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
;
76 typedef int (get_block_t
)(struct inode
*inode
, sector_t iblock
,
77 struct buffer_head
*bh_result
, int create
);
78 typedef int (dio_iodone_t
)(struct kiocb
*iocb
, loff_t offset
,
79 ssize_t bytes
, void *private);
81 #define MAY_EXEC 0x00000001
82 #define MAY_WRITE 0x00000002
83 #define MAY_READ 0x00000004
84 #define MAY_APPEND 0x00000008
85 #define MAY_ACCESS 0x00000010
86 #define MAY_OPEN 0x00000020
87 #define MAY_CHDIR 0x00000040
88 /* called from RCU mode, don't block */
89 #define MAY_NOT_BLOCK 0x00000080
92 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
93 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
96 /* file is open for reading */
97 #define FMODE_READ ((__force fmode_t)0x1)
98 /* file is open for writing */
99 #define FMODE_WRITE ((__force fmode_t)0x2)
100 /* file is seekable */
101 #define FMODE_LSEEK ((__force fmode_t)0x4)
102 /* file can be accessed using pread */
103 #define FMODE_PREAD ((__force fmode_t)0x8)
104 /* file can be accessed using pwrite */
105 #define FMODE_PWRITE ((__force fmode_t)0x10)
106 /* File is opened for execution with sys_execve / sys_uselib */
107 #define FMODE_EXEC ((__force fmode_t)0x20)
108 /* File is opened with O_NDELAY (only set for block devices) */
109 #define FMODE_NDELAY ((__force fmode_t)0x40)
110 /* File is opened with O_EXCL (only set for block devices) */
111 #define FMODE_EXCL ((__force fmode_t)0x80)
112 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
113 (specialy hack for floppy.c) */
114 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
115 /* 32bit hashes as llseek() offset (for directories) */
116 #define FMODE_32BITHASH ((__force fmode_t)0x200)
117 /* 64bit hashes as llseek() offset (for directories) */
118 #define FMODE_64BITHASH ((__force fmode_t)0x400)
121 * Don't update ctime and mtime.
123 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
124 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
126 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
128 /* Expect random access pattern */
129 #define FMODE_RANDOM ((__force fmode_t)0x1000)
131 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
132 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
134 /* File is opened with O_PATH; almost nothing can be done with it */
135 #define FMODE_PATH ((__force fmode_t)0x4000)
137 /* File needs atomic accesses to f_pos */
138 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
139 /* Write access to underlying fs */
140 #define FMODE_WRITER ((__force fmode_t)0x10000)
141 /* Has read method(s) */
142 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
143 /* Has write method(s) */
144 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
146 /* File was opened by fanotify and shouldn't generate fanotify events */
147 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
149 /* File is capable of returning -EAGAIN if AIO will block */
150 #define FMODE_AIO_NOWAIT ((__force fmode_t)0x8000000)
153 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
154 * that indicates that they should check the contents of the iovec are
155 * valid, but not check the memory that the iovec elements
158 #define CHECK_IOVEC_ONLY -1
161 * Attribute flags. These should be or-ed together to figure out what
164 #define ATTR_MODE (1 << 0)
165 #define ATTR_UID (1 << 1)
166 #define ATTR_GID (1 << 2)
167 #define ATTR_SIZE (1 << 3)
168 #define ATTR_ATIME (1 << 4)
169 #define ATTR_MTIME (1 << 5)
170 #define ATTR_CTIME (1 << 6)
171 #define ATTR_ATIME_SET (1 << 7)
172 #define ATTR_MTIME_SET (1 << 8)
173 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
174 #define ATTR_ATTR_FLAG (1 << 10)
175 #define ATTR_KILL_SUID (1 << 11)
176 #define ATTR_KILL_SGID (1 << 12)
177 #define ATTR_FILE (1 << 13)
178 #define ATTR_KILL_PRIV (1 << 14)
179 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
180 #define ATTR_TIMES_SET (1 << 16)
181 #define ATTR_TOUCH (1 << 17)
184 * Whiteout is represented by a char device. The following constants define the
185 * mode and device number to use.
187 #define WHITEOUT_MODE 0
188 #define WHITEOUT_DEV 0
191 * This is the Inode Attributes structure, used for notify_change(). It
192 * uses the above definitions as flags, to know which values have changed.
193 * Also, in this manner, a Filesystem can look at only the values it cares
194 * about. Basically, these are the attributes that the VFS layer can
195 * request to change from the FS layer.
197 * Derek Atkins <warlord@MIT.EDU> 94-10-20
200 unsigned int ia_valid
;
205 struct timespec ia_atime
;
206 struct timespec ia_mtime
;
207 struct timespec ia_ctime
;
210 * Not an attribute, but an auxiliary info for filesystems wanting to
211 * implement an ftruncate() like method. NOTE: filesystem should
212 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
214 struct file
*ia_file
;
218 * Includes for diskquotas.
220 #include <linux/quota.h>
223 * Maximum number of layers of fs stack. Needs to be limited to
224 * prevent kernel stack overflow
226 #define FILESYSTEM_MAX_STACK_DEPTH 2
229 * enum positive_aop_returns - aop return codes with specific semantics
231 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
232 * completed, that the page is still locked, and
233 * should be considered active. The VM uses this hint
234 * to return the page to the active list -- it won't
235 * be a candidate for writeback again in the near
236 * future. Other callers must be careful to unlock
237 * the page if they get this return. Returned by
240 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
241 * unlocked it and the page might have been truncated.
242 * The caller should back up to acquiring a new page and
243 * trying again. The aop will be taking reasonable
244 * precautions not to livelock. If the caller held a page
245 * reference, it should drop it before retrying. Returned
248 * address_space_operation functions return these large constants to indicate
249 * special semantics to the caller. These are much larger than the bytes in a
250 * page to allow for functions that return the number of bytes operated on in a
254 enum positive_aop_returns
{
255 AOP_WRITEPAGE_ACTIVATE
= 0x80000,
256 AOP_TRUNCATED_PAGE
= 0x80001,
259 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
260 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
261 * helper code (eg buffer layer)
262 * to clear GFP_FS from alloc */
265 * oh the beauties of C type declarations.
268 struct address_space
;
269 struct writeback_control
;
272 * Write life time hint values.
275 WRITE_LIFE_NOT_SET
= 0,
276 WRITE_LIFE_NONE
= RWH_WRITE_LIFE_NONE
,
277 WRITE_LIFE_SHORT
= RWH_WRITE_LIFE_SHORT
,
278 WRITE_LIFE_MEDIUM
= RWH_WRITE_LIFE_MEDIUM
,
279 WRITE_LIFE_LONG
= RWH_WRITE_LIFE_LONG
,
280 WRITE_LIFE_EXTREME
= RWH_WRITE_LIFE_EXTREME
,
283 #define IOCB_EVENTFD (1 << 0)
284 #define IOCB_APPEND (1 << 1)
285 #define IOCB_DIRECT (1 << 2)
286 #define IOCB_HIPRI (1 << 3)
287 #define IOCB_DSYNC (1 << 4)
288 #define IOCB_SYNC (1 << 5)
289 #define IOCB_WRITE (1 << 6)
290 #define IOCB_NOWAIT (1 << 7)
293 struct file
*ki_filp
;
295 void (*ki_complete
)(struct kiocb
*iocb
, long ret
, long ret2
);
298 enum rw_hint ki_hint
;
301 static inline bool is_sync_kiocb(struct kiocb
*kiocb
)
303 return kiocb
->ki_complete
== NULL
;
307 * "descriptor" for what we're up to with a read.
308 * This allows us to use the same read code yet
309 * have multiple different users of the data that
310 * we read from a file.
312 * The simplest case just copies the data to user
325 typedef int (*read_actor_t
)(read_descriptor_t
*, struct page
*,
326 unsigned long, unsigned long);
328 struct address_space_operations
{
329 int (*writepage
)(struct page
*page
, struct writeback_control
*wbc
);
330 int (*readpage
)(struct file
*, struct page
*);
332 /* Write back some dirty pages from this mapping. */
333 int (*writepages
)(struct address_space
*, struct writeback_control
*);
335 /* Set a page dirty. Return true if this dirtied it */
336 int (*set_page_dirty
)(struct page
*page
);
338 int (*readpages
)(struct file
*filp
, struct address_space
*mapping
,
339 struct list_head
*pages
, unsigned nr_pages
);
341 int (*write_begin
)(struct file
*, struct address_space
*mapping
,
342 loff_t pos
, unsigned len
, unsigned flags
,
343 struct page
**pagep
, void **fsdata
);
344 int (*write_end
)(struct file
*, struct address_space
*mapping
,
345 loff_t pos
, unsigned len
, unsigned copied
,
346 struct page
*page
, void *fsdata
);
348 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
349 sector_t (*bmap
)(struct address_space
*, sector_t
);
350 void (*invalidatepage
) (struct page
*, unsigned int, unsigned int);
351 int (*releasepage
) (struct page
*, gfp_t
);
352 void (*freepage
)(struct page
*);
353 ssize_t (*direct_IO
)(struct kiocb
*, struct iov_iter
*iter
);
355 * migrate the contents of a page to the specified target. If
356 * migrate_mode is MIGRATE_ASYNC, it must not block.
358 int (*migratepage
) (struct address_space
*,
359 struct page
*, struct page
*, enum migrate_mode
);
360 bool (*isolate_page
)(struct page
*, isolate_mode_t
);
361 void (*putback_page
)(struct page
*);
362 int (*launder_page
) (struct page
*);
363 int (*is_partially_uptodate
) (struct page
*, unsigned long,
365 void (*is_dirty_writeback
) (struct page
*, bool *, bool *);
366 int (*error_remove_page
)(struct address_space
*, struct page
*);
368 /* swapfile support */
369 int (*swap_activate
)(struct swap_info_struct
*sis
, struct file
*file
,
371 void (*swap_deactivate
)(struct file
*file
);
374 extern const struct address_space_operations empty_aops
;
377 * pagecache_write_begin/pagecache_write_end must be used by general code
378 * to write into the pagecache.
380 int pagecache_write_begin(struct file
*, struct address_space
*mapping
,
381 loff_t pos
, unsigned len
, unsigned flags
,
382 struct page
**pagep
, void **fsdata
);
384 int pagecache_write_end(struct file
*, struct address_space
*mapping
,
385 loff_t pos
, unsigned len
, unsigned copied
,
386 struct page
*page
, void *fsdata
);
388 struct address_space
{
389 struct inode
*host
; /* owner: inode, block_device */
390 struct radix_tree_root page_tree
; /* radix tree of all pages */
391 spinlock_t tree_lock
; /* and lock protecting it */
392 atomic_t i_mmap_writable
;/* count VM_SHARED mappings */
393 struct rb_root i_mmap
; /* tree of private and shared mappings */
394 struct rw_semaphore i_mmap_rwsem
; /* protect tree, count, list */
395 /* Protected by tree_lock together with the radix tree */
396 unsigned long nrpages
; /* number of total pages */
397 /* number of shadow or DAX exceptional entries */
398 unsigned long nrexceptional
;
399 pgoff_t writeback_index
;/* writeback starts here */
400 const struct address_space_operations
*a_ops
; /* methods */
401 unsigned long flags
; /* error bits */
402 spinlock_t private_lock
; /* for use by the address_space */
403 gfp_t gfp_mask
; /* implicit gfp mask for allocations */
404 struct list_head private_list
; /* ditto */
405 void *private_data
; /* ditto */
407 } __attribute__((aligned(sizeof(long))));
409 * On most architectures that alignment is already the case; but
410 * must be enforced here for CRIS, to let the least significant bit
411 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
413 struct request_queue
;
415 struct block_device
{
416 dev_t bd_dev
; /* not a kdev_t - it's a search key */
418 struct inode
* bd_inode
; /* will die */
419 struct super_block
* bd_super
;
420 struct mutex bd_mutex
; /* open/close mutex */
424 bool bd_write_holder
;
426 struct list_head bd_holder_disks
;
428 struct block_device
* bd_contains
;
429 unsigned bd_block_size
;
430 struct hd_struct
* bd_part
;
431 /* number of times partitions within this device have been opened. */
432 unsigned bd_part_count
;
434 struct gendisk
* bd_disk
;
435 struct request_queue
* bd_queue
;
436 struct backing_dev_info
*bd_bdi
;
437 struct list_head bd_list
;
439 * Private data. You must have bd_claim'ed the block_device
440 * to use this. NOTE: bd_claim allows an owner to claim
441 * the same device multiple times, the owner must take special
442 * care to not mess up bd_private for that case.
