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/xarray.h>
17 #include <linux/rbtree.h>
18 #include <linux/init.h>
19 #include <linux/pid.h>
20 #include <linux/bug.h>
21 #include <linux/mutex.h>
22 #include <linux/rwsem.h>
23 #include <linux/mm_types.h>
24 #include <linux/capability.h>
25 #include <linux/semaphore.h>
26 #include <linux/fcntl.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>
38 #include <linux/ioprio.h>
39 #include <linux/fs_types.h>
40 #include <linux/build_bug.h>
41 #include <linux/stddef.h>
42 #include <linux/mount.h>
43 #include <linux/cred.h>
45 #include <asm/byteorder.h>
46 #include <uapi/linux/fs.h>
48 struct backing_dev_info
;
51 struct export_operations
;
52 struct fiemap_extent_info
;
57 struct pipe_inode_info
;
58 struct poll_table_struct
;
60 struct vm_area_struct
;
63 struct swap_info_struct
;
65 struct workqueue_struct
;
68 struct fscrypt_operations
;
70 struct fsverity_operations
;
72 struct fs_parameter_spec
;
75 extern void __init
inode_init(void);
76 extern void __init
inode_init_early(void);
77 extern void __init
files_init(void);
78 extern void __init
files_maxfiles_init(void);
80 extern struct files_stat_struct files_stat
;
81 extern unsigned long get_max_files(void);
82 extern unsigned int sysctl_nr_open
;
83 extern struct inodes_stat_t inodes_stat
;
84 extern int leases_enable
, lease_break_time
;
85 extern int sysctl_protected_symlinks
;
86 extern int sysctl_protected_hardlinks
;
87 extern int sysctl_protected_fifos
;
88 extern int sysctl_protected_regular
;
90 typedef __kernel_rwf_t rwf_t
;
93 typedef int (get_block_t
)(struct inode
*inode
, sector_t iblock
,
94 struct buffer_head
*bh_result
, int create
);
95 typedef int (dio_iodone_t
)(struct kiocb
*iocb
, loff_t offset
,
96 ssize_t bytes
, void *private);
98 #define MAY_EXEC 0x00000001
99 #define MAY_WRITE 0x00000002
100 #define MAY_READ 0x00000004
101 #define MAY_APPEND 0x00000008
102 #define MAY_ACCESS 0x00000010
103 #define MAY_OPEN 0x00000020
104 #define MAY_CHDIR 0x00000040
105 /* called from RCU mode, don't block */
106 #define MAY_NOT_BLOCK 0x00000080
109 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
110 * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
113 /* file is open for reading */
114 #define FMODE_READ ((__force fmode_t)0x1)
115 /* file is open for writing */
116 #define FMODE_WRITE ((__force fmode_t)0x2)
117 /* file is seekable */
118 #define FMODE_LSEEK ((__force fmode_t)0x4)
119 /* file can be accessed using pread */
120 #define FMODE_PREAD ((__force fmode_t)0x8)
121 /* file can be accessed using pwrite */
122 #define FMODE_PWRITE ((__force fmode_t)0x10)
123 /* File is opened for execution with sys_execve / sys_uselib */
124 #define FMODE_EXEC ((__force fmode_t)0x20)
125 /* File is opened with O_NDELAY (only set for block devices) */
126 #define FMODE_NDELAY ((__force fmode_t)0x40)
127 /* File is opened with O_EXCL (only set for block devices) */
128 #define FMODE_EXCL ((__force fmode_t)0x80)
129 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
130 (specialy hack for floppy.c) */
131 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
132 /* 32bit hashes as llseek() offset (for directories) */
133 #define FMODE_32BITHASH ((__force fmode_t)0x200)
134 /* 64bit hashes as llseek() offset (for directories) */
135 #define FMODE_64BITHASH ((__force fmode_t)0x400)
138 * Don't update ctime and mtime.
140 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
141 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
143 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
145 /* Expect random access pattern */
146 #define FMODE_RANDOM ((__force fmode_t)0x1000)
148 /* File is huge (eg. /dev/mem): treat loff_t as unsigned */
149 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
151 /* File is opened with O_PATH; almost nothing can be done with it */
152 #define FMODE_PATH ((__force fmode_t)0x4000)
154 /* File needs atomic accesses to f_pos */
155 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
156 /* Write access to underlying fs */
157 #define FMODE_WRITER ((__force fmode_t)0x10000)
158 /* Has read method(s) */
159 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
160 /* Has write method(s) */
161 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
163 #define FMODE_OPENED ((__force fmode_t)0x80000)
164 #define FMODE_CREATED ((__force fmode_t)0x100000)
166 /* File is stream-like */
167 #define FMODE_STREAM ((__force fmode_t)0x200000)
169 /* File was opened by fanotify and shouldn't generate fanotify events */
170 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
172 /* File is capable of returning -EAGAIN if I/O will block */
173 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
175 /* File represents mount that needs unmounting */
176 #define FMODE_NEED_UNMOUNT ((__force fmode_t)0x10000000)
178 /* File does not contribute to nr_files count */
179 #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000)
181 /* File supports async buffered reads */
182 #define FMODE_BUF_RASYNC ((__force fmode_t)0x40000000)
185 * Attribute flags. These should be or-ed together to figure out what
188 #define ATTR_MODE (1 << 0)
189 #define ATTR_UID (1 << 1)
190 #define ATTR_GID (1 << 2)
191 #define ATTR_SIZE (1 << 3)
192 #define ATTR_ATIME (1 << 4)
193 #define ATTR_MTIME (1 << 5)
194 #define ATTR_CTIME (1 << 6)
195 #define ATTR_ATIME_SET (1 << 7)
196 #define ATTR_MTIME_SET (1 << 8)
197 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
198 #define ATTR_KILL_SUID (1 << 11)
199 #define ATTR_KILL_SGID (1 << 12)
200 #define ATTR_FILE (1 << 13)
201 #define ATTR_KILL_PRIV (1 << 14)
202 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
203 #define ATTR_TIMES_SET (1 << 16)
204 #define ATTR_TOUCH (1 << 17)
207 * Whiteout is represented by a char device. The following constants define the
208 * mode and device number to use.
210 #define WHITEOUT_MODE 0
211 #define WHITEOUT_DEV 0
214 * This is the Inode Attributes structure, used for notify_change(). It
215 * uses the above definitions as flags, to know which values have changed.
216 * Also, in this manner, a Filesystem can look at only the values it cares
217 * about. Basically, these are the attributes that the VFS layer can
218 * request to change from the FS layer.
220 * Derek Atkins <warlord@MIT.EDU> 94-10-20
223 unsigned int ia_valid
;
228 struct timespec64 ia_atime
;
229 struct timespec64 ia_mtime
;
230 struct timespec64 ia_ctime
;
233 * Not an attribute, but an auxiliary info for filesystems wanting to
234 * implement an ftruncate() like method. NOTE: filesystem should
235 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
237 struct file
*ia_file
;
241 * Includes for diskquotas.
243 #include <linux/quota.h>
246 * Maximum number of layers of fs stack. Needs to be limited to
247 * prevent kernel stack overflow
249 #define FILESYSTEM_MAX_STACK_DEPTH 2
252 * enum positive_aop_returns - aop return codes with specific semantics
254 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
255 * completed, that the page is still locked, and
256 * should be considered active. The VM uses this hint
257 * to return the page to the active list -- it won't
258 * be a candidate for writeback again in the near
259 * future. Other callers must be careful to unlock
260 * the page if they get this return. Returned by
263 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
264 * unlocked it and the page might have been truncated.
265 * The caller should back up to acquiring a new page and
266 * trying again. The aop will be taking reasonable
267 * precautions not to livelock. If the caller held a page
268 * reference, it should drop it before retrying. Returned
271 * address_space_operation functions return these large constants to indicate
272 * special semantics to the caller. These are much larger than the bytes in a
273 * page to allow for functions that return the number of bytes operated on in a
277 enum positive_aop_returns
{
278 AOP_WRITEPAGE_ACTIVATE
= 0x80000,
279 AOP_TRUNCATED_PAGE
= 0x80001,
282 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
283 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
284 * helper code (eg buffer layer)
285 * to clear GFP_FS from alloc */
288 * oh the beauties of C type declarations.
291 struct address_space
;
292 struct writeback_control
;
293 struct readahead_control
;
296 * Write life time hint values.
297 * Stored in struct inode as u8.
300 WRITE_LIFE_NOT_SET
= 0,
301 WRITE_LIFE_NONE
= RWH_WRITE_LIFE_NONE
,
302 WRITE_LIFE_SHORT
= RWH_WRITE_LIFE_SHORT
,
303 WRITE_LIFE_MEDIUM
= RWH_WRITE_LIFE_MEDIUM
,
304 WRITE_LIFE_LONG
= RWH_WRITE_LIFE_LONG
,
305 WRITE_LIFE_EXTREME
= RWH_WRITE_LIFE_EXTREME
,
308 /* Match RWF_* bits to IOCB bits */
309 #define IOCB_HIPRI (__force int) RWF_HIPRI
310 #define IOCB_DSYNC (__force int) RWF_DSYNC
311 #define IOCB_SYNC (__force int) RWF_SYNC
312 #define IOCB_NOWAIT (__force int) RWF_NOWAIT
313 #define IOCB_APPEND (__force int) RWF_APPEND
315 /* non-RWF related bits - start at 16 */
316 #define IOCB_EVENTFD (1 << 16)
317 #define IOCB_DIRECT (1 << 17)
318 #define IOCB_WRITE (1 << 18)
319 /* iocb->ki_waitq is valid */
320 #define IOCB_WAITQ (1 << 19)
321 #define IOCB_NOIO (1 << 20)
324 struct file
*ki_filp
;
326 /* The 'ki_filp' pointer is shared in a union for aio */
327 randomized_struct_fields_start
330 void (*ki_complete
)(struct kiocb
*iocb
, long ret
, long ret2
);
334 u16 ki_ioprio
; /* See linux/ioprio.h */
336 unsigned int ki_cookie
; /* for ->iopoll */
337 struct wait_page_queue
*ki_waitq
; /* for async buffered IO */
340 randomized_struct_fields_end
343 static inline bool is_sync_kiocb(struct kiocb
*kiocb
)
345 return kiocb
->ki_complete
== NULL
;
349 * "descriptor" for what we're up to with a read.
350 * This allows us to use the same read code yet
351 * have multiple different users of the data that
352 * we read from a file.
354 * The simplest case just copies the data to user
367 typedef int (*read_actor_t
)(read_descriptor_t
*, struct page
*,
368 unsigned long, unsigned long);
370 struct address_space_operations
{
371 int (*writepage
)(struct page
*page
, struct writeback_control
*wbc
);
372 int (*readpage
)(struct file
*, struct page
*);
374 /* Write back some dirty pages from this mapping. */
375 int (*writepages
)(struct address_space
*, struct writeback_control
*);
377 /* Set a page dirty. Return true if this dirtied it */
378 int (*set_page_dirty
)(struct page
*page
);
381 * Reads in the requested pages. Unlike ->readpage(), this is
382 * PURELY used for read-ahead!.
384 int (*readpages
)(struct file
*filp
, struct address_space
*mapping
,
385 struct list_head
*pages
, unsigned nr_pages
);
386 void (*readahead
)(struct readahead_control
*);
388 int (*write_begin
)(struct file
*, struct address_space
*mapping
,
389 loff_t pos
, unsigned len
, unsigned flags
,
390 struct page
**pagep
, void **fsdata
);
391 int (*write_end
)(struct file
*, struct address_space
*mapping
,
392 loff_t pos
, unsigned len
, unsigned copied
,
393 struct page
*page
, void *fsdata
);
395 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
396 sector_t (*bmap
)(struct address_space
*, sector_t
);
397 void (*invalidatepage
) (struct page
*, unsigned int, unsigned int);
398 int (*releasepage
) (struct page
*, gfp_t
);
399 void (*freepage
)(struct page
*);
400 ssize_t (*direct_IO
)(struct kiocb
*, struct iov_iter
*iter
);
402 * migrate the contents of a page to the specified target. If
403 * migrate_mode is MIGRATE_ASYNC, it must not block.
