4 #include <linux/linkage.h>
5 #include <linux/wait_bit.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/mm_types.h>
22 #include <linux/capability.h>
23 #include <linux/semaphore.h>
24 #include <linux/fcntl.h>
25 #include <linux/fiemap.h>
26 #include <linux/rculist_bl.h>
27 #include <linux/atomic.h>
28 #include <linux/shrinker.h>
29 #include <linux/migrate_mode.h>
30 #include <linux/uidgid.h>
31 #include <linux/lockdep.h>
32 #include <linux/percpu-rwsem.h>
33 #include <linux/workqueue.h>
34 #include <linux/delayed_call.h>
35 #include <linux/uuid.h>
36 #include <linux/errseq.h>
38 #include <asm/byteorder.h>
39 #include <uapi/linux/fs.h>
41 struct backing_dev_info
;
44 struct export_operations
;
49 struct pipe_inode_info
;
50 struct poll_table_struct
;
52 struct vm_area_struct
;
55 struct swap_info_struct
;
57 struct workqueue_struct
;
60 struct fscrypt_operations
;
62 extern void __init
inode_init(void);
63 extern void __init
inode_init_early(void);
64 extern void __init
files_init(void);
65 extern void __init
files_maxfiles_init(void);
67 extern struct files_stat_struct files_stat
;
68 extern unsigned long get_max_files(void);
69 extern unsigned int sysctl_nr_open
;
70 extern struct inodes_stat_t inodes_stat
;
71 extern int leases_enable
, lease_break_time
;
72 extern int sysctl_protected_symlinks
;
73 extern int sysctl_protected_hardlinks
;
75 typedef __kernel_rwf_t rwf_t
;
78 typedef int (get_block_t
)(struct inode
*inode
, sector_t iblock
,
79 struct buffer_head
*bh_result
, int create
);
80 typedef int (dio_iodone_t
)(struct kiocb
*iocb
, loff_t offset
,
81 ssize_t bytes
, void *private);
83 #define MAY_EXEC 0x00000001
84 #define MAY_WRITE 0x00000002
85 #define MAY_READ 0x00000004
86 #define MAY_APPEND 0x00000008
87 #define MAY_ACCESS 0x00000010
88 #define MAY_OPEN 0x00000020
89 #define MAY_CHDIR 0x00000040
90 /* called from RCU mode, don't block */
91 #define MAY_NOT_BLOCK 0x00000080
94 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
95 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
98 /* file is open for reading */
99 #define FMODE_READ ((__force fmode_t)0x1)
100 /* file is open for writing */
101 #define FMODE_WRITE ((__force fmode_t)0x2)
102 /* file is seekable */
103 #define FMODE_LSEEK ((__force fmode_t)0x4)
104 /* file can be accessed using pread */
105 #define FMODE_PREAD ((__force fmode_t)0x8)
106 /* file can be accessed using pwrite */
107 #define FMODE_PWRITE ((__force fmode_t)0x10)
108 /* File is opened for execution with sys_execve / sys_uselib */
109 #define FMODE_EXEC ((__force fmode_t)0x20)
110 /* File is opened with O_NDELAY (only set for block devices) */
111 #define FMODE_NDELAY ((__force fmode_t)0x40)
112 /* File is opened with O_EXCL (only set for block devices) */
113 #define FMODE_EXCL ((__force fmode_t)0x80)
114 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
115 (specialy hack for floppy.c) */
116 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
117 /* 32bit hashes as llseek() offset (for directories) */
118 #define FMODE_32BITHASH ((__force fmode_t)0x200)
119 /* 64bit hashes as llseek() offset (for directories) */
120 #define FMODE_64BITHASH ((__force fmode_t)0x400)
123 * Don't update ctime and mtime.
125 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
126 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
128 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
130 /* Expect random access pattern */
131 #define FMODE_RANDOM ((__force fmode_t)0x1000)
133 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
134 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
136 /* File is opened with O_PATH; almost nothing can be done with it */
137 #define FMODE_PATH ((__force fmode_t)0x4000)
139 /* File needs atomic accesses to f_pos */
140 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
141 /* Write access to underlying fs */
142 #define FMODE_WRITER ((__force fmode_t)0x10000)
143 /* Has read method(s) */
144 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
145 /* Has write method(s) */
146 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
148 /* File was opened by fanotify and shouldn't generate fanotify events */
149 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
151 /* File is capable of returning -EAGAIN if AIO will block */
152 #define FMODE_AIO_NOWAIT ((__force fmode_t)0x8000000)
155 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
156 * that indicates that they should check the contents of the iovec are
157 * valid, but not check the memory that the iovec elements
160 #define CHECK_IOVEC_ONLY -1
163 * Attribute flags. These should be or-ed together to figure out what
166 #define ATTR_MODE (1 << 0)
167 #define ATTR_UID (1 << 1)
168 #define ATTR_GID (1 << 2)
169 #define ATTR_SIZE (1 << 3)
170 #define ATTR_ATIME (1 << 4)
171 #define ATTR_MTIME (1 << 5)
172 #define ATTR_CTIME (1 << 6)
173 #define ATTR_ATIME_SET (1 << 7)
174 #define ATTR_MTIME_SET (1 << 8)
175 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
176 #define ATTR_ATTR_FLAG (1 << 10)
177 #define ATTR_KILL_SUID (1 << 11)
178 #define ATTR_KILL_SGID (1 << 12)
179 #define ATTR_FILE (1 << 13)
180 #define ATTR_KILL_PRIV (1 << 14)
181 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
182 #define ATTR_TIMES_SET (1 << 16)
183 #define ATTR_TOUCH (1 << 17)
186 * Whiteout is represented by a char device. The following constants define the
187 * mode and device number to use.
189 #define WHITEOUT_MODE 0
190 #define WHITEOUT_DEV 0
193 * This is the Inode Attributes structure, used for notify_change(). It
194 * uses the above definitions as flags, to know which values have changed.
195 * Also, in this manner, a Filesystem can look at only the values it cares
196 * about. Basically, these are the attributes that the VFS layer can
197 * request to change from the FS layer.
199 * Derek Atkins <warlord@MIT.EDU> 94-10-20
202 unsigned int ia_valid
;
207 struct timespec ia_atime
;
208 struct timespec ia_mtime
;
209 struct timespec ia_ctime
;
212 * Not an attribute, but an auxiliary info for filesystems wanting to
213 * implement an ftruncate() like method. NOTE: filesystem should
214 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
216 struct file
*ia_file
;
220 * Includes for diskquotas.
222 #include <linux/quota.h>
225 * Maximum number of layers of fs stack. Needs to be limited to
226 * prevent kernel stack overflow
228 #define FILESYSTEM_MAX_STACK_DEPTH 2
231 * enum positive_aop_returns - aop return codes with specific semantics
233 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
234 * completed, that the page is still locked, and
235 * should be considered active. The VM uses this hint
236 * to return the page to the active list -- it won't
237 * be a candidate for writeback again in the near
238 * future. Other callers must be careful to unlock
239 * the page if they get this return. Returned by
242 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
243 * unlocked it and the page might have been truncated.
244 * The caller should back up to acquiring a new page and
245 * trying again. The aop will be taking reasonable
246 * precautions not to livelock. If the caller held a page
247 * reference, it should drop it before retrying. Returned
250 * address_space_operation functions return these large constants to indicate
251 * special semantics to the caller. These are much larger than the bytes in a
252 * page to allow for functions that return the number of bytes operated on in a
256 enum positive_aop_returns
{
257 AOP_WRITEPAGE_ACTIVATE
= 0x80000,
258 AOP_TRUNCATED_PAGE
= 0x80001,
261 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
262 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
263 * helper code (eg buffer layer)
264 * to clear GFP_FS from alloc */
267 * oh the beauties of C type declarations.
270 struct address_space
;
271 struct writeback_control
;
274 * Write life time hint values.
277 WRITE_LIFE_NOT_SET
= 0,
278 WRITE_LIFE_NONE
= RWH_WRITE_LIFE_NONE
,
279 WRITE_LIFE_SHORT
= RWH_WRITE_LIFE_SHORT
,
280 WRITE_LIFE_MEDIUM
= RWH_WRITE_LIFE_MEDIUM
,
281 WRITE_LIFE_LONG
= RWH_WRITE_LIFE_LONG
,
282 WRITE_LIFE_EXTREME
= RWH_WRITE_LIFE_EXTREME
,
285 #define IOCB_EVENTFD (1 << 0)
286 #define IOCB_APPEND (1 << 1)
287 #define IOCB_DIRECT (1 << 2)
288 #define IOCB_HIPRI (1 << 3)
289 #define IOCB_DSYNC (1 << 4)
290 #define IOCB_SYNC (1 << 5)
291 #define IOCB_WRITE (1 << 6)
292 #define IOCB_NOWAIT (1 << 7)
295 struct file
*ki_filp
;
297 void (*ki_complete
)(struct kiocb
*iocb
, long ret
, long ret2
);
300 enum rw_hint ki_hint
;
301 } __randomize_layout
;
303 static inline bool is_sync_kiocb(struct kiocb
*kiocb
)
305 return kiocb
->ki_complete
== NULL
;
309 * "descriptor" for what we're up to with a read.
310 * This allows us to use the same read code yet
311 * have multiple different users of the data that
312 * we read from a file.
314 * The simplest case just copies the data to user
327 typedef int (*read_actor_t
)(read_descriptor_t
*, struct page
*,
328 unsigned long, unsigned long);
330 struct address_space_operations
{
331 int (*writepage
)(struct page
*page
, struct writeback_control
*wbc
);
332 int (*readpage
)(struct file
*, struct page
*);
334 /* Write back some dirty pages from this mapping. */
335 int (*writepages
)(struct address_space
*, struct writeback_control
*);
337 /* Set a page dirty. Return true if this dirtied it */
338 int (*set_page_dirty
)(struct page
*page
);
340 int (*readpages
)(struct file
*filp
, struct address_space
*mapping
,
341 struct list_head
*pages
, unsigned nr_pages
);
343 int (*write_begin
)(struct file
*, struct address_space
*mapping
,
344 loff_t pos
, unsigned len
, unsigned flags
,
345 struct page
**pagep
, void **fsdata
);
346 int (*write_end
)(struct file
*, struct address_space
*mapping
,
347 loff_t pos
, unsigned len
, unsigned copied
,
348 struct page
*page
, void *fsdata
);
350 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
351 sector_t (*bmap
)(struct address_space
*, sector_t
);
352 void (*invalidatepage
) (struct page
*, unsigned int, unsigned int);
353 int (*releasepage
) (struct page
*, gfp_t
);
354 void (*freepage
)(struct page
*);
355 ssize_t (*direct_IO
)(struct kiocb
*, struct iov_iter
*iter
);
357 * migrate the contents of a page to the specified target. If
358 * migrate_mode is MIGRATE_ASYNC, it must not block.
360 int (*migratepage
) (struct address_space
*,
361 struct page
*, struct page
*, enum migrate_mode
);
362 bool (*isolate_page
)(struct page
*, isolate_mode_t
);
363 void (*putback_page
)(struct page
*);
364 int (*launder_page
) (struct page
*);
365 int (*is_partially_uptodate
) (struct page
*, unsigned long,
367 void (*is_dirty_writeback
) (struct page
*, bool *, bool *);
368 int (*error_remove_page
)(struct address_space
*, struct page
*);
370 /* swapfile support */
371 int (*swap_activate
)(struct swap_info_struct
*sis
, struct file
*file
,
373 void (*swap_deactivate
)(struct file
*file
);
376 extern const struct address_space_operations empty_aops
;
379 * pagecache_write_begin/pagecache_write_end must be used by general code
380 * to write into the pagecache.
382 int pagecache_write_begin(struct file
*, struct address_space
*mapping
,
383 loff_t pos
, unsigned len
, unsigned flags
,
384 struct page
**pagep
, void **fsdata
);
386 int pagecache_write_end(struct file
*, struct address_space
*mapping
,
387 loff_t pos
, unsigned len
, unsigned copied
,
388 struct page
*page
, void *fsdata
);
390 struct address_space
{
391 struct inode
*host
; /* owner: inode, block_device */
392 struct radix_tree_root page_tree
; /* radix tree of all pages */
393 spinlock_t tree_lock
; /* and lock protecting it */
394 atomic_t i_mmap_writable
;/* count VM_SHARED mappings */
395 struct rb_root i_mmap
; /* tree of private and shared mappings */
396 struct rw_semaphore i_mmap_rwsem
; /* protect tree, count, list */
397 /* Protected by tree_lock together with the radix tree */
398 unsigned long nrpages
; /* number of total pages */
399 /* number of shadow or DAX exceptional entries */
400 unsigned long nrexceptional
;
401 pgoff_t writeback_index
;/* writeback starts here */
402 const struct address_space_operations
*a_ops
; /* methods */
403 unsigned long flags
; /* error bits */
404 spinlock_t private_lock
; /* for use by the address_space */
405 gfp_t gfp_mask
; /* implicit gfp mask for allocations */
406 struct list_head private_list
; /* ditto */
407 void *private_data
; /* ditto */
409 } __attribute__((aligned(sizeof(long)))) __randomize_layout
;
411 * On most architectures that alignment is already the case; but
412 * must be enforced here for CRIS, to let the least significant bit
413 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
415 struct request_queue
;
417 struct block_device
{
418 dev_t bd_dev
; /* not a kdev_t - it's a search key */
420 struct inode
* bd_inode
; /* will die */
421 struct super_block
* bd_super
;
422 struct mutex bd_mutex
; /* open/close mutex */
426 bool bd_write_holder
;
428 struct list_head bd_holder_disks
;
430 struct block_device
* bd_contains
;
431 unsigned bd_block_size
;
432 struct hd_struct
* bd_part
;
433 /* number of times partitions within this device have been opened. */
434 unsigned bd_part_count
;
436 struct gendisk
* bd_disk
;
437 struct request_queue
* bd_queue
;
438 struct backing_dev_info
*bd_bdi
;
439 struct list_head bd_list
;
441 * Private data. You must have bd_claim'ed the block_device
442 * to use this. NOTE: bd_claim allows an owner to claim
443 * the same device multiple times, the owner must take special
444 * care to not mess up bd_private for that case.