444 unsigned long bd_private
;
446 /* The counter of freeze processes */
447 int bd_fsfreeze_count
;
448 /* Mutex for freeze */
449 struct mutex bd_fsfreeze_mutex
;
453 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
456 #define PAGECACHE_TAG_DIRTY 0
457 #define PAGECACHE_TAG_WRITEBACK 1
458 #define PAGECACHE_TAG_TOWRITE 2
460 int mapping_tagged(struct address_space
*mapping
, int tag
);
462 static inline void i_mmap_lock_write(struct address_space
*mapping
)
464 down_write(&mapping
->i_mmap_rwsem
);
467 static inline void i_mmap_unlock_write(struct address_space
*mapping
)
469 up_write(&mapping
->i_mmap_rwsem
);
472 static inline void i_mmap_lock_read(struct address_space
*mapping
)
474 down_read(&mapping
->i_mmap_rwsem
);
477 static inline void i_mmap_unlock_read(struct address_space
*mapping
)
479 up_read(&mapping
->i_mmap_rwsem
);
483 * Might pages of this file be mapped into userspace?
485 static inline int mapping_mapped(struct address_space
*mapping
)
487 return !RB_EMPTY_ROOT(&mapping
->i_mmap
);
491 * Might pages of this file have been modified in userspace?
492 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
493 * marks vma as VM_SHARED if it is shared, and the file was opened for
494 * writing i.e. vma may be mprotected writable even if now readonly.
496 * If i_mmap_writable is negative, no new writable mappings are allowed. You
497 * can only deny writable mappings, if none exists right now.
499 static inline int mapping_writably_mapped(struct address_space
*mapping
)
501 return atomic_read(&mapping
->i_mmap_writable
) > 0;
504 static inline int mapping_map_writable(struct address_space
*mapping
)
506 return atomic_inc_unless_negative(&mapping
->i_mmap_writable
) ?
510 static inline void mapping_unmap_writable(struct address_space
*mapping
)
512 atomic_dec(&mapping
->i_mmap_writable
);
515 static inline int mapping_deny_writable(struct address_space
*mapping
)
517 return atomic_dec_unless_positive(&mapping
->i_mmap_writable
) ?
521 static inline void mapping_allow_writable(struct address_space
*mapping
)
523 atomic_inc(&mapping
->i_mmap_writable
);
527 * Use sequence counter to get consistent i_size on 32-bit processors.
529 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
530 #include <linux/seqlock.h>
531 #define __NEED_I_SIZE_ORDERED
532 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
534 #define i_size_ordered_init(inode) do { } while (0)
538 #define ACL_NOT_CACHED ((void *)(-1))
539 #define ACL_DONT_CACHE ((void *)(-3))
541 static inline struct posix_acl
*
542 uncached_acl_sentinel(struct task_struct
*task
)
544 return (void *)task
+ 1;
548 is_uncached_acl(struct posix_acl
*acl
)
550 return (long)acl
& 1;
553 #define IOP_FASTPERM 0x0001
554 #define IOP_LOOKUP 0x0002
555 #define IOP_NOFOLLOW 0x0004
556 #define IOP_XATTR 0x0008
557 #define IOP_DEFAULT_READLINK 0x0010
559 struct fsnotify_mark_connector
;
562 * Keep mostly read-only and often accessed (especially for
563 * the RCU path lookup and 'stat' data) fields at the beginning
564 * of the 'struct inode'
568 unsigned short i_opflags
;
571 unsigned int i_flags
;
573 #ifdef CONFIG_FS_POSIX_ACL
574 struct posix_acl
*i_acl
;
575 struct posix_acl
*i_default_acl
;
578 const struct inode_operations
*i_op
;
579 struct super_block
*i_sb
;
580 struct address_space
*i_mapping
;
582 #ifdef CONFIG_SECURITY
586 /* Stat data, not accessed from path walking */
589 * Filesystems may only read i_nlink directly. They shall use the
590 * following functions for modification:
592 * (set|clear|inc|drop)_nlink
593 * inode_(inc|dec)_link_count
596 const unsigned int i_nlink
;
597 unsigned int __i_nlink
;
601 struct timespec i_atime
;
602 struct timespec i_mtime
;
603 struct timespec i_ctime
;
604 spinlock_t i_lock
; /* i_blocks, i_bytes, maybe i_size */
605 unsigned short i_bytes
;
606 unsigned int i_blkbits
;
607 enum rw_hint i_write_hint
;
610 #ifdef __NEED_I_SIZE_ORDERED
611 seqcount_t i_size_seqcount
;
615 unsigned long i_state
;
616 struct rw_semaphore i_rwsem
;
618 unsigned long dirtied_when
; /* jiffies of first dirtying */
619 unsigned long dirtied_time_when
;
621 struct hlist_node i_hash
;
622 struct list_head i_io_list
; /* backing dev IO list */
623 #ifdef CONFIG_CGROUP_WRITEBACK
624 struct bdi_writeback
*i_wb
; /* the associated cgroup wb */
626 /* foreign inode detection, see wbc_detach_inode() */
628 u16 i_wb_frn_avg_time
;
629 u16 i_wb_frn_history
;
631 struct list_head i_lru
; /* inode LRU list */
632 struct list_head i_sb_list
;
633 struct list_head i_wb_list
; /* backing dev writeback list */
635 struct hlist_head i_dentry
;
636 struct rcu_head i_rcu
;
640 atomic_t i_dio_count
;
641 atomic_t i_writecount
;
643 atomic_t i_readcount
; /* struct files open RO */
645 const struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
646 struct file_lock_context
*i_flctx
;
647 struct address_space i_data
;
648 struct list_head i_devices
;
650 struct pipe_inode_info
*i_pipe
;
651 struct block_device
*i_bdev
;
659 #ifdef CONFIG_FSNOTIFY
660 __u32 i_fsnotify_mask
; /* all events this inode cares about */
661 struct fsnotify_mark_connector __rcu
*i_fsnotify_marks
;
664 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
665 struct fscrypt_info
*i_crypt_info
;
668 void *i_private
; /* fs or device private pointer */
671 static inline unsigned int i_blocksize(const struct inode
*node
)
673 return (1 << node
->i_blkbits
);
676 static inline int inode_unhashed(struct inode
*inode
)
678 return hlist_unhashed(&inode
->i_hash
);
682 * inode->i_mutex nesting subclasses for the lock validator:
684 * 0: the object of the current VFS operation
688 * 4: second non-directory
689 * 5: second parent (when locking independent directories in rename)
691 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
692 * non-directories at once.
694 * The locking order between these classes is
695 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
697 enum inode_i_mutex_lock_class
707 static inline void inode_lock(struct inode
*inode
)
709 down_write(&inode
->i_rwsem
);
712 static inline void inode_unlock(struct inode
*inode
)
714 up_write(&inode
->i_rwsem
);
717 static inline void inode_lock_shared(struct inode
*inode
)
719 down_read(&inode
->i_rwsem
);
722 static inline void inode_unlock_shared(struct inode
*inode
)
724 up_read(&inode
->i_rwsem
);
727 static inline int inode_trylock(struct inode
*inode
)
729 return down_write_trylock(&inode
->i_rwsem
);
732 static inline int inode_trylock_shared(struct inode
*inode
)
734 return down_read_trylock(&inode
->i_rwsem
);
737 static inline int inode_is_locked(struct inode
*inode
)
739 return rwsem_is_locked(&inode
->i_rwsem
);
742 static inline void inode_lock_nested(struct inode
*inode
, unsigned subclass
)
744 down_write_nested(&inode
->i_rwsem
, subclass
);
747 void lock_two_nondirectories(struct inode
*, struct inode
*);
748 void unlock_two_nondirectories(struct inode
*, struct inode
*);
751 * NOTE: in a 32bit arch with a preemptable kernel and
752 * an UP compile the i_size_read/write must be atomic
753 * with respect to the local cpu (unlike with preempt disabled),
754 * but they don't need to be atomic with respect to other cpus like in
755 * true SMP (so they need either to either locally disable irq around
756 * the read or for example on x86 they can be still implemented as a
757 * cmpxchg8b without the need of the lock prefix). For SMP compiles
758 * and 64bit archs it makes no difference if preempt is enabled or not.
760 static inline loff_t
i_size_read(const struct inode
*inode
)
762 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
767 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
768 i_size
= inode
->i_size
;
769 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
771 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
775 i_size
= inode
->i_size
;
779 return inode
->i_size
;
784 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
785 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
786 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
788 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
790 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
792 write_seqcount_begin(&inode
->i_size_seqcount
);
793 inode
->i_size
= i_size
;
794 write_seqcount_end(&inode
->i_size_seqcount
);
796 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
798 inode
->i_size
= i_size
;
801 inode
->i_size
= i_size
;
805 static inline unsigned iminor(const struct inode
*inode
)
807 return MINOR(inode
->i_rdev
);
810 static inline unsigned imajor(const struct inode
*inode
)
812 return MAJOR(inode
->i_rdev
);
815 extern struct block_device
*I_BDEV(struct inode
*inode
);
818 rwlock_t lock
; /* protects pid, uid, euid fields */
819 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
820 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
821 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
822 int signum
; /* posix.1b rt signal to be delivered on IO */
826 * Track a single file's readahead state
828 struct file_ra_state
{
829 pgoff_t start
; /* where readahead started */
830 unsigned int size
; /* # of readahead pages */
831 unsigned int async_size
; /* do asynchronous readahead when
832 there are only # of pages ahead */
834 unsigned int ra_pages
; /* Maximum readahead window */
835 unsigned int mmap_miss
; /* Cache miss stat for mmap accesses */
836 loff_t prev_pos
; /* Cache last read() position */
840 * Check if @index falls in the readahead windows.
842 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
844 return (index
>= ra
->start
&&
845 index
< ra
->start
+ ra
->size
);
850 struct llist_node fu_llist
;
851 struct rcu_head fu_rcuhead
;
854 struct inode
*f_inode
; /* cached value */
855 const struct file_operations
*f_op
;
858 * Protects f_ep_links, f_flags.
859 * Must not be taken from IRQ context.
862 enum rw_hint f_write_hint
;
863 atomic_long_t f_count
;
864 unsigned int f_flags
;
866 struct mutex f_pos_lock
;
868 struct fown_struct f_owner
;
869 const struct cred
*f_cred
;
870 struct file_ra_state f_ra
;
873 #ifdef CONFIG_SECURITY
876 /* needed for tty driver, and maybe others */
880 /* Used by fs/eventpoll.c to link all the hooks to this file */
881 struct list_head f_ep_links
;
882 struct list_head f_tfile_llink
;
883 #endif /* #ifdef CONFIG_EPOLL */
884 struct address_space
*f_mapping
;
886 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
891 /* file identifier */
892 unsigned char f_handle
[0];
895 static inline struct file
*get_file(struct file
*f
)
897 atomic_long_inc(&f
->f_count
);
900 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
901 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
902 #define file_count(x) atomic_long_read(&(x)->f_count)
904 #define MAX_NON_LFS ((1UL<<31) - 1)
906 /* Page cache limit. The filesystems should put that into their s_maxbytes
907 limits, otherwise bad things can happen in VM. */
908 #if BITS_PER_LONG==32
909 #define MAX_LFS_FILESIZE (((loff_t)PAGE_SIZE << (BITS_PER_LONG-1))-1)
910 #elif BITS_PER_LONG==64
911 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
916 #define FL_DELEG 4 /* NFSv4 delegation */
917 #define FL_ACCESS 8 /* not trying to lock, just looking */
918 #define FL_EXISTS 16 /* when unlocking, test for existence */
919 #define FL_LEASE 32 /* lease held on this file */
920 #define FL_CLOSE 64 /* unlock on close */
921 #define FL_SLEEP 128 /* A blocking lock */
922 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
923 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
924 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
925 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
927 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
930 * Special return value from posix_lock_file() and vfs_lock_file() for
931 * asynchronous locking.
933 #define FILE_LOCK_DEFERRED 1
935 /* legacy typedef, should eventually be removed */
936 typedef void *fl_owner_t
;
940 struct file_lock_operations
{
941 void (*fl_copy_lock
)(struct file_lock
*, struct file_lock
*);
942 void (*fl_release_private
)(struct file_lock
*);
945 struct lock_manager_operations
{
946 int (*lm_compare_owner
)(struct file_lock
*, struct file_lock
*);
947 unsigned long (*lm_owner_key
)(struct file_lock
*);
948 fl_owner_t (*lm_get_owner
)(fl_owner_t
);
949 void (*lm_put_owner
)(fl_owner_t
);
950 void (*lm_notify
)(struct file_lock
*); /* unblock callback */
951 int (*lm_grant
)(struct file_lock
*, int);
952 bool (*lm_break
)(struct file_lock
*);
953 int (*lm_change
)(struct file_lock
*, int, struct list_head
*);
954 void (*lm_setup
)(struct file_lock
*, void **);
957 struct lock_manager
{
958 struct list_head list
;
960 * NFSv4 and up also want opens blocked during the grace period;
967 void locks_start_grace(struct net
*, struct lock_manager
*);
968 void locks_end_grace(struct lock_manager
*);
969 int locks_in_grace(struct net
*);
970 int opens_in_grace(struct net
*);
972 /* that will die - we need it for nfs_lock_info */
973 #include <linux/nfs_fs_i.h>
976 * struct file_lock represents a generic "file lock". It's used to represent
977 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
978 * note that the same struct is used to represent both a request for a lock and
979 * the lock itself, but the same object is never used for both.
