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 bool (*isolate_page
)(struct page
*, isolate_mode_t
);
402 void (*putback_page
)(struct page
*);
403 int (*launder_page
) (struct page
*);
404 int (*is_partially_uptodate
) (struct page
*, unsigned long,
406 void (*is_dirty_writeback
) (struct page
*, bool *, bool *);
407 int (*error_remove_page
)(struct address_space
*, struct page
*);
409 /* swapfile support */
410 int (*swap_activate
)(struct swap_info_struct
*sis
, struct file
*file
,
412 void (*swap_deactivate
)(struct file
*file
);
415 extern const struct address_space_operations empty_aops
;
418 * pagecache_write_begin/pagecache_write_end must be used by general code
419 * to write into the pagecache.
421 int pagecache_write_begin(struct file
*, struct address_space
*mapping
,
422 loff_t pos
, unsigned len
, unsigned flags
,
423 struct page
**pagep
, void **fsdata
);
425 int pagecache_write_end(struct file
*, struct address_space
*mapping
,
426 loff_t pos
, unsigned len
, unsigned copied
,
427 struct page
*page
, void *fsdata
);
429 struct address_space
{
430 struct inode
*host
; /* owner: inode, block_device */
431 struct radix_tree_root page_tree
; /* radix tree of all pages */
432 spinlock_t tree_lock
; /* and lock protecting it */
433 atomic_t i_mmap_writable
;/* count VM_SHARED mappings */
434 struct rb_root i_mmap
; /* tree of private and shared mappings */
435 struct rw_semaphore i_mmap_rwsem
; /* protect tree, count, list */
436 /* Protected by tree_lock together with the radix tree */
437 unsigned long nrpages
; /* number of total pages */
438 /* number of shadow or DAX exceptional entries */
439 unsigned long nrexceptional
;
440 pgoff_t writeback_index
;/* writeback starts here */
441 const struct address_space_operations
*a_ops
; /* methods */
442 unsigned long flags
; /* error bits/gfp mask */
443 spinlock_t private_lock
; /* for use by the address_space */
444 struct list_head private_list
; /* ditto */
445 void *private_data
; /* ditto */
446 } __attribute__((aligned(sizeof(long))));
448 * On most architectures that alignment is already the case; but
449 * must be enforced here for CRIS, to let the least significant bit
450 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
452 struct request_queue
;
454 struct block_device
{
455 dev_t bd_dev
; /* not a kdev_t - it's a search key */
457 struct inode
* bd_inode
; /* will die */
458 struct super_block
* bd_super
;
459 struct mutex bd_mutex
; /* open/close mutex */
463 bool bd_write_holder
;
465 struct list_head bd_holder_disks
;
467 struct block_device
* bd_contains
;
468 unsigned bd_block_size
;
469 struct hd_struct
* bd_part
;
470 /* number of times partitions within this device have been opened. */
471 unsigned bd_part_count
;
473 struct gendisk
* bd_disk
;
474 struct request_queue
* bd_queue
;
475 struct list_head bd_list
;
477 * Private data. You must have bd_claim'ed the block_device
478 * to use this. NOTE: bd_claim allows an owner to claim
479 * the same device multiple times, the owner must take special
480 * care to not mess up bd_private for that case.
482 unsigned long bd_private
;
484 /* The counter of freeze processes */
485 int bd_fsfreeze_count
;
486 /* Mutex for freeze */
487 struct mutex bd_fsfreeze_mutex
;
491 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
494 #define PAGECACHE_TAG_DIRTY 0
495 #define PAGECACHE_TAG_WRITEBACK 1
496 #define PAGECACHE_TAG_TOWRITE 2
498 int mapping_tagged(struct address_space
*mapping
, int tag
);
500 static inline void i_mmap_lock_write(struct address_space
*mapping
)
502 down_write(&mapping
->i_mmap_rwsem
);
505 static inline void i_mmap_unlock_write(struct address_space
*mapping
)
507 up_write(&mapping
->i_mmap_rwsem
);
510 static inline void i_mmap_lock_read(struct address_space
*mapping
)
512 down_read(&mapping
->i_mmap_rwsem
);
515 static inline void i_mmap_unlock_read(struct address_space
*mapping
)
517 up_read(&mapping
->i_mmap_rwsem
);
521 * Might pages of this file be mapped into userspace?
523 static inline int mapping_mapped(struct address_space
*mapping
)
525 return !RB_EMPTY_ROOT(&mapping
->i_mmap
);
529 * Might pages of this file have been modified in userspace?
530 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
531 * marks vma as VM_SHARED if it is shared, and the file was opened for
532 * writing i.e. vma may be mprotected writable even if now readonly.
534 * If i_mmap_writable is negative, no new writable mappings are allowed. You
535 * can only deny writable mappings, if none exists right now.
537 static inline int mapping_writably_mapped(struct address_space
*mapping
)
539 return atomic_read(&mapping
->i_mmap_writable
) > 0;
542 static inline int mapping_map_writable(struct address_space
*mapping
)
544 return atomic_inc_unless_negative(&mapping
->i_mmap_writable
) ?
548 static inline void mapping_unmap_writable(struct address_space
*mapping
)
550 atomic_dec(&mapping
->i_mmap_writable
);
553 static inline int mapping_deny_writable(struct address_space
*mapping
)
555 return atomic_dec_unless_positive(&mapping
->i_mmap_writable
) ?
559 static inline void mapping_allow_writable(struct address_space
*mapping
)
561 atomic_inc(&mapping
->i_mmap_writable
);
565 * Use sequence counter to get consistent i_size on 32-bit processors.
567 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
568 #include <linux/seqlock.h>
569 #define __NEED_I_SIZE_ORDERED
570 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
572 #define i_size_ordered_init(inode) do { } while (0)
576 #define ACL_NOT_CACHED ((void *)(-1))
578 static inline struct posix_acl
*
579 uncached_acl_sentinel(struct task_struct
*task
)
581 return (void *)task
+ 1;
585 is_uncached_acl(struct posix_acl
*acl
)
587 return (long)acl
& 1;
590 #define IOP_FASTPERM 0x0001
591 #define IOP_LOOKUP 0x0002
592 #define IOP_NOFOLLOW 0x0004
595 * Keep mostly read-only and often accessed (especially for
596 * the RCU path lookup and 'stat' data) fields at the beginning
597 * of the 'struct inode'
601 unsigned short i_opflags
;
604 unsigned int i_flags
;
606 #ifdef CONFIG_FS_POSIX_ACL
607 struct posix_acl
*i_acl
;
608 struct posix_acl
*i_default_acl
;
611 const struct inode_operations
*i_op
;
612 struct super_block
*i_sb
;
613 struct address_space
*i_mapping
;
615 #ifdef CONFIG_SECURITY
619 /* Stat data, not accessed from path walking */
622 * Filesystems may only read i_nlink directly. They shall use the
623 * following functions for modification:
625 * (set|clear|inc|drop)_nlink
626 * inode_(inc|dec)_link_count
629 const unsigned int i_nlink
;
630 unsigned int __i_nlink
;
634 struct timespec i_atime
;
635 struct timespec i_mtime
;
636 struct timespec i_ctime
;
637 spinlock_t i_lock
; /* i_blocks, i_bytes, maybe i_size */
638 unsigned short i_bytes
;
639 unsigned int i_blkbits
;
642 #ifdef __NEED_I_SIZE_ORDERED
643 seqcount_t i_size_seqcount
;
647 unsigned long i_state
;
648 struct rw_semaphore i_rwsem
;
650 unsigned long dirtied_when
; /* jiffies of first dirtying */
651 unsigned long dirtied_time_when
;
653 struct hlist_node i_hash
;
654 struct list_head i_io_list
; /* backing dev IO list */
655 #ifdef CONFIG_CGROUP_WRITEBACK
656 struct bdi_writeback
*i_wb
; /* the associated cgroup wb */
658 /* foreign inode detection, see wbc_detach_inode() */
660 u16 i_wb_frn_avg_time
;
661 u16 i_wb_frn_history
;
663 struct list_head i_lru
; /* inode LRU list */
664 struct list_head i_sb_list
;
665 struct list_head i_wb_list
; /* backing dev writeback list */
667 struct hlist_head i_dentry
;
668 struct rcu_head i_rcu
;
672 atomic_t i_dio_count
;
673 atomic_t i_writecount
;
675 atomic_t i_readcount
; /* struct files open RO */
677 const struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
678 struct file_lock_context
*i_flctx
;
679 struct address_space i_data
;
680 struct list_head i_devices
;
682 struct pipe_inode_info
*i_pipe
;
683 struct block_device
*i_bdev
;
691 #ifdef CONFIG_FSNOTIFY
692 __u32 i_fsnotify_mask
; /* all events this inode cares about */
693 struct hlist_head i_fsnotify_marks
;
696 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
697 struct fscrypt_info
*i_crypt_info
;
700 void *i_private
; /* fs or device private pointer */
703 static inline int inode_unhashed(struct inode
*inode
)
705 return hlist_unhashed(&inode
->i_hash
);
709 * inode->i_mutex nesting subclasses for the lock validator:
711 * 0: the object of the current VFS operation
715 * 4: second non-directory
716 * 5: second parent (when locking independent directories in rename)
718 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
719 * non-directories at once.
721 * The locking order between these classes is
722 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
724 enum inode_i_mutex_lock_class
734 static inline void inode_lock(struct inode
*inode
)
736 down_write(&inode
->i_rwsem
);
739 static inline void inode_unlock(struct inode
*inode
)
741 up_write(&inode
->i_rwsem
);
744 static inline void inode_lock_shared(struct inode
*inode
)
746 down_read(&inode
->i_rwsem
);
749 static inline void inode_unlock_shared(struct inode
*inode
)
751 up_read(&inode
->i_rwsem
);
754 static inline int inode_trylock(struct inode
*inode
)
756 return down_write_trylock(&inode
->i_rwsem
);
759 static inline int inode_trylock_shared(struct inode
*inode
)
761 return down_read_trylock(&inode
->i_rwsem
);
764 static inline int inode_is_locked(struct inode
*inode
)
766 return rwsem_is_locked(&inode
->i_rwsem
);
769 static inline void inode_lock_nested(struct inode
*inode
, unsigned subclass
)
771 down_write_nested(&inode
->i_rwsem
, subclass
);
774 void lock_two_nondirectories(struct inode
*, struct inode
*);
775 void unlock_two_nondirectories(struct inode
*, struct inode
*);
778 * NOTE: in a 32bit arch with a preemptable kernel and
779 * an UP compile the i_size_read/write must be atomic
780 * with respect to the local cpu (unlike with preempt disabled),
781 * but they don't need to be atomic with respect to other cpus like in
782 * true SMP (so they need either to either locally disable irq around
783 * the read or for example on x86 they can be still implemented as a
784 * cmpxchg8b without the need of the lock prefix). For SMP compiles
785 * and 64bit archs it makes no difference if preempt is enabled or not.
