1 /* SPDX-License-Identifier: GPL-2.0 */
5 #include <linux/linkage.h>
6 #include <linux/wait_bit.h>
7 #include <linux/kdev_t.h>
8 #include <linux/dcache.h>
9 #include <linux/path.h>
10 #include <linux/stat.h>
11 #include <linux/cache.h>
12 #include <linux/list.h>
13 #include <linux/list_lru.h>
14 #include <linux/llist.h>
15 #include <linux/radix-tree.h>
16 #include <linux/rbtree.h>
17 #include <linux/init.h>
18 #include <linux/pid.h>
19 #include <linux/bug.h>
20 #include <linux/mutex.h>
21 #include <linux/rwsem.h>
22 #include <linux/mm_types.h>
23 #include <linux/capability.h>
24 #include <linux/semaphore.h>
25 #include <linux/fcntl.h>
26 #include <linux/fiemap.h>
27 #include <linux/rculist_bl.h>
28 #include <linux/atomic.h>
29 #include <linux/shrinker.h>
30 #include <linux/migrate_mode.h>
31 #include <linux/uidgid.h>
32 #include <linux/lockdep.h>
33 #include <linux/percpu-rwsem.h>
34 #include <linux/workqueue.h>
35 #include <linux/delayed_call.h>
36 #include <linux/uuid.h>
37 #include <linux/errseq.h>
39 #include <asm/byteorder.h>
40 #include <uapi/linux/fs.h>
42 struct backing_dev_info
;
45 struct export_operations
;
50 struct pipe_inode_info
;
51 struct poll_table_struct
;
53 struct vm_area_struct
;
56 struct swap_info_struct
;
58 struct workqueue_struct
;
61 struct fscrypt_operations
;
63 extern void __init
inode_init(void);
64 extern void __init
inode_init_early(void);
65 extern void __init
files_init(void);
66 extern void __init
files_maxfiles_init(void);
68 extern struct files_stat_struct files_stat
;
69 extern unsigned long get_max_files(void);
70 extern unsigned int sysctl_nr_open
;
71 extern struct inodes_stat_t inodes_stat
;
72 extern int leases_enable
, lease_break_time
;
73 extern int sysctl_protected_symlinks
;
74 extern int sysctl_protected_hardlinks
;
75 extern int sysctl_protected_fifos
;
76 extern int sysctl_protected_regular
;
78 typedef __kernel_rwf_t rwf_t
;
81 typedef int (get_block_t
)(struct inode
*inode
, sector_t iblock
,
82 struct buffer_head
*bh_result
, int create
);
83 typedef int (dio_iodone_t
)(struct kiocb
*iocb
, loff_t offset
,
84 ssize_t bytes
, void *private);
86 #define MAY_EXEC 0x00000001
87 #define MAY_WRITE 0x00000002
88 #define MAY_READ 0x00000004
89 #define MAY_APPEND 0x00000008
90 #define MAY_ACCESS 0x00000010
91 #define MAY_OPEN 0x00000020
92 #define MAY_CHDIR 0x00000040
93 /* called from RCU mode, don't block */
94 #define MAY_NOT_BLOCK 0x00000080
97 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
98 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
101 /* file is open for reading */
102 #define FMODE_READ ((__force fmode_t)0x1)
103 /* file is open for writing */
104 #define FMODE_WRITE ((__force fmode_t)0x2)
105 /* file is seekable */
106 #define FMODE_LSEEK ((__force fmode_t)0x4)
107 /* file can be accessed using pread */
108 #define FMODE_PREAD ((__force fmode_t)0x8)
109 /* file can be accessed using pwrite */
110 #define FMODE_PWRITE ((__force fmode_t)0x10)
111 /* File is opened for execution with sys_execve / sys_uselib */
112 #define FMODE_EXEC ((__force fmode_t)0x20)
113 /* File is opened with O_NDELAY (only set for block devices) */
114 #define FMODE_NDELAY ((__force fmode_t)0x40)
115 /* File is opened with O_EXCL (only set for block devices) */
116 #define FMODE_EXCL ((__force fmode_t)0x80)
117 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
118 (specialy hack for floppy.c) */
119 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
120 /* 32bit hashes as llseek() offset (for directories) */
121 #define FMODE_32BITHASH ((__force fmode_t)0x200)
122 /* 64bit hashes as llseek() offset (for directories) */
123 #define FMODE_64BITHASH ((__force fmode_t)0x400)
126 * Don't update ctime and mtime.
128 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
129 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
131 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
133 /* Expect random access pattern */
134 #define FMODE_RANDOM ((__force fmode_t)0x1000)
136 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
137 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
139 /* File is opened with O_PATH; almost nothing can be done with it */
140 #define FMODE_PATH ((__force fmode_t)0x4000)
142 /* File needs atomic accesses to f_pos */
143 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
144 /* Write access to underlying fs */
145 #define FMODE_WRITER ((__force fmode_t)0x10000)
146 /* Has read method(s) */
147 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
148 /* Has write method(s) */
149 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
151 /* File is stream-like */
152 #define FMODE_STREAM ((__force fmode_t)0x200000)
154 /* File was opened by fanotify and shouldn't generate fanotify events */
155 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
157 /* File is capable of returning -EAGAIN if I/O will block */
158 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
161 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
162 * that indicates that they should check the contents of the iovec are
163 * valid, but not check the memory that the iovec elements
166 #define CHECK_IOVEC_ONLY -1
169 * Attribute flags. These should be or-ed together to figure out what
172 #define ATTR_MODE (1 << 0)
173 #define ATTR_UID (1 << 1)
174 #define ATTR_GID (1 << 2)
175 #define ATTR_SIZE (1 << 3)
176 #define ATTR_ATIME (1 << 4)
177 #define ATTR_MTIME (1 << 5)
178 #define ATTR_CTIME (1 << 6)
179 #define ATTR_ATIME_SET (1 << 7)
180 #define ATTR_MTIME_SET (1 << 8)
181 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
182 #define ATTR_ATTR_FLAG (1 << 10)
183 #define ATTR_KILL_SUID (1 << 11)
184 #define ATTR_KILL_SGID (1 << 12)
185 #define ATTR_FILE (1 << 13)
186 #define ATTR_KILL_PRIV (1 << 14)
187 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
188 #define ATTR_TIMES_SET (1 << 16)
189 #define ATTR_TOUCH (1 << 17)
192 * Whiteout is represented by a char device. The following constants define the
193 * mode and device number to use.
195 #define WHITEOUT_MODE 0
196 #define WHITEOUT_DEV 0
199 * This is the Inode Attributes structure, used for notify_change(). It
200 * uses the above definitions as flags, to know which values have changed.
201 * Also, in this manner, a Filesystem can look at only the values it cares
202 * about. Basically, these are the attributes that the VFS layer can
203 * request to change from the FS layer.
205 * Derek Atkins <warlord@MIT.EDU> 94-10-20
208 unsigned int ia_valid
;
213 struct timespec ia_atime
;
214 struct timespec ia_mtime
;
215 struct timespec ia_ctime
;
218 * Not an attribute, but an auxiliary info for filesystems wanting to
219 * implement an ftruncate() like method. NOTE: filesystem should
220 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
222 struct file
*ia_file
;
226 * Includes for diskquotas.
228 #include <linux/quota.h>
231 * Maximum number of layers of fs stack. Needs to be limited to
232 * prevent kernel stack overflow
234 #define FILESYSTEM_MAX_STACK_DEPTH 2
237 * enum positive_aop_returns - aop return codes with specific semantics
239 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
240 * completed, that the page is still locked, and
241 * should be considered active. The VM uses this hint
242 * to return the page to the active list -- it won't
243 * be a candidate for writeback again in the near
244 * future. Other callers must be careful to unlock
245 * the page if they get this return. Returned by
248 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
249 * unlocked it and the page might have been truncated.
250 * The caller should back up to acquiring a new page and
251 * trying again. The aop will be taking reasonable
252 * precautions not to livelock. If the caller held a page
253 * reference, it should drop it before retrying. Returned
256 * address_space_operation functions return these large constants to indicate
257 * special semantics to the caller. These are much larger than the bytes in a
258 * page to allow for functions that return the number of bytes operated on in a
262 enum positive_aop_returns
{
263 AOP_WRITEPAGE_ACTIVATE
= 0x80000,
264 AOP_TRUNCATED_PAGE
= 0x80001,
267 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
268 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
269 * helper code (eg buffer layer)
270 * to clear GFP_FS from alloc */
273 * oh the beauties of C type declarations.
276 struct address_space
;
277 struct writeback_control
;
280 * Write life time hint values.
283 WRITE_LIFE_NOT_SET
= 0,
284 WRITE_LIFE_NONE
= RWH_WRITE_LIFE_NONE
,
285 WRITE_LIFE_SHORT
= RWH_WRITE_LIFE_SHORT
,
286 WRITE_LIFE_MEDIUM
= RWH_WRITE_LIFE_MEDIUM
,
287 WRITE_LIFE_LONG
= RWH_WRITE_LIFE_LONG
,
288 WRITE_LIFE_EXTREME
= RWH_WRITE_LIFE_EXTREME
,
291 #define IOCB_EVENTFD (1 << 0)
292 #define IOCB_APPEND (1 << 1)
293 #define IOCB_DIRECT (1 << 2)
294 #define IOCB_HIPRI (1 << 3)
295 #define IOCB_DSYNC (1 << 4)
296 #define IOCB_SYNC (1 << 5)
297 #define IOCB_WRITE (1 << 6)
298 #define IOCB_NOWAIT (1 << 7)
301 struct file
*ki_filp
;
303 void (*ki_complete
)(struct kiocb
*iocb
, long ret
, long ret2
);
306 enum rw_hint ki_hint
;
307 } __randomize_layout
;
309 static inline bool is_sync_kiocb(struct kiocb
*kiocb
)
311 return kiocb
->ki_complete
== NULL
;
315 * "descriptor" for what we're up to with a read.
316 * This allows us to use the same read code yet
317 * have multiple different users of the data that
318 * we read from a file.
320 * The simplest case just copies the data to user
333 typedef int (*read_actor_t
)(read_descriptor_t
*, struct page
*,
334 unsigned long, unsigned long);
336 struct address_space_operations
{
337 int (*writepage
)(struct page
*page
, struct writeback_control
*wbc
);
338 int (*readpage
)(struct file
*, struct page
*);
340 /* Write back some dirty pages from this mapping. */
341 int (*writepages
)(struct address_space
*, struct writeback_control
*);
343 /* Set a page dirty. Return true if this dirtied it */
344 int (*set_page_dirty
)(struct page
*page
);
346 int (*readpages
)(struct file
*filp
, struct address_space
*mapping
,
347 struct list_head
*pages
, unsigned nr_pages
);
349 int (*write_begin
)(struct file
*, struct address_space
*mapping
,
350 loff_t pos
, unsigned len
, unsigned flags
,
351 struct page
**pagep
, void **fsdata
);
352 int (*write_end
)(struct file
*, struct address_space
*mapping
,
353 loff_t pos
, unsigned len
, unsigned copied
,
354 struct page
*page
, void *fsdata
);
356 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
357 sector_t (*bmap
)(struct address_space
*, sector_t
);
358 void (*invalidatepage
) (struct page
*, unsigned int, unsigned int);
359 int (*releasepage
) (struct page
*, gfp_t
);
360 void (*freepage
)(struct page
*);
361 ssize_t (*direct_IO
)(struct kiocb
*, struct iov_iter
*iter
);
363 * migrate the contents of a page to the specified target. If
364 * migrate_mode is MIGRATE_ASYNC, it must not block.
366 int (*migratepage
) (struct address_space
*,
367 struct page
*, struct page
*, enum migrate_mode
);
368 bool (*isolate_page
)(struct page
*, isolate_mode_t
);
369 void (*putback_page
)(struct page
*);
370 int (*launder_page
) (struct page
*);
371 int (*is_partially_uptodate
) (struct page
*, unsigned long,
373 void (*is_dirty_writeback
) (struct page
*, bool *, bool *);
374 int (*error_remove_page
)(struct address_space
*, struct page
*);
376 /* swapfile support */
377 int (*swap_activate
)(struct swap_info_struct
*sis
, struct file
*file
,
379 void (*swap_deactivate
)(struct file
*file
);
382 extern const struct address_space_operations empty_aops
;
385 * pagecache_write_begin/pagecache_write_end must be used by general code
386 * to write into the pagecache.
388 int pagecache_write_begin(struct file
*, struct address_space
*mapping
,
389 loff_t pos
, unsigned len
, unsigned flags
,
390 struct page
**pagep
, void **fsdata
);
392 int pagecache_write_end(struct file
*, struct address_space
*mapping
,
393 loff_t pos
, unsigned len
, unsigned copied
,
394 struct page
*page
, void *fsdata
);
396 struct address_space
{
397 struct inode
*host
; /* owner: inode, block_device */
398 struct radix_tree_root page_tree
; /* radix tree of all pages */
399 spinlock_t tree_lock
; /* and lock protecting it */
400 atomic_t i_mmap_writable
;/* count VM_SHARED mappings */
401 struct rb_root_cached i_mmap
; /* tree of private and shared mappings */
402 struct rw_semaphore i_mmap_rwsem
; /* protect tree, count, list */
403 /* Protected by tree_lock together with the radix tree */
404 unsigned long nrpages
; /* number of total pages */
405 /* number of shadow or DAX exceptional entries */
406 unsigned long nrexceptional
;
407 pgoff_t writeback_index
;/* writeback starts here */
408 const struct address_space_operations
*a_ops
; /* methods */
409 unsigned long flags
; /* error bits */
410 spinlock_t private_lock
; /* for use by the address_space */
411 gfp_t gfp_mask
; /* implicit gfp mask for allocations */
412 struct list_head private_list
; /* for use by the address_space */
413 void *private_data
; /* ditto */
415 } __attribute__((aligned(sizeof(long)))) __randomize_layout
;
417 * On most architectures that alignment is already the case; but
418 * must be enforced here for CRIS, to let the least significant bit
419 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
421 struct request_queue
;
423 struct block_device
{
424 dev_t bd_dev
; /* not a kdev_t - it's a search key */
426 struct inode
* bd_inode
; /* will die */
427 struct super_block
* bd_super
;
428 struct mutex bd_mutex
; /* open/close mutex */
432 bool bd_write_holder
;
434 struct list_head bd_holder_disks
;
436 struct block_device
* bd_contains
;
437 unsigned bd_block_size
;
439 struct hd_struct
* bd_part
;
440 /* number of times partitions within this device have been opened. */
441 unsigned bd_part_count
;
443 struct gendisk
* bd_disk
;
444 struct request_queue
* bd_queue
;
445 struct backing_dev_info
*bd_bdi
;
446 struct list_head bd_list
;
448 * Private data. You must have bd_claim'ed the block_device
449 * to use this. NOTE: bd_claim allows an owner to claim
450 * the same device multiple times, the owner must take special
451 * care to not mess up bd_private for that case.