981 * FIXME: should we create a separate "struct lock_request" to help distinguish
984 * The varous i_flctx lists are ordered by:
987 * 2) lock range start
990 * Obviously, the last two criteria only matter for POSIX locks.
993 struct file_lock
*fl_next
; /* singly linked list for this inode */
994 struct list_head fl_list
; /* link into file_lock_context */
995 struct hlist_node fl_link
; /* node in global lists */
996 struct list_head fl_block
; /* circular list of blocked processes */
998 unsigned int fl_flags
;
999 unsigned char fl_type
;
1000 unsigned int fl_pid
;
1001 int fl_link_cpu
; /* what cpu's list is this on? */
1002 struct pid
*fl_nspid
;
1003 wait_queue_head_t fl_wait
;
1004 struct file
*fl_file
;
1008 struct fasync_struct
* fl_fasync
; /* for lease break notifications */
1009 /* for lease breaks: */
1010 unsigned long fl_break_time
;
1011 unsigned long fl_downgrade_time
;
1013 const struct file_lock_operations
*fl_ops
; /* Callbacks for filesystems */
1014 const struct lock_manager_operations
*fl_lmops
; /* Callbacks for lockmanagers */
1016 struct nfs_lock_info nfs_fl
;
1017 struct nfs4_lock_info nfs4_fl
;
1019 struct list_head link
; /* link in AFS vnode's pending_locks list */
1020 int state
; /* state of grant or error if -ve */
1025 struct file_lock_context
{
1026 spinlock_t flc_lock
;
1027 struct list_head flc_flock
;
1028 struct list_head flc_posix
;
1029 struct list_head flc_lease
;
1032 /* The following constant reflects the upper bound of the file/locking space */
1034 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1035 #define OFFSET_MAX INT_LIMIT(loff_t)
1036 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1039 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
1042 * Return the inode to use for locking
1044 * For overlayfs this should be the overlay inode, not the real inode returned
1045 * by file_inode(). For any other fs file_inode(filp) and locks_inode(filp) are
1048 static inline struct inode
*locks_inode(const struct file
*f
)
1050 return f
->f_path
.dentry
->d_inode
;
1053 #ifdef CONFIG_FILE_LOCKING
1054 extern int fcntl_getlk(struct file
*, unsigned int, struct flock
*);
1055 extern int fcntl_setlk(unsigned int, struct file
*, unsigned int,
1058 #if BITS_PER_LONG == 32
1059 extern int fcntl_getlk64(struct file
*, unsigned int, struct flock64
*);
1060 extern int fcntl_setlk64(unsigned int, struct file
*, unsigned int,
1064 extern int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
);
1065 extern int fcntl_getlease(struct file
*filp
);
1068 void locks_free_lock_context(struct inode
*inode
);
1069 void locks_free_lock(struct file_lock
*fl
);
1070 extern void locks_init_lock(struct file_lock
*);
1071 extern struct file_lock
* locks_alloc_lock(void);
1072 extern void locks_copy_lock(struct file_lock
*, struct file_lock
*);
1073 extern void locks_copy_conflock(struct file_lock
*, struct file_lock
*);
1074 extern void locks_remove_posix(struct file
*, fl_owner_t
);
1075 extern void locks_remove_file(struct file
*);
1076 extern void locks_release_private(struct file_lock
*);
1077 extern void posix_test_lock(struct file
*, struct file_lock
*);
1078 extern int posix_lock_file(struct file
*, struct file_lock
*, struct file_lock
*);
1079 extern int posix_unblock_lock(struct file_lock
*);
1080 extern int vfs_test_lock(struct file
*, struct file_lock
*);
1081 extern int vfs_lock_file(struct file
*, unsigned int, struct file_lock
*, struct file_lock
*);
1082 extern int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
);
1083 extern int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
);
1084 extern int __break_lease(struct inode
*inode
, unsigned int flags
, unsigned int type
);
1085 extern void lease_get_mtime(struct inode
*, struct timespec
*time
);
1086 extern int generic_setlease(struct file
*, long, struct file_lock
**, void **priv
);
1087 extern int vfs_setlease(struct file
*, long, struct file_lock
**, void **);
1088 extern int lease_modify(struct file_lock
*, int, struct list_head
*);
1089 struct files_struct
;
1090 extern void show_fd_locks(struct seq_file
*f
,
1091 struct file
*filp
, struct files_struct
*files
);
1092 #else /* !CONFIG_FILE_LOCKING */
1093 static inline int fcntl_getlk(struct file
*file
, unsigned int cmd
,
1094 struct flock __user
*user
)
1099 static inline int fcntl_setlk(unsigned int fd
, struct file
*file
,
1100 unsigned int cmd
, struct flock __user
*user
)
1105 #if BITS_PER_LONG == 32
1106 static inline int fcntl_getlk64(struct file
*file
, unsigned int cmd
,
1107 struct flock64 __user
*user
)
1112 static inline int fcntl_setlk64(unsigned int fd
, struct file
*file
,
1113 unsigned int cmd
, struct flock64 __user
*user
)
1118 static inline int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
)
1123 static inline int fcntl_getlease(struct file
*filp
)
1129 locks_free_lock_context(struct inode
*inode
)
1133 static inline void locks_init_lock(struct file_lock
*fl
)
1138 static inline void locks_copy_conflock(struct file_lock
*new, struct file_lock
*fl
)
1143 static inline void locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
1148 static inline void locks_remove_posix(struct file
*filp
, fl_owner_t owner
)
1153 static inline void locks_remove_file(struct file
*filp
)
1158 static inline void posix_test_lock(struct file
*filp
, struct file_lock
*fl
)
1163 static inline int posix_lock_file(struct file
*filp
, struct file_lock
*fl
,
1164 struct file_lock
*conflock
)
1169 static inline int posix_unblock_lock(struct file_lock
*waiter
)
1174 static inline int vfs_test_lock(struct file
*filp
, struct file_lock
*fl
)
1179 static inline int vfs_lock_file(struct file
*filp
, unsigned int cmd
,
1180 struct file_lock
*fl
, struct file_lock
*conf
)
1185 static inline int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
)
1190 static inline int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
)
1195 static inline int __break_lease(struct inode
*inode
, unsigned int mode
, unsigned int type
)
1200 static inline void lease_get_mtime(struct inode
*inode
, struct timespec
*time
)
1205 static inline int generic_setlease(struct file
*filp
, long arg
,
1206 struct file_lock
**flp
, void **priv
)
1211 static inline int vfs_setlease(struct file
*filp
, long arg
,
1212 struct file_lock
**lease
, void **priv
)
1217 static inline int lease_modify(struct file_lock
*fl
, int arg
,
1218 struct list_head
*dispose
)
1223 struct files_struct
;
1224 static inline void show_fd_locks(struct seq_file
*f
,
1225 struct file
*filp
, struct files_struct
*files
) {}
1226 #endif /* !CONFIG_FILE_LOCKING */
1228 static inline struct inode
*file_inode(const struct file
*f
)
1233 static inline struct dentry
*file_dentry(const struct file
*file
)
1235 return d_real(file
->f_path
.dentry
, file_inode(file
), 0);
1238 static inline int locks_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1240 return locks_lock_inode_wait(locks_inode(filp
), fl
);
1243 struct fasync_struct
{
1247 struct fasync_struct
*fa_next
; /* singly linked list */
1248 struct file
*fa_file
;
1249 struct rcu_head fa_rcu
;
1252 #define FASYNC_MAGIC 0x4601
1254 /* SMP safe fasync helpers: */
1255 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1256 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1257 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1258 extern struct fasync_struct
*fasync_alloc(void);
1259 extern void fasync_free(struct fasync_struct
*);
1261 /* can be called from interrupts */
1262 extern void kill_fasync(struct fasync_struct
**, int, int);
1264 extern void __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1265 extern int f_setown(struct file
*filp
, unsigned long arg
, int force
);
1266 extern void f_delown(struct file
*filp
);
1267 extern pid_t
f_getown(struct file
*filp
);
1268 extern int send_sigurg(struct fown_struct
*fown
);
1276 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1277 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1278 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1279 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1280 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1283 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1284 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1285 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1287 /* sb->s_iflags to limit user namespace mounts */
1288 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1290 /* Possible states of 'frozen' field */
1292 SB_UNFROZEN
= 0, /* FS is unfrozen */
1293 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1294 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1295 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1296 * internal threads if needed) */
1297 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1300 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1303 int frozen
; /* Is sb frozen? */
1304 wait_queue_head_t wait_unfrozen
; /* for get_super_thawed() */
1305 struct percpu_rw_semaphore rw_sem
[SB_FREEZE_LEVELS
];
1308 struct super_block
{
1309 struct list_head s_list
; /* Keep this first */
1310 dev_t s_dev
; /* search index; _not_ kdev_t */
1311 unsigned char s_blocksize_bits
;
1312 unsigned long s_blocksize
;
1313 loff_t s_maxbytes
; /* Max file size */
1314 struct file_system_type
*s_type
;
1315 const struct super_operations
*s_op
;
1316 const struct dquot_operations
*dq_op
;
1317 const struct quotactl_ops
*s_qcop
;
1318 const struct export_operations
*s_export_op
;
1319 unsigned long s_flags
;
1320 unsigned long s_iflags
; /* internal SB_I_* flags */
1321 unsigned long s_magic
;
1322 struct dentry
*s_root
;
1323 struct rw_semaphore s_umount
;
1326 #ifdef CONFIG_SECURITY
1329 const struct xattr_handler
**s_xattr
;
1331 const struct fscrypt_operations
*s_cop
;
1333 struct hlist_bl_head s_anon
; /* anonymous dentries for (nfs) exporting */
1334 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1335 struct block_device
*s_bdev
;
1336 struct backing_dev_info
*s_bdi
;
1337 struct mtd_info
*s_mtd
;
1338 struct hlist_node s_instances
;
1339 unsigned int s_quota_types
; /* Bitmask of supported quota types */
1340 struct quota_info s_dquot
; /* Diskquota specific options */
1342 struct sb_writers s_writers
;
1344 char s_id
[32]; /* Informational name */
1345 uuid_t s_uuid
; /* UUID */
1347 void *s_fs_info
; /* Filesystem private info */
1348 unsigned int s_max_links
;
1351 /* Granularity of c/m/atime in ns.
1352 Cannot be worse than a second */
1356 * The next field is for VFS *only*. No filesystems have any business
1357 * even looking at it. You had been warned.
1359 struct mutex s_vfs_rename_mutex
; /* Kludge */
1362 * Filesystem subtype. If non-empty the filesystem type field
1363 * in /proc/mounts will be "type.subtype"
1367 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1370 * Saved pool identifier for cleancache (-1 means none)
1372 int cleancache_poolid
;
1374 struct shrinker s_shrink
; /* per-sb shrinker handle */
1376 /* Number of inodes with nlink == 0 but still referenced */
1377 atomic_long_t s_remove_count
;
1379 /* Being remounted read-only */
1380 int s_readonly_remount
;
1382 /* AIO completions deferred from interrupt context */
1383 struct workqueue_struct
*s_dio_done_wq
;
1384 struct hlist_head s_pins
;
1387 * Owning user namespace and default context in which to
1388 * interpret filesystem uids, gids, quotas, device nodes,
1389 * xattrs and security labels.
1391 struct user_namespace
*s_user_ns
;
1394 * Keep the lru lists last in the structure so they always sit on their
1395 * own individual cachelines.
1397 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp
;
1398 struct list_lru s_inode_lru ____cacheline_aligned_in_smp
;
1399 struct rcu_head rcu
;
1400 struct work_struct destroy_work
;
1402 struct mutex s_sync_lock
; /* sync serialisation lock */
1405 * Indicates how deep in a filesystem stack this SB is
1409 /* s_inode_list_lock protects s_inodes */
1410 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp
;
1411 struct list_head s_inodes
; /* all inodes */
1413 spinlock_t s_inode_wblist_lock
;
1414 struct list_head s_inodes_wb
; /* writeback inodes */
1417 /* Helper functions so that in most cases filesystems will
1418 * not need to deal directly with kuid_t and kgid_t and can
1419 * instead deal with the raw numeric values that are stored
1420 * in the filesystem.
1422 static inline uid_t
i_uid_read(const struct inode
*inode
)
1424 return from_kuid(inode
->i_sb
->s_user_ns
, inode
->i_uid
);
1427 static inline gid_t
i_gid_read(const struct inode
*inode
)
1429 return from_kgid(inode
->i_sb
->s_user_ns
, inode
->i_gid
);
1432 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
1434 inode
->i_uid
= make_kuid(inode
->i_sb
->s_user_ns
, uid
);
1437 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
1439 inode
->i_gid
= make_kgid(inode
->i_sb
->s_user_ns
, gid
);
1442 extern struct timespec
current_time(struct inode
*inode
);
1445 * Snapshotting support.