787 static inline loff_t
i_size_read(const struct inode
*inode
)
789 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
794 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
795 i_size
= inode
->i_size
;
796 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
798 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
802 i_size
= inode
->i_size
;
806 return inode
->i_size
;
811 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
812 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
813 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
815 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
817 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
819 write_seqcount_begin(&inode
->i_size_seqcount
);
820 inode
->i_size
= i_size
;
821 write_seqcount_end(&inode
->i_size_seqcount
);
823 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
825 inode
->i_size
= i_size
;
828 inode
->i_size
= i_size
;
832 static inline unsigned iminor(const struct inode
*inode
)
834 return MINOR(inode
->i_rdev
);
837 static inline unsigned imajor(const struct inode
*inode
)
839 return MAJOR(inode
->i_rdev
);
842 extern struct block_device
*I_BDEV(struct inode
*inode
);
845 rwlock_t lock
; /* protects pid, uid, euid fields */
846 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
847 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
848 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
849 int signum
; /* posix.1b rt signal to be delivered on IO */
853 * Track a single file's readahead state
855 struct file_ra_state
{
856 pgoff_t start
; /* where readahead started */
857 unsigned int size
; /* # of readahead pages */
858 unsigned int async_size
; /* do asynchronous readahead when
859 there are only # of pages ahead */
861 unsigned int ra_pages
; /* Maximum readahead window */
862 unsigned int mmap_miss
; /* Cache miss stat for mmap accesses */
863 loff_t prev_pos
; /* Cache last read() position */
867 * Check if @index falls in the readahead windows.
869 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
871 return (index
>= ra
->start
&&
872 index
< ra
->start
+ ra
->size
);
877 struct llist_node fu_llist
;
878 struct rcu_head fu_rcuhead
;
881 struct inode
*f_inode
; /* cached value */
882 const struct file_operations
*f_op
;
885 * Protects f_ep_links, f_flags.
886 * Must not be taken from IRQ context.
889 atomic_long_t f_count
;
890 unsigned int f_flags
;
892 struct mutex f_pos_lock
;
894 struct fown_struct f_owner
;
895 const struct cred
*f_cred
;
896 struct file_ra_state f_ra
;
899 #ifdef CONFIG_SECURITY
902 /* needed for tty driver, and maybe others */
906 /* Used by fs/eventpoll.c to link all the hooks to this file */
907 struct list_head f_ep_links
;
908 struct list_head f_tfile_llink
;
909 #endif /* #ifdef CONFIG_EPOLL */
910 struct address_space
*f_mapping
;
911 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
916 /* file identifier */
917 unsigned char f_handle
[0];
920 static inline struct file
*get_file(struct file
*f
)
922 atomic_long_inc(&f
->f_count
);
925 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
926 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
927 #define file_count(x) atomic_long_read(&(x)->f_count)
929 #define MAX_NON_LFS ((1UL<<31) - 1)
931 /* Page cache limit. The filesystems should put that into their s_maxbytes
932 limits, otherwise bad things can happen in VM. */
933 #if BITS_PER_LONG==32
934 #define MAX_LFS_FILESIZE (((loff_t)PAGE_SIZE << (BITS_PER_LONG-1))-1)
935 #elif BITS_PER_LONG==64
936 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
941 #define FL_DELEG 4 /* NFSv4 delegation */
942 #define FL_ACCESS 8 /* not trying to lock, just looking */
943 #define FL_EXISTS 16 /* when unlocking, test for existence */
944 #define FL_LEASE 32 /* lease held on this file */
945 #define FL_CLOSE 64 /* unlock on close */
946 #define FL_SLEEP 128 /* A blocking lock */
947 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
948 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
949 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
950 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
953 * Special return value from posix_lock_file() and vfs_lock_file() for
954 * asynchronous locking.
956 #define FILE_LOCK_DEFERRED 1
958 /* legacy typedef, should eventually be removed */
959 typedef void *fl_owner_t
;
963 struct file_lock_operations
{
964 void (*fl_copy_lock
)(struct file_lock
*, struct file_lock
*);
965 void (*fl_release_private
)(struct file_lock
*);
968 struct lock_manager_operations
{
969 int (*lm_compare_owner
)(struct file_lock
*, struct file_lock
*);
970 unsigned long (*lm_owner_key
)(struct file_lock
*);
971 fl_owner_t (*lm_get_owner
)(fl_owner_t
);
972 void (*lm_put_owner
)(fl_owner_t
);
973 void (*lm_notify
)(struct file_lock
*); /* unblock callback */
974 int (*lm_grant
)(struct file_lock
*, int);
975 bool (*lm_break
)(struct file_lock
*);
976 int (*lm_change
)(struct file_lock
*, int, struct list_head
*);
977 void (*lm_setup
)(struct file_lock
*, void **);
980 struct lock_manager
{
981 struct list_head list
;
983 * NFSv4 and up also want opens blocked during the grace period;
990 void locks_start_grace(struct net
*, struct lock_manager
*);
991 void locks_end_grace(struct lock_manager
*);
992 int locks_in_grace(struct net
*);
993 int opens_in_grace(struct net
*);
995 /* that will die - we need it for nfs_lock_info */
996 #include <linux/nfs_fs_i.h>
999 * struct file_lock represents a generic "file lock". It's used to represent
1000 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1001 * note that the same struct is used to represent both a request for a lock and
1002 * the lock itself, but the same object is never used for both.
1004 * FIXME: should we create a separate "struct lock_request" to help distinguish
1007 * The varous i_flctx lists are ordered by:
1010 * 2) lock range start
1013 * Obviously, the last two criteria only matter for POSIX locks.
1016 struct file_lock
*fl_next
; /* singly linked list for this inode */
1017 struct list_head fl_list
; /* link into file_lock_context */
1018 struct hlist_node fl_link
; /* node in global lists */
1019 struct list_head fl_block
; /* circular list of blocked processes */
1020 fl_owner_t fl_owner
;
1021 unsigned int fl_flags
;
1022 unsigned char fl_type
;
1023 unsigned int fl_pid
;
1024 int fl_link_cpu
; /* what cpu's list is this on? */
1025 struct pid
*fl_nspid
;
1026 wait_queue_head_t fl_wait
;
1027 struct file
*fl_file
;
1031 struct fasync_struct
* fl_fasync
; /* for lease break notifications */
1032 /* for lease breaks: */
1033 unsigned long fl_break_time
;
1034 unsigned long fl_downgrade_time
;
1036 const struct file_lock_operations
*fl_ops
; /* Callbacks for filesystems */
1037 const struct lock_manager_operations
*fl_lmops
; /* Callbacks for lockmanagers */
1039 struct nfs_lock_info nfs_fl
;
1040 struct nfs4_lock_info nfs4_fl
;
1042 struct list_head link
; /* link in AFS vnode's pending_locks list */
1043 int state
; /* state of grant or error if -ve */
1048 struct file_lock_context
{
1049 spinlock_t flc_lock
;
1050 struct list_head flc_flock
;
1051 struct list_head flc_posix
;
1052 struct list_head flc_lease
;
1055 /* The following constant reflects the upper bound of the file/locking space */
1057 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1058 #define OFFSET_MAX INT_LIMIT(loff_t)
1059 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1062 #include <linux/fcntl.h>
1064 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
1066 #ifdef CONFIG_FILE_LOCKING
1067 extern int fcntl_getlk(struct file
*, unsigned int, struct flock __user
*);
1068 extern int fcntl_setlk(unsigned int, struct file
*, unsigned int,
1069 struct flock __user
*);
1071 #if BITS_PER_LONG == 32
1072 extern int fcntl_getlk64(struct file
*, unsigned int, struct flock64 __user
*);
1073 extern int fcntl_setlk64(unsigned int, struct file
*, unsigned int,
1074 struct flock64 __user
*);
1077 extern int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
);
1078 extern int fcntl_getlease(struct file
*filp
);
1081 void locks_free_lock_context(struct inode
*inode
);
1082 void locks_free_lock(struct file_lock
*fl
);
1083 extern void locks_init_lock(struct file_lock
*);
1084 extern struct file_lock
* locks_alloc_lock(void);
1085 extern void locks_copy_lock(struct file_lock
*, struct file_lock
*);
1086 extern void locks_copy_conflock(struct file_lock
*, struct file_lock
*);
1087 extern void locks_remove_posix(struct file
*, fl_owner_t
);
1088 extern void locks_remove_file(struct file
*);
1089 extern void locks_release_private(struct file_lock
*);
1090 extern void posix_test_lock(struct file
*, struct file_lock
*);
1091 extern int posix_lock_file(struct file
*, struct file_lock
*, struct file_lock
*);
1092 extern int posix_unblock_lock(struct file_lock
*);
1093 extern int vfs_test_lock(struct file
*, struct file_lock
*);
1094 extern int vfs_lock_file(struct file
*, unsigned int, struct file_lock
*, struct file_lock
*);
1095 extern int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
);
1096 extern int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
);
1097 extern int __break_lease(struct inode
*inode
, unsigned int flags
, unsigned int type
);
1098 extern void lease_get_mtime(struct inode
*, struct timespec
*time
);
1099 extern int generic_setlease(struct file
*, long, struct file_lock
**, void **priv
);
1100 extern int vfs_setlease(struct file
*, long, struct file_lock
**, void **);
1101 extern int lease_modify(struct file_lock
*, int, struct list_head
*);
1102 struct files_struct
;
1103 extern void show_fd_locks(struct seq_file
*f
,
1104 struct file
*filp
, struct files_struct
*files
);
1105 #else /* !CONFIG_FILE_LOCKING */
1106 static inline int fcntl_getlk(struct file
*file
, unsigned int cmd
,
1107 struct flock __user
*user
)
1112 static inline int fcntl_setlk(unsigned int fd
, struct file
*file
,
1113 unsigned int cmd
, struct flock __user
*user
)
1118 #if BITS_PER_LONG == 32
1119 static inline int fcntl_getlk64(struct file
*file
, unsigned int cmd
,
1120 struct flock64 __user
*user
)
1125 static inline int fcntl_setlk64(unsigned int fd
, struct file
*file
,
1126 unsigned int cmd
, struct flock64 __user
*user
)
1131 static inline int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
)
1136 static inline int fcntl_getlease(struct file
*filp
)
1142 locks_free_lock_context(struct inode
*inode
)
1146 static inline void locks_init_lock(struct file_lock
*fl
)
1151 static inline void locks_copy_conflock(struct file_lock
*new, struct file_lock
*fl
)
1156 static inline void locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
1161 static inline void locks_remove_posix(struct file
*filp
, fl_owner_t owner
)
1166 static inline void locks_remove_file(struct file
*filp
)
1171 static inline void posix_test_lock(struct file
*filp
, struct file_lock
*fl
)
1176 static inline int posix_lock_file(struct file
*filp
, struct file_lock
*fl
,
1177 struct file_lock
*conflock
)
1182 static inline int posix_unblock_lock(struct file_lock
*waiter
)
1187 static inline int vfs_test_lock(struct file
*filp
, struct file_lock
*fl
)
1192 static inline int vfs_lock_file(struct file
*filp
, unsigned int cmd
,
1193 struct file_lock
*fl
, struct file_lock
*conf
)
1198 static inline int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
)
1203 static inline int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
)
1208 static inline int __break_lease(struct inode
*inode
, unsigned int mode
, unsigned int type
)
1213 static inline void lease_get_mtime(struct inode
*inode
, struct timespec
*time
)
1218 static inline int generic_setlease(struct file
*filp
, long arg
,
1219 struct file_lock
**flp
, void **priv
)
1224 static inline int vfs_setlease(struct file