405 int (*migratepage
) (struct address_space
*,
406 struct page
*, struct page
*, enum migrate_mode
);
407 bool (*isolate_page
)(struct page
*, isolate_mode_t
);
408 void (*putback_page
)(struct page
*);
409 int (*launder_page
) (struct page
*);
410 int (*is_partially_uptodate
) (struct page
*, unsigned long,
412 void (*is_dirty_writeback
) (struct page
*, bool *, bool *);
413 int (*error_remove_page
)(struct address_space
*, struct page
*);
415 /* swapfile support */
416 int (*swap_activate
)(struct swap_info_struct
*sis
, struct file
*file
,
418 void (*swap_deactivate
)(struct file
*file
);
421 extern const struct address_space_operations empty_aops
;
424 * pagecache_write_begin/pagecache_write_end must be used by general code
425 * to write into the pagecache.
427 int pagecache_write_begin(struct file
*, struct address_space
*mapping
,
428 loff_t pos
, unsigned len
, unsigned flags
,
429 struct page
**pagep
, void **fsdata
);
431 int pagecache_write_end(struct file
*, struct address_space
*mapping
,
432 loff_t pos
, unsigned len
, unsigned copied
,
433 struct page
*page
, void *fsdata
);
436 * struct address_space - Contents of a cacheable, mappable object.
437 * @host: Owner, either the inode or the block_device.
438 * @i_pages: Cached pages.
439 * @invalidate_lock: Guards coherency between page cache contents and
440 * file offset->disk block mappings in the filesystem during invalidates.
441 * It is also used to block modification of page cache contents through
443 * @gfp_mask: Memory allocation flags to use for allocating pages.
444 * @i_mmap_writable: Number of VM_SHARED mappings.
445 * @nr_thps: Number of THPs in the pagecache (non-shmem only).
446 * @i_mmap: Tree of private and shared mappings.
447 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
448 * @nrpages: Number of page entries, protected by the i_pages lock.
449 * @writeback_index: Writeback starts here.
451 * @flags: Error bits and flags (AS_*).
452 * @wb_err: The most recent error which has occurred.
453 * @private_lock: For use by the owner of the address_space.
454 * @private_list: For use by the owner of the address_space.
455 * @private_data: For use by the owner of the address_space.
457 struct address_space
{
459 struct xarray i_pages
;
460 struct rw_semaphore invalidate_lock
;
462 atomic_t i_mmap_writable
;
463 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
464 /* number of thp, only for non-shmem files */
467 struct rb_root_cached i_mmap
;
468 struct rw_semaphore i_mmap_rwsem
;
469 unsigned long nrpages
;
470 pgoff_t writeback_index
;
471 const struct address_space_operations
*a_ops
;
474 spinlock_t private_lock
;
475 struct list_head private_list
;
477 } __attribute__((aligned(sizeof(long)))) __randomize_layout
;
479 * On most architectures that alignment is already the case; but
480 * must be enforced here for CRIS, to let the least significant bit
481 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
484 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
485 #define PAGECACHE_TAG_DIRTY XA_MARK_0
486 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
487 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
490 * Returns true if any of the pages in the mapping are marked with the tag.
492 static inline bool mapping_tagged(struct address_space
*mapping
, xa_mark_t tag
)
494 return xa_marked(&mapping
->i_pages
, tag
);
497 static inline void i_mmap_lock_write(struct address_space
*mapping
)
499 down_write(&mapping
->i_mmap_rwsem
);
502 static inline int i_mmap_trylock_write(struct address_space
*mapping
)
504 return down_write_trylock(&mapping
->i_mmap_rwsem
);
507 static inline void i_mmap_unlock_write(struct address_space
*mapping
)
509 up_write(&mapping
->i_mmap_rwsem
);
512 static inline void i_mmap_lock_read(struct address_space
*mapping
)
514 down_read(&mapping
->i_mmap_rwsem
);
517 static inline void i_mmap_unlock_read(struct address_space
*mapping
)
519 up_read(&mapping
->i_mmap_rwsem
);
522 static inline void i_mmap_assert_locked(struct address_space
*mapping
)
524 lockdep_assert_held(&mapping
->i_mmap_rwsem
);
527 static inline void i_mmap_assert_write_locked(struct address_space
*mapping
)
529 lockdep_assert_held_write(&mapping
->i_mmap_rwsem
);
533 * Might pages of this file be mapped into userspace?
535 static inline int mapping_mapped(struct address_space
*mapping
)
537 return !RB_EMPTY_ROOT(&mapping
->i_mmap
.rb_root
);
541 * Might pages of this file have been modified in userspace?
542 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap
543 * marks vma as VM_SHARED if it is shared, and the file was opened for
544 * writing i.e. vma may be mprotected writable even if now readonly.
546 * If i_mmap_writable is negative, no new writable mappings are allowed. You
547 * can only deny writable mappings, if none exists right now.
549 static inline int mapping_writably_mapped(struct address_space
*mapping
)
551 return atomic_read(&mapping
->i_mmap_writable
) > 0;
554 static inline int mapping_map_writable(struct address_space
*mapping
)
556 return atomic_inc_unless_negative(&mapping
->i_mmap_writable
) ?
560 static inline void mapping_unmap_writable(struct address_space
*mapping
)
562 atomic_dec(&mapping
->i_mmap_writable
);
565 static inline int mapping_deny_writable(struct address_space
*mapping
)
567 return atomic_dec_unless_positive(&mapping
->i_mmap_writable
) ?
571 static inline void mapping_allow_writable(struct address_space
*mapping
)
573 atomic_inc(&mapping
->i_mmap_writable
);
577 * Use sequence counter to get consistent i_size on 32-bit processors.
579 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
580 #include <linux/seqlock.h>
581 #define __NEED_I_SIZE_ORDERED
582 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
584 #define i_size_ordered_init(inode) do { } while (0)
588 #define ACL_NOT_CACHED ((void *)(-1))
589 #define ACL_DONT_CACHE ((void *)(-3))
591 static inline struct posix_acl
*
592 uncached_acl_sentinel(struct task_struct
*task
)
594 return (void *)task
+ 1;
598 is_uncached_acl(struct posix_acl
*acl
)
600 return (long)acl
& 1;
603 #define IOP_FASTPERM 0x0001
604 #define IOP_LOOKUP 0x0002
605 #define IOP_NOFOLLOW 0x0004
606 #define IOP_XATTR 0x0008
607 #define IOP_DEFAULT_READLINK 0x0010
609 struct fsnotify_mark_connector
;
612 * Keep mostly read-only and often accessed (especially for
613 * the RCU path lookup and 'stat' data) fields at the beginning
614 * of the 'struct inode'
618 unsigned short i_opflags
;
621 unsigned int i_flags
;
623 #ifdef CONFIG_FS_POSIX_ACL
624 struct posix_acl
*i_acl
;
625 struct posix_acl
*i_default_acl
;
628 const struct inode_operations
*i_op
;
629 struct super_block
*i_sb
;
630 struct address_space
*i_mapping
;
632 #ifdef CONFIG_SECURITY
636 /* Stat data, not accessed from path walking */
639 * Filesystems may only read i_nlink directly. They shall use the
640 * following functions for modification:
642 * (set|clear|inc|drop)_nlink
643 * inode_(inc|dec)_link_count
646 const unsigned int i_nlink
;
647 unsigned int __i_nlink
;
651 struct timespec64 i_atime
;
652 struct timespec64 i_mtime
;
653 struct timespec64 i_ctime
;
654 spinlock_t i_lock
; /* i_blocks, i_bytes, maybe i_size */
655 unsigned short i_bytes
;
660 #ifdef __NEED_I_SIZE_ORDERED
661 seqcount_t i_size_seqcount
;
665 unsigned long i_state
;
666 struct rw_semaphore i_rwsem
;
668 unsigned long dirtied_when
; /* jiffies of first dirtying */
669 unsigned long dirtied_time_when
;
671 struct hlist_node i_hash
;
672 struct list_head i_io_list
; /* backing dev IO list */
673 #ifdef CONFIG_CGROUP_WRITEBACK
674 struct bdi_writeback
*i_wb
; /* the associated cgroup wb */
676 /* foreign inode detection, see wbc_detach_inode() */
678 u16 i_wb_frn_avg_time
;
679 u16 i_wb_frn_history
;
681 struct list_head i_lru
; /* inode LRU list */
682 struct list_head i_sb_list
;
683 struct list_head i_wb_list
; /* backing dev writeback list */
685 struct hlist_head i_dentry
;
686 struct rcu_head i_rcu
;
688 atomic64_t i_version
;
689 atomic64_t i_sequence
; /* see futex */
691 atomic_t i_dio_count
;
692 atomic_t i_writecount
;
693 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
694 atomic_t i_readcount
; /* struct files open RO */
697 const struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
698 void (*free_inode
)(struct inode
*);
700 struct file_lock_context
*i_flctx
;
701 struct address_space i_data
;
702 struct list_head i_devices
;
704 struct pipe_inode_info
*i_pipe
;
712 #ifdef CONFIG_FSNOTIFY
713 __u32 i_fsnotify_mask
; /* all events this inode cares about */
714 struct fsnotify_mark_connector __rcu
*i_fsnotify_marks
;
717 #ifdef CONFIG_FS_ENCRYPTION
718 struct fscrypt_info
*i_crypt_info
;
721 #ifdef CONFIG_FS_VERITY
722 struct fsverity_info
*i_verity_info
;
725 void *i_private
; /* fs or device private pointer */
726 } __randomize_layout
;
728 struct timespec64
timestamp_truncate(struct timespec64 t
, struct inode
*inode
);
730 static inline unsigned int i_blocksize(const struct inode
*node
)
732 return (1 << node
->i_blkbits
);
735 static inline int inode_unhashed(struct inode
*inode
)
737 return hlist_unhashed(&inode
->i_hash
);
741 * __mark_inode_dirty expects inodes to be hashed. Since we don't
742 * want special inodes in the fileset inode space, we make them
743 * appear hashed, but do not put on any lists. hlist_del()
744 * will work fine and require no locking.
746 static inline void inode_fake_hash(struct inode
*inode
)
748 hlist_add_fake(&inode
->i_hash
);
752 * inode->i_mutex nesting subclasses for the lock validator:
754 * 0: the object of the current VFS operation
758 * 4: second non-directory
759 * 5: second parent (when locking independent directories in rename)
761 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
762 * non-directories at once.
764 * The locking order between these classes is
765 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
767 enum inode_i_mutex_lock_class
777 static inline void inode_lock(struct inode
*inode
)
779 down_write(&inode
->i_rwsem
);
782 static inline void inode_unlock(struct inode
*inode
)
784 up_write(&inode
->i_rwsem
);
787 static inline void inode_lock_shared(struct inode
*inode
)
789 down_read(&inode
->i_rwsem
);
792 static inline void inode_unlock_shared(struct inode
*inode
)
794 up_read(&inode
->i_rwsem
);
797 static inline int inode_trylock(struct inode
*inode
)
799 return down_write_trylock(&inode
->i_rwsem
);
802 static inline int inode_trylock_shared(struct inode
*inode
)
804 return down_read_trylock(&inode
->i_rwsem
);
807 static inline int inode_is_locked(struct inode
*inode
)
809 return rwsem_is_locked(&inode
->i_rwsem
);
812 static inline void inode_lock_nested(struct inode
*inode
, unsigned subclass
)
814 down_write_nested(&inode
->i_rwsem
, subclass
);
817 static inline void inode_lock_shared_nested(struct inode
*inode
, unsigned subclass
)
819 down_read_nested(&inode
->i_rwsem
, subclass
);
822 static inline void filemap_invalidate_lock(struct address_space
*mapping
)
824 down_write(&mapping
->invalidate_lock
);
827 static inline void filemap_invalidate_unlock(struct address_space
*mapping
)
829 up_write(&mapping
->invalidate_lock
);
832 static inline void filemap_invalidate_lock_shared(struct address_space
*mapping
)
834 down_read(&mapping
->invalidate_lock
);
837 static inline int filemap_invalidate_trylock_shared(
838 struct address_space
*mapping
)
840 return down_read_trylock(&mapping
->invalidate_lock
);
843 static inline void filemap_invalidate_unlock_shared(
844 struct address_space
*mapping
)
846 up_read(&mapping
->invalidate_lock
);
849 void lock_two_nondirectories(struct inode
*, struct inode
*);
850 void unlock_two_nondirectories(struct inode
*, struct inode
*);
852 void filemap_invalidate_lock_two(struct address_space
*mapping1
,
853 struct address_space
*mapping2
);
854 void filemap_invalidate_unlock_two(struct address_space
*mapping1
,
855 struct address_space
*mapping2
);
859 * NOTE: in a 32bit arch with a preemptable kernel and
860 * an UP compile the i_size_read/write must be atomic
861 * with respect to the local cpu (unlike with preempt disabled),
862 * but they don't need to be atomic with respect to other cpus like in
863 * true SMP (so they need either to either locally disable irq around
864 * the read or for example on x86 they can be still implemented as a
865 * cmpxchg8b without the need of the lock prefix). For SMP compiles
866 * and 64bit archs it makes no difference if preempt is enabled or not.