1448 void __sb_end_write(struct super_block
*sb
, int level
);
1449 int __sb_start_write(struct super_block
*sb
, int level
, bool wait
);
1451 #define __sb_writers_acquired(sb, lev) \
1452 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1453 #define __sb_writers_release(sb, lev) \
1454 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1457 * sb_end_write - drop write access to a superblock
1458 * @sb: the super we wrote to
1460 * Decrement number of writers to the filesystem. Wake up possible waiters
1461 * wanting to freeze the filesystem.
1463 static inline void sb_end_write(struct super_block
*sb
)
1465 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1469 * sb_end_pagefault - drop write access to a superblock from a page fault
1470 * @sb: the super we wrote to
1472 * Decrement number of processes handling write page fault to the filesystem.
1473 * Wake up possible waiters wanting to freeze the filesystem.
1475 static inline void sb_end_pagefault(struct super_block
*sb
)
1477 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1481 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1482 * @sb: the super we wrote to
1484 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1485 * waiters wanting to freeze the filesystem.
1487 static inline void sb_end_intwrite(struct super_block
*sb
)
1489 __sb_end_write(sb
, SB_FREEZE_FS
);
1493 * sb_start_write - get write access to a superblock
1494 * @sb: the super we write to
1496 * When a process wants to write data or metadata to a file system (i.e. dirty
1497 * a page or an inode), it should embed the operation in a sb_start_write() -
1498 * sb_end_write() pair to get exclusion against file system freezing. This
1499 * function increments number of writers preventing freezing. If the file
1500 * system is already frozen, the function waits until the file system is
1503 * Since freeze protection behaves as a lock, users have to preserve
1504 * ordering of freeze protection and other filesystem locks. Generally,
1505 * freeze protection should be the outermost lock. In particular, we have:
1508 * -> i_mutex (write path, truncate, directory ops, ...)
1509 * -> s_umount (freeze_super, thaw_super)
1511 static inline void sb_start_write(struct super_block
*sb
)
1513 __sb_start_write(sb
, SB_FREEZE_WRITE
, true);
1516 static inline int sb_start_write_trylock(struct super_block
*sb
)
1518 return __sb_start_write(sb
, SB_FREEZE_WRITE
, false);
1522 * sb_start_pagefault - get write access to a superblock from a page fault
1523 * @sb: the super we write to
1525 * When a process starts handling write page fault, it should embed the
1526 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1527 * exclusion against file system freezing. This is needed since the page fault
1528 * is going to dirty a page. This function increments number of running page
1529 * faults preventing freezing. If the file system is already frozen, the
1530 * function waits until the file system is thawed.
1532 * Since page fault freeze protection behaves as a lock, users have to preserve
1533 * ordering of freeze protection and other filesystem locks. It is advised to
1534 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1535 * handling code implies lock dependency:
1538 * -> sb_start_pagefault
1540 static inline void sb_start_pagefault(struct super_block
*sb
)
1542 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
, true);
1546 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1547 * @sb: the super we write to
1549 * This is the third level of protection against filesystem freezing. It is
1550 * free for use by a filesystem. The only requirement is that it must rank
1551 * below sb_start_pagefault.
1553 * For example filesystem can call sb_start_intwrite() when starting a
1554 * transaction which somewhat eases handling of freezing for internal sources
1555 * of filesystem changes (internal fs threads, discarding preallocation on file
1558 static inline void sb_start_intwrite(struct super_block
*sb
)
1560 __sb_start_write(sb
, SB_FREEZE_FS
, true);
1564 extern bool inode_owner_or_capable(const struct inode
*inode
);
1567 * VFS helper functions..
1569 extern int vfs_create(struct inode
*, struct dentry
*, umode_t
, bool);
1570 extern int vfs_mkdir(struct inode
*, struct dentry
*, umode_t
);
1571 extern int vfs_mknod(struct inode
*, struct dentry
*, umode_t
, dev_t
);
1572 extern int vfs_symlink(struct inode
*, struct dentry
*, const char *);
1573 extern int vfs_link(struct dentry
*, struct inode
*, struct dentry
*, struct inode
**);
1574 extern int vfs_rmdir(struct inode
*, struct dentry
*);
1575 extern int vfs_unlink(struct inode
*, struct dentry
*, struct inode
**);
1576 extern int vfs_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*, struct inode
**, unsigned int);
1577 extern int vfs_whiteout(struct inode
*, struct dentry
*);
1579 extern struct dentry
*vfs_tmpfile(struct dentry
*dentry
, umode_t mode
,
1583 * VFS file helper functions.
1585 extern void inode_init_owner(struct inode
*inode
, const struct inode
*dir
,
1587 extern bool may_open_dev(const struct path
*path
);
1589 * VFS FS_IOC_FIEMAP helper definitions.
1591 struct fiemap_extent_info
{
1592 unsigned int fi_flags
; /* Flags as passed from user */
1593 unsigned int fi_extents_mapped
; /* Number of mapped extents */
1594 unsigned int fi_extents_max
; /* Size of fiemap_extent array */
1595 struct fiemap_extent __user
*fi_extents_start
; /* Start of
1596 fiemap_extent array */
1598 int fiemap_fill_next_extent(struct fiemap_extent_info
*info
, u64 logical
,
1599 u64 phys
, u64 len
, u32 flags
);
1600 int fiemap_check_flags(struct fiemap_extent_info
*fieinfo
, u32 fs_flags
);
1605 * NOTE! These match bits 12..15 of stat.st_mode
1606 * (ie "(i_mode >> 12) & 15").
1608 #define DT_UNKNOWN 0
1619 * This is the "filldir" function type, used by readdir() to let
1620 * the kernel specify what kind of dirent layout it wants to have.
1621 * This allows the kernel to read directories into kernel space or
1622 * to have different dirent layouts depending on the binary type.
1625 typedef int (*filldir_t
)(struct dir_context
*, const char *, int, loff_t
, u64
,
1628 struct dir_context
{
1629 const filldir_t actor
;
1633 struct block_device_operations
;
1635 /* These macros are for out of kernel modules to test that
1636 * the kernel supports the unlocked_ioctl and compat_ioctl
1637 * fields in struct file_operations. */
1638 #define HAVE_COMPAT_IOCTL 1
1639 #define HAVE_UNLOCKED_IOCTL 1
1642 * These flags let !MMU mmap() govern direct device mapping vs immediate
1643 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1645 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1646 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1647 * NOMMU_MAP_READ: Can be mapped for reading
1648 * NOMMU_MAP_WRITE: Can be mapped for writing
1649 * NOMMU_MAP_EXEC: Can be mapped for execution
1651 #define NOMMU_MAP_COPY 0x00000001
1652 #define NOMMU_MAP_DIRECT 0x00000008
1653 #define NOMMU_MAP_READ VM_MAYREAD
1654 #define NOMMU_MAP_WRITE VM_MAYWRITE
1655 #define NOMMU_MAP_EXEC VM_MAYEXEC
1657 #define NOMMU_VMFLAGS \
1658 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1663 struct file_operations
{
1664 struct module
*owner
;
1665 loff_t (*llseek
) (struct file
*, loff_t
, int);
1666 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
1667 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
1668 ssize_t (*read_iter
) (struct kiocb
*, struct iov_iter
*);
1669 ssize_t (*write_iter
) (struct kiocb
*, struct iov_iter
*);
1670 int (*iterate
) (struct file
*, struct dir_context
*);
1671 int (*iterate_shared
) (struct file
*, struct dir_context
*);
1672 unsigned int (*poll
) (struct file
*, struct poll_table_struct
*);
1673 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
1674 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
1675 int (*mmap
) (struct file
*, struct vm_area_struct
*);
1676 int (*open
) (struct inode
*, struct file
*);
1677 int (*flush
) (struct file
*, fl_owner_t id
);
1678 int (*release
) (struct inode
*, struct file
*);
1679 int (*fsync
) (struct file
*, loff_t
, loff_t
, int datasync
);
1680 int (*fasync
) (int, struct file
*, int);
1681 int (*lock
) (struct file
*, int, struct file_lock
*);
1682 ssize_t (*sendpage
) (struct file
*, struct page
*, int, size_t, loff_t
*, int);
1683 unsigned long (*get_unmapped_area
)(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1684 int (*check_flags
)(int);
1685 int (*flock
) (struct file
*, int, struct file_lock
*);
1686 ssize_t (*splice_write
)(struct pipe_inode_info
*, struct file
*, loff_t
*, size_t, unsigned int);
1687 ssize_t (*splice_read
)(struct file
*, loff_t
*, struct pipe_inode_info
*, size_t, unsigned int);
1688 int (*setlease
)(struct file
*, long, struct file_lock
**, void **);
1689 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
1691 void (*show_fdinfo
)(struct seq_file
*m
, struct file
*f
);
1693 unsigned (*mmap_capabilities
)(struct file
*);
1695 ssize_t (*copy_file_range
)(struct file
*, loff_t
, struct file
*,
1696 loff_t
, size_t, unsigned int);
1697 int (*clone_file_range
)(struct file
*, loff_t
, struct file
*, loff_t
,
1699 ssize_t (*dedupe_file_range
)(struct file
*, u64
, u64
, struct file
*,
1703 struct inode_operations
{
1704 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*, unsigned int);
1705 const char * (*get_link
) (struct dentry
*, struct inode
*, struct delayed_call
*);
1706 int (*permission
) (struct inode
*, int);
1707 struct posix_acl
* (*get_acl
)(struct inode
*, int);
1709 int (*readlink
) (struct dentry
*, char __user
*,int);
1711 int (*create
) (struct inode
*,struct dentry
*, umode_t
, bool);
1712 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
1713 int (*unlink
) (struct inode
*,struct dentry
*);
1714 int (*symlink
) (struct inode
*,struct dentry
*,const char *);
1715 int (*mkdir
) (struct inode
*,struct dentry
*,umode_t
);
1716 int (*rmdir
) (struct inode
*,struct dentry
*);
1717 int (*mknod
) (struct inode
*,struct dentry
*,umode_t
,dev_t
);
1718 int (*rename
) (struct inode
*, struct dentry
*,
1719 struct inode
*, struct dentry
*, unsigned int);
1720 int (*setattr
) (struct dentry
*, struct iattr
*);
1721 int (*getattr
) (const struct path
*, struct kstat
*, u32
, unsigned int);
1722 ssize_t (*listxattr
) (struct dentry
*, char *, size_t);
1723 int (*fiemap
)(struct inode
*, struct fiemap_extent_info
*, u64 start
,
1725 int (*update_time
)(struct inode
*, struct timespec
*, int);
1726 int (*atomic_open
)(struct inode
*, struct dentry
*,
1727 struct file
*, unsigned open_flag
,
1728 umode_t create_mode
, int *opened
);
1729 int (*tmpfile
) (struct inode
*, struct dentry
*, umode_t
);
1730 int (*set_acl
)(struct inode
*, struct posix_acl
*, int);
1731 } ____cacheline_aligned
;
1733 static inline ssize_t
call_read_iter(struct file
*file
, struct kiocb
*kio
,
1734 struct iov_iter
*iter
)
1736 return file
->f_op
->read_iter(kio
, iter
);
1739 static inline ssize_t
call_write_iter(struct file
*file
, struct kiocb
*kio
,
1740 struct iov_iter
*iter
)
1742 return file
->f_op
->write_iter(kio
, iter
);
1745 static inline int call_mmap(struct file
*file
, struct vm_area_struct
*vma
)
1747 return file
->f_op
->mmap(file
, vma
);
1750 ssize_t
rw_copy_check_uvector(int type
, const struct iovec __user
* uvector
,
1751 unsigned long nr_segs
, unsigned long fast_segs
,
1752 struct iovec
*fast_pointer
,
1753 struct iovec
**ret_pointer
);
1755 extern ssize_t
__vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1756 extern ssize_t
__vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1757 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1758 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1759 extern ssize_t
vfs_readv(struct file
*, const struct iovec __user
*,
1760 unsigned long, loff_t
*, int);
1761 extern ssize_t
vfs_writev(struct file
*, const struct iovec __user
*,
1762 unsigned long, loff_t
*, int);
1763 extern ssize_t
vfs_copy_file_range(struct file
*, loff_t
, struct file
*,
1764 loff_t
, size_t, unsigned int);
1765 extern int vfs_clone_file_prep_inodes(struct inode
*inode_in
, loff_t pos_in
,
1766 struct inode
*inode_out
, loff_t pos_out
,
1767 u64
*len
, bool is_dedupe
);
1768 extern int vfs_clone_file_range(struct file
*file_in
, loff_t pos_in
,
1769 struct file
*file_out
, loff_t pos_out
, u64 len
);
1770 extern int vfs_dedupe_file_range_compare(struct inode
*src
, loff_t srcoff
,
1771 struct inode
*dest
, loff_t destoff
,
1772 loff_t len
, bool *is_same
);
1773 extern int vfs_dedupe_file_range(struct file
*file
,
1774 struct file_dedupe_range
*same
);
1776 struct super_operations
{
1777 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
1778 void (*destroy_inode
)(struct inode
*);
1780 void (*dirty_inode
) (struct inode
*, int flags
);
1781 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
1782 int (*drop_inode
) (struct inode
*);
1783 void (*evict_inode
) (struct inode
*);
1784 void (*put_super
) (struct super_block
*);
1785 int (*sync_fs
)(struct super_block
*sb
, int wait
);
1786 int (*freeze_super
) (struct super_block
*);
1787 int (*freeze_fs
) (struct super_block
*);
1788 int (*thaw_super
) (struct super_block
*);
1789 int (*unfreeze_fs
) (struct super_block
*);
1790 int (*statfs
) (struct dentry
*, struct kstatfs
*);
1791 int (*remount_fs
) (struct super_block
*, int *, char *);
1792 void (*umount_begin
) (struct super_block
*);
1794 int (*show_options
)(struct seq_file
*, struct dentry
*);
1795 int (*show_devname
)(struct seq_file
*, struct dentry
*);
1796 int (*show_path
)(struct seq_file
*, struct dentry
*);
1797 int (*show_stats
)(struct seq_file
*, struct dentry
*);
1799 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
1800 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
1801 struct dquot
**(*get_dquots
)(struct inode
*);
1803 int (*bdev_try_to_free_page
)(struct super_block
*, struct page
*, gfp_t
);
1804 long (*nr_cached_objects
)(struct super_block
*,
1805 struct shrink_control
*);
1806 long (*free_cached_objects
)(struct super_block
*,
1807 struct shrink_control
*);
1811 * Inode flags - they have no relation to superblock flags now
1813 #define S_SYNC 1 /* Writes are synced at once */
1814 #define S_NOATIME 2 /* Do not update access times */
1815 #define S_APPEND 4 /* Append-only file */
1816 #define S_IMMUTABLE 8 /* Immutable file */
1817 #define S_DEAD 16 /* removed, but still open directory */
1818 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1819 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1820 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1821 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1822 #define S_PRIVATE 512 /* Inode is fs-internal */
1823 #define S_IMA 1024 /* Inode has an associated IMA struct */
1824 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1825 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1826 #ifdef CONFIG_FS_DAX
1827 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1829 #define S_DAX 0 /* Make all the DAX code disappear */
1833 * Note that nosuid etc flags are inode-specific: setting some file-system
1834 * flags just means all the inodes inherit those flags by default. It might be
1835 * possible to override it selectively if you really wanted to with some
1836 * ioctl() that is not currently implemented.