453 unsigned long bd_private
;
455 /* The counter of freeze processes */
456 int bd_fsfreeze_count
;
457 /* Mutex for freeze */
458 struct mutex bd_fsfreeze_mutex
;
459 } __randomize_layout
;
462 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
465 #define PAGECACHE_TAG_DIRTY 0
466 #define PAGECACHE_TAG_WRITEBACK 1
467 #define PAGECACHE_TAG_TOWRITE 2
469 int mapping_tagged(struct address_space
*mapping
, int tag
);
471 static inline void i_mmap_lock_write(struct address_space
*mapping
)
473 down_write(&mapping
->i_mmap_rwsem
);
476 static inline void i_mmap_unlock_write(struct address_space
*mapping
)
478 up_write(&mapping
->i_mmap_rwsem
);
481 static inline void i_mmap_lock_read(struct address_space
*mapping
)
483 down_read(&mapping
->i_mmap_rwsem
);
486 static inline void i_mmap_unlock_read(struct address_space
*mapping
)
488 up_read(&mapping
->i_mmap_rwsem
);
492 * Might pages of this file be mapped into userspace?
494 static inline int mapping_mapped(struct address_space
*mapping
)
496 return !RB_EMPTY_ROOT(&mapping
->i_mmap
.rb_root
);
500 * Might pages of this file have been modified in userspace?
501 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
502 * marks vma as VM_SHARED if it is shared, and the file was opened for
503 * writing i.e. vma may be mprotected writable even if now readonly.
505 * If i_mmap_writable is negative, no new writable mappings are allowed. You
506 * can only deny writable mappings, if none exists right now.
508 static inline int mapping_writably_mapped(struct address_space
*mapping
)
510 return atomic_read(&mapping
->i_mmap_writable
) > 0;
513 static inline int mapping_map_writable(struct address_space
*mapping
)
515 return atomic_inc_unless_negative(&mapping
->i_mmap_writable
) ?
519 static inline void mapping_unmap_writable(struct address_space
*mapping
)
521 atomic_dec(&mapping
->i_mmap_writable
);
524 static inline int mapping_deny_writable(struct address_space
*mapping
)
526 return atomic_dec_unless_positive(&mapping
->i_mmap_writable
) ?
530 static inline void mapping_allow_writable(struct address_space
*mapping
)
532 atomic_inc(&mapping
->i_mmap_writable
);
536 * Use sequence counter to get consistent i_size on 32-bit processors.
538 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
539 #include <linux/seqlock.h>
540 #define __NEED_I_SIZE_ORDERED
541 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
543 #define i_size_ordered_init(inode) do { } while (0)
547 #define ACL_NOT_CACHED ((void *)(-1))
548 #define ACL_DONT_CACHE ((void *)(-3))
550 static inline struct posix_acl
*
551 uncached_acl_sentinel(struct task_struct
*task
)
553 return (void *)task
+ 1;
557 is_uncached_acl(struct posix_acl
*acl
)
559 return (long)acl
& 1;
562 #define IOP_FASTPERM 0x0001
563 #define IOP_LOOKUP 0x0002
564 #define IOP_NOFOLLOW 0x0004
565 #define IOP_XATTR 0x0008
566 #define IOP_DEFAULT_READLINK 0x0010
568 struct fsnotify_mark_connector
;
571 * Keep mostly read-only and often accessed (especially for
572 * the RCU path lookup and 'stat' data) fields at the beginning
573 * of the 'struct inode'
577 unsigned short i_opflags
;
580 unsigned int i_flags
;
582 #ifdef CONFIG_FS_POSIX_ACL
583 struct posix_acl
*i_acl
;
584 struct posix_acl
*i_default_acl
;
587 const struct inode_operations
*i_op
;
588 struct super_block
*i_sb
;
589 struct address_space
*i_mapping
;
591 #ifdef CONFIG_SECURITY
595 /* Stat data, not accessed from path walking */
598 * Filesystems may only read i_nlink directly. They shall use the
599 * following functions for modification:
601 * (set|clear|inc|drop)_nlink
602 * inode_(inc|dec)_link_count
605 const unsigned int i_nlink
;
606 unsigned int __i_nlink
;
610 struct timespec i_atime
;
611 struct timespec i_mtime
;
612 struct timespec i_ctime
;
613 spinlock_t i_lock
; /* i_blocks, i_bytes, maybe i_size */
614 unsigned short i_bytes
;
615 unsigned int i_blkbits
;
616 enum rw_hint i_write_hint
;
619 #ifdef __NEED_I_SIZE_ORDERED
620 seqcount_t i_size_seqcount
;
624 unsigned long i_state
;
625 struct rw_semaphore i_rwsem
;
627 unsigned long dirtied_when
; /* jiffies of first dirtying */
628 unsigned long dirtied_time_when
;
630 struct hlist_node i_hash
;
631 struct list_head i_io_list
; /* backing dev IO list */
632 #ifdef CONFIG_CGROUP_WRITEBACK
633 struct bdi_writeback
*i_wb
; /* the associated cgroup wb */
635 /* foreign inode detection, see wbc_detach_inode() */
637 u16 i_wb_frn_avg_time
;
638 u16 i_wb_frn_history
;
640 struct list_head i_lru
; /* inode LRU list */
641 struct list_head i_sb_list
;
642 struct list_head i_wb_list
; /* backing dev writeback list */
644 struct hlist_head i_dentry
;
645 struct rcu_head i_rcu
;
649 atomic_t i_dio_count
;
650 atomic_t i_writecount
;
652 atomic_t i_readcount
; /* struct files open RO */
654 const struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
655 struct file_lock_context
*i_flctx
;
656 struct address_space i_data
;
657 struct list_head i_devices
;
659 struct pipe_inode_info
*i_pipe
;
660 struct block_device
*i_bdev
;
668 #ifdef CONFIG_FSNOTIFY
669 __u32 i_fsnotify_mask
; /* all events this inode cares about */
670 struct fsnotify_mark_connector __rcu
*i_fsnotify_marks
;
673 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
674 struct fscrypt_info
*i_crypt_info
;
677 void *i_private
; /* fs or device private pointer */
678 } __randomize_layout
;
680 static inline unsigned int i_blocksize(const struct inode
*node
)
682 return (1 << node
->i_blkbits
);
685 static inline int inode_unhashed(struct inode
*inode
)
687 return hlist_unhashed(&inode
->i_hash
);
691 * inode->i_mutex nesting subclasses for the lock validator:
693 * 0: the object of the current VFS operation
697 * 4: second non-directory
698 * 5: second parent (when locking independent directories in rename)
700 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
701 * non-directories at once.
703 * The locking order between these classes is
704 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
706 enum inode_i_mutex_lock_class
716 static inline void inode_lock(struct inode
*inode
)
718 down_write(&inode
->i_rwsem
);
721 static inline void inode_unlock(struct inode
*inode
)
723 up_write(&inode
->i_rwsem
);
726 static inline void inode_lock_shared(struct inode
*inode
)
728 down_read(&inode
->i_rwsem
);
731 static inline void inode_unlock_shared(struct inode
*inode
)
733 up_read(&inode
->i_rwsem
);
736 static inline int inode_trylock(struct inode
*inode
)
738 return down_write_trylock(&inode
->i_rwsem
);
741 static inline int inode_trylock_shared(struct inode
*inode
)
743 return down_read_trylock(&inode
->i_rwsem
);
746 static inline int inode_is_locked(struct inode
*inode
)
748 return rwsem_is_locked(&inode
->i_rwsem
);
751 static inline void inode_lock_nested(struct inode
*inode
, unsigned subclass
)
753 down_write_nested(&inode
->i_rwsem
, subclass
);
756 void lock_two_nondirectories(struct inode
*, struct inode
*);
757 void unlock_two_nondirectories(struct inode
*, struct inode
*);
760 * NOTE: in a 32bit arch with a preemptable kernel and
761 * an UP compile the i_size_read/write must be atomic
762 * with respect to the local cpu (unlike with preempt disabled),
763 * but they don't need to be atomic with respect to other cpus like in
764 * true SMP (so they need either to either locally disable irq around
765 * the read or for example on x86 they can be still implemented as a
766 * cmpxchg8b without the need of the lock prefix). For SMP compiles
767 * and 64bit archs it makes no difference if preempt is enabled or not.
769 static inline loff_t
i_size_read(const struct inode
*inode
)
771 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
776 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
777 i_size
= inode
->i_size
;
778 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
780 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
784 i_size
= inode
->i_size
;
788 return inode
->i_size
;
793 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
794 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
795 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
797 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
799 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
801 write_seqcount_begin(&inode
->i_size_seqcount
);
802 inode
->i_size
= i_size
;
803 write_seqcount_end(&inode
->i_size_seqcount
);
805 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
807 inode
->i_size
= i_size
;
810 inode
->i_size
= i_size
;
814 static inline unsigned iminor(const struct inode
*inode
)
816 return MINOR(inode
->i_rdev
);
819 static inline unsigned imajor(const struct inode
*inode
)
821 return MAJOR(inode
->i_rdev
);
824 extern struct block_device
*I_BDEV(struct inode
*inode
);
827 rwlock_t lock
; /* protects pid, uid, euid fields */
828 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
829 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
830 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
831 int signum
; /* posix.1b rt signal to be delivered on IO */
835 * Track a single file's readahead state
837 struct file_ra_state
{
838 pgoff_t start
; /* where readahead started */
839 unsigned int size
; /* # of readahead pages */
840 unsigned int async_size
; /* do asynchronous readahead when
841 there are only # of pages ahead */
843 unsigned int ra_pages
; /* Maximum readahead window */
844 unsigned int mmap_miss
; /* Cache miss stat for mmap accesses */
845 loff_t prev_pos
; /* Cache last read() position */
849 * Check if @index falls in the readahead windows.
851 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
853 return (index
>= ra
->start
&&
854 index
< ra
->start
+ ra
->size
);
859 struct llist_node fu_llist
;
860 struct rcu_head fu_rcuhead
;
863 struct inode
*f_inode
; /* cached value */
864 const struct file_operations
*f_op
;
867 * Protects f_ep_links, f_flags.
868 * Must not be taken from IRQ context.
871 enum rw_hint f_write_hint
;
872 atomic_long_t f_count
;
873 unsigned int f_flags
;
875 struct mutex f_pos_lock
;
877 struct fown_struct f_owner
;
878 const struct cred
*f_cred
;
879 struct file_ra_state f_ra
;
882 #ifdef CONFIG_SECURITY
885 /* needed for tty driver, and maybe others */
889 /* Used by fs/eventpoll.c to link all the hooks to this file */
890 struct list_head f_ep_links
;
891 struct list_head f_tfile_llink
;
892 #endif /* #ifdef CONFIG_EPOLL */
893 struct address_space
*f_mapping
;
896 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
901 /* file identifier */
902 unsigned char f_handle
[0];
905 static inline struct file
*get_file(struct file
*f
)
907 atomic_long_inc(&f
->f_count
);
910 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
911 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
912 #define file_count(x) atomic_long_read(&(x)->f_count)
914 #define MAX_NON_LFS ((1UL<<31) - 1)
916 /* Page cache limit. The filesystems should put that into their s_maxbytes
917 limits, otherwise bad things can happen in VM. */
918 #if BITS_PER_LONG==32
919 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
920 #elif BITS_PER_LONG==64
921 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
926 #define FL_DELEG 4 /* NFSv4 delegation */
927 #define FL_ACCESS 8 /* not trying to lock, just looking */
928 #define FL_EXISTS 16 /* when unlocking, test for existence */
929 #define FL_LEASE 32 /* lease held on this file */
930 #define FL_CLOSE 64 /* unlock on close */
931 #define FL_SLEEP 128 /* A blocking lock */
932 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
933 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
934 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
935 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
937 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
940 * Special return value from posix_lock_file() and vfs_lock_file() for
941 * asynchronous locking.