868 static inline loff_t
i_size_read(const struct inode
*inode
)
870 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
875 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
876 i_size
= inode
->i_size
;
877 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
879 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
883 i_size
= inode
->i_size
;
887 return inode
->i_size
;
892 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
893 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
894 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
896 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
898 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
900 write_seqcount_begin(&inode
->i_size_seqcount
);
901 inode
->i_size
= i_size
;
902 write_seqcount_end(&inode
->i_size_seqcount
);
904 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
906 inode
->i_size
= i_size
;
909 inode
->i_size
= i_size
;
913 static inline unsigned iminor(const struct inode
*inode
)
915 return MINOR(inode
->i_rdev
);
918 static inline unsigned imajor(const struct inode
*inode
)
920 return MAJOR(inode
->i_rdev
);
924 rwlock_t lock
; /* protects pid, uid, euid fields */
925 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
926 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
927 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
928 int signum
; /* posix.1b rt signal to be delivered on IO */
932 * struct file_ra_state - Track a file's readahead state.
933 * @start: Where the most recent readahead started.
934 * @size: Number of pages read in the most recent readahead.
935 * @async_size: Start next readahead when this many pages are left.
936 * @ra_pages: Maximum size of a readahead request.
937 * @mmap_miss: How many mmap accesses missed in the page cache.
938 * @prev_pos: The last byte in the most recent read request.
940 struct file_ra_state
{
943 unsigned int async_size
;
944 unsigned int ra_pages
;
945 unsigned int mmap_miss
;
950 * Check if @index falls in the readahead windows.
952 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
954 return (index
>= ra
->start
&&
955 index
< ra
->start
+ ra
->size
);
960 struct llist_node fu_llist
;
961 struct rcu_head fu_rcuhead
;
964 struct inode
*f_inode
; /* cached value */
965 const struct file_operations
*f_op
;
968 * Protects f_ep, f_flags.
969 * Must not be taken from IRQ context.
972 enum rw_hint f_write_hint
;
973 atomic_long_t f_count
;
974 unsigned int f_flags
;
976 struct mutex f_pos_lock
;
978 struct fown_struct f_owner
;
979 const struct cred
*f_cred
;
980 struct file_ra_state f_ra
;
983 #ifdef CONFIG_SECURITY
986 /* needed for tty driver, and maybe others */
990 /* Used by fs/eventpoll.c to link all the hooks to this file */
991 struct hlist_head
*f_ep
;
992 #endif /* #ifdef CONFIG_EPOLL */
993 struct address_space
*f_mapping
;
995 errseq_t f_sb_err
; /* for syncfs */
997 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
1002 /* file identifier */
1003 unsigned char f_handle
[];
1006 static inline struct file
*get_file(struct file
*f
)
1008 atomic_long_inc(&f
->f_count
);
1011 #define get_file_rcu_many(x, cnt) \
1012 atomic_long_add_unless(&(x)->f_count, (cnt), 0)
1013 #define get_file_rcu(x) get_file_rcu_many((x), 1)
1014 #define file_count(x) atomic_long_read(&(x)->f_count)
1016 #define MAX_NON_LFS ((1UL<<31) - 1)
1018 /* Page cache limit. The filesystems should put that into their s_maxbytes
1019 limits, otherwise bad things can happen in VM. */
1020 #if BITS_PER_LONG==32
1021 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
1022 #elif BITS_PER_LONG==64
1023 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
1028 #define FL_DELEG 4 /* NFSv4 delegation */
1029 #define FL_ACCESS 8 /* not trying to lock, just looking */
1030 #define FL_EXISTS 16 /* when unlocking, test for existence */
1031 #define FL_LEASE 32 /* lease held on this file */
1032 #define FL_CLOSE 64 /* unlock on close */
1033 #define FL_SLEEP 128 /* A blocking lock */
1034 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
1035 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
1036 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
1037 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
1039 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
1042 * Special return value from posix_lock_file() and vfs_lock_file() for
1043 * asynchronous locking.
1045 #define FILE_LOCK_DEFERRED 1
1047 /* legacy typedef, should eventually be removed */
1048 typedef void *fl_owner_t
;
1052 struct file_lock_operations
{
1053 void (*fl_copy_lock
)(struct file_lock
*, struct file_lock
*);
1054 void (*fl_release_private
)(struct file_lock
*);
1057 struct lock_manager_operations
{
1058 fl_owner_t (*lm_get_owner
)(fl_owner_t
);
1059 void (*lm_put_owner
)(fl_owner_t
);
1060 void (*lm_notify
)(struct file_lock
*); /* unblock callback */
1061 int (*lm_grant
)(struct file_lock
*, int);
1062 bool (*lm_break
)(struct file_lock
*);
1063 int (*lm_change
)(struct file_lock
*, int, struct list_head
*);
1064 void (*lm_setup
)(struct file_lock
*, void **);
1065 bool (*lm_breaker_owns_lease
)(struct file_lock
*);
1068 struct lock_manager
{
1069 struct list_head list
;
1071 * NFSv4 and up also want opens blocked during the grace period;
1078 void locks_start_grace(struct net
*, struct lock_manager
*);
1079 void locks_end_grace(struct lock_manager
*);
1080 bool locks_in_grace(struct net
*);
1081 bool opens_in_grace(struct net
*);
1083 /* that will die - we need it for nfs_lock_info */
1084 #include <linux/nfs_fs_i.h>
1087 * struct file_lock represents a generic "file lock". It's used to represent
1088 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1089 * note that the same struct is used to represent both a request for a lock and
1090 * the lock itself, but the same object is never used for both.
1092 * FIXME: should we create a separate "struct lock_request" to help distinguish
1095 * The varous i_flctx lists are ordered by:
1098 * 2) lock range start
1101 * Obviously, the last two criteria only matter for POSIX locks.
1104 struct file_lock
*fl_blocker
; /* The lock, that is blocking us */
1105 struct list_head fl_list
; /* link into file_lock_context */
1106 struct hlist_node fl_link
; /* node in global lists */
1107 struct list_head fl_blocked_requests
; /* list of requests with
1108 * ->fl_blocker pointing here
1110 struct list_head fl_blocked_member
; /* node in
1111 * ->fl_blocker->fl_blocked_requests
1113 fl_owner_t fl_owner
;
1114 unsigned int fl_flags
;
1115 unsigned char fl_type
;
1116 unsigned int fl_pid
;
1117 int fl_link_cpu
; /* what cpu's list is this on? */
1118 wait_queue_head_t fl_wait
;
1119 struct file
*fl_file
;
1123 struct fasync_struct
* fl_fasync
; /* for lease break notifications */
1124 /* for lease breaks: */
1125 unsigned long fl_break_time
;
1126 unsigned long fl_downgrade_time
;
1128 const struct file_lock_operations
*fl_ops
; /* Callbacks for filesystems */
1129 const struct lock_manager_operations
*fl_lmops
; /* Callbacks for lockmanagers */
1131 struct nfs_lock_info nfs_fl
;
1132 struct nfs4_lock_info nfs4_fl
;
1134 struct list_head link
; /* link in AFS vnode's pending_locks list */
1135 int state
; /* state of grant or error if -ve */
1136 unsigned int debug_id
;
1139 } __randomize_layout
;
1141 struct file_lock_context
{
1142 spinlock_t flc_lock
;
1143 struct list_head flc_flock
;
1144 struct list_head flc_posix
;
1145 struct list_head flc_lease
;
1148 /* The following constant reflects the upper bound of the file/locking space */
1150 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1151 #define OFFSET_MAX INT_LIMIT(loff_t)
1152 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1155 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
1157 #define locks_inode(f) file_inode(f)
1159 #ifdef CONFIG_FILE_LOCKING
1160 extern int fcntl_getlk(struct file
*, unsigned int, struct flock
*);
1161 extern int fcntl_setlk(unsigned int, struct file
*, unsigned int,
1164 #if BITS_PER_LONG == 32
1165 extern int fcntl_getlk64(struct file
*, unsigned int, struct flock64
*);
1166 extern int fcntl_setlk64(unsigned int, struct file
*, unsigned int,
1170 extern int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
);
1171 extern int fcntl_getlease(struct file
*filp
);
1174 void locks_free_lock_context(struct inode
*inode
);
1175 void locks_free_lock(struct file_lock
*fl
);
1176 extern void locks_init_lock(struct file_lock
*);
1177 extern struct file_lock
* locks_alloc_lock(void);
1178 extern void locks_copy_lock(struct file_lock
*, struct file_lock
*);
1179 extern void locks_copy_conflock(struct file_lock
*, struct file_lock
*);
1180 extern void locks_remove_posix(struct file
*, fl_owner_t
);
1181 extern void locks_remove_file(struct file
*);
1182 extern void locks_release_private(struct file_lock
*);
1183 extern void posix_test_lock(struct file
*, struct file_lock
*);
1184 extern int posix_lock_file(struct file
*, struct file_lock
*, struct file_lock
*);
1185 extern int locks_delete_block(struct file_lock
*);
1186 extern int vfs_test_lock(struct file
*, struct file_lock
*);
1187 extern int vfs_lock_file(struct file
*, unsigned int, struct file_lock
*, struct file_lock
*);
1188 extern int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
);
1189 extern int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
);
1190 extern int __break_lease(struct inode
*inode
, unsigned int flags
, unsigned int type
);
1191 extern void lease_get_mtime(struct inode
*, struct timespec64
*time
);
1192 extern int generic_setlease(struct file
*, long, struct file_lock
**, void **priv
);
1193 extern int vfs_setlease(struct file
*, long, struct file_lock
**, void **);
1194 extern int lease_modify(struct file_lock
*, int, struct list_head
*);
1196 struct notifier_block
;
1197 extern int lease_register_notifier(struct notifier_block
*);
1198 extern void lease_unregister_notifier(struct notifier_block
*);
1200 struct files_struct
;
1201 extern void show_fd_locks(struct seq_file
*f
,
1202 struct file
*filp
, struct files_struct
*files
);
1203 #else /* !CONFIG_FILE_LOCKING */
1204 static inline int fcntl_getlk(struct file
*file
, unsigned int cmd
,
1205 struct flock __user
*user
)
1210 static inline int fcntl_setlk(unsigned int fd
, struct file
*file
,
1211 unsigned int cmd
, struct flock __user
*user
)
1216 #if BITS_PER_LONG == 32
1217 static inline int fcntl_getlk64(struct file
*file
, unsigned int cmd
,
1218 struct flock64 __user
*user
)
1223 static inline int fcntl_setlk64(unsigned int fd
, struct file
*file
,
1224 unsigned int cmd
, struct flock64 __user
*user
)
1229 static inline int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
)
1234 static inline int fcntl_getlease(struct file
*filp
)
1240 locks_free_lock_context(struct inode
*inode
)
1244 static inline void locks_init_lock(struct file_lock
*fl
)
1249 static inline void locks_copy_conflock(struct file_lock
*new, struct file_lock
*fl
)
1254 static inline void locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
1259 static inline void locks_remove_posix(struct file
*filp
, fl_owner_t owner
)
1264 static inline void locks_remove_file(struct file
*filp
)
1269 static inline void posix_test_lock(struct file
*filp
, struct file_lock
*fl
)
1274 static inline int posix_lock_file(struct file
*filp
, struct file_lock
*fl
,
1275 struct file_lock
*conflock
)
1280 static inline int locks_delete_block(struct file_lock
*waiter
)
1285 static inline int vfs_test_lock(struct file
*filp
, struct file_lock
*fl
)
1290 static inline int vfs_lock_file(struct file
*filp
, unsigned int cmd
,
1291 struct file_lock
*fl
, struct file_lock
*conf
)
1296 static inline int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
)
1301 static inline int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
)
1306 static inline int __break_lease(struct inode
*inode
, unsigned int mode
, unsigned int type
)
1311 static inline void lease_get_mtime(struct inode
*inode
,
1312 struct timespec64
*time
)
1317 static inline int generic_setlease(struct file
*filp
, long arg
,
1318 struct file_lock
**flp
, void **priv
)
1323 static inline int vfs_setlease(struct file
*filp
, long arg
,
1324 struct file_lock
**lease
, void **priv
)
1329 static inline int lease_modify(struct file_lock
*fl
, int arg
,
1330 struct list_head
*dispose
)
1335 struct files_struct
;
1336 static inline void show_fd_locks(struct seq_file
*f
,
1337 struct file
*filp
, struct files_struct
*files
) {}
1338 #endif /* !CONFIG_FILE_LOCKING */
1340 static inline struct inode
*file_inode(const struct file
*f
)
1345 static inline struct dentry
*file_dentry(const struct file
*file
)
1347 return d_real(file
->f_path
.dentry
, file_inode(file
));
1350 static inline int locks_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1352 return locks_lock_inode_wait(locks_inode(filp
), fl
);
1355 struct fasync_struct
{
1359 struct fasync_struct
*fa_next
; /* singly linked list */
1360 struct file
*fa_file
;
1361 struct rcu_head fa_rcu
;
1364 #define FASYNC_MAGIC 0x4601
1366 /* SMP safe fasync helpers: */
1367 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1368 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1369 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1370 extern struct fasync_struct
*fasync_alloc(void);
1371 extern void fasync_free(struct fasync_struct
*);
1373 /* can be called from interrupts */
1374 extern void kill_fasync(struct fasync_struct
**, int, int);
1376 extern void __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1377 extern int f_setown(struct file
*filp
, unsigned long arg
, int force
);
1378 extern void f_delown(struct file
*filp
);
1379 extern pid_t
f_getown(struct file
*filp
);
1380 extern int send_sigurg(struct fown_struct
*fown
);
1383 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1384 * represented in both.
