845 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1846 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1847 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1849 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1850 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1851 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1852 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1853 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1854 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1855 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1856 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1858 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1859 (inode)->i_rdev == WHITEOUT_DEV)
1862 * Inode state bits. Protected by inode->i_lock
1864 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1865 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1867 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1868 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1869 * various stages of removing an inode.
1871 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1873 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1874 * fdatasync(). i_atime is the usual cause.
1875 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1876 * these changes separately from I_DIRTY_SYNC so that we
1877 * don't have to write inode on fdatasync() when only
1878 * mtime has changed in it.
1879 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1880 * I_NEW Serves as both a mutex and completion notification.
1881 * New inodes set I_NEW. If two processes both create
1882 * the same inode, one of them will release its inode and
1883 * wait for I_NEW to be released before returning.
1884 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1885 * also cause waiting on I_NEW, without I_NEW actually
1886 * being set. find_inode() uses this to prevent returning
1887 * nearly-dead inodes.
1888 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1889 * is zero. I_FREEING must be set when I_WILL_FREE is
1891 * I_FREEING Set when inode is about to be freed but still has dirty
1892 * pages or buffers attached or the inode itself is still
1894 * I_CLEAR Added by clear_inode(). In this state the inode is
1895 * clean and can be destroyed. Inode keeps I_FREEING.
1897 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1898 * prohibited for many purposes. iget() must wait for
1899 * the inode to be completely released, then create it
1900 * anew. Other functions will just ignore such inodes,
1901 * if appropriate. I_NEW is used for waiting.
1903 * I_SYNC Writeback of inode is running. The bit is set during
1904 * data writeback, and cleared with a wakeup on the bit
1905 * address once it is done. The bit is also used to pin
1906 * the inode in memory for flusher thread.
1908 * I_REFERENCED Marks the inode as recently references on the LRU list.
1910 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1912 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1913 * synchronize competing switching instances and to tell
1914 * wb stat updates to grab mapping->tree_lock. See
1915 * inode_switch_wb_work_fn() for details.
1917 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1919 #define I_DIRTY_SYNC (1 << 0)
1920 #define I_DIRTY_DATASYNC (1 << 1)
1921 #define I_DIRTY_PAGES (1 << 2)
1923 #define I_NEW (1 << __I_NEW)
1924 #define I_WILL_FREE (1 << 4)
1925 #define I_FREEING (1 << 5)
1926 #define I_CLEAR (1 << 6)
1928 #define I_SYNC (1 << __I_SYNC)
1929 #define I_REFERENCED (1 << 8)
1930 #define __I_DIO_WAKEUP 9
1931 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1932 #define I_LINKABLE (1 << 10)
1933 #define I_DIRTY_TIME (1 << 11)
1934 #define __I_DIRTY_TIME_EXPIRED 12
1935 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1936 #define I_WB_SWITCH (1 << 13)
1938 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1939 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1941 extern void __mark_inode_dirty(struct inode
*, int);
1942 static inline void mark_inode_dirty(struct inode
*inode
)
1944 __mark_inode_dirty(inode
, I_DIRTY
);
1947 static inline void mark_inode_dirty_sync(struct inode
*inode
)
1949 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
1952 extern void inc_nlink(struct inode
*inode
);
1953 extern void drop_nlink(struct inode
*inode
);
1954 extern void clear_nlink(struct inode
*inode
);
1955 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
1957 static inline void inode_inc_link_count(struct inode
*inode
)
1960 mark_inode_dirty(inode
);
1963 static inline void inode_dec_link_count(struct inode
*inode
)
1966 mark_inode_dirty(inode
);
1970 * inode_inc_iversion - increments i_version
1971 * @inode: inode that need to be updated
1973 * Every time the inode is modified, the i_version field will be incremented.
1974 * The filesystem has to be mounted with i_version flag
1977 static inline void inode_inc_iversion(struct inode
*inode
)
1979 spin_lock(&inode
->i_lock
);
1981 spin_unlock(&inode
->i_lock
);
1984 enum file_time_flags
{
1991 extern bool atime_needs_update(const struct path
*, struct inode
*);