943 #define FILE_LOCK_DEFERRED 1
945 /* legacy typedef, should eventually be removed */
946 typedef void *fl_owner_t
;
950 struct file_lock_operations
{
951 void (*fl_copy_lock
)(struct file_lock
*, struct file_lock
*);
952 void (*fl_release_private
)(struct file_lock
*);
955 struct lock_manager_operations
{
956 int (*lm_compare_owner
)(struct file_lock
*, struct file_lock
*);
957 unsigned long (*lm_owner_key
)(struct file_lock
*);
958 fl_owner_t (*lm_get_owner
)(fl_owner_t
);
959 void (*lm_put_owner
)(fl_owner_t
);
960 void (*lm_notify
)(struct file_lock
*); /* unblock callback */
961 int (*lm_grant
)(struct file_lock
*, int);
962 bool (*lm_break
)(struct file_lock
*);
963 int (*lm_change
)(struct file_lock
*, int, struct list_head
*);
964 void (*lm_setup
)(struct file_lock
*, void **);
967 struct lock_manager
{
968 struct list_head list
;
970 * NFSv4 and up also want opens blocked during the grace period;
977 void locks_start_grace(struct net
*, struct lock_manager
*);
978 void locks_end_grace(struct lock_manager
*);
979 bool locks_in_grace(struct net
*);
980 bool opens_in_grace(struct net
*);
982 /* that will die - we need it for nfs_lock_info */
983 #include <linux/nfs_fs_i.h>
986 * struct file_lock represents a generic "file lock". It's used to represent
987 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
988 * note that the same struct is used to represent both a request for a lock and
989 * the lock itself, but the same object is never used for both.
991 * FIXME: should we create a separate "struct lock_request" to help distinguish
994 * The varous i_flctx lists are ordered by:
997 * 2) lock range start
1000 * Obviously, the last two criteria only matter for POSIX locks.
1003 struct file_lock
*fl_next
; /* singly linked list for this inode */
1004 struct list_head fl_list
; /* link into file_lock_context */
1005 struct hlist_node fl_link
; /* node in global lists */
1006 struct list_head fl_block
; /* circular list of blocked processes */
1007 fl_owner_t fl_owner
;
1008 unsigned int fl_flags
;
1009 unsigned char fl_type
;
1010 unsigned int fl_pid
;
1011 int fl_link_cpu
; /* what cpu's list is this on? */
1012 wait_queue_head_t fl_wait
;
1013 struct file
*fl_file
;
1017 struct fasync_struct
* fl_fasync
; /* for lease break notifications */
1018 /* for lease breaks: */
1019 unsigned long fl_break_time
;
1020 unsigned long fl_downgrade_time
;
1022 const struct file_lock_operations
*fl_ops
; /* Callbacks for filesystems */
1023 const struct lock_manager_operations
*fl_lmops
; /* Callbacks for lockmanagers */
1025 struct nfs_lock_info nfs_fl
;
1026 struct nfs4_lock_info nfs4_fl
;
1028 struct list_head link
; /* link in AFS vnode's pending_locks list */
1029 int state
; /* state of grant or error if -ve */
1032 } __randomize_layout
;
1034 struct file_lock_context
{
1035 spinlock_t flc_lock
;
1036 struct list_head flc_flock
;
1037 struct list_head flc_posix
;
1038 struct list_head flc_lease
;
1041 /* The following constant reflects the upper bound of the file/locking space */
1043 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1044 #define OFFSET_MAX INT_LIMIT(loff_t)
1045 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1048 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
1051 * Return the inode to use for locking
1053 * For overlayfs this should be the overlay inode, not the real inode returned
1054 * by file_inode(). For any other fs file_inode(filp) and locks_inode(filp) are
1057 static inline struct inode
*locks_inode(const struct file
*f
)
1059 return f
->f_path
.dentry
->d_inode
;
1062 #ifdef CONFIG_FILE_LOCKING
1063 extern int fcntl_getlk(struct file
*, unsigned int, struct flock
*);
1064 extern int fcntl_setlk(unsigned int, struct file
*, unsigned int,
1067 #if BITS_PER_LONG == 32
1068 extern int fcntl_getlk64(struct file
*, unsigned int, struct flock64
*);
1069 extern int fcntl_setlk64(unsigned int, struct file
*, unsigned int,
1073 extern int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
);
1074 extern int fcntl_getlease(struct file
*filp
);
1077 void locks_free_lock_context(struct inode
*inode
);
1078 void locks_free_lock(struct file_lock
*fl
);
1079 extern void locks_init_lock(struct file_lock
*);
1080 extern struct file_lock
* locks_alloc_lock(void);
1081 extern void locks_copy_lock(struct file_lock
*, struct file_lock
*);
1082 extern void locks_copy_conflock(struct file_lock
*, struct file_lock
*);
1083 extern void locks_remove_posix(struct file
*, fl_owner_t
);
1084 extern void locks_remove_file(struct file
*);
1085 extern void locks_release_private(struct file_lock
*);
1086 extern void posix_test_lock(struct file
*, struct file_lock
*);
1087 extern int posix_lock_file(struct file
*, struct file_lock
*, struct file_lock
*);
1088 extern int posix_unblock_lock(struct file_lock
*);
1089 extern int vfs_test_lock(struct file
*, struct file_lock
*);
1090 extern int vfs_lock_file(struct file
*, unsigned int, struct file_lock
*, struct file_lock
*);
1091 extern int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
);
1092 extern int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
);
1093 extern int __break_lease(struct inode
*inode
, unsigned int flags
, unsigned int type
);
1094 extern void lease_get_mtime(struct inode
*, struct timespec
*time
);
1095 extern int generic_setlease(struct file
*, long, struct file_lock
**, void **priv
);
1096 extern int vfs_setlease(struct file
*, long, struct file_lock
**, void **);
1097 extern int lease_modify(struct file_lock
*, int, struct list_head
*);
1098 struct files_struct
;
1099 extern void show_fd_locks(struct seq_file
*f
,
1100 struct file
*filp
, struct files_struct
*files
);
1101 #else /* !CONFIG_FILE_LOCKING */
1102 static inline int fcntl_getlk(struct file
*file
, unsigned int cmd
,
1103 struct flock __user
*user
)
1108 static inline int fcntl_setlk(unsigned int fd
, struct file
*file
,
1109 unsigned int cmd
, struct flock __user
*user
)
1114 #if BITS_PER_LONG == 32
1115 static inline int fcntl_getlk64(struct file
*file
, unsigned int cmd
,
1116 struct flock64 __user
*user
)
1121 static inline int fcntl_setlk64(unsigned int fd
, struct file
*file
,
1122 unsigned int cmd
, struct flock64 __user
*user
)
1127 static inline int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
)
1132 static inline int fcntl_getlease(struct file
*filp
)
1138 locks_free_lock_context(struct inode
*inode
)
1142 static inline void locks_init_lock(struct file_lock
*fl
)
1147 static inline void locks_copy_conflock(struct file_lock
*new, struct file_lock
*fl
)
1152 static inline void locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
1157 static inline void locks_remove_posix(struct file
*filp
, fl_owner_t owner
)
1162 static inline void locks_remove_file(struct file
*filp
)
1167 static inline void posix_test_lock(struct file
*filp
, struct file_lock
*fl
)
1172 static inline int posix_lock_file(struct file
*filp
, struct file_lock
*fl
,
1173 struct file_lock
*conflock
)
1178 static inline int posix_unblock_lock(struct file_lock
*waiter
)
1183 static inline int vfs_test_lock(struct file
*filp
, struct file_lock
*fl
)
1188 static inline int vfs_lock_file(struct file
*filp
, unsigned int cmd
,
1189 struct file_lock
*fl
, struct file_lock
*conf
)
1194 static inline int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
)
1199 static inline int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
)
1204 static inline int __break_lease(struct inode
*inode
, unsigned int mode
, unsigned int type
)
1209 static inline void lease_get_mtime(struct inode
*inode
, struct timespec
*time
)
1214 static inline int generic_setlease(struct file
*filp
, long arg
,
1215 struct file_lock
**flp
, void **priv
)
1220 static inline int vfs_setlease(struct file
*filp
, long arg
,
1221 struct file_lock
**lease
, void **priv
)
1226 static inline int lease_modify(struct file_lock
*fl
, int arg
,
1227 struct list_head
*dispose
)
1232 struct files_struct
;
1233 static inline void show_fd_locks(struct seq_file
*f
,
1234 struct file
*filp
, struct files_struct
*files
) {}
1235 #endif /* !CONFIG_FILE_LOCKING */
1237 static inline struct inode
*file_inode(const struct file
*f
)
1242 static inline struct dentry
*file_dentry(const struct file
*file
)
1244 return d_real(file
->f_path
.dentry
, file_inode(file
), 0, 0);
1247 static inline int locks_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1249 return locks_lock_inode_wait(locks_inode(filp
), fl
);
1252 struct fasync_struct
{
1256 struct fasync_struct
*fa_next
; /* singly linked list */
1257 struct file
*fa_file
;
1258 struct rcu_head fa_rcu
;
1261 #define FASYNC_MAGIC 0x4601
1263 /* SMP safe fasync helpers: */
1264 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1265 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1266 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1267 extern struct fasync_struct
*fasync_alloc(void);
1268 extern void fasync_free(struct fasync_struct
*);
1270 /* can be called from interrupts */
1271 extern void kill_fasync(struct fasync_struct
**, int, int);
1273 extern int setfl(int fd
, struct file
* filp
, unsigned long arg
);
1274 extern void __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1275 extern int f_setown(struct file
*filp
, unsigned long arg
, int force
);
1276 extern void f_delown(struct file
*filp
);
1277 extern pid_t
f_getown(struct file
*filp
);
1278 extern int send_sigurg(struct fown_struct
*fown
);
1281 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1282 * represented in both.
1284 #define SB_RDONLY 1 /* Mount read-only */
1285 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1286 #define SB_NODEV 4 /* Disallow access to device special files */
1287 #define SB_NOEXEC 8 /* Disallow program execution */
1288 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1289 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1290 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1291 #define SB_NOATIME 1024 /* Do not update access times. */
1292 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1293 #define SB_SILENT 32768
1294 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1295 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1296 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1297 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1299 /* These sb flags are internal to the kernel */
1300 #define SB_SUBMOUNT (1<<26)
1301 #define SB_NOREMOTELOCK (1<<27)
1302 #define SB_NOSEC (1<<28)
1303 #define SB_BORN (1<<29)
1304 #define SB_ACTIVE (1<<30)
1305 #define SB_NOUSER (1<<31)
1311 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1312 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1313 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1314 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1315 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1318 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1319 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1320 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1321 #define SB_I_NOSUID 0x00000008 /* Ignore suid on this fs */
1322 #define SB_I_MULTIROOT 0x00000010 /* Multiple roots to the dentry tree */
1324 /* sb->s_iflags to limit user namespace mounts */
1325 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1327 /* Possible states of 'frozen' field */
1329 SB_UNFROZEN
= 0, /* FS is unfrozen */
1330 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1331 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1332 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1333 * internal threads if needed) */
1334 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1337 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1340 int frozen
; /* Is sb frozen? */
1341 wait_queue_head_t wait_unfrozen
; /* for get_super_thawed() */
1342 struct percpu_rw_semaphore rw_sem
[SB_FREEZE_LEVELS
];
1345 struct super_block
{
1346 struct list_head s_list
; /* Keep this first */
1347 dev_t s_dev
; /* search index; _not_ kdev_t */
1348 unsigned char s_blocksize_bits
;
1349 unsigned long s_blocksize
;
1350 loff_t s_maxbytes
; /* Max file size */
1351 struct file_system_type
*s_type
;
1352 const struct super_operations
*s_op
;
1353 const struct dquot_operations
*dq_op
;
1354 const struct quotactl_ops
*s_qcop
;
1355 const struct export_operations
*s_export_op
;
1356 unsigned long s_flags
;
1357 unsigned long s_iflags
; /* internal SB_I_* flags */
1358 unsigned long s_magic
;
1359 struct dentry
*s_root
;
1360 struct rw_semaphore s_umount
;
1363 #ifdef CONFIG_SECURITY
1366 const struct xattr_handler
**s_xattr
;
1368 const struct fscrypt_operations
*s_cop
;
1370 struct hlist_bl_head s_anon
; /* anonymous dentries for (nfs) exporting */
1371 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1372 struct block_device
*s_bdev
;
1373 struct backing_dev_info
*s_bdi
;
1374 struct mtd_info
*s_mtd
;
1375 struct hlist_node s_instances
;
1376 unsigned int s_quota_types
; /* Bitmask of supported quota types */
1377 struct quota_info s_dquot
; /* Diskquota specific options */
1379 struct sb_writers s_writers
;
1381 char s_id
[32]; /* Informational name */
1382 uuid_t s_uuid
; /* UUID */
1384 void *s_fs_info
; /* Filesystem private info */
1385 unsigned int s_max_links
;
1388 /* Granularity of c/m/atime in ns.
1389 Cannot be worse than a second */
1393 * The next field is for VFS *only*. No filesystems have any business
1394 * even looking at it. You had been warned.
1396 struct mutex s_vfs_rename_mutex
; /* Kludge */
1399 * Filesystem subtype. If non-empty the filesystem type field
1400 * in /proc/mounts will be "type.subtype"
1404 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1407 * Saved pool identifier for cleancache (-1 means none)
1409 int cleancache_poolid
;
1411 struct shrinker s_shrink
; /* per-sb shrinker handle */
1413 /* Number of inodes with nlink == 0 but still referenced */
1414 atomic_long_t s_remove_count
;
1416 /* Being remounted read-only */
1417 int s_readonly_remount
;
1419 /* AIO completions deferred from interrupt context */
1420 struct workqueue_struct
*s_dio_done_wq
;
1421 struct hlist_head s_pins
;
1424 * Owning user namespace and default context in which to
1425 * interpret filesystem uids, gids, quotas, device nodes,
1426 * xattrs and security labels.
1428 struct user_namespace
*s_user_ns
;
1431 * Keep the lru lists last in the structure so they always sit on their
1432 * own individual cachelines.