*filp
, long arg
,
1225 struct file_lock
**lease
, void **priv
)
1230 static inline int lease_modify(struct file_lock
*fl
, int arg
,
1231 struct list_head
*dispose
)
1236 struct files_struct
;
1237 static inline void show_fd_locks(struct seq_file
*f
,
1238 struct file
*filp
, struct files_struct
*files
) {}
1239 #endif /* !CONFIG_FILE_LOCKING */
1241 static inline struct inode
*file_inode(const struct file
*f
)
1246 static inline struct dentry
*file_dentry(const struct file
*file
)
1248 return d_real(file
->f_path
.dentry
, file_inode(file
), 0);
1251 static inline int locks_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1253 return locks_lock_inode_wait(file_inode(filp
), fl
);
1256 struct fasync_struct
{
1260 struct fasync_struct
*fa_next
; /* singly linked list */
1261 struct file
*fa_file
;
1262 struct rcu_head fa_rcu
;
1265 #define FASYNC_MAGIC 0x4601
1267 /* SMP safe fasync helpers: */
1268 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1269 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1270 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1271 extern struct fasync_struct
*fasync_alloc(void);
1272 extern void fasync_free(struct fasync_struct
*);
1274 /* can be called from interrupts */
1275 extern void kill_fasync(struct fasync_struct
**, int, int);
1277 extern void __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1278 extern void f_setown(struct file
*filp
, unsigned long arg
, int force
);
1279 extern void f_delown(struct file
*filp
);
1280 extern pid_t
f_getown(struct file
*filp
);
1281 extern int send_sigurg(struct fown_struct
*fown
);
1289 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1290 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1291 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1292 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1293 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1296 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1297 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1298 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1300 /* sb->s_iflags to limit user namespace mounts */
1301 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1303 /* Possible states of 'frozen' field */
1305 SB_UNFROZEN
= 0, /* FS is unfrozen */
1306 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1307 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1308 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1309 * internal threads if needed) */
1310 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1313 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1316 int frozen
; /* Is sb frozen? */
1317 wait_queue_head_t wait_unfrozen
; /* for get_super_thawed() */
1318 struct percpu_rw_semaphore rw_sem
[SB_FREEZE_LEVELS
];
1321 struct super_block
{
1322 struct list_head s_list
; /* Keep this first */
1323 dev_t s_dev
; /* search index; _not_ kdev_t */
1324 unsigned char s_blocksize_bits
;
1325 unsigned long s_blocksize
;
1326 loff_t s_maxbytes
; /* Max file size */
1327 struct file_system_type
*s_type
;
1328 const struct super_operations
*s_op
;
1329 const struct dquot_operations
*dq_op
;
1330 const struct quotactl_ops
*s_qcop
;
1331 const struct export_operations
*s_export_op
;
1332 unsigned long s_flags
;
1333 unsigned long s_iflags
; /* internal SB_I_* flags */
1334 unsigned long s_magic
;
1335 struct dentry
*s_root
;
1336 struct rw_semaphore s_umount
;
1339 #ifdef CONFIG_SECURITY
1342 const struct xattr_handler
**s_xattr
;
1344 const struct fscrypt_operations
*s_cop
;
1346 struct hlist_bl_head s_anon
; /* anonymous dentries for (nfs) exporting */
1347 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1348 struct block_device
*s_bdev
;
1349 struct backing_dev_info
*s_bdi
;
1350 struct mtd_info
*s_mtd
;
1351 struct hlist_node s_instances
;
1352 unsigned int s_quota_types
; /* Bitmask of supported quota types */
1353 struct quota_info s_dquot
; /* Diskquota specific options */
1355 struct sb_writers s_writers
;
1357 char s_id
[32]; /* Informational name */
1358 u8 s_uuid
[16]; /* UUID */
1360 void *s_fs_info
; /* Filesystem private info */
1361 unsigned int s_max_links
;
1364 /* Granularity of c/m/atime in ns.
1365 Cannot be worse than a second */
1369 * The next field is for VFS *only*. No filesystems have any business
1370 * even looking at it. You had been warned.
1372 struct mutex s_vfs_rename_mutex
; /* Kludge */
1375 * Filesystem subtype. If non-empty the filesystem type field
1376 * in /proc/mounts will be "type.subtype"
1381 * Saved mount options for lazy filesystems using
1382 * generic_show_options()
1384 char __rcu
*s_options
;
1385 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1388 * Saved pool identifier for cleancache (-1 means none)
1390 int cleancache_poolid
;
1392 struct shrinker s_shrink
; /* per-sb shrinker handle */
1394 /* Number of inodes with nlink == 0 but still referenced */
1395 atomic_long_t s_remove_count
;
1397 /* Being remounted read-only */
1398 int s_readonly_remount
;
1400 /* AIO completions deferred from interrupt context */
1401 struct workqueue_struct
*s_dio_done_wq
;
1402 struct hlist_head s_pins
;
1405 * Owning user namespace and default context in which to
1406 * interpret filesystem uids, gids, quotas, device nodes,
1407 * xattrs and security labels.
1409 struct user_namespace
*s_user_ns
;
1412 * Keep the lru lists last in the structure so they always sit on their
1413 * own individual cachelines.
1415 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp
;
1416 struct list_lru s_inode_lru ____cacheline_aligned_in_smp
;
1417 struct rcu_head rcu
;
1418 struct work_struct destroy_work
;
1420 struct mutex s_sync_lock
; /* sync serialisation lock */
1423 * Indicates how deep in a filesystem stack this SB is
1427 /* s_inode_list_lock protects s_inodes */
1428 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp
;
1429 struct list_head s_inodes
; /* all inodes */
1431 spinlock_t s_inode_wblist_lock
;
1432 struct list_head s_inodes_wb
; /* writeback inodes */
1435 /* Helper functions so that in most cases filesystems will
1436 * not need to deal directly with kuid_t and kgid_t and can
1437 * instead deal with the raw numeric values that are stored
1438 * in the filesystem.
1440 static inline uid_t
i_uid_read(const struct inode
*inode
)
1442 return from_kuid(inode
->i_sb
->s_user_ns
, inode
->i_uid
);
1445 static inline gid_t
i_gid_read(const struct inode
*inode
)
1447 return from_kgid(inode
->i_sb
->s_user_ns
, inode
->i_gid
);
1450 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
1452 inode
->i_uid
= make_kuid(inode
->i_sb
->s_user_ns
, uid
);
1455 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
1457 inode
->i_gid
= make_kgid(inode
->i_sb
->s_user_ns
, gid
);
1460 extern struct timespec
current_fs_time(struct super_block
*sb
);
1463 * Snapshotting support.
1466 void __sb_end_write(struct super_block
*sb
, int level
);
1467 int __sb_start_write(struct super_block
*sb
, int level
, bool wait
);
1469 #define __sb_writers_acquired(sb, lev) \
1470 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1471 #define __sb_writers_release(sb, lev) \
1472 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1475 * sb_end_write - drop write access to a superblock
1476 * @sb: the super we wrote to
1478 * Decrement number of writers to the filesystem. Wake up possible waiters
1479 * wanting to freeze the filesystem.
1481 static inline void sb_end_write(struct super_block
*sb
)
1483 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1487 * sb_end_pagefault - drop write access to a superblock from a page fault
1488 * @sb: the super we wrote to
1490 * Decrement number of processes handling write page fault to the filesystem.
1491 * Wake up possible waiters wanting to freeze the filesystem.
1493 static inline void sb_end_pagefault(struct super_block
*sb
)
1495 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1499 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1500 * @sb: the super we wrote to
1502 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1503 * waiters wanting to freeze the filesystem.
1505 static inline void sb_end_intwrite(struct super_block
*sb
)
1507 __sb_end_write(sb
, SB_FREEZE_FS
);
1511 * sb_start_write - get write access to a superblock
1512 * @sb: the super we write to
1514 * When a process wants to write data or metadata to a file system (i.e. dirty
1515 * a page or an inode), it should embed the operation in a sb_start_write() -
1516 * sb_end_write() pair to get exclusion against file system freezing. This
1517 * function increments number of writers preventing freezing. If the file
1518 * system is already frozen, the function waits until the file system is
1521 * Since freeze protection behaves as a lock, users have to preserve
1522 * ordering of freeze protection and other filesystem locks. Generally,
1523 * freeze protection should be the outermost lock. In particular, we have:
1526 * -> i_mutex (write path, truncate, directory ops, ...)
1527 * -> s_umount (freeze_super, thaw_super)
1529 static inline void sb_start_write(struct super_block
*sb
)
1531 __sb_start_write(sb
, SB_FREEZE_WRITE
, true);
1534 static inline int sb_start_write_trylock(struct super_block
*sb
)
1536 return __sb_start_write(sb
, SB_FREEZE_WRITE
, false);
1540 * sb_start_pagefault - get write access to a superblock from a page fault
1541 * @sb: the super we write to
1543 * When a process starts handling write page fault, it should embed the
1544 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1545 * exclusion against file system freezing. This is needed since the page fault
1546 * is going to dirty a page. This function increments number of running page
1547 * faults preventing freezing. If the file system is already frozen, the
1548 * function waits until the file system is thawed.
1550 * Since page fault freeze protection behaves as a lock, users have to preserve
1551 * ordering of freeze protection and other filesystem locks. It is advised to
1552 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1553 * handling code implies lock dependency:
1556 * -> sb_start_pagefault
1558 static inline void sb_start_pagefault(struct super_block
*sb
)
1560 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
, true);
1564 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1565 * @sb: the super we write to
1567 * This is the third level of protection against filesystem freezing. It is
1568 * free for use by a filesystem. The only requirement is that it must rank
1569 * below sb_start_pagefault.
1571 * For example filesystem can call sb_start_intwrite() when starting a
1572 * transaction which somewhat eases handling of freezing for internal sources
1573 * of filesystem changes (internal fs threads, discarding preallocation on file
1576 static inline void sb_start_intwrite(struct super_block
*sb
)
1578 __sb_start_write(sb
, SB_FREEZE_FS
, true);
1582 extern bool inode_owner_or_capable(const struct inode
*inode
);
1585 * VFS helper functions..
1587 extern int vfs_create(struct inode
*, struct dentry
*, umode_t
, bool);
1588 extern int vfs_mkdir(struct inode
*, struct dentry
*, umode_t
);
1589 extern int vfs_mknod(struct inode
*, struct dentry
*, umode_t
, dev_t
);
1590 extern int vfs_symlink(struct inode
*, struct dentry
*, const char *);
1591 extern int vfs_link(struct dentry
*, struct inode
*, struct dentry
*, struct inode
**);
1592 extern int vfs_rmdir(struct inode
*, struct dentry
*);
1593 extern int vfs_unlink(struct inode
*, struct dentry
*, struct inode
**);
1594 extern int vfs_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*, struct inode
**, unsigned int);
1595 extern int vfs_whiteout(struct inode
*, struct dentry
*);
1598 * VFS file helper functions.