1992 extern void touch_atime(const struct path
*);
1993 static inline void file_accessed(struct file
*file
)
1995 if (!(file
->f_flags
& O_NOATIME
))
1996 touch_atime(&file
->f_path
);
1999 int sync_inode(struct inode
*inode
, struct writeback_control
*wbc
);
2000 int sync_inode_metadata(struct inode
*inode
, int wait
);
2002 struct file_system_type
{
2005 #define FS_REQUIRES_DEV 1
2006 #define FS_BINARY_MOUNTDATA 2
2007 #define FS_HAS_SUBTYPE 4
2008 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2009 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
2010 #define FS_USERNS_VISIBLE 32 /* FS must already be visible */
2011 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2012 struct dentry
*(*mount
) (struct file_system_type
*, int,
2013 const char *, void *);
2014 void (*kill_sb
) (struct super_block
*);
2015 struct module
*owner
;
2016 struct file_system_type
* next
;
2017 struct hlist_head fs_supers
;
2019 struct lock_class_key s_lock_key
;
2020 struct lock_class_key s_umount_key
;
2021 struct lock_class_key s_vfs_rename_key
;
2022 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
2024 struct lock_class_key i_lock_key
;
2025 struct lock_class_key i_mutex_key
;
2026 struct lock_class_key i_mutex_dir_key
;
2029 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2031 extern struct dentry
*mount_ns(struct file_system_type
*fs_type
, int flags
,
2032 void *data
, int (*fill_super
)(struct super_block
*, void *, int));
2033 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
2034 int flags
, const char *dev_name
, void *data
,
2035 int (*fill_super
)(struct super_block
*, void *, int));
2036 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
2037 int flags
, void *data
,
2038 int (*fill_super
)(struct super_block
*, void *, int));
2039 extern struct dentry
*mount_nodev(struct file_system_type
*fs_type
,
2040 int flags
, void *data
,
2041 int (*fill_super
)(struct super_block
*, void *, int));
2042 extern struct dentry
*mount_subtree(struct vfsmount
*mnt
, const char *path
);
2043 void generic_shutdown_super(struct super_block
*sb
);
2044 void kill_block_super(struct super_block
*sb
);
2045 void kill_anon_super(struct super_block
*sb
);
2046 void kill_litter_super(struct super_block
*sb
);
2047 void deactivate_super(struct super_block
*sb
);
2048 void deactivate_locked_super(struct super_block
*sb
);
2049 int set_anon_super(struct super_block
*s
, void *data
);
2050 int get_anon_bdev(dev_t
*);
2051 void free_anon_bdev(dev_t
);
2052 struct super_block
*sget(struct file_system_type
*type
,
2053 int (*test
)(struct super_block
*,void *),
2054 int (*set
)(struct super_block
*,void *),
2055 int flags
, void *data
);
2056 extern struct dentry
*mount_pseudo(struct file_system_type
*, char *,
2057 const struct super_operations
*ops
,
2058 const struct dentry_operations
*dops
,
2061 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2062 #define fops_get(fops) \
2063 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2064 #define fops_put(fops) \
2065 do { if (fops) module_put((fops)->owner); } while(0)
2067 * This one is to be used *ONLY* from ->open() instances.
2068 * fops must be non-NULL, pinned down *and* module dependencies
2069 * should be sufficient to pin the caller down as well.
2071 #define replace_fops(f, fops) \
2073 struct file *__file = (f); \
2074 fops_put(__file->f_op); \
2075 BUG_ON(!(__file->f_op = (fops))); \
2078 extern int register_filesystem(struct file_system_type
*);
2079 extern int unregister_filesystem(struct file_system_type
*);
2080 extern struct vfsmount
*kern_mount_data(struct file_system_type
*, void *data
);
2081 #define kern_mount(type) kern_mount_data(type, NULL)
2082 extern void kern_unmount(struct vfsmount
*mnt
);
2083 extern int may_umount_tree(struct vfsmount
*);
2084 extern int may_umount(struct vfsmount
*);
2085 extern long do_mount(const char *, const char __user
*,
2086 const char *, unsigned long, void *);
2087 extern struct vfsmount
*collect_mounts(struct path
*);
2088 extern void drop_collected_mounts(struct vfsmount
*);
2089 extern int iterate_mounts(int (*)(struct vfsmount
*, void *), void *,
2091 extern int vfs_statfs(struct path
*, struct kstatfs
*);
2092 extern int user_statfs(const char __user
*, struct kstatfs
*);
2093 extern int fd_statfs(int, struct kstatfs
*);
2094 extern int vfs_ustat(dev_t
, struct kstatfs
*);
2095 extern int freeze_super(struct super_block
*super
);
2096 extern int thaw_super(struct super_block
*super
);
2097 extern bool our_mnt(struct vfsmount
*mnt
);
2099 extern int current_umask(void);
2101 extern void ihold(struct inode
* inode
);
2102 extern void iput(struct inode
*);
2103 extern int generic_update_time(struct inode
*, struct timespec
*, int);
2106 extern struct kobject
*fs_kobj
;
2108 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2110 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2111 extern int locks_mandatory_locked(struct file
*);
2112 extern int locks_mandatory_area(struct inode
*, struct file
*, loff_t
, loff_t
, unsigned char);
2115 * Candidates for mandatory locking have the setgid bit set
2116 * but no group execute bit - an otherwise meaningless combination.