1434 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp
;
1435 struct list_lru s_inode_lru ____cacheline_aligned_in_smp
;
1436 struct rcu_head rcu
;
1437 struct work_struct destroy_work
;
1439 struct mutex s_sync_lock
; /* sync serialisation lock */
1442 * Indicates how deep in a filesystem stack this SB is
1446 /* s_inode_list_lock protects s_inodes */
1447 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp
;
1448 struct list_head s_inodes
; /* all inodes */
1450 spinlock_t s_inode_wblist_lock
;
1451 struct list_head s_inodes_wb
; /* writeback inodes */
1452 } __randomize_layout
;
1454 /* Helper functions so that in most cases filesystems will
1455 * not need to deal directly with kuid_t and kgid_t and can
1456 * instead deal with the raw numeric values that are stored
1457 * in the filesystem.
1459 static inline uid_t
i_uid_read(const struct inode
*inode
)
1461 return from_kuid(inode
->i_sb
->s_user_ns
, inode
->i_uid
);
1464 static inline gid_t
i_gid_read(const struct inode
*inode
)
1466 return from_kgid(inode
->i_sb
->s_user_ns
, inode
->i_gid
);
1469 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
1471 inode
->i_uid
= make_kuid(inode
->i_sb
->s_user_ns
, uid
);
1474 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
1476 inode
->i_gid
= make_kgid(inode
->i_sb
->s_user_ns
, gid
);
1479 extern struct timespec
current_time(struct inode
*inode
);
1482 * Snapshotting support.
1485 void __sb_end_write(struct super_block
*sb
, int level
);
1486 int __sb_start_write(struct super_block
*sb
, int level
, bool wait
);
1488 #define __sb_writers_acquired(sb, lev) \
1489 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1490 #define __sb_writers_release(sb, lev) \
1491 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1494 * sb_end_write - drop write access to a superblock
1495 * @sb: the super we wrote to
1497 * Decrement number of writers to the filesystem. Wake up possible waiters
1498 * wanting to freeze the filesystem.
1500 static inline void sb_end_write(struct super_block
*sb
)
1502 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1506 * sb_end_pagefault - drop write access to a superblock from a page fault
1507 * @sb: the super we wrote to
1509 * Decrement number of processes handling write page fault to the filesystem.
1510 * Wake up possible waiters wanting to freeze the filesystem.
1512 static inline void sb_end_pagefault(struct super_block
*sb
)
1514 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1518 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1519 * @sb: the super we wrote to
1521 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1522 * waiters wanting to freeze the filesystem.
1524 static inline void sb_end_intwrite(struct super_block
*sb
)
1526 __sb_end_write(sb
, SB_FREEZE_FS
);
1530 * sb_start_write - get write access to a superblock
1531 * @sb: the super we write to
1533 * When a process wants to write data or metadata to a file system (i.e. dirty
1534 * a page or an inode), it should embed the operation in a sb_start_write() -
1535 * sb_end_write() pair to get exclusion against file system freezing. This
1536 * function increments number of writers preventing freezing. If the file
1537 * system is already frozen, the function waits until the file system is
1540 * Since freeze protection behaves as a lock, users have to preserve
1541 * ordering of freeze protection and other filesystem locks. Generally,
1542 * freeze protection should be the outermost lock. In particular, we have:
1545 * -> i_mutex (write path, truncate, directory ops, ...)
1546 * -> s_umount (freeze_super, thaw_super)
1548 static inline void sb_start_write(struct super_block
*sb
)
1550 __sb_start_write(sb
, SB_FREEZE_WRITE
, true);
1553 static inline int sb_start_write_trylock(struct super_block
*sb
)
1555 return __sb_start_write(sb
, SB_FREEZE_WRITE
, false);
1559 * sb_start_pagefault - get write access to a superblock from a page fault
1560 * @sb: the super we write to
1562 * When a process starts handling write page fault, it should embed the
1563 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1564 * exclusion against file system freezing. This is needed since the page fault
1565 * is going to dirty a page. This function increments number of running page
1566 * faults preventing freezing. If the file system is already frozen, the
1567 * function waits until the file system is thawed.
1569 * Since page fault freeze protection behaves as a lock, users have to preserve
1570 * ordering of freeze protection and other filesystem locks. It is advised to
1571 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1572 * handling code implies lock dependency:
1575 * -> sb_start_pagefault
1577 static inline void sb_start_pagefault(struct super_block
*sb
)
1579 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
, true);
1583 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1584 * @sb: the super we write to
1586 * This is the third level of protection against filesystem freezing. It is
1587 * free for use by a filesystem. The only requirement is that it must rank
1588 * below sb_start_pagefault.
1590 * For example filesystem can call sb_start_intwrite() when starting a
1591 * transaction which somewhat eases handling of freezing for internal sources
1592 * of filesystem changes (internal fs threads, discarding preallocation on file
1595 static inline void sb_start_intwrite(struct super_block
*sb
)
1597 __sb_start_write(sb
, SB_FREEZE_FS
, true);
1601 extern bool inode_owner_or_capable(const struct inode
*inode
);
1604 * VFS helper functions..
1606 extern int vfs_create(struct inode
*, struct dentry
*, umode_t
, bool);
1607 extern int vfs_mkdir(struct inode
*, struct dentry
*, umode_t
);
1608 extern int vfs_mknod(struct inode
*, struct dentry
*, umode_t
, dev_t
);
1609 extern int vfs_symlink(struct inode
*, struct dentry
*, const char *);
1610 extern int vfs_link(struct dentry
*, struct inode
*, struct dentry
*, struct inode
**);
1611 extern int vfs_rmdir(struct inode
*, struct dentry
*);
1612 extern int vfs_unlink(struct inode
*, struct dentry
*, struct inode
**);
1613 extern int vfs_rename(struct inode
*, struct dentry
*, struct inode
*, struct dentry
*, struct inode
**, unsigned int);
1614 extern int vfs_whiteout(struct inode
*, struct dentry
*);
1616 extern struct dentry
*vfs_tmpfile(struct dentry
*dentry
, umode_t mode
,
1620 * VFS file helper functions.
1622 extern void inode_init_owner(struct inode
*inode
, const struct inode
*dir
,
1624 extern bool may_open_dev(const struct path
*path
);
1626 * VFS FS_IOC_FIEMAP helper definitions.
1628 struct fiemap_extent_info
{
1629 unsigned int fi_flags
; /* Flags as passed from user */
1630 unsigned int fi_extents_mapped
; /* Number of mapped extents */
1631 unsigned int fi_extents_max
; /* Size of fiemap_extent array */
1632 struct fiemap_extent __user
*fi_extents_start
; /* Start of
1633 fiemap_extent array */
1635 int fiemap_fill_next_extent(struct fiemap_extent_info
*info
, u64 logical
,
1636 u64 phys
, u64 len
, u32 flags
);
1637 int fiemap_check_flags(struct fiemap_extent_info
*fieinfo
, u32 fs_flags
);
1642 * NOTE! These match bits 12..15 of stat.st_mode
1643 * (ie "(i_mode >> 12) & 15").
1645 #define DT_UNKNOWN 0
1656 * This is the "filldir" function type, used by readdir() to let
1657 * the kernel specify what kind of dirent layout it wants to have.
1658 * This allows the kernel to read directories into kernel space or
1659 * to have different dirent layouts depending on the binary type.
1662 typedef int (*filldir_t
)(struct dir_context
*, const char *, int, loff_t
, u64
,
1665 struct dir_context
{
1666 const filldir_t actor
;
1670 struct block_device_operations
;
1672 /* These macros are for out of kernel modules to test that
1673 * the kernel supports the unlocked_ioctl and compat_ioctl
1674 * fields in struct file_operations. */
1675 #define HAVE_COMPAT_IOCTL 1
1676 #define HAVE_UNLOCKED_IOCTL 1
1679 * These flags let !MMU mmap() govern direct device mapping vs immediate
1680 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1682 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1683 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1684 * NOMMU_MAP_READ: Can be mapped for reading
1685 * NOMMU_MAP_WRITE: Can be mapped for writing
1686 * NOMMU_MAP_EXEC: Can be mapped for execution
1688 #define NOMMU_MAP_COPY 0x00000001
1689 #define NOMMU_MAP_DIRECT 0x00000008
1690 #define NOMMU_MAP_READ VM_MAYREAD
1691 #define NOMMU_MAP_WRITE VM_MAYWRITE
1692 #define NOMMU_MAP_EXEC VM_MAYEXEC
1694 #define NOMMU_VMFLAGS \
1695 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1700 struct file_operations
{
1701 struct module
*owner
;
1702 loff_t (*llseek
) (struct file
*, loff_t
, int);
1703 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
1704 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
1705 ssize_t (*read_iter
) (struct kiocb
*, struct iov_iter
*);
1706 ssize_t (*write_iter
) (struct kiocb
*, struct iov_iter
*);
1707 int (*iterate
) (struct file
*, struct dir_context
*);
1708 int (*iterate_shared
) (struct file
*, struct dir_context
*);
1709 unsigned int (*poll
) (struct file
*, struct poll_table_struct
*);
1710 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
1711 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
1712 int (*mmap
) (struct file
*, struct vm_area_struct
*);
1713 unsigned long mmap_supported_flags
;
1714 int (*open
) (struct inode
*, struct file
*);
1715 int (*flush
) (struct file
*, fl_owner_t id
);
1716 int (*release
) (struct inode
*, struct file
*);
1717 int (*fsync
) (struct file
*, loff_t
, loff_t
, int datasync
);
1718 int (*fasync
) (int, struct file
*, int);
1719 int (*lock
) (struct file
*, int, struct file_lock
*);
1720 ssize_t (*sendpage
) (struct file
*, struct page
*, int, size_t, loff_t
*, int);
1721 unsigned long (*get_unmapped_area
)(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1722 int (*check_flags
)(int);
1723 int (*setfl
)(struct file
*, unsigned long);
1724 int (*flock
) (struct file
*, int, struct file_lock
*);
1725 ssize_t (*splice_write
)(struct pipe_inode_info
*, struct file
*, loff_t
*, size_t, unsigned int);
1726 ssize_t (*splice_read
)(struct file
*, loff_t
*, struct pipe_inode_info
*, size_t, unsigned int);
1727 int (*setlease
)(struct file
*, long, struct file_lock
**, void **);
1728 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
1730 void (*show_fdinfo
)(struct seq_file
*m
, struct file
*f
);
1732 unsigned (*mmap_capabilities
)(struct file
*);
1734 ssize_t (*copy_file_range
)(struct file
*, loff_t
, struct file
*,
1735 loff_t
, size_t, unsigned int);
1736 int (*clone_file_range
)(struct file
*, loff_t
, struct file
*, loff_t
,
1738 ssize_t (*dedupe_file_range
)(struct file
*, u64
, u64
, struct file
*,
1740 } __randomize_layout
;
1742 struct inode_operations
{
1743 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*, unsigned int);
1744 const char * (*get_link
) (struct dentry
*, struct inode
*, struct delayed_call
*);
1745 int (*permission
) (struct inode
*, int);
1746 struct posix_acl
* (*get_acl
)(struct inode
*, int);
1748 int (*readlink
) (struct dentry
*, char __user
*,int);
1750 int (*create
) (struct inode
*,struct dentry
*, umode_t
, bool);
1751 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
1752 int (*unlink
) (struct inode
*,struct dentry
*);
1753 int (*symlink
) (struct inode
*,struct dentry
*,const char *);
1754 int (*mkdir
) (struct inode
*,struct dentry
*,umode_t
);
1755 int (*rmdir
) (struct inode
*,struct dentry
*);
1756 int (*mknod
) (struct inode
*,struct dentry
*,umode_t
,dev_t
);
1757 int (*rename
) (struct inode
*, struct dentry
*,
1758 struct inode
*, struct dentry
*, unsigned int);
1759 int (*setattr
) (struct dentry
*, struct iattr
*);
1760 int (*getattr
) (const struct path
*, struct kstat
*, u32
, unsigned int);
1761 ssize_t (*listxattr
) (struct dentry
*, char *, size_t);
1762 int (*fiemap
)(struct inode
*, struct fiemap_extent_info
*, u64 start
,
1764 int (*update_time
)(struct inode
*, struct timespec
*, int);
1765 int (*atomic_open
)(struct inode
*, struct dentry
*,
1766 struct file
*, unsigned open_flag
,
1767 umode_t create_mode
, int *opened
);
1768 int (*tmpfile
) (struct inode
*, struct dentry
*, umode_t
);
1769 int (*set_acl
)(struct inode
*, struct posix_acl
*, int);
1770 } ____cacheline_aligned
;
1772 static inline ssize_t
call_read_iter(struct file
*file
, struct kiocb
*kio
,
1773 struct iov_iter
*iter
)
1775 return file
->f_op
->read_iter(kio
, iter
);
1778 static inline ssize_t
call_write_iter(struct file
*file
, struct kiocb
*kio
,
1779 struct iov_iter
*iter
)
1781 return file
->f_op
->write_iter(kio
, iter
);
1784 static inline int call_mmap(struct file
*file
, struct vm_area_struct
*vma
)
1786 return file
->f_op
->mmap(file
, vma
);
1789 ssize_t
rw_copy_check_uvector(int type
, const struct iovec __user
* uvector
,
1790 unsigned long nr_segs
, unsigned long fast_segs
,
1791 struct iovec
*fast_pointer
,
1792 struct iovec
**ret_pointer
);
1794 typedef ssize_t (*vfs_readf_t
)(struct file
*, char __user
*, size_t, loff_t
*);
1795 typedef ssize_t (*vfs_writef_t
)(struct file
*, const char __user
*, size_t,
1797 vfs_readf_t
vfs_readf(struct file
*file
);
1798 vfs_writef_t
vfs_writef(struct file
*file
);
1800 extern ssize_t
__vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1801 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
1802 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
1803 extern ssize_t
vfs_readv(struct file
*, const struct iovec __user
*,
1804 unsigned long, loff_t
*, rwf_t
);
1805 extern ssize_t
vfs_copy_file_range(struct file
*, loff_t
, struct file
*,
1806 loff_t
, size_t, unsigned int);
1807 extern int vfs_clone_file_prep_inodes(struct inode
*inode_in
, loff_t pos_in
,
1808 struct inode
*inode_out
, loff_t pos_out
,
1809 u64
*len
, bool is_dedupe
);
1810 extern int do_clone_file_range(struct file
*file_in
, loff_t pos_in
,
1811 struct file
*file_out
, loff_t pos_out
, u64 len
);
1812 extern int vfs_clone_file_range(struct file
*file_in
, loff_t pos_in
,
1813 struct file
*file_out
, loff_t pos_out
, u64 len
);
1814 extern int vfs_dedupe_file_range_compare(struct inode
*src
, loff_t srcoff
,
1815 struct inode
*dest
, loff_t destoff
,
1816 loff_t len
, bool *is_same
);
1817 extern int vfs_dedupe_file_range(struct file
*file
,
1818 struct file_dedupe_range
*same
);
1820 struct super_operations
{
1821 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
1822 void (*destroy_inode
)(struct inode
*);
1824 void (*dirty_inode
) (struct inode
*, int flags
);
1825 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
1826 int (*drop_inode
) (struct inode
*);
1827 void (*evict_inode
) (struct inode
*);
1828 void (*put_super
) (struct super_block
*);
1829 int (*sync_fs
)(struct super_block
*sb
, int wait
);
1830 int (*freeze_super
) (struct super_block
*);
1831 int (*freeze_fs
) (struct super_block
*);
1832 int (*thaw_super
) (struct super_block
*);
1833 int (*unfreeze_fs
) (struct super_block
*);
1834 int (*statfs
) (struct dentry
*, struct kstatfs
*);
1835 int (*remount_fs
) (struct super_block
*, int *, char *);
1836 void (*umount_begin
) (struct super_block
*);
1838 int (*show_options
)(struct seq_file
*, struct dentry
*);
1839 int (*show_devname
)(struct seq_file
*, struct dentry
*);
1840 int (*show_path
)(struct seq_file
*, struct dentry
*);
1841 int (*show_stats
)(struct seq_file
*, struct dentry
*);
1843 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
1844 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
1845 struct dquot
**(*get_dquots
)(struct inode
*);
1847 int (*bdev_try_to_free_page
)(struct super_block
*, struct page
*, gfp_t
);
1848 long (*nr_cached_objects
)(struct super_block
*,
1849 struct shrink_control
*);
1850 long (*free_cached_objects
)(struct super_block
*,
1851 struct shrink_control
*);
1852 #if defined(CONFIG_BLK_DEV_LOOP) || defined(CONFIG_BLK_DEV_LOOP_MODULE)
1854 struct file
*(*real_loop
)(struct file
*);
1859 * Inode flags - they have no relation to superblock flags now
1861 #define S_SYNC 1 /* Writes are synced at once */
1862 #define S_NOATIME 2 /* Do not update access times */
1863 #define S_APPEND 4 /* Append-only file */
1864 #define S_IMMUTABLE 8 /* Immutable file */
1865 #define S_DEAD 16 /* removed, but still open directory */
1866 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1867 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1868 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1869 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1870 #define S_PRIVATE 512 /* Inode is fs-internal */
1871 #define S_IMA 1024 /* Inode has an associated IMA struct */
1872 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1873 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1874 #ifdef CONFIG_FS_DAX
1875 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1877 #define S_DAX 0 /* Make all the DAX code disappear */
1879 #define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */
1882 * Note that nosuid etc flags are inode-specific: setting some file-system
1883 * flags just means all the inodes inherit those flags by default. It might be
1884 * possible to override it selectively if you really wanted to with some
1885 * ioctl() that is not currently implemented.