1600 extern void inode_init_owner(struct inode
*inode
, const struct inode
*dir
,
1602 extern bool may_open_dev(const struct path
*path
);
1604 * VFS FS_IOC_FIEMAP helper definitions.
1606 struct fiemap_extent_info
{
1607 unsigned int fi_flags
; /* Flags as passed from user */
1608 unsigned int fi_extents_mapped
; /* Number of mapped extents */
1609 unsigned int fi_extents_max
; /* Size of fiemap_extent array */
1610 struct fiemap_extent __user
*fi_extents_start
; /* Start of
1611 fiemap_extent array */
1613 int fiemap_fill_next_extent(struct fiemap_extent_info
*info
, u64 logical
,
1614 u64 phys
, u64 len
, u32 flags
);
1615 int fiemap_check_flags(struct fiemap_extent_info
*fieinfo
, u32 fs_flags
);
1620 * NOTE! These match bits 12..15 of stat.st_mode
1621 * (ie "(i_mode >> 12) & 15").
1623 #define DT_UNKNOWN 0
1634 * This is the "filldir" function type, used by readdir() to let
1635 * the kernel specify what kind of dirent layout it wants to have.
1636 * This allows the kernel to read directories into kernel space or
1637 * to have different dirent layouts depending on the binary type.
1640 typedef int (*filldir_t
)(struct dir_context
*, const char *, int, loff_t
, u64
,
1643 struct dir_context
{
1644 const filldir_t actor
;
1648 struct block_device_operations
;
1650 /* These macros are for out of kernel modules to test that
1651 * the kernel supports the unlocked_ioctl and compat_ioctl
1652 * fields in struct file_operations. */
1653 #define HAVE_COMPAT_IOCTL 1
1654 #define HAVE_UNLOCKED_IOCTL 1
1657 * These flags let !MMU mmap() govern direct device mapping vs immediate
1658 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1660 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1661 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1662 * NOMMU_MAP_READ: Can be mapped for reading
1663 * NOMMU_MAP_WRITE: Can be mapped for writing
1664 * NOMMU_MAP_EXEC: Can be mapped for execution
1666 #define NOMMU_MAP_COPY 0x00000001
1667 #define NOMMU_MAP_DIRECT 0x00000008
1668 #define NOMMU_MAP_READ VM_MAYREAD
1669 #define NOMMU_MAP_WRITE VM_MAYWRITE
1670 #define NOMMU_MAP_EXEC VM_MAYEXEC
1672 #define NOMMU_VMFLAGS \
1673 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1678 struct file_operations
{
1679 struct module
*owner
;
1680 loff_t (*llseek
) (struct file
*, loff_t
, int);
1681 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
1682 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
1683 ssize_t (*read_iter
) (struct kiocb
*, struct iov_iter
*);
1684 ssize_t (*write_iter
) (struct kiocb
*, struct iov_iter
*);
1685 int (*iterate
) (struct file
*, struct dir_context
*);
1686 int (*iterate_shared
) (struct file
*, struct dir_context
*);
1687 unsigned int (*poll
) (struct file
*, struct poll_table_struct
*);
1688 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
1689 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
1690 int (*mmap
) (struct file
*, struct vm_area_struct
*);
1691 int (*open
) (struct inode
*, struct file
*);
1692 int (*flush
) (struct file
*, fl_owner_t id
);
1693 int (*release
) (struct inode
*, struct file
*);
1694 int (*fsync
) (struct file
*, loff_t
, loff_t
, int datasync
);
1695 int (*aio_fsync
) (struct kiocb
*, int datasync
);
1696 int (*fasync
) (int, struct file
*, int);
1697 int (*lock
) (struct file
*, int, struct file_lock
*);
1698 ssize_t (*sendpage
) (struct file
*, struct page
*, int, size_t, loff_t
*, int);
1699 unsigned long (*get_unmapped_area
)(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1700 int (*check_flags
)(int);
1701 int (*flock
) (struct file
*, int, struct file_lock
*);
1702 ssize_t (*splice_write
)(struct pipe_inode_info
*, struct file
*, loff_t
*, size_t, unsigned int);
1703 ssize_t (*splice_read
)(struct file
*, loff_t
*, struct pipe_inode_info
*, size_t, unsigned int);
1704 int (*setlease
)(struct file
*, long, struct file_lock
**, void **);
1705 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
1707 void (*show_fdinfo
)(struct seq_file
*m
, struct file
*f
);
1709 unsigned (*mmap_capabilities
)(struct file
*);
1711 ssize_t (*copy_file_range
)(struct file
*, loff_t
, struct file
*,
1712 loff_t
, size_t, unsigned int);
1713 int (*clone_file_range
)(struct file
*, loff_t
, struct file
*, loff_t
,
1715 ssize_t (*dedupe_file_range
)(struct file
*, u64
, u64
, struct file
*,
1719 struct inode_operations
{
1720 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*, unsigned int);
1721 const char * (*get_link
) (struct dentry
*, struct inode
*, struct delayed_call
*);
1722 int (*permission
) (struct inode
*, int);
1723 struct posix_acl
* (*get_acl
)(struct inode
*, int);
1725 int (*readlink
) (struct dentry
*, char __user
*,int);
1727 int (*create
) (struct inode
*,struct dentry
*, umode_t
, bool);
1728 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
1729 int (*unlink
) (struct inode
*,struct dentry
*);
1730 int (*symlink
) (struct inode
*,struct dentry
*,const char *);
1731 int (*mkdir
) (struct inode
*,struct dentry
*,umode_t
);
1732 int (*rmdir
) (struct inode
*,struct dentry
*);
1733 int (*mknod
) (struct inode
*,struct dentry
*,umode_t
,dev_t
);
1734 int (*rename
) (struct inode
*, struct dentry
*,
1735 struct inode
*, struct dentry
*);
1736 int (*rename2
) (struct inode
*, struct dentry
*,
1737 struct inode
*, struct dentry
*, unsigned int);
1738 int (*setattr
) (struct dentry
*, struct iattr
*);
1739 int (*getattr
) (struct vfsmount
*mnt
, struct dentry
*, struct kstat
*);
1740 int (*setxattr
) (struct dentry
*, struct inode
*,
1741 const char *, const void *, size_t, int);
1742 ssize_t (*getxattr
) (struct dentry
*, struct inode
*,
1743 const char *, void *, size_t);
1744 ssize_t (*listxattr
) (struct dentry
*, char *, size_t);
1745 int (*removexattr
) (struct dentry
*, const char *);
1746 int (*fiemap
)(struct inode
*, struct fiemap_extent_info
*, u64 start
,
1748 int (*update_time
)(struct inode
*, struct timespec
*, int);
1749 int (*atomic_open
)(struct inode
*, struct dentry
*,
1750 struct file
*, unsigned open_flag
,
1751 umode_t create_mode
, int *opened
);
1752 int (*tmpfile
) (struct inode
*, struct dentry
*, umode_t
);
1753 int (*set_acl
)(struct inode
*, struct posix_acl
*, int);
1754 } ____cacheline_aligned
;
1756 ssize_t
rw_copy_check_uvector(int type
, const struct iovec __user
* uvector
,
1757 unsigned long nr_segs
, unsigned long fast_segs
,
1758 struct iovec
*fast_pointer
,
1759 struct iovec
**ret_pointer
);
1761 extern ssize_t
__vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1762 extern ssize_t
__vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1763 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1764 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1765 extern ssize_t
vfs_readv(struct file
*, const struct iovec __user
*,
1766 unsigned long, loff_t
*, int);
1767 extern ssize_t
vfs_writev(struct file
*, const struct iovec __user
*,
1768 unsigned long, loff_t
*, int);
1769 extern ssize_t
vfs_copy_file_range(struct file
*, loff_t
, struct file
*,
1770 loff_t
, size_t, unsigned int);
1771 extern int vfs_clone_file_range(struct file
*file_in
, loff_t pos_in
,
1772 struct file
*file_out
, loff_t pos_out
, u64 len
);
1773 extern int vfs_dedupe_file_range(struct file
*file
,
1774 struct file_dedupe_range
*same
);
1776 struct super_operations
{
1777 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
1778 void (*destroy_inode
)(struct inode
*);
1780 void (*dirty_inode
) (struct inode
*, int flags
);
1781 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
1782 int (*drop_inode
) (struct inode
*);
1783 void (*evict_inode
) (struct inode
*);
1784 void (*put_super
) (struct super_block
*);
1785 int (*sync_fs
)(struct super_block
*sb
, int wait
);
1786 int (*freeze_super
) (struct super_block
*);
1787 int (*freeze_fs
) (struct super_block
*);
1788 int (*thaw_super
) (struct super_block
*);
1789 int (*unfreeze_fs
) (struct super_block
*);
1790 int (*statfs
) (struct dentry
*, struct kstatfs
*);
1791 int (*remount_fs
) (struct super_block
*, int *, char *);
1792 void (*umount_begin
) (struct super_block
*);
1794 int (*show_options
)(struct seq_file
*, struct dentry
*);
1795 int (*show_devname
)(struct seq_file
*, struct dentry
*);
1796 int (*show_path
)(struct seq_file
*, struct dentry
*);
1797 int (*show_stats
)(struct seq_file
*, struct dentry
*);
1799 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
1800 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
1801 struct dquot
**(*get_dquots
)(struct inode
*);
1803 int (*bdev_try_to_free_page
)(struct super_block
*, struct page
*, gfp_t
);
1804 long (*nr_cached_objects
)(struct super_block
*,
1805 struct shrink_control
*);
1806 long (*free_cached_objects
)(struct super_block
*,
1807 struct shrink_control
*);
1811 * Inode flags - they have no relation to superblock flags now
1813 #define S_SYNC 1 /* Writes are synced at once */
1814 #define S_NOATIME 2 /* Do not update access times */
1815 #define S_APPEND 4 /* Append-only file */
1816 #define S_IMMUTABLE 8 /* Immutable file */
1817 #define S_DEAD 16 /* removed, but still open directory */
1818 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1819 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1820 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1821 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1822 #define S_PRIVATE 512 /* Inode is fs-internal */
1823 #define S_IMA 1024 /* Inode has an associated IMA struct */
1824 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1825 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1826 #ifdef CONFIG_FS_DAX
1827 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1829 #define S_DAX 0 /* Make all the DAX code disappear */
1833 * Note that nosuid etc flags are inode-specific: setting some file-system
1834 * flags just means all the inodes inherit those flags by default. It might be
1835 * possible to override it selectively if you really wanted to with some
1836 * ioctl() that is not currently implemented.
1838 * Exception: MS_RDONLY is always applied to the entire file system.