1386 #define SB_RDONLY 1 /* Mount read-only */
1387 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1388 #define SB_NODEV 4 /* Disallow access to device special files */
1389 #define SB_NOEXEC 8 /* Disallow program execution */
1390 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1391 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1392 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1393 #define SB_NOATIME 1024 /* Do not update access times. */
1394 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1395 #define SB_SILENT 32768
1396 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1397 #define SB_INLINECRYPT (1<<17) /* Use blk-crypto for encrypted files */
1398 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1399 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1400 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1402 /* These sb flags are internal to the kernel */
1403 #define SB_SUBMOUNT (1<<26)
1404 #define SB_FORCE (1<<27)
1405 #define SB_NOSEC (1<<28)
1406 #define SB_BORN (1<<29)
1407 #define SB_ACTIVE (1<<30)
1408 #define SB_NOUSER (1<<31)
1410 /* These flags relate to encoding and casefolding */
1411 #define SB_ENC_STRICT_MODE_FL (1 << 0)
1413 #define sb_has_strict_encoding(sb) \
1414 (sb->s_encoding_flags & SB_ENC_STRICT_MODE_FL)
1420 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1421 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1422 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1423 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1424 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1427 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1428 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1429 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1430 #define SB_I_STABLE_WRITES 0x00000008 /* don't modify blks until WB is done */
1432 /* sb->s_iflags to limit user namespace mounts */
1433 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1434 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1435 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1437 #define SB_I_SKIP_SYNC 0x00000100 /* Skip superblock at global sync */
1439 /* Possible states of 'frozen' field */
1441 SB_UNFROZEN
= 0, /* FS is unfrozen */
1442 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1443 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1444 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1445 * internal threads if needed) */
1446 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1449 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1452 int frozen
; /* Is sb frozen? */
1453 wait_queue_head_t wait_unfrozen
; /* wait for thaw */
1454 struct percpu_rw_semaphore rw_sem
[SB_FREEZE_LEVELS
];
1457 struct super_block
{
1458 struct list_head s_list
; /* Keep this first */
1459 dev_t s_dev
; /* search index; _not_ kdev_t */
1460 unsigned char s_blocksize_bits
;
1461 unsigned long s_blocksize
;
1462 loff_t s_maxbytes
; /* Max file size */
1463 struct file_system_type
*s_type
;
1464 const struct super_operations
*s_op
;
1465 const struct dquot_operations
*dq_op
;
1466 const struct quotactl_ops
*s_qcop
;
1467 const struct export_operations
*s_export_op
;
1468 unsigned long s_flags
;
1469 unsigned long s_iflags
; /* internal SB_I_* flags */
1470 unsigned long s_magic
;
1471 struct dentry
*s_root
;
1472 struct rw_semaphore s_umount
;
1475 #ifdef CONFIG_SECURITY
1478 const struct xattr_handler
**s_xattr
;
1479 #ifdef CONFIG_FS_ENCRYPTION
1480 const struct fscrypt_operations
*s_cop
;
1481 struct key
*s_master_keys
; /* master crypto keys in use */
1483 #ifdef CONFIG_FS_VERITY
1484 const struct fsverity_operations
*s_vop
;
1486 #ifdef CONFIG_UNICODE
1487 struct unicode_map
*s_encoding
;
1488 __u16 s_encoding_flags
;
1490 struct hlist_bl_head s_roots
; /* alternate root dentries for NFS */
1491 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1492 struct block_device
*s_bdev
;
1493 struct backing_dev_info
*s_bdi
;
1494 struct mtd_info
*s_mtd
;
1495 struct hlist_node s_instances
;
1496 unsigned int s_quota_types
; /* Bitmask of supported quota types */
1497 struct quota_info s_dquot
; /* Diskquota specific options */
1499 struct sb_writers s_writers
;
1502 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1503 * s_fsnotify_marks together for cache efficiency. They are frequently
1504 * accessed and rarely modified.
1506 void *s_fs_info
; /* Filesystem private info */
1508 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1510 /* Time limits for c/m/atime in seconds */
1511 time64_t s_time_min
;
1512 time64_t s_time_max
;
1513 #ifdef CONFIG_FSNOTIFY
1514 __u32 s_fsnotify_mask
;
1515 struct fsnotify_mark_connector __rcu
*s_fsnotify_marks
;
1518 char s_id
[32]; /* Informational name */
1519 uuid_t s_uuid
; /* UUID */
1521 unsigned int s_max_links
;
1525 * The next field is for VFS *only*. No filesystems have any business
1526 * even looking at it. You had been warned.
1528 struct mutex s_vfs_rename_mutex
; /* Kludge */
1531 * Filesystem subtype. If non-empty the filesystem type field
1532 * in /proc/mounts will be "type.subtype"
1534 const char *s_subtype
;
1536 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1539 * Saved pool identifier for cleancache (-1 means none)
1541 int cleancache_poolid
;
1543 struct shrinker s_shrink
; /* per-sb shrinker handle */
1545 /* Number of inodes with nlink == 0 but still referenced */
1546 atomic_long_t s_remove_count
;
1548 /* Pending fsnotify inode refs */
1549 atomic_long_t s_fsnotify_inode_refs
;
1551 /* Being remounted read-only */
1552 int s_readonly_remount
;
1554 /* per-sb errseq_t for reporting writeback errors via syncfs */
1557 /* AIO completions deferred from interrupt context */
1558 struct workqueue_struct
*s_dio_done_wq
;
1559 struct hlist_head s_pins
;
1562 * Owning user namespace and default context in which to
1563 * interpret filesystem uids, gids, quotas, device nodes,
1564 * xattrs and security labels.
1566 struct user_namespace
*s_user_ns
;
1569 * The list_lru structure is essentially just a pointer to a table
1570 * of per-node lru lists, each of which has its own spinlock.
1571 * There is no need to put them into separate cachelines.
1573 struct list_lru s_dentry_lru
;
1574 struct list_lru s_inode_lru
;
1575 struct rcu_head rcu
;
1576 struct work_struct destroy_work
;
1578 struct mutex s_sync_lock
; /* sync serialisation lock */
1581 * Indicates how deep in a filesystem stack this SB is
1585 /* s_inode_list_lock protects s_inodes */
1586 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp
;
1587 struct list_head s_inodes
; /* all inodes */
1589 spinlock_t s_inode_wblist_lock
;
1590 struct list_head s_inodes_wb
; /* writeback inodes */
1591 } __randomize_layout
;
1593 /* Helper functions so that in most cases filesystems will
1594 * not need to deal directly with kuid_t and kgid_t and can
1595 * instead deal with the raw numeric values that are stored
1596 * in the filesystem.
1598 static inline uid_t
i_uid_read(const struct inode
*inode
)
1600 return from_kuid(inode
->i_sb
->s_user_ns
, inode
->i_uid
);
1603 static inline gid_t
i_gid_read(const struct inode
*inode
)
1605 return from_kgid(inode
->i_sb
->s_user_ns
, inode
->i_gid
);
1608 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
1610 inode
->i_uid
= make_kuid(inode
->i_sb
->s_user_ns
, uid
);
1613 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
1615 inode
->i_gid
= make_kgid(inode
->i_sb
->s_user_ns
, gid
);
1619 * kuid_into_mnt - map a kuid down into a mnt_userns
1620 * @mnt_userns: user namespace of the relevant mount
1621 * @kuid: kuid to be mapped
1623 * Return: @kuid mapped according to @mnt_userns.
1624 * If @kuid has no mapping INVALID_UID is returned.
1626 static inline kuid_t
kuid_into_mnt(struct user_namespace
*mnt_userns
,
1629 return make_kuid(mnt_userns
, __kuid_val(kuid
));
1633 * kgid_into_mnt - map a kgid down into a mnt_userns
1634 * @mnt_userns: user namespace of the relevant mount
1635 * @kgid: kgid to be mapped
1637 * Return: @kgid mapped according to @mnt_userns.
1638 * If @kgid has no mapping INVALID_GID is returned.
1640 static inline kgid_t
kgid_into_mnt(struct user_namespace
*mnt_userns
,
1643 return make_kgid(mnt_userns
, __kgid_val(kgid
));
1647 * i_uid_into_mnt - map an inode's i_uid down into a mnt_userns
1648 * @mnt_userns: user namespace of the mount the inode was found from
1649 * @inode: inode to map
1651 * Return: the inode's i_uid mapped down according to @mnt_userns.
1652 * If the inode's i_uid has no mapping INVALID_UID is returned.
1654 static inline kuid_t
i_uid_into_mnt(struct user_namespace
*mnt_userns
,
1655 const struct inode
*inode
)
1657 return kuid_into_mnt(mnt_userns
, inode
->i_uid
);
1661 * i_gid_into_mnt - map an inode's i_gid down into a mnt_userns
1662 * @mnt_userns: user namespace of the mount the inode was found from
1663 * @inode: inode to map
1665 * Return: the inode's i_gid mapped down according to @mnt_userns.
1666 * If the inode's i_gid has no mapping INVALID_GID is returned.
1668 static inline kgid_t
i_gid_into_mnt(struct user_namespace
*mnt_userns
,
1669 const struct inode
*inode
)
1671 return kgid_into_mnt(mnt_userns
, inode
->i_gid
);
1675 * kuid_from_mnt - map a kuid up into a mnt_userns
1676 * @mnt_userns: user namespace of the relevant mount
1677 * @kuid: kuid to be mapped
1679 * Return: @kuid mapped up according to @mnt_userns.
1680 * If @kuid has no mapping INVALID_UID is returned.
1682 static inline kuid_t
kuid_from_mnt(struct user_namespace
*mnt_userns
,
1685 return KUIDT_INIT(from_kuid(mnt_userns
, kuid
));
1689 * kgid_from_mnt - map a kgid up into a mnt_userns
1690 * @mnt_userns: user namespace of the relevant mount
1691 * @kgid: kgid to be mapped
1693 * Return: @kgid mapped up according to @mnt_userns.
1694 * If @kgid has no mapping INVALID_GID is returned.
1696 static inline kgid_t
kgid_from_mnt(struct user_namespace
*mnt_userns
,
1699 return KGIDT_INIT(from_kgid(mnt_userns
, kgid
));
1703 * mapped_fsuid - return caller's fsuid mapped up into a mnt_userns
1704 * @mnt_userns: user namespace of the relevant mount
1706 * Use this helper to initialize a new vfs or filesystem object based on
1707 * the caller's fsuid. A common example is initializing the i_uid field of
1708 * a newly allocated inode triggered by a creation event such as mkdir or
1709 * O_CREAT. Other examples include the allocation of quotas for a specific
1712 * Return: the caller's current fsuid mapped up according to @mnt_userns.