1838 * Exception: MS_RDONLY is always applied to the entire file system.
1840 * Unfortunately, it is possible to change a filesystems flags with it mounted
1841 * with files in use. This means that all of the inodes will not have their
1842 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1843 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1845 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1847 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1848 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1849 ((inode)->i_flags & S_SYNC))
1850 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1851 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1852 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1853 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1854 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1856 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1857 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1858 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1859 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1861 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1862 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1863 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1864 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1865 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1866 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1867 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1868 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1870 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1871 (inode)->i_rdev == WHITEOUT_DEV)
1873 static inline bool HAS_UNMAPPED_ID(struct inode
*inode
)
1875 return !uid_valid(inode
->i_uid
) || !gid_valid(inode
->i_gid
);
1878 static inline enum rw_hint
file_write_hint(struct file
*file
)
1880 if (file
->f_write_hint
!= WRITE_LIFE_NOT_SET
)
1881 return file
->f_write_hint
;
1883 return file_inode(file
)->i_write_hint
;
1886 static inline int iocb_flags(struct file
*file
);
1888 static inline void init_sync_kiocb(struct kiocb
*kiocb
, struct file
*filp
)
1890 *kiocb
= (struct kiocb
) {
1892 .ki_flags
= iocb_flags(filp
),
1893 .ki_hint
= file_write_hint(filp
),
1898 * Inode state bits. Protected by inode->i_lock
1900 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1901 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1903 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1904 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1905 * various stages of removing an inode.
1907 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1909 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1910 * fdatasync(). i_atime is the usual cause.
1911 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1912 * these changes separately from I_DIRTY_SYNC so that we
1913 * don't have to write inode on fdatasync() when only
1914 * mtime has changed in it.
1915 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1916 * I_NEW Serves as both a mutex and completion notification.
1917 * New inodes set I_NEW. If two processes both create
1918 * the same inode, one of them will release its inode and
1919 * wait for I_NEW to be released before returning.
1920 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1921 * also cause waiting on I_NEW, without I_NEW actually
1922 * being set. find_inode() uses this to prevent returning
1923 * nearly-dead inodes.
1924 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1925 * is zero. I_FREEING must be set when I_WILL_FREE is
1927 * I_FREEING Set when inode is about to be freed but still has dirty
1928 * pages or buffers attached or the inode itself is still
1930 * I_CLEAR Added by clear_inode(). In this state the inode is
1931 * clean and can be destroyed. Inode keeps I_FREEING.
1933 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1934 * prohibited for many purposes. iget() must wait for
1935 * the inode to be completely released, then create it
1936 * anew. Other functions will just ignore such inodes,
1937 * if appropriate. I_NEW is used for waiting.
1939 * I_SYNC Writeback of inode is running. The bit is set during
1940 * data writeback, and cleared with a wakeup on the bit
1941 * address once it is done. The bit is also used to pin
1942 * the inode in memory for flusher thread.
1944 * I_REFERENCED Marks the inode as recently references on the LRU list.
1946 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1948 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1949 * synchronize competing switching instances and to tell
1950 * wb stat updates to grab mapping->tree_lock. See
1951 * inode_switch_wb_work_fn() for details.
1953 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
1954 * and work dirs among overlayfs mounts.
1956 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1958 #define I_DIRTY_SYNC (1 << 0)
1959 #define I_DIRTY_DATASYNC (1 << 1)
1960 #define I_DIRTY_PAGES (1 << 2)
1962 #define I_NEW (1 << __I_NEW)
1963 #define I_WILL_FREE (1 << 4)
1964 #define I_FREEING (1 << 5)
1965 #define I_CLEAR (1 << 6)
1967 #define I_SYNC (1 << __I_SYNC)
1968 #define I_REFERENCED (1 << 8)
1969 #define __I_DIO_WAKEUP 9
1970 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1971 #define I_LINKABLE (1 << 10)
1972 #define I_DIRTY_TIME (1 << 11)
1973 #define __I_DIRTY_TIME_EXPIRED 12
1974 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1975 #define I_WB_SWITCH (1 << 13)
1976 #define I_OVL_INUSE (1 << 14)
1978 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1979 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1981 extern void __mark_inode_dirty(struct inode
*, int);
1982 static inline void mark_inode_dirty(struct inode
*inode
)
1984 __mark_inode_dirty(inode
, I_DIRTY
);
1987 static inline void mark_inode_dirty_sync(struct inode
*inode
)
1989 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
1992 extern void inc_nlink(struct inode
*inode
);
1993 extern void drop_nlink(struct inode
*inode
);
1994 extern void clear_nlink(struct inode
*inode
);
1995 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
1997 static inline void inode_inc_link_count(struct inode
*inode
)
2000 mark_inode_dirty(inode
);
2003 static inline void inode_dec_link_count(struct inode
*inode
)
2006 mark_inode_dirty(inode
);
2010 * inode_inc_iversion - increments i_version
2011 * @inode: inode that need to be updated
2013 * Every time the inode is modified, the i_version field will be incremented.
2014 * The filesystem has to be mounted with i_version flag
2017 static inline void inode_inc_iversion(struct inode
*inode
)
2019 spin_lock(&inode
->i_lock
);
2021 spin_unlock(&inode
->i_lock
);
2024 enum file_time_flags
{
2031 extern void touch_atime(const struct path
*);
2032 static inline void file_accessed(struct file
*file
)
2034 if (!(file
->f_flags
& O_NOATIME
))
2035 touch_atime(&file
->f_path
);
2038 int sync_inode(struct inode
*inode
, struct writeback_control
*wbc
);
2039 int sync_inode_metadata(struct inode
*inode
, int wait
);
2041 struct file_system_type
{
2044 #define FS_REQUIRES_DEV 1
2045 #define FS_BINARY_MOUNTDATA 2
2046 #define FS_HAS_SUBTYPE 4
2047 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2048 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2049 struct dentry
*(*mount
) (struct file_system_type
*, int,
2050 const char *, void *);
2051 void (*kill_sb
) (struct super_block
*);
2052 struct module
*owner
;
2053 struct file_system_type
* next
;
2054 struct hlist_head fs_supers
;
2056 struct lock_class_key s_lock_key
;
2057 struct lock_class_key s_umount_key
;
2058 struct lock_class_key s_vfs_rename_key
;
2059 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
2061 struct lock_class_key i_lock_key
;
2062 struct lock_class_key i_mutex_key
;
2063 struct lock_class_key i_mutex_dir_key
;
2066 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2068 extern struct dentry
*mount_ns(struct file_system_type
*fs_type
,
2069 int flags
, void *data
, void *ns
, struct user_namespace
*user_ns
,
2070 int (*fill_super
)(struct super_block
*, void *, int));
2071 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
2072 int flags
, const char *dev_name
, void *data
,
2073 int (*fill_super
)(struct super_block
*, void *, int));
2074 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
2075 int flags
, void *data
,
2076 int (*fill_super
)(struct super_block
*, void *, int));
2077 extern struct dentry
*mount_nodev(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_subtree(struct vfsmount
*mnt
, const char *path
);
2081 void generic_shutdown_super(struct super_block
*sb
);
2082 void kill_block_super(struct super_block
*sb
);
2083 void kill_anon_super(struct super_block
*sb
);
2084 void kill_litter_super(struct super_block
*sb
);
2085 void deactivate_super(struct super_block
*sb
);
2086 void deactivate_locked_super(struct super_block
*sb
);
2087 int set_anon_super(struct super_block
*s
, void *data
);
2088 int get_anon_bdev(dev_t
*);
2089 void free_anon_bdev(dev_t
);
2090 struct super_block
*sget_userns(struct file_system_type
*type
,
2091 int (*test
)(struct super_block
*,void *),
2092 int (*set
)(struct super_block
*,void *),
2093 int flags
, struct user_namespace
*user_ns
,
2095 struct super_block
*sget(struct file_system_type
*type
,
2096 int (*test
)(struct super_block
*,void *),
2097 int (*set
)(struct super_block
*,void *),
2098 int flags
, void *data
);
2099 extern struct dentry
*mount_pseudo_xattr(struct file_system_type
*, char *,
2100 const struct super_operations
*ops
,
2101 const struct xattr_handler
**xattr
,
2102 const struct dentry_operations
*dops
,
2105 static inline struct dentry
*
2106 mount_pseudo(struct file_system_type
*fs_type
, char *name
,
2107 const struct super_operations
*ops
,
2108 const struct dentry_operations
*dops
, unsigned long magic
)
2110 return mount_pseudo_xattr(fs_type
, name
, ops
, NULL
, dops
, magic
);
2113 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2114 #define fops_get(fops) \
2115 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2116 #define fops_put(fops) \
2117 do { if (fops) module_put((fops)->owner); } while(0)
2119 * This one is to be used *ONLY* from ->open() instances.
2120 * fops must be non-NULL, pinned down *and* module dependencies
2121 * should be sufficient to pin the caller down as well.
2123 #define replace_fops(f, fops) \
2125 struct file *__file = (f); \
2126 fops_put(__file->f_op); \
2127 BUG_ON(!(__file->f_op = (fops))); \
2130 extern int register_filesystem(struct file_system_type
*);
2131 extern int unregister_filesystem(struct file_system_type
*);
2132 extern struct vfsmount
*kern_mount_data(struct file_system_type
*, void *data
);
2133 #define kern_mount(type) kern_mount_data(type, NULL)
2134 extern void kern_unmount(struct vfsmount
*mnt
);
2135 extern int may_umount_tree(struct vfsmount
*);
2136 extern int may_umount(struct vfsmount
*);
2137 extern long do_mount(const char *, const char __user
*,
2138 const char *, unsigned long, void *);
2139 extern struct vfsmount
*collect_mounts(const struct path
*);
2140 extern void drop_collected_mounts(struct vfsmount
*);
2141 extern int iterate_mounts(int (*)(struct vfsmount
*, void *), void *,
2143 extern int vfs_statfs(const struct path
*, struct kstatfs
*);
2144 extern int user_statfs(const char __user
*, struct kstatfs
*);
2145 extern int fd_statfs(int, struct kstatfs
*);
2146 extern int vfs_ustat(dev_t
, struct kstatfs
*);
2147 extern int freeze_super(struct super_block
*super
);
2148 extern int thaw_super(struct super_block
*super
);
2149 extern bool our_mnt(struct vfsmount
*mnt
);
2150 extern __printf(2, 3)
2151 int super_setup_bdi_name(struct super_block
*sb
, char *fmt
, ...);
2152 extern int super_setup_bdi(struct super_block
*sb
);
2154 extern int current_umask(void);
2156 extern void ihold(struct inode
* inode
);
2157 extern void iput(struct inode
*);
2158 extern int generic_update_time(struct inode
*, struct timespec
*, int);
2161 extern struct kobject
*fs_kobj
;
2163 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2165 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2166 extern int locks_mandatory_locked(struct file
*);
2167 extern int locks_mandatory_area(struct inode
*, struct file
*, loff_t
, loff_t
, unsigned char);
2170 * Candidates for mandatory locking have the setgid bit set
2171 * but no group execute bit - an otherwise meaningless combination.