1840 * Unfortunately, it is possible to change a filesystems flags with it mounted
1841 * with files in use. This means that all of the inodes will not have their
1842 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1843 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1845 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1847 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1848 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1849 ((inode)->i_flags & S_SYNC))
1850 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1851 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1852 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1853 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1854 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1856 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1857 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1858 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1859 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1861 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1862 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1863 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1864 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1865 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1866 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1867 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1868 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1870 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1871 (inode)->i_rdev == WHITEOUT_DEV)
1873 static inline bool HAS_UNMAPPED_ID(struct inode
*inode
)
1875 return !uid_valid(inode
->i_uid
) || !gid_valid(inode
->i_gid
);
1879 * Inode state bits. Protected by inode->i_lock
1881 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1882 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1884 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1885 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1886 * various stages of removing an inode.
1888 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1890 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1891 * fdatasync(). i_atime is the usual cause.
1892 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1893 * these changes separately from I_DIRTY_SYNC so that we
1894 * don't have to write inode on fdatasync() when only
1895 * mtime has changed in it.
1896 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1897 * I_NEW Serves as both a mutex and completion notification.
1898 * New inodes set I_NEW. If two processes both create
1899 * the same inode, one of them will release its inode and
1900 * wait for I_NEW to be released before returning.
1901 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1902 * also cause waiting on I_NEW, without I_NEW actually
1903 * being set. find_inode() uses this to prevent returning
1904 * nearly-dead inodes.
1905 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1906 * is zero. I_FREEING must be set when I_WILL_FREE is
1908 * I_FREEING Set when inode is about to be freed but still has dirty
1909 * pages or buffers attached or the inode itself is still
1911 * I_CLEAR Added by clear_inode(). In this state the inode is
1912 * clean and can be destroyed. Inode keeps I_FREEING.
1914 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1915 * prohibited for many purposes. iget() must wait for
1916 * the inode to be completely released, then create it
1917 * anew. Other functions will just ignore such inodes,
1918 * if appropriate. I_NEW is used for waiting.
1920 * I_SYNC Writeback of inode is running. The bit is set during
1921 * data writeback, and cleared with a wakeup on the bit
1922 * address once it is done. The bit is also used to pin
1923 * the inode in memory for flusher thread.
1925 * I_REFERENCED Marks the inode as recently references on the LRU list.
1927 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1929 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1930 * synchronize competing switching instances and to tell
1931 * wb stat updates to grab mapping->tree_lock. See
1932 * inode_switch_wb_work_fn() for details.
1934 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1936 #define I_DIRTY_SYNC (1 << 0)
1937 #define I_DIRTY_DATASYNC (1 << 1)
1938 #define I_DIRTY_PAGES (1 << 2)
1940 #define I_NEW (1 << __I_NEW)
1941 #define I_WILL_FREE (1 << 4)
1942 #define I_FREEING (1 << 5)
1943 #define I_CLEAR (1 << 6)
1945 #define I_SYNC (1 << __I_SYNC)
1946 #define I_REFERENCED (1 << 8)
1947 #define __I_DIO_WAKEUP 9
1948 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1949 #define I_LINKABLE (1 << 10)
1950 #define I_DIRTY_TIME (1 << 11)
1951 #define __I_DIRTY_TIME_EXPIRED 12
1952 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1953 #define I_WB_SWITCH (1 << 13)
1955 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1956 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1958 extern void __mark_inode_dirty(struct inode
*, int);
1959 static inline void mark_inode_dirty(struct inode
*inode
)
1961 __mark_inode_dirty(inode
, I_DIRTY
);
1964 static inline void mark_inode_dirty_sync(struct inode
*inode
)
1966 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
1969 extern void inc_nlink(struct inode
*inode
);
1970 extern void drop_nlink(struct inode
*inode
);
1971 extern void clear_nlink(struct inode
*inode
);
1972 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
1974 static inline void inode_inc_link_count(struct inode
*inode
)
1977 mark_inode_dirty(inode
);
1980 static inline void inode_dec_link_count(struct inode
*inode
)
1983 mark_inode_dirty(inode
);
1987 * inode_inc_iversion - increments i_version
1988 * @inode: inode that need to be updated
1990 * Every time the inode is modified, the i_version field will be incremented.
1991 * The filesystem has to be mounted with i_version flag
1994 static inline void inode_inc_iversion(struct inode
*inode
)
1996 spin_lock(&inode
->i_lock
);
1998 spin_unlock(&inode
->i_lock
);
2001 enum file_time_flags
{
2008 extern bool atime_needs_update(const struct path
*, struct inode
*);
2009 extern void touch_atime(const struct path
*);
2010 static inline void file_accessed(struct file
*file
)
2012 if (!(file
->f_flags
& O_NOATIME
))
2013 touch_atime(&file
->f_path
);
2016 int sync_inode(struct inode
*inode
, struct writeback_control
*wbc
);
2017 int sync_inode_metadata(struct inode
*inode
, int wait
);
2019 struct file_system_type
{
2022 #define FS_REQUIRES_DEV 1
2023 #define FS_BINARY_MOUNTDATA 2
2024 #define FS_HAS_SUBTYPE 4
2025 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2026 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2027 struct dentry
*(*mount
) (struct file_system_type
*, int,
2028 const char *, void *);
2029 void (*kill_sb
) (struct super_block
*);
2030 struct module
*owner
;
2031 struct file_system_type
* next
;
2032 struct hlist_head fs_supers
;
2034 struct lock_class_key s_lock_key
;
2035 struct lock_class_key s_umount_key
;
2036 struct lock_class_key s_vfs_rename_key
;
2037 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
2039 struct lock_class_key i_lock_key
;
2040 struct lock_class_key i_mutex_key
;
2041 struct lock_class_key i_mutex_dir_key
;
2044 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2046 extern struct dentry
*mount_ns(struct file_system_type
*fs_type
,
2047 int flags
, void *data
, void *ns
, struct user_namespace
*user_ns
,
2048 int (*fill_super
)(struct super_block
*, void *, int));
2049 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
2050 int flags
, const char *dev_name
, void *data
,
2051 int (*fill_super
)(struct super_block
*, void *, int));
2052 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
2053 int flags
, void *data
,
2054 int (*fill_super
)(struct super_block
*, void *, int));
2055 extern struct dentry
*mount_nodev(struct file_system_type
*fs_type
,
2056 int flags
, void *data
,
2057 int (*fill_super
)(struct super_block
*, void *, int));
2058 extern struct dentry
*mount_subtree(struct vfsmount
*mnt
, const char *path
);
2059 void generic_shutdown_super(struct super_block
*sb
);
2060 void kill_block_super(struct super_block
*sb
);
2061 void kill_anon_super(struct super_block
*sb
);
2062 void kill_litter_super(struct super_block
*sb
);
2063 void deactivate_super(struct super_block
*sb
);
2064 void deactivate_locked_super(struct super_block
*sb
);
2065 int set_anon_super(struct super_block
*s
, void *data
);
2066 int get_anon_bdev(dev_t
*);
2067 void free_anon_bdev(dev_t
);
2068 struct super_block
*sget_userns(struct file_system_type
*type
,
2069 int (*test
)(struct super_block
*,void *),
2070 int (*set
)(struct super_block
*,void *),
2071 int flags
, struct user_namespace
*user_ns
,
2073 struct super_block
*sget(struct file_system_type
*type
,
2074 int (*test
)(struct super_block
*,void *),
2075 int (*set
)(struct super_block
*,void *),
2076 int flags
, void *data
);
2077 extern struct dentry
*mount_pseudo(struct file_system_type
*, char *,
2078 const struct super_operations
*ops
,
2079 const struct dentry_operations
*dops
,
2082 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2083 #define fops_get(fops) \
2084 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2085 #define fops_put(fops) \
2086 do { if (fops) module_put((fops)->owner); } while(0)
2088 * This one is to be used *ONLY* from ->open() instances.
2089 * fops must be non-NULL, pinned down *and* module dependencies
2090 * should be sufficient to pin the caller down as well.
2092 #define replace_fops(f, fops) \
2094 struct file *__file = (f); \
2095 fops_put(__file->f_op); \
2096 BUG_ON(!(__file->f_op = (fops))); \
2099 extern int register_filesystem(struct file_system_type
*);
2100 extern int unregister_filesystem(struct file_system_type
*);
2101 extern struct vfsmount
*kern_mount_data(struct file_system_type
*, void *data
);
2102 #define kern_mount(type) kern_mount_data(type, NULL)
2103 extern void kern_unmount(struct vfsmount
*mnt
);
2104 extern int may_umount_tree(struct vfsmount
*);
2105 extern int may_umount(struct vfsmount
*);
2106 extern long do_mount(const char *, const char __user
*,
2107 const char *, unsigned long, void *);
2108 extern struct vfsmount
*collect_mounts(struct path
*);
2109 extern void drop_collected_mounts(struct vfsmount
*);
2110 extern int iterate_mounts(int (*)(struct vfsmount
*, void *), void *,
2112 extern int vfs_statfs(struct path
*, struct kstatfs
*);
2113 extern int user_statfs(const char __user
*, struct kstatfs
*);
2114 extern int fd_statfs(int, struct kstatfs
*);
2115 extern int vfs_ustat(dev_t
, struct kstatfs
*);
2116 extern int freeze_super(struct super_block
*super
);
2117 extern int thaw_super(struct super_block
*super
);
2118 extern bool our_mnt(struct vfsmount
*mnt
);
2120 extern int current_umask(void);
2122 extern void ihold(struct inode
* inode
);
2123 extern void iput(struct inode
*);
2124 extern int generic_update_time(struct inode
*, struct timespec
*, int);
2127 extern struct kobject
*fs_kobj
;
2129 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2131 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2132 extern int locks_mandatory_locked(struct file
*);
2133 extern int locks_mandatory_area(struct inode
*, struct file
*, loff_t
, loff_t
, unsigned char);
2136 * Candidates for mandatory locking have the setgid bit set
2137 * but no group execute bit - an otherwise meaningless combination.