2119 static inline int __mandatory_lock(struct inode
*ino
)
2121 return (ino
->i_mode
& (S_ISGID
| S_IXGRP
)) == S_ISGID
;
2125 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2126 * otherwise these will be advisory locks
2129 static inline int mandatory_lock(struct inode
*ino
)
2131 return IS_MANDLOCK(ino
) && __mandatory_lock(ino
);
2134 static inline int locks_verify_locked(struct file
*file
)
2136 if (mandatory_lock(file_inode(file
)))
2137 return locks_mandatory_locked(file
);
2141 static inline int locks_verify_truncate(struct inode
*inode
,
2145 if (!inode
->i_flctx
|| !mandatory_lock(inode
))
2148 if (size
< inode
->i_size
) {
2149 return locks_mandatory_area(inode
, f
, size
, inode
->i_size
- 1,
2152 return locks_mandatory_area(inode
, f
, inode
->i_size
, size
- 1,
2157 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2159 static inline int locks_mandatory_locked(struct file
*file
)
2164 static inline int locks_mandatory_area(struct inode
*inode
, struct file
*filp
,
2165 loff_t start
, loff_t end
, unsigned char type
)
2170 static inline int __mandatory_lock(struct inode
*inode
)
2175 static inline int mandatory_lock(struct inode
*inode
)
2180 static inline int locks_verify_locked(struct file
*file
)
2185 static inline int locks_verify_truncate(struct inode
*inode
, struct file
*filp
,
2191 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2194 #ifdef CONFIG_FILE_LOCKING
2195 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2198 * Since this check is lockless, we must ensure that any refcounts
2199 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2200 * could end up racing with tasks trying to set a new lease on this
2204 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2205 return __break_lease(inode
, mode
, FL_LEASE
);
2209 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2212 * Since this check is lockless, we must ensure that any refcounts
2213 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2214 * could end up racing with tasks trying to set a new lease on this
2218 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2219 return __break_lease(inode
, mode
, FL_DELEG
);
2223 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2227 ret
= break_deleg(inode
, O_WRONLY
|O_NONBLOCK
);
2228 if (ret
== -EWOULDBLOCK
&& delegated_inode
) {
2229 *delegated_inode
= inode
;
2235 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2239 ret
= break_deleg(*delegated_inode
, O_WRONLY
);
2240 iput(*delegated_inode
);
2241 *delegated_inode
= NULL
;
2245 static inline int break_layout(struct inode
*inode
, bool wait
)
2248 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2249 return __break_lease(inode
,
2250 wait
? O_WRONLY
: O_WRONLY
| O_NONBLOCK
,
2255 #else /* !CONFIG_FILE_LOCKING */
2256 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2261 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2266 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2271 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2277 static inline int break_layout(struct inode
*inode
, bool wait
)
2282 #endif /* CONFIG_FILE_LOCKING */
2287 const char *name
; /* pointer to actual string */
2288 const __user
char *uptr
; /* original userland pointer */
2289 struct audit_names
*aname
;
2294 extern long vfs_truncate(const struct path
*, loff_t
);
2295 extern int do_truncate(struct dentry
*, loff_t start
, unsigned int time_attrs
,
2297 extern int vfs_fallocate(struct file
*file
, int mode
, loff_t offset
,
2299 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
2301 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
2302 extern struct file
*filp_open(const char *, int, umode_t
);
2303 extern struct file
*file_open_root(struct dentry
*, struct vfsmount
*,
2304 const char *, int, umode_t
);
2305 extern struct file
* dentry_open(const struct path
*, int, const struct cred
*);
2306 extern int filp_close(struct file
*, fl_owner_t id
);
2308 extern struct filename
*getname_flags(const char __user
*, int, int *);
2309 extern struct filename
*getname(const char __user
*);
2310 extern struct filename
*getname_kernel(const char *);
2311 extern void putname(struct filename
*name
);
2317 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
2318 int (*open
)(struct inode
*, struct file
*),
2320 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
2324 extern int ioctl_preallocate(struct file
*filp
, void __user
*argp
);
2327 extern void __init
vfs_caches_init_early(void);
2328 extern void __init
vfs_caches_init(void);
2330 extern struct kmem_cache
*names_cachep
;
2332 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2333 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2336 extern int register_blkdev(unsigned int, const char *);
2337 extern void unregister_blkdev(unsigned int, const char *);
2338 extern struct block_device
*bdget(dev_t
);
2339 extern struct block_device
*bdgrab(struct block_device
*bdev
);
2340 extern void bd_set_size(struct block_device
*, loff_t size
);
2341 extern void bd_forget(struct inode
*inode
);
2342 extern void bdput(struct block_device
*);
2343 extern void invalidate_bdev(struct block_device
*);
2344 extern void iterate_bdevs(void (*)(struct block_device
*, void *), void *);
2345 extern int sync_blockdev(struct block_device
*bdev
);
2346 extern void kill_bdev(struct block_device
*);
2347 extern struct super_block
*freeze_bdev(struct block_device
*);
2348 extern void emergency_thaw_all(void);
2349 extern int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
);
2350 extern int fsync_bdev(struct block_device
*);
2352 extern struct super_block
*blockdev_superblock
;
2354 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2356 return sb
== blockdev_superblock
;
2359 static inline void bd_forget(struct inode
*inode
) {}
2360 static inline int sync_blockdev(struct block_device
*bdev
) { return 0; }
2361 static inline void kill_bdev(struct block_device
*bdev
) {}
2362 static inline void invalidate_bdev(struct block_device
*bdev
) {}
2364 static inline struct super_block
*freeze_bdev(struct block_device
*sb
)
2369 static inline int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
)
2374 static inline void iterate_bdevs(void (*f
)(struct block_device
*, void *), void *arg
)
2378 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2383 extern int sync_filesystem(struct super_block
*);
2384 extern const struct file_operations def_blk_fops
;