1887 * Exception: SB_RDONLY is always applied to the entire file system.
1889 * Unfortunately, it is possible to change a filesystems flags with it mounted
1890 * with files in use. This means that all of the inodes will not have their
1891 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1892 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1894 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1896 static inline bool sb_rdonly(const struct super_block
*sb
) { return sb
->s_flags
& SB_RDONLY
; }
1897 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
1898 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
1899 ((inode)->i_flags & S_SYNC))
1900 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
1901 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1902 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
1903 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
1904 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
1906 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1907 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1908 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1909 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
1911 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1912 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1913 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1914 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1915 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1916 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1917 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1918 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1919 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
1921 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1922 (inode)->i_rdev == WHITEOUT_DEV)
1924 static inline bool HAS_UNMAPPED_ID(struct inode
*inode
)
1926 return !uid_valid(inode
->i_uid
) || !gid_valid(inode
->i_gid
);
1929 static inline enum rw_hint
file_write_hint(struct file
*file
)
1931 if (file
->f_write_hint
!= WRITE_LIFE_NOT_SET
)
1932 return file
->f_write_hint
;
1934 return file_inode(file
)->i_write_hint
;
1937 static inline int iocb_flags(struct file
*file
);
1939 static inline void init_sync_kiocb(struct kiocb
*kiocb
, struct file
*filp
)
1941 *kiocb
= (struct kiocb
) {
1943 .ki_flags
= iocb_flags(filp
),
1944 .ki_hint
= file_write_hint(filp
),
1949 * Inode state bits. Protected by inode->i_lock
1951 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1952 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1954 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1955 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1956 * various stages of removing an inode.
1958 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1960 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1961 * fdatasync(). i_atime is the usual cause.
1962 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1963 * these changes separately from I_DIRTY_SYNC so that we
1964 * don't have to write inode on fdatasync() when only
1965 * mtime has changed in it.
1966 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1967 * I_NEW Serves as both a mutex and completion notification.
1968 * New inodes set I_NEW. If two processes both create
1969 * the same inode, one of them will release its inode and
1970 * wait for I_NEW to be released before returning.
1971 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1972 * also cause waiting on I_NEW, without I_NEW actually
1973 * being set. find_inode() uses this to prevent returning
1974 * nearly-dead inodes.
1975 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1976 * is zero. I_FREEING must be set when I_WILL_FREE is
1978 * I_FREEING Set when inode is about to be freed but still has dirty
1979 * pages or buffers attached or the inode itself is still
1981 * I_CLEAR Added by clear_inode(). In this state the inode is
1982 * clean and can be destroyed. Inode keeps I_FREEING.
1984 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1985 * prohibited for many purposes. iget() must wait for
1986 * the inode to be completely released, then create it
1987 * anew. Other functions will just ignore such inodes,
1988 * if appropriate. I_NEW is used for waiting.
1990 * I_SYNC Writeback of inode is running. The bit is set during
1991 * data writeback, and cleared with a wakeup on the bit
1992 * address once it is done. The bit is also used to pin
1993 * the inode in memory for flusher thread.
1995 * I_REFERENCED Marks the inode as recently references on the LRU list.
1997 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1999 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2000 * synchronize competing switching instances and to tell
2001 * wb stat updates to grab mapping->tree_lock. See
2002 * inode_switch_wb_work_fn() for details.
2004 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2005 * and work dirs among overlayfs mounts.
2007 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2009 #define I_DIRTY_SYNC (1 << 0)
2010 #define I_DIRTY_DATASYNC (1 << 1)
2011 #define I_DIRTY_PAGES (1 << 2)
2013 #define I_NEW (1 << __I_NEW)
2014 #define I_WILL_FREE (1 << 4)
2015 #define I_FREEING (1 << 5)
2016 #define I_CLEAR (1 << 6)
2018 #define I_SYNC (1 << __I_SYNC)
2019 #define I_REFERENCED (1 << 8)
2020 #define __I_DIO_WAKEUP 9
2021 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2022 #define I_LINKABLE (1 << 10)
2023 #define I_DIRTY_TIME (1 << 11)
2024 #define __I_DIRTY_TIME_EXPIRED 12
2025 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
2026 #define I_WB_SWITCH (1 << 13)
2027 #define I_OVL_INUSE (1 << 14)
2029 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
2030 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2032 extern void __mark_inode_dirty(struct inode
*, int);
2033 static inline void mark_inode_dirty(struct inode
*inode
)
2035 __mark_inode_dirty(inode
, I_DIRTY
);
2038 static inline void mark_inode_dirty_sync(struct inode
*inode
)
2040 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
2043 extern void inc_nlink(struct inode
*inode
);
2044 extern void drop_nlink(struct inode
*inode
);
2045 extern void clear_nlink(struct inode
*inode
);
2046 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
2048 static inline void inode_inc_link_count(struct inode
*inode
)
2051 mark_inode_dirty(inode
);
2054 static inline void inode_dec_link_count(struct inode
*inode
)
2057 mark_inode_dirty(inode
);
2061 * inode_inc_iversion - increments i_version
2062 * @inode: inode that need to be updated
2064 * Every time the inode is modified, the i_version field will be incremented.
2065 * The filesystem has to be mounted with i_version flag
2068 static inline void inode_inc_iversion(struct inode
*inode
)
2070 spin_lock(&inode
->i_lock
);
2072 spin_unlock(&inode
->i_lock
);
2075 enum file_time_flags
{
2082 extern void touch_atime(const struct path
*);
2083 static inline void file_accessed(struct file
*file
)
2085 if (!(file
->f_flags
& O_NOATIME
))
2086 touch_atime(&file
->f_path
);
2089 int sync_inode(struct inode
*inode
, struct writeback_control
*wbc
);
2090 int sync_inode_metadata(struct inode
*inode
, int wait
);
2092 struct file_system_type
{
2095 #define FS_REQUIRES_DEV 1
2096 #define FS_BINARY_MOUNTDATA 2
2097 #define FS_HAS_SUBTYPE 4
2098 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2099 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2100 struct dentry
*(*mount
) (struct file_system_type
*, int,
2101 const char *, void *);
2102 void (*kill_sb
) (struct super_block
*);
2103 struct module
*owner
;
2104 struct file_system_type
* next
;
2105 struct hlist_head fs_supers
;
2107 struct lock_class_key s_lock_key
;
2108 struct lock_class_key s_umount_key
;
2109 struct lock_class_key s_vfs_rename_key
;
2110 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
2112 struct lock_class_key i_lock_key
;
2113 struct lock_class_key i_mutex_key
;
2114 struct lock_class_key i_mutex_dir_key
;
2117 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2119 extern struct dentry
*mount_ns(struct file_system_type
*fs_type
,
2120 int flags
, void *data
, void *ns
, struct user_namespace
*user_ns
,
2121 int (*fill_super
)(struct super_block
*, void *, int));
2123 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
2124 int flags
, const char *dev_name
, void *data
,
2125 int (*fill_super
)(struct super_block
*, void *, int));
2127 static inline struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
2128 int flags
, const char *dev_name
, void *data
,
2129 int (*fill_super
)(struct super_block
*, void *, int))
2131 return ERR_PTR(-ENODEV
);
2134 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
2135 int flags
, void *data
,
2136 int (*fill_super
)(struct super_block
*, void *, int));
2137 extern struct dentry
*mount_nodev(struct file_system_type
*fs_type
,
2138 int flags
, void *data
,
2139 int (*fill_super
)(struct super_block
*, void *, int));
2140 extern struct dentry
*mount_subtree(struct vfsmount
*mnt
, const char *path
);
2141 void generic_shutdown_super(struct super_block
*sb
);
2143 void kill_block_super(struct super_block
*sb
);
2145 static inline void kill_block_super(struct super_block
*sb
)
2150 void kill_anon_super(struct super_block
*sb
);
2151 void kill_litter_super(struct super_block
*sb
);
2152 void deactivate_super(struct super_block
*sb
);
2153 void deactivate_locked_super(struct super_block
*sb
);
2154 int set_anon_super(struct super_block
*s
, void *data
);
2155 int get_anon_bdev(dev_t
*);
2156 void free_anon_bdev(dev_t
);
2157 struct super_block
*sget_userns(struct file_system_type
*type
,
2158 int (*test
)(struct super_block
*,void *),
2159 int (*set
)(struct super_block
*,void *),
2160 int flags
, struct user_namespace
*user_ns
,
2162 struct super_block
*sget(struct file_system_type
*type
,
2163 int (*test
)(struct super_block
*,void *),
2164 int (*set
)(struct super_block
*,void *),
2165 int flags
, void *data
);
2166 extern struct dentry
*mount_pseudo_xattr(struct file_system_type
*, char *,
2167 const struct super_operations
*ops
,
2168 const struct xattr_handler
**xattr
,
2169 const struct dentry_operations
*dops
,
2172 static inline struct dentry
*
2173 mount_pseudo(struct file_system_type
*fs_type
, char *name
,
2174 const struct super_operations
*ops
,
2175 const struct dentry_operations
*dops
, unsigned long magic
)
2177 return mount_pseudo_xattr(fs_type
, name
, ops
, NULL
, dops
, magic
);
2180 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2181 #define fops_get(fops) \
2182 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2183 #define fops_put(fops) \
2184 do { if (fops) module_put((fops)->owner); } while(0)
2186 * This one is to be used *ONLY* from ->open() instances.
2187 * fops must be non-NULL, pinned down *and* module dependencies
2188 * should be sufficient to pin the caller down as well.
