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/workqueue.h>
32 #include <linux/percpu-rwsem.h>
33 #include <linux/delayed_call.h>
35 #include <asm/byteorder.h>
36 #include <uapi/linux/fs.h>
38 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 unsigned 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 * Attribute flags. These should be or-ed together to figure out what
158 #define ATTR_MODE (1 << 0)
159 #define ATTR_UID (1 << 1)
160 #define ATTR_GID (1 << 2)
161 #define ATTR_SIZE (1 << 3)
162 #define ATTR_ATIME (1 << 4)
163 #define ATTR_MTIME (1 << 5)
164 #define ATTR_CTIME (1 << 6)
165 #define ATTR_ATIME_SET (1 << 7)
166 #define ATTR_MTIME_SET (1 << 8)
167 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
168 #define ATTR_ATTR_FLAG (1 << 10)
169 #define ATTR_KILL_SUID (1 << 11)
170 #define ATTR_KILL_SGID (1 << 12)
171 #define ATTR_FILE (1 << 13)
172 #define ATTR_KILL_PRIV (1 << 14)
173 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
174 #define ATTR_TIMES_SET (1 << 16)
175 #define ATTR_TOUCH (1 << 17)
178 * Whiteout is represented by a char device. The following constants define the
179 * mode and device number to use.
181 #define WHITEOUT_MODE 0
182 #define WHITEOUT_DEV 0
185 * This is the Inode Attributes structure, used for notify_change(). It
186 * uses the above definitions as flags, to know which values have changed.
187 * Also, in this manner, a Filesystem can look at only the values it cares
188 * about. Basically, these are the attributes that the VFS layer can
189 * request to change from the FS layer.
191 * Derek Atkins <warlord@MIT.EDU> 94-10-20
194 unsigned int ia_valid
;
199 struct timespec ia_atime
;
200 struct timespec ia_mtime
;
201 struct timespec ia_ctime
;
204 * Not an attribute, but an auxiliary info for filesystems wanting to
205 * implement an ftruncate() like method. NOTE: filesystem should
206 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
208 struct file
*ia_file
;
212 * Includes for diskquotas.
214 #include <linux/quota.h>
217 * Maximum number of layers of fs stack. Needs to be limited to
218 * prevent kernel stack overflow
220 #define FILESYSTEM_MAX_STACK_DEPTH 2
223 * enum positive_aop_returns - aop return codes with specific semantics
225 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
226 * completed, that the page is still locked, and
227 * should be considered active. The VM uses this hint
228 * to return the page to the active list -- it won't
229 * be a candidate for writeback again in the near
230 * future. Other callers must be careful to unlock
231 * the page if they get this return. Returned by
234 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
235 * unlocked it and the page might have been truncated.
236 * The caller should back up to acquiring a new page and
237 * trying again. The aop will be taking reasonable
238 * precautions not to livelock. If the caller held a page
239 * reference, it should drop it before retrying. Returned
242 * address_space_operation functions return these large constants to indicate
243 * special semantics to the caller. These are much larger than the bytes in a
244 * page to allow for functions that return the number of bytes operated on in a
248 enum positive_aop_returns
{
249 AOP_WRITEPAGE_ACTIVATE
= 0x80000,
250 AOP_TRUNCATED_PAGE
= 0x80001,
253 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
254 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
255 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
256 * helper code (eg buffer layer)
257 * to clear GFP_FS from alloc */
260 * oh the beauties of C type declarations.
263 struct address_space
;
264 struct writeback_control
;
266 #define IOCB_EVENTFD (1 << 0)
267 #define IOCB_APPEND (1 << 1)
268 #define IOCB_DIRECT (1 << 2)
269 #define IOCB_HIPRI (1 << 3)
270 #define IOCB_DSYNC (1 << 4)
271 #define IOCB_SYNC (1 << 5)
272 #define IOCB_WRITE (1 << 6)
275 struct file
*ki_filp
;
277 void (*ki_complete
)(struct kiocb
*iocb
, long ret
, long ret2
);
282 static inline bool is_sync_kiocb(struct kiocb
*kiocb
)
284 return kiocb
->ki_complete
== NULL
;
287 static inline int iocb_flags(struct file
*file
);
289 static inline void init_sync_kiocb(struct kiocb
*kiocb
, struct file
*filp
)
291 *kiocb
= (struct kiocb
) {
293 .ki_flags
= iocb_flags(filp
),
298 * "descriptor" for what we're up to with a read.
299 * This allows us to use the same read code yet
300 * have multiple different users of the data that
301 * we read from a file.
303 * The simplest case just copies the data to user
316 typedef int (*read_actor_t
)(read_descriptor_t
*, struct page
*,
317 unsigned long, unsigned long);
319 struct address_space_operations
{
320 int (*writepage
)(struct page
*page
, struct writeback_control
*wbc
);
321 int (*readpage
)(struct file
*, struct page
*);
323 /* Write back some dirty pages from this mapping. */
324 int (*writepages
)(struct address_space
*, struct writeback_control
*);
326 /* Set a page dirty. Return true if this dirtied it */
327 int (*set_page_dirty
)(struct page
*page
);
329 int (*readpages
)(struct file
*filp
, struct address_space
*mapping
,
330 struct list_head
*pages
, unsigned nr_pages
);
332 int (*write_begin
)(struct file
*, struct address_space
*mapping
,
333 loff_t pos
, unsigned len
, unsigned flags
,
334 struct page
**pagep
, void **fsdata
);
335 int (*write_end
)(struct file
*, struct address_space
*mapping
,
336 loff_t pos
, unsigned len
, unsigned copied
,
337 struct page
*page
, void *fsdata
);
339 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
340 sector_t (*bmap
)(struct address_space
*, sector_t
);
341 void (*invalidatepage
) (struct page
*, unsigned int, unsigned int);
342 int (*releasepage
) (struct page
*, gfp_t
);
343 void (*freepage
)(struct page
*);
344 ssize_t (*direct_IO
)(struct kiocb
*, struct iov_iter
*iter
);
346 * migrate the contents of a page to the specified target. If
347 * migrate_mode is MIGRATE_ASYNC, it must not block.
349 int (*migratepage
) (struct address_space
*,
350 struct page
*, struct page
*, enum migrate_mode
);
351 bool (*isolate_page
)(struct page
*, isolate_mode_t
);
352 void (*putback_page
)(struct page
*);
353 int (*launder_page
) (struct page
*);
354 int (*is_partially_uptodate
) (struct page
*, unsigned long,
356 void (*is_dirty_writeback
) (struct page
*, bool *, bool *);
357 int (*error_remove_page
)(struct address_space
*, struct page
*);
359 /* swapfile support */
360 int (*swap_activate
)(struct swap_info_struct
*sis
, struct file
*file
,
362 void (*swap_deactivate
)(struct file
*file
);
365 extern const struct address_space_operations empty_aops
;
368 * pagecache_write_begin/pagecache_write_end must be used by general code
369 * to write into the pagecache.
371 int pagecache_write_begin(struct file
*, struct address_space
*mapping
,
372 loff_t pos
, unsigned len
, unsigned flags
,
373 struct page
**pagep
, void **fsdata
);
375 int pagecache_write_end(struct file
*, struct address_space
*mapping
,
376 loff_t pos
, unsigned len
, unsigned copied
,
377 struct page
*page
, void *fsdata
);
379 struct address_space
{
380 struct inode
*host
; /* owner: inode, block_device */
381 struct radix_tree_root page_tree
; /* radix tree of all pages */
382 spinlock_t tree_lock
; /* and lock protecting it */
383 atomic_t i_mmap_writable
;/* count VM_SHARED mappings */
384 struct rb_root i_mmap
; /* tree of private and shared mappings */
385 struct rw_semaphore i_mmap_rwsem
; /* protect tree, count, list */
386 /* Protected by tree_lock together with the radix tree */
387 unsigned long nrpages
; /* number of total pages */
388 /* number of shadow or DAX exceptional entries */
389 unsigned long nrexceptional
;
390 pgoff_t writeback_index
;/* writeback starts here */
391 const struct address_space_operations
*a_ops
; /* methods */
392 unsigned long flags
; /* error bits */
393 spinlock_t private_lock
; /* for use by the address_space */
394 gfp_t gfp_mask
; /* implicit gfp mask for allocations */
395 struct list_head private_list
; /* ditto */
396 void *private_data
; /* ditto */
397 } __attribute__((aligned(sizeof(long))));
399 * On most architectures that alignment is already the case; but
400 * must be enforced here for CRIS, to let the least significant bit
401 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
403 struct request_queue
;
405 struct block_device
{
406 dev_t bd_dev
; /* not a kdev_t - it's a search key */
408 struct inode
* bd_inode
; /* will die */
409 struct super_block
* bd_super
;
410 struct mutex bd_mutex
; /* open/close mutex */
414 bool bd_write_holder
;
416 struct list_head bd_holder_disks
;
418 struct block_device
* bd_contains
;
419 unsigned bd_block_size
;
420 struct hd_struct
* bd_part
;
421 /* number of times partitions within this device have been opened. */
422 unsigned bd_part_count
;
424 struct gendisk
* bd_disk
;
425 struct request_queue
* bd_queue
;
426 struct list_head bd_list
;
428 * Private data. You must have bd_claim'ed the block_device
429 * to use this. NOTE: bd_claim allows an owner to claim
430 * the same device multiple times, the owner must take special
431 * care to not mess up bd_private for that case.
433 unsigned long bd_private
;
435 /* The counter of freeze processes */
436 int bd_fsfreeze_count
;
437 /* Mutex for freeze */
438 struct mutex bd_fsfreeze_mutex
;
442 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
445 #define PAGECACHE_TAG_DIRTY 0
446 #define PAGECACHE_TAG_WRITEBACK 1
447 #define PAGECACHE_TAG_TOWRITE 2
449 int mapping_tagged(struct address_space
*mapping
, int tag
);
451 static inline void i_mmap_lock_write(struct address_space
*mapping
)
453 down_write(&mapping
->i_mmap_rwsem
);
456 static inline void i_mmap_unlock_write(struct address_space
*mapping
)
458 up_write(&mapping
->i_mmap_rwsem
);
461 static inline void i_mmap_lock_read(struct address_space
*mapping
)
463 down_read(&mapping
->i_mmap_rwsem
);
466 static inline void i_mmap_unlock_read(struct address_space
*mapping
)
468 up_read(&mapping
->i_mmap_rwsem
);
472 * Might pages of this file be mapped into userspace?
474 static inline int mapping_mapped(struct address_space
*mapping
)
476 return !RB_EMPTY_ROOT(&mapping
->i_mmap
);
480 * Might pages of this file have been modified in userspace?
481 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
482 * marks vma as VM_SHARED if it is shared, and the file was opened for
483 * writing i.e. vma may be mprotected writable even if now readonly.
485 * If i_mmap_writable is negative, no new writable mappings are allowed. You
486 * can only deny writable mappings, if none exists right now.
488 static inline int mapping_writably_mapped(struct address_space
*mapping
)
490 return atomic_read(&mapping
->i_mmap_writable
) > 0;
493 static inline int mapping_map_writable(struct address_space
*mapping
)
495 return atomic_inc_unless_negative(&mapping
->i_mmap_writable
) ?
499 static inline void mapping_unmap_writable(struct address_space
*mapping
)
501 atomic_dec(&mapping
->i_mmap_writable
);
504 static inline int mapping_deny_writable(struct address_space
*mapping
)
506 return atomic_dec_unless_positive(&mapping
->i_mmap_writable
) ?
510 static inline void mapping_allow_writable(struct address_space
*mapping
)
512 atomic_inc(&mapping
->i_mmap_writable
);
516 * Use sequence counter to get consistent i_size on 32-bit processors.