1714 static inline kuid_t
mapped_fsuid(struct user_namespace
*mnt_userns
)
1716 return kuid_from_mnt(mnt_userns
, current_fsuid());
1720 * mapped_fsgid - return caller's fsgid mapped up into a mnt_userns
1721 * @mnt_userns: user namespace of the relevant mount
1723 * Use this helper to initialize a new vfs or filesystem object based on
1724 * the caller's fsgid. A common example is initializing the i_gid field of
1725 * a newly allocated inode triggered by a creation event such as mkdir or
1726 * O_CREAT. Other examples include the allocation of quotas for a specific
1729 * Return: the caller's current fsgid mapped up according to @mnt_userns.
1731 static inline kgid_t
mapped_fsgid(struct user_namespace
*mnt_userns
)
1733 return kgid_from_mnt(mnt_userns
, current_fsgid());
1737 * inode_fsuid_set - initialize inode's i_uid field with callers fsuid
1738 * @inode: inode to initialize
1739 * @mnt_userns: user namespace of the mount the inode was found from
1741 * Initialize the i_uid field of @inode. If the inode was found/created via
1742 * an idmapped mount map the caller's fsuid according to @mnt_users.
1744 static inline void inode_fsuid_set(struct inode
*inode
,
1745 struct user_namespace
*mnt_userns
)
1747 inode
->i_uid
= mapped_fsuid(mnt_userns
);
1751 * inode_fsgid_set - initialize inode's i_gid field with callers fsgid
1752 * @inode: inode to initialize
1753 * @mnt_userns: user namespace of the mount the inode was found from
1755 * Initialize the i_gid field of @inode. If the inode was found/created via
1756 * an idmapped mount map the caller's fsgid according to @mnt_users.
1758 static inline void inode_fsgid_set(struct inode
*inode
,
1759 struct user_namespace
*mnt_userns
)
1761 inode
->i_gid
= mapped_fsgid(mnt_userns
);
1765 * fsuidgid_has_mapping() - check whether caller's fsuid/fsgid is mapped
1766 * @sb: the superblock we want a mapping in
1767 * @mnt_userns: user namespace of the relevant mount
1769 * Check whether the caller's fsuid and fsgid have a valid mapping in the
1770 * s_user_ns of the superblock @sb. If the caller is on an idmapped mount map
1771 * the caller's fsuid and fsgid according to the @mnt_userns first.
1773 * Return: true if fsuid and fsgid is mapped, false if not.
1775 static inline bool fsuidgid_has_mapping(struct super_block
*sb
,
1776 struct user_namespace
*mnt_userns
)
1778 struct user_namespace
*s_user_ns
= sb
->s_user_ns
;
1780 return kuid_has_mapping(s_user_ns
, mapped_fsuid(mnt_userns
)) &&
1781 kgid_has_mapping(s_user_ns
, mapped_fsgid(mnt_userns
));
1784 extern struct timespec64
current_time(struct inode
*inode
);
1787 * Snapshotting support.
1791 * These are internal functions, please use sb_start_{write,pagefault,intwrite}
1794 static inline void __sb_end_write(struct super_block
*sb
, int level
)
1796 percpu_up_read(sb
->s_writers
.rw_sem
+ level
-1);
1799 static inline void __sb_start_write(struct super_block
*sb
, int level
)
1801 percpu_down_read(sb
->s_writers
.rw_sem
+ level
- 1);
1804 static inline bool __sb_start_write_trylock(struct super_block
*sb
, int level
)
1806 return percpu_down_read_trylock(sb
->s_writers
.rw_sem
+ level
- 1);
1809 #define __sb_writers_acquired(sb, lev) \
1810 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1811 #define __sb_writers_release(sb, lev) \
1812 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1815 * sb_end_write - drop write access to a superblock
1816 * @sb: the super we wrote to
1818 * Decrement number of writers to the filesystem. Wake up possible waiters
1819 * wanting to freeze the filesystem.
1821 static inline void sb_end_write(struct super_block
*sb
)
1823 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1827 * sb_end_pagefault - drop write access to a superblock from a page fault
1828 * @sb: the super we wrote to
1830 * Decrement number of processes handling write page fault to the filesystem.
1831 * Wake up possible waiters wanting to freeze the filesystem.
1833 static inline void sb_end_pagefault(struct super_block
*sb
)
1835 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1839 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1840 * @sb: the super we wrote to
1842 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1843 * waiters wanting to freeze the filesystem.
1845 static inline void sb_end_intwrite(struct super_block
*sb
)
1847 __sb_end_write(sb
, SB_FREEZE_FS
);
1851 * sb_start_write - get write access to a superblock
1852 * @sb: the super we write to
1854 * When a process wants to write data or metadata to a file system (i.e. dirty
1855 * a page or an inode), it should embed the operation in a sb_start_write() -
1856 * sb_end_write() pair to get exclusion against file system freezing. This
1857 * function increments number of writers preventing freezing. If the file
1858 * system is already frozen, the function waits until the file system is
1861 * Since freeze protection behaves as a lock, users have to preserve
1862 * ordering of freeze protection and other filesystem locks. Generally,
1863 * freeze protection should be the outermost lock. In particular, we have:
1866 * -> i_mutex (write path, truncate, directory ops, ...)
1867 * -> s_umount (freeze_super, thaw_super)
1869 static inline void sb_start_write(struct super_block
*sb
)
1871 __sb_start_write(sb
, SB_FREEZE_WRITE
);
1874 static inline bool sb_start_write_trylock(struct super_block
*sb
)
1876 return __sb_start_write_trylock(sb
, SB_FREEZE_WRITE
);
1880 * sb_start_pagefault - get write access to a superblock from a page fault
1881 * @sb: the super we write to
1883 * When a process starts handling write page fault, it should embed the
1884 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1885 * exclusion against file system freezing. This is needed since the page fault
1886 * is going to dirty a page. This function increments number of running page
1887 * faults preventing freezing. If the file system is already frozen, the
1888 * function waits until the file system is thawed.
1890 * Since page fault freeze protection behaves as a lock, users have to preserve
1891 * ordering of freeze protection and other filesystem locks. It is advised to
1892 * put sb_start_pagefault() close to mmap_lock in lock ordering. Page fault
1893 * handling code implies lock dependency:
1896 * -> sb_start_pagefault
1898 static inline void sb_start_pagefault(struct super_block
*sb
)
1900 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
);
1904 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1905 * @sb: the super we write to
1907 * This is the third level of protection against filesystem freezing. It is
1908 * free for use by a filesystem. The only requirement is that it must rank
1909 * below sb_start_pagefault.
1911 * For example filesystem can call sb_start_intwrite() when starting a
1912 * transaction which somewhat eases handling of freezing for internal sources
1913 * of filesystem changes (internal fs threads, discarding preallocation on file
1916 static inline void sb_start_intwrite(struct super_block
*sb
)
1918 __sb_start_write(sb
, SB_FREEZE_FS
);
1921 static inline bool sb_start_intwrite_trylock(struct super_block
*sb
)
1923 return __sb_start_write_trylock(sb
, SB_FREEZE_FS
);
1926 bool inode_owner_or_capable(struct user_namespace
*mnt_userns
,
1927 const struct inode
*inode
);
1930 * VFS helper functions..
1932 int vfs_create(struct user_namespace
*, struct inode
*,
1933 struct dentry
*, umode_t
, bool);
1934 int vfs_mkdir(struct user_namespace
*, struct inode
*,
1935 struct dentry
*, umode_t
);
1936 int vfs_mknod(struct user_namespace
*, struct inode
*, struct dentry
*,
1938 int vfs_symlink(struct user_namespace
*, struct inode
*,
1939 struct dentry
*, const char *);
1940 int vfs_link(struct dentry
*, struct user_namespace
*, struct inode
*,
1941 struct dentry
*, struct inode
**);
1942 int vfs_rmdir(struct user_namespace
*, struct inode
*, struct dentry
*);
1943 int vfs_unlink(struct user_namespace
*, struct inode
*, struct dentry
*,
1947 * struct renamedata - contains all information required for renaming
1948 * @old_mnt_userns: old user namespace of the mount the inode was found from
1949 * @old_dir: parent of source
1950 * @old_dentry: source
1951 * @new_mnt_userns: new user namespace of the mount the inode was found from
1952 * @new_dir: parent of destination
1953 * @new_dentry: destination
1954 * @delegated_inode: returns an inode needing a delegation break
1955 * @flags: rename flags
1958 struct user_namespace
*old_mnt_userns
;
1959 struct inode
*old_dir
;
1960 struct dentry
*old_dentry
;
1961 struct user_namespace
*new_mnt_userns
;
1962 struct inode
*new_dir
;
1963 struct dentry
*new_dentry
;
1964 struct inode
**delegated_inode
;
1966 } __randomize_layout
;
1968 int vfs_rename(struct renamedata
*);
1970 static inline int vfs_whiteout(struct user_namespace
*mnt_userns
,
1971 struct inode
*dir
, struct dentry
*dentry
)
1973 return vfs_mknod(mnt_userns
, dir
, dentry
, S_IFCHR
| WHITEOUT_MODE
,
1977 struct dentry
*vfs_tmpfile(struct user_namespace
*mnt_userns
,
1978 struct dentry
*dentry
, umode_t mode
, int open_flag
);
1980 int vfs_mkobj(struct dentry
*, umode_t
,
1981 int (*f
)(struct dentry
*, umode_t
, void *),
1984 int vfs_fchown(struct file
*file
, uid_t user
, gid_t group
);
1985 int vfs_fchmod(struct file
*file
, umode_t mode
);
1986 int vfs_utimes(const struct path
*path
, struct timespec64
*times
);
1988 extern long vfs_ioctl(struct file
*file
, unsigned int cmd
, unsigned long arg
);
1990 #ifdef CONFIG_COMPAT
1991 extern long compat_ptr_ioctl(struct file
*file
, unsigned int cmd
,
1994 #define compat_ptr_ioctl NULL
1998 * VFS file helper functions.
2000 void inode_init_owner(struct user_namespace
*mnt_userns
, struct inode
*inode
,
2001 const struct inode
*dir
, umode_t mode
);
2002 extern bool may_open_dev(const struct path
*path
);
2005 * This is the "filldir" function type, used by readdir() to let
2006 * the kernel specify what kind of dirent layout it wants to have.
2007 * This allows the kernel to read directories into kernel space or
2008 * to have different dirent layouts depending on the binary type.
2011 typedef int (*filldir_t
)(struct dir_context
*, const char *, int, loff_t
, u64
,
2014 struct dir_context
{
2020 * These flags let !MMU mmap() govern direct device mapping vs immediate
2021 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
2023 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
2024 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
2025 * NOMMU_MAP_READ: Can be mapped for reading
2026 * NOMMU_MAP_WRITE: Can be mapped for writing
2027 * NOMMU_MAP_EXEC: Can be mapped for execution
2029 #define NOMMU_MAP_COPY 0x00000001
2030 #define NOMMU_MAP_DIRECT 0x00000008
2031 #define NOMMU_MAP_READ VM_MAYREAD
2032 #define NOMMU_MAP_WRITE VM_MAYWRITE
2033 #define NOMMU_MAP_EXEC VM_MAYEXEC
2035 #define NOMMU_VMFLAGS \
2036 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
2039 * These flags control the behavior of the remap_file_range function pointer.
2040 * If it is called with len == 0 that means "remap to end of source file".
2041 * See Documentation/filesystems/vfs.rst for more details about this call.
2043 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
2044 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
2046 #define REMAP_FILE_DEDUP (1 << 0)
2047 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
2050 * These flags signal that the caller is ok with altering various aspects of
2051 * the behavior of the remap operation. The changes must be made by the
2052 * implementation; the vfs remap helper functions can take advantage of them.
2053 * Flags in this category exist to preserve the quirky behavior of the hoisted
2054 * btrfs clone/dedupe ioctls.