2140 static inline int __mandatory_lock(struct inode
*ino
)
2142 return (ino
->i_mode
& (S_ISGID
| S_IXGRP
)) == S_ISGID
;
2146 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2147 * otherwise these will be advisory locks
2150 static inline int mandatory_lock(struct inode
*ino
)
2152 return IS_MANDLOCK(ino
) && __mandatory_lock(ino
);
2155 static inline int locks_verify_locked(struct file
*file
)
2157 if (mandatory_lock(file_inode(file
)))
2158 return locks_mandatory_locked(file
);
2162 static inline int locks_verify_truncate(struct inode
*inode
,
2166 if (!inode
->i_flctx
|| !mandatory_lock(inode
))
2169 if (size
< inode
->i_size
) {
2170 return locks_mandatory_area(inode
, f
, size
, inode
->i_size
- 1,
2173 return locks_mandatory_area(inode
, f
, inode
->i_size
, size
- 1,
2178 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2180 static inline int locks_mandatory_locked(struct file
*file
)
2185 static inline int locks_mandatory_area(struct inode
*inode
, struct file
*filp
,
2186 loff_t start
, loff_t end
, unsigned char type
)
2191 static inline int __mandatory_lock(struct inode
*inode
)
2196 static inline int mandatory_lock(struct inode
*inode
)
2201 static inline int locks_verify_locked(struct file
*file
)
2206 static inline int locks_verify_truncate(struct inode
*inode
, struct file
*filp
,
2212 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2215 #ifdef CONFIG_FILE_LOCKING
2216 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2219 * Since this check is lockless, we must ensure that any refcounts
2220 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2221 * could end up racing with tasks trying to set a new lease on this
2225 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2226 return __break_lease(inode
, mode
, FL_LEASE
);
2230 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2233 * Since this check is lockless, we must ensure that any refcounts
2234 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2235 * could end up racing with tasks trying to set a new lease on this
2239 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2240 return __break_lease(inode
, mode
, FL_DELEG
);
2244 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2248 ret
= break_deleg(inode
, O_WRONLY
|O_NONBLOCK
);
2249 if (ret
== -EWOULDBLOCK
&& delegated_inode
) {
2250 *delegated_inode
= inode
;
2256 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2260 ret
= break_deleg(*delegated_inode
, O_WRONLY
);
2261 iput(*delegated_inode
);
2262 *delegated_inode
= NULL
;
2266 static inline int break_layout(struct inode
*inode
, bool wait
)
2269 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2270 return __break_lease(inode
,
2271 wait
? O_WRONLY
: O_WRONLY
| O_NONBLOCK
,
2276 #else /* !CONFIG_FILE_LOCKING */
2277 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2282 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2287 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2292 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2298 static inline int break_layout(struct inode
*inode
, bool wait
)
2303 #endif /* CONFIG_FILE_LOCKING */
2308 const char *name
; /* pointer to actual string */
2309 const __user
char *uptr
; /* original userland pointer */
2310 struct audit_names
*aname
;
2315 extern long vfs_truncate(const struct path
*, loff_t
);
2316 extern int do_truncate(struct dentry
*, loff_t start
, unsigned int time_attrs
,
2318 extern int vfs_fallocate(struct file
*file
, int mode
, loff_t offset
,
2320 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
2322 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
2323 extern struct file
*filp_open(const char *, int, umode_t
);
2324 extern struct file
*file_open_root(struct dentry
*, struct vfsmount
*,
2325 const char *, int, umode_t
);
2326 extern struct file
* dentry_open(const struct path
*, int, const struct cred
*);
2327 extern int filp_close(struct file
*, fl_owner_t id
);
2329 extern struct filename
*getname_flags(const char __user
*, int, int *);
2330 extern struct filename
*getname(const char __user
*);
2331 extern struct filename
*getname_kernel(const char *);
2332 extern void putname(struct filename
*name
);
2338 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
2339 int (*open
)(struct inode
*, struct file
*),
2341 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
2345 extern int ioctl_preallocate(struct file
*filp
, void __user
*argp
);
2348 extern void __init
vfs_caches_init_early(void);
2349 extern void __init
vfs_caches_init(void);
2351 extern struct kmem_cache
*names_cachep
;
2353 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2354 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2357 extern int register_blkdev(unsigned int, const char *);
2358 extern void unregister_blkdev(unsigned int, const char *);
2359 extern struct block_device
*bdget(dev_t
);
2360 extern struct block_device
*bdgrab(struct block_device
*bdev
);
2361 extern void bd_set_size(struct block_device
*, loff_t size
);
2362 extern void bd_forget(struct inode
*inode
);
2363 extern void bdput(struct block_device
*);
2364 extern void invalidate_bdev(struct block_device
*);
2365 extern void iterate_bdevs(void (*)(struct block_device
*, void *), void *);
2366 extern int sync_blockdev(struct block_device
*bdev
);
2367 extern void kill_bdev(struct block_device
*);
2368 extern struct super_block
*freeze_bdev(struct block_device
*);
2369 extern void emergency_thaw_all(void);
2370 extern int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
);
2371 extern int fsync_bdev(struct block_device
*);
2373 extern struct super_block
*blockdev_superblock
;
2375 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2377 return sb
== blockdev_superblock
;
2380 static inline void bd_forget(struct inode
*inode
) {}
2381 static inline int sync_blockdev(struct block_device
*bdev
) { return 0; }
2382 static inline void kill_bdev(struct block_device
*bdev
) {}
2383 static inline void invalidate_bdev(struct block_device
*bdev
) {}
2385 static inline struct super_block
*freeze_bdev(struct block_device
*sb
)
2390 static inline int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
)
2395 static inline void iterate_bdevs(void (*f
)(struct block_device
*, void *), void *arg
)
2399 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2404 extern int sync_filesystem(struct super_block
*);
2405 extern const struct file_operations def_blk_fops
;
2406 extern const struct file_operations def_chr_fops
;
2408 extern int ioctl_by_bdev(struct block_device
*, unsigned, unsigned long);
2409 extern int blkdev_ioctl(struct block_device
*, fmode_t
, unsigned, unsigned long);
2410 extern long compat_blkdev_ioctl(struct file
*, unsigned, unsigned long);
2411 extern int blkdev_get(struct block_device
*bdev
, fmode_t mode
, void *holder
);
2412 extern struct block_device
*blkdev_get_by_path(const char *path
, fmode_t mode
,
2414 extern struct block_device
*blkdev_get_by_dev(dev_t dev
, fmode_t mode
,
2416 extern void blkdev_put(struct block_device
*bdev
, fmode_t mode
);
2417 extern int __blkdev_reread_part(struct block_device
*bdev
);
2418 extern int blkdev_reread_part(struct block_device
*bdev
);
2421 extern int bd_link_disk_holder(struct block_device
*bdev
, struct gendisk
*disk
);
2422 extern void bd_unlink_disk_holder(struct block_device
*bdev
,
2423 struct gendisk
*disk
);
2425 static inline int bd_link_disk_holder(struct block_device
*bdev
,
2426 struct gendisk
*disk
)
2430 static inline void bd_unlink_disk_holder(struct block_device
*bdev
,
2431 struct gendisk
*disk
)
2438 #define CHRDEV_MAJOR_HASH_SIZE 255
2439 /* Marks the bottom of the first segment of free char majors */
2440 #define CHRDEV_MAJOR_DYN_END 234
2441 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
2442 extern int register_chrdev_region(dev_t
, unsigned, const char *);
2443 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
2444 unsigned int count
, const char *name
,
2445 const struct file_operations
*fops
);
2446 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
2447 unsigned int count
, const char *name
);
2448 extern void unregister_chrdev_region(dev_t
, unsigned);
2449 extern void chrdev_show(struct seq_file
*,off_t
);
2451 static inline int register_chrdev(unsigned int major
, const char *name
,
2452 const struct file_operations
*fops
)
2454 return __register_chrdev(major
, 0, 256, name
, fops
);
2457 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2459 __unregister_chrdev(major
, 0, 256, name
);
2462 /* fs/block_dev.c */
2463 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2464 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2467 #define BLKDEV_MAJOR_HASH_SIZE 255
2468 extern const char *__bdevname(dev_t
, char *buffer
);
2469 extern const char *bdevname(struct block_device
*bdev
, char *buffer
);
2470 extern struct block_device
*lookup_bdev(const char *);
2471 extern void blkdev_show(struct seq_file
*,off_t
);
2474 #define BLKDEV_MAJOR_HASH_SIZE 0
2477 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2479 /* Invalid inode operations -- fs/bad_inode.c */
2480 extern void make_bad_inode(struct inode
*);
2481 extern bool is_bad_inode(struct inode
*);
2483 static inline bool op_is_write(unsigned int op
)
2485 return op
== REQ_OP_READ
? false : true;
2491 * return data direction, READ or WRITE
2493 static inline int bio_data_dir(struct bio
*bio
)
2495 return op_is_write(bio_op(bio
)) ? WRITE
: READ
;
2498 extern void check_disk_size_change(struct gendisk
*disk
,
2499 struct block_device
*bdev
);
2500 extern int revalidate_disk(struct gendisk
*);
2501 extern int check_disk_change(struct block_device
*);
2502 extern int __invalidate_device(struct block_device
*, bool);
2503 extern int invalidate_partition(struct gendisk
*, int);
2505 unsigned long invalidate_mapping_pages(struct address_space
*mapping
,
2506 pgoff_t start
, pgoff_t end
);
2508 static inline void invalidate_remote_inode(struct inode
*inode
)
2510 if (S_ISREG(inode
->i_mode
) || S_ISDIR(inode
->i_mode
) ||
2511 S_ISLNK(inode
->i_mode
))
2512 invalidate_mapping_pages(inode
->i_mapping
, 0, -1);
2514 extern int invalidate_inode_pages2(struct address_space
*mapping
);
2515 extern int invalidate_inode_pages2_range(struct address_space
*mapping
,
2516 pgoff_t start
, pgoff_t end
);
2517 extern int write_inode_now(struct inode
*, int);
2518 extern int filemap_fdatawrite(struct address_space
*);
2519 extern int filemap_flush(struct address_space
*);
2520 extern int filemap_fdatawait(struct address_space
*);
2521 extern void filemap_fdatawait_keep_errors(struct address_space
*);
2522 extern int filemap_fdatawait_range(struct address_space
*, loff_t lstart
,
2524 extern int filemap_write_and_wait(struct address_space
*mapping
);
2525 extern int filemap_write_and_wait_range(struct address_space
*mapping
,
2526 loff_t lstart
, loff_t lend
);
2527 extern int __filemap_fdatawrite_range(struct address_space
*mapping
,
2528 loff_t start
, loff_t end
, int sync_mode
);
2529 extern int filemap_fdatawrite_range(struct address_space
*mapping
,
2530 loff_t start
, loff_t end
);
2531 extern int filemap_check_errors(struct address_space
*mapping
);
2533 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
2535 extern int vfs_fsync(struct file
*file
, int datasync
);
2538 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2539 * to already be updated for the write, and will return either the amount
2540 * of bytes passed in, or an error if syncing the file failed.
2542 static inline ssize_t
generic_write_sync(struct kiocb
*iocb
, ssize_t count
)
2544 if (iocb
->ki_flags
& IOCB_DSYNC
) {
2545 int ret
= vfs_fsync_range(iocb
->ki_filp
,
2546 iocb
->ki_pos
- count
, iocb
->ki_pos
- 1,
2547 (iocb
->ki_flags
& IOCB_SYNC
) ? 0 : 1);
2555 extern void emergency_sync(void);
2556 extern void emergency_remount(void);
2558 extern sector_t
bmap(struct inode
*, sector_t
);
2560 extern int notify_change(struct dentry
*, struct iattr
*, struct inode
**);
2561 extern int inode_permission(struct inode
*, int);
2562 extern int __inode_permission(struct inode
*, int);
2563 extern int generic_permission(struct inode
*, int);
2564 extern int __check_sticky(struct inode
*dir
, struct inode
*inode
);
2566 static inline bool execute_ok(struct inode
*inode
)
2568 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
2571 static inline void file_start_write(struct file
*file
)
2573 if (!S_ISREG(file_inode(file
)->i_mode
))
2575 __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, true);
2578 static inline bool file_start_write_trylock(struct file
*file
)
2580 if (!S_ISREG(file_inode(file
)->i_mode
))
2582 return __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, false);
2585 static inline void file_end_write(struct file
*file
)
2587 if (!S_ISREG(file_inode(file
)->i_mode
))
2589 __sb_end_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
);
2593 * get_write_access() gets write permission for a file.