518 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
519 #include <linux/seqlock.h>
520 #define __NEED_I_SIZE_ORDERED
521 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
523 #define i_size_ordered_init(inode) do { } while (0)
527 #define ACL_NOT_CACHED ((void *)(-1))
528 #define ACL_DONT_CACHE ((void *)(-3))
530 static inline struct posix_acl
*
531 uncached_acl_sentinel(struct task_struct
*task
)
533 return (void *)task
+ 1;
537 is_uncached_acl(struct posix_acl
*acl
)
539 return (long)acl
& 1;
542 #define IOP_FASTPERM 0x0001
543 #define IOP_LOOKUP 0x0002
544 #define IOP_NOFOLLOW 0x0004
545 #define IOP_XATTR 0x0008
546 #define IOP_DEFAULT_READLINK 0x0010
549 * Keep mostly read-only and often accessed (especially for
550 * the RCU path lookup and 'stat' data) fields at the beginning
551 * of the 'struct inode'
555 unsigned short i_opflags
;
558 unsigned int i_flags
;
560 #ifdef CONFIG_FS_POSIX_ACL
561 struct posix_acl
*i_acl
;
562 struct posix_acl
*i_default_acl
;
565 const struct inode_operations
*i_op
;
566 struct super_block
*i_sb
;
567 struct address_space
*i_mapping
;
569 #ifdef CONFIG_SECURITY
573 /* Stat data, not accessed from path walking */
576 * Filesystems may only read i_nlink directly. They shall use the
577 * following functions for modification:
579 * (set|clear|inc|drop)_nlink
580 * inode_(inc|dec)_link_count
583 const unsigned int i_nlink
;
584 unsigned int __i_nlink
;
588 struct timespec i_atime
;
589 struct timespec i_mtime
;
590 struct timespec i_ctime
;
591 spinlock_t i_lock
; /* i_blocks, i_bytes, maybe i_size */
592 unsigned short i_bytes
;
593 unsigned int i_blkbits
;
596 #ifdef __NEED_I_SIZE_ORDERED
597 seqcount_t i_size_seqcount
;
601 unsigned long i_state
;
602 struct rw_semaphore i_rwsem
;
604 unsigned long dirtied_when
; /* jiffies of first dirtying */
605 unsigned long dirtied_time_when
;
607 struct hlist_node i_hash
;
608 struct list_head i_io_list
; /* backing dev IO list */
609 #ifdef CONFIG_CGROUP_WRITEBACK
610 struct bdi_writeback
*i_wb
; /* the associated cgroup wb */
612 /* foreign inode detection, see wbc_detach_inode() */
614 u16 i_wb_frn_avg_time
;
615 u16 i_wb_frn_history
;
617 struct list_head i_lru
; /* inode LRU list */
618 struct list_head i_sb_list
;
619 struct list_head i_wb_list
; /* backing dev writeback list */
621 struct hlist_head i_dentry
;
622 struct rcu_head i_rcu
;
626 atomic_t i_dio_count
;
627 atomic_t i_writecount
;
629 atomic_t i_readcount
; /* struct files open RO */
631 const struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
632 struct file_lock_context
*i_flctx
;
633 struct address_space i_data
;
634 struct list_head i_devices
;
636 struct pipe_inode_info
*i_pipe
;
637 struct block_device
*i_bdev
;
645 #ifdef CONFIG_FSNOTIFY
646 __u32 i_fsnotify_mask
; /* all events this inode cares about */
647 struct hlist_head i_fsnotify_marks
;
650 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
651 struct fscrypt_info
*i_crypt_info
;
654 void *i_private
; /* fs or device private pointer */
657 static inline int inode_unhashed(struct inode
*inode
)
659 return hlist_unhashed(&inode
->i_hash
);
663 * inode->i_mutex nesting subclasses for the lock validator:
665 * 0: the object of the current VFS operation
669 * 4: second non-directory
670 * 5: second parent (when locking independent directories in rename)
672 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
673 * non-directories at once.
675 * The locking order between these classes is
676 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
678 enum inode_i_mutex_lock_class
688 static inline void inode_lock(struct inode
*inode
)
690 down_write(&inode
->i_rwsem
);
693 static inline void inode_unlock(struct inode
*inode
)
695 up_write(&inode
->i_rwsem
);
698 static inline void inode_lock_shared(struct inode
*inode
)
700 down_read(&inode
->i_rwsem
);
703 static inline void inode_unlock_shared(struct inode
*inode
)
705 up_read(&inode
->i_rwsem
);
708 static inline int inode_trylock(struct inode
*inode
)
710 return down_write_trylock(&inode
->i_rwsem
);
713 static inline int inode_trylock_shared(struct inode
*inode
)
715 return down_read_trylock(&inode
->i_rwsem
);
718 static inline int inode_is_locked(struct inode
*inode
)
720 return rwsem_is_locked(&inode
->i_rwsem
);
723 static inline void inode_lock_nested(struct inode
*inode
, unsigned subclass
)
725 down_write_nested(&inode
->i_rwsem
, subclass
);
728 void lock_two_nondirectories(struct inode
*, struct inode
*);
729 void unlock_two_nondirectories(struct inode
*, struct inode
*);
732 * NOTE: in a 32bit arch with a preemptable kernel and
733 * an UP compile the i_size_read/write must be atomic
734 * with respect to the local cpu (unlike with preempt disabled),
735 * but they don't need to be atomic with respect to other cpus like in
736 * true SMP (so they need either to either locally disable irq around
737 * the read or for example on x86 they can be still implemented as a
738 * cmpxchg8b without the need of the lock prefix). For SMP compiles
739 * and 64bit archs it makes no difference if preempt is enabled or not.
741 static inline loff_t
i_size_read(const struct inode
*inode
)
743 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
748 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
749 i_size
= inode
->i_size
;
750 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
752 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
756 i_size
= inode
->i_size
;
760 return inode
->i_size
;
765 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
766 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
767 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
769 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
771 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
773 write_seqcount_begin(&inode
->i_size_seqcount
);
774 inode
->i_size
= i_size
;
775 write_seqcount_end(&inode
->i_size_seqcount
);
777 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
779 inode
->i_size
= i_size
;
782 inode
->i_size
= i_size
;
786 static inline unsigned iminor(const struct inode
*inode
)
788 return MINOR(inode
->i_rdev
);
791 static inline unsigned imajor(const struct inode
*inode
)
793 return MAJOR(inode
->i_rdev
);
796 extern struct block_device
*I_BDEV(struct inode
*inode
);
799 rwlock_t lock
; /* protects pid, uid, euid fields */
800 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
801 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
802 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
803 int signum
; /* posix.1b rt signal to be delivered on IO */
807 * Track a single file's readahead state
809 struct file_ra_state
{
810 pgoff_t start
; /* where readahead started */
811 unsigned int size
; /* # of readahead pages */
812 unsigned int async_size
; /* do asynchronous readahead when
813 there are only # of pages ahead */
815 unsigned int ra_pages
; /* Maximum readahead window */
816 unsigned int mmap_miss
; /* Cache miss stat for mmap accesses */
817 loff_t prev_pos
; /* Cache last read() position */
821 * Check if @index falls in the readahead windows.
823 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
825 return (index
>= ra
->start
&&
826 index
< ra
->start
+ ra
->size
);
831 struct llist_node fu_llist
;
832 struct rcu_head fu_rcuhead
;
835 struct inode
*f_inode
; /* cached value */
836 const struct file_operations
*f_op
;
839 * Protects f_ep_links, f_flags.
840 * Must not be taken from IRQ context.
843 atomic_long_t f_count
;
844 unsigned int f_flags
;
846 struct mutex f_pos_lock
;
848 struct fown_struct f_owner
;
849 const struct cred
*f_cred
;
850 struct file_ra_state f_ra
;
853 #ifdef CONFIG_SECURITY
856 /* needed for tty driver, and maybe others */
860 /* Used by fs/eventpoll.c to link all the hooks to this file */
861 struct list_head f_ep_links
;
862 struct list_head f_tfile_llink
;
863 #endif /* #ifdef CONFIG_EPOLL */
864 struct address_space
*f_mapping
;
865 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
870 /* file identifier */
871 unsigned char f_handle
[0];
874 static inline struct file
*get_file(struct file
*f
)
876 atomic_long_inc(&f
->f_count
);
879 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
880 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
881 #define file_count(x) atomic_long_read(&(x)->f_count)
883 #define MAX_NON_LFS ((1UL<<31) - 1)
885 /* Page cache limit. The filesystems should put that into their s_maxbytes
886 limits, otherwise bad things can happen in VM. */
887 #if BITS_PER_LONG==32
888 #define MAX_LFS_FILESIZE (((loff_t)PAGE_SIZE << (BITS_PER_LONG-1))-1)
889 #elif BITS_PER_LONG==64
890 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
895 #define FL_DELEG 4 /* NFSv4 delegation */
896 #define FL_ACCESS 8 /* not trying to lock, just looking */
897 #define FL_EXISTS 16 /* when unlocking, test for existence */
898 #define FL_LEASE 32 /* lease held on this file */
899 #define FL_CLOSE 64 /* unlock on close */
900 #define FL_SLEEP 128 /* A blocking lock */
901 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
902 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
903 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
904 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
907 * Special return value from posix_lock_file() and vfs_lock_file() for
908 * asynchronous locking.
910 #define FILE_LOCK_DEFERRED 1
912 /* legacy typedef, should eventually be removed */
913 typedef void *fl_owner_t
;
917 struct file_lock_operations
{
918 void (*fl_copy_lock
)(struct file_lock
*, struct file_lock
*);
919 void (*fl_release_private
)(struct file_lock
*);
922 struct lock_manager_operations
{
923 int (*lm_compare_owner
)(struct file_lock
*, struct file_lock
*);
924 unsigned long (*lm_owner_key
)(struct file_lock
*);
925 fl_owner_t (*lm_get_owner
)(fl_owner_t
);
926 void (*lm_put_owner
)(fl_owner_t
);
927 void (*lm_notify
)(struct file_lock
*); /* unblock callback */
928 int (*lm_grant
)(struct file_lock
*, int);
929 bool (*lm_break
)(struct file_lock
*);
930 int (*lm_change
)(struct file_lock
*, int, struct list_head
*);
931 void (*lm_setup
)(struct file_lock
*, void **);
934 struct lock_manager
{
935 struct list_head list
;
937 * NFSv4 and up also want opens blocked during the grace period;
944 void locks_start_grace(struct net
*, struct lock_manager
*);
945 void locks_end_grace(struct lock_manager
*);
946 int locks_in_grace(struct net
*);
947 int opens_in_grace(struct net
*);
949 /* that will die - we need it for nfs_lock_info */
950 #include <linux/nfs_fs_i.h>
953 * struct file_lock represents a generic "file lock". It's used to represent
954 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
955 * note that the same struct is used to represent both a request for a lock and
956 * the lock itself, but the same object is never used for both.
958 * FIXME: should we create a separate "struct lock_request" to help distinguish
961 * The varous i_flctx lists are ordered by:
964 * 2) lock range start
967 * Obviously, the last two criteria only matter for POSIX locks.