2056 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
2060 struct file_operations
{
2061 struct module
*owner
;
2062 loff_t (*llseek
) (struct file
*, loff_t
, int);
2063 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
2064 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
2065 ssize_t (*read_iter
) (struct kiocb
*, struct iov_iter
*);
2066 ssize_t (*write_iter
) (struct kiocb
*, struct iov_iter
*);
2067 int (*iopoll
)(struct kiocb
*kiocb
, bool spin
);
2068 int (*iterate
) (struct file
*, struct dir_context
*);
2069 int (*iterate_shared
) (struct file
*, struct dir_context
*);
2070 __poll_t (*poll
) (struct file
*, struct poll_table_struct
*);
2071 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
2072 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
2073 int (*mmap
) (struct file
*, struct vm_area_struct
*);
2074 unsigned long mmap_supported_flags
;
2075 int (*open
) (struct inode
*, struct file
*);
2076 int (*flush
) (struct file
*, fl_owner_t id
);
2077 int (*release
) (struct inode
*, struct file
*);
2078 int (*fsync
) (struct file
*, loff_t
, loff_t
, int datasync
);
2079 int (*fasync
) (int, struct file
*, int);
2080 int (*lock
) (struct file
*, int, struct file_lock
*);
2081 ssize_t (*sendpage
) (struct file
*, struct page
*, int, size_t, loff_t
*, int);
2082 unsigned long (*get_unmapped_area
)(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
2083 int (*check_flags
)(int);
2084 int (*flock
) (struct file
*, int, struct file_lock
*);
2085 ssize_t (*splice_write
)(struct pipe_inode_info
*, struct file
*, loff_t
*, size_t, unsigned int);
2086 ssize_t (*splice_read
)(struct file
*, loff_t
*, struct pipe_inode_info
*, size_t, unsigned int);
2087 int (*setlease
)(struct file
*, long, struct file_lock
**, void **);
2088 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
2090 void (*show_fdinfo
)(struct seq_file
*m
, struct file
*f
);
2092 unsigned (*mmap_capabilities
)(struct file
*);
2094 ssize_t (*copy_file_range
)(struct file
*, loff_t
, struct file
*,
2095 loff_t
, size_t, unsigned int);
2096 loff_t (*remap_file_range
)(struct file
*file_in
, loff_t pos_in
,
2097 struct file
*file_out
, loff_t pos_out
,
2098 loff_t len
, unsigned int remap_flags
);
2099 int (*fadvise
)(struct file
*, loff_t
, loff_t
, int);
2100 } __randomize_layout
;
2102 struct inode_operations
{
2103 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*, unsigned int);
2104 const char * (*get_link
) (struct dentry
*, struct inode
*, struct delayed_call
*);
2105 int (*permission
) (struct user_namespace
*, struct inode
*, int);
2106 struct posix_acl
* (*get_acl
)(struct inode
*, int);
2108 int (*readlink
) (struct dentry
*, char __user
*,int);
2110 int (*create
) (struct user_namespace
*, struct inode
*,struct dentry
*,
2112 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
2113 int (*unlink
) (struct inode
*,struct dentry
*);
2114 int (*symlink
) (struct user_namespace
*, struct inode
*,struct dentry
*,
2116 int (*mkdir
) (struct user_namespace
*, struct inode
*,struct dentry
*,
2118 int (*rmdir
) (struct inode
*,struct dentry
*);
2119 int (*mknod
) (struct user_namespace
*, struct inode
*,struct dentry
*,
2121 int (*rename
) (struct user_namespace
*, struct inode
*, struct dentry
*,
2122 struct inode
*, struct dentry
*, unsigned int);
2123 int (*setattr
) (struct user_namespace
*, struct dentry
*,
2125 int (*getattr
) (struct user_namespace
*, const struct path
*,
2126 struct kstat
*, u32
, unsigned int);
2127 ssize_t (*listxattr
) (struct dentry
*, char *, size_t);
2128 int (*fiemap
)(struct inode
*, struct fiemap_extent_info
*, u64 start
,
2130 int (*update_time
)(struct inode
*, struct timespec64
*, int);
2131 int (*atomic_open
)(struct inode
*, struct dentry
*,
2132 struct file
*, unsigned open_flag
,
2133 umode_t create_mode
);
2134 int (*tmpfile
) (struct user_namespace
*, struct inode
*,
2135 struct dentry
*, umode_t
);
2136 int (*set_acl
)(struct user_namespace
*, struct inode
*,
2137 struct posix_acl
*, int);
2138 int (*fileattr_set
)(struct user_namespace
*mnt_userns
,
2139 struct dentry
*dentry
, struct fileattr
*fa
);
2140 int (*fileattr_get
)(struct dentry
*dentry
, struct fileattr
*fa
);
2141 } ____cacheline_aligned
;
2143 static inline ssize_t
call_read_iter(struct file
*file
, struct kiocb
*kio
,
2144 struct iov_iter
*iter
)
2146 return file
->f_op
->read_iter(kio
, iter
);
2149 static inline ssize_t
call_write_iter(struct file
*file
, struct kiocb
*kio
,
2150 struct iov_iter
*iter
)
2152 return file
->f_op
->write_iter(kio
, iter
);
2155 static inline int call_mmap(struct file
*file
, struct vm_area_struct
*vma
)
2157 return file
->f_op
->mmap(file
, vma
);
2160 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
2161 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
2162 extern ssize_t
vfs_copy_file_range(struct file
*, loff_t
, struct file
*,
2163 loff_t
, size_t, unsigned int);
2164 extern ssize_t
generic_copy_file_range(struct file
*file_in
, loff_t pos_in
,
2165 struct file
*file_out
, loff_t pos_out
,
2166 size_t len
, unsigned int flags
);
2167 extern int generic_remap_file_range_prep(struct file
*file_in
, loff_t pos_in
,
2168 struct file
*file_out
, loff_t pos_out
,
2170 unsigned int remap_flags
);
2171 extern loff_t
do_clone_file_range(struct file
*file_in
, loff_t pos_in
,
2172 struct file
*file_out
, loff_t pos_out
,
2173 loff_t len
, unsigned int remap_flags
);
2174 extern loff_t
vfs_clone_file_range(struct file
*file_in
, loff_t pos_in
,
2175 struct file
*file_out
, loff_t pos_out
,
2176 loff_t len
, unsigned int remap_flags
);
2177 extern int vfs_dedupe_file_range(struct file
*file
,
2178 struct file_dedupe_range
*same
);
2179 extern loff_t
vfs_dedupe_file_range_one(struct file
*src_file
, loff_t src_pos
,
2180 struct file
*dst_file
, loff_t dst_pos
,
2181 loff_t len
, unsigned int remap_flags
);
2184 struct super_operations
{
2185 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
2186 void (*destroy_inode
)(struct inode
*);
2187 void (*free_inode
)(struct inode
*);
2189 void (*dirty_inode
) (struct inode
*, int flags
);
2190 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
2191 int (*drop_inode
) (struct inode
*);
2192 void (*evict_inode
) (struct inode
*);
2193 void (*put_super
) (struct super_block
*);
2194 int (*sync_fs
)(struct super_block
*sb
, int wait
);
2195 int (*freeze_super
) (struct super_block
*);
2196 int (*freeze_fs
) (struct super_block
*);
2197 int (*thaw_super
) (struct super_block
*);
2198 int (*unfreeze_fs
) (struct super_block
*);
2199 int (*statfs
) (struct dentry
*, struct kstatfs
*);
2200 int (*remount_fs
) (struct super_block
*, int *, char *);
2201 void (*umount_begin
) (struct super_block
*);
2203 int (*show_options
)(struct seq_file
*, struct dentry
*);
2204 int (*show_devname
)(struct seq_file
*, struct dentry
*);
2205 int (*show_path
)(struct seq_file
*, struct dentry
*);
2206 int (*show_stats
)(struct seq_file
*, struct dentry
*);
2208 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
2209 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
2210 struct dquot
**(*get_dquots
)(struct inode
*);
2212 long (*nr_cached_objects
)(struct super_block
*,
2213 struct shrink_control
*);
2214 long (*free_cached_objects
)(struct super_block
*,
2215 struct shrink_control
*);
2219 * Inode flags - they have no relation to superblock flags now
2221 #define S_SYNC (1 << 0) /* Writes are synced at once */
2222 #define S_NOATIME (1 << 1) /* Do not update access times */
2223 #define S_APPEND (1 << 2) /* Append-only file */
2224 #define S_IMMUTABLE (1 << 3) /* Immutable file */
2225 #define S_DEAD (1 << 4) /* removed, but still open directory */
2226 #define S_NOQUOTA (1 << 5) /* Inode is not counted to quota */
2227 #define S_DIRSYNC (1 << 6) /* Directory modifications are synchronous */
2228 #define S_NOCMTIME (1 << 7) /* Do not update file c/mtime */
2229 #define S_SWAPFILE (1 << 8) /* Do not truncate: swapon got its bmaps */
2230 #define S_PRIVATE (1 << 9) /* Inode is fs-internal */
2231 #define S_IMA (1 << 10) /* Inode has an associated IMA struct */
2232 #define S_AUTOMOUNT (1 << 11) /* Automount/referral quasi-directory */
2233 #define S_NOSEC (1 << 12) /* no suid or xattr security attributes */
2234 #ifdef CONFIG_FS_DAX
2235 #define S_DAX (1 << 13) /* Direct Access, avoiding the page cache */
2237 #define S_DAX 0 /* Make all the DAX code disappear */
2239 #define S_ENCRYPTED (1 << 14) /* Encrypted file (using fs/crypto/) */
2240 #define S_CASEFOLD (1 << 15) /* Casefolded file */
2241 #define S_VERITY (1 << 16) /* Verity file (using fs/verity/) */
2244 * Note that nosuid etc flags are inode-specific: setting some file-system
2245 * flags just means all the inodes inherit those flags by default. It might be
2246 * possible to override it selectively if you really wanted to with some
2247 * ioctl() that is not currently implemented.
2249 * Exception: SB_RDONLY is always applied to the entire file system.
2251 * Unfortunately, it is possible to change a filesystems flags with it mounted
2252 * with files in use. This means that all of the inodes will not have their
2253 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
2254 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
2256 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
2258 static inline bool sb_rdonly(const struct super_block
*sb
) { return sb
->s_flags
& SB_RDONLY
; }
2259 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
2260 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
2261 ((inode)->i_flags & S_SYNC))
2262 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
2263 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
2264 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
2265 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
2266 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
2268 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
2269 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
2270 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
2271 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
2273 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
2274 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
2275 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
2276 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
2277 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
2278 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2279 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2280 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2281 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2282 #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD)
2283 #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY)
2285 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2286 (inode)->i_rdev == WHITEOUT_DEV)
2288 static inline bool HAS_UNMAPPED_ID(struct user_namespace
*mnt_userns
,
2289 struct inode
*inode
)
2291 return !uid_valid(i_uid_into_mnt(mnt_userns
, inode
)) ||
2292 !gid_valid(i_gid_into_mnt(mnt_userns
, inode
));
2295 static inline enum rw_hint
file_write_hint(struct file
*file
)
2297 if (file
->f_write_hint
!= WRITE_LIFE_NOT_SET
)
2298 return file
->f_write_hint
;
2300 return file_inode(file
)->i_write_hint
;
2303 static inline int iocb_flags(struct file
*file
);
2305 static inline u16
ki_hint_validate(enum rw_hint hint
)
2307 typeof(((struct kiocb
*)0)->ki_hint
) max_hint
= -1;
2309 if (hint
<= max_hint
)
2314 static inline void init_sync_kiocb(struct kiocb
*kiocb
, struct file
*filp
)
2316 *kiocb
= (struct kiocb
) {
2318 .ki_flags
= iocb_flags(filp
),
2319 .ki_hint
= ki_hint_validate(file_write_hint(filp
)),
2320 .ki_ioprio
= get_current_ioprio(),
2324 static inline void kiocb_clone(struct kiocb
*kiocb
, struct kiocb
*kiocb_src
,
2327 *kiocb
= (struct kiocb
) {
2329 .ki_flags
= kiocb_src
->ki_flags
,
2330 .ki_hint
= kiocb_src
->ki_hint
,
2331 .ki_ioprio
= kiocb_src
->ki_ioprio
,
2332 .ki_pos
= kiocb_src
->ki_pos
,
2337 * Inode state bits. Protected by inode->i_lock
2339 * Four bits determine the dirty state of the inode: I_DIRTY_SYNC,
2340 * I_DIRTY_DATASYNC, I_DIRTY_PAGES, and I_DIRTY_TIME.
2342 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2343 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2344 * various stages of removing an inode.
2346 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2348 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2349 * fdatasync() (unless I_DIRTY_DATASYNC is also set).