446 unsigned long bd_private
;
448 /* The counter of freeze processes */
449 int bd_fsfreeze_count
;
450 /* Mutex for freeze */
451 struct mutex bd_fsfreeze_mutex
;
452 } __randomize_layout
;
455 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
458 #define PAGECACHE_TAG_DIRTY 0
459 #define PAGECACHE_TAG_WRITEBACK 1
460 #define PAGECACHE_TAG_TOWRITE 2
462 int mapping_tagged(struct address_space
*mapping
, int tag
);
464 static inline void i_mmap_lock_write(struct address_space
*mapping
)
466 down_write(&mapping
->i_mmap_rwsem
);
469 static inline void i_mmap_unlock_write(struct address_space
*mapping
)
471 up_write(&mapping
->i_mmap_rwsem
);
474 static inline void i_mmap_lock_read(struct address_space
*mapping
)
476 down_read(&mapping
->i_mmap_rwsem
);
479 static inline void i_mmap_unlock_read(struct address_space
*mapping
)
481 up_read(&mapping
->i_mmap_rwsem
);
485 * Might pages of this file be mapped into userspace?
487 static inline int mapping_mapped(struct address_space
*mapping
)
489 return !RB_EMPTY_ROOT(&mapping
->i_mmap
);
493 * Might pages of this file have been modified in userspace?
494 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
495 * marks vma as VM_SHARED if it is shared, and the file was opened for
496 * writing i.e. vma may be mprotected writable even if now readonly.
498 * If i_mmap_writable is negative, no new writable mappings are allowed. You
499 * can only deny writable mappings, if none exists right now.
501 static inline int mapping_writably_mapped(struct address_space
*mapping
)
503 return atomic_read(&mapping
->i_mmap_writable
) > 0;
506 static inline int mapping_map_writable(struct address_space
*mapping
)
508 return atomic_inc_unless_negative(&mapping
->i_mmap_writable
) ?
512 static inline void mapping_unmap_writable(struct address_space
*mapping
)
514 atomic_dec(&mapping
->i_mmap_writable
);
517 static inline int mapping_deny_writable(struct address_space
*mapping
)
519 return atomic_dec_unless_positive(&mapping
->i_mmap_writable
) ?
523 static inline void mapping_allow_writable(struct address_space
*mapping
)
525 atomic_inc(&mapping
->i_mmap_writable
);
529 * Use sequence counter to get consistent i_size on 32-bit processors.
531 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
532 #include <linux/seqlock.h>
533 #define __NEED_I_SIZE_ORDERED
534 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
536 #define i_size_ordered_init(inode) do { } while (0)
540 #define ACL_NOT_CACHED ((void *)(-1))
541 #define ACL_DONT_CACHE ((void *)(-3))
543 static inline struct posix_acl
*
544 uncached_acl_sentinel(struct task_struct
*task
)
546 return (void *)task
+ 1;
550 is_uncached_acl(struct posix_acl
*acl
)
552 return (long)acl
& 1;
555 #define IOP_FASTPERM 0x0001
556 #define IOP_LOOKUP 0x0002
557 #define IOP_NOFOLLOW 0x0004
558 #define IOP_XATTR 0x0008
559 #define IOP_DEFAULT_READLINK 0x0010
561 struct fsnotify_mark_connector
;
564 * Keep mostly read-only and often accessed (especially for
565 * the RCU path lookup and 'stat' data) fields at the beginning
566 * of the 'struct inode'
570 unsigned short i_opflags
;
573 unsigned int i_flags
;
575 #ifdef CONFIG_FS_POSIX_ACL
576 struct posix_acl
*i_acl
;
577 struct posix_acl
*i_default_acl
;
580 const struct inode_operations
*i_op
;
581 struct super_block
*i_sb
;
582 struct address_space
*i_mapping
;
584 #ifdef CONFIG_SECURITY
588 /* Stat data, not accessed from path walking */
591 * Filesystems may only read i_nlink directly. They shall use the
592 * following functions for modification:
594 * (set|clear|inc|drop)_nlink
595 * inode_(inc|dec)_link_count
598 const unsigned int i_nlink
;
599 unsigned int __i_nlink
;
603 struct timespec i_atime
;
604 struct timespec i_mtime
;
605 struct timespec i_ctime
;
606 spinlock_t i_lock
; /* i_blocks, i_bytes, maybe i_size */
607 unsigned short i_bytes
;
608 unsigned int i_blkbits
;
609 enum rw_hint i_write_hint
;
612 #ifdef __NEED_I_SIZE_ORDERED
613 seqcount_t i_size_seqcount
;
617 unsigned long i_state
;
618 struct rw_semaphore i_rwsem
;
620 unsigned long dirtied_when
; /* jiffies of first dirtying */
621 unsigned long dirtied_time_when
;
623 struct hlist_node i_hash
;
624 struct list_head i_io_list
; /* backing dev IO list */
625 #ifdef CONFIG_CGROUP_WRITEBACK
626 struct bdi_writeback
*i_wb
; /* the associated cgroup wb */
628 /* foreign inode detection, see wbc_detach_inode() */
630 u16 i_wb_frn_avg_time
;
631 u16 i_wb_frn_history
;
633 struct list_head i_lru
; /* inode LRU list */
634 struct list_head i_sb_list
;
635 struct list_head i_wb_list
; /* backing dev writeback list */
637 struct hlist_head i_dentry
;
638 struct rcu_head i_rcu
;
642 atomic_t i_dio_count
;
643 atomic_t i_writecount
;
645 atomic_t i_readcount
; /* struct files open RO */
647 const struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
648 struct file_lock_context
*i_flctx
;
649 struct address_space i_data
;
650 struct list_head i_devices
;
652 struct pipe_inode_info
*i_pipe
;
653 struct block_device
*i_bdev
;
661 #ifdef CONFIG_FSNOTIFY
662 __u32 i_fsnotify_mask
; /* all events this inode cares about */
663 struct fsnotify_mark_connector __rcu
*i_fsnotify_marks
;
666 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
667 struct fscrypt_info
*i_crypt_info
;
670 void *i_private
; /* fs or device private pointer */
671 } __randomize_layout
;
673 static inline unsigned int i_blocksize(const struct inode
*node
)
675 return (1 << node
->i_blkbits
);
678 static inline int inode_unhashed(struct inode
*inode
)
680 return hlist_unhashed(&inode
->i_hash
);
684 * inode->i_mutex nesting subclasses for the lock validator:
686 * 0: the object of the current VFS operation
690 * 4: second non-directory
691 * 5: second parent (when locking independent directories in rename)
693 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
694 * non-directories at once.
696 * The locking order between these classes is
697 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
699 enum inode_i_mutex_lock_class
709 static inline void inode_lock(struct inode
*inode
)
711 down_write(&inode
->i_rwsem
);
714 static inline void inode_unlock(struct inode
*inode
)
716 up_write(&inode
->i_rwsem
);
719 static inline void inode_lock_shared(struct inode
*inode
)
721 down_read(&inode
->i_rwsem
);
724 static inline void inode_unlock_shared(struct inode
*inode
)
726 up_read(&inode
->i_rwsem
);
729 static inline int inode_trylock(struct inode
*inode
)
731 return down_write_trylock(&inode
->i_rwsem
);
734 static inline int inode_trylock_shared(struct inode
*inode
)
736 return down_read_trylock(&inode
->i_rwsem
);
739 static inline int inode_is_locked(struct inode
*inode
)
741 return rwsem_is_locked(&inode
->i_rwsem
);
744 static inline void inode_lock_nested(struct inode
*inode
, unsigned subclass
)
746 down_write_nested(&inode
->i_rwsem
, subclass
);
749 void lock_two_nondirectories(struct inode
*, struct inode
*);
750 void unlock_two_nondirectories(struct inode
*, struct inode
*);
753 * NOTE: in a 32bit arch with a preemptable kernel and
754 * an UP compile the i_size_read/write must be atomic
755 * with respect to the local cpu (unlike with preempt disabled),
756 * but they don't need to be atomic with respect to other cpus like in
757 * true SMP (so they need either to either locally disable irq around
758 * the read or for example on x86 they can be still implemented as a
759 * cmpxchg8b without the need of the lock prefix). For SMP compiles
760 * and 64bit archs it makes no difference if preempt is enabled or not.
762 static inline loff_t
i_size_read(const struct inode
*inode
)
764 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
769 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
770 i_size
= inode
->i_size
;
771 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
773 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
777 i_size
= inode
->i_size
;
781 return inode
->i_size
;
786 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
787 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
788 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
790 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
792 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
794 write_seqcount_begin(&inode
->i_size_seqcount
);
795 inode
->i_size
= i_size
;
796 write_seqcount_end(&inode
->i_size_seqcount
);
798 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
800 inode
->i_size
= i_size
;
803 inode
->i_size
= i_size
;
807 static inline unsigned iminor(const struct inode
*inode
)
809 return MINOR(inode
->i_rdev
);
812 static inline unsigned imajor(const struct inode
*inode
)
814 return MAJOR(inode
->i_rdev
);
817 extern struct block_device
*I_BDEV(struct inode
*inode
);
820 rwlock_t lock
; /* protects pid, uid, euid fields */
821 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
822 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
823 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
824 int signum
; /* posix.1b rt signal to be delivered on IO */
828 * Track a single file's readahead state
830 struct file_ra_state
{
831 pgoff_t start
; /* where readahead started */
832 unsigned int size
; /* # of readahead pages */
833 unsigned int async_size
; /* do asynchronous readahead when
834 there are only # of pages ahead */
836 unsigned int ra_pages
; /* Maximum readahead window */
837 unsigned int mmap_miss
; /* Cache miss stat for mmap accesses */
838 loff_t prev_pos
; /* Cache last read() position */
842 * Check if @index falls in the readahead windows.
844 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
846 return (index
>= ra
->start
&&
847 index
< ra
->start
+ ra
->size
);
852 struct llist_node fu_llist
;
853 struct rcu_head fu_rcuhead
;
856 struct inode
*f_inode
; /* cached value */
857 const struct file_operations
*f_op
;
860 * Protects f_ep_links, f_flags.
861 * Must not be taken from IRQ context.
864 enum rw_hint f_write_hint
;
865 atomic_long_t f_count
;
866 unsigned int f_flags
;
868 struct mutex f_pos_lock
;
870 struct fown_struct f_owner
;
871 const struct cred
*f_cred
;
872 struct file_ra_state f_ra
;
875 #ifdef CONFIG_SECURITY
878 /* needed for tty driver, and maybe others */
882 /* Used by fs/eventpoll.c to link all the hooks to this file */
883 struct list_head f_ep_links
;
884 struct list_head f_tfile_llink
;
885 #endif /* #ifdef CONFIG_EPOLL */
886 struct address_space
*f_mapping
;
889 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
894 /* file identifier */
895 unsigned char f_handle
[0];
898 static inline struct file
*get_file(struct file
*f
)
900 atomic_long_inc(&f
->f_count
);
903 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
904 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
905 #define file_count(x) atomic_long_read(&(x)->f_count)
907 #define MAX_NON_LFS ((1UL<<31) - 1)
909 /* Page cache limit. The filesystems should put that into their s_maxbytes
910 limits, otherwise bad things can happen in VM. */
911 #if BITS_PER_LONG==32
912 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
913 #elif BITS_PER_LONG==64
914 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
919 #define FL_DELEG 4 /* NFSv4 delegation */
920 #define FL_ACCESS 8 /* not trying to lock, just looking */
921 #define FL_EXISTS 16 /* when unlocking, test for existence */
922 #define FL_LEASE 32 /* lease held on this file */
923 #define FL_CLOSE 64 /* unlock on close */
924 #define FL_SLEEP 128 /* A blocking lock */
925 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
926 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
927 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
928 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
930 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
933 * Special return value from posix_lock_file() and vfs_lock_file() for
934 * asynchronous locking.