970 struct file_lock
*fl_next
; /* singly linked list for this inode */
971 struct list_head fl_list
; /* link into file_lock_context */
972 struct hlist_node fl_link
; /* node in global lists */
973 struct list_head fl_block
; /* circular list of blocked processes */
975 unsigned int fl_flags
;
976 unsigned char fl_type
;
978 int fl_link_cpu
; /* what cpu's list is this on? */
979 struct pid
*fl_nspid
;
980 wait_queue_head_t fl_wait
;
981 struct file
*fl_file
;
985 struct fasync_struct
* fl_fasync
; /* for lease break notifications */
986 /* for lease breaks: */
987 unsigned long fl_break_time
;
988 unsigned long fl_downgrade_time
;
990 const struct file_lock_operations
*fl_ops
; /* Callbacks for filesystems */
991 const struct lock_manager_operations
*fl_lmops
; /* Callbacks for lockmanagers */
993 struct nfs_lock_info nfs_fl
;
994 struct nfs4_lock_info nfs4_fl
;
996 struct list_head link
; /* link in AFS vnode's pending_locks list */
997 int state
; /* state of grant or error if -ve */
1002 struct file_lock_context
{
1003 spinlock_t flc_lock
;
1004 struct list_head flc_flock
;
1005 struct list_head flc_posix
;
1006 struct list_head flc_lease
;
1009 /* The following constant reflects the upper bound of the file/locking space */
1011 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1012 #define OFFSET_MAX INT_LIMIT(loff_t)
1013 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1016 #include <linux/fcntl.h>
1018 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
1021 * Return the inode to use for locking
1023 * For overlayfs this should be the overlay inode, not the real inode returned
1024 * by file_inode(). For any other fs file_inode(filp) and locks_inode(filp) are
1027 static inline struct inode
*locks_inode(const struct file
*f
)
1029 return f
->f_path
.dentry
->d_inode
;
1032 #ifdef CONFIG_FILE_LOCKING
1033 extern int fcntl_getlk(struct file
*, unsigned int, struct flock __user
*);
1034 extern int fcntl_setlk(unsigned int, struct file
*, unsigned int,
1035 struct flock __user
*);
1037 #if BITS_PER_LONG == 32
1038 extern int fcntl_getlk64(struct file
*, unsigned int, struct flock64 __user
*);
1039 extern int fcntl_setlk64(unsigned int, struct file
*, unsigned int,
1040 struct flock64 __user
*);
1043 extern int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
);
1044 extern int fcntl_getlease(struct file
*filp
);
1047 void locks_free_lock_context(struct inode
*inode
);
1048 void locks_free_lock(struct file_lock
*fl
);
1049 extern void locks_init_lock(struct file_lock
*);
1050 extern struct file_lock
* locks_alloc_lock(void);
1051 extern void locks_copy_lock(struct file_lock
*, struct file_lock
*);
1052 extern void locks_copy_conflock(struct file_lock
*, struct file_lock
*);
1053 extern void locks_remove_posix(struct file
*, fl_owner_t
);
1054 extern void locks_remove_file(struct file
*);
1055 extern void locks_release_private(struct file_lock
*);
1056 extern void posix_test_lock(struct file
*, struct file_lock
*);
1057 extern int posix_lock_file(struct file
*, struct file_lock
*, struct file_lock
*);
1058 extern int posix_unblock_lock(struct file_lock
*);
1059 extern int vfs_test_lock(struct file
*, struct file_lock
*);
1060 extern int vfs_lock_file(struct file
*, unsigned int, struct file_lock
*, struct file_lock
*);
1061 extern int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
);
1062 extern int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
);
1063 extern int __break_lease(struct inode
*inode
, unsigned int flags
, unsigned int type
);
1064 extern void lease_get_mtime(struct inode
*, struct timespec
*time
);
1065 extern int generic_setlease(struct file
*, long, struct file_lock
**, void **priv
);
1066 extern int vfs_setlease(struct file
*, long, struct file_lock
**, void **);
1067 extern int lease_modify(struct file_lock
*, int, struct list_head
*);
1068 struct files_struct
;
1069 extern void show_fd_locks(struct seq_file
*f
,
1070 struct file
*filp
, struct files_struct
*files
);
1071 #else /* !CONFIG_FILE_LOCKING */
1072 static inline int fcntl_getlk(struct file
*file
, unsigned int cmd
,
1073 struct flock __user
*user
)
1078 static inline int fcntl_setlk(unsigned int fd
, struct file
*file
,
1079 unsigned int cmd
, struct flock __user
*user
)
1084 #if BITS_PER_LONG == 32
1085 static inline int fcntl_getlk64(struct file
*file
, unsigned int cmd
,
1086 struct flock64 __user
*user
)
1091 static inline int fcntl_setlk64(unsigned int fd
, struct file
*file
,
1092 unsigned int cmd
, struct flock64 __user
*user
)
1097 static inline int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
)
1102 static inline int fcntl_getlease(struct file
*filp
)
1108 locks_free_lock_context(struct inode
*inode
)
1112 static inline void locks_init_lock(struct file_lock
*fl
)
1117 static inline void locks_copy_conflock(struct file_lock
*new, struct file_lock
*fl
)
1122 static inline void locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
1127 static inline void locks_remove_posix(struct file
*filp
, fl_owner_t owner
)
1132 static inline void locks_remove_file(struct file
*filp
)
1137 static inline void posix_test_lock(struct file
*filp
, struct file_lock
*fl
)
1142 static inline int posix_lock_file(struct file
*filp
, struct file_lock
*fl
,
1143 struct file_lock
*conflock
)
1148 static inline int posix_unblock_lock(struct file_lock
*waiter
)
1153 static inline int vfs_test_lock(struct file
*filp
, struct file_lock
*fl
)
1158 static inline int vfs_lock_file(struct file
*filp
, unsigned int cmd
,
1159 struct file_lock
*fl
, struct file_lock
*conf
)
1164 static inline int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
)
1169 static inline int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
)
1174 static inline int __break_lease(struct inode
*inode
, unsigned int mode
, unsigned int type
)
1179 static inline void lease_get_mtime(struct inode
*inode
, struct timespec
*time
)
1184 static inline int generic_setlease(struct file
*filp
, long arg
,
1185 struct file_lock
**flp
, void **priv
)
1190 static inline int vfs_setlease(struct file
*filp
, long arg
,
1191 struct file_lock
**lease
, void **priv
)
1196 static inline int lease_modify(struct file_lock
*fl
, int arg
,
1197 struct list_head
*dispose
)
1202 struct files_struct
;
1203 static inline void show_fd_locks(struct seq_file
*f
,
1204 struct file
*filp
, struct files_struct
*files
) {}
1205 #endif /* !CONFIG_FILE_LOCKING */
1207 static inline struct inode
*file_inode(const struct file
*f
)
1212 static inline struct dentry
*file_dentry(const struct file
*file
)
1214 return d_real(file
->f_path
.dentry
, file_inode(file
), 0);
1217 static inline int locks_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1219 return locks_lock_inode_wait(locks_inode(filp
), fl
);
1222 struct fasync_struct
{
1226 struct fasync_struct
*fa_next
; /* singly linked list */
1227 struct file
*fa_file
;
1228 struct rcu_head fa_rcu
;
1231 #define FASYNC_MAGIC 0x4601
1233 /* SMP safe fasync helpers: */
1234 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1235 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1236 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1237 extern struct fasync_struct
*fasync_alloc(void);
1238 extern void fasync_free(struct fasync_struct
*);
1240 /* can be called from interrupts */
1241 extern void kill_fasync(struct fasync_struct
**, int, int);
1243 extern void __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1244 extern void f_setown(struct file
*filp
, unsigned long arg
, int force
);
1245 extern void f_delown(struct file
*filp
);
1246 extern pid_t
f_getown(struct file
*filp
);
1247 extern int send_sigurg(struct fown_struct
*fown
);
1255 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1256 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1257 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1258 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1259 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1262 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1263 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1264 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1266 /* sb->s_iflags to limit user namespace mounts */
1267 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1269 /* Possible states of 'frozen' field */
1271 SB_UNFROZEN
= 0, /* FS is unfrozen */
1272 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1273 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1274 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1275 * internal threads if needed) */
1276 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1279 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1282 int frozen
; /* Is sb frozen? */
1283 wait_queue_head_t wait_unfrozen
; /* for get_super_thawed() */
1284 struct percpu_rw_semaphore rw_sem
[SB_FREEZE_LEVELS
];
1287 struct super_block
{
1288 struct list_head s_list
; /* Keep this first */
1289 dev_t s_dev
; /* search index; _not_ kdev_t */
1290 unsigned char s_blocksize_bits
;
1291 unsigned long s_blocksize
;
1292 loff_t s_maxbytes
; /* Max file size */
1293 struct file_system_type
*s_type
;
1294 const struct super_operations
*s_op
;
1295 const struct dquot_operations
*dq_op
;
1296 const struct quotactl_ops
*s_qcop
;
1297 const struct export_operations
*s_export_op
;
1298 unsigned long s_flags
;
1299 unsigned long s_iflags
; /* internal SB_I_* flags */
1300 unsigned long s_magic
;
1301 struct dentry
*s_root
;
1302 struct rw_semaphore s_umount
;
1305 #ifdef CONFIG_SECURITY
1308 const struct xattr_handler
**s_xattr
;
1310 const struct fscrypt_operations
*s_cop
;
1312 struct hlist_bl_head s_anon
; /* anonymous dentries for (nfs) exporting */
1313 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1314 struct block_device
*s_bdev
;
1315 struct backing_dev_info
*s_bdi
;
1316 struct mtd_info
*s_mtd
;
1317 struct hlist_node s_instances
;
1318 unsigned int s_quota_types
; /* Bitmask of supported quota types */
1319 struct quota_info s_dquot
; /* Diskquota specific options */
1321 struct sb_writers s_writers
;
1323 char s_id
[32]; /* Informational name */
1324 u8 s_uuid
[16]; /* UUID */
1326 void *s_fs_info
; /* Filesystem private info */
1327 unsigned int s_max_links
;
1330 /* Granularity of c/m/atime in ns.
1331 Cannot be worse than a second */
1335 * The next field is for VFS *only*. No filesystems have any business
1336 * even looking at it. You had been warned.
1338 struct mutex s_vfs_rename_mutex
; /* Kludge */
1341 * Filesystem subtype. If non-empty the filesystem type field
1342 * in /proc/mounts will be "type.subtype"
1347 * Saved mount options for lazy filesystems using
1348 * generic_show_options()
1350 char __rcu
*s_options
;
1351 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1354 * Saved pool identifier for cleancache (-1 means none)
1356 int cleancache_poolid
;
1358 struct shrinker s_shrink
; /* per-sb shrinker handle */
1360 /* Number of inodes with nlink == 0 but still referenced */
1361 atomic_long_t s_remove_count
;
1363 /* Being remounted read-only */
1364 int s_readonly_remount
;
1366 /* AIO completions deferred from interrupt context */
1367 struct workqueue_struct
*s_dio_done_wq
;
1368 struct hlist_head s_pins
;
1371 * Owning user namespace and default context in which to
1372 * interpret filesystem uids, gids, quotas, device nodes,
1373 * xattrs and security labels.
1375 struct user_namespace
*s_user_ns
;
1378 * Keep the lru lists last in the structure so they always sit on their
1379 * own individual cachelines.
1381 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp
;
1382 struct list_lru s_inode_lru ____cacheline_aligned_in_smp
;
1383 struct rcu_head rcu
;
1384 struct work_struct destroy_work
;
1386 struct mutex s_sync_lock
; /* sync serialisation lock */
1389 * Indicates how deep in a filesystem stack this SB is
1393 /* s_inode_list_lock protects s_inodes */
1394 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp
;
1395 struct list_head s_inodes
; /* all inodes */
1397 spinlock_t s_inode_wblist_lock
;
1398 struct list_head s_inodes_wb
; /* writeback inodes */
1401 /* Helper functions so that in most cases filesystems will
1402 * not need to deal directly with kuid_t and kgid_t and can
1403 * instead deal with the raw numeric values that are stored
1404 * in the filesystem.
1406 static inline uid_t
i_uid_read(const struct inode
*inode
)
1408 return from_kuid(inode
->i_sb
->s_user_ns
, inode
->i_uid
);
1411 static inline gid_t
i_gid_read(const struct inode
*inode
)
1413 return from_kgid(inode
->i_sb
->s_user_ns
, inode
->i_gid
);
1416 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
1418 inode
->i_uid
= make_kuid(inode
->i_sb
->s_user_ns
, uid
);
1421 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
1423 inode
->i_gid
= make_kgid(inode
->i_sb
->s_user_ns
, gid
);
1426 extern struct timespec
current_fs_time(struct super_block
*sb
);
1427 extern struct timespec
current_time(struct inode
*inode
);
1430 * Snapshotting support.