2350 * Timestamp updates are the usual cause.
2351 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2352 * these changes separately from I_DIRTY_SYNC so that we
2353 * don't have to write inode on fdatasync() when only
2354 * e.g. the timestamps have changed.
2355 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2356 * I_DIRTY_TIME The inode itself only has dirty timestamps, and the
2357 * lazytime mount option is enabled. We keep track of this
2358 * separately from I_DIRTY_SYNC in order to implement
2359 * lazytime. This gets cleared if I_DIRTY_INODE
2360 * (I_DIRTY_SYNC and/or I_DIRTY_DATASYNC) gets set. I.e.
2361 * either I_DIRTY_TIME *or* I_DIRTY_INODE can be set in
2362 * i_state, but not both. I_DIRTY_PAGES may still be set.
2363 * I_NEW Serves as both a mutex and completion notification.
2364 * New inodes set I_NEW. If two processes both create
2365 * the same inode, one of them will release its inode and
2366 * wait for I_NEW to be released before returning.
2367 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2368 * also cause waiting on I_NEW, without I_NEW actually
2369 * being set. find_inode() uses this to prevent returning
2370 * nearly-dead inodes.
2371 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2372 * is zero. I_FREEING must be set when I_WILL_FREE is
2374 * I_FREEING Set when inode is about to be freed but still has dirty
2375 * pages or buffers attached or the inode itself is still
2377 * I_CLEAR Added by clear_inode(). In this state the inode is
2378 * clean and can be destroyed. Inode keeps I_FREEING.
2380 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2381 * prohibited for many purposes. iget() must wait for
2382 * the inode to be completely released, then create it
2383 * anew. Other functions will just ignore such inodes,
2384 * if appropriate. I_NEW is used for waiting.
2386 * I_SYNC Writeback of inode is running. The bit is set during
2387 * data writeback, and cleared with a wakeup on the bit
2388 * address once it is done. The bit is also used to pin
2389 * the inode in memory for flusher thread.
2391 * I_REFERENCED Marks the inode as recently references on the LRU list.
2393 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2395 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2396 * synchronize competing switching instances and to tell
2397 * wb stat updates to grab the i_pages lock. See
2398 * inode_switch_wbs_work_fn() for details.
2400 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2401 * and work dirs among overlayfs mounts.
2403 * I_CREATING New object's inode in the middle of setting up.
2405 * I_DONTCACHE Evict inode as soon as it is not used anymore.
2407 * I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists.
2408 * Used to detect that mark_inode_dirty() should not move
2409 * inode between dirty lists.
2411 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2413 #define I_DIRTY_SYNC (1 << 0)
2414 #define I_DIRTY_DATASYNC (1 << 1)
2415 #define I_DIRTY_PAGES (1 << 2)
2417 #define I_NEW (1 << __I_NEW)
2418 #define I_WILL_FREE (1 << 4)
2419 #define I_FREEING (1 << 5)
2420 #define I_CLEAR (1 << 6)
2422 #define I_SYNC (1 << __I_SYNC)
2423 #define I_REFERENCED (1 << 8)
2424 #define __I_DIO_WAKEUP 9
2425 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2426 #define I_LINKABLE (1 << 10)
2427 #define I_DIRTY_TIME (1 << 11)
2428 #define I_WB_SWITCH (1 << 13)
2429 #define I_OVL_INUSE (1 << 14)
2430 #define I_CREATING (1 << 15)
2431 #define I_DONTCACHE (1 << 16)
2432 #define I_SYNC_QUEUED (1 << 17)
2434 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2435 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2436 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2438 extern void __mark_inode_dirty(struct inode
*, int);
2439 static inline void mark_inode_dirty(struct inode
*inode
)
2441 __mark_inode_dirty(inode
, I_DIRTY
);
2444 static inline void mark_inode_dirty_sync(struct inode
*inode
)
2446 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
2450 * Returns true if the given inode itself only has dirty timestamps (its pages
2451 * may still be dirty) and isn't currently being allocated or freed.
2452 * Filesystems should call this if when writing an inode when lazytime is
2453 * enabled, they want to opportunistically write the timestamps of other inodes
2454 * located very nearby on-disk, e.g. in the same inode block. This returns true
2455 * if the given inode is in need of such an opportunistic update. Requires
2456 * i_lock, or at least later re-checking under i_lock.
2458 static inline bool inode_is_dirtytime_only(struct inode
*inode
)
2460 return (inode
->i_state
& (I_DIRTY_TIME
| I_NEW
|
2461 I_FREEING
| I_WILL_FREE
)) == I_DIRTY_TIME
;
2464 extern void inc_nlink(struct inode
*inode
);
2465 extern void drop_nlink(struct inode
*inode
);
2466 extern void clear_nlink(struct inode
*inode
);
2467 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
2469 static inline void inode_inc_link_count(struct inode
*inode
)
2472 mark_inode_dirty(inode
);
2475 static inline void inode_dec_link_count(struct inode
*inode
)
2478 mark_inode_dirty(inode
);
2481 enum file_time_flags
{
2488 extern bool atime_needs_update(const struct path
*, struct inode
*);
2489 extern void touch_atime(const struct path
*);
2490 static inline void file_accessed(struct file
*file
)
2492 if (!(file
->f_flags
& O_NOATIME
))
2493 touch_atime(&file
->f_path
);
2496 extern int file_modified(struct file
*file
);
2498 int sync_inode(struct inode
*inode
, struct writeback_control
*wbc
);
2499 int sync_inode_metadata(struct inode
*inode
, int wait
);
2501 struct file_system_type
{
2504 #define FS_REQUIRES_DEV 1
2505 #define FS_BINARY_MOUNTDATA 2
2506 #define FS_HAS_SUBTYPE 4
2507 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2508 #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */
2509 #define FS_ALLOW_IDMAP 32 /* FS has been updated to handle vfs idmappings. */
2510 #define FS_THP_SUPPORT 8192 /* Remove once all fs converted */
2511 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2512 int (*init_fs_context
)(struct fs_context
*);
2513 const struct fs_parameter_spec
*parameters
;
2514 struct dentry
*(*mount
) (struct file_system_type
*, int,
2515 const char *, void *);
2516 void (*kill_sb
) (struct super_block
*);
2517 struct module
*owner
;
2518 struct file_system_type
* next
;
2519 struct hlist_head fs_supers
;
2521 struct lock_class_key s_lock_key
;
2522 struct lock_class_key s_umount_key
;
2523 struct lock_class_key s_vfs_rename_key
;
2524 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
2526 struct lock_class_key i_lock_key
;
2527 struct lock_class_key i_mutex_key
;
2528 struct lock_class_key invalidate_lock_key
;
2529 struct lock_class_key i_mutex_dir_key
;
2532 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2534 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
2535 int flags
, const char *dev_name
, void *data
,
2536 int (*fill_super
)(struct super_block
*, void *, int));
2537 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
2538 int flags
, void *data
,
2539 int (*fill_super
)(struct super_block
*, void *, int));
2540 extern struct dentry
*mount_nodev(struct file_system_type
*fs_type
,
2541 int flags
, void *data
,
2542 int (*fill_super
)(struct super_block
*, void *, int));
2543 extern struct dentry
*mount_subtree(struct vfsmount
*mnt
, const char *path
);
2544 void generic_shutdown_super(struct super_block
*sb
);
2545 void kill_block_super(struct super_block
*sb
);
2546 void kill_anon_super(struct super_block
*sb
);
2547 void kill_litter_super(struct super_block
*sb
);
2548 void deactivate_super(struct super_block
*sb
);
2549 void deactivate_locked_super(struct super_block
*sb
);
2550 int set_anon_super(struct super_block
*s
, void *data
);
2551 int set_anon_super_fc(struct super_block
*s
, struct fs_context
*fc
);
2552 int get_anon_bdev(dev_t
*);
2553 void free_anon_bdev(dev_t
);
2554 struct super_block
*sget_fc(struct fs_context
*fc
,
2555 int (*test
)(struct super_block
*, struct fs_context
*),
2556 int (*set
)(struct super_block
*, struct fs_context
*));
2557 struct super_block
*sget(struct file_system_type
*type
,
2558 int (*test
)(struct super_block
*,void *),
2559 int (*set
)(struct super_block
*,void *),
2560 int flags
, void *data
);
2562 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2563 #define fops_get(fops) \
2564 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2565 #define fops_put(fops) \
2566 do { if (fops) module_put((fops)->owner); } while(0)
2568 * This one is to be used *ONLY* from ->open() instances.
2569 * fops must be non-NULL, pinned down *and* module dependencies
2570 * should be sufficient to pin the caller down as well.
2572 #define replace_fops(f, fops) \
2574 struct file *__file = (f); \
2575 fops_put(__file->f_op); \
2576 BUG_ON(!(__file->f_op = (fops))); \
2579 extern int register_filesystem(struct file_system_type
*);
2580 extern int unregister_filesystem(struct file_system_type
*);
2581 extern struct vfsmount
*kern_mount(struct file_system_type
*);
2582 extern void kern_unmount(struct vfsmount
*mnt
);
2583 extern int may_umount_tree(struct vfsmount
*);
2584 extern int may_umount(struct vfsmount
*);
2585 extern long do_mount(const char *, const char __user
*,
2586 const char *, unsigned long, void *);
2587 extern struct vfsmount
*collect_mounts(const struct path
*);
2588 extern void drop_collected_mounts(struct vfsmount
*);
2589 extern int iterate_mounts(int (*)(struct vfsmount
*, void *), void *,
2591 extern int vfs_statfs(const struct path
*, struct kstatfs
*);
2592 extern int user_statfs(const char __user
*, struct kstatfs
*);
2593 extern int fd_statfs(int, struct kstatfs
*);
2594 extern int freeze_super(struct super_block
*super
);
2595 extern int thaw_super(struct super_block
*super
);
2596 extern bool our_mnt(struct vfsmount
*mnt
);
2597 extern __printf(2, 3)
2598 int super_setup_bdi_name(struct super_block
*sb
, char *fmt
, ...);
2599 extern int super_setup_bdi(struct super_block
*sb
);
2601 extern int current_umask(void);
2603 extern void ihold(struct inode
* inode
);
2604 extern void iput(struct inode
*);
2605 extern int generic_update_time(struct inode
*, struct timespec64
*, int);
2608 extern struct kobject
*fs_kobj
;
2610 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2612 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2613 extern int locks_mandatory_locked(struct file
*);
2614 extern int locks_mandatory_area(struct inode
*, struct file
*, loff_t
, loff_t
, unsigned char);
2617 * Candidates for mandatory locking have the setgid bit set
2618 * but no group execute bit - an otherwise meaningless combination.