936 #define FILE_LOCK_DEFERRED 1
938 /* legacy typedef, should eventually be removed */
939 typedef void *fl_owner_t
;
943 struct file_lock_operations
{
944 void (*fl_copy_lock
)(struct file_lock
*, struct file_lock
*);
945 void (*fl_release_private
)(struct file_lock
*);
948 struct lock_manager_operations
{
949 int (*lm_compare_owner
)(struct file_lock
*, struct file_lock
*);
950 unsigned long (*lm_owner_key
)(struct file_lock
*);
951 fl_owner_t (*lm_get_owner
)(fl_owner_t
);
952 void (*lm_put_owner
)(fl_owner_t
);
953 void (*lm_notify
)(struct file_lock
*); /* unblock callback */
954 int (*lm_grant
)(struct file_lock
*, int);
955 bool (*lm_break
)(struct file_lock
*);
956 int (*lm_change
)(struct file_lock
*, int, struct list_head
*);
957 void (*lm_setup
)(struct file_lock
*, void **);
960 struct lock_manager
{
961 struct list_head list
;
963 * NFSv4 and up also want opens blocked during the grace period;
970 void locks_start_grace(struct net
*, struct lock_manager
*);
971 void locks_end_grace(struct lock_manager
*);
972 int locks_in_grace(struct net
*);
973 int opens_in_grace(struct net
*);
975 /* that will die - we need it for nfs_lock_info */
976 #include <linux/nfs_fs_i.h>
979 * struct file_lock represents a generic "file lock". It's used to represent
980 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
981 * note that the same struct is used to represent both a request for a lock and
982 * the lock itself, but the same object is never used for both.
984 * FIXME: should we create a separate "struct lock_request" to help distinguish
987 * The varous i_flctx lists are ordered by:
990 * 2) lock range start
993 * Obviously, the last two criteria only matter for POSIX locks.
996 struct file_lock
*fl_next
; /* singly linked list for this inode */
997 struct list_head fl_list
; /* link into file_lock_context */
998 struct hlist_node fl_link
; /* node in global lists */
999 struct list_head fl_block
; /* circular list of blocked processes */
1000 fl_owner_t fl_owner
;
1001 unsigned int fl_flags
;
1002 unsigned char fl_type
;
1003 unsigned int fl_pid
;
1004 int fl_link_cpu
; /* what cpu's list is this on? */
1005 struct pid
*fl_nspid
;
1006 wait_queue_head_t fl_wait
;
1007 struct file
*fl_file
;
1011 struct fasync_struct
* fl_fasync
; /* for lease break notifications */
1012 /* for lease breaks: */
1013 unsigned long fl_break_time
;
1014 unsigned long fl_downgrade_time
;
1016 const struct file_lock_operations
*fl_ops
; /* Callbacks for filesystems */
1017 const struct lock_manager_operations
*fl_lmops
; /* Callbacks for lockmanagers */
1019 struct nfs_lock_info nfs_fl
;
1020 struct nfs4_lock_info nfs4_fl
;
1022 struct list_head link
; /* link in AFS vnode's pending_locks list */
1023 int state
; /* state of grant or error if -ve */
1026 } __randomize_layout
;
1028 struct file_lock_context
{
1029 spinlock_t flc_lock
;
1030 struct list_head flc_flock
;
1031 struct list_head flc_posix
;
1032 struct list_head flc_lease
;
1035 /* The following constant reflects the upper bound of the file/locking space */
1037 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1038 #define OFFSET_MAX INT_LIMIT(loff_t)
1039 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1042 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
1045 * Return the inode to use for locking
1047 * For overlayfs this should be the overlay inode, not the real inode returned
1048 * by file_inode(). For any other fs file_inode(filp) and locks_inode(filp) are
1051 static inline struct inode
*locks_inode(const struct file
*f
)
1053 return f
->f_path
.dentry
->d_inode
;
1056 #ifdef CONFIG_FILE_LOCKING
1057 extern int fcntl_getlk(struct file
*, unsigned int, struct flock
*);
1058 extern int fcntl_setlk(unsigned int, struct file
*, unsigned int,
1061 #if BITS_PER_LONG == 32
1062 extern int fcntl_getlk64(struct file
*, unsigned int, struct flock64
*);
1063 extern int fcntl_setlk64(unsigned int, struct file
*, unsigned int,
1067 extern int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
);
1068 extern int fcntl_getlease(struct file
*filp
);
1071 void locks_free_lock_context(struct inode
*inode
);
1072 void locks_free_lock(struct file_lock
*fl
);
1073 extern void locks_init_lock(struct file_lock
*);
1074 extern struct file_lock
* locks_alloc_lock(void);
1075 extern void locks_copy_lock(struct file_lock
*, struct file_lock
*);
1076 extern void locks_copy_conflock(struct file_lock
*, struct file_lock
*);
1077 extern void locks_remove_posix(struct file
*, fl_owner_t
);
1078 extern void locks_remove_file(struct file
*);
1079 extern void locks_release_private(struct file_lock
*);
1080 extern void posix_test_lock(struct file
*, struct file_lock
*);
1081 extern int posix_lock_file(struct file
*, struct file_lock
*, struct file_lock
*);
1082 extern int posix_unblock_lock(struct file_lock
*);
1083 extern int vfs_test_lock(struct file
*, struct file_lock
*);
1084 extern int vfs_lock_file(struct file
*, unsigned int, struct file_lock
*, struct file_lock
*);
1085 extern int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
);
1086 extern int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
);
1087 extern int __break_lease(struct inode
*inode
, unsigned int flags
, unsigned int type
);
1088 extern void lease_get_mtime(struct inode
*, struct timespec
*time
);
1089 extern int generic_setlease(struct file
*, long, struct file_lock
**, void **priv
);
1090 extern int vfs_setlease(struct file
*, long, struct file_lock
**, void **);
1091 extern int lease_modify(struct file_lock
*, int, struct list_head
*);
1092 struct files_struct
;
1093 extern void show_fd_locks(struct seq_file
*f
,
1094 struct file
*filp
, struct files_struct
*files
);
1095 #else /* !CONFIG_FILE_LOCKING */
1096 static inline int fcntl_getlk(struct file
*file
, unsigned int cmd
,
1097 struct flock __user
*user
)
1102 static inline int fcntl_setlk(unsigned int fd
, struct file
*file
,
1103 unsigned int cmd
, struct flock __user
*user
)
1108 #if BITS_PER_LONG == 32
1109 static inline int fcntl_getlk64(struct file
*file
, unsigned int cmd
,
1110 struct flock64 __user
*user
)
1115 static inline int fcntl_setlk64(unsigned int fd
, struct file
*file
,
1116 unsigned int cmd
, struct flock64 __user
*user
)
1121 static inline int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
)
1126 static inline int fcntl_getlease(struct file
*filp
)
1132 locks_free_lock_context(struct inode
*inode
)
1136 static inline void locks_init_lock(struct file_lock
*fl
)
1141 static inline void locks_copy_conflock(struct file_lock
*new, struct file_lock
*fl
)
1146 static inline void locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
1151 static inline void locks_remove_posix(struct file
*filp
, fl_owner_t owner
)
1156 static inline void locks_remove_file(struct file
*filp
)
1161 static inline void posix_test_lock(struct file
*filp
, struct file_lock
*fl
)
1166 static inline int posix_lock_file(struct file
*filp
, struct file_lock
*fl
,
1167 struct file_lock
*conflock
)
1172 static inline int posix_unblock_lock(struct file_lock
*waiter
)
1177 static inline int vfs_test_lock(struct file
*filp
, struct file_lock
*fl
)
1182 static inline int vfs_lock_file(struct file
*filp
, unsigned int cmd
,
1183 struct file_lock
*fl
, struct file_lock
*conf
)
1188 static inline int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
)
1193 static inline int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
)
1198 static inline int __break_lease(struct inode
*inode
, unsigned int mode
, unsigned int type
)
1203 static inline void lease_get_mtime(struct inode
*inode
, struct timespec
*time
)
1208 static inline int generic_setlease(struct file
*filp
, long arg
,
1209 struct file_lock
**flp
, void **priv
)
1214 static inline int vfs_setlease(struct file
*filp
, long arg
,
1215 struct file_lock
**lease
, void **priv
)
1220 static inline int lease_modify(struct file_lock
*fl
, int arg
,
1221 struct list_head
*dispose
)
1226 struct files_struct
;
1227 static inline void show_fd_locks(struct seq_file
*f
,
1228 struct file
*filp
, struct files_struct
*files
) {}
1229 #endif /* !CONFIG_FILE_LOCKING */
1231 static inline struct inode
*file_inode(const struct file
*f
)
1236 static inline struct dentry
*file_dentry(const struct file
*file
)
1238 return d_real(file
->f_path
.dentry
, file_inode(file
), 0);
1241 static inline int locks_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1243 return locks_lock_inode_wait(locks_inode(filp
), fl
);
1246 struct fasync_struct
{
1250 struct fasync_struct
*fa_next
; /* singly linked list */
1251 struct file
*fa_file
;
1252 struct rcu_head fa_rcu
;
1255 #define FASYNC_MAGIC 0x4601
1257 /* SMP safe fasync helpers: */
1258 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1259 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1260 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1261 extern struct fasync_struct
*fasync_alloc(void);
1262 extern void fasync_free(struct fasync_struct
*);
1264 /* can be called from interrupts */
1265 extern void kill_fasync(struct fasync_struct
**, int, int);
1267 extern void __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1268 extern int f_setown(struct file
*filp
, unsigned long arg
, int force
);
1269 extern void f_delown(struct file
*filp
);
1270 extern pid_t
f_getown(struct file
*filp
);
1271 extern int send_sigurg(struct fown_struct
*fown
);
1279 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1280 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1281 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1282 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1283 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1286 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1287 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1288 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1290 /* sb->s_iflags to limit user namespace mounts */
1291 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1293 /* Possible states of 'frozen' field */
1295 SB_UNFROZEN
= 0, /* FS is unfrozen */
1296 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1297 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1298 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1299 * internal threads if needed) */
1300 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1303 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1306 int frozen
; /* Is sb frozen? */
1307 wait_queue_head_t wait_unfrozen
; /* for get_super_thawed() */
1308 struct percpu_rw_semaphore rw_sem
[SB_FREEZE_LEVELS
];
1311 struct super_block
{
1312 struct list_head s_list
; /* Keep this first */
1313 dev_t s_dev
; /* search index; _not_ kdev_t */
1314 unsigned char s_blocksize_bits
;
1315 unsigned long s_blocksize
;
1316 loff_t s_maxbytes
; /* Max file size */
1317 struct file_system_type
*s_type
;
1318 const struct super_operations
*s_op
;
1319 const struct dquot_operations
*dq_op
;
1320 const struct quotactl_ops
*s_qcop
;
1321 const struct export_operations
*s_export_op
;
1322 unsigned long s_flags
;
1323 unsigned long s_iflags
; /* internal SB_I_* flags */
1324 unsigned long s_magic
;
1325 struct dentry
*s_root
;
1326 struct rw_semaphore s_umount
;
1329 #ifdef CONFIG_SECURITY
1332 const struct xattr_handler
**s_xattr
;
1334 const struct fscrypt_operations
*s_cop
;
1336 struct hlist_bl_head s_anon
; /* anonymous dentries for (nfs) exporting */
1337 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1338 struct block_device
*s_bdev
;
1339 struct backing_dev_info
*s_bdi
;
1340 struct mtd_info
*s_mtd
;
1341 struct hlist_node s_instances
;
1342 unsigned int s_quota_types
; /* Bitmask of supported quota types */
1343 struct quota_info s_dquot
; /* Diskquota specific options */
1345 struct sb_writers s_writers
;
1347 char s_id
[32]; /* Informational name */
1348 uuid_t s_uuid
; /* UUID */
1350 void *s_fs_info
; /* Filesystem private info */
1351 unsigned int s_max_links
;
1354 /* Granularity of c/m/atime in ns.
1355 Cannot be worse than a second */
1359 * The next field is for VFS *only*. No filesystems have any business
1360 * even looking at it. You had been warned.
1362 struct mutex s_vfs_rename_mutex
; /* Kludge */
1365 * Filesystem subtype. If non-empty the filesystem type field
1366 * in /proc/mounts will be "type.subtype"
1370 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1373 * Saved pool identifier for cleancache (-1 means none)
1375 int cleancache_poolid
;
1377 struct shrinker s_shrink
; /* per-sb shrinker handle */
1379 /* Number of inodes with nlink == 0 but still referenced */
1380 atomic_long_t s_remove_count
;
1382 /* Being remounted read-only */
1383 int s_readonly_remount
;
1385 /* AIO completions deferred from interrupt context */
1386 struct workqueue_struct
*s_dio_done_wq
;
1387 struct hlist_head s_pins
;
1390 * Owning user namespace and default context in which to
1391 * interpret filesystem uids, gids, quotas, device nodes,
1392 * xattrs and security labels.
1394 struct user_namespace
*s_user_ns
;
1397 * Keep the lru lists last in the structure so they always sit on their
1398 * own individual cachelines.
1400 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp
;
1401 struct list_lru s_inode_lru ____cacheline_aligned_in_smp
;
1402 struct rcu_head rcu
;
1403 struct work_struct destroy_work
;
1405 struct mutex s_sync_lock
; /* sync serialisation lock */
1408 * Indicates how deep in a filesystem stack this SB is
1412 /* s_inode_list_lock protects s_inodes */
1413 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp
;
1414 struct list_head s_inodes
; /* all inodes */
1416 spinlock_t s_inode_wblist_lock
;
1417 struct list_head s_inodes_wb
; /* writeback inodes */
1418 } __randomize_layout
;
1420 /* Helper functions so that in most cases filesystems will
1421 * not need to deal directly with kuid_t and kgid_t and can
1422 * instead deal with the raw numeric values that are stored
1423 * in the filesystem.