2385 extern const struct file_operations def_chr_fops
;
2387 extern int ioctl_by_bdev(struct block_device
*, unsigned, unsigned long);
2388 extern int blkdev_ioctl(struct block_device
*, fmode_t
, unsigned, unsigned long);
2389 extern long compat_blkdev_ioctl(struct file
*, unsigned, unsigned long);
2390 extern int blkdev_get(struct block_device
*bdev
, fmode_t mode
, void *holder
);
2391 extern struct block_device
*blkdev_get_by_path(const char *path
, fmode_t mode
,
2393 extern struct block_device
*blkdev_get_by_dev(dev_t dev
, fmode_t mode
,
2395 extern void blkdev_put(struct block_device
*bdev
, fmode_t mode
);
2396 extern int __blkdev_reread_part(struct block_device
*bdev
);
2397 extern int blkdev_reread_part(struct block_device
*bdev
);
2400 extern int bd_link_disk_holder(struct block_device
*bdev
, struct gendisk
*disk
);
2401 extern void bd_unlink_disk_holder(struct block_device
*bdev
,
2402 struct gendisk
*disk
);
2404 static inline int bd_link_disk_holder(struct block_device
*bdev
,
2405 struct gendisk
*disk
)
2409 static inline void bd_unlink_disk_holder(struct block_device
*bdev
,
2410 struct gendisk
*disk
)
2417 #define CHRDEV_MAJOR_HASH_SIZE 255
2418 /* Marks the bottom of the first segment of free char majors */
2419 #define CHRDEV_MAJOR_DYN_END 234
2420 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
2421 extern int register_chrdev_region(dev_t
, unsigned, const char *);
2422 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
2423 unsigned int count
, const char *name
,
2424 const struct file_operations
*fops
);
2425 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
2426 unsigned int count
, const char *name
);
2427 extern void unregister_chrdev_region(dev_t
, unsigned);
2428 extern void chrdev_show(struct seq_file
*,off_t
);
2430 static inline int register_chrdev(unsigned int major
, const char *name
,
2431 const struct file_operations
*fops
)
2433 return __register_chrdev(major
, 0, 256, name
, fops
);
2436 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2438 __unregister_chrdev(major
, 0, 256, name
);
2441 /* fs/block_dev.c */
2442 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2443 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2446 #define BLKDEV_MAJOR_HASH_SIZE 255
2447 extern const char *__bdevname(dev_t
, char *buffer
);
2448 extern const char *bdevname(struct block_device
*bdev
, char *buffer
);
2449 extern struct block_device
*lookup_bdev(const char *);
2450 extern void blkdev_show(struct seq_file
*,off_t
);
2453 #define BLKDEV_MAJOR_HASH_SIZE 0
2456 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2458 /* Invalid inode operations -- fs/bad_inode.c */
2459 extern void make_bad_inode(struct inode
*);
2460 extern bool is_bad_inode(struct inode
*);
2463 static inline bool op_is_write(unsigned int op
)
2465 return op
== REQ_OP_READ
? false : true;
2469 * return data direction, READ or WRITE
2471 static inline int bio_data_dir(struct bio
*bio
)
2473 return op_is_write(bio_op(bio
)) ? WRITE
: READ
;
2476 extern void check_disk_size_change(struct gendisk
*disk
,
2477 struct block_device
*bdev
);
2478 extern int revalidate_disk(struct gendisk
*);
2479 extern int check_disk_change(struct block_device
*);
2480 extern int __invalidate_device(struct block_device
*, bool);
2481 extern int invalidate_partition(struct gendisk
*, int);
2483 unsigned long invalidate_mapping_pages(struct address_space
*mapping
,
2484 pgoff_t start
, pgoff_t end
);
2486 static inline void invalidate_remote_inode(struct inode
*inode
)
2488 if (S_ISREG(inode
->i_mode
) || S_ISDIR(inode
->i_mode
) ||
2489 S_ISLNK(inode
->i_mode
))
2490 invalidate_mapping_pages(inode
->i_mapping
, 0, -1);
2492 extern int invalidate_inode_pages2(struct address_space
*mapping
);
2493 extern int invalidate_inode_pages2_range(struct address_space
*mapping
,
2494 pgoff_t start
, pgoff_t end
);
2495 extern int write_inode_now(struct inode
*, int);
2496 extern int filemap_fdatawrite(struct address_space
*);
2497 extern int filemap_flush(struct address_space
*);
2498 extern int filemap_fdatawait(struct address_space
*);
2499 extern void filemap_fdatawait_keep_errors(struct address_space
*);
2500 extern int filemap_fdatawait_range(struct address_space
*, loff_t lstart
,
2502 extern int filemap_write_and_wait(struct address_space
*mapping
);
2503 extern int filemap_write_and_wait_range(struct address_space
*mapping
,
2504 loff_t lstart
, loff_t lend
);
2505 extern int __filemap_fdatawrite_range(struct address_space
*mapping
,
2506 loff_t start
, loff_t end
, int sync_mode
);
2507 extern int filemap_fdatawrite_range(struct address_space
*mapping
,
2508 loff_t start
, loff_t end
);
2510 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
2512 extern int vfs_fsync(struct file
*file
, int datasync
);
2515 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2516 * to already be updated for the write, and will return either the amount
2517 * of bytes passed in, or an error if syncing the file failed.
2519 static inline ssize_t
generic_write_sync(struct kiocb
*iocb
, ssize_t count
)
2521 if (iocb
->ki_flags
& IOCB_DSYNC
) {
2522 int ret
= vfs_fsync_range(iocb
->ki_filp
,
2523 iocb
->ki_pos
- count
, iocb
->ki_pos
- 1,
2524 (iocb
->ki_flags
& IOCB_SYNC
) ? 0 : 1);
2532 extern void emergency_sync(void);
2533 extern void emergency_remount(void);
2535 extern sector_t
bmap(struct inode
*, sector_t
);
2537 extern int notify_change(struct dentry
*, struct iattr
*, struct inode
**);
2538 extern int inode_permission(struct inode
*, int);
2539 extern int __inode_permission(struct inode
*, int);
2540 extern int generic_permission(struct inode
*, int);
2541 extern int __check_sticky(struct inode
*dir
, struct inode
*inode
);
2543 static inline bool execute_ok(struct inode
*inode
)
2545 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
2548 static inline void file_start_write(struct file
*file
)
2550 if (!S_ISREG(file_inode(file
)->i_mode
))
2552 __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, true);
2555 static inline bool file_start_write_trylock(struct file
*file
)
2557 if (!S_ISREG(file_inode(file
)->i_mode
))
2559 return __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, false);
2562 static inline void file_end_write(struct file
*file
)
2564 if (!S_ISREG(file_inode(file
)->i_mode
))
2566 __sb_end_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
);
2570 * get_write_access() gets write permission for a file.