2621 static inline int __mandatory_lock(struct inode
*ino
)
2623 return (ino
->i_mode
& (S_ISGID
| S_IXGRP
)) == S_ISGID
;
2627 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2628 * otherwise these will be advisory locks
2631 static inline int mandatory_lock(struct inode
*ino
)
2633 return IS_MANDLOCK(ino
) && __mandatory_lock(ino
);
2636 static inline int locks_verify_locked(struct file
*file
)
2638 if (mandatory_lock(locks_inode(file
)))
2639 return locks_mandatory_locked(file
);
2643 static inline int locks_verify_truncate(struct inode
*inode
,
2647 if (!inode
->i_flctx
|| !mandatory_lock(inode
))
2650 if (size
< inode
->i_size
) {
2651 return locks_mandatory_area(inode
, f
, size
, inode
->i_size
- 1,
2654 return locks_mandatory_area(inode
, f
, inode
->i_size
, size
- 1,
2659 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2661 static inline int locks_mandatory_locked(struct file
*file
)
2666 static inline int locks_mandatory_area(struct inode
*inode
, struct file
*filp
,
2667 loff_t start
, loff_t end
, unsigned char type
)
2672 static inline int __mandatory_lock(struct inode
*inode
)
2677 static inline int mandatory_lock(struct inode
*inode
)
2682 static inline int locks_verify_locked(struct file
*file
)
2687 static inline int locks_verify_truncate(struct inode
*inode
, struct file
*filp
,
2693 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2696 #ifdef CONFIG_FILE_LOCKING
2697 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2700 * Since this check is lockless, we must ensure that any refcounts
2701 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2702 * could end up racing with tasks trying to set a new lease on this
2706 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2707 return __break_lease(inode
, mode
, FL_LEASE
);
2711 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2714 * Since this check is lockless, we must ensure that any refcounts
2715 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2716 * could end up racing with tasks trying to set a new lease on this
2720 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2721 return __break_lease(inode
, mode
, FL_DELEG
);
2725 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2729 ret
= break_deleg(inode
, O_WRONLY
|O_NONBLOCK
);
2730 if (ret
== -EWOULDBLOCK
&& delegated_inode
) {
2731 *delegated_inode
= inode
;
2737 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2741 ret
= break_deleg(*delegated_inode
, O_WRONLY
);
2742 iput(*delegated_inode
);
2743 *delegated_inode
= NULL
;
2747 static inline int break_layout(struct inode
*inode
, bool wait
)
2750 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2751 return __break_lease(inode
,
2752 wait
? O_WRONLY
: O_WRONLY
| O_NONBLOCK
,
2757 #else /* !CONFIG_FILE_LOCKING */
2758 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2763 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2768 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2773 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2779 static inline int break_layout(struct inode
*inode
, bool wait
)
2784 #endif /* CONFIG_FILE_LOCKING */
2789 const char *name
; /* pointer to actual string */
2790 const __user
char *uptr
; /* original userland pointer */
2792 struct audit_names
*aname
;
2795 static_assert(offsetof(struct filename
, iname
) % sizeof(long) == 0);
2797 static inline struct user_namespace
*file_mnt_user_ns(struct file
*file
)
2799 return mnt_user_ns(file
->f_path
.mnt
);
2801 extern long vfs_truncate(const struct path
*, loff_t
);
2802 int do_truncate(struct user_namespace
*, struct dentry
*, loff_t start
,
2803 unsigned int time_attrs
, struct file
*filp
);
2804 extern int vfs_fallocate(struct file
*file
, int mode
, loff_t offset
,
2806 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
2808 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
2809 extern struct file
*filp_open(const char *, int, umode_t
);
2810 extern struct file
*file_open_root(const struct path
*,
2811 const char *, int, umode_t
);
2812 static inline struct file
*file_open_root_mnt(struct vfsmount
*mnt
,
2813 const char *name
, int flags
, umode_t mode
)
2815 return file_open_root(&(struct path
){.mnt
= mnt
, .dentry
= mnt
->mnt_root
},
2818 extern struct file
* dentry_open(const struct path
*, int, const struct cred
*);
2819 extern struct file
* open_with_fake_path(const struct path
*, int,
2820 struct inode
*, const struct cred
*);
2821 static inline struct file
*file_clone_open(struct file
*file
)
2823 return dentry_open(&file
->f_path
, file
->f_flags
, file
->f_cred
);
2825 extern int filp_close(struct file
*, fl_owner_t id
);
2827 extern struct filename
*getname_flags(const char __user
*, int, int *);
2828 extern struct filename
*getname(const char __user
*);
2829 extern struct filename
*getname_kernel(const char *);
2830 extern void putname(struct filename
*name
);
2832 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
2833 int (*open
)(struct inode
*, struct file
*));
2834 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
2837 extern void __init
vfs_caches_init_early(void);
2838 extern void __init
vfs_caches_init(void);
2840 extern struct kmem_cache
*names_cachep
;
2842 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2843 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2845 extern struct super_block
*blockdev_superblock
;
2846 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2848 return IS_ENABLED(CONFIG_BLOCK
) && sb
== blockdev_superblock
;
2851 void emergency_thaw_all(void);
2852 extern int sync_filesystem(struct super_block
*);
2853 extern const struct file_operations def_blk_fops
;
2854 extern const struct file_operations def_chr_fops
;
2857 #define CHRDEV_MAJOR_MAX 512
2858 /* Marks the bottom of the first segment of free char majors */
2859 #define CHRDEV_MAJOR_DYN_END 234
2860 /* Marks the top and bottom of the second segment of free char majors */
2861 #define CHRDEV_MAJOR_DYN_EXT_START 511
2862 #define CHRDEV_MAJOR_DYN_EXT_END 384
2864 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
2865 extern int register_chrdev_region(dev_t
, unsigned, const char *);
2866 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
2867 unsigned int count
, const char *name
,
2868 const struct file_operations
*fops
);
2869 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
2870 unsigned int count
, const char *name
);
2871 extern void unregister_chrdev_region(dev_t
, unsigned);
2872 extern void chrdev_show(struct seq_file
*,off_t
);
2874 static inline int register_chrdev(unsigned int major
, const char *name
,
2875 const struct file_operations
*fops
)
2877 return __register_chrdev(major
, 0, 256, name
, fops
);
2880 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2882 __unregister_chrdev(major
, 0, 256, name
);
2885 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2887 /* Invalid inode operations -- fs/bad_inode.c */
2888 extern void make_bad_inode(struct inode
*);
2889 extern bool is_bad_inode(struct inode
*);
2891 unsigned long invalidate_mapping_pages(struct address_space
*mapping
,
2892 pgoff_t start
, pgoff_t end
);
2894 void invalidate_mapping_pagevec(struct address_space
*mapping
,
2895 pgoff_t start
, pgoff_t end
,
2896 unsigned long *nr_pagevec
);
2898 static inline void invalidate_remote_inode(struct inode
*inode
)
2900 if (S_ISREG(inode
->i_mode
) || S_ISDIR(inode
->i_mode
) ||
2901 S_ISLNK(inode
->i_mode
))
2902 invalidate_mapping_pages(inode
->i_mapping
, 0, -1);
2904 extern int invalidate_inode_pages2(struct address_space
*mapping
);
2905 extern int invalidate_inode_pages2_range(struct address_space
*mapping
,
2906 pgoff_t start
, pgoff_t end
);
2907 extern int write_inode_now(struct inode
*, int);
2908 extern int filemap_fdatawrite(struct address_space
*);
2909 extern int filemap_flush(struct address_space
*);
2910 extern int filemap_fdatawait_keep_errors(struct address_space
*mapping
);
2911 extern int filemap_fdatawait_range(struct address_space
*, loff_t lstart
,
2913 extern int filemap_fdatawait_range_keep_errors(struct address_space
*mapping
,
2914 loff_t start_byte
, loff_t end_byte
);
2916 static inline int filemap_fdatawait(struct address_space
*mapping
)
2918 return filemap_fdatawait_range(mapping
, 0, LLONG_MAX
);
2921 extern bool filemap_range_has_page(struct address_space
*, loff_t lstart
,
2923 extern bool filemap_range_needs_writeback(struct address_space
*,
2924 loff_t lstart
, loff_t lend
);
2925 extern int filemap_write_and_wait_range(struct address_space
*mapping
,
2926 loff_t lstart
, loff_t lend
);
2927 extern int __filemap_fdatawrite_range(struct address_space
*mapping
,
2928 loff_t start
, loff_t end
, int sync_mode
);
2929 extern int filemap_fdatawrite_range(struct address_space
*mapping
,
2930 loff_t start
, loff_t end
);
2931 extern int filemap_check_errors(struct address_space
*mapping
);
2932 extern void __filemap_set_wb_err(struct address_space
*mapping
, int err
);
2934 static inline int filemap_write_and_wait(struct address_space
*mapping
)
2936 return filemap_write_and_wait_range(mapping
, 0, LLONG_MAX
);
2939 extern int __must_check
file_fdatawait_range(struct file
*file
, loff_t lstart
,
2941 extern int __must_check
file_check_and_advance_wb_err(struct file
*file
);
2942 extern int __must_check
file_write_and_wait_range(struct file
*file
,
2943 loff_t start
, loff_t end
);
2945 static inline int file_write_and_wait(struct file
*file
)
2947 return file_write_and_wait_range(file
, 0, LLONG_MAX
);
2951 * filemap_set_wb_err - set a writeback error on an address_space
2952 * @mapping: mapping in which to set writeback error
2953 * @err: error to be set in mapping
2955 * When writeback fails in some way, we must record that error so that
2956 * userspace can be informed when fsync and the like are called. We endeavor
2957 * to report errors on any file that was open at the time of the error. Some
2958 * internal callers also need to know when writeback errors have occurred.
2960 * When a writeback error occurs, most filesystems will want to call
2961 * filemap_set_wb_err to record the error in the mapping so that it will be
2962 * automatically reported whenever fsync is called on the file.
2964 static inline void filemap_set_wb_err(struct address_space
*mapping
, int err
)
2966 /* Fastpath for common case of no error */
2968 __filemap_set_wb_err(mapping
, err
);
2972 * filemap_check_wb_err - has an error occurred since the mark was sampled?
2973 * @mapping: mapping to check for writeback errors
2974 * @since: previously-sampled errseq_t
2976 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2977 * the given value was sampled.
2979 * If it has then report the latest error set, otherwise return 0.
2981 static inline int filemap_check_wb_err(struct address_space
*mapping
,
2984 return errseq_check(&mapping
->wb_err
, since
);
2988 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2989 * @mapping: mapping to be sampled
2991 * Writeback errors are always reported relative to a particular sample point
2992 * in the past. This function provides those sample points.
2994 static inline errseq_t
filemap_sample_wb_err(struct address_space
*mapping
)
2996 return errseq_sample(&mapping
->wb_err
);
3000 * file_sample_sb_err - sample the current errseq_t to test for later errors
3001 * @file: file pointer to be sampled
3003 * Grab the most current superblock-level errseq_t value for the given
3006 static inline errseq_t
file_sample_sb_err(struct file
*file
)
3008 return errseq_sample(&file
->f_path
.dentry
->d_sb
->s_wb_err
);
3011 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
3013 extern int vfs_fsync(struct file
*file
, int datasync
);
3015 extern int sync_file_range(struct file
*file
, loff_t offset
, loff_t nbytes
,
3016 unsigned int flags
);
3019 * Sync the bytes written if this was a synchronous write. Expect ki_pos
3020 * to already be updated for the write, and will return either the amount
3021 * of bytes passed in, or an error if syncing the file failed.
3023 static inline ssize_t
generic_write_sync(struct kiocb
*iocb
, ssize_t count
)
3025 if (iocb
->ki_flags
& IOCB_DSYNC
) {
3026 int ret
= vfs_fsync_range(iocb
->ki_filp
,
3027 iocb
->ki_pos
- count
, iocb
->ki_pos
- 1,
3028 (iocb
->ki_flags
& IOCB_SYNC
) ? 0 : 1);
3036 extern void emergency_sync(void);
3037 extern void emergency_remount(void);
3040 extern int bmap(struct inode
*inode
, sector_t
*block
);
3042 static inline int bmap(struct inode
*inode
, sector_t
*block
)
3048 int notify_change(struct user_namespace
*, struct dentry
*,
3049 struct iattr
*, struct inode
**);
3050 int inode_permission(struct user_namespace
*, struct inode
*, int);
3051 int generic_permission(struct user_namespace
*, struct inode
*, int);
3052 static inline int file_permission(struct file
*file
, int mask
)
3054 return inode_permission(file_mnt_user_ns(file
),
3055 file_inode(file
), mask
);
3057 static inline int path_permission(const struct path
*path
, int mask
)
3059 return inode_permission(mnt_user_ns(path
->mnt
),
3060 d_inode(path
->dentry
), mask
);
3062 int __check_sticky(struct user_namespace
*mnt_userns
, struct inode
*dir
,
3063 struct inode
*inode
);
3065 static inline bool execute_ok(struct inode
*inode
)
3067 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
3070 static inline bool inode_wrong_type(const struct inode
*inode
, umode_t mode
)
3072 return (inode
->i_mode
^ mode
) & S_IFMT
;
3075 static inline void file_start_write(struct file
*file
)
3077 if (!S_ISREG(file_inode(file
)->i_mode
))
3079 sb_start_write(file_inode(file
)->i_sb
);
3082 static inline bool file_start_write_trylock(struct file
*file
)
3084 if (!S_ISREG(file_inode(file
)->i_mode
))
3086 return sb_start_write_trylock(file_inode(file
)->i_sb
);
3089 static inline void file_end_write(struct file
*file
)
3091 if (!S_ISREG(file_inode(file
)->i_mode
))
3093 __sb_end_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
);
3097 * get_write_access() gets write permission for a file.
3098 * put_write_access() releases this write permission.
3099 * This is used for regular files.
3100 * We cannot support write (and maybe mmap read-write shared) accesses and
3101 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
3102 * can have the following values:
3103 * 0: no writers, no VM_DENYWRITE mappings
3104 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
3105 * > 0: (i