1425 static inline uid_t
i_uid_read(const struct inode
*inode
)
1427 return from_kuid(inode
->i_sb
->s_user_ns
, inode
->i_uid
);
1430 static inline gid_t
i_gid_read(const struct inode
*inode
)
1432 return from_kgid(inode
->i_sb
->s_user_ns
, inode
->i_gid
);
1435 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
1437 inode
->i_uid
= make_kuid(inode
->i_sb
->s_user_ns
, uid
);
1440 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
1442 inode
->i_gid
= make_kgid(inode
->i_sb
->s_user_ns
, gid
);
1445 extern struct timespec
current_time(struct inode
*inode
);
1448 * Snapshotting support.
1451 void __sb_end_write(struct super_block
*sb
, int level
);
1452 int __sb_start_write(struct super_block
*sb
, int level
, bool wait
);
1454 #define __sb_writers_acquired(sb, lev) \
1455 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1456 #define __sb_writers_release(sb, lev) \
1457 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1460 * sb_end_write - drop write access to a superblock
1461 * @sb: the super we wrote to
1463 * Decrement number of writers to the filesystem. Wake up possible waiters
1464 * wanting to freeze the filesystem.
1466 static inline void sb_end_write(struct super_block
*sb
)
1468 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1472 * sb_end_pagefault - drop write access to a superblock from a page fault
1473 * @sb: the super we wrote to
1475 * Decrement number of processes handling write page fault to the filesystem.
1476 * Wake up possible waiters wanting to freeze the filesystem.
1478 static inline void sb_end_pagefault(struct super_block
*sb
)
1480 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1484 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1485 * @sb: the super we wrote to
1487 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1488 * waiters wanting to freeze the filesystem.
1490 static inline void sb_end_intwrite(struct super_block
*sb
)
1492 __sb_end_write(sb
, SB_FREEZE_FS
);
1496 * sb_start_write - get write access to a superblock
1497 * @sb: the super we write to
1499 * When a process wants to write data or metadata to a file system (i.e. dirty
1500 * a page or an inode), it should embed the operation in a sb_start_write() -
1501 * sb_end_write() pair to get exclusion against file system freezing. This
1502 * function increments number of writers preventing freezing. If the file
1503 * system is already frozen, the function waits until the file system is
1506 * Since freeze protection behaves as a lock, users have to preserve
1507 * ordering of freeze protection and other filesystem locks. Generally,
1508 * freeze protection should be the outermost lock. In particular, we have:
1511 * -> i_mutex (write path, truncate, directory ops, ...)
1512 * -> s_umount (freeze_super, thaw_super)
1514 static inline void sb_start_write(struct super_block
*sb
)
1516 __sb_start_write(sb
, SB_FREEZE_WRITE
, true);
1519 static inline int sb_start_write_trylock(struct super_block
*sb
)
1521 return __sb_start_write(sb
, SB_FREEZE_WRITE
, false);
1525 * sb_start_pagefault - get write access to a superblock from a page fault
1526 * @sb: the super we write to
1528 * When a process starts handling write page fault, it should embed the
1529 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1530 * exclusion against file system freezing. This is needed since the page fault
1531 * is going to dirty a page. This function increments number of running page
1532 * faults preventing freezing. If the file system is already frozen, the
1533 * function waits until the file system is thawed.
1535 * Since page fault freeze protection behaves as a lock, users have to preserve
1536 * ordering of freeze protection and other filesystem locks. It is advised to
1537 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1538 * handling code implies lock dependency:
1541 * -> sb_start_pagefault
1543 static inline void sb_start_pagefault(struct super_block
*sb
)
1545 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
, true);
1549 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1550 * @sb: the super we write to
1552 * This is the third level of protection against filesystem freezing. It is
1553 * free for use by a filesystem. The only requirement is that it must rank
1554 * below sb_start_pagefault.
1556 * For example filesystem can call sb_start_intwrite() when starting a
1557 * transaction which somewhat eases handling of freezing for internal sources
1558 * of filesystem changes (internal fs threads, discarding preallocation on file
1561 static inline void sb_start_intwrite(struct super_block
*sb
)
1563 __sb_start_write(sb
, SB_FREEZE_FS
, true);
1567 extern bool inode_owner_or_capable(const struct inode
*inode
);
1570 * VFS helper functions..
1572 extern int vfs_create(struct inode
*, struct dentry
*, umode_t
, bool);
1573 extern int vfs_mkdir(struct inode
*, struct dentry
*, umode_t
);
1574 extern int vfs_mknod(struct inode
*, struct dentry
*, umode_t
, dev_t
);
1575 extern int vfs_symlink(struct inode
*, struct dentry
*, const char *);
1576 extern int vfs_link(struct dentry
*, struct inode
*, struct dentry
*, struct inode
**);
1577 extern int vfs_rmdir(struct inode
*, struct dentry
*);
1578 extern int vfs_unlink(struct inode
*, struct dentry
*, struct inode
**);
1579 extern int vfs_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*, struct inode
**, unsigned int);
1580 extern int vfs_whiteout(struct inode
*, struct dentry
*);
1582 extern struct dentry
*vfs_tmpfile(struct dentry
*dentry
, umode_t mode
,
1586 * VFS file helper functions.
1588 extern void inode_init_owner(struct inode
*inode
, const struct inode
*dir
,
1590 extern bool may_open_dev(const struct path
*path
);
1592 * VFS FS_IOC_FIEMAP helper definitions.
1594 struct fiemap_extent_info
{
1595 unsigned int fi_flags
; /* Flags as passed from user */
1596 unsigned int fi_extents_mapped
; /* Number of mapped extents */
1597 unsigned int fi_extents_max
; /* Size of fiemap_extent array */
1598 struct fiemap_extent __user
*fi_extents_start
; /* Start of
1599 fiemap_extent array */
1601 int fiemap_fill_next_extent(struct fiemap_extent_info
*info
, u64 logical
,
1602 u64 phys
, u64 len
, u32 flags
);
1603 int fiemap_check_flags(struct fiemap_extent_info
*fieinfo
, u32 fs_flags
);
1608 * NOTE! These match bits 12..15 of stat.st_mode
1609 * (ie "(i_mode >> 12) & 15").
1611 #define DT_UNKNOWN 0
1622 * This is the "filldir" function type, used by readdir() to let
1623 * the kernel specify what kind of dirent layout it wants to have.
1624 * This allows the kernel to read directories into kernel space or
1625 * to have different dirent layouts depending on the binary type.
1628 typedef int (*filldir_t
)(struct dir_context
*, const char *, int, loff_t
, u64
,
1631 struct dir_context
{
1632 const filldir_t actor
;
1636 struct block_device_operations
;
1638 /* These macros are for out of kernel modules to test that
1639 * the kernel supports the unlocked_ioctl and compat_ioctl
1640 * fields in struct file_operations. */
1641 #define HAVE_COMPAT_IOCTL 1
1642 #define HAVE_UNLOCKED_IOCTL 1
1645 * These flags let !MMU mmap() govern direct device mapping vs immediate
1646 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1648 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1649 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1650 * NOMMU_MAP_READ: Can be mapped for reading
1651 * NOMMU_MAP_WRITE: Can be mapped for writing
1652 * NOMMU_MAP_EXEC: Can be mapped for execution
1654 #define NOMMU_MAP_COPY 0x00000001
1655 #define NOMMU_MAP_DIRECT 0x00000008
1656 #define NOMMU_MAP_READ VM_MAYREAD
1657 #define NOMMU_MAP_WRITE VM_MAYWRITE
1658 #define NOMMU_MAP_EXEC VM_MAYEXEC
1660 #define NOMMU_VMFLAGS \
1661 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1666 struct file_operations
{
1667 struct module
*owner
;
1668 loff_t (*llseek
) (struct file
*, loff_t
, int);
1669 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
1670 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
1671 ssize_t (*read_iter
) (struct kiocb
*, struct iov_iter
*);
1672 ssize_t (*write_iter
) (struct kiocb
*, struct iov_iter
*);
1673 int (*iterate
) (struct file
*, struct dir_context
*);
1674 int (*iterate_shared
) (struct file
*, struct dir_context
*);
1675 unsigned int (*poll
) (struct file
*, struct poll_table_struct
*);
1676 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
1677 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
1678 int (*mmap
) (struct file
*, struct vm_area_struct
*);
1679 int (*open
) (struct inode
*, struct file
*);
1680 int (*flush
) (struct file
*, fl_owner_t id
);
1681 int (*release
) (struct inode
*, struct file
*);
1682 int (*fsync
) (struct file
*, loff_t
, loff_t
, int datasync
);
1683 int (*fasync
) (int, struct file
*, int);
1684 int (*lock
) (struct file
*, int, struct file_lock
*);
1685 ssize_t (*sendpage
) (struct file
*, struct page
*, int, size_t, loff_t
*, int);
1686 unsigned long (*get_unmapped_area
)(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1687 int (*check_flags
)(int);
1688 int (*flock
) (struct file
*, int, struct file_lock
*);
1689 ssize_t (*splice_write
)(struct pipe_inode_info
*, struct file
*, loff_t
*, size_t, unsigned int);
1690 ssize_t (*splice_read
)(struct file
*, loff_t
*, struct pipe_inode_info
*, size_t, unsigned int);
1691 int (*setlease
)(struct file
*, long, struct file_lock
**, void **);
1692 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
1694 void (*show_fdinfo
)(struct seq_file
*m
, struct file
*f
);
1696 unsigned (*mmap_capabilities
)(struct file
*);
1698 ssize_t (*copy_file_range
)(struct file
*, loff_t
, struct file
*,
1699 loff_t
, size_t, unsigned int);
1700 int (*clone_file_range
)(struct file
*, loff_t
, struct file
*, loff_t
,
1702 ssize_t (*dedupe_file_range
)(struct file
*, u64
, u64
, struct file
*,
1704 } __randomize_layout
;
1706 struct inode_operations
{
1707 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*, unsigned int);
1708 const char * (*get_link
) (struct dentry
*, struct inode
*, struct delayed_call
*);
1709 int (*permission
) (struct inode
*, int);
1710 struct posix_acl
* (*get_acl
)(struct inode
*, int);
1712 int (*readlink
) (struct dentry
*, char __user
*,int);
1714 int (*create
) (struct inode
*,struct dentry
*, umode_t
, bool);
1715 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
1716 int (*unlink
) (struct inode
*,struct dentry
*);
1717 int (*symlink
) (struct inode
*,struct dentry
*,const char *);
1718 int (*mkdir
) (struct inode
*,struct dentry
*,umode_t
);
1719 int (*rmdir
) (struct inode
*,struct dentry
*);
1720 int (*mknod
) (struct inode
*,struct dentry
*,umode_t
,dev_t
);
1721 int (*rename
) (struct inode
*, struct dentry
*,
1722 struct inode
*, struct dentry
*, unsigned int);
1723 int (*setattr
) (struct dentry
*, struct iattr
*);
1724 int (*getattr
) (const struct path
*, struct kstat
*, u32
, unsigned int);
1725 ssize_t (*listxattr
) (struct dentry
*, char *, size_t);
1726 int (*fiemap
)(struct inode
*, struct fiemap_extent_info
*, u64 start
,
1728 int (*update_time
)(struct inode
*, struct timespec
*, int);
1729 int (*atomic_open
)(struct inode
*, struct dentry
*,
1730 struct file
*, unsigned open_flag
,
1731 umode_t create_mode
, int *opened
);
1732 int (*tmpfile
) (struct inode
*, struct dentry
*, umode_t
);
1733 int (*set_acl
)(struct inode
*, struct posix_acl
*, int);
1734 } ____cacheline_aligned
;
1736 static inline ssize_t
call_read_iter(struct file
*file
, struct kiocb
*kio
,
1737 struct iov_iter
*iter
)
1739 return file
->f_op
->read_iter(kio
, iter
);
1742 static inline ssize_t
call_write_iter(struct file
*file
, struct kiocb
*kio
,
1743 struct iov_iter
*iter
)
1745 return file
->f_op
->write_iter(kio
, iter
);
1748 static inline int call_mmap(struct file
*file
, struct vm_area_struct
*vma
)
1750 return file
->f_op
->mmap(file
, vma
);
1753 ssize_t
rw_copy_check_uvector(int type
, const struct iovec __user
* uvector
,
1754 unsigned long nr_segs
, unsigned long fast_segs
,
1755 struct iovec
*fast_pointer
,
1756 struct iovec
**ret_pointer
);
1758 extern ssize_t
__vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1759 extern ssize_t
__vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1760 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1761 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1762 extern ssize_t
vfs_readv(struct file
*, const struct iovec __user
*,
1763 unsigned long, loff_t
*, rwf_t
);
1764 extern ssize_t
vfs_writev(struct file
*, const struct iovec __user
*,
1765 unsigned long, loff_t
*, rwf_t
);
1766 extern ssize_t
vfs_copy_file_range(struct file
*, loff_t
, struct file
*,
1767 loff_t
, size_t, unsigned int);
1768 extern int vfs_clone_file_prep_inodes(struct inode
*inode_in
, loff_t pos_in
,
1769 struct inode
*inode_out
, loff_t pos_out
,
1770 u64
*len
, bool is_dedupe
);
1771 extern int vfs_clone_file_range(struct file
*file_in
, loff_t pos_in
,
1772 struct file
*file_out
, loff_t pos_out
, u64 len
);
1773 extern int vfs_dedupe_file_range_compare(struct inode
*src
, loff_t srcoff
,
1774 struct inode
*dest
, loff_t destoff
,
1775 loff_t len
, bool *is_same
);
1776 extern int vfs_dedupe_file_range(struct file
*file
,
1777 struct file_dedupe_range
*same
);
1779 struct super_operations
{
1780 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
1781 void (*destroy_inode
)(struct inode
*);
1783 void (*dirty_inode
) (struct inode
*, int flags
);
1784 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
1785 int (*drop_inode
) (struct inode
*);
1786 void (*evict_inode
) (struct inode
*);
1787 void (*put_super
) (struct super_block
*);
1788 int (*sync_fs
)(struct super_block
*sb
, int wait
);
1789 int (*freeze_super
) (struct super_block
*);
1790 int (*freeze_fs
) (struct super_block
*);
1791 int (*thaw_super
) (struct super_block
*);
1792 int (*unfreeze_fs
) (struct super_block
*);
1793 int (*statfs
) (struct dentry
*, struct kstatfs
*);
1794 int (*remount_fs
) (struct super_block
*, int *, char *);
1795 void (*umount_begin
) (struct super_block
*);
1797 int (*show_options
)(struct seq_file
*, struct dentry
*);
1798 int (*show_devname
)(struct seq_file
*, struct dentry
*);
1799 int (*show_path
)(struct seq_file
*, struct dentry
*);
1800 int (*show_stats
)(struct seq_file
*, struct dentry
*);
1802 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
1803 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
1804 struct dquot
**(*get_dquots
)(struct inode
*);
1806 int (*bdev_try_to_free_page
)(struct super_block
*, struct page
*, gfp_t
);
1807 long (*nr_cached_objects
)(struct super_block
*,
1808 struct shrink_control
*);
1809 long (*free_cached_objects
)(struct super_block
*,
1810 struct shrink_control
*);
1814 * Inode flags - they have no relation to superblock flags now
1816 #define S_SYNC 1 /* Writes are synced at once */
1817 #define S_NOATIME 2 /* Do not update access times */
1818 #define S_APPEND 4 /* Append-only file */
1819 #define S_IMMUTABLE 8 /* Immutable file */
1820 #define S_DEAD 16 /* removed, but still open directory */
1821 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1822 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1823 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1824 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1825 #define S_PRIVATE 512 /* Inode is fs-internal */
1826 #define S_IMA 1024 /* Inode has an associated IMA struct */
1827 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1828 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1829 #ifdef CONFIG_FS_DAX
1830 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1832 #define S_DAX 0 /* Make all the DAX code disappear */
1836 * Note that nosuid etc flags are inode-specific: setting some file-system
1837 * flags just means all the inodes inherit those flags by default. It might be
1838 * possible to override it selectively if you really wanted to with some
1839 * ioctl() that is not currently implemented.