2174 static inline int __mandatory_lock(struct inode
*ino
)
2176 return (ino
->i_mode
& (S_ISGID
| S_IXGRP
)) == S_ISGID
;
2180 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2181 * otherwise these will be advisory locks
2184 static inline int mandatory_lock(struct inode
*ino
)
2186 return IS_MANDLOCK(ino
) && __mandatory_lock(ino
);
2189 static inline int locks_verify_locked(struct file
*file
)
2191 if (mandatory_lock(locks_inode(file
)))
2192 return locks_mandatory_locked(file
);
2196 static inline int locks_verify_truncate(struct inode
*inode
,
2200 if (!inode
->i_flctx
|| !mandatory_lock(inode
))
2203 if (size
< inode
->i_size
) {
2204 return locks_mandatory_area(inode
, f
, size
, inode
->i_size
- 1,
2207 return locks_mandatory_area(inode
, f
, inode
->i_size
, size
- 1,
2212 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2214 static inline int locks_mandatory_locked(struct file
*file
)
2219 static inline int locks_mandatory_area(struct inode
*inode
, struct file
*filp
,
2220 loff_t start
, loff_t end
, unsigned char type
)
2225 static inline int __mandatory_lock(struct inode
*inode
)
2230 static inline int mandatory_lock(struct inode
*inode
)
2235 static inline int locks_verify_locked(struct file
*file
)
2240 static inline int locks_verify_truncate(struct inode
*inode
, struct file
*filp
,
2246 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2249 #ifdef CONFIG_FILE_LOCKING
2250 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2253 * Since this check is lockless, we must ensure that any refcounts
2254 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2255 * could end up racing with tasks trying to set a new lease on this
2259 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2260 return __break_lease(inode
, mode
, FL_LEASE
);
2264 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2267 * Since this check is lockless, we must ensure that any refcounts
2268 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2269 * could end up racing with tasks trying to set a new lease on this
2273 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2274 return __break_lease(inode
, mode
, FL_DELEG
);
2278 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2282 ret
= break_deleg(inode
, O_WRONLY
|O_NONBLOCK
);
2283 if (ret
== -EWOULDBLOCK
&& delegated_inode
) {
2284 *delegated_inode
= inode
;
2290 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2294 ret
= break_deleg(*delegated_inode
, O_WRONLY
);
2295 iput(*delegated_inode
);
2296 *delegated_inode
= NULL
;
2300 static inline int break_layout(struct inode
*inode
, bool wait
)
2303 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2304 return __break_lease(inode
,
2305 wait
? O_WRONLY
: O_WRONLY
| O_NONBLOCK
,
2310 #else /* !CONFIG_FILE_LOCKING */
2311 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2316 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2321 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2326 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2332 static inline int break_layout(struct inode
*inode
, bool wait
)
2337 #endif /* CONFIG_FILE_LOCKING */
2342 const char *name
; /* pointer to actual string */
2343 const __user
char *uptr
; /* original userland pointer */
2344 struct audit_names
*aname
;
2349 extern long vfs_truncate(const struct path
*, loff_t
);
2350 extern int do_truncate(struct dentry
*, loff_t start
, unsigned int time_attrs
,
2352 extern int vfs_fallocate(struct file
*file
, int mode
, loff_t offset
,
2354 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
2356 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
2357 extern struct file
*filp_open(const char *, int, umode_t
);
2358 extern struct file
*file_open_root(struct dentry
*, struct vfsmount
*,
2359 const char *, int, umode_t
);
2360 extern struct file
* dentry_open(const struct path
*, int, const struct cred
*);
2361 extern int filp_close(struct file
*, fl_owner_t id
);
2363 extern struct filename
*getname_flags(const char __user
*, int, int *);
2364 extern struct filename
*getname(const char __user
*);
2365 extern struct filename
*getname_kernel(const char *);
2366 extern void putname(struct filename
*name
);
2372 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
2373 int (*open
)(struct inode
*, struct file
*),
2375 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
2379 extern int ioctl_preallocate(struct file
*filp
, void __user
*argp
);
2382 extern void __init
vfs_caches_init_early(void);
2383 extern void __init
vfs_caches_init(void);
2385 extern struct kmem_cache
*names_cachep
;
2387 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2388 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2391 extern int register_blkdev(unsigned int, const char *);
2392 extern void unregister_blkdev(unsigned int, const char *);
2393 extern void bdev_unhash_inode(dev_t dev
);
2394 extern struct block_device
*bdget(dev_t
);
2395 extern struct block_device
*bdgrab(struct block_device
*bdev
);
2396 extern void bd_set_size(struct block_device
*, loff_t size
);
2397 extern void bd_forget(struct inode
*inode
);
2398 extern void bdput(struct block_device
*);
2399 extern void invalidate_bdev(struct block_device
*);
2400 extern void iterate_bdevs(void (*)(struct block_device
*, void *), void *);
2401 extern int sync_blockdev(struct block_device
*bdev
);
2402 extern void kill_bdev(struct block_device
*);
2403 extern struct super_block
*freeze_bdev(struct block_device
*);
2404 extern void emergency_thaw_all(void);
2405 extern int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
);
2406 extern int fsync_bdev(struct block_device
*);
2408 extern struct super_block
*blockdev_superblock
;
2410 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2412 return sb
== blockdev_superblock
;
2415 static inline void bd_forget(struct inode
*inode
) {}
2416 static inline int sync_blockdev(struct block_device
*bdev
) { return 0; }
2417 static inline void kill_bdev(struct block_device
*bdev
) {}
2418 static inline void invalidate_bdev(struct block_device
*bdev
) {}
2420 static inline struct super_block
*freeze_bdev(struct block_device
*sb
)
2425 static inline int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
)
2430 static inline void iterate_bdevs(void (*f
)(struct block_device
*, void *), void *arg
)
2434 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2439 extern int sync_filesystem(struct super_block
*);
2440 extern const struct file_operations def_blk_fops
;
2441 extern const struct file_operations def_chr_fops
;
2443 extern int ioctl_by_bdev(struct block_device
*, unsigned, unsigned long);
2444 extern int blkdev_ioctl(struct block_device
*, fmode_t
, unsigned, unsigned long);
2445 extern long compat_blkdev_ioctl(struct file
*, unsigned, unsigned long);
2446 extern int blkdev_get(struct block_device
*bdev
, fmode_t mode
, void *holder
);
2447 extern struct block_device
*blkdev_get_by_path(const char *path
, fmode_t mode
,
2449 extern struct block_device
*blkdev_get_by_dev(dev_t dev
, fmode_t mode
,
2451 extern void blkdev_put(struct block_device
*bdev
, fmode_t mode
);
2452 extern int __blkdev_reread_part(struct block_device
*bdev
);
2453 extern int blkdev_reread_part(struct block_device
*bdev
);
2456 extern int bd_link_disk_holder(struct block_device
*bdev
, struct gendisk
*disk
);
2457 extern void bd_unlink_disk_holder(struct block_device
*bdev
,
2458 struct gendisk
*disk
);
2460 static inline int bd_link_disk_holder(struct block_device
*bdev
,
2461 struct gendisk
*disk
)
2465 static inline void bd_unlink_disk_holder(struct block_device
*bdev
,
2466 struct gendisk
*disk
)
2473 #define CHRDEV_MAJOR_HASH_SIZE 255
2474 /* Marks the bottom of the first segment of free char majors */
2475 #define CHRDEV_MAJOR_DYN_END 234
2476 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
2477 extern int register_chrdev_region(dev_t
, unsigned, const char *);
2478 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
2479 unsigned int count
, const char *name
,
2480 const struct file_operations
*fops
);
2481 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
2482 unsigned int count
, const char *name
);
2483 extern void unregister_chrdev_region(dev_t
, unsigned);
2484 extern void chrdev_show(struct seq_file
*,off_t
);
2486 static inline int register_chrdev(unsigned int major
, const char *name
,
2487 const struct file_operations
*fops
)
2489 return __register_chrdev(major
, 0, 256, name
, fops
);
2492 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2494 __unregister_chrdev(major
, 0, 256, name
);
2497 /* fs/block_dev.c */
2498 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2499 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2502 #define BLKDEV_MAJOR_HASH_SIZE 255
2503 extern const char *__bdevname(dev_t
, char *buffer
);
2504 extern const char *bdevname(struct block_device
*bdev
, char *buffer
);
2505 extern struct block_device
*lookup_bdev(const char *);
2506 extern void blkdev_show(struct seq_file
*,off_t
);
2509 #define BLKDEV_MAJOR_HASH_SIZE 0
2512 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2514 /* Invalid inode operations -- fs/bad_inode.c */
2515 extern void make_bad_inode(struct inode
*);
2516 extern bool is_bad_inode(struct inode
*);
2519 extern void check_disk_size_change(struct gendisk
*disk
,
2520 struct block_device
*bdev
);
2521 extern int revalidate_disk(struct gendisk
*);
2522 extern int check_disk_change(struct block_device
*);
2523 extern int __invalidate_device(struct block_device
*, bool);
2524 extern int invalidate_partition(struct gendisk
*, int);
2526 unsigned long invalidate_mapping_pages(struct address_space
*mapping
,
2527 pgoff_t start
, pgoff_t end
);
2529 static inline void invalidate_remote_inode(struct inode
*inode
)
2531 if (S_ISREG(inode
->i_mode
) || S_ISDIR(inode
->i_mode
) ||
2532 S_ISLNK(inode
->i_mode
))
2533 invalidate_mapping_pages(inode
->i_mapping
, 0, -1);
2535 extern int invalidate_inode_pages2(struct address_space
*mapping
);
2536 extern int invalidate_inode_pages2_range(struct address_space
*mapping
,
2537 pgoff_t start
, pgoff_t end
);
2538 extern int write_inode_now(struct inode
*, int);
2539 extern int filemap_fdatawrite(struct address_space
*);
2540 extern int filemap_flush(struct address_space
*);
2541 extern int filemap_fdatawait(struct address_space
*);
2542 extern int filemap_fdatawait_keep_errors(struct address_space
*mapping
);
2543 extern int filemap_fdatawait_range(struct address_space
*, loff_t lstart
,
2545 extern bool filemap_range_has_page(struct address_space
*, loff_t lstart
,
2547 extern int filemap_write_and_wait(struct address_space
*mapping
);
2548 extern int filemap_write_and_wait_range(struct address_space
*mapping
,
2549 loff_t lstart
, loff_t lend
);
2550 extern int __filemap_fdatawrite_range(struct address_space
*mapping
,
2551 loff_t start
, loff_t end
, int sync_mode
);
2552 extern int filemap_fdatawrite_range(struct address_space
*mapping
,
2553 loff_t start
, loff_t end
);
2554 extern int filemap_check_errors(struct address_space
*mapping
);
2556 extern void __filemap_set_wb_err(struct address_space
*mapping
, int err
);
2557 extern int __must_check
file_check_and_advance_wb_err(struct file
*file
);
2558 extern int __must_check
file_write_and_wait_range(struct file
*file
,
2559 loff_t start
, loff_t end
);
2562 * filemap_set_wb_err - set a writeback error on an address_space
2563 * @mapping: mapping in which to set writeback error
2564 * @err: error to be set in mapping
2566 * When writeback fails in some way, we must record that error so that
2567 * userspace can be informed when fsync and the like are called. We endeavor
2568 * to report errors on any file that was open at the time of the error. Some
2569 * internal callers also need to know when writeback errors have occurred.
2571 * When a writeback error occurs, most filesystems will want to call
2572 * filemap_set_wb_err to record the error in the mapping so that it will be
2573 * automatically reported whenever fsync is called on the file.
2575 * FIXME: mention FS_* flag here?
2577 static inline void filemap_set_wb_err(struct address_space
*mapping
, int err
)
2579 /* Fastpath for common case of no error */
2581 __filemap_set_wb_err(mapping
, err
);
2585 * filemap_check_wb_error - has an error occurred since the mark was sampled?
2586 * @mapping: mapping to check for writeback errors
2587 * @since: previously-sampled errseq_t
2589 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2590 * the given value was sampled.
2592 * If it has then report the latest error set, otherwise return 0.