2190 #define replace_fops(f, fops) \
2192 struct file *__file = (f); \
2193 fops_put(__file->f_op); \
2194 BUG_ON(!(__file->f_op = (fops))); \
2197 extern int register_filesystem(struct file_system_type
*);
2198 extern int unregister_filesystem(struct file_system_type
*);
2199 extern struct vfsmount
*kern_mount_data(struct file_system_type
*, void *data
);
2200 #define kern_mount(type) kern_mount_data(type, NULL)
2201 extern void kern_unmount(struct vfsmount
*mnt
);
2202 extern int may_umount_tree(struct vfsmount
*);
2203 extern int may_umount(struct vfsmount
*);
2204 extern long do_mount(const char *, const char __user
*,
2205 const char *, unsigned long, void *);
2206 extern struct vfsmount
*collect_mounts(const struct path
*);
2207 extern void drop_collected_mounts(struct vfsmount
*);
2208 extern int iterate_mounts(int (*)(struct vfsmount
*, void *), void *,
2210 extern int vfs_statfs(const struct path
*, struct kstatfs
*);
2211 extern int user_statfs(const char __user
*, struct kstatfs
*);
2212 extern int fd_statfs(int, struct kstatfs
*);
2213 extern int freeze_super(struct super_block
*super
);
2214 extern int thaw_super(struct super_block
*super
);
2215 extern bool our_mnt(struct vfsmount
*mnt
);
2216 extern __printf(2, 3)
2217 int super_setup_bdi_name(struct super_block
*sb
, char *fmt
, ...);
2218 extern int super_setup_bdi(struct super_block
*sb
);
2220 extern int current_umask(void);
2222 extern void ihold(struct inode
* inode
);
2223 extern void iput(struct inode
*);
2224 extern int generic_update_time(struct inode
*, struct timespec
*, int);
2225 extern int update_time(struct inode
*, struct timespec
*, int);
2228 extern struct kobject
*fs_kobj
;
2230 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2232 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2233 extern int locks_mandatory_locked(struct file
*);
2234 extern int locks_mandatory_area(struct inode
*, struct file
*, loff_t
, loff_t
, unsigned char);
2237 * Candidates for mandatory locking have the setgid bit set
2238 * but no group execute bit - an otherwise meaningless combination.
2241 static inline int __mandatory_lock(struct inode
*ino
)
2243 return (ino
->i_mode
& (S_ISGID
| S_IXGRP
)) == S_ISGID
;
2247 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2248 * otherwise these will be advisory locks
2251 static inline int mandatory_lock(struct inode
*ino
)
2253 return IS_MANDLOCK(ino
) && __mandatory_lock(ino
);
2256 static inline int locks_verify_locked(struct file
*file
)
2258 if (mandatory_lock(locks_inode(file
)))
2259 return locks_mandatory_locked(file
);
2263 static inline int locks_verify_truncate(struct inode
*inode
,
2267 if (!inode
->i_flctx
|| !mandatory_lock(inode
))
2270 if (size
< inode
->i_size
) {
2271 return locks_mandatory_area(inode
, f
, size
, inode
->i_size
- 1,
2274 return locks_mandatory_area(inode
, f
, inode
->i_size
, size
- 1,
2279 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2281 static inline int locks_mandatory_locked(struct file
*file
)
2286 static inline int locks_mandatory_area(struct inode
*inode
, struct file
*filp
,
2287 loff_t start
, loff_t end
, unsigned char type
)
2292 static inline int __mandatory_lock(struct inode
*inode
)
2297 static inline int mandatory_lock(struct inode
*inode
)
2302 static inline int locks_verify_locked(struct file
*file
)
2307 static inline int locks_verify_truncate(struct inode
*inode
, struct file
*filp
,
2313 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2316 #ifdef CONFIG_FILE_LOCKING
2317 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2320 * Since this check is lockless, we must ensure that any refcounts
2321 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2322 * could end up racing with tasks trying to set a new lease on this
2326 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2327 return __break_lease(inode
, mode
, FL_LEASE
);
2331 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2334 * Since this check is lockless, we must ensure that any refcounts
2335 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2336 * could end up racing with tasks trying to set a new lease on this
2340 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2341 return __break_lease(inode
, mode
, FL_DELEG
);
2345 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2349 ret
= break_deleg(inode
, O_WRONLY
|O_NONBLOCK
);
2350 if (ret
== -EWOULDBLOCK
&& delegated_inode
) {
2351 *delegated_inode
= inode
;
2357 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2361 ret
= break_deleg(*delegated_inode
, O_WRONLY
);
2362 iput(*delegated_inode
);
2363 *delegated_inode
= NULL
;
2367 static inline int break_layout(struct inode
*inode
, bool wait
)
2370 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2371 return __break_lease(inode
,
2372 wait
? O_WRONLY
: O_WRONLY
| O_NONBLOCK
,
2377 #else /* !CONFIG_FILE_LOCKING */
2378 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2383 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2388 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2393 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2399 static inline int break_layout(struct inode
*inode
, bool wait
)
2404 #endif /* CONFIG_FILE_LOCKING */
2409 const char *name
; /* pointer to actual string */
2410 const __user
char *uptr
; /* original userland pointer */
2411 struct audit_names
*aname
;
2416 extern long vfs_truncate(const struct path
*, loff_t
);
2417 extern int do_truncate(struct dentry
*, loff_t start
, unsigned int time_attrs
,
2419 extern int vfs_fallocate(struct file
*file
, int mode
, loff_t offset
,
2421 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
2423 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
2424 extern struct file
*filp_open(const char *, int, umode_t
);
2425 extern struct file
*file_open_root(struct dentry
*, struct vfsmount
*,
2426 const char *, int, umode_t
);
2427 extern struct file
* dentry_open(const struct path
*, int, const struct cred
*);
2428 extern struct file
*filp_clone_open(struct file
*);
2429 extern int filp_close(struct file
*, fl_owner_t id
);
2431 extern struct filename
*getname_flags(const char __user
*, int, int *);
2432 extern struct filename
*getname(const char __user
*);
2433 extern struct filename
*getname_kernel(const char *);
2434 extern void putname(struct filename
*name
);
2440 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
2441 int (*open
)(struct inode
*, struct file
*),
2443 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
2447 extern int ioctl_preallocate(struct file
*filp
, void __user
*argp
);
2450 extern void __init
vfs_caches_init_early(void);
2451 extern void __init
vfs_caches_init(void);
2453 extern struct kmem_cache
*names_cachep
;
2455 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2456 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2459 extern int register_blkdev(unsigned int, const char *);
2460 extern void unregister_blkdev(unsigned int, const char *);
2461 extern void bdev_unhash_inode(dev_t dev
);
2462 extern struct block_device
*bdget(dev_t
);
2463 extern struct block_device
*bdgrab(struct block_device
*bdev
);
2464 extern void bd_set_size(struct block_device
*, loff_t size
);
2465 extern void bd_forget(struct inode
*inode
);
2466 extern void bdput(struct block_device
*);
2467 extern void invalidate_bdev(struct block_device
*);
2468 extern void iterate_bdevs(void (*)(struct block_device
*, void *), void *);
2469 extern int sync_blockdev(struct block_device
*bdev
);
2470 extern void kill_bdev(struct block_device
*);
2471 extern struct super_block
*freeze_bdev(struct block_device
*);
2472 extern void emergency_thaw_all(void);
2473 extern int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
);
2474 extern int fsync_bdev(struct block_device
*);
2476 extern struct super_block
*blockdev_superblock
;
2478 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2480 return sb
== blockdev_superblock
;
2483 static inline void bd_forget(struct inode
*inode
) {}
2484 static inline int sync_blockdev(struct block_device
*bdev
) { return 0; }
2485 static inline void kill_bdev(struct block_device
*bdev
) {}
2486 static inline void invalidate_bdev(struct block_device
*bdev
) {}
2488 static inline struct super_block
*freeze_bdev(struct block_device
*sb
)
2493 static inline int thaw_bdev(struct block_device
*bdev
, struct super_block
*sb
)
2498 static inline void iterate_bdevs(void (*f
)(struct block_device
*, void *), void *arg
)
2502 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2507 extern int __sync_filesystem(struct super_block
*, int);
2508 extern int sync_filesystem(struct super_block
*);
2509 extern const struct file_operations def_blk_fops
;
2510 extern const struct file_operations def_chr_fops
;
2512 extern int ioctl_by_bdev(struct block_device
*, unsigned, unsigned long);
2513 extern int blkdev_ioctl(struct block_device
*, fmode_t
, unsigned, unsigned long);
2514 extern long compat_blkdev_ioctl(struct file
*, unsigned, unsigned long);
2515 extern int blkdev_get(struct block_device
*bdev
, fmode_t mode
, void *holder
);
2516 extern struct block_device
*blkdev_get_by_path(const char *path
, fmode_t mode
,
2518 extern struct block_device
*blkdev_get_by_dev(dev_t dev
, fmode_t mode
,
2520 extern void blkdev_put(struct block_device
*bdev
, fmode_t mode
);
2521 extern int __blkdev_reread_part(struct block_device
*bdev
);
2522 extern int blkdev_reread_part(struct block_device
*bdev
);
2525 extern int bd_link_disk_holder(struct block_device
*bdev
, struct gendisk
*disk
);
2526 extern void bd_unlink_disk_holder(struct block_device
*bdev
,
2527 struct gendisk
*disk
);
2529 static inline int bd_link_disk_holder(struct block_device
*bdev
,
2530 struct gendisk
*disk
)
2534 static inline void bd_unlink_disk_holder(struct block_device
*bdev
,
2535 struct gendisk
*disk
)
2542 #define CHRDEV_MAJOR_MAX 512
2543 /* Marks the bottom of the first segment of free char majors */
2544 #define CHRDEV_MAJOR_DYN_END 234
2545 /* Marks the top and bottom of the second segment of free char majors */
2546 #define CHRDEV_MAJOR_DYN_EXT_START 511
2547 #define CHRDEV_MAJOR_DYN_EXT_END 384
2549 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
2550 extern int register_chrdev_region(dev_t
, unsigned, const char *);
2551 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
2552 unsigned int count
, const char *name
,
2553 const struct file_operations
*fops
);
2554 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
2555 unsigned int count
, const char *name
);
2556 extern void unregister_chrdev_region(dev_t
, unsigned);
2557 extern void chrdev_show(struct seq_file
*,off_t
);
2559 static inline int register_chrdev(unsigned int major
, const char *name
,
2560 const struct file_operations
*fops
)
2562 return __register_chrdev(major
, 0, 256, name
, fops
);
2565 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2567 __unregister_chrdev(major
, 0, 256, name
);
2570 /* fs/block_dev.c */
2571 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2572 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2575 #define BLKDEV_MAJOR_MAX 512
2576 extern const char *__bdevname(dev_t
, char *buffer
);
2577 extern const char *bdevname(struct block_device
*bdev
, char *buffer
);
2578 extern struct block_device
*lookup_bdev(const char *, int mask
);
2579 extern void blkdev_show(struct seq_file
*,off_t
);
2582 #define BLKDEV_MAJOR_MAX 0
2585 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2587 /* Invalid inode operations -- fs/bad_inode.c */
2588 extern void make_bad_inode(struct inode
*);
2589 extern bool is_bad_inode(struct inode
*);
2592 extern void check_disk_size_change(struct gendisk
*disk
,
2593 struct block_device
*bdev
);
2594 extern int revalidate_disk(struct gendisk
*);
2595 extern int check_disk_change(struct block_device
*);
2596 extern int __invalidate_device(struct block_device
*, bool);
2597 extern int invalidate_partition(struct gendisk
*, int);
2599 unsigned long invalidate_mapping_pages(struct address_space
*mapping
,
2600 pgoff_t start
, pgoff_t end
);
2602 static inline void invalidate_remote_inode(struct inode
*inode
)
2604 if (S_ISREG(inode
->i_mode
) || S_ISDIR(inode
->i_mode
) ||
2605 S_ISLNK(inode
->i_mode
))
2606 invalidate_mapping_pages(inode
->i_mapping
, 0, -1);
2608 extern int invalidate_inode_pages2(struct address_space
*mapping
);
2609 extern int invalidate_inode_pages2_range(struct address_space
*mapping
,
2610 pgoff_t start
, pgoff_t end
);
2611 extern int write_inode_now(struct inode
*, int);
2612 extern int filemap_fdatawrite(struct address_space
*);
2613 extern int filemap_flush(struct address_space
*);
2614 extern int filemap_fdatawait_keep_errors(struct address_space
*mapping
);
2615 extern int filemap_fdatawait_range(struct address_space
*, loff_t lstart
,
2618 static inline int filemap_fdatawait(struct address_space
*mapping
)
2620 return filemap_fdatawait_range(mapping
, 0, LLONG_MAX
);
2623 extern bool filemap_range_has_page(struct address_space
*, loff_t lstart
,
2625 extern int __must_check
file_fdatawait_range(struct file
*file
, loff_t lstart
,
2627 extern int filemap_write_and_wait(struct address_space
*mapping
);
2628 extern int filemap_write_and_wait_range(struct address_space
*mapping
,
2629 loff_t lstart
, loff_t lend
);
2630 extern int __filemap_fdatawrite_range(struct address_space
*mapping
,
2631 loff_t start
, loff_t end
, int sync_mode
);
2632 extern int filemap_fdatawrite_range(struct address_space
*mapping
,
2633 loff_t start
, loff_t end
);
2634 extern int filemap_check_errors(struct address_space
*mapping
);
2635 extern void __filemap_set_wb_err(struct address_space
*mapping
, int err
);
2637 extern int __must_check
file_fdatawait_range(struct file
*file
, loff_t lstart
,
2639 extern int __must_check
file_check_and_advance_wb_err(struct file
*file
);
2640 extern int __must_check
file_write_and_wait_range(struct file
*file
,
2641 loff_t start
, loff_t end
);
2643 static inline int file_write_and_wait(struct file
*file
)
2645 return file_write_and_wait_range(file
, 0, LLONG_MAX
);
2649 * filemap_set_wb_err - set a writeback error on an address_space
2650 * @mapping: mapping in which to set writeback error
2651 * @err: error to be set in mapping
2653 * When writeback fails in some way, we must record that error so that
2654 * userspace can be informed when fsync and the like are called. We endeavor
2655 * to report errors on any file that was open at the time of the error. Some
2656 * internal callers also need to know when writeback errors have occurred.