1841 * Exception: MS_RDONLY is always applied to the entire file system.
1843 * Unfortunately, it is possible to change a filesystems flags with it mounted
1844 * with files in use. This means that all of the inodes will not have their
1845 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1846 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1848 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1850 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1851 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1852 ((inode)->i_flags & S_SYNC))
1853 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1854 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1855 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1856 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1857 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1859 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1860 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1861 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1862 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1864 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1865 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1866 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1867 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1868 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1869 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1870 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1871 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1873 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1874 (inode)->i_rdev == WHITEOUT_DEV)
1876 static inline bool HAS_UNMAPPED_ID(struct inode
*inode
)
1878 return !uid_valid(inode
->i_uid
) || !gid_valid(inode
->i_gid
);
1881 static inline enum rw_hint
file_write_hint(struct file
*file
)
1883 if (file
->f_write_hint
!= WRITE_LIFE_NOT_SET
)
1884 return file
->f_write_hint
;
1886 return file_inode(file
)->i_write_hint
;
1889 static inline int iocb_flags(struct file
*file
);
1891 static inline void init_sync_kiocb(struct kiocb
*kiocb
, struct file
*filp
)
1893 *kiocb
= (struct kiocb
) {
1895 .ki_flags
= iocb_flags(filp
),
1896 .ki_hint
= file_write_hint(filp
),
1901 * Inode state bits. Protected by inode->i_lock
1903 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1904 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1906 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1907 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1908 * various stages of removing an inode.
1910 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1912 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1913 * fdatasync(). i_atime is the usual cause.
1914 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1915 * these changes separately from I_DIRTY_SYNC so that we
1916 * don't have to write inode on fdatasync() when only
1917 * mtime has changed in it.
1918 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1919 * I_NEW Serves as both a mutex and completion notification.
1920 * New inodes set I_NEW. If two processes both create
1921 * the same inode, one of them will release its inode and
1922 * wait for I_NEW to be released before returning.
1923 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1924 * also cause waiting on I_NEW, without I_NEW actually
1925 * being set. find_inode() uses this to prevent returning
1926 * nearly-dead inodes.
1927 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1928 * is zero. I_FREEING must be set when I_WILL_FREE is
1930 * I_FREEING Set when inode is about to be freed but still has dirty
1931 * pages or buffers attached or the inode itself is still
1933 * I_CLEAR Added by clear_inode(). In this state the inode is
1934 * clean and can be destroyed. Inode keeps I_FREEING.
1936 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1937 * prohibited for many purposes. iget() must wait for
1938 * the inode to be completely released, then create it
1939 * anew. Other functions will just ignore such inodes,
1940 * if appropriate. I_NEW is used for waiting.
1942 * I_SYNC Writeback of inode is running. The bit is set during
1943 * data writeback, and cleared with a wakeup on the bit
1944 * address once it is done. The bit is also used to pin
1945 * the inode in memory for flusher thread.
1947 * I_REFERENCED Marks the inode as recently references on the LRU list.
1949 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1951 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1952 * synchronize competing switching instances and to tell
1953 * wb stat updates to grab mapping->tree_lock. See
1954 * inode_switch_wb_work_fn() for details.
1956 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
1957 * and work dirs among overlayfs mounts.
1959 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1961 #define I_DIRTY_SYNC (1 << 0)
1962 #define I_DIRTY_DATASYNC (1 << 1)
1963 #define I_DIRTY_PAGES (1 << 2)
1965 #define I_NEW (1 << __I_NEW)
1966 #define I_WILL_FREE (1 << 4)
1967 #define I_FREEING (1 << 5)
1968 #define I_CLEAR (1 << 6)
1970 #define I_SYNC (1 << __I_SYNC)
1971 #define I_REFERENCED (1 << 8)
1972 #define __I_DIO_WAKEUP 9
1973 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1974 #define I_LINKABLE (1 << 10)
1975 #define I_DIRTY_TIME (1 << 11)
1976 #define __I_DIRTY_TIME_EXPIRED 12
1977 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1978 #define I_WB_SWITCH (1 << 13)
1979 #define I_OVL_INUSE (1 << 14)
1981 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1982 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1984 extern void __mark_inode_dirty(struct inode
*, int);
1985 static inline void mark_inode_dirty(struct inode
*inode
)
1987 __mark_inode_dirty(inode
, I_DIRTY
);
1990 static inline void mark_inode_dirty_sync(struct inode
*inode
)
1992 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
1995 extern void inc_nlink(struct inode
*inode
);
1996 extern void drop_nlink(struct inode
*inode
);
1997 extern void clear_nlink(struct inode
*inode
);
1998 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
2000 static inline void inode_inc_link_count(struct inode
*inode
)
2003 mark_inode_dirty(inode
);
2006 static inline void inode_dec_link_count(struct inode
*inode
)
2009 mark_inode_dirty(inode
);
2013 * inode_inc_iversion - increments i_version
2014 * @inode: inode that need to be updated
2016 * Every time the inode is modified, the i_version field will be incremented.
2017 * The filesystem has to be mounted with i_version flag
2020 static inline void inode_inc_iversion(struct inode
*inode
)
2022 spin_lock(&inode
->i_lock
);
2024 spin_unlock(&inode
->i_lock
);
2027 enum file_time_flags
{
2034 extern void touch_atime(const struct path
*);
2035 static inline void file_accessed(struct file
*file
)
2037 if (!(file
->f_flags
& O_NOATIME
))
2038 touch_atime(&file
->f_path
);
2041 int sync_inode(struct inode
*inode
, struct writeback_control
*wbc
);
2042 int sync_inode_metadata(struct inode
*inode
, int wait
);
2044 struct file_system_type
{
2047 #define FS_REQUIRES_DEV 1
2048 #define FS_BINARY_MOUNTDATA 2
2049 #define FS_HAS_SUBTYPE 4
2050 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2051 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2052 struct dentry
*(*mount
) (struct file_system_type
*, int,
2053 const char *, void *);
2054 void (*kill_sb
) (struct super_block
*);
2055 struct module
*owner
;
2056 struct file_system_type
* next
;
2057 struct hlist_head fs_supers
;
2059 struct lock_class_key s_lock_key
;
2060 struct lock_class_key s_umount_key
;
2061 struct lock_class_key s_vfs_rename_key
;
2062 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
2064 struct lock_class_key i_lock_key
;
2065 struct lock_class_key i_mutex_key
;
2066 struct lock_class_key i_mutex_dir_key
;
2069 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2071 extern struct dentry
*mount_ns(struct file_system_type
*fs_type
,
2072 int flags
, void *data
, void *ns
, struct user_namespace
*user_ns
,
2073 int (*fill_super
)(struct super_block
*, void *, int));
2074 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
2075 int flags
, const char *dev_name
, void *data
,
2076 int (*fill_super
)(struct super_block
*, void *, int));
2077 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
2078 int flags
, void *data
,
2079 int (*fill_super
)(struct super_block
*, void *, int));
2080 extern struct dentry
*mount_nodev(struct file_system_type
*fs_type
,
2081 int flags
, void *data
,
2082 int (*fill_super
)(struct super_block
*, void *, int));
2083 extern struct dentry
*mount_subtree(struct vfsmount
*mnt
, const char *path
);
2084 void generic_shutdown_super(struct super_block
*sb
);
2085 void kill_block_super(struct super_block
*sb
);
2086 void kill_anon_super(struct super_block
*sb
);
2087 void kill_litter_super(struct super_block
*sb
);
2088 void deactivate_super(struct super_block
*sb
);
2089 void deactivate_locked_super(struct super_block
*sb
);
2090 int set_anon_super(struct super_block
*s
, void *data
);
2091 int get_anon_bdev(dev_t
*);
2092 void free_anon_bdev(dev_t
);
2093 struct super_block
*sget_userns(struct file_system_type
*type
,
2094 int (*test
)(struct super_block
*,void *),
2095 int (*set
)(struct super_block
*,void *),
2096 int flags
, struct user_namespace
*user_ns
,
2098 struct super_block
*sget(struct file_system_type
*type
,
2099 int (*test
)(struct super_block
*,void *),
2100 int (*set
)(struct super_block
*,void *),
2101 int flags
, void *data
);
2102 extern struct dentry
*mount_pseudo_xattr(struct file_system_type
*, char *,
2103 const struct super_operations
*ops
,
2104 const struct xattr_handler
**xattr
,
2105 const struct dentry_operations
*dops
,
2108 static inline struct dentry
*
2109 mount_pseudo(struct file_system_type
*fs_type
, char *name
,
2110 const struct super_operations
*ops
,
2111 const struct dentry_operations
*dops
, unsigned long magic
)
2113 return mount_pseudo_xattr(fs_type
, name
, ops
, NULL
, dops
, magic
);
2116 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2117 #define fops_get(fops) \
2118 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2119 #define fops_put(fops) \
2120 do { if (fops) module_put((fops)->owner); } while(0)
2122 * This one is to be used *ONLY* from ->open() instances.
2123 * fops must be non-NULL, pinned down *and* module dependencies
2124 * should be sufficient to pin the caller down as well.