2594 static inline int filemap_check_wb_err(struct address_space
*mapping
,
2597 return errseq_check(&mapping
->wb_err
, since
);
2601 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2602 * @mapping: mapping to be sampled
2604 * Writeback errors are always reported relative to a particular sample point
2605 * in the past. This function provides those sample points.
2607 static inline errseq_t
filemap_sample_wb_err(struct address_space
*mapping
)
2609 return errseq_sample(&mapping
->wb_err
);
2612 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
2614 extern int vfs_fsync(struct file
*file
, int datasync
);
2617 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2618 * to already be updated for the write, and will return either the amount
2619 * of bytes passed in, or an error if syncing the file failed.
2621 static inline ssize_t
generic_write_sync(struct kiocb
*iocb
, ssize_t count
)
2623 if (iocb
->ki_flags
& IOCB_DSYNC
) {
2624 int ret
= vfs_fsync_range(iocb
->ki_filp
,
2625 iocb
->ki_pos
- count
, iocb
->ki_pos
- 1,
2626 (iocb
->ki_flags
& IOCB_SYNC
) ? 0 : 1);
2634 extern void emergency_sync(void);
2635 extern void emergency_remount(void);
2637 extern sector_t
bmap(struct inode
*, sector_t
);
2639 extern int notify_change(struct dentry
*, struct iattr
*, struct inode
**);
2640 extern int inode_permission(struct inode
*, int);
2641 extern int __inode_permission(struct inode
*, int);
2642 extern int generic_permission(struct inode
*, int);
2643 extern int __check_sticky(struct inode
*dir
, struct inode
*inode
);
2645 static inline bool execute_ok(struct inode
*inode
)
2647 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
2650 static inline void file_start_write(struct file
*file
)
2652 if (!S_ISREG(file_inode(file
)->i_mode
))
2654 __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, true);
2657 static inline bool file_start_write_trylock(struct file
*file
)
2659 if (!S_ISREG(file_inode(file
)->i_mode
))
2661 return __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, false);
2664 static inline void file_end_write(struct file
*file
)
2666 if (!S_ISREG(file_inode(file
)->i_mode
))
2668 __sb_end_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
);
2671 static inline int do_clone_file_range(struct file
*file_in
, loff_t pos_in
,
2672 struct file
*file_out
, loff_t pos_out
,
2677 file_start_write(file_out
);
2678 ret
= vfs_clone_file_range(file_in
, pos_in
, file_out
, pos_out
, len
);
2679 file_end_write(file_out
);
2685 * get_write_access() gets write permission for a file.
2686 * put_write_access() releases this write permission.
2687 * This is used for regular files.
2688 * We cannot support write (and maybe mmap read-write shared) accesses and
2689 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2690 * can have the following values:
2691 * 0: no writers, no VM_DENYWRITE mappings
2692 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2693 * > 0: (i_writecount) users are writing to the file.
2695 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2696 * except for the cases where we don't hold i_writecount yet. Then we need to
2697 * use {get,deny}_write_access() - these functions check the sign and refuse
2698 * to do the change if sign is wrong.
2700 static inline int get_write_access(struct inode
*inode
)
2702 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2704 static inline int deny_write_access(struct file
*file
)
2706 struct inode
*inode
= file_inode(file
);
2707 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2709 static inline void put_write_access(struct inode
* inode
)
2711 atomic_dec(&inode
->i_writecount
);
2713 static inline void allow_write_access(struct file
*file
)
2716 atomic_inc(&file_inode(file
)->i_writecount
);
2718 static inline bool inode_is_open_for_write(const struct inode
*inode
)
2720 return atomic_read(&inode
->i_writecount
) > 0;
2724 static inline void i_readcount_dec(struct inode
*inode
)
2726 BUG_ON(!atomic_read(&inode
->i_readcount
));
2727 atomic_dec(&inode
->i_readcount
);
2729 static inline void i_readcount_inc(struct inode
*inode
)
2731 atomic_inc(&inode
->i_readcount
);
2734 static inline void i_readcount_dec(struct inode
*inode
)
2738 static inline void i_readcount_inc(struct inode
*inode
)
2743 extern int do_pipe_flags(int *, int);
2745 #define __kernel_read_file_id(id) \
2746 id(UNKNOWN, unknown) \
2747 id(FIRMWARE, firmware) \
2748 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2749 id(MODULE, kernel-module) \
2750 id(KEXEC_IMAGE, kexec-image) \
2751 id(KEXEC_INITRAMFS, kexec-initramfs) \
2752 id(POLICY, security-policy) \
2755 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2756 #define __fid_stringify(dummy, str) #str,
2758 enum kernel_read_file_id
{
2759 __kernel_read_file_id(__fid_enumify
)
2762 static const char * const kernel_read_file_str
[] = {
2763 __kernel_read_file_id(__fid_stringify
)
2766 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id
)
2768 if ((unsigned)id
>= READING_MAX_ID
)
2769 return kernel_read_file_str
[READING_UNKNOWN
];
2771 return kernel_read_file_str
[id
];
2774 extern int kernel_read(struct file
*, loff_t
, char *, unsigned long);
2775 extern int kernel_read_file(struct file
*, void **, loff_t
*, loff_t
,
2776 enum kernel_read_file_id
);
2777 extern int kernel_read_file_from_path(char *, void **, loff_t
*, loff_t
,
2778 enum kernel_read_file_id
);
2779 extern int kernel_read_file_from_fd(int, void **, loff_t
*, loff_t
,
2780 enum kernel_read_file_id
);
2781 extern ssize_t
kernel_write(struct file
*, const char *, size_t, loff_t
);
2782 extern ssize_t
__kernel_write(struct file
*, const char *, size_t, loff_t
*);
2783 extern struct file
* open_exec(const char *);
2785 /* fs/dcache.c -- generic fs support functions */
2786 extern bool is_subdir(struct dentry
*, struct dentry
*);
2787 extern bool path_is_under(const struct path
*, const struct path
*);
2789 extern char *file_path(struct file
*, char *, int);
2791 #include <linux/err.h>
2793 /* needed for stackable file system support */
2794 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int whence
);
2796 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int whence
);
2798 extern int inode_init_always(struct super_block
*, struct inode
*);
2799 extern void inode_init_once(struct inode
*);
2800 extern void address_space_init_once(struct address_space
*mapping
);
2801 extern struct inode
* igrab(struct inode
*);
2802 extern ino_t
iunique(struct super_block
*, ino_t
);
2803 extern int inode_needs_sync(struct inode
*inode
);
2804 extern int generic_delete_inode(struct inode
*inode
);
2805 static inline int generic_drop_inode(struct inode
*inode
)
2807 return !inode
->i_nlink
|| inode_unhashed(inode
);
2810 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
2811 unsigned long hashval
, int (*test
)(struct inode
*, void *),
2813 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
2814 int (*test
)(struct inode
*, void *), void *data
);
2815 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
2817 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
2818 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
2819 extern struct inode
*find_inode_nowait(struct super_block
*,
2821 int (*match
)(struct inode
*,
2822 unsigned long, void *),
2824 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
2825 extern int insert_inode_locked(struct inode
*);
2826 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2827 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
2829 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
2831 extern void unlock_new_inode(struct inode
*);
2832 extern unsigned int get_next_ino(void);
2834 extern void __iget(struct inode
* inode
);
2835 extern void iget_failed(struct inode
*);
2836 extern void clear_inode(struct inode
*);
2837 extern void __destroy_inode(struct inode
*);
2838 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
2839 extern struct inode
*new_inode(struct super_block
*sb
);
2840 extern void free_inode_nonrcu(struct inode
*inode
);
2841 extern int should_remove_suid(struct dentry
*);
2842 extern int file_remove_privs(struct file
*);
2844 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
2845 static inline void insert_inode_hash(struct inode
*inode
)
2847 __insert_inode_hash(inode
, inode
->i_ino
);
2850 extern void __remove_inode_hash(struct inode
*);
2851 static inline void remove_inode_hash(struct inode
*inode
)
2853 if (!inode_unhashed(inode
) && !hlist_fake(&inode
->i_hash
))
2854 __remove_inode_hash(inode
);
2857 extern void inode_sb_list_add(struct inode
*inode
);
2860 extern int bdev_read_only(struct block_device
*);
2862 extern int set_blocksize(struct block_device
*, int);
2863 extern int sb_set_blocksize(struct super_block
*, int);
2864 extern int sb_min_blocksize(struct super_block
*, int);
2866 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
2867 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
2868 extern ssize_t
generic_write_checks(struct kiocb
*, struct iov_iter
*);
2869 extern ssize_t
generic_file_read_iter(struct kiocb
*, struct iov_iter
*);
2870 extern ssize_t
__generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2871 extern ssize_t
generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2872 extern ssize_t
generic_file_direct_write(struct kiocb
*, struct iov_iter
*);
2873 extern ssize_t
generic_perform_write(struct file
*, struct iov_iter
*, loff_t
);
2875 ssize_t
vfs_iter_read(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
2877 ssize_t
vfs_iter_write(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
2880 /* fs/block_dev.c */
2881 extern ssize_t
blkdev_read_iter(struct kiocb
*iocb
, struct iov_iter
*to
);
2882 extern ssize_t
blkdev_write_iter(struct kiocb
*iocb
, struct iov_iter
*from
);
2883 extern int blkdev_fsync(struct file
*filp
, loff_t start
, loff_t end
,
2885 extern void block_sync_page(struct page
*page
);
2888 extern ssize_t
generic_file_splice_read(struct file
*, loff_t
*,
2889 struct pipe_inode_info
*, size_t, unsigned int);
2890 extern ssize_t
iter_file_splice_write(struct pipe_inode_info
*,
2891 struct file
*, loff_t
*, size_t, unsigned int);
2892 extern ssize_t
generic_splice_sendpage(struct pipe_inode_info
*pipe
,
2893 struct file
*out
, loff_t
*, size_t len
, unsigned int flags
);
2894 extern long do_splice_direct(struct file
*in
, loff_t
*ppos
, struct file
*out
,
2895 loff_t
*opos
, size_t len
, unsigned int flags
);
2899 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
2900 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int whence
);
2901 extern loff_t
no_llseek(struct file
*file
, loff_t offset
, int whence
);
2902 extern loff_t
vfs_setpos(struct file
*file
, loff_t offset
, loff_t maxsize
);
2903 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int whence
);
2904 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
2905 int whence
, loff_t maxsize
, loff_t eof
);
2906 extern loff_t
fixed_size_llseek(struct file
*file
, loff_t offset
,
2907 int whence
, loff_t size
);
2908 extern loff_t
no_seek_end_llseek_size(struct file
*, loff_t
, int, loff_t
);
2909 extern loff_t
no_seek_end_llseek(struct file
*, loff_t
, int);
2910 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
2911 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
2914 typedef void (dio_submit_t
)(struct bio
*bio
, struct inode
*inode
,
2915 loff_t file_offset
);
2918 /* need locking between buffered and direct access */
2921 /* filesystem does not support filling holes */
2922 DIO_SKIP_HOLES
= 0x02,
2924 /* filesystem can handle aio writes beyond i_size */
2925 DIO_ASYNC_EXTEND
= 0x04,
2927 /* inode/fs/bdev does not need truncate protection */
2928 DIO_SKIP_DIO_COUNT
= 0x08,
2931 void dio_end_io(struct bio
*bio
);
2933 ssize_t
__blockdev_direct_IO(struct kiocb
*iocb
, struct inode
*inode
,
2934 struct block_device
*bdev
, struct iov_iter
*iter
,
2935 get_block_t get_block
,
2936 dio_iodone_t end_io
, dio_submit_t submit_io
,
2939 static inline ssize_t
blockdev_direct_IO(struct kiocb
*iocb
,
2940 struct inode
*inode
,
2941 struct iov_iter
*iter
,
2942 get_block_t get_block
)
2944 return __blockdev_direct_IO(iocb
, inode
, inode
->i_sb
->s_bdev
, iter
,
2945 get_block
, NULL
, NULL
, DIO_LOCKING
| DIO_SKIP_HOLES
);
2949 void inode_dio_wait(struct inode
*inode
);
2952 * inode_dio_begin - signal start of a direct I/O requests
2953 * @inode: inode the direct I/O happens on
2955 * This is called once we've finished processing a direct I/O request,
2956 * and is used to wake up callers waiting for direct I/O to be quiesced.
2958 static inline void inode_dio_begin(struct inode
*inode
)
2960 atomic_inc(&inode
->i_dio_count
);
2964 * inode_dio_end - signal finish of a direct I/O requests
2965 * @inode: inode the direct I/O happens on
2967 * This is called once we've finished processing a direct I/O request,
2968 * and is used to wake up callers waiting for direct I/O to be quiesced.