2571 * put_write_access() releases this write permission.
2572 * This is used for regular files.
2573 * We cannot support write (and maybe mmap read-write shared) accesses and
2574 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2575 * can have the following values:
2576 * 0: no writers, no VM_DENYWRITE mappings
2577 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2578 * > 0: (i_writecount) users are writing to the file.
2580 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2581 * except for the cases where we don't hold i_writecount yet. Then we need to
2582 * use {get,deny}_write_access() - these functions check the sign and refuse
2583 * to do the change if sign is wrong.
2585 static inline int get_write_access(struct inode
*inode
)
2587 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2589 static inline int deny_write_access(struct file
*file
)
2591 struct inode
*inode
= file_inode(file
);
2592 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2594 static inline void put_write_access(struct inode
* inode
)
2596 atomic_dec(&inode
->i_writecount
);
2598 static inline void allow_write_access(struct file
*file
)
2601 atomic_inc(&file_inode(file
)->i_writecount
);
2603 static inline bool inode_is_open_for_write(const struct inode
*inode
)
2605 return atomic_read(&inode
->i_writecount
) > 0;
2609 static inline void i_readcount_dec(struct inode
*inode
)
2611 BUG_ON(!atomic_read(&inode
->i_readcount
));
2612 atomic_dec(&inode
->i_readcount
);
2614 static inline void i_readcount_inc(struct inode
*inode
)
2616 atomic_inc(&inode
->i_readcount
);
2619 static inline void i_readcount_dec(struct inode
*inode
)
2623 static inline void i_readcount_inc(struct inode
*inode
)
2628 extern int do_pipe_flags(int *, int);
2630 #define __kernel_read_file_id(id) \
2631 id(UNKNOWN, unknown) \
2632 id(FIRMWARE, firmware) \
2633 id(MODULE, kernel-module) \
2634 id(KEXEC_IMAGE, kexec-image) \
2635 id(KEXEC_INITRAMFS, kexec-initramfs) \
2636 id(POLICY, security-policy) \
2639 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2640 #define __fid_stringify(dummy, str) #str,
2642 enum kernel_read_file_id
{
2643 __kernel_read_file_id(__fid_enumify
)
2646 static const char * const kernel_read_file_str
[] = {
2647 __kernel_read_file_id(__fid_stringify
)
2650 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id
)
2652 if (id
< 0 || id
>= READING_MAX_ID
)
2653 return kernel_read_file_str
[READING_UNKNOWN
];
2655 return kernel_read_file_str
[id
];
2658 extern int kernel_read(struct file
*, loff_t
, char *, unsigned long);
2659 extern int kernel_read_file(struct file
*, void **, loff_t
*, loff_t
,
2660 enum kernel_read_file_id
);
2661 extern int kernel_read_file_from_path(char *, void **, loff_t
*, loff_t
,
2662 enum kernel_read_file_id
);
2663 extern int kernel_read_file_from_fd(int, void **, loff_t
*, loff_t
,
2664 enum kernel_read_file_id
);
2665 extern ssize_t
kernel_write(struct file
*, const char *, size_t, loff_t
);
2666 extern ssize_t
__kernel_write(struct file
*, const char *, size_t, loff_t
*);
2667 extern struct file
* open_exec(const char *);
2669 /* fs/dcache.c -- generic fs support functions */
2670 extern bool is_subdir(struct dentry
*, struct dentry
*);
2671 extern bool path_is_under(struct path
*, struct path
*);
2673 extern char *file_path(struct file
*, char *, int);
2675 #include <linux/err.h>
2677 /* needed for stackable file system support */
2678 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int whence
);
2680 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int whence
);
2682 extern int inode_init_always(struct super_block
*, struct inode
*);
2683 extern void inode_init_once(struct inode
*);
2684 extern void address_space_init_once(struct address_space
*mapping
);
2685 extern struct inode
* igrab(struct inode
*);
2686 extern ino_t
iunique(struct super_block
*, ino_t
);
2687 extern int inode_needs_sync(struct inode
*inode
);
2688 extern int generic_delete_inode(struct inode
*inode
);
2689 static inline int generic_drop_inode(struct inode
*inode
)
2691 return !inode
->i_nlink
|| inode_unhashed(inode
);
2694 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
2695 unsigned long hashval
, int (*test
)(struct inode
*, void *),
2697 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
2698 int (*test
)(struct inode
*, void *), void *data
);
2699 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
2701 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
2702 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
2703 extern struct inode
*find_inode_nowait(struct super_block
*,
2705 int (*match
)(struct inode
*,
2706 unsigned long, void *),
2708 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
2709 extern int insert_inode_locked(struct inode
*);
2710 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2711 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
2713 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
2715 extern void unlock_new_inode(struct inode
*);
2716 extern unsigned int get_next_ino(void);
2718 extern void __iget(struct inode
* inode
);
2719 extern void iget_failed(struct inode
*);
2720 extern void clear_inode(struct inode
*);
2721 extern void __destroy_inode(struct inode
*);
2722 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