1433 void __sb_end_write(struct super_block
*sb
, int level
);
1434 int __sb_start_write(struct super_block
*sb
, int level
, bool wait
);
1436 #define __sb_writers_acquired(sb, lev) \
1437 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1438 #define __sb_writers_release(sb, lev) \
1439 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1442 * sb_end_write - drop write access to a superblock
1443 * @sb: the super we wrote to
1445 * Decrement number of writers to the filesystem. Wake up possible waiters
1446 * wanting to freeze the filesystem.
1448 static inline void sb_end_write(struct super_block
*sb
)
1450 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1454 * sb_end_pagefault - drop write access to a superblock from a page fault
1455 * @sb: the super we wrote to
1457 * Decrement number of processes handling write page fault to the filesystem.
1458 * Wake up possible waiters wanting to freeze the filesystem.
1460 static inline void sb_end_pagefault(struct super_block
*sb
)
1462 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1466 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1467 * @sb: the super we wrote to
1469 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1470 * waiters wanting to freeze the filesystem.
1472 static inline void sb_end_intwrite(struct super_block
*sb
)
1474 __sb_end_write(sb
, SB_FREEZE_FS
);
1478 * sb_start_write - get write access to a superblock
1479 * @sb: the super we write to
1481 * When a process wants to write data or metadata to a file system (i.e. dirty
1482 * a page or an inode), it should embed the operation in a sb_start_write() -
1483 * sb_end_write() pair to get exclusion against file system freezing. This
1484 * function increments number of writers preventing freezing. If the file
1485 * system is already frozen, the function waits until the file system is
1488 * Since freeze protection behaves as a lock, users have to preserve
1489 * ordering of freeze protection and other filesystem locks. Generally,
1490 * freeze protection should be the outermost lock. In particular, we have:
1493 * -> i_mutex (write path, truncate, directory ops, ...)
1494 * -> s_umount (freeze_super, thaw_super)
1496 static inline void sb_start_write(struct super_block
*sb
)
1498 __sb_start_write(sb
, SB_FREEZE_WRITE
, true);
1501 static inline int sb_start_write_trylock(struct super_block
*sb
)
1503 return __sb_start_write(sb
, SB_FREEZE_WRITE
, false);
1507 * sb_start_pagefault - get write access to a superblock from a page fault
1508 * @sb: the super we write to
1510 * When a process starts handling write page fault, it should embed the
1511 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1512 * exclusion against file system freezing. This is needed since the page fault
1513 * is going to dirty a page. This function increments number of running page
1514 * faults preventing freezing. If the file system is already frozen, the
1515 * function waits until the file system is thawed.
1517 * Since page fault freeze protection behaves as a lock, users have to preserve
1518 * ordering of freeze protection and other filesystem locks. It is advised to
1519 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1520 * handling code implies lock dependency:
1523 * -> sb_start_pagefault
1525 static inline void sb_start_pagefault(struct super_block
*sb
)
1527 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
, true);
1531 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1532 * @sb: the super we write to
1534 * This is the third level of protection against filesystem freezing. It is
1535 * free for use by a filesystem. The only requirement is that it must rank
1536 * below sb_start_pagefault.
1538 * For example filesystem can call sb_start_intwrite() when starting a
1539 * transaction which somewhat eases handling of freezing for internal sources
1540 * of filesystem changes (internal fs threads, discarding preallocation on file
1543 static inline void sb_start_intwrite(struct super_block
*sb
)
1545 __sb_start_write(sb
, SB_FREEZE_FS
, true);
1549 extern bool inode_owner_or_capable(const struct inode
*inode
);
1552 * VFS helper functions..
1554 extern int vfs_create(struct inode
*, struct dentry
*, umode_t
, bool);
1555 extern int vfs_mkdir(struct inode
*, struct dentry
*, umode_t
);
1556 extern int vfs_mknod(struct inode
*, struct dentry
*, umode_t
, dev_t
);
1557 extern int vfs_symlink(struct inode
*, struct dentry
*, const char *);
1558 extern int vfs_link(struct dentry
*, struct inode
*, struct dentry
*, struct inode
**);
1559 extern int vfs_rmdir(struct inode
*, struct dentry
*);
1560 extern int vfs_unlink(struct inode
*, struct dentry
*, struct inode
**);
1561 extern int vfs_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*, struct inode
**, unsigned int);
1562 extern int vfs_whiteout(struct inode
*, struct dentry
*);
1565 * VFS file helper functions.
1567 extern void inode_init_owner(struct inode
*inode
, const struct inode
*dir
,
1569 extern bool may_open_dev(const struct path
*path
);
1571 * VFS FS_IOC_FIEMAP helper definitions.
1573 struct fiemap_extent_info
{
1574 unsigned int fi_flags
; /* Flags as passed from user */
1575 unsigned int fi_extents_mapped
; /* Number of mapped extents */
1576 unsigned int fi_extents_max
; /* Size of fiemap_extent array */
1577 struct fiemap_extent __user
*fi_extents_start
; /* Start of
1578 fiemap_extent array */
1580 int fiemap_fill_next_extent(struct fiemap_extent_info
*info
, u64 logical
,
1581 u64 phys
, u64 len
, u32 flags
);
1582 int fiemap_check_flags(struct fiemap_extent_info
*fieinfo
, u32 fs_flags
);
1587 * NOTE! These match bits 12..15 of stat.st_mode
1588 * (ie "(i_mode >> 12) & 15").
1590 #define DT_UNKNOWN 0
1601 * This is the "filldir" function type, used by readdir() to let
1602 * the kernel specify what kind of dirent layout it wants to have.
1603 * This allows the kernel to read directories into kernel space or
1604 * to have different dirent layouts depending on the binary type.
1607 typedef int (*filldir_t
)(struct dir_context
*, const char *, int, loff_t
, u64
,
1610 struct dir_context
{
1611 const filldir_t actor
;
1615 struct block_device_operations
;
1617 /* These macros are for out of kernel modules to test that
1618 * the kernel supports the unlocked_ioctl and compat_ioctl
1619 * fields in struct file_operations. */
1620 #define HAVE_COMPAT_IOCTL 1
1621 #define HAVE_UNLOCKED_IOCTL 1
1624 * These flags let !MMU mmap() govern direct device mapping vs immediate
1625 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1627 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1628 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1629 * NOMMU_MAP_READ: Can be mapped for reading
1630 * NOMMU_MAP_WRITE: Can be mapped for writing
1631 * NOMMU_MAP_EXEC: Can be mapped for execution
1633 #define NOMMU_MAP_COPY 0x00000001
1634 #define NOMMU_MAP_DIRECT 0x00000008
1635 #define NOMMU_MAP_READ VM_MAYREAD
1636 #define NOMMU_MAP_WRITE VM_MAYWRITE
1637 #define NOMMU_MAP_EXEC VM_MAYEXEC
1639 #define NOMMU_VMFLAGS \
1640 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1645 struct file_operations
{
1646 struct module
*owner
;
1647 loff_t (*llseek
) (struct file
*, loff_t
, int);
1648 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
1649 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
1650 ssize_t (*read_iter
) (struct kiocb
*, struct iov_iter
*);
1651 ssize_t (*write_iter
) (struct kiocb
*, struct iov_iter
*);
1652 int (*iterate
) (struct file
*, struct dir_context
*);
1653 int (*iterate_shared
) (struct file
*, struct dir_context
*);
1654 unsigned int (*poll
) (struct file
*, struct poll_table_struct
*);
1655 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
1656 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
1657 int (*mmap
) (struct file
*, struct vm_area_struct
*);
1658 int (*open
) (struct inode
*, struct file
*);
1659 int (*flush
) (struct file
*, fl_owner_t id
);
1660 int (*release
) (struct inode
*, struct file
*);
1661 int (*fsync
) (struct file
*, loff_t
, loff_t
, int datasync
);
1662 int (*fasync
) (int, struct file
*, int);
1663 int (*lock
) (struct file
*, int, struct file_lock
*);
1664 ssize_t (*sendpage
) (struct file
*, struct page
*, int, size_t, loff_t
*, int);
1665 unsigned long (*get_unmapped_area
)(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1666 int (*check_flags
)(int);
1667 int (*flock
) (struct file
*, int, struct file_lock
*);
1668 ssize_t (*splice_write
)(struct pipe_inode_info
*, struct file
*, loff_t
*, size_t, unsigned int);
1669 ssize_t (*splice_read
)(struct file
*, loff_t
*, struct pipe_inode_info
*, size_t, unsigned int);
1670 int (*setlease
)(struct file
*, long, struct file_lock
**, void **);
1671 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
1673 void (*show_fdinfo
)(struct seq_file
*m
, struct file
*f
);
1675 unsigned (*mmap_capabilities
)(struct file
*);
1677 ssize_t (*copy_file_range
)(struct file
*, loff_t
, struct file
*,
1678 loff_t
, size_t, unsigned int);
1679 int (*clone_file_range
)(struct file
*, loff_t
, struct file
*, loff_t
,
1681 ssize_t (*dedupe_file_range
)(struct file
*, u64
, u64
, struct file
*,
1685 struct inode_operations
{
1686 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*, unsigned int);
1687 const char * (*get_link
) (struct dentry
*, struct inode
*, struct delayed_call
*);
1688 int (*permission
) (struct inode
*, int);
1689 struct posix_acl
* (*get_acl
)(struct inode
*, int);
1691 int (*readlink
) (struct dentry
*, char __user
*,int);
1693 int (*create
) (struct inode
*,struct dentry
*, umode_t
, bool);
1694 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
1695 int (*unlink
) (struct inode
*,struct dentry
*);
1696 int (*symlink
) (struct inode
*,struct dentry
*,const char *);
1697 int (*mkdir
) (struct inode
*,struct dentry
*,umode_t
);
1698 int (*rmdir
) (struct inode
*,struct dentry
*);
1699 int (*mknod
) (struct inode
*,struct dentry
*,umode_t
,dev_t
);
1700 int (*rename
) (struct inode
*, struct dentry
*,
1701 struct inode
*, struct dentry
*, unsigned int);
1702 int (*setattr
) (struct dentry
*, struct iattr
*);
1703 int (*getattr
) (struct vfsmount
*mnt
, struct dentry
*, struct kstat
*);
1704 ssize_t (*listxattr
) (struct dentry
*, char *, size_t);
1705 int (*fiemap
)(struct inode
*, struct fiemap_extent_info
*, u64 start
,
1707 int (*update_time
)(struct inode
*, struct timespec
*, int);
1708 int (*atomic_open
)(struct inode
*, struct dentry
*,
1709 struct file
*, unsigned open_flag
,
1710 umode_t create_mode
, int *opened
);
1711 int (*tmpfile
) (struct inode
*, struct dentry
*, umode_t
);
1712 int (*set_acl
)(struct inode
*, struct posix_acl
*, int);
1713 } ____cacheline_aligned
;
1715 ssize_t
rw_copy_check_uvector(int type
, const struct iovec __user
* uvector
,
1716 unsigned long nr_segs
, unsigned long fast_segs
,
1717 struct iovec
*fast_pointer
,
1718 struct iovec
**ret_pointer
);
1720 extern ssize_t
__vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1721 extern ssize_t
__vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1722 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1723 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1724 extern ssize_t
vfs_readv(struct file
*, const struct iovec __user
*,
1725 unsigned long, loff_t
*, int);
1726 extern ssize_t
vfs_writev(struct file
*, const struct iovec __user
*,
1727 unsigned long, loff_t
*, int);
1728 extern ssize_t
vfs_copy_file_range(struct file
*, loff_t
, struct file
*,
1729 loff_t
, size_t, unsigned int);
1730 extern int vfs_clone_file_prep_inodes(struct inode
*inode_in
, loff_t pos_in
,
1731 struct inode
*inode_out
, loff_t pos_out
,
1732 u64
*len
, bool is_dedupe
);
1733 extern int vfs_clone_file_range(struct file
*file_in
, loff_t pos_in
,
1734 struct file
*file_out
, loff_t pos_out
, u64 len
);
1735 extern int vfs_dedupe_file_range_compare(struct inode
*src
, loff_t srcoff
,
1736 struct inode
*dest
, loff_t destoff
,
1737 loff_t len
, bool *is_same
);
1738 extern int vfs_dedupe_file_range(struct file
*file
,
1739 struct file_dedupe_range
*same
);
1741 static inline int do_clone_file_range(struct file
*file_in
, loff_t pos_in
,
1742 struct file
*file_out
, loff_t pos_out
,
1747 sb_start_write(file_inode(file_out
)->i_sb
);
1748 ret
= vfs_clone_file_range(file_in
, pos_in
, file_out
, pos_out
, len
);
1749 sb_end_write(file_inode(file_out
)->i_sb
);
1754 struct super_operations
{
1755 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
1756 void (*destroy_inode
)(struct inode
*);
1758 void (*dirty_inode
) (struct inode
*, int flags
);
1759 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
1760 int (*drop_inode
) (struct inode
*);
1761 void (*evict_inode
) (struct inode
*);
1762 void (*put_super
) (struct super_block
*);
1763 int (*sync_fs
)(struct super_block
*sb
, int wait
);
1764 int (*freeze_super
) (struct super_block
*);
1765 int (*freeze_fs
) (struct super_block
*);
1766 int (*thaw_super
) (struct super_block
*);
1767 int (*unfreeze_fs
) (struct super_block
*);
1768 int (*statfs
) (struct dentry
*, struct kstatfs
*);
1769 int (*remount_fs
) (struct super_block
*, int *, char *);
1770 void (*umount_begin
) (struct super_block
*);
1772 int (*show_options
)(struct seq_file
*, struct dentry
*);
1773 int (*show_devname
)(struct seq_file
*, struct dentry
*);
1774 int (*show_path
)(struct seq_file
*, struct dentry
*);
1775 int (*show_stats
)(struct seq_file
*, struct dentry
*);
1777 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
1778 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
1779 struct dquot
**(*get_dquots
)(struct inode
*);
1781 int (*bdev_try_to_free_page
)(struct super_block
*, struct page
*, gfp_t
);
1782 long (*nr_cached_objects
)(struct super_block
*,
1783 struct shrink_control
*);
1784 long (*free_cached_objects
)(struct super_block
*,
1785 struct shrink_control
*);
1789 * Inode flags - they have no relation to superblock flags now
1791 #define S_SYNC 1 /* Writes are synced at once */
1792 #define S_NOATIME 2 /* Do not update access times */
1793 #define S_APPEND 4 /* Append-only file */
1794 #define S_IMMUTABLE 8 /* Immutable file */
1795 #define S_DEAD 16 /* removed, but still open directory */
1796 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1797 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1798 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1799 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1800 #define S_PRIVATE 512 /* Inode is fs-internal */
1801 #define S_IMA 1024 /* Inode has an associated IMA struct */
1802 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1803 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1804 #ifdef CONFIG_FS_DAX
1805 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1807 #define S_DAX 0 /* Make all the DAX code disappear */
1811 * Note that nosuid etc flags are inode-specific: setting some file-system
1812 * flags just means all the inodes inherit those flags by default. It might be
1813 * possible to override it selectively if you really wanted to with some
1814 * ioctl() that is not currently implemented.