2126 #define replace_fops(f, fops) \
2128 struct file *__file = (f); \
2129 fops_put(__file->f_op); \
2130 BUG_ON(!(__file->f_op = (fops))); \
2133 extern int register_filesystem(struct file_system_type
*);
2134 extern int unregister_filesystem(struct file_system_type
*);
2135 extern struct vfsmount
*kern_mount_data(struct file_system_type
*, void *data
);
2136 #define kern_mount(type) kern_mount_data(type, NULL)
2137 extern void kern_unmount(struct vfsmount
*mnt
);
2138 extern int may_umount_tree(struct vfsmount
*);
2139 extern int may_umount(struct vfsmount
*);
2140 extern long do_mount(const char *, const char __user
*,
2141 const char *, unsigned long, void *);
2142 extern struct vfsmount
*collect_mounts(const struct path
*);
2143 extern void drop_collected_mounts(struct vfsmount
*);
2144 extern int iterate_mounts(int (*)(struct vfsmount
*, void *), void *,
2146 extern int vfs_statfs(const struct path
*, struct kstatfs
*);
2147 extern int user_statfs(const char __user
*, struct kstatfs
*);
2148 extern int fd_statfs(int, struct kstatfs
*);
2149 extern int vfs_ustat(dev_t
, struct kstatfs
*);
2150 extern int freeze_super(struct super_block
*super
);
2151 extern int thaw_super(struct super_block
*super
);
2152 extern bool our_mnt(struct vfsmount
*mnt
);
2153 extern __printf(2, 3)
2154 int super_setup_bdi_name(struct super_block
*sb
, char *fmt
, ...);
2155 extern int super_setup_bdi(struct super_block
*sb
);
2157 extern int current_umask(void);
2159 extern void ihold(struct inode
* inode
);
2160 extern void iput(struct inode
*);
2161 extern int generic_update_time(struct inode
*, struct timespec
*, int);
2164 extern struct kobject
*fs_kobj
;
2166 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2168 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2169 extern int locks_mandatory_locked(struct file
*);
2170 extern int locks_mandatory_area(struct inode
*, struct file
*, loff_t
, loff_t
, unsigned char);
2173 * Candidates for mandatory locking have the setgid bit set
2174 * but no group execute bit - an otherwise meaningless combination.
2177 static inline int __mandatory_lock(struct inode
*ino
)
2179 return (ino
->i_mode
& (S_ISGID
| S_IXGRP
)) == S_ISGID
;
2183 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2184 * otherwise these will be advisory locks
2187 static inline int mandatory_lock(struct inode
*ino
)
2189 return IS_MANDLOCK(ino
) && __mandatory_lock(ino
);
2192 static inline int locks_verify_locked(struct file
*file
)
2194 if (mandatory_lock(locks_inode(file
)))
2195 return locks_mandatory_locked(file
);
2199 static inline int locks_verify_truncate(struct inode
*inode
,
2203 if (!inode
->i_flctx
|| !mandatory_lock(inode
))
2206 if (size
< inode
->i_size
) {
2207 return locks_mandatory_area(inode
, f
, size
, inode
->i_size
- 1,
2210 return locks_mandatory_area(inode
, f
, inode
->i_size
, size
- 1,
2215 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2217 static inline int locks_mandatory_locked(struct file
*file
)
2222 static inline int locks_mandatory_area(struct inode
*inode
, struct file
*filp
,
2223 loff_t start
, loff_t end
, unsigned char type
)
2228 static inline int __mandatory_lock(struct inode
*inode
)
2233 static inline int mandatory_lock(struct inode
*inode
)
2238 static inline int locks_verify_locked(struct file
*file
)
2243 static inline int locks_verify_truncate(struct inode
*inode
, struct file
*filp
,
2249 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2252 #ifdef CONFIG_FILE_LOCKING
2253 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2256 * Since this check is lockless, we must ensure that any refcounts
2257 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2258 * could end up racing with tasks trying to set a new lease on this
2262 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2263 return __break_lease(inode
, mode
, FL_LEASE
);
2267 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2270 * Since this check is lockless, we must ensure that any refcounts
2271 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2272 * could end up racing with tasks trying to set a new lease on this
2276 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2277 return __break_lease(inode
, mode
, FL_DELEG
);
2281 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2285 ret
= break_deleg(inode
, O_WRONLY
|O_NONBLOCK
);
2286 if (ret
== -EWOULDBLOCK
&& delegated_inode
) {
2287 *delegated_inode
= inode
;
2293 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2297 ret
= break_deleg(*delegated_inode
, O_WRONLY
);
2298 iput(*delegated_inode
);
2299 *delegated_inode
= NULL
;
2303 static inline int break_layout(struct inode
*inode
, bool wait
)
2306 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2307 return __break_lease(inode
,
2308 wait
? O_WRONLY
: O_WRONLY
| O_NONBLOCK
,
2313 #else /* !CONFIG_FILE_LOCKING */
2314 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2319 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2324 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2329 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2335 static inline int break_layout(struct inode
*inode
, bool wait
)
2340 #endif /* CONFIG_FILE_LOCKING */
2345 const char *name
; /* pointer to actual string */
2346 const __user
char *uptr
; /* original userland pointer */
2347 struct audit_names
*aname
;
2352 extern long vfs_truncate(const struct path
*, loff_t
);
2353 extern int do_truncate(struct dentry
*, loff_t start
, unsigned int time_attrs
,
2355 extern int vfs_fallocate(struct file
*file
, int mode
, loff_t offset
,
2357 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
2359 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
2360 extern struct file
*filp_open(const char *, int, umode_t
);
2361 extern struct file
*file_open_root(struct dentry
*, struct vfsmount
*,
2362 const char *, int, umode_t
);
2363 extern struct file
* dentry_open(const struct path
*, int, const struct cred
*);
2364 extern int filp_close(struct file
*, fl_owner_t id
);
2366 extern struct filename
*getname_flags(const char __user
*, int, int *);
2367 extern struct filename
*getname(const char __user
*);
2368 extern struct filename
*getname_kernel(const char *);
2369 extern void putname(struct filename
*name
);
2375 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
2376 int (*open
)(struct inode
*, struct file
*),
2378 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
2382 extern int ioctl_preallocate(struct file
*filp
, void __user
*argp
);
2385 extern void __init
vfs_caches_init_early(void);
2386 extern void __init
vfs_caches_init(void);
2388 extern struct kmem_cache
*names_cachep
;
2390 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2391 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2394 extern int register_blkdev(unsigned int, const char *);
2395 extern void unregister_blkdev(unsigned int, const char *);
2396 extern void bdev_unhash_inode(dev_t dev
);
2397 extern struct block_device
*bdget(dev_t
);
2398 extern struct block_device
*bdgrab(struct block_device
*bdev
);
2399 extern void bd_set_size(struct block_device
*, loff_t size
);
2400 extern void bd_forget(struct inode
*inode
);
2401 extern void bdput(struct block_device
*);
2402 extern void invalidate_bdev(struct block_device
*);
2403 extern void iterate_bdevs(void (*)(struct block_device
*, void *), void *);
2404 extern int sync_blockdev(struct block_device
*bdev
);
2405 extern void kill_bdev(struct block_device
*);
2406 extern struct super_block
*freeze_bdev(struct block_device
*);
2407 extern void emergency_thaw_all(void);
2408 extern int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
);
2409 extern int fsync_bdev(struct block_device
*);
2411 extern struct super_block
*blockdev_superblock
;
2413 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2415 return sb
== blockdev_superblock
;
2418 static inline void bd_forget(struct inode
*inode
) {}
2419 static inline int sync_blockdev(struct block_device
*bdev
) { return 0; }
2420 static inline void kill_bdev(struct block_device
*bdev
) {}
2421 static inline void invalidate_bdev(struct block_device
*bdev
) {}
2423 static inline struct super_block
*freeze_bdev(struct block_device
*sb
)
2428 static inline int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
)
2433 static inline void iterate_bdevs(void (*f
)(struct block_device
*, void *), void *arg
)
2437 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2442 extern int sync_filesystem(struct super_block
*);
2443 extern const struct file_operations def_blk_fops
;
2444 extern const struct file_operations def_chr_fops
;
2446 extern int ioctl_by_bdev(struct block_device
*, unsigned, unsigned long);
2447 extern int blkdev_ioctl(struct block_device
*, fmode_t
, unsigned, unsigned long);
2448 extern long compat_blkdev_ioctl(struct file
*, unsigned, unsigned long);
2449 extern int blkdev_get(struct block_device
*bdev
, fmode_t mode
, void *holder
);
2450 extern struct block_device
*blkdev_get_by_path(const char *path
, fmode_t mode
,
2452 extern struct block_device
*blkdev_get_by_dev(dev_t dev
, fmode_t mode
,
2454 extern void blkdev_put(struct block_device
*bdev
, fmode_t mode
);
2455 extern int __blkdev_reread_part(struct block_device
*bdev
);
2456 extern int blkdev_reread_part(struct block_device
*bdev
);
2459 extern int bd_link_disk_holder(struct block_device
*bdev
, struct gendisk
*disk
);
2460 extern void bd_unlink_disk_holder(struct block_device
*bdev
,
2461 struct gendisk
*disk
);
2463 static inline int bd_link_disk_holder(struct block_device
*bdev
,
2464 struct gendisk
*disk
)
2468 static inline void bd_unlink_disk_holder(struct block_device
*bdev
,
2469 struct gendisk
*disk
)
2476 #define CHRDEV_MAJOR_MAX 512
2477 /* Marks the bottom of the first segment of free char majors */
2478 #define CHRDEV_MAJOR_DYN_END 234
2479 /* Marks the top and bottom of the second segment of free char majors */
2480 #define CHRDEV_MAJOR_DYN_EXT_START 511
2481 #define CHRDEV_MAJOR_DYN_EXT_END 384
2483 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
2484 extern int register_chrdev_region(dev_t
, unsigned, const char *);
2485 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
2486 unsigned int count
, const char *name
,
2487 const struct file_operations
*fops
);
2488 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
2489 unsigned int count
, const char *name
);
2490 extern void unregister_chrdev_region(dev_t
, unsigned);
2491 extern void chrdev_show(struct seq_file
*,off_t
);
2493 static inline int register_chrdev(unsigned int major
, const char *name
,
2494 const struct file_operations
*fops
)
2496 return __register_chrdev(major
, 0, 256, name
, fops
);
2499 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2501 __unregister_chrdev(major
, 0, 256, name
);
2504 /* fs/block_dev.c */
2505 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2506 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2509 #define BLKDEV_MAJOR_MAX 512
2510 extern const char *__bdevname(dev_t
, char *buffer
);
2511 extern const char *bdevname(struct block_device
*bdev
, char *buffer
);
2512 extern struct block_device
*lookup_bdev(const char *);
2513 extern void blkdev_show(struct seq_file
*,off_t
);
2516 #define BLKDEV_MAJOR_MAX 0
2519 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2521 /* Invalid inode operations -- fs/bad_inode.c */
2522 extern void make_bad_inode(struct inode
*);
2523 extern bool is_bad_inode(struct inode
*);
2526 extern void check_disk_size_change(struct gendisk
*disk
,
2527 struct block_device
*bdev
);
2528 extern int revalidate_disk(struct gendisk
*);
2529 extern int check_disk_change(struct block_device
*);
2530 extern int __invalidate_device(struct block_device
*, bool);
2531 extern int invalidate_partition(struct gendisk
*, int);
2533 unsigned long invalidate_mapping_pages(struct address_space
*mapping
,
2534 pgoff_t start
, pgoff_t end
);
2536 static inline void invalidate_remote_inode(struct inode
*inode
)
2538 if (S_ISREG(inode
->i_mode
) || S_ISDIR(inode
->i_mode
) ||
2539 S_ISLNK(inode
->i_mode
))
2540 invalidate_mapping_pages(inode
->i_mapping
, 0, -1);
2542 extern int invalidate_inode_pages2(struct address_space
*mapping
);
2543 extern int invalidate_inode_pages2_range(struct address_space
*mapping
,
2544 pgoff_t start
, pgoff_t end
);
2545 extern int write_inode_now(struct inode
*, int);
2546 extern int filemap_fdatawrite(struct address_space
*);
2547 extern int filemap_flush(struct address_space
*);
2548 extern int filemap_fdatawait(struct address_space
*);
2549 extern int filemap_fdatawait_keep_errors(struct address_space
*mapping
);
2550 extern int filemap_fdatawait_range(struct address_space
*, loff_t lstart
,
2552 extern bool filemap_range_has_page(struct address_space
*, loff_t lstart
,
2554 extern int filemap_write_and_wait(struct address_space
*mapping
);
2555 extern int filemap_write_and_wait_range(struct address_space
*mapping
,
2556 loff_t lstart
, loff_t lend
);
2557 extern int __filemap_fdatawrite_range(struct address_space
*mapping
,
2558 loff_t start
, loff_t end
, int sync_mode
);
2559 extern int filemap_fdatawrite_range(struct address_space
*mapping
,
2560 loff_t start
, loff_t end
);
2561 extern int filemap_check_errors(struct address_space
*mapping
);
2563 extern void __filemap_set_wb_err(struct address_space
*mapping
, int err
);
2564 extern int __must_check
file_check_and_advance_wb_err(struct file
*file
);
2565 extern int __must_check
file_write_and_wait_range(struct file
*file
,
2566 loff_t start
, loff_t end
);
2569 * filemap_set_wb_err - set a writeback error on an address_space
2570 * @mapping: mapping in which to set writeback error
2571 * @err: error to be set in mapping
2573 * When writeback fails in some way, we must record that error so that
2574 * userspace can be informed when fsync and the like are called. We endeavor
2575 * to report errors on any file that was open at the time of the error. Some
2576 * internal callers also need to know when writeback errors have occurred.
2578 * When a writeback error occurs, most filesystems will want to call
2579 * filemap_set_wb_err to record the error in the mapping so that it will be
2580 * automatically reported whenever fsync is called on the file.
2582 * FIXME: mention FS_* flag here?