2658 * When a writeback error occurs, most filesystems will want to call
2659 * filemap_set_wb_err to record the error in the mapping so that it will be
2660 * automatically reported whenever fsync is called on the file.
2662 static inline void filemap_set_wb_err(struct address_space
*mapping
, int err
)
2664 /* Fastpath for common case of no error */
2666 __filemap_set_wb_err(mapping
, err
);
2670 * filemap_check_wb_error - has an error occurred since the mark was sampled?
2671 * @mapping: mapping to check for writeback errors
2672 * @since: previously-sampled errseq_t
2674 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2675 * the given value was sampled.
2677 * If it has then report the latest error set, otherwise return 0.
2679 static inline int filemap_check_wb_err(struct address_space
*mapping
,
2682 return errseq_check(&mapping
->wb_err
, since
);
2686 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2687 * @mapping: mapping to be sampled
2689 * Writeback errors are always reported relative to a particular sample point
2690 * in the past. This function provides those sample points.
2692 static inline errseq_t
filemap_sample_wb_err(struct address_space
*mapping
)
2694 return errseq_sample(&mapping
->wb_err
);
2697 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
2699 extern int vfs_fsync(struct file
*file
, int datasync
);
2702 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2703 * to already be updated for the write, and will return either the amount
2704 * of bytes passed in, or an error if syncing the file failed.
2706 static inline ssize_t
generic_write_sync(struct kiocb
*iocb
, ssize_t count
)
2708 if (iocb
->ki_flags
& IOCB_DSYNC
) {
2709 int ret
= vfs_fsync_range(iocb
->ki_filp
,
2710 iocb
->ki_pos
- count
, iocb
->ki_pos
- 1,
2711 (iocb
->ki_flags
& IOCB_SYNC
) ? 0 : 1);
2719 extern void emergency_sync(void);
2720 extern void emergency_remount(void);
2722 extern sector_t
bmap(struct inode
*, sector_t
);
2724 extern int notify_change(struct dentry
*, struct iattr
*, struct inode
**);
2725 extern int inode_permission(struct inode
*, int);
2726 extern int __inode_permission(struct inode
*, int);
2727 extern int generic_permission(struct inode
*, int);
2728 extern int __check_sticky(struct inode
*dir
, struct inode
*inode
);
2730 static inline bool execute_ok(struct inode
*inode
)
2732 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
2735 static inline void file_start_write(struct file
*file
)
2737 if (!S_ISREG(file_inode(file
)->i_mode
))
2739 __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, true);
2742 static inline bool file_start_write_trylock(struct file
*file
)
2744 if (!S_ISREG(file_inode(file
)->i_mode
))
2746 return __sb_start_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
, false);
2749 static inline void file_end_write(struct file
*file
)
2751 if (!S_ISREG(file_inode(file
)->i_mode
))
2753 __sb_end_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
);
2757 * get_write_access() gets write permission for a file.
2758 * put_write_access() releases this write permission.
2759 * This is used for regular files.
2760 * We cannot support write (and maybe mmap read-write shared) accesses and
2761 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2762 * can have the following values:
2763 * 0: no writers, no VM_DENYWRITE mappings
2764 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2765 * > 0: (i_writecount) users are writing to the file.
2767 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2768 * except for the cases where we don't hold i_writecount yet. Then we need to
2769 * use {get,deny}_write_access() - these functions check the sign and refuse
2770 * to do the change if sign is wrong.
2772 static inline int get_write_access(struct inode
*inode
)
2774 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2776 static inline int deny_write_access(struct file
*file
)
2778 struct inode
*inode
= file_inode(file
);
2779 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
2781 static inline void put_write_access(struct inode
* inode
)
2783 atomic_dec(&inode
->i_writecount
);
2785 static inline void allow_write_access(struct file
*file
)
2788 atomic_inc(&file_inode(file
)->i_writecount
);
2790 static inline bool inode_is_open_for_write(const struct inode
*inode
)
2792 return atomic_read(&inode
->i_writecount
) > 0;
2796 static inline void i_readcount_dec(struct inode
*inode
)
2798 BUG_ON(!atomic_read(&inode
->i_readcount
));
2799 atomic_dec(&inode
->i_readcount
);
2801 static inline void i_readcount_inc(struct inode
*inode
)
2803 atomic_inc(&inode
->i_readcount
);
2806 static inline void i_readcount_dec(struct inode
*inode
)
2810 static inline void i_readcount_inc(struct inode
*inode
)
2815 extern int do_pipe_flags(int *, int);
2817 #define __kernel_read_file_id(id) \
2818 id(UNKNOWN, unknown) \
2819 id(FIRMWARE, firmware) \
2820 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2821 id(MODULE, kernel-module) \
2822 id(KEXEC_IMAGE, kexec-image) \
2823 id(KEXEC_INITRAMFS, kexec-initramfs) \
2824 id(POLICY, security-policy) \
2825 id(X509_CERTIFICATE, x509-certificate) \
2828 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2829 #define __fid_stringify(dummy, str) #str,
2831 enum kernel_read_file_id
{
2832 __kernel_read_file_id(__fid_enumify
)
2835 static const char * const kernel_read_file_str
[] = {
2836 __kernel_read_file_id(__fid_stringify
)
2839 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id
)
2841 if ((unsigned)id
>= READING_MAX_ID
)
2842 return kernel_read_file_str
[READING_UNKNOWN
];
2844 return kernel_read_file_str
[id
];
2847 extern int kernel_read_file(struct file
*, void **, loff_t
*, loff_t
,
2848 enum kernel_read_file_id
);
2849 extern int kernel_read_file_from_path(const char *, void **, loff_t
*, loff_t
,
2850 enum kernel_read_file_id
);
2851 extern int kernel_read_file_from_fd(int, void **, loff_t
*, loff_t
,
2852 enum kernel_read_file_id
);
2853 extern ssize_t
kernel_read(struct file
*, void *, size_t, loff_t
*);
2854 extern ssize_t
kernel_write(struct file
*, const void *, size_t, loff_t
*);
2855 extern ssize_t
__kernel_write(struct file
*, const void *, size_t, loff_t
*);
2856 extern struct file
* open_exec(const char *);
2858 /* fs/dcache.c -- generic fs support functions */
2859 extern bool is_subdir(struct dentry
*, struct dentry
*);
2860 extern bool path_is_under(const struct path
*, const struct path
*);
2862 extern char *file_path(struct file
*, char *, int);
2864 #include <linux/err.h>
2866 /* needed for stackable file system support */
2867 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int whence
);
2869 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int whence
);
2871 extern int inode_init_always(struct super_block
*, struct inode
*);
2872 extern void inode_init_once(struct inode
*);
2873 extern void address_space_init_once(struct address_space
*mapping
);
2874 extern struct inode
* igrab(struct inode
*);
2875 extern ino_t
iunique(struct super_block
*, ino_t
);
2876 extern int inode_needs_sync(struct inode
*inode
);
2877 extern int generic_delete_inode(struct inode
*inode
);
2878 static inline int generic_drop_inode(struct inode
*inode
)
2880 return !inode
->i_nlink
|| inode_unhashed(inode
);
2883 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
2884 unsigned long hashval
, int (*test
)(struct inode
*, void *),
2886 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
2887 int (*test
)(struct inode
*, void *), void *data
);
2888 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
2890 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
2891 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
2892 extern struct inode
*find_inode_nowait(struct super_block
*,
2894 int (*match
)(struct inode
*,
2895 unsigned long, void *),
2897 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
2898 extern int insert_inode_locked(struct inode
*);
2899 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2900 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
2902 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
2904 extern void unlock_new_inode(struct inode
*);
2905 extern unsigned int get_next_ino(void);
2906 extern void evict_inodes(struct super_block
*sb
);
2908 extern void __iget(struct inode
* inode
);
2909 extern void iget_failed(struct inode
*);
2910 extern void clear_inode(struct inode
*);
2911 extern void __destroy_inode(struct inode
*);
2912 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
2913 extern struct inode
*new_inode(struct super_block
*sb
);
2914 extern void free_inode_nonrcu(struct inode
*inode
);
2915 extern int should_remove_suid(struct dentry
*);
2916 extern int file_remove_privs(struct file
*);
2918 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
2919 static inline void insert_inode_hash(struct inode
*inode
)
2921 __insert_inode_hash(inode
, inode
->i_ino
);
2924 extern void __remove_inode_hash(struct inode
*);
2925 static inline void remove_inode_hash(struct inode
*inode
)
2927 if (!inode_unhashed(inode
) && !hlist_fake(&inode
->i_hash
))
2928 __remove_inode_hash(inode
);
2931 extern void inode_sb_list_add(struct inode
*inode
);
2934 extern int bdev_read_only(struct block_device
*);
2936 extern int set_blocksize(struct block_device
*, int);
2937 extern int sb_set_blocksize(struct super_block
*, int);
2938 extern int sb_min_blocksize(struct super_block
*, int);
2940 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
2941 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
2942 extern ssize_t
generic_write_checks(struct kiocb
*, struct iov_iter
*);
2943 extern ssize_t
generic_file_read_iter(struct kiocb
*, struct iov_iter
*);
2944 extern ssize_t
__generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2945 extern ssize_t
generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
2946 extern ssize_t
generic_file_direct_write(struct kiocb
*, struct iov_iter
*);
2947 extern ssize_t
generic_perform_write(struct file
*, struct iov_iter
*, loff_t
);
2949 ssize_t
vfs_iter_read(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
2951 ssize_t
vfs_iter_write(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
2954 /* fs/block_dev.c */
2955 extern ssize_t
blkdev_read_iter(struct kiocb
*iocb
, struct iov_iter
*to
);
2956 extern ssize_t
blkdev_write_iter(struct kiocb
*iocb
, struct iov_iter
*from
);
2957 extern int blkdev_fsync(struct file
*filp
, loff_t start
, loff_t end
,
2959 extern void block_sync_page(struct page
*page
);
2962 extern ssize_t
generic_file_splice_read(struct file
*, loff_t
*,
2963 struct pipe_inode_info
*, size_t, unsigned int);
2964 extern ssize_t
iter_file_splice_write(struct pipe_inode_info
*,
2965 struct file
*, loff_t
*, size_t, unsigned int);
2966 extern ssize_t
generic_splice_sendpage(struct pipe_inode_info
*pipe
,
2967 struct file
*out
, loff_t
*, size_t len
, unsigned int flags
);
2968 extern long do_splice_direct(struct file
*in
, loff_t
*ppos
, struct file
*out
,
2969 loff_t
*opos
, size_t len
, unsigned int flags
);
2973 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
2974 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int whence
);
2975 extern loff_t
no_llseek(struct file
*file
, loff_t offset
, int whence
);
2976 extern loff_t
vfs_setpos(struct file
*file
, loff_t offset
, loff_t maxsize
);
2977 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int whence
);
2978 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
2979 int whence
, loff_t maxsize
, loff_t eof
);
2980 extern loff_t
fixed_size_llseek(struct file
*file
, loff_t offset
,
2981 int whence
, loff_t size
);
2982 extern loff_t
no_seek_end_llseek_size(struct file
*, loff_t
, int, loff_t
);
2983 extern loff_t
no_seek_end_llseek(struct file
*, loff_t
, int);
2984 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
2985 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
2986 extern int stream_open(struct inode
* inode
, struct file
* filp
);
2989 typedef void (dio_submit_t
)(struct bio
*bio
, struct inode
*inode
,
2990 loff_t file_offset
);
2993 /* need locking between buffered and direct access */
2996 /* filesystem does not support filling holes */
2997 DIO_SKIP_HOLES
= 0x02,
2999 /* filesystem can handle aio writes beyond i_size */
3000 DIO_ASYNC_EXTEND
= 0x04,
3002 /* inode/fs/bdev does not need truncate protection */
3003 DIO_SKIP_DIO_COUNT
= 0x08,
3006 void dio_end_io(struct bio
*bio
);
3007 void dio_warn_stale_pagecache(struct file
*filp
);
3009 ssize_t
__blockdev_direct_IO(struct kiocb
*iocb
, struct inode
*inode
,
3010 struct block_device
*bdev
, struct iov_iter
*iter
,
3011 get_block_t get_block
,
3012 dio_iodone_t end_io
, dio_submit_t submit_io
,
3015 static inline ssize_t
blockdev_direct_IO(struct kiocb
*iocb
,
3016 struct inode
*inode
,
3017 struct iov_iter
*iter
,
3018 get_block_t get_block
)
3020 return __blockdev_direct_IO(iocb
, inode
, inode
->i_sb
->s_bdev
, iter
,
3021 get_block
, NULL
, NULL
, DIO_LOCKING
| DIO_SKIP_HOLES
);
3025 void inode_dio_wait(struct inode
*inode
);
3028 * inode_dio_begin - signal start of a direct I/O requests
3029 * @inode: inode the direct I/O happens on
3031 * This is called once we've finished processing a direct I/O request,
3032 * and is used to wake up callers waiting for direct I/O to be quiesced.
3034 static inline void inode_dio_begin(struct inode
*inode
)
3036 atomic_inc(&inode
->i_dio_count
);
3040 * inode_dio_end - signal finish of a direct I/O requests
3041 * @inode: inode the direct I/O happens on
3043 * This is called once we've finished processing a direct I/O request,
3044 * and is used to wake up callers waiting for direct I/O to be quiesced.