1816 * Exception: MS_RDONLY is always applied to the entire file system.
1818 * Unfortunately, it is possible to change a filesystems flags with it mounted
1819 * with files in use. This means that all of the inodes will not have their
1820 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1821 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1823 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1825 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1826 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1827 ((inode)->i_flags & S_SYNC))
1828 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1829 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1830 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1831 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1832 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1834 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1835 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1836 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1837 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1839 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1840 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1841 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1842 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1843 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1844 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1845 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1846 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1848 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1849 (inode)->i_rdev == WHITEOUT_DEV)
1851 static inline bool HAS_UNMAPPED_ID(struct inode
*inode
)
1853 return !uid_valid(inode
->i_uid
) || !gid_valid(inode
->i_gid
);
1857 * Inode state bits. Protected by inode->i_lock
1859 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1860 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1862 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1863 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1864 * various stages of removing an inode.
1866 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1868 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1869 * fdatasync(). i_atime is the usual cause.
1870 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1871 * these changes separately from I_DIRTY_SYNC so that we
1872 * don't have to write inode on fdatasync() when only
1873 * mtime has changed in it.
1874 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1875 * I_NEW Serves as both a mutex and completion notification.
1876 * New inodes set I_NEW. If two processes both create
1877 * the same inode, one of them will release its inode and
1878 * wait for I_NEW to be released before returning.
1879 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1880 * also cause waiting on I_NEW, without I_NEW actually
1881 * being set. find_inode() uses this to prevent returning
1882 * nearly-dead inodes.
1883 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1884 * is zero. I_FREEING must be set when I_WILL_FREE is
1886 * I_FREEING Set when inode is about to be freed but still has dirty
1887 * pages or buffers attached or the inode itself is still
1889 * I_CLEAR Added by clear_inode(). In this state the inode is
1890 * clean and can be destroyed. Inode keeps I_FREEING.
1892 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1893 * prohibited for many purposes. iget() must wait for
1894 * the inode to be completely released, then create it
1895 * anew. Other functions will just ignore such inodes,
1896 * if appropriate. I_NEW is used for waiting.
1898 * I_SYNC Writeback of inode is running. The bit is set during
1899 * data writeback, and cleared with a wakeup on the bit
1900 * address once it is done. The bit is also used to pin
1901 * the inode in memory for flusher thread.
1903 * I_REFERENCED Marks the inode as recently references on the LRU list.
1905 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1907 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1908 * synchronize competing switching instances and to tell
1909 * wb stat updates to grab mapping->tree_lock. See
1910 * inode_switch_wb_work_fn() for details.
1912 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1914 #define I_DIRTY_SYNC (1 << 0)
1915 #define I_DIRTY_DATASYNC (1 << 1)
1916 #define I_DIRTY_PAGES (1 << 2)
1918 #define I_NEW (1 << __I_NEW)
1919 #define I_WILL_FREE (1 << 4)
1920 #define I_FREEING (1 << 5)
1921 #define I_CLEAR (1 << 6)
1923 #define I_SYNC (1 << __I_SYNC)
1924 #define I_REFERENCED (1 << 8)
1925 #define __I_DIO_WAKEUP 9
1926 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1927 #define I_LINKABLE (1 << 10)
1928 #define I_DIRTY_TIME (1 << 11)
1929 #define __I_DIRTY_TIME_EXPIRED 12
1930 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1931 #define I_WB_SWITCH (1 << 13)
1933 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1934 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1936 extern void __mark_inode_dirty(struct inode
*, int);
1937 static inline void mark_inode_dirty(struct inode
*inode
)
1939 __mark_inode_dirty(inode
, I_DIRTY
);
1942 static inline void mark_inode_dirty_sync(struct inode
*inode
)
1944 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
1947 extern void inc_nlink(struct inode
*inode
);
1948 extern void drop_nlink(struct inode
*inode
);
1949 extern void clear_nlink(struct inode
*inode
);
1950 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
1952 static inline void inode_inc_link_count(struct inode
*inode
)
1955 mark_inode_dirty(inode
);
1958 static inline void inode_dec_link_count(struct inode
*inode
)
1961 mark_inode_dirty(inode
);
1965 * inode_inc_iversion - increments i_version
1966 * @inode: inode that need to be updated
1968 * Every time the inode is modified, the i_version field will be incremented.
1969 * The filesystem has to be mounted with i_version flag
1972 static inline void inode_inc_iversion(struct inode
*inode
)
1974 spin_lock(&inode
->i_lock
);
1976 spin_unlock(&inode
->i_lock
);
1979 enum file_time_flags
{
1986 extern void touch_atime(const struct path
*);
1987 static inline void file_accessed(struct file
*file
)
1989 if (!(file
->f_flags
& O_NOATIME
))
1990 touch_atime(&file
->f_path
);
1993 int sync_inode(struct inode
*inode
, struct writeback_control
*wbc
);
1994 int sync_inode_metadata(struct inode
*inode
, int wait
);
1996 struct file_system_type
{
1999 #define FS_REQUIRES_DEV 1
2000 #define FS_BINARY_MOUNTDATA 2
2001 #define FS_HAS_SUBTYPE 4
2002 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2003 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2004 struct dentry
*(*mount
) (struct file_system_type
*, int,
2005 const char *, void *);
2006 void (*kill_sb
) (struct super_block
*);
2007 struct module
*owner
;
2008 struct file_system_type
* next
;
2009 struct hlist_head fs_supers
;
2011 struct lock_class_key s_lock_key
;
2012 struct lock_class_key s_umount_key
;
2013 struct lock_class_key s_vfs_rename_key
;
2014 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
2016 struct lock_class_key i_lock_key
;
2017 struct lock_class_key i_mutex_key
;
2018 struct lock_class_key i_mutex_dir_key
;
2021 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2023 extern struct dentry
*mount_ns(struct file_system_type
*fs_type
,
2024 int flags
, void *data
, void *ns
, struct user_namespace
*user_ns
,
2025 int (*fill_super
)(struct super_block
*, void *, int));
2026 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
2027 int flags
, const char *dev_name
, void *data
,
2028 int (*fill_super
)(struct super_block
*, void *, int));
2029 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
2030 int flags
, void *data
,
2031 int (*fill_super
)(struct super_block
*, void *, int));
2032 extern struct dentry
*mount_nodev(struct file_system_type
*fs_type
,
2033 int flags
, void *data
,
2034 int (*fill_super
)(struct super_block
*, void *, int));
2035 extern struct dentry
*mount_subtree(struct vfsmount
*mnt
, const char *path
);
2036 void generic_shutdown_super(struct super_block
*sb
);
2037 void kill_block_super(struct super_block
*sb
);
2038 void kill_anon_super(struct super_block
*sb
);
2039 void kill_litter_super(struct super_block
*sb
);
2040 void deactivate_super(struct super_block
*sb
);
2041 void deactivate_locked_super(struct super_block
*sb
);
2042 int set_anon_super(struct super_block
*s
, void *data
);
2043 int get_anon_bdev(dev_t
*);
2044 void free_anon_bdev(dev_t
);
2045 struct super_block
*sget_userns(struct file_system_type
*type
,
2046 int (*test
)(struct super_block
*,void *),
2047 int (*set
)(struct super_block
*,void *),
2048 int flags
, struct user_namespace
*user_ns
,
2050 struct super_block
*sget(struct file_system_type
*type
,
2051 int (*test
)(struct super_block
*,void *),
2052 int (*set
)(struct super_block
*,void *),
2053 int flags
, void *data
);
2054 extern struct dentry
*mount_pseudo_xattr(struct file_system_type
*, char *,
2055 const struct super_operations
*ops
,
2056 const struct xattr_handler
**xattr
,
2057 const struct dentry_operations
*dops
,
2060 static inline struct dentry
*
2061 mount_pseudo(struct file_system_type
*fs_type
, char *name
,
2062 const struct super_operations
*ops
,
2063 const struct dentry_operations
*dops
, unsigned long magic
)
2065 return mount_pseudo_xattr(fs_type
, name
, ops
, NULL
, dops
, magic
);
2068 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2069 #define fops_get(fops) \
2070 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2071 #define fops_put(fops) \
2072 do { if (fops) module_put((fops)->owner); } while(0)
2074 * This one is to be used *ONLY* from ->open() instances.
2075 * fops must be non-NULL, pinned down *and* module dependencies
2076 * should be sufficient to pin the caller down as well.