2584 static inline void filemap_set_wb_err(struct address_space
*mapping
, int err
)
2586 /* Fastpath for common case of no error */
2588 __filemap_set_wb_err(mapping
, err
);
2592 * filemap_check_wb_error - has an error occurred since the mark was sampled?
2593 * @mapping: mapping to check for writeback errors
2594 * @since: previously-sampled errseq_t
2596 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2597 * the given value was sampled.
2599 * If it has then report the latest error set, otherwise return 0.
2601 static inline int filemap_check_wb_err(struct address_space
*mapping
,
2604 return errseq_check(&mapping
->wb_err
, since
);
2608 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2609 * @mapping: mapping to be sampled
2611 * Writeback errors are always reported relative to a particular sample point
2612 * in the past. This function provides those sample points.
2614 static inline errseq_t
filemap_sample_wb_err(struct address_space
*mapping
)
2616 return errseq_sample(&mapping
->wb_err
);
2619 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
2621 extern int vfs_fsync(struct file
*file
, int datasync
);
2624 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2625 * to already be updated for the write, and will return either the amount
2626 * of bytes passed in, or an error if syncing the file failed.
2628 static inline ssize_t
generic_write_sync(struct kiocb
*iocb
, ssize_t count
)
2630 if (iocb
->ki_flags
& IOCB_DSYNC
) {
2631 int ret
= vfs_fsync_range(iocb
->ki_filp
,
2632 iocb
->ki_pos
- count
, iocb
->ki_pos
- 1,
2633 (iocb
->ki_flags
& IOCB_SYNC
) ? 0 : 1);
2641 extern void emergency_sync(void);
2642 extern void emergency_remount(void);
2644 extern sector_t
bmap(struct inode
*, sector_t
);
2646 extern int notify_change(struct dentry
*, struct iattr
*, struct inode
**);
2647 extern int inode_permission(struct inode
*, int);
2648 extern int __inode_permission(struct inode
*, int);
2649 extern int generic_permission(struct inode
*, int);
2650 extern int __check_sticky(struct inode
*dir
, struct inode
*inode
);
2652 static inline bool execute_ok(struct inode
*inode
)
2654 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
2657 static inline void file_start_write(struct file
*file
)
2659 if (!S_ISREG(file_inode(file
)->i_mode
))
2661 __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, true);
2664 static inline bool file_start_write_trylock(struct file
*file
)
2666 if (!S_ISREG(file_inode(file
)->i_mode
))
2668 return __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, false);
2671 static inline void file_end_write(struct file
*file
)
2673 if (!S_ISREG(file_inode(file
)->i_mode
))
2675 __sb_end_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
);
2678 static inline int do_clone_file_range(struct file
*file_in
, loff_t pos_in
,
2679 struct file
*file_out
, loff_t pos_out
,
2684 file_start_write(file_out
);
2685 ret
= vfs_clone_file_range(file_in
, pos_in
, file_out
, pos_out
, len
);
2686 file_end_write(file_out
);
2692 * get_write_access() gets write permission for a file.
2693 * put_write_access() releases this write permission.
2694 * This is used for regular files.
2695 * We cannot support write (and maybe mmap read-write shared) accesses and
2696 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2697 * can have the following values:
2698 * 0: no writers, no VM_DENYWRITE mappings
2699 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2700 * > 0: (i_writecount) users are writing to the file.
2702 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2703 * except for the cases where we don't hold i_writecount yet. Then we need to
2704 * use {get,deny}_write_access() - these functions check the sign and refuse
2705 * to do the change if sign is wrong.
2707 static inline int get_write_access(struct inode
*inode
)
2709 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2711 static inline int deny_write_access(struct file
*file
)
2713 struct inode
*inode
= file_inode(file
);
2714 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2716 static inline void put_write_access(struct inode
* inode
)
2718 atomic_dec(&inode
->i_writecount
);
2720 static inline void allow_write_access(struct file
*file
)
2723 atomic_inc(&file_inode(file
)->i_writecount
);
2725 static inline bool inode_is_open_for_write(const struct inode
*inode
)
2727 return atomic_read(&inode
->i_writecount
) > 0;
2731 static inline void i_readcount_dec(struct inode
*inode
)
2733 BUG_ON(!atomic_read(&inode
->i_readcount
));
2734 atomic_dec(&inode
->i_readcount
);
2736 static inline void i_readcount_inc(struct inode
*inode
)
2738 atomic_inc(&inode
->i_readcount
);
2741 static inline void i_readcount_dec(struct inode
*inode
)
2745 static inline void i_readcount_inc(struct inode
*inode
)
2750 extern int do_pipe_flags(int *, int);
2752 #define __kernel_read_file_id(id) \
2753 id(UNKNOWN, unknown) \
2754 id(FIRMWARE, firmware) \
2755 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2756 id(MODULE, kernel-module) \
2757 id(KEXEC_IMAGE, kexec-image) \
2758 id(KEXEC_INITRAMFS, kexec-initramfs) \
2759 id(POLICY, security-policy) \
2762 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2763 #define __fid_stringify(dummy, str) #str,
2765 enum kernel_read_file_id
{
2766 __kernel_read_file_id(__fid_enumify
)
2769 static const char * const kernel_read_file_str
[] = {
2770 __kernel_read_file_id(__fid_stringify
)
2773 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id
)
2775 if ((unsigned)id
>= READING_MAX_ID
)
2776 return kernel_read_file_str
[READING_UNKNOWN
];
2778 return kernel_read_file_str
[id
];
2781 extern int kernel_read(struct file
*, loff_t
, char *, unsigned long);
2782 extern int kernel_read_file(struct file
*, void **, loff_t
*, loff_t
,
2783 enum kernel_read_file_id
);
2784 extern int kernel_read_file_from_path(char *, void **, loff_t
*, loff_t
,
2785 enum kernel_read_file_id
);
2786 extern int kernel_read_file_from_fd(int, void **, loff_t
*, loff_t
,
2787 enum kernel_read_file_id
);
2788 extern ssize_t
kernel_write(struct file
*, const char *, size_t, loff_t
);
2789 extern ssize_t
__kernel_write(struct file
*, const char *, size_t, loff_t
*);
2790 extern struct file
* open_exec(const char *);
2792 /* fs/dcache.c -- generic fs support functions */
2793 extern bool is_subdir(struct dentry
*, struct dentry
*);
2794 extern bool path_is_under(const struct path
*, const struct path
*);
2796 extern char *file_path(struct file
*, char *, int);
2798 #include <linux/err.h>
2800 /* needed for stackable file system support */
2801 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int whence
);
2803 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int whence
);
2805 extern int inode_init_always(struct super_block
*, struct inode
*);
2806 extern void inode_init_once(struct inode
*);
2807 extern void address_space_init_once(struct address_space
*mapping
);
2808 extern struct inode
* igrab(struct inode
*);
2809 extern ino_t
iunique(struct super_block
*, ino_t
);
2810 extern int inode_needs_sync(struct inode
*inode
);
2811 extern int generic_delete_inode(struct inode
*inode
);
2812 static inline int generic_drop_inode(struct inode
*inode
)
2814 return !inode
->i_nlink
|| inode_unhashed(inode
);
2817 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
2818 unsigned long hashval
, int (*test
)(struct inode
*, void *),
2820 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
2821 int (*test
)(struct inode
*, void *), void *data
);
2822 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
2824 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
2825 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
2826 extern struct inode
*find_inode_nowait(struct super_block
*,
2828 int (*match
)(struct inode
*,
2829 unsigned long, void *),
2831 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
2832 extern int insert_inode_locked(struct inode
*);
2833 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2834 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
2836 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
2838 extern void unlock_new_inode(struct inode
*);
2839 extern unsigned int get_next_ino(void);
2841 extern void __iget(struct inode
* inode
);
2842 extern void iget_failed(struct inode
*);
2843 extern void clear_inode(struct inode
*);
2844 extern void __destroy_inode(struct inode
*);
2845 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
2846 extern struct inode
*new_inode(struct super_block
*sb
);
2847 extern void free_inode_nonrcu(struct inode
*inode
);
2848 extern int should_remove_suid(struct dentry
*);
2849 extern int file_remove_privs(struct file
*);
2851 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
2852 static inline void insert_inode_hash(struct inode
*inode
)
2854 __insert_inode_hash(inode
, inode
->i_ino
);
2857 extern void __remove_inode_hash(struct inode
*);
2858 static inline void remove_inode_hash(struct inode
*inode
)
2860 if (!inode_unhashed(inode
) && !hlist_fake(&inode
->i_hash
))
2861 __remove_inode_hash(inode
);
2864 extern void inode_sb_list_add(struct inode
*inode
);
2867 extern int bdev_read_only(struct block_device
*);
2869 extern int set_blocksize(struct block_device
*, int);
2870 extern int sb_set_blocksize(struct super_block
*, int);
2871 extern int sb_min_blocksize(struct super_block
*, int);
2873 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
2874 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
2875 extern ssize_t
generic_write_checks(struct kiocb
*, struct iov_iter
*);
2876 extern ssize_t
generic_file_read_iter(struct kiocb
*, struct iov_iter
*);
2877 extern ssize_t
__generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2878 extern ssize_t
generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2879 extern ssize_t
generic_file_direct_write(struct kiocb
*, struct iov_iter
*);
2880 extern ssize_t
generic_perform_write(struct file
*, struct iov_iter
*, loff_t
);
2882 ssize_t
vfs_iter_read(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
2884 ssize_t
vfs_iter_write(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
2887 /* fs/block_dev.c */
2888 extern ssize_t
blkdev_read_iter(struct kiocb
*iocb
, struct iov_iter
*to
);
2889 extern ssize_t
blkdev_write_iter(struct kiocb
*iocb
, struct iov_iter
*from
);
2890 extern int blkdev_fsync(struct file
*filp
, loff_t start
, loff_t end
,
2892 extern void block_sync_page(struct page
*page
);
2895 extern ssize_t
generic_file_splice_read(struct file
*, loff_t
*,
2896 struct pipe_inode_info
*, size_t, unsigned int);
2897 extern ssize_t
iter_file_splice_write(struct pipe_inode_info
*,
2898 struct file
*, loff_t
*, size_t, unsigned int);
2899 extern ssize_t
generic_splice_sendpage(struct pipe_inode_info
*pipe
,
2900 struct file
*out
, loff_t
*, size_t len
, unsigned int flags
);
2901 extern long do_splice_direct(struct file
*in
, loff_t
*ppos
, struct file
*out
,
2902 loff_t
*opos
, size_t len
, unsigned int flags
);
2906 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
2907 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int whence
);
2908 extern loff_t
no_llseek(struct file
*file
, loff_t offset
, int whence
);
2909 extern loff_t
vfs_setpos(struct file
*file
, loff_t offset
, loff_t maxsize
);
2910 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int whence
);
2911 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
2912 int whence
, loff_t maxsize
, loff_t eof
);
2913 extern loff_t
fixed_size_llseek(struct file
*file
, loff_t offset
,
2914 int whence
, loff_t size
);
2915 extern loff_t
no_seek_end_llseek_size(struct file
*, loff_t
, int, loff_t
);
2916 extern loff_t
no_seek_end_llseek(struct file
*, loff_t
, int);
2917 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
2918 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
2921 typedef void (dio_submit_t
)(struct bio
*bio
, struct inode
*inode
,
2922 loff_t file_offset
);
2925 /* need locking between buffered and direct access */
2928 /* filesystem does not support filling holes */
2929 DIO_SKIP_HOLES
= 0x02,
2931 /* filesystem can handle aio writes beyond i_size */
2932 DIO_ASYNC_EXTEND
= 0x04,
2934 /* inode/fs/bdev does not need truncate protection */
2935 DIO_SKIP_DIO_COUNT
= 0x08,
2938 void dio_end_io(struct bio
*bio
);
2940 ssize_t
__blockdev_direct_IO(struct kiocb
*iocb
, struct inode
*inode
,
2941 struct block_device
*bdev
, struct iov_iter
*iter
,
2942 get_block_t get_block
,
2943 dio_iodone_t end_io
, dio_submit_t submit_io
,
2946 static inline ssize_t
blockdev_direct_IO(struct kiocb
*iocb
,
2947 struct inode
*inode
,
2948 struct iov_iter
*iter
,
2949 get_block_t get_block
)
2951 return __blockdev_direct_IO(iocb
, inode
, inode
->i_sb
->s_bdev
, iter
,
2952 get_block
, NULL
, NULL
, DIO_LOCKING
| DIO_SKIP_HOLES
);
2956 void inode_dio_wait(struct inode
*inode
);
2959 * inode_dio_begin - signal start of a direct I/O requests
2960 * @inode: inode the direct I/O happens on
2962 * This is called once we've finished processing a direct I/O request,
2963 * and is used to wake up callers waiting for direct I/O to be quiesced.
2965 static inline void inode_dio_begin(struct inode
*inode
)
2967 atomic_inc(&inode
->i_dio_count
);
2971 * inode_dio_end - signal finish of a direct I/O requests
2972 * @inode: inode the direct I/O happens on
2974 * This is called once we've finished processing a direct I/O request,
2975 * and is used to wake up callers waiting for direct I/O to be quiesced.