2723 extern struct inode
*new_inode(struct super_block
*sb
);
2724 extern void free_inode_nonrcu(struct inode
*inode
);
2725 extern int should_remove_suid(struct dentry
*);
2726 extern int file_remove_privs(struct file
*);
2727 extern int dentry_needs_remove_privs(struct dentry
*dentry
);
2728 static inline int file_needs_remove_privs(struct file
*file
)
2730 return dentry_needs_remove_privs(file
->f_path
.dentry
);
2733 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
2734 static inline void insert_inode_hash(struct inode
*inode
)
2736 __insert_inode_hash(inode
, inode
->i_ino
);
2739 extern void __remove_inode_hash(struct inode
*);
2740 static inline void remove_inode_hash(struct inode
*inode
)
2742 if (!inode_unhashed(inode
) && !hlist_fake(&inode
->i_hash
))
2743 __remove_inode_hash(inode
);
2746 extern void inode_sb_list_add(struct inode
*inode
);
2749 extern blk_qc_t
submit_bio(struct bio
*);
2750 extern int bdev_read_only(struct block_device
*);
2752 extern int set_blocksize(struct block_device
*, int);
2753 extern int sb_set_blocksize(struct super_block
*, int);
2754 extern int sb_min_blocksize(struct super_block
*, int);
2756 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
2757 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
2758 extern ssize_t
generic_write_checks(struct kiocb
*, struct iov_iter
*);
2759 extern ssize_t
generic_file_read_iter(struct kiocb
*, struct iov_iter
*);
2760 extern ssize_t
__generic_file_write_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_direct_write(struct kiocb
*, struct iov_iter
*);
2763 extern ssize_t
generic_perform_write(struct file
*, struct iov_iter
*, loff_t
);
2765 ssize_t
vfs_iter_read(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
);
2766 ssize_t
vfs_iter_write(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
);
2768 /* fs/block_dev.c */
2769 extern ssize_t
blkdev_read_iter(struct kiocb
*iocb
, struct iov_iter
*to
);
2770 extern ssize_t
blkdev_write_iter(struct kiocb
*iocb
, struct iov_iter
*from
);
2771 extern int blkdev_fsync(struct file
*filp
, loff_t start
, loff_t end
,
2773 extern void block_sync_page(struct page
*page
);
2776 extern ssize_t
generic_file_splice_read(struct file
*, loff_t
*,
2777 struct pipe_inode_info
*, size_t, unsigned int);
2778 extern ssize_t
default_file_splice_read(struct file
*, loff_t
*,
2779 struct pipe_inode_info
*, size_t, unsigned int);
2780 extern ssize_t
iter_file_splice_write(struct pipe_inode_info
*,
2781 struct file
*, loff_t
*, size_t, unsigned int);
2782 extern ssize_t
generic_splice_sendpage(struct pipe_inode_info
*pipe
,
2783 struct file
*out
, loff_t
*, size_t len
, unsigned int flags
);
2784 extern long do_splice_direct(struct file
*in
, loff_t
*ppos
, struct file
*out
,
2785 loff_t
*opos
, size_t len
, unsigned int flags
);
2789 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
2790 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int whence
);
2791 extern loff_t
no_llseek(struct file
*file
, loff_t offset
, int whence
);
2792 extern loff_t
vfs_setpos(struct file
*file
, loff_t offset
, loff_t maxsize
);
2793 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int whence
);
2794 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
2795 int whence
, loff_t maxsize
, loff_t eof
);
2796 extern loff_t
fixed_size_llseek(struct file
*file
, loff_t offset
,
2797 int whence
, loff_t size
);
2798 extern loff_t
no_seek_end_llseek_size(struct file
*, loff_t
, int, loff_t
);
2799 extern loff_t
no_seek_end_llseek(struct file
*, loff_t
, int);
2800 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
2801 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
2804 typedef void (dio_submit_t
)(struct bio
*bio
, struct inode
*inode
,
2805 loff_t file_offset
);
2808 /* need locking between buffered and direct access */
2811 /* filesystem does not support filling holes */
2812 DIO_SKIP_HOLES
= 0x02,
2814 /* filesystem can handle aio writes beyond i_size */
2815 DIO_ASYNC_EXTEND
= 0x04,
2817 /* inode/fs/bdev does not need truncate protection */
2818 DIO_SKIP_DIO_COUNT
= 0x08,
2821 void dio_end_io(struct bio
*bio
, int error
);
2823 ssize_t
__blockdev_direct_IO(struct kiocb
*iocb
, struct inode
*inode
,
2824 struct block_device
*bdev
, struct iov_iter
*iter
,
2825 get_block_t get_block
,
2826 dio_iodone_t end_io
, dio_submit_t submit_io
,
2829 static inline ssize_t
blockdev_direct_IO(struct kiocb
*iocb
,
2830 struct inode
*inode
,
2831 struct iov_iter
*iter
,
2832 get_block_t get_block
)
2834 return __blockdev_direct_IO(iocb
, inode
, inode
->i_sb
->s_bdev
, iter
,
2835 get_block
, NULL
, NULL
, DIO_LOCKING
| DIO_SKIP_HOLES
);
2839 void inode_dio_wait(struct inode
*inode
);
2842 * inode_dio_begin - signal start of a direct I/O requests
2843 * @inode: inode the direct I/O happens on
2845 * This is called once we've finished processing a direct I/O request,
2846 * and is used to wake up callers waiting for direct I/O to be quiesced.
2848 static inline void inode_dio_begin(struct inode
*inode
)
2850 atomic_inc(&inode
->i_dio_count
);
2854 * inode_dio_end - signal finish of a direct I/O requests
2855 * @inode: inode the direct I/O happens on
2857 * This is called once we've finished processing a direct I/O request,
2858 * and is used to wake up callers waiting for direct I/O to be quiesced.