2594 * put_write_access() releases this write permission.
2595 * This is used for regular files.
2596 * We cannot support write (and maybe mmap read-write shared) accesses and
2597 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2598 * can have the following values:
2599 * 0: no writers, no VM_DENYWRITE mappings
2600 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2601 * > 0: (i_writecount) users are writing to the file.
2603 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2604 * except for the cases where we don't hold i_writecount yet. Then we need to
2605 * use {get,deny}_write_access() - these functions check the sign and refuse
2606 * to do the change if sign is wrong.
2608 static inline int get_write_access(struct inode
*inode
)
2610 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2612 static inline int deny_write_access(struct file
*file
)
2614 struct inode
*inode
= file_inode(file
);
2615 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2617 static inline void put_write_access(struct inode
* inode
)
2619 atomic_dec(&inode
->i_writecount
);
2621 static inline void allow_write_access(struct file
*file
)
2624 atomic_inc(&file_inode(file
)->i_writecount
);
2626 static inline bool inode_is_open_for_write(const struct inode
*inode
)
2628 return atomic_read(&inode
->i_writecount
) > 0;
2632 static inline void i_readcount_dec(struct inode
*inode
)
2634 BUG_ON(!atomic_read(&inode
->i_readcount
));
2635 atomic_dec(&inode
->i_readcount
);
2637 static inline void i_readcount_inc(struct inode
*inode
)
2639 atomic_inc(&inode
->i_readcount
);
2642 static inline void i_readcount_dec(struct inode
*inode
)
2646 static inline void i_readcount_inc(struct inode
*inode
)
2651 extern int do_pipe_flags(int *, int);
2653 #define __kernel_read_file_id(id) \
2654 id(UNKNOWN, unknown) \
2655 id(FIRMWARE, firmware) \
2656 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2657 id(MODULE, kernel-module) \
2658 id(KEXEC_IMAGE, kexec-image) \
2659 id(KEXEC_INITRAMFS, kexec-initramfs) \
2660 id(POLICY, security-policy) \
2663 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2664 #define __fid_stringify(dummy, str) #str,
2666 enum kernel_read_file_id
{
2667 __kernel_read_file_id(__fid_enumify
)
2670 static const char * const kernel_read_file_str
[] = {
2671 __kernel_read_file_id(__fid_stringify
)
2674 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id
)
2676 if (id
< 0 || id
>= READING_MAX_ID
)
2677 return kernel_read_file_str
[READING_UNKNOWN
];
2679 return kernel_read_file_str
[id
];
2682 extern int kernel_read(struct file
*, loff_t
, char *, unsigned long);
2683 extern int kernel_read_file(struct file
*, void **, loff_t
*, loff_t
,
2684 enum kernel_read_file_id
);
2685 extern int kernel_read_file_from_path(char *, void **, loff_t
*, loff_t
,
2686 enum kernel_read_file_id
);
2687 extern int kernel_read_file_from_fd(int, void **, loff_t
*, loff_t
,
2688 enum kernel_read_file_id
);
2689 extern ssize_t
kernel_write(struct file
*, const char *, size_t, loff_t
);
2690 extern ssize_t
__kernel_write(struct file
*, const char *, size_t, loff_t
*);
2691 extern struct file
* open_exec(const char *);
2693 /* fs/dcache.c -- generic fs support functions */
2694 extern bool is_subdir(struct dentry
*, struct dentry
*);
2695 extern bool path_is_under(struct path
*, struct path
*);
2697 extern char *file_path(struct file
*, char *, int);
2699 #include <linux/err.h>
2701 /* needed for stackable file system support */
2702 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int whence
);
2704 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int whence
);
2706 extern int inode_init_always(struct super_block
*, struct inode
*);
2707 extern void inode_init_once(struct inode
*);
2708 extern void address_space_init_once(struct address_space
*mapping
);
2709 extern struct inode
* igrab(struct inode
*);
2710 extern ino_t
iunique(struct super_block
*, ino_t
);
2711 extern int inode_needs_sync(struct inode
*inode
);
2712 extern int generic_delete_inode(struct inode
*inode
);
2713 static inline int generic_drop_inode(struct inode
*inode
)
2715 return !inode
->i_nlink
|| inode_unhashed(inode
);
2718 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
2719 unsigned long hashval
, int (*test
)(struct inode
*, void *),
2721 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
2722 int (*test
)(struct inode
*, void *), void *data
);
2723 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
2725 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
2726 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
2727 extern struct inode
*find_inode_nowait(struct super_block
*,
2729 int (*match
)(struct inode
*,
2730 unsigned long, void *),
2732 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
2733 extern int insert_inode_locked(struct inode
*);
2734 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2735 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
2737 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
2739 extern void unlock_new_inode(struct inode
*);
2740 extern unsigned int get_next_ino(void);
2742 extern void __iget(struct inode
* inode
);
2743 extern void iget_failed(struct inode
*);
2744 extern void clear_inode(struct inode
*);
2745 extern void __destroy_inode(struct inode
*);
2746 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
2747 extern struct inode
*new_inode(struct super_block
*sb
);
2748 extern void free_inode_nonrcu(struct inode
*inode
);
2749 extern int should_remove_suid(struct dentry
*);
2750 extern int file_remove_privs(struct file
*);
2751 extern int dentry_needs_remove_privs(struct dentry
*dentry
);
2752 static inline int file_needs_remove_privs(struct file
*file
)
2754 return dentry_needs_remove_privs(file
->f_path
.dentry
);
2757 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
2758 static inline void insert_inode_hash(struct inode
*inode
)
2760 __insert_inode_hash(inode
, inode
->i_ino
);
2763 extern void __remove_inode_hash(struct inode
*);
2764 static inline void remove_inode_hash(struct inode
*inode
)
2766 if (!inode_unhashed(inode
) && !hlist_fake(&inode
->i_hash
))
2767 __remove_inode_hash(inode
);
2770 extern void inode_sb_list_add(struct inode
*inode
);
2773 extern blk_qc_t
submit_bio(struct bio
*);
2774 extern int bdev_read_only(struct block_device
*);
2776 extern int set_blocksize(struct block_device
*, int);
2777 extern int sb_set_blocksize(struct super_block
*, int);
2778 extern int sb_min_blocksize(struct super_block
*, int);
2780 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
2781 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
2782 extern ssize_t
generic_write_checks(struct kiocb
*, struct iov_iter
*);
2783 extern ssize_t
generic_file_read_iter(struct kiocb
*, struct iov_iter
*);
2784 extern ssize_t
__generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2785 extern ssize_t
generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2786 extern ssize_t
generic_file_direct_write(struct kiocb
*, struct iov_iter
*);
2787 extern ssize_t
generic_perform_write(struct file
*, struct iov_iter
*, loff_t
);
2789 ssize_t
vfs_iter_read(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
);
2790 ssize_t
vfs_iter_write(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
);
2792 /* fs/block_dev.c */
2793 extern ssize_t
blkdev_read_iter(struct kiocb
*iocb
, struct iov_iter
*to
);
2794 extern ssize_t
blkdev_write_iter(struct kiocb
*iocb
, struct iov_iter
*from
);
2795 extern int blkdev_fsync(struct file
*filp
, loff_t start
, loff_t end
,
2797 extern void block_sync_page(struct page
*page
);
2800 extern ssize_t
generic_file_splice_read(struct file
*, loff_t
*,
2801 struct pipe_inode_info
*, size_t, unsigned int);
2802 extern ssize_t
default_file_splice_read(struct file
*, loff_t
*,
2803 struct pipe_inode_info
*, size_t, unsigned int);
2804 extern ssize_t
iter_file_splice_write(struct pipe_inode_info
*,
2805 struct file
*, loff_t
*, size_t, unsigned int);
2806 extern ssize_t
generic_splice_sendpage(struct pipe_inode_info
*pipe
,
2807 struct file
*out
, loff_t
*, size_t len
, unsigned int flags
);
2808 extern long do_splice_direct(struct file
*in
, loff_t
*ppos
, struct file
*out
,
2809 loff_t
*opos
, size_t len
, unsigned int flags
);
2813 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
2814 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int whence
);
2815 extern loff_t
no_llseek(struct file
*file
, loff_t offset
, int whence
);
2816 extern loff_t
vfs_setpos(struct file
*file
, loff_t offset
, loff_t maxsize
);
2817 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int whence
);
2818 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
2819 int whence
, loff_t maxsize
, loff_t eof
);
2820 extern loff_t
fixed_size_llseek(struct file
*file
, loff_t offset
,
2821 int whence
, loff_t size
);
2822 extern loff_t
no_seek_end_llseek_size(struct file
*, loff_t
, int, loff_t
);
2823 extern loff_t
no_seek_end_llseek(struct file
*, loff_t
, int);
2824 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
2825 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
2828 typedef void (dio_submit_t
)(struct bio
*bio
, struct inode
*inode
,
2829 loff_t file_offset
);
2832 /* need locking between buffered and direct access */
2835 /* filesystem does not support filling holes */
2836 DIO_SKIP_HOLES
= 0x02,
2838 /* filesystem can handle aio writes beyond i_size */
2839 DIO_ASYNC_EXTEND
= 0x04,
2841 /* inode/fs/bdev does not need truncate protection */
2842 DIO_SKIP_DIO_COUNT
= 0x08,
2845 void dio_end_io(struct bio
*bio
, int error
);
2847 ssize_t
__blockdev_direct_IO(struct kiocb
*iocb
, struct inode
*inode
,
2848 struct block_device
*bdev
, struct iov_iter
*iter
,
2849 get_block_t get_block
,
2850 dio_iodone_t end_io
, dio_submit_t submit_io
,
2853 static inline ssize_t
blockdev_direct_IO(struct kiocb
*iocb
,
2854 struct inode
*inode
,
2855 struct iov_iter
*iter
,
2856 get_block_t get_block
)
2858 return __blockdev_direct_IO(iocb
, inode
, inode
->i_sb
->s_bdev
, iter
,
2859 get_block
, NULL
, NULL
, DIO_LOCKING
| DIO_SKIP_HOLES
);
2863 void inode_dio_wait(struct inode
*inode
);
2866 * inode_dio_begin - signal start of a direct I/O requests
2867 * @inode: inode the direct I/O happens on
2869 * This is called once we've finished processing a direct I/O request,
2870 * and is used to wake up callers waiting for direct I/O to be quiesced.
2872 static inline void inode_dio_begin(struct inode
*inode
)
2874 atomic_inc(&inode
->i_dio_count
);
2878 * inode_dio_end - signal finish of a direct I/O requests
2879 * @inode: inode the direct I/O happens on
2881 * This is called once we've finished processing a direct I/O request,
2882 * and is used to wake up callers waiting for direct I/O to be quiesced.