2078 #define replace_fops(f, fops) \
2080 struct file *__file = (f); \
2081 fops_put(__file->f_op); \
2082 BUG_ON(!(__file->f_op = (fops))); \
2085 extern int register_filesystem(struct file_system_type
*);
2086 extern int unregister_filesystem(struct file_system_type
*);
2087 extern struct vfsmount
*kern_mount_data(struct file_system_type
*, void *data
);
2088 #define kern_mount(type) kern_mount_data(type, NULL)
2089 extern void kern_unmount(struct vfsmount
*mnt
);
2090 extern int may_umount_tree(struct vfsmount
*);
2091 extern int may_umount(struct vfsmount
*);
2092 extern long do_mount(const char *, const char __user
*,
2093 const char *, unsigned long, void *);
2094 extern struct vfsmount
*collect_mounts(const struct path
*);
2095 extern void drop_collected_mounts(struct vfsmount
*);
2096 extern int iterate_mounts(int (*)(struct vfsmount
*, void *), void *,
2098 extern int vfs_statfs(const struct path
*, struct kstatfs
*);
2099 extern int user_statfs(const char __user
*, struct kstatfs
*);
2100 extern int fd_statfs(int, struct kstatfs
*);
2101 extern int vfs_ustat(dev_t
, struct kstatfs
*);
2102 extern int freeze_super(struct super_block
*super
);
2103 extern int thaw_super(struct super_block
*super
);
2104 extern bool our_mnt(struct vfsmount
*mnt
);
2106 extern int current_umask(void);
2108 extern void ihold(struct inode
* inode
);
2109 extern void iput(struct inode
*);
2110 extern int generic_update_time(struct inode
*, struct timespec
*, int);
2113 extern struct kobject
*fs_kobj
;
2115 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2117 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2118 extern int locks_mandatory_locked(struct file
*);
2119 extern int locks_mandatory_area(struct inode
*, struct file
*, loff_t
, loff_t
, unsigned char);
2122 * Candidates for mandatory locking have the setgid bit set
2123 * but no group execute bit - an otherwise meaningless combination.
2126 static inline int __mandatory_lock(struct inode
*ino
)
2128 return (ino
->i_mode
& (S_ISGID
| S_IXGRP
)) == S_ISGID
;
2132 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2133 * otherwise these will be advisory locks
2136 static inline int mandatory_lock(struct inode
*ino
)
2138 return IS_MANDLOCK(ino
) && __mandatory_lock(ino
);
2141 static inline int locks_verify_locked(struct file
*file
)
2143 if (mandatory_lock(locks_inode(file
)))
2144 return locks_mandatory_locked(file
);
2148 static inline int locks_verify_truncate(struct inode
*inode
,
2152 if (!inode
->i_flctx
|| !mandatory_lock(inode
))
2155 if (size
< inode
->i_size
) {
2156 return locks_mandatory_area(inode
, f
, size
, inode
->i_size
- 1,
2159 return locks_mandatory_area(inode
, f
, inode
->i_size
, size
- 1,
2164 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2166 static inline int locks_mandatory_locked(struct file
*file
)
2171 static inline int locks_mandatory_area(struct inode
*inode
, struct file
*filp
,
2172 loff_t start
, loff_t end
, unsigned char type
)
2177 static inline int __mandatory_lock(struct inode
*inode
)
2182 static inline int mandatory_lock(struct inode
*inode
)
2187 static inline int locks_verify_locked(struct file
*file
)
2192 static inline int locks_verify_truncate(struct inode
*inode
, struct file
*filp
,
2198 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2201 #ifdef CONFIG_FILE_LOCKING
2202 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2205 * Since this check is lockless, we must ensure that any refcounts
2206 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2207 * could end up racing with tasks trying to set a new lease on this
2211 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2212 return __break_lease(inode
, mode
, FL_LEASE
);
2216 static inline int break_deleg(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_DELEG
);
2230 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2234 ret
= break_deleg(inode
, O_WRONLY
|O_NONBLOCK
);
2235 if (ret
== -EWOULDBLOCK
&& delegated_inode
) {
2236 *delegated_inode
= inode
;
2242 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2246 ret
= break_deleg(*delegated_inode
, O_WRONLY
);
2247 iput(*delegated_inode
);
2248 *delegated_inode
= NULL
;
2252 static inline int break_layout(struct inode
*inode
, bool wait
)
2255 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2256 return __break_lease(inode
,
2257 wait
? O_WRONLY
: O_WRONLY
| O_NONBLOCK
,
2262 #else /* !CONFIG_FILE_LOCKING */
2263 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2268 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2273 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2278 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2284 static inline int break_layout(struct inode
*inode
, bool wait
)
2289 #endif /* CONFIG_FILE_LOCKING */
2294 const char *name
; /* pointer to actual string */
2295 const __user
char *uptr
; /* original userland pointer */
2296 struct audit_names
*aname
;
2301 extern long vfs_truncate(const struct path
*, loff_t
);
2302 extern int do_truncate(struct dentry
*, loff_t start
, unsigned int time_attrs
,
2304 extern int vfs_fallocate(struct file
*file
, int mode
, loff_t offset
,
2306 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
2308 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
2309 extern struct file
*filp_open(const char *, int, umode_t
);
2310 extern struct file
*file_open_root(struct dentry
*, struct vfsmount
*,
2311 const char *, int, umode_t
);
2312 extern struct file
* dentry_open(const struct path
*, int, const struct cred
*);
2313 extern int filp_close(struct file
*, fl_owner_t id
);
2315 extern struct filename
*getname_flags(const char __user
*, int, int *);
2316 extern struct filename
*getname(const char __user
*);
2317 extern struct filename
*getname_kernel(const char *);
2318 extern void putname(struct filename
*name
);
2324 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
2325 int (*open
)(struct inode
*, struct file
*),
2327 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
2331 extern int ioctl_preallocate(struct file
*filp
, void __user
*argp
);
2334 extern void __init
vfs_caches_init_early(void);
2335 extern void __init
vfs_caches_init(void);
2337 extern struct kmem_cache
*names_cachep
;
2339 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2340 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2343 extern int register_blkdev(unsigned int, const char *);
2344 extern void unregister_blkdev(unsigned int, const char *);
2345 extern struct block_device
*bdget(dev_t
);
2346 extern struct block_device
*bdgrab(struct block_device
*bdev
);
2347 extern void bd_set_size(struct block_device
*, loff_t size
);
2348 extern void bd_forget(struct inode
*inode
);
2349 extern void bdput(struct block_device
*);
2350 extern void invalidate_bdev(struct block_device
*);
2351 extern void iterate_bdevs(void (*)(struct block_device
*, void *), void *);
2352 extern int sync_blockdev(struct block_device
*bdev
);
2353 extern void kill_bdev(struct block_device
*);
2354 extern struct super_block
*freeze_bdev(struct block_device
*);
2355 extern void emergency_thaw_all(void);
2356 extern int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
);
2357 extern int fsync_bdev(struct block_device
*);
2359 extern struct super_block
*blockdev_superblock
;
2361 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2363 return sb
== blockdev_superblock
;
2366 static inline void bd_forget(struct inode
*inode
) {}
2367 static inline int sync_blockdev(struct block_device
*bdev
) { return 0; }
2368 static inline void kill_bdev(struct block_device
*bdev
) {}
2369 static inline void invalidate_bdev(struct block_device
*bdev
) {}
2371 static inline struct super_block
*freeze_bdev(struct block_device
*sb
)
2376 static inline int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
)
2381 static inline void iterate_bdevs(void (*f
)(struct block_device
*, void *), void *arg
)
2385 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2390 extern int sync_filesystem(struct super_block
*);
2391 extern const struct file_operations def_blk_fops
;
2392 extern const struct file_operations def_chr_fops
;
2394 extern int ioctl_by_bdev(struct block_device
*, unsigned, unsigned long);
2395 extern int blkdev_ioctl(struct block_device
*, fmode_t
, unsigned, unsigned long);
2396 extern long compat_blkdev_ioctl(struct file
*, unsigned, unsigned long);
2397 extern int blkdev_get(struct block_device
*bdev
, fmode_t mode
, void *holder
);
2398 extern struct block_device
*blkdev_get_by_path(const char *path
, fmode_t mode
,
2400 extern struct block_device
*blkdev_get_by_dev(dev_t dev
, fmode_t mode
,
2402 extern void blkdev_put(struct block_device
*bdev
, fmode_t mode
);
2403 extern int __blkdev_reread_part(struct block_device
*bdev
);
2404 extern int blkdev_reread_part(struct block_device
*bdev
);
2407 extern int bd_link_disk_holder(struct block_device
*bdev
, struct gendisk
*disk
);
2408 extern void bd_unlink_disk_holder(struct block_device
*bdev
,
2409 struct gendisk
*disk
);
2411 static inline int bd_link_disk_holder(struct block_device
*bdev
,
2412 struct gendisk
*disk
)
2416 static inline void bd_unlink_disk_holder(struct block_device
*bdev
,
2417 struct gendisk
*disk
)
2424 #define CHRDEV_MAJOR_HASH_SIZE 255
2425 /* Marks the bottom of the first segment of free char majors */
2426 #define CHRDEV_MAJOR_DYN_END 234
2427 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
2428 extern int register_chrdev_region(dev_t
, unsigned, const char *);
2429 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
2430 unsigned int count
, const char *name
,
2431 const struct file_operations
*fops
);
2432 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
2433 unsigned int count
, const char *name
);
2434 extern void unregister_chrdev_region(dev_t
, unsigned);
2435 extern void chrdev_show(struct seq_file
*,off_t
);
2437 static inline int register_chrdev(unsigned int major
, const char *name
,
2438 const struct file_operations
*fops
)
2440 return __register_chrdev(major
, 0, 256, name
, fops
);
2443 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2445 __unregister_chrdev(major
, 0, 256, name
);
2448 /* fs/block_dev.c */
2449 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2450 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2453 #define BLKDEV_MAJOR_HASH_SIZE 255
2454 extern const char *__bdevname(dev_t
, char *buffer
);
2455 extern const char *bdevname(struct block_device
*bdev
, char *buffer
);
2456 extern struct block_device
*lookup_bdev(const char *);
2457 extern void blkdev_show(struct seq_file
*,off_t
);
2460 #define BLKDEV_MAJOR_HASH_SIZE 0
2463 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2465 /* Invalid inode operations -- fs/bad_inode.c */
2466 extern void make_bad_inode(struct inode
*);
2467 extern bool is_bad_inode(struct inode
*);
2470 extern void check_disk_size_change(struct gendisk
*disk
,
2471 struct block_device
*bdev
);
2472 extern int revalidate_disk(struct gendisk
*);
2473 extern int check_disk_change(struct block_device
*);
2474 extern int __invalidate_device(struct block_device
*, bool);
2475 extern int invalidate_partition(struct gendisk
*, int);
2477 unsigned long invalidate_mapping_pages(struct address_space
*mapping
,
2478 pgoff_t start
, pgoff_t end
);
2480 static inline void invalidate_remote_inode(struct inode
*inode
)
2482 if (S_ISREG(inode
->i_mode
) || S_ISDIR(inode
->i_mode
) ||
2483 S_ISLNK(inode
->i_mode
))
2484 invalidate_mapping_pages(inode
->i_mapping
, 0, -1);
2486 extern int invalidate_inode_pages2(struct address_space
*mapping
);
2487 extern int invalidate_inode_pages2_range(struct address_space
*mapping
,
2488 pgoff_t start
, pgoff_t end
);
2489 extern int write_inode_now(struct inode
*, int);
2490 extern int filemap_fdatawrite(struct address_space
*);
2491 extern int filemap_flush(struct address_space
*);
2492 extern int filemap_fdatawait(struct address_space
*);
2493 extern void filemap_fdatawait_keep_errors(struct address_space
*);
2494 extern int filemap_fdatawait_range(struct address_space
*, loff_t lstart
,
2496 extern int filemap_write_and_wait(struct address_space
*mapping
);
2497 extern int filemap_write_and_wait_range(struct address_space
*mapping
,
2498 loff_t lstart
, loff_t lend
);
2499 extern int __filemap_fdatawrite_range(struct address_space
*mapping
,
2500 loff_t start
, loff_t end
, int sync_mode
);
2501 extern int filemap_fdatawrite_range(struct address_space
*mapping
,
2502 loff_t start
, loff_t end
);
2503 extern int filemap_check_errors(struct address_space
*mapping
);
2505 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
2507 extern int vfs_fsync(struct file
*file
, int datasync
);
2510 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2511 * to already be updated for the write, and will return either the amount
2512 * of bytes passed in, or an error if syncing the file failed.