2860 static inline void inode_dio_end(struct inode
*inode
)
2862 if (atomic_dec_and_test(&inode
->i_dio_count
))
2863 wake_up_bit(&inode
->i_state
, __I_DIO_WAKEUP
);
2866 extern void inode_set_flags(struct inode
*inode
, unsigned int flags
,
2869 extern const struct file_operations generic_ro_fops
;
2871 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2873 extern int readlink_copy(char __user
*, int, const char *);
2874 extern int page_readlink(struct dentry
*, char __user
*, int);
2875 extern const char *page_get_link(struct dentry
*, struct inode
*,
2876 struct delayed_call
*);
2877 extern void page_put_link(void *);
2878 extern int __page_symlink(struct inode
*inode
, const char *symname
, int len
,
2880 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
2881 extern const struct inode_operations page_symlink_inode_operations
;
2882 extern void kfree_link(void *);
2883 extern int generic_readlink(struct dentry
*, char __user
*, int);
2884 extern void generic_fillattr(struct inode
*, struct kstat
*);
2885 int vfs_getattr_nosec(struct path
*path
, struct kstat
*stat
);
2886 extern int vfs_getattr(struct path
*, struct kstat
*);
2887 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2888 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2889 void __inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2890 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2891 loff_t
inode_get_bytes(struct inode
*inode
);
2892 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
2893 const char *simple_get_link(struct dentry
*, struct inode
*,
2894 struct delayed_call
*);
2895 extern const struct inode_operations simple_symlink_inode_operations
;
2897 extern int iterate_dir(struct file
*, struct dir_context
*);
2899 extern int vfs_stat(const char __user
*, struct kstat
*);
2900 extern int vfs_lstat(const char __user
*, struct kstat
*);
2901 extern int vfs_fstat(unsigned int, struct kstat
*);
2902 extern int vfs_fstatat(int , const char __user
*, struct kstat
*, int);
2904 extern int __generic_block_fiemap(struct inode
*inode
,
2905 struct fiemap_extent_info
*fieinfo
,
2906 loff_t start
, loff_t len
,
2907 get_block_t
*get_block
);
2908 extern int generic_block_fiemap(struct inode
*inode
,
2909 struct fiemap_extent_info
*fieinfo
, u64 start
,
2910 u64 len
, get_block_t
*get_block
);
2912 extern void get_filesystem(struct file_system_type
*fs
);
2913 extern void put_filesystem(struct file_system_type
*fs
);
2914 extern struct file_system_type
*get_fs_type(const char *name
);
2915 extern struct super_block
*get_super(struct block_device
*);
2916 extern struct super_block
*get_super_thawed(struct block_device
*);
2917 extern struct super_block
*get_active_super(struct block_device
*bdev
);
2918 extern void drop_super(struct super_block
*sb
);
2919 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
2920 extern void iterate_supers_type(struct file_system_type
*,
2921 void (*)(struct super_block
*, void *), void *);
2923 extern int dcache_dir_open(struct inode
*, struct file
*);
2924 extern int dcache_dir_close(struct inode
*, struct file
*);
2925 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
2926 extern int dcache_readdir(struct file
*, struct dir_context
*);
2927 extern int simple_setattr(struct dentry
*, struct iattr
*);
2928 extern int simple_getattr(struct vfsmount
*, struct dentry
*, struct kstat
*);
2929 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
2930 extern int simple_open(struct inode
*inode
, struct file
*file
);
2931 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
2932 extern int simple_unlink(struct inode
*, struct dentry
*);
2933 extern int simple_rmdir(struct inode
*, struct dentry
*);
2934 extern int simple_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*);
2935 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
2936 extern int simple_empty(struct dentry
*);
2937 extern int simple_readpage(struct file
*file
, struct page
*page
);
2938 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
2939 loff_t pos
, unsigned len
, unsigned flags
,
2940 struct page
**pagep
, void **fsdata
);
2941 extern int simple_write_end(struct file
*file
, struct address_space
*mapping
,
2942 loff_t pos
, unsigned len
, unsigned copied
,
2943 struct page
*page
, void *fsdata
);
2944 extern int always_delete_dentry(const struct dentry
*);
2945 extern struct inode
*alloc_anon_inode(struct super_block
*);
2946 extern int simple_nosetlease(struct file
*, long, struct file_lock
**, void **);
2947 extern const struct dentry_operations simple_dentry_operations
;
2949 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
2950 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
2951 extern const struct file_operations simple_dir_operations
;
2952 extern const struct inode_operations simple_dir_inode_operations
;
2953 extern void make_empty_dir_inode(struct inode
*inode
);
2954 extern bool is_empty_dir_inode(struct inode
*inode
);
2955 struct tree_descr
{ char *name
; const struct file_operations
*ops
; int mode
; };
2956 struct dentry
*d_alloc_name(struct dentry
*, const char *);
2957 extern int simple_fill_super(struct super_block
*, unsigned long, struct tree_descr
*);
2958 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
2959 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
2961 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
2962 loff_t
*ppos
, const void *from
, size_t available
);
2963 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
2964 const void __user
*from
, size_t count
);
2966 extern int __generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
2967 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
2969 extern int generic_check_addressable(unsigned, u64
);
2971 #ifdef CONFIG_MIGRATION
2972 extern int buffer_migrate_page(struct address_space
*,
2973 struct page
*, struct page
*,
2976 #define buffer_migrate_page NULL
2979 extern int inode_change_ok(const struct inode
*, struct iattr
*);
2980 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
2981 extern void setattr_copy(struct inode
*inode
, const struct iattr
*attr
);
2983 extern int file_update_time(struct file
*file
);
2985 extern int generic_show_options(struct seq_file
*m
, struct dentry
*root
);
2986 extern void save_mount_options(struct super_block
*sb
, char *options
);
2987 extern void replace_mount_options(struct super_block
*sb
, char *options
);
2989 static inline bool io_is_direct(struct file
*filp
)
2991 return (filp
->f_flags
& O_DIRECT
) || IS_DAX(filp
->f_mapping
->host
);
2994 static inline int iocb_flags(struct file
*file
)
2997 if (file
->f_flags
& O_APPEND
)
2999 if (io_is_direct(file
))
3001 if ((file
->f_flags
& O_DSYNC
) || IS_SYNC(file
->f_mapping
->host
))
3003 if (file
->f_flags
& __O_SYNC
)
3008 static inline ino_t
parent_ino(struct dentry
*dentry
)
3013 * Don't strictly need d_lock here? If the parent ino could change
3014 * then surely we'd have a deeper race in the caller?