2884 static inline void inode_dio_end(struct inode
*inode
)
2886 if (atomic_dec_and_test(&inode
->i_dio_count
))
2887 wake_up_bit(&inode
->i_state
, __I_DIO_WAKEUP
);
2890 extern void inode_set_flags(struct inode
*inode
, unsigned int flags
,
2893 extern const struct file_operations generic_ro_fops
;
2895 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2897 extern int readlink_copy(char __user
*, int, const char *);
2898 extern int page_readlink(struct dentry
*, char __user
*, int);
2899 extern const char *page_get_link(struct dentry
*, struct inode
*,
2900 struct delayed_call
*);
2901 extern void page_put_link(void *);
2902 extern int __page_symlink(struct inode
*inode
, const char *symname
, int len
,
2904 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
2905 extern const struct inode_operations page_symlink_inode_operations
;
2906 extern void kfree_link(void *);
2907 extern int generic_readlink(struct dentry
*, char __user
*, int);
2908 extern void generic_fillattr(struct inode
*, struct kstat
*);
2909 int vfs_getattr_nosec(struct path
*path
, struct kstat
*stat
);
2910 extern int vfs_getattr(struct path
*, struct kstat
*);
2911 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2912 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2913 void __inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2914 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2915 loff_t
inode_get_bytes(struct inode
*inode
);
2916 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
2917 const char *simple_get_link(struct dentry
*, struct inode
*,
2918 struct delayed_call
*);
2919 extern const struct inode_operations simple_symlink_inode_operations
;
2921 extern int iterate_dir(struct file
*, struct dir_context
*);
2923 extern int vfs_stat(const char __user
*, struct kstat
*);
2924 extern int vfs_lstat(const char __user
*, struct kstat
*);
2925 extern int vfs_fstat(unsigned int, struct kstat
*);
2926 extern int vfs_fstatat(int , const char __user
*, struct kstat
*, int);
2928 extern int __generic_block_fiemap(struct inode
*inode
,
2929 struct fiemap_extent_info
*fieinfo
,
2930 loff_t start
, loff_t len
,
2931 get_block_t
*get_block
);
2932 extern int generic_block_fiemap(struct inode
*inode
,
2933 struct fiemap_extent_info
*fieinfo
, u64 start
,
2934 u64 len
, get_block_t
*get_block
);
2936 extern void get_filesystem(struct file_system_type
*fs
);
2937 extern void put_filesystem(struct file_system_type
*fs
);
2938 extern struct file_system_type
*get_fs_type(const char *name
);
2939 extern struct super_block
*get_super(struct block_device
*);
2940 extern struct super_block
*get_super_thawed(struct block_device
*);
2941 extern struct super_block
*get_active_super(struct block_device
*bdev
);
2942 extern void drop_super(struct super_block
*sb
);
2943 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
2944 extern void iterate_supers_type(struct file_system_type
*,
2945 void (*)(struct super_block
*, void *), void *);
2947 extern int dcache_dir_open(struct inode
*, struct file
*);
2948 extern int dcache_dir_close(struct inode
*, struct file
*);
2949 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
2950 extern int dcache_readdir(struct file
*, struct dir_context
*);
2951 extern int simple_setattr(struct dentry
*, struct iattr
*);
2952 extern int simple_getattr(struct vfsmount
*, struct dentry
*, struct kstat
*);
2953 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
2954 extern int simple_open(struct inode
*inode
, struct file
*file
);
2955 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
2956 extern int simple_unlink(struct inode
*, struct dentry
*);
2957 extern int simple_rmdir(struct inode
*, struct dentry
*);
2958 extern int simple_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*);
2959 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
2960 extern int simple_empty(struct dentry
*);
2961 extern int simple_readpage(struct file
*file
, struct page
*page
);
2962 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
2963 loff_t pos
, unsigned len
, unsigned flags
,
2964 struct page
**pagep
, void **fsdata
);
2965 extern int simple_write_end(struct file
*file
, struct address_space
*mapping
,
2966 loff_t pos
, unsigned len
, unsigned copied
,
2967 struct page
*page
, void *fsdata
);
2968 extern int always_delete_dentry(const struct dentry
*);
2969 extern struct inode
*alloc_anon_inode(struct super_block
*);
2970 extern int simple_nosetlease(struct file
*, long, struct file_lock
**, void **);
2971 extern const struct dentry_operations simple_dentry_operations
;
2973 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
2974 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
2975 extern const struct file_operations simple_dir_operations
;
2976 extern const struct inode_operations simple_dir_inode_operations
;
2977 extern void make_empty_dir_inode(struct inode
*inode
);
2978 extern bool is_empty_dir_inode(struct inode
*inode
);
2979 struct tree_descr
{ char *name
; const struct file_operations
*ops
; int mode
; };
2980 struct dentry
*d_alloc_name(struct dentry
*, const char *);
2981 extern int simple_fill_super(struct super_block
*, unsigned long, struct tree_descr
*);
2982 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
2983 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
2985 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
2986 loff_t
*ppos
, const void *from
, size_t available
);
2987 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
2988 const void __user
*from
, size_t count
);
2990 extern int __generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
2991 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
2993 extern int generic_check_addressable(unsigned, u64
);
2995 #ifdef CONFIG_MIGRATION
2996 extern int buffer_migrate_page(struct address_space
*,
2997 struct page
*, struct page
*,
3000 #define buffer_migrate_page NULL
3003 extern int inode_change_ok(const struct inode
*, struct iattr
*);
3004 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
3005 extern void setattr_copy(struct inode
*inode
, const struct iattr
*attr
);
3007 extern int file_update_time(struct file
*file
);
3009 extern int generic_show_options(struct seq_file
*m
, struct dentry
*root
);
3010 extern void save_mount_options(struct super_block
*sb
, char *options
);
3011 extern void replace_mount_options(struct super_block
*sb
, char *options
);
3013 static inline bool io_is_direct(struct file
*filp
)
3015 return (filp
->f_flags
& O_DIRECT
) || IS_DAX(filp
->f_mapping
->host
);
3018 static inline int iocb_flags(struct file
*file
)
3021 if (file
->f_flags
& O_APPEND
)
3023 if (io_is_direct(file
))
3025 if ((file
->f_flags
& O_DSYNC
) || IS_SYNC(file
->f_mapping
->host
))
3027 if (file
->f_flags
& __O_SYNC
)
3032 static inline ino_t
parent_ino(struct dentry
*dentry
)
3037 * Don't strictly need d_lock here? If the parent ino could change
3038 * then surely we'd have a deeper race in the caller?
3040 spin_lock(&dentry
->d_lock
);
3041 res
= dentry
->d_parent
->d_inode
->i_ino
;
3042 spin_unlock(&dentry
->d_lock
);
3046 /* Transaction based IO helpers */
3049 * An argresp is stored in an allocated page and holds the
3050 * size of the argument or response, along with its content
3052 struct simple_transaction_argresp
{
3057 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3059 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
3061 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
3062 size_t size
, loff_t
*pos
);
3063 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
3065 void simple_transaction_set(struct file
*file
, size_t n
);
3068 * simple attribute files
3070 * These attributes behave similar to those in sysfs:
3072 * Writing to an attribute immediately sets a value, an open file can be
3073 * written to multiple times.
3075 * Reading from an attribute creates a buffer from the value that might get
3076 * read with multiple read calls. When the attribute has been read
3077 * completely, no further read calls are possible until the file is opened
3080 * All attributes contain a text representation of a numeric value
3081 * that are accessed with the get() and set() functions.
3083 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3084 static int __fops ## _open(struct inode *inode, struct file *file) \
3086 __simple_attr_check_format(__fmt, 0ull); \
3087 return simple_attr_open(inode, file, __get, __set, __fmt); \
3089 static const struct file_operations __fops = { \
3090 .owner = THIS_MODULE, \
3091 .open = __fops ## _open, \
3092 .release = simple_attr_release, \
3093 .read = simple_attr_read, \
3094 .write = simple_attr_write, \
3095 .llseek = generic_file_llseek, \
3098 static inline __printf(1, 2)
3099 void __simple_attr_check_format(const char *fmt
, ...)
3101 /* don't do anything, just let the compiler check the arguments; */
3104 int simple_attr_open(struct inode
*inode
, struct file
*file
,
3105 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
3107 int simple_attr_release(struct inode
*inode
, struct file
*file
);
3108 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
3109 size_t len
, loff_t
*ppos
);
3110 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
3111 size_t len
, loff_t
*ppos
);
3114 int proc_nr_files(struct ctl_table
*table
, int write
,
3115 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3116 int proc_nr_dentry(struct ctl_table
*table
, int write
,
3117 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3118 int proc_nr_inodes(struct ctl_table
*table
, int write
,
3119 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3120 int __init
get_filesystem_list(char *buf
);
3122 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3123 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3125 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3126 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3127 (flag & __FMODE_NONOTIFY)))
3129 static inline bool is_sxid(umode_t mode
)
3131 return (mode
& S_ISUID
) || ((mode
& S_ISGID
) && (mode
& S_IXGRP
));
3134 static inline int check_sticky(struct inode
*dir
, struct inode
*inode
)
3136 if (!(dir
->i_mode
& S_ISVTX
))
3139 return __check_sticky(dir
, inode
);
3142 static inline void inode_has_no_xattr(struct inode
*inode
)
3144 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& MS_NOSEC
))
3145 inode
->i_flags
|= S_NOSEC
;
3148 static inline bool is_root_inode(struct inode
*inode
)
3150 return inode
== inode
->i_sb
->s_root
->d_inode
;
3153 static inline bool dir_emit(struct dir_context
*ctx
,
3154 const char *name
, int namelen
,
3155 u64 ino
, unsigned type
)
3157 return ctx
->actor(ctx
, name
, namelen
, ctx
->pos
, ino
, type
) == 0;
3159 static inline bool dir_emit_dot(struct file
*file
, struct dir_context
*ctx
)
3161 return ctx
->actor(ctx
, ".", 1, ctx
->pos
,
3162 file
->f_path
.dentry
->d_inode
->i_ino
, DT_DIR
) == 0;
3164 static inline bool dir_emit_dotdot(struct file
*file
, struct dir_context
*ctx
)
3166 return ctx
->actor(ctx
, "..", 2, ctx
->pos
,
3167 parent_ino(file
->f_path
.dentry
), DT_DIR
) == 0;
3169 static inline bool dir_emit_dots(struct file
*file
, struct dir_context
*ctx
)
3171 if (ctx
->pos
== 0) {
3172 if (!dir_emit_dot(file
, ctx
))
3176 if (ctx
->pos
== 1) {
3177 if (!dir_emit_dotdot(file
, ctx
))
3183 static inline bool dir_relax(struct inode
*inode
)
3185 inode_unlock(inode
);
3187 return !IS_DEADDIR(inode
);
3190 static inline bool dir_relax_shared(struct inode
*inode
)
3192 inode_unlock_shared(inode
);
3193 inode_lock_shared(inode
);
3194 return !IS_DEADDIR(inode
);
3197 extern bool path_noexec(const struct path
*path
);
3198 extern void inode_nohighmem(struct inode
*inode
);
3200 #endif /* _LINUX_FS_H */