2514 static inline ssize_t
generic_write_sync(struct kiocb
*iocb
, ssize_t count
)
2516 if (iocb
->ki_flags
& IOCB_DSYNC
) {
2517 int ret
= vfs_fsync_range(iocb
->ki_filp
,
2518 iocb
->ki_pos
- count
, iocb
->ki_pos
- 1,
2519 (iocb
->ki_flags
& IOCB_SYNC
) ? 0 : 1);
2527 extern void emergency_sync(void);
2528 extern void emergency_remount(void);
2530 extern sector_t
bmap(struct inode
*, sector_t
);
2532 extern int notify_change(struct dentry
*, struct iattr
*, struct inode
**);
2533 extern int inode_permission(struct inode
*, int);
2534 extern int __inode_permission(struct inode
*, int);
2535 extern int generic_permission(struct inode
*, int);
2536 extern int __check_sticky(struct inode
*dir
, struct inode
*inode
);
2538 static inline bool execute_ok(struct inode
*inode
)
2540 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
2543 static inline void file_start_write(struct file
*file
)
2545 if (!S_ISREG(file_inode(file
)->i_mode
))
2547 __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, true);
2550 static inline bool file_start_write_trylock(struct file
*file
)
2552 if (!S_ISREG(file_inode(file
)->i_mode
))
2554 return __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, false);
2557 static inline void file_end_write(struct file
*file
)
2559 if (!S_ISREG(file_inode(file
)->i_mode
))
2561 __sb_end_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
);
2565 * get_write_access() gets write permission for a file.
2566 * put_write_access() releases this write permission.
2567 * This is used for regular files.
2568 * We cannot support write (and maybe mmap read-write shared) accesses and
2569 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2570 * can have the following values:
2571 * 0: no writers, no VM_DENYWRITE mappings
2572 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2573 * > 0: (i_writecount) users are writing to the file.
2575 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2576 * except for the cases where we don't hold i_writecount yet. Then we need to
2577 * use {get,deny}_write_access() - these functions check the sign and refuse
2578 * to do the change if sign is wrong.
2580 static inline int get_write_access(struct inode
*inode
)
2582 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2584 static inline int deny_write_access(struct file
*file
)
2586 struct inode
*inode
= file_inode(file
);
2587 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2589 static inline void put_write_access(struct inode
* inode
)
2591 atomic_dec(&inode
->i_writecount
);
2593 static inline void allow_write_access(struct file
*file
)
2596 atomic_inc(&file_inode(file
)->i_writecount
);
2598 static inline bool inode_is_open_for_write(const struct inode
*inode
)
2600 return atomic_read(&inode
->i_writecount
) > 0;
2604 static inline void i_readcount_dec(struct inode
*inode
)
2606 BUG_ON(!atomic_read(&inode
->i_readcount
));
2607 atomic_dec(&inode
->i_readcount
);
2609 static inline void i_readcount_inc(struct inode
*inode
)
2611 atomic_inc(&inode
->i_readcount
);
2614 static inline void i_readcount_dec(struct inode
*inode
)
2618 static inline void i_readcount_inc(struct inode
*inode
)
2623 extern int do_pipe_flags(int *, int);
2625 #define __kernel_read_file_id(id) \
2626 id(UNKNOWN, unknown) \
2627 id(FIRMWARE, firmware) \
2628 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2629 id(MODULE, kernel-module) \
2630 id(KEXEC_IMAGE, kexec-image) \
2631 id(KEXEC_INITRAMFS, kexec-initramfs) \
2632 id(POLICY, security-policy) \
2635 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2636 #define __fid_stringify(dummy, str) #str,
2638 enum kernel_read_file_id
{
2639 __kernel_read_file_id(__fid_enumify
)
2642 static const char * const kernel_read_file_str
[] = {
2643 __kernel_read_file_id(__fid_stringify
)
2646 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id
)
2648 if (id
< 0 || id
>= READING_MAX_ID
)
2649 return kernel_read_file_str
[READING_UNKNOWN
];
2651 return kernel_read_file_str
[id
];
2654 extern int kernel_read(struct file
*, loff_t
, char *, unsigned long);
2655 extern int kernel_read_file(struct file
*, void **, loff_t
*, loff_t
,
2656 enum kernel_read_file_id
);
2657 extern int kernel_read_file_from_path(char *, void **, loff_t
*, loff_t
,
2658 enum kernel_read_file_id
);
2659 extern int kernel_read_file_from_fd(int, void **, loff_t
*, loff_t
,
2660 enum kernel_read_file_id
);
2661 extern ssize_t
kernel_write(struct file
*, const char *, size_t, loff_t
);
2662 extern ssize_t
__kernel_write(struct file
*, const char *, size_t, loff_t
*);
2663 extern struct file
* open_exec(const char *);
2665 /* fs/dcache.c -- generic fs support functions */
2666 extern bool is_subdir(struct dentry
*, struct dentry
*);
2667 extern bool path_is_under(const struct path
*, const struct path
*);
2669 extern char *file_path(struct file
*, char *, int);
2671 #include <linux/err.h>
2673 /* needed for stackable file system support */
2674 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int whence
);
2676 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int whence
);
2678 extern int inode_init_always(struct super_block
*, struct inode
*);
2679 extern void inode_init_once(struct inode
*);
2680 extern void address_space_init_once(struct address_space
*mapping
);
2681 extern struct inode
* igrab(struct inode
*);
2682 extern ino_t
iunique(struct super_block
*, ino_t
);
2683 extern int inode_needs_sync(struct inode
*inode
);
2684 extern int generic_delete_inode(struct inode
*inode
);
2685 static inline int generic_drop_inode(struct inode
*inode
)
2687 return !inode
->i_nlink
|| inode_unhashed(inode
);
2690 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
2691 unsigned long hashval
, int (*test
)(struct inode
*, void *),
2693 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
2694 int (*test
)(struct inode
*, void *), void *data
);
2695 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
2697 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
2698 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
2699 extern struct inode
*find_inode_nowait(struct super_block
*,
2701 int (*match
)(struct inode
*,
2702 unsigned long, void *),
2704 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
2705 extern int insert_inode_locked(struct inode
*);
2706 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2707 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
2709 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
2711 extern void unlock_new_inode(struct inode
*);
2712 extern unsigned int get_next_ino(void);
2714 extern void __iget(struct inode
* inode
);
2715 extern void iget_failed(struct inode
*);
2716 extern void clear_inode(struct inode
*);
2717 extern void __destroy_inode(struct inode
*);
2718 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
2719 extern struct inode
*new_inode(struct super_block
*sb
);
2720 extern void free_inode_nonrcu(struct inode
*inode
);
2721 extern int should_remove_suid(struct dentry
*);
2722 extern int file_remove_privs(struct file
*);
2724 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
2725 static inline void insert_inode_hash(struct inode
*inode
)
2727 __insert_inode_hash(inode
, inode
->i_ino
);
2730 extern void __remove_inode_hash(struct inode
*);
2731 static inline void remove_inode_hash(struct inode
*inode
)
2733 if (!inode_unhashed(inode
) && !hlist_fake(&inode
->i_hash
))
2734 __remove_inode_hash(inode
);
2737 extern void inode_sb_list_add(struct inode
*inode
);
2740 extern int bdev_read_only(struct block_device
*);
2742 extern int set_blocksize(struct block_device
*, int);
2743 extern int sb_set_blocksize(struct super_block
*, int);
2744 extern int sb_min_blocksize(struct super_block
*, int);
2746 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
2747 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
2748 extern ssize_t
generic_write_checks(struct kiocb
*, struct iov_iter
*);
2749 extern ssize_t
generic_file_read_iter(struct kiocb
*, struct iov_iter
*);
2750 extern ssize_t
__generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2751 extern ssize_t
generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2752 extern ssize_t
generic_file_direct_write(struct kiocb
*, struct iov_iter
*);
2753 extern ssize_t
generic_perform_write(struct file
*, struct iov_iter
*, loff_t
);
2755 ssize_t
vfs_iter_read(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
);
2756 ssize_t
vfs_iter_write(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
);
2758 /* fs/block_dev.c */
2759 extern ssize_t
blkdev_read_iter(struct kiocb
*iocb
, struct iov_iter
*to
);
2760 extern ssize_t
blkdev_write_iter(struct kiocb
*iocb
, struct iov_iter
*from
);
2761 extern int blkdev_fsync(struct file
*filp
, loff_t start
, loff_t end
,
2763 extern void block_sync_page(struct page
*page
);
2766 extern ssize_t
generic_file_splice_read(struct file
*, loff_t
*,
2767 struct pipe_inode_info
*, size_t, unsigned int);
2768 extern ssize_t
iter_file_splice_write(struct pipe_inode_info
*,
2769 struct file
*, loff_t
*, size_t, unsigned int);
2770 extern ssize_t
generic_splice_sendpage(struct pipe_inode_info
*pipe
,
2771 struct file
*out
, loff_t
*, size_t len
, unsigned int flags
);
2772 extern long do_splice_direct(struct file
*in
, loff_t
*ppos
, struct file
*out
,
2773 loff_t
*opos
, size_t len
, unsigned int flags
);
2777 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
2778 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int whence
);
2779 extern loff_t
no_llseek(struct file
*file
, loff_t offset
, int whence
);
2780 extern loff_t
vfs_setpos(struct file
*file
, loff_t offset
, loff_t maxsize
);
2781 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int whence
);
2782 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
2783 int whence
, loff_t maxsize
, loff_t eof
);
2784 extern loff_t
fixed_size_llseek(struct file
*file
, loff_t offset
,
2785 int whence
, loff_t size
);
2786 extern loff_t
no_seek_end_llseek_size(struct file
*, loff_t
, int, loff_t
);
2787 extern loff_t
no_seek_end_llseek(struct file
*, loff_t
, int);
2788 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
2789 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
2792 typedef void (dio_submit_t
)(struct bio
*bio
, struct inode
*inode
,
2793 loff_t file_offset
);
2796 /* need locking between buffered and direct access */
2799 /* filesystem does not support filling holes */
2800 DIO_SKIP_HOLES
= 0x02,
2802 /* filesystem can handle aio writes beyond i_size */
2803 DIO_ASYNC_EXTEND
= 0x04,
2805 /* inode/fs/bdev does not need truncate protection */
2806 DIO_SKIP_DIO_COUNT
= 0x08,
2809 void dio_end_io(struct bio
*bio
, int error
);
2811 ssize_t
__blockdev_direct_IO(struct kiocb
*iocb
, struct inode
*inode
,
2812 struct block_device
*bdev
, struct iov_iter
*iter
,
2813 get_block_t get_block
,
2814 dio_iodone_t end_io
, dio_submit_t submit_io
,
2817 static inline ssize_t
blockdev_direct_IO(struct kiocb
*iocb
,
2818 struct inode
*inode
,
2819 struct iov_iter
*iter
,
2820 get_block_t get_block
)
2822 return __blockdev_direct_IO(iocb
, inode
, inode
->i_sb
->s_bdev
, iter
,
2823 get_block
, NULL
, NULL
, DIO_LOCKING
| DIO_SKIP_HOLES
);
2827 void inode_dio_wait(struct inode
*inode
);
2830 * inode_dio_begin - signal start of a direct I/O requests
2831 * @inode: inode the direct I/O happens on
2833 * This is called once we've finished processing a direct I/O request,
2834 * and is used to wake up callers waiting for direct I/O to be quiesced.