2977 static inline void inode_dio_end(struct inode
*inode
)
2979 if (atomic_dec_and_test(&inode
->i_dio_count
))
2980 wake_up_bit(&inode
->i_state
, __I_DIO_WAKEUP
);
2983 extern void inode_set_flags(struct inode
*inode
, unsigned int flags
,
2986 extern const struct file_operations generic_ro_fops
;
2988 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2990 extern int readlink_copy(char __user
*, int, const char *);
2991 extern int page_readlink(struct dentry
*, char __user
*, int);
2992 extern const char *page_get_link(struct dentry
*, struct inode
*,
2993 struct delayed_call
*);
2994 extern void page_put_link(void *);
2995 extern int __page_symlink(struct inode
*inode
, const char *symname
, int len
,
2997 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
2998 extern const struct inode_operations page_symlink_inode_operations
;
2999 extern void kfree_link(void *);
3000 extern void generic_fillattr(struct inode
*, struct kstat
*);
3001 extern int vfs_getattr_nosec(const struct path
*, struct kstat
*, u32
, unsigned int);
3002 extern int vfs_getattr(const struct path
*, struct kstat
*, u32
, unsigned int);
3003 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
3004 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
3005 void __inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
3006 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
3007 loff_t
inode_get_bytes(struct inode
*inode
);
3008 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
3009 const char *simple_get_link(struct dentry
*, struct inode
*,
3010 struct delayed_call
*);
3011 extern const struct inode_operations simple_symlink_inode_operations
;
3013 extern int iterate_dir(struct file
*, struct dir_context
*);
3015 extern int vfs_statx(int, const char __user
*, int, struct kstat
*, u32
);
3016 extern int vfs_statx_fd(unsigned int, struct kstat
*, u32
, unsigned int);
3018 static inline int vfs_stat(const char __user
*filename
, struct kstat
*stat
)
3020 return vfs_statx(AT_FDCWD
, filename
, AT_NO_AUTOMOUNT
,
3021 stat
, STATX_BASIC_STATS
);
3023 static inline int vfs_lstat(const char __user
*name
, struct kstat
*stat
)
3025 return vfs_statx(AT_FDCWD
, name
, AT_SYMLINK_NOFOLLOW
| AT_NO_AUTOMOUNT
,
3026 stat
, STATX_BASIC_STATS
);
3028 static inline int vfs_fstatat(int dfd
, const char __user
*filename
,
3029 struct kstat
*stat
, int flags
)
3031 return vfs_statx(dfd
, filename
, flags
| AT_NO_AUTOMOUNT
,
3032 stat
, STATX_BASIC_STATS
);
3034 static inline int vfs_fstat(int fd
, struct kstat
*stat
)
3036 return vfs_statx_fd(fd
, stat
, STATX_BASIC_STATS
, 0);
3040 extern const char *vfs_get_link(struct dentry
*, struct delayed_call
*);
3041 extern int vfs_readlink(struct dentry
*, char __user
*, int);
3043 extern int __generic_block_fiemap(struct inode
*inode
,
3044 struct fiemap_extent_info
*fieinfo
,
3045 loff_t start
, loff_t len
,
3046 get_block_t
*get_block
);
3047 extern int generic_block_fiemap(struct inode
*inode
,
3048 struct fiemap_extent_info
*fieinfo
, u64 start
,
3049 u64 len
, get_block_t
*get_block
);
3051 extern struct file_system_type
*get_filesystem(struct file_system_type
*fs
);
3052 extern void put_filesystem(struct file_system_type
*fs
);
3053 extern struct file_system_type
*get_fs_type(const char *name
);
3054 extern struct super_block
*get_super(struct block_device
*);
3055 extern struct super_block
*get_super_thawed(struct block_device
*);
3056 extern struct super_block
*get_super_exclusive_thawed(struct block_device
*bdev
);
3057 extern struct super_block
*get_active_super(struct block_device
*bdev
);
3058 extern void drop_super(struct super_block
*sb
);
3059 extern void drop_super_exclusive(struct super_block
*sb
);
3060 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
3061 extern void iterate_supers_type(struct file_system_type
*,
3062 void (*)(struct super_block
*, void *), void *);
3064 extern int dcache_dir_open(struct inode
*, struct file
*);
3065 extern int dcache_dir_close(struct inode
*, struct file
*);
3066 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
3067 extern int dcache_readdir(struct file
*, struct dir_context
*);
3068 extern int simple_setattr(struct dentry
*, struct iattr
*);
3069 extern int simple_getattr(const struct path
*, struct kstat
*, u32
, unsigned int);
3070 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
3071 extern int simple_open(struct inode
*inode
, struct file
*file
);
3072 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
3073 extern int simple_unlink(struct inode
*, struct dentry
*);
3074 extern int simple_rmdir(struct inode
*, struct dentry
*);
3075 extern int simple_rename(struct inode
*, struct dentry
*,
3076 struct inode
*, struct dentry
*, unsigned int);
3077 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
3078 extern int simple_empty(struct dentry
*);
3079 extern int simple_readpage(struct file
*file
, struct page
*page
);
3080 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
3081 loff_t pos
, unsigned len
, unsigned flags
,
3082 struct page
**pagep
, void **fsdata
);
3083 extern int simple_write_end(struct file
*file
, struct address_space
*mapping
,
3084 loff_t pos
, unsigned len
, unsigned copied
,
3085 struct page
*page
, void *fsdata
);
3086 extern int always_delete_dentry(const struct dentry
*);
3087 extern struct inode
*alloc_anon_inode(struct super_block
*);
3088 extern int simple_nosetlease(struct file
*, long, struct file_lock
**, void **);
3089 extern const struct dentry_operations simple_dentry_operations
;
3091 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
3092 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
3093 extern const struct file_operations simple_dir_operations
;
3094 extern const struct inode_operations simple_dir_inode_operations
;
3095 extern void make_empty_dir_inode(struct inode
*inode
);
3096 extern bool is_empty_dir_inode(struct inode
*inode
);
3097 struct tree_descr
{ const char *name
; const struct file_operations
*ops
; int mode
; };
3098 struct dentry
*d_alloc_name(struct dentry
*, const char *);
3099 extern int simple_fill_super(struct super_block
*, unsigned long,
3100 const struct tree_descr
*);
3101 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
3102 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
3104 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
3105 loff_t
*ppos
, const void *from
, size_t available
);
3106 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
3107 const void __user
*from
, size_t count
);
3109 extern int __generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3110 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3112 extern int generic_check_addressable(unsigned, u64
);
3114 #ifdef CONFIG_MIGRATION
3115 extern int buffer_migrate_page(struct address_space
*,
3116 struct page
*, struct page
*,
3119 #define buffer_migrate_page NULL
3122 extern int setattr_prepare(struct dentry
*, struct iattr
*);
3123 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
3124 extern void setattr_copy(struct inode
*inode
, const struct iattr
*attr
);
3126 extern int file_update_time(struct file
*file
);
3128 static inline bool io_is_direct(struct file
*filp
)
3130 return (filp
->f_flags
& O_DIRECT
) || IS_DAX(filp
->f_mapping
->host
);
3133 static inline bool vma_is_dax(struct vm_area_struct
*vma
)
3135 return vma
->vm_file
&& IS_DAX(vma
->vm_file
->f_mapping
->host
);
3138 static inline int iocb_flags(struct file
*file
)
3141 if (file
->f_flags
& O_APPEND
)
3143 if (io_is_direct(file
))
3145 if ((file
->f_flags
& O_DSYNC
) || IS_SYNC(file
->f_mapping
->host
))
3147 if (file
->f_flags
& __O_SYNC
)
3152 static inline int kiocb_set_rw_flags(struct kiocb
*ki
, rwf_t flags
)
3154 if (unlikely(flags
& ~RWF_SUPPORTED
))
3157 if (flags
& RWF_NOWAIT
) {
3158 if (!(ki
->ki_filp
->f_mode
& FMODE_AIO_NOWAIT
))
3160 ki
->ki_flags
|= IOCB_NOWAIT
;
3162 if (flags
& RWF_HIPRI
)
3163 ki
->ki_flags
|= IOCB_HIPRI
;
3164 if (flags
& RWF_DSYNC
)
3165 ki
->ki_flags
|= IOCB_DSYNC
;
3166 if (flags
& RWF_SYNC
)
3167 ki
->ki_flags
|= (IOCB_DSYNC
| IOCB_SYNC
);
3171 static inline ino_t
parent_ino(struct dentry
*dentry
)
3176 * Don't strictly need d_lock here? If the parent ino could change
3177 * then surely we'd have a deeper race in the caller?
3179 spin_lock(&dentry
->d_lock
);
3180 res
= dentry
->d_parent
->d_inode
->i_ino
;
3181 spin_unlock(&dentry
->d_lock
);
3185 /* Transaction based IO helpers */
3188 * An argresp is stored in an allocated page and holds the
3189 * size of the argument or response, along with its content
3191 struct simple_transaction_argresp
{
3196 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3198 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
3200 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
3201 size_t size
, loff_t
*pos
);
3202 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
3204 void simple_transaction_set(struct file
*file
, size_t n
);
3207 * simple attribute files
3209 * These attributes behave similar to those in sysfs:
3211 * Writing to an attribute immediately sets a value, an open file can be
3212 * written to multiple times.
3214 * Reading from an attribute creates a buffer from the value that might get
3215 * read with multiple read calls. When the attribute has been read
3216 * completely, no further read calls are possible until the file is opened
3219 * All attributes contain a text representation of a numeric value
3220 * that are accessed with the get() and set() functions.
3222 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3223 static int __fops ## _open(struct inode *inode, struct file *file) \
3225 __simple_attr_check_format(__fmt, 0ull); \
3226 return simple_attr_open(inode, file, __get, __set, __fmt); \
3228 static const struct file_operations __fops = { \
3229 .owner = THIS_MODULE, \
3230 .open = __fops ## _open, \
3231 .release = simple_attr_release, \
3232 .read = simple_attr_read, \
3233 .write = simple_attr_write, \
3234 .llseek = generic_file_llseek, \
3237 static inline __printf(1, 2)
3238 void __simple_attr_check_format(const char *fmt
, ...)
3240 /* don't do anything, just let the compiler check the arguments; */
3243 int simple_attr_open(struct inode
*inode
, struct file
*file
,
3244 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
3246 int simple_attr_release(struct inode
*inode
, struct file
*file
);
3247 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
3248 size_t len
, loff_t
*ppos
);
3249 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
3250 size_t len
, loff_t
*ppos
);
3253 int proc_nr_files(struct ctl_table
*table
, int write
,
3254 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3255 int proc_nr_dentry(struct ctl_table
*table
, int write
,
3256 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3257 int proc_nr_inodes(struct ctl_table
*table
, int write
,
3258 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3259 int __init
get_filesystem_list(char *buf
);
3261 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3262 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3264 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3265 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3266 (flag & __FMODE_NONOTIFY)))
3268 static inline bool is_sxid(umode_t mode
)
3270 return (mode
& S_ISUID
) || ((mode
& S_ISGID
) && (mode
& S_IXGRP
));
3273 static inline int check_sticky(struct inode
*dir
, struct inode
*inode
)
3275 if (!(dir
->i_mode
& S_ISVTX
))
3278 return __check_sticky(dir
, inode
);
3281 static inline void inode_has_no_xattr(struct inode
*inode
)
3283 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& MS_NOSEC
))
3284 inode
->i_flags
|= S_NOSEC
;
3287 static inline bool is_root_inode(struct inode
*inode
)
3289 return inode
== inode
->i_sb
->s_root
->d_inode
;
3292 static inline bool dir_emit(struct dir_context
*ctx
,
3293 const char *name
, int namelen
,
3294 u64 ino
, unsigned type
)
3296 return ctx
->actor(ctx
, name
, namelen
, ctx
->pos
, ino
, type
) == 0;
3298 static inline bool dir_emit_dot(struct file
*file
, struct dir_context
*ctx
)
3300 return ctx
->actor(ctx
, ".", 1, ctx
->pos
,
3301 file
->f_path
.dentry
->d_inode
->i_ino
, DT_DIR
) == 0;
3303 static inline bool dir_emit_dotdot(struct file
*file
, struct dir_context
*ctx
)
3305 return ctx
->actor(ctx
, "..", 2, ctx
->pos
,
3306 parent_ino(file
->f_path
.dentry
), DT_DIR
) == 0;
3308 static inline bool dir_emit_dots(struct file
*file
, struct dir_context
*ctx
)
3310 if (ctx
->pos
== 0) {
3311 if (!dir_emit_dot(file
, ctx
))
3315 if (ctx
->pos
== 1) {
3316 if (!dir_emit_dotdot(file
, ctx
))
3322 static inline bool dir_relax(struct inode
*inode
)
3324 inode_unlock(inode
);
3326 return !IS_DEADDIR(inode
);
3329 static inline bool dir_relax_shared(struct inode
*inode
)
3331 inode_unlock_shared(inode
);
3332 inode_lock_shared(inode
);
3333 return !IS_DEADDIR(inode
);
3336 extern bool path_noexec(const struct path
*path
);
3337 extern void inode_nohighmem(struct inode
*inode
);
3339 #endif /* _LINUX_FS_H */