2970 static inline void inode_dio_end(struct inode
*inode
)
2972 if (atomic_dec_and_test(&inode
->i_dio_count
))
2973 wake_up_bit(&inode
->i_state
, __I_DIO_WAKEUP
);
2976 extern void inode_set_flags(struct inode
*inode
, unsigned int flags
,
2979 extern const struct file_operations generic_ro_fops
;
2981 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2983 extern int readlink_copy(char __user
*, int, const char *);
2984 extern int page_readlink(struct dentry
*, char __user
*, int);
2985 extern const char *page_get_link(struct dentry
*, struct inode
*,
2986 struct delayed_call
*);
2987 extern void page_put_link(void *);
2988 extern int __page_symlink(struct inode
*inode
, const char *symname
, int len
,
2990 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
2991 extern const struct inode_operations page_symlink_inode_operations
;
2992 extern void kfree_link(void *);
2993 extern void generic_fillattr(struct inode
*, struct kstat
*);
2994 extern int vfs_getattr_nosec(const struct path
*, struct kstat
*, u32
, unsigned int);
2995 extern int vfs_getattr(const struct path
*, struct kstat
*, u32
, unsigned int);
2996 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2997 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
2998 void __inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
2999 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
3000 loff_t
inode_get_bytes(struct inode
*inode
);
3001 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
3002 const char *simple_get_link(struct dentry
*, struct inode
*,
3003 struct delayed_call
*);
3004 extern const struct inode_operations simple_symlink_inode_operations
;
3006 extern int iterate_dir(struct file
*, struct dir_context
*);
3008 extern int vfs_statx(int, const char __user
*, int, struct kstat
*, u32
);
3009 extern int vfs_statx_fd(unsigned int, struct kstat
*, u32
, unsigned int);
3011 static inline int vfs_stat(const char __user
*filename
, struct kstat
*stat
)
3013 return vfs_statx(AT_FDCWD
, filename
, AT_NO_AUTOMOUNT
,
3014 stat
, STATX_BASIC_STATS
);
3016 static inline int vfs_lstat(const char __user
*name
, struct kstat
*stat
)
3018 return vfs_statx(AT_FDCWD
, name
, AT_SYMLINK_NOFOLLOW
| AT_NO_AUTOMOUNT
,
3019 stat
, STATX_BASIC_STATS
);
3021 static inline int vfs_fstatat(int dfd
, const char __user
*filename
,
3022 struct kstat
*stat
, int flags
)
3024 return vfs_statx(dfd
, filename
, flags
| AT_NO_AUTOMOUNT
,
3025 stat
, STATX_BASIC_STATS
);
3027 static inline int vfs_fstat(int fd
, struct kstat
*stat
)
3029 return vfs_statx_fd(fd
, stat
, STATX_BASIC_STATS
, 0);
3033 extern const char *vfs_get_link(struct dentry
*, struct delayed_call
*);
3034 extern int vfs_readlink(struct dentry
*, char __user
*, int);
3036 extern int __generic_block_fiemap(struct inode
*inode
,
3037 struct fiemap_extent_info
*fieinfo
,
3038 loff_t start
, loff_t len
,
3039 get_block_t
*get_block
);
3040 extern int generic_block_fiemap(struct inode
*inode
,
3041 struct fiemap_extent_info
*fieinfo
, u64 start
,
3042 u64 len
, get_block_t
*get_block
);
3044 extern struct file_system_type
*get_filesystem(struct file_system_type
*fs
);
3045 extern void put_filesystem(struct file_system_type
*fs
);
3046 extern struct file_system_type
*get_fs_type(const char *name
);
3047 extern struct super_block
*get_super(struct block_device
*);
3048 extern struct super_block
*get_super_thawed(struct block_device
*);
3049 extern struct super_block
*get_super_exclusive_thawed(struct block_device
*bdev
);
3050 extern struct super_block
*get_active_super(struct block_device
*bdev
);
3051 extern void drop_super(struct super_block
*sb
);
3052 extern void drop_super_exclusive(struct super_block
*sb
);
3053 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
3054 extern void iterate_supers_type(struct file_system_type
*,
3055 void (*)(struct super_block
*, void *), void *);
3057 extern int dcache_dir_open(struct inode
*, struct file
*);
3058 extern int dcache_dir_close(struct inode
*, struct file
*);
3059 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
3060 extern int dcache_readdir(struct file
*, struct dir_context
*);
3061 extern int simple_setattr(struct dentry
*, struct iattr
*);
3062 extern int simple_getattr(const struct path
*, struct kstat
*, u32
, unsigned int);
3063 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
3064 extern int simple_open(struct inode
*inode
, struct file
*file
);
3065 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
3066 extern int simple_unlink(struct inode
*, struct dentry
*);
3067 extern int simple_rmdir(struct inode
*, struct dentry
*);
3068 extern int simple_rename(struct inode
*, struct dentry
*,
3069 struct inode
*, struct dentry
*, unsigned int);
3070 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
3071 extern int simple_empty(struct dentry
*);
3072 extern int simple_readpage(struct file
*file
, struct page
*page
);
3073 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
3074 loff_t pos
, unsigned len
, unsigned flags
,
3075 struct page
**pagep
, void **fsdata
);
3076 extern int simple_write_end(struct file
*file
, struct address_space
*mapping
,
3077 loff_t pos
, unsigned len
, unsigned copied
,
3078 struct page
*page
, void *fsdata
);
3079 extern int always_delete_dentry(const struct dentry
*);
3080 extern struct inode
*alloc_anon_inode(struct super_block
*);
3081 extern int simple_nosetlease(struct file
*, long, struct file_lock
**, void **);
3082 extern const struct dentry_operations simple_dentry_operations
;
3084 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
3085 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
3086 extern const struct file_operations simple_dir_operations
;
3087 extern const struct inode_operations simple_dir_inode_operations
;
3088 extern void make_empty_dir_inode(struct inode
*inode
);
3089 extern bool is_empty_dir_inode(struct inode
*inode
);
3090 struct tree_descr
{ const char *name
; const struct file_operations
*ops
; int mode
; };
3091 struct dentry
*d_alloc_name(struct dentry
*, const char *);
3092 extern int simple_fill_super(struct super_block
*, unsigned long,
3093 const struct tree_descr
*);
3094 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
3095 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
3097 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
3098 loff_t
*ppos
, const void *from
, size_t available
);
3099 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
3100 const void __user
*from
, size_t count
);
3102 extern int __generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3103 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3105 extern int generic_check_addressable(unsigned, u64
);
3107 #ifdef CONFIG_MIGRATION
3108 extern int buffer_migrate_page(struct address_space
*,
3109 struct page
*, struct page
*,
3112 #define buffer_migrate_page NULL
3115 extern int setattr_prepare(struct dentry
*, struct iattr
*);
3116 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
3117 extern void setattr_copy(struct inode
*inode
, const struct iattr
*attr
);
3119 extern int file_update_time(struct file
*file
);
3121 static inline bool io_is_direct(struct file
*filp
)
3123 return (filp
->f_flags
& O_DIRECT
) || IS_DAX(filp
->f_mapping
->host
);
3126 static inline bool vma_is_dax(struct vm_area_struct
*vma
)
3128 return vma
->vm_file
&& IS_DAX(vma
->vm_file
->f_mapping
->host
);
3131 static inline int iocb_flags(struct file
*file
)
3134 if (file
->f_flags
& O_APPEND
)
3136 if (io_is_direct(file
))
3138 if ((file
->f_flags
& O_DSYNC
) || IS_SYNC(file
->f_mapping
->host
))
3140 if (file
->f_flags
& __O_SYNC
)
3145 static inline int kiocb_set_rw_flags(struct kiocb
*ki
, int flags
)
3147 if (unlikely(flags
& ~RWF_SUPPORTED
))
3150 if (flags
& RWF_NOWAIT
) {
3151 if (!(ki
->ki_filp
->f_mode
& FMODE_AIO_NOWAIT
))
3153 ki
->ki_flags
|= IOCB_NOWAIT
;
3155 if (flags
& RWF_HIPRI
)
3156 ki
->ki_flags
|= IOCB_HIPRI
;
3157 if (flags
& RWF_DSYNC
)
3158 ki
->ki_flags
|= IOCB_DSYNC
;
3159 if (flags
& RWF_SYNC
)
3160 ki
->ki_flags
|= (IOCB_DSYNC
| IOCB_SYNC
);
3164 static inline ino_t
parent_ino(struct dentry
*dentry
)
3169 * Don't strictly need d_lock here? If the parent ino could change
3170 * then surely we'd have a deeper race in the caller?
3172 spin_lock(&dentry
->d_lock
);
3173 res
= dentry
->d_parent
->d_inode
->i_ino
;
3174 spin_unlock(&dentry
->d_lock
);
3178 /* Transaction based IO helpers */
3181 * An argresp is stored in an allocated page and holds the
3182 * size of the argument or response, along with its content
3184 struct simple_transaction_argresp
{
3189 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3191 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
3193 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
3194 size_t size
, loff_t
*pos
);
3195 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
3197 void simple_transaction_set(struct file
*file
, size_t n
);
3200 * simple attribute files
3202 * These attributes behave similar to those in sysfs:
3204 * Writing to an attribute immediately sets a value, an open file can be
3205 * written to multiple times.
3207 * Reading from an attribute creates a buffer from the value that might get
3208 * read with multiple read calls. When the attribute has been read
3209 * completely, no further read calls are possible until the file is opened
3212 * All attributes contain a text representation of a numeric value
3213 * that are accessed with the get() and set() functions.
3215 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3216 static int __fops ## _open(struct inode *inode, struct file *file) \
3218 __simple_attr_check_format(__fmt, 0ull); \
3219 return simple_attr_open(inode, file, __get, __set, __fmt); \
3221 static const struct file_operations __fops = { \
3222 .owner = THIS_MODULE, \
3223 .open = __fops ## _open, \
3224 .release = simple_attr_release, \
3225 .read = simple_attr_read, \
3226 .write = simple_attr_write, \
3227 .llseek = generic_file_llseek, \
3230 static inline __printf(1, 2)
3231 void __simple_attr_check_format(const char *fmt
, ...)
3233 /* don't do anything, just let the compiler check the arguments; */
3236 int simple_attr_open(struct inode
*inode
, struct file
*file
,
3237 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
3239 int simple_attr_release(struct inode
*inode
, struct file
*file
);
3240 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
3241 size_t len
, loff_t
*ppos
);
3242 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
3243 size_t len
, loff_t
*ppos
);
3246 int proc_nr_files(struct ctl_table
*table
, int write
,
3247 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3248 int proc_nr_dentry(struct ctl_table
*table
, int write
,
3249 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3250 int proc_nr_inodes(struct ctl_table
*table
, int write
,
3251 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3252 int __init
get_filesystem_list(char *buf
);
3254 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3255 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3257 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3258 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3259 (flag & __FMODE_NONOTIFY)))
3261 static inline bool is_sxid(umode_t mode
)
3263 return (mode
& S_ISUID
) || ((mode
& S_ISGID
) && (mode
& S_IXGRP
));
3266 static inline int check_sticky(struct inode
*dir
, struct inode
*inode
)
3268 if (!(dir
->i_mode
& S_ISVTX
))
3271 return __check_sticky(dir
, inode
);
3274 static inline void inode_has_no_xattr(struct inode
*inode
)
3276 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& MS_NOSEC
))
3277 inode
->i_flags
|= S_NOSEC
;
3280 static inline bool is_root_inode(struct inode
*inode
)
3282 return inode
== inode
->i_sb
->s_root
->d_inode
;
3285 static inline bool dir_emit(struct dir_context
*ctx
,
3286 const char *name
, int namelen
,
3287 u64 ino
, unsigned type
)
3289 return ctx
->actor(ctx
, name
, namelen
, ctx
->pos
, ino
, type
) == 0;
3291 static inline bool dir_emit_dot(struct file
*file
, struct dir_context
*ctx
)
3293 return ctx
->actor(ctx
, ".", 1, ctx
->pos
,
3294 file
->f_path
.dentry
->d_inode
->i_ino
, DT_DIR
) == 0;
3296 static inline bool dir_emit_dotdot(struct file
*file
, struct dir_context
*ctx
)
3298 return ctx
->actor(ctx
, "..", 2, ctx
->pos
,
3299 parent_ino(file
->f_path
.dentry
), DT_DIR
) == 0;
3301 static inline bool dir_emit_dots(struct file
*file
, struct dir_context
*ctx
)
3303 if (ctx
->pos
== 0) {
3304 if (!dir_emit_dot(file
, ctx
))
3308 if (ctx
->pos
== 1) {
3309 if (!dir_emit_dotdot(file
, ctx
))
3315 static inline bool dir_relax(struct inode
*inode
)
3317 inode_unlock(inode
);
3319 return !IS_DEADDIR(inode
);
3322 static inline bool dir_relax_shared(struct inode
*inode
)
3324 inode_unlock_shared(inode
);
3325 inode_lock_shared(inode
);
3326 return !IS_DEADDIR(inode
);
3329 extern bool path_noexec(const struct path
*path
);
3330 extern void inode_nohighmem(struct inode
*inode
);
3332 #endif /* _LINUX_FS_H */