3046 static inline void inode_dio_end(struct inode
*inode
)
3048 if (atomic_dec_and_test(&inode
->i_dio_count
))
3049 wake_up_bit(&inode
->i_state
, __I_DIO_WAKEUP
);
3052 extern void inode_set_flags(struct inode
*inode
, unsigned int flags
,
3055 extern const struct file_operations generic_ro_fops
;
3057 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3059 extern int readlink_copy(char __user
*, int, const char *);
3060 extern int page_readlink(struct dentry
*, char __user
*, int);
3061 extern const char *page_get_link(struct dentry
*, struct inode
*,
3062 struct delayed_call
*);
3063 extern void page_put_link(void *);
3064 extern int __page_symlink(struct inode
*inode
, const char *symname
, int len
,
3066 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
3067 extern const struct inode_operations page_symlink_inode_operations
;
3068 extern void kfree_link(void *);
3069 extern void generic_fillattr(struct inode
*, struct kstat
*);
3070 extern int vfs_getattr_nosec(const struct path
*, struct kstat
*, u32
, unsigned int);
3071 extern int vfs_getattr(const struct path
*, struct kstat
*, u32
, unsigned int);
3072 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
3073 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
3074 void __inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
3075 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
3076 static inline loff_t
__inode_get_bytes(struct inode
*inode
)
3078 return (((loff_t
)inode
->i_blocks
) << 9) + inode
->i_bytes
;
3080 loff_t
inode_get_bytes(struct inode
*inode
);
3081 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
3082 const char *simple_get_link(struct dentry
*, struct inode
*,
3083 struct delayed_call
*);
3084 extern const struct inode_operations simple_symlink_inode_operations
;
3086 extern int iterate_dir(struct file
*, struct dir_context
*);
3088 extern int vfs_statx(int, const char __user
*, int, struct kstat
*, u32
);
3089 extern int vfs_statx_fd(unsigned int, struct kstat
*, u32
, unsigned int);
3091 static inline int vfs_stat(const char __user
*filename
, struct kstat
*stat
)
3093 return vfs_statx(AT_FDCWD
, filename
, AT_NO_AUTOMOUNT
,
3094 stat
, STATX_BASIC_STATS
);
3096 static inline int vfs_lstat(const char __user
*name
, struct kstat
*stat
)
3098 return vfs_statx(AT_FDCWD
, name
, AT_SYMLINK_NOFOLLOW
| AT_NO_AUTOMOUNT
,
3099 stat
, STATX_BASIC_STATS
);
3101 static inline int vfs_fstatat(int dfd
, const char __user
*filename
,
3102 struct kstat
*stat
, int flags
)
3104 return vfs_statx(dfd
, filename
, flags
| AT_NO_AUTOMOUNT
,
3105 stat
, STATX_BASIC_STATS
);
3107 static inline int vfs_fstat(int fd
, struct kstat
*stat
)
3109 return vfs_statx_fd(fd
, stat
, STATX_BASIC_STATS
, 0);
3113 extern const char *vfs_get_link(struct dentry
*, struct delayed_call
*);
3114 extern int vfs_readlink(struct dentry
*, char __user
*, int);
3116 extern int __generic_block_fiemap(struct inode
*inode
,
3117 struct fiemap_extent_info
*fieinfo
,
3118 loff_t start
, loff_t len
,
3119 get_block_t
*get_block
);
3120 extern int generic_block_fiemap(struct inode
*inode
,
3121 struct fiemap_extent_info
*fieinfo
, u64 start
,
3122 u64 len
, get_block_t
*get_block
);
3124 extern struct file_system_type
*get_filesystem(struct file_system_type
*fs
);
3125 extern void put_filesystem(struct file_system_type
*fs
);
3126 extern struct file_system_type
*get_fs_type(const char *name
);
3127 extern struct super_block
*get_super(struct block_device
*);
3128 extern struct super_block
*get_super_thawed(struct block_device
*);
3129 extern struct super_block
*get_super_exclusive_thawed(struct block_device
*bdev
);
3130 extern struct super_block
*get_active_super(struct block_device
*bdev
);
3131 extern void drop_super(struct super_block
*sb
);
3132 extern void drop_super_exclusive(struct super_block
*sb
);
3133 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
3134 extern void iterate_supers_type(struct file_system_type
*,
3135 void (*)(struct super_block
*, void *), void *);
3137 extern int dcache_dir_open(struct inode
*, struct file
*);
3138 extern int dcache_dir_close(struct inode
*, struct file
*);
3139 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
3140 extern int dcache_readdir(struct file
*, struct dir_context
*);
3141 extern int simple_setattr(struct dentry
*, struct iattr
*);
3142 extern int simple_getattr(const struct path
*, struct kstat
*, u32
, unsigned int);
3143 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
3144 extern int simple_open(struct inode
*inode
, struct file
*file
);
3145 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
3146 extern int simple_unlink(struct inode
*, struct dentry
*);
3147 extern int simple_rmdir(struct inode
*, struct dentry
*);
3148 extern int simple_rename(struct inode
*, struct dentry
*,
3149 struct inode
*, struct dentry
*, unsigned int);
3150 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
3151 extern int simple_empty(struct dentry
*);
3152 extern int simple_readpage(struct file
*file
, struct page
*page
);
3153 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
3154 loff_t pos
, unsigned len
, unsigned flags
,
3155 struct page
**pagep
, void **fsdata
);
3156 extern int simple_write_end(struct file
*file
, struct address_space
*mapping
,
3157 loff_t pos
, unsigned len
, unsigned copied
,
3158 struct page
*page
, void *fsdata
);
3159 extern int always_delete_dentry(const struct dentry
*);
3160 extern struct inode
*alloc_anon_inode(struct super_block
*);
3161 extern int simple_nosetlease(struct file
*, long, struct file_lock
**, void **);
3162 extern const struct dentry_operations simple_dentry_operations
;
3164 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
3165 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
3166 extern const struct file_operations simple_dir_operations
;
3167 extern const struct inode_operations simple_dir_inode_operations
;
3168 extern void make_empty_dir_inode(struct inode
*inode
);
3169 extern bool is_empty_dir_inode(struct inode
*inode
);
3170 struct tree_descr
{ const char *name
; const struct file_operations
*ops
; int mode
; };
3171 struct dentry
*d_alloc_name(struct dentry
*, const char *);
3172 extern int simple_fill_super(struct super_block
*, unsigned long,
3173 const struct tree_descr
*);
3174 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
3175 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
3177 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
3178 loff_t
*ppos
, const void *from
, size_t available
);
3179 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
3180 const void __user
*from
, size_t count
);
3182 extern int __generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3183 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3185 extern int generic_check_addressable(unsigned, u64
);
3187 #ifdef CONFIG_MIGRATION
3188 extern int buffer_migrate_page(struct address_space
*,
3189 struct page
*, struct page
*,
3192 #define buffer_migrate_page NULL
3195 extern int setattr_prepare(struct dentry
*, struct iattr
*);
3196 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
3197 extern void setattr_copy(struct inode
*inode
, const struct iattr
*attr
);
3199 extern int file_update_time(struct file
*file
);
3201 static inline bool io_is_direct(struct file
*filp
)
3203 return (filp
->f_flags
& O_DIRECT
) || IS_DAX(filp
->f_mapping
->host
);
3206 static inline bool vma_is_dax(struct vm_area_struct
*vma
)
3208 return vma
->vm_file
&& IS_DAX(vma
->vm_file
->f_mapping
->host
);
3211 static inline bool vma_is_fsdax(struct vm_area_struct
*vma
)
3213 struct inode
*inode
;
3217 if (!vma_is_dax(vma
))
3219 inode
= file_inode(vma
->vm_file
);
3220 if (S_ISCHR(inode
->i_mode
))
3221 return false; /* device-dax */
3225 static inline int iocb_flags(struct file
*file
)
3228 if (file
->f_flags
& O_APPEND
)
3230 if (io_is_direct(file
))
3232 if ((file
->f_flags
& O_DSYNC
) || IS_SYNC(file
->f_mapping
->host
))
3234 if (file
->f_flags
& __O_SYNC
)
3239 static inline int kiocb_set_rw_flags(struct kiocb
*ki
, rwf_t flags
)
3241 if (unlikely(flags
& ~RWF_SUPPORTED
))
3244 if (flags
& RWF_NOWAIT
) {
3245 if (!(ki
->ki_filp
->f_mode
& FMODE_NOWAIT
))
3247 ki
->ki_flags
|= IOCB_NOWAIT
;
3249 if (flags
& RWF_HIPRI
)
3250 ki
->ki_flags
|= IOCB_HIPRI
;
3251 if (flags
& RWF_DSYNC
)
3252 ki
->ki_flags
|= IOCB_DSYNC
;
3253 if (flags
& RWF_SYNC
)
3254 ki
->ki_flags
|= (IOCB_DSYNC
| IOCB_SYNC
);
3258 static inline ino_t
parent_ino(struct dentry
*dentry
)
3263 * Don't strictly need d_lock here? If the parent ino could change
3264 * then surely we'd have a deeper race in the caller?
3266 spin_lock(&dentry
->d_lock
);
3267 res
= dentry
->d_parent
->d_inode
->i_ino
;
3268 spin_unlock(&dentry
->d_lock
);
3272 /* Transaction based IO helpers */
3275 * An argresp is stored in an allocated page and holds the
3276 * size of the argument or response, along with its content
3278 struct simple_transaction_argresp
{
3283 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3285 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
3287 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
3288 size_t size
, loff_t
*pos
);
3289 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
3291 void simple_transaction_set(struct file
*file
, size_t n
);
3294 * simple attribute files
3296 * These attributes behave similar to those in sysfs:
3298 * Writing to an attribute immediately sets a value, an open file can be
3299 * written to multiple times.
3301 * Reading from an attribute creates a buffer from the value that might get
3302 * read with multiple read calls. When the attribute has been read
3303 * completely, no further read calls are possible until the file is opened
3306 * All attributes contain a text representation of a numeric value
3307 * that are accessed with the get() and set() functions.
3309 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3310 static int __fops ## _open(struct inode *inode, struct file *file) \
3312 __simple_attr_check_format(__fmt, 0ull); \
3313 return simple_attr_open(inode, file, __get, __set, __fmt); \
3315 static const struct file_operations __fops = { \
3316 .owner = THIS_MODULE, \
3317 .open = __fops ## _open, \
3318 .release = simple_attr_release, \
3319 .read = simple_attr_read, \
3320 .write = simple_attr_write, \
3321 .llseek = generic_file_llseek, \
3324 static inline __printf(1, 2)
3325 void __simple_attr_check_format(const char *fmt
, ...)
3327 /* don't do anything, just let the compiler check the arguments; */
3330 int simple_attr_open(struct inode
*inode
, struct file
*file
,
3331 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
3333 int simple_attr_release(struct inode
*inode
, struct file
*file
);
3334 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
3335 size_t len
, loff_t
*ppos
);
3336 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
3337 size_t len
, loff_t
*ppos
);
3340 int proc_nr_files(struct ctl_table
*table
, int write
,
3341 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3342 int proc_nr_dentry(struct ctl_table
*table
, int write
,
3343 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3344 int proc_nr_inodes(struct ctl_table
*table
, int write
,
3345 void __user
*buffer
, size_t *lenp
, loff_t
*ppos
);
3346 int __init
get_filesystem_list(char *buf
);
3348 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3349 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3351 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3352 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3353 (flag & __FMODE_NONOTIFY)))
3355 static inline bool is_sxid(umode_t mode
)
3357 return (mode
& S_ISUID
) || ((mode
& S_ISGID
) && (mode
& S_IXGRP
));
3360 static inline int check_sticky(struct inode
*dir
, struct inode
*inode
)
3362 if (!(dir
->i_mode
& S_ISVTX
))
3365 return __check_sticky(dir
, inode
);
3368 static inline void inode_has_no_xattr(struct inode
*inode
)
3370 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& SB_NOSEC
))
3371 inode
->i_flags
|= S_NOSEC
;
3374 static inline bool is_root_inode(struct inode
*inode
)
3376 return inode
== inode
->i_sb
->s_root
->d_inode
;
3379 static inline bool dir_emit(struct dir_context
*ctx
,
3380 const char *name
, int namelen
,
3381 u64 ino
, unsigned type
)
3383 return ctx
->actor(ctx
, name
, namelen
, ctx
->pos
, ino
, type
) == 0;
3385 static inline bool dir_emit_dot(struct file
*file
, struct dir_context
*ctx
)
3387 return ctx
->actor(ctx
, ".", 1, ctx
->pos
,
3388 file
->f_path
.dentry
->d_inode
->i_ino
, DT_DIR
) == 0;
3390 static inline bool dir_emit_dotdot(struct file
*file
, struct dir_context
*ctx
)
3392 return ctx
->actor(ctx
, "..", 2, ctx
->pos
,
3393 parent_ino(file
->f_path
.dentry
), DT_DIR
) == 0;
3395 static inline bool dir_emit_dots(struct file
*file
, struct dir_context
*ctx
)
3397 if (ctx
->pos
== 0) {
3398 if (!dir_emit_dot(file
, ctx
))
3402 if (ctx
->pos
== 1) {
3403 if (!dir_emit_dotdot(file
, ctx
))
3409 static inline bool dir_relax(struct inode
*inode
)
3411 inode_unlock(inode
);
3413 return !IS_DEADDIR(inode
);
3416 static inline bool dir_relax_shared(struct inode
*inode
)
3418 inode_unlock_shared(inode
);
3419 inode_lock_shared(inode
);
3420 return !IS_DEADDIR(inode
);
3423 extern bool path_noexec(const struct path
*path
);
3424 extern bool path_nosuid(const struct path
*path
);
3425 extern void inode_nohighmem(struct inode
*inode
);
3427 #endif /* _LINUX_FS_H */