3016 spin_lock(&dentry
->d_lock
);
3017 res
= dentry
->d_parent
->d_inode
->i_ino
;
3018 spin_unlock(&dentry
->d_lock
);
3022 /* Transaction based IO helpers */
3025 * An argresp is stored in an allocated page and holds the
3026 * size of the argument or response, along with its content
3028 struct simple_transaction_argresp
{
3033 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3035 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
3037 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
3038 size_t size
, loff_t
*pos
);
3039 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
3041 void simple_transaction_set(struct file
*file
, size_t n
);
3044 * simple attribute files
3046 * These attributes behave similar to those in sysfs:
3048 * Writing to an attribute immediately sets a value, an open file can be
3049 * written to multiple times.
3051 * Reading from an attribute creates a buffer from the value that might get
3052 * read with multiple read calls. When the attribute has been read
3053 * completely, no further read calls are possible until the file is opened
3056 * All attributes contain a text representation of a numeric value
3057 * that are accessed with the get() and set() functions.
3059 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3060 static int __fops ## _open(struct inode *inode, struct file *file) \
3062 __simple_attr_check_format(__fmt, 0ull); \
3063 return simple_attr_open(inode, file, __get, __set, __fmt); \
3065 static const struct file_operations __fops = { \
3066 .owner = THIS_MODULE, \
3067 .open = __fops ## _open, \
3068 .release = simple_attr_release, \
3069 .read = simple_attr_read, \
3070 .write = simple_attr_write, \
3071 .llseek = generic_file_llseek, \
3074 static inline __printf(1, 2)
3075 void __simple_attr_check_format(const char *fmt
, ...)
3077 /* don't do anything, just let the compiler check the arguments; */
3080 int simple_attr_open(struct inode
*inode
, struct file
*file
,
3081 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
3083 int simple_attr_release(struct inode
*inode
, struct file
*file
);
3084 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
3085 size_t len
, loff_t
*ppos
);
3086 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
3087 size_t len
, loff_t
*ppos
);
3090 int proc_nr_files(struct ctl_table
*table
, int write
,
3091 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3092 int proc_nr_dentry(struct ctl_table
*table
, int write
,
3093 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3094 int proc_nr_inodes(struct ctl_table
*table
, int write
,
3095 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3096 int __init
get_filesystem_list(char *buf
);
3098 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3099 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3101 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3102 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3103 (flag & __FMODE_NONOTIFY)))
3105 static inline bool is_sxid(umode_t mode
)
3107 return (mode
& S_ISUID
) || ((mode
& S_ISGID
) && (mode
& S_IXGRP
));
3110 static inline int check_sticky(struct inode
*dir
, struct inode
*inode
)
3112 if (!(dir
->i_mode
& S_ISVTX
))
3115 return __check_sticky(dir
, inode
);
3118 static inline void inode_has_no_xattr(struct inode
*inode
)
3120 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& MS_NOSEC
))
3121 inode
->i_flags
|= S_NOSEC
;
3124 static inline bool is_root_inode(struct inode
*inode
)
3126 return inode
== inode
->i_sb
->s_root
->d_inode
;
3129 static inline bool dir_emit(struct dir_context
*ctx
,
3130 const char *name
, int namelen
,
3131 u64 ino
, unsigned type
)
3133 return ctx
->actor(ctx
, name
, namelen
, ctx
->pos
, ino
, type
) == 0;
3135 static inline bool dir_emit_dot(struct file
*file
, struct dir_context
*ctx
)
3137 return ctx
->actor(ctx
, ".", 1, ctx
->pos
,
3138 file
->f_path
.dentry
->d_inode
->i_ino
, DT_DIR
) == 0;
3140 static inline bool dir_emit_dotdot(struct file
*file
, struct dir_context
*ctx
)
3142 return ctx
->actor(ctx
, "..", 2, ctx
->pos
,
3143 parent_ino(file
->f_path
.dentry
), DT_DIR
) == 0;
3145 static inline bool dir_emit_dots(struct file
*file
, struct dir_context
*ctx
)
3147 if (ctx
->pos
== 0) {
3148 if (!dir_emit_dot(file
, ctx
))
3152 if (ctx
->pos
== 1) {
3153 if (!dir_emit_dotdot(file
, ctx
))
3159 static inline bool dir_relax(struct inode
*inode
)
3161 inode_unlock(inode
);
3163 return !IS_DEADDIR(inode
);
3166 static inline bool dir_relax_shared(struct inode
*inode
)
3168 inode_unlock_shared(inode
);
3169 inode_lock_shared(inode
);
3170 return !IS_DEADDIR(inode
);
3173 extern bool path_noexec(const struct path
*path
);
3174 extern void inode_nohighmem(struct inode
*inode
);
3176 #endif /* _LINUX_FS_H */