2836 static inline void inode_dio_begin(struct inode
*inode
)
2838 atomic_inc(&inode
->i_dio_count
);
2842 * inode_dio_end - signal finish of a direct I/O requests
2843 * @inode: inode the direct I/O happens on
2845 * This is called once we've finished processing a direct I/O request,
2846 * and is used to wake up callers waiting for direct I/O to be quiesced.
2848 static inline void inode_dio_end(struct inode
*inode
)
2850 if (atomic_dec_and_test(&inode
->i_dio_count
))
2851 wake_up_bit(&inode
->i_state
, __I_DIO_WAKEUP
);
2854 extern void inode_set_flags(struct inode
*inode
, unsigned int flags
,
2857 extern const struct file_operations generic_ro_fops
;
2859 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2861 extern int readlink_copy(char __user
*, int, const char *);
2862 extern int page_readlink(struct dentry
*, char __user
*, int);
2863 extern const char *page_get_link(struct dentry
*, struct inode
*,
2864 struct delayed_call
*);
2865 extern void page_put_link(void *);
2866 extern int __page_symlink(struct inode
*inode
, const char *symname
, int len
,
2868 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
2869 extern const struct inode_operations page_symlink_inode_operations
;
2870 extern void kfree_link(void *);
2871 extern void generic_fillattr(struct inode
*, struct kstat
*);
2872 int vfs_getattr_nosec(struct path
*path
, struct kstat
*stat
);
2873 extern int vfs_getattr(struct path
*, struct kstat
*);
2874 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2875 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2876 void __inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2877 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2878 loff_t
inode_get_bytes(struct inode
*inode
);
2879 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
2880 const char *simple_get_link(struct dentry
*, struct inode
*,
2881 struct delayed_call
*);
2882 extern const struct inode_operations simple_symlink_inode_operations
;
2884 extern int iterate_dir(struct file
*, struct dir_context
*);
2886 extern int vfs_stat(const char __user
*, struct kstat
*);
2887 extern int vfs_lstat(const char __user
*, struct kstat
*);
2888 extern int vfs_fstat(unsigned int, struct kstat
*);
2889 extern int vfs_fstatat(int , const char __user
*, struct kstat
*, int);
2890 extern const char *vfs_get_link(struct dentry
*, struct delayed_call
*);
2891 extern int vfs_readlink(struct dentry
*, char __user
*, int);
2893 extern int __generic_block_fiemap(struct inode
*inode
,
2894 struct fiemap_extent_info
*fieinfo
,
2895 loff_t start
, loff_t len
,
2896 get_block_t
*get_block
);
2897 extern int generic_block_fiemap(struct inode
*inode
,
2898 struct fiemap_extent_info
*fieinfo
, u64 start
,
2899 u64 len
, get_block_t
*get_block
);
2901 extern void get_filesystem(struct file_system_type
*fs
);
2902 extern void put_filesystem(struct file_system_type
*fs
);
2903 extern struct file_system_type
*get_fs_type(const char *name
);
2904 extern struct super_block
*get_super(struct block_device
*);
2905 extern struct super_block
*get_super_thawed(struct block_device
*);
2906 extern struct super_block
*get_super_exclusive_thawed(struct block_device
*bdev
);
2907 extern struct super_block
*get_active_super(struct block_device
*bdev
);
2908 extern void drop_super(struct super_block
*sb
);
2909 extern void drop_super_exclusive(struct super_block
*sb
);
2910 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
2911 extern void iterate_supers_type(struct file_system_type
*,
2912 void (*)(struct super_block
*, void *), void *);
2914 extern int dcache_dir_open(struct inode
*, struct file
*);
2915 extern int dcache_dir_close(struct inode
*, struct file
*);
2916 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
2917 extern int dcache_readdir(struct file
*, struct dir_context
*);
2918 extern int simple_setattr(struct dentry
*, struct iattr
*);
2919 extern int simple_getattr(struct vfsmount
*, struct dentry
*, struct kstat
*);
2920 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
2921 extern int simple_open(struct inode
*inode
, struct file
*file
);
2922 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
2923 extern int simple_unlink(struct inode
*, struct dentry
*);
2924 extern int simple_rmdir(struct inode
*, struct dentry
*);
2925 extern int simple_rename(struct inode
*, struct dentry
*,
2926 struct inode
*, struct dentry
*, unsigned int);
2927 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
2928 extern int simple_empty(struct dentry
*);
2929 extern int simple_readpage(struct file
*file
, struct page
*page
);
2930 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
2931 loff_t pos
, unsigned len
, unsigned flags
,
2932 struct page
**pagep
, void **fsdata
);
2933 extern int simple_write_end(struct file
*file
, struct address_space
*mapping
,
2934 loff_t pos
, unsigned len
, unsigned copied
,
2935 struct page
*page
, void *fsdata
);
2936 extern int always_delete_dentry(const struct dentry
*);
2937 extern struct inode
*alloc_anon_inode(struct super_block
*);
2938 extern int simple_nosetlease(struct file
*, long, struct file_lock
**, void **);
2939 extern const struct dentry_operations simple_dentry_operations
;
2941 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
2942 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
2943 extern const struct file_operations simple_dir_operations
;
2944 extern const struct inode_operations simple_dir_inode_operations
;
2945 extern void make_empty_dir_inode(struct inode
*inode
);
2946 extern bool is_empty_dir_inode(struct inode
*inode
);
2947 struct tree_descr
{ char *name
; const struct file_operations
*ops
; int mode
; };
2948 struct dentry
*d_alloc_name(struct dentry
*, const char *);
2949 extern int simple_fill_super(struct super_block
*, unsigned long, struct tree_descr
*);
2950 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
2951 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
2953 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
2954 loff_t
*ppos
, const void *from
, size_t available
);
2955 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
2956 const void __user
*from
, size_t count
);
2958 extern int __generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
2959 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
2961 extern int generic_check_addressable(unsigned, u64
);
2963 #ifdef CONFIG_MIGRATION
2964 extern int buffer_migrate_page(struct address_space
*,
2965 struct page
*, struct page
*,
2968 #define buffer_migrate_page NULL
2971 extern int setattr_prepare(struct dentry
*, struct iattr
*);
2972 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
2973 extern void setattr_copy(struct inode
*inode
, const struct iattr
*attr
);
2975 extern int file_update_time(struct file
*file
);
2977 extern int generic_show_options(struct seq_file
*m
, struct dentry
*root
);
2978 extern void save_mount_options(struct super_block
*sb
, char *options
);
2979 extern void replace_mount_options(struct super_block
*sb
, char *options
);
2981 static inline bool io_is_direct(struct file
*filp
)
2983 return (filp
->f_flags
& O_DIRECT
) || IS_DAX(filp
->f_mapping
->host
);
2986 static inline int iocb_flags(struct file
*file
)
2989 if (file
->f_flags
& O_APPEND
)
2991 if (io_is_direct(file
))
2993 if ((file
->f_flags
& O_DSYNC
) || IS_SYNC(file
->f_mapping
->host
))
2995 if (file
->f_flags
& __O_SYNC
)
3000 static inline ino_t
parent_ino(struct dentry
*dentry
)
3005 * Don't strictly need d_lock here? If the parent ino could change
3006 * then surely we'd have a deeper race in the caller?
3008 spin_lock(&dentry
->d_lock
);
3009 res
= dentry
->d_parent
->d_inode
->i_ino
;
3010 spin_unlock(&dentry
->d_lock
);
3014 /* Transaction based IO helpers */
3017 * An argresp is stored in an allocated page and holds the
3018 * size of the argument or response, along with its content
3020 struct simple_transaction_argresp
{
3025 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3027 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
3029 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
3030 size_t size
, loff_t
*pos
);
3031 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
3033 void simple_transaction_set(struct file
*file
, size_t n
);
3036 * simple attribute files
3038 * These attributes behave similar to those in sysfs:
3040 * Writing to an attribute immediately sets a value, an open file can be
3041 * written to multiple times.
3043 * Reading from an attribute creates a buffer from the value that might get
3044 * read with multiple read calls. When the attribute has been read
3045 * completely, no further read calls are possible until the file is opened
3048 * All attributes contain a text representation of a numeric value
3049 * that are accessed with the get() and set() functions.
3051 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3052 static int __fops ## _open(struct inode *inode, struct file *file) \
3054 __simple_attr_check_format(__fmt, 0ull); \
3055 return simple_attr_open(inode, file, __get, __set, __fmt); \
3057 static const struct file_operations __fops = { \
3058 .owner = THIS_MODULE, \
3059 .open = __fops ## _open, \
3060 .release = simple_attr_release, \
3061 .read = simple_attr_read, \
3062 .write = simple_attr_write, \
3063 .llseek = generic_file_llseek, \
3066 static inline __printf(1, 2)
3067 void __simple_attr_check_format(const char *fmt
, ...)
3069 /* don't do anything, just let the compiler check the arguments; */
3072 int simple_attr_open(struct inode
*inode
, struct file
*file
,
3073 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
3075 int simple_attr_release(struct inode
*inode
, struct file
*file
);
3076 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
3077 size_t len
, loff_t
*ppos
);
3078 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
3079 size_t len
, loff_t
*ppos
);
3082 int proc_nr_files(struct ctl_table
*table
, int write
,
3083 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3084 int proc_nr_dentry(struct ctl_table
*table
, int write
,
3085 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3086 int proc_nr_inodes(struct ctl_table
*table
, int write
,
3087 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3088 int __init
get_filesystem_list(char *buf
);
3090 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3091 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3093 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3094 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3095 (flag & __FMODE_NONOTIFY)))
3097 static inline bool is_sxid(umode_t mode
)
3099 return (mode
& S_ISUID
) || ((mode
& S_ISGID
) && (mode
& S_IXGRP
));
3102 static inline int check_sticky(struct inode
*dir
, struct inode
*inode
)
3104 if (!(dir
->i_mode
& S_ISVTX
))
3107 return __check_sticky(dir
, inode
);
3110 static inline void inode_has_no_xattr(struct inode
*inode
)
3112 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& MS_NOSEC
))
3113 inode
->i_flags
|= S_NOSEC
;
3116 static inline bool is_root_inode(struct inode
*inode
)
3118 return inode
== inode
->i_sb
->s_root
->d_inode
;
3121 static inline bool dir_emit(struct dir_context
*ctx
,
3122 const char *name
, int namelen
,
3123 u64 ino
, unsigned type
)
3125 return ctx
->actor(ctx
, name
, namelen
, ctx
->pos
, ino
, type
) == 0;
3127 static inline bool dir_emit_dot(struct file
*file
, struct dir_context
*ctx
)
3129 return ctx
->actor(ctx
, ".", 1, ctx
->pos
,
3130 file
->f_path
.dentry
->d_inode
->i_ino
, DT_DIR
) == 0;
3132 static inline bool dir_emit_dotdot(struct file
*file
, struct dir_context
*ctx
)
3134 return ctx
->actor(ctx
, "..", 2, ctx
->pos
,
3135 parent_ino(file
->f_path
.dentry
), DT_DIR
) == 0;
3137 static inline bool dir_emit_dots(struct file
*file
, struct dir_context
*ctx
)
3139 if (ctx
->pos
== 0) {
3140 if (!dir_emit_dot(file
, ctx
))
3144 if (ctx
->pos
== 1) {
3145 if (!dir_emit_dotdot(file
, ctx
))
3151 static inline bool dir_relax(struct inode
*inode
)
3153 inode_unlock(inode
);
3155 return !IS_DEADDIR(inode
);
3158 static inline bool dir_relax_shared(struct inode
*inode
)
3160 inode_unlock_shared(inode
);
3161 inode_lock_shared(inode
);
3162 return !IS_DEADDIR(inode
);
3165 extern bool path_noexec(const struct path
*path
);
3166 extern void inode_nohighmem(struct inode
*inode
);
3168 #endif /* _LINUX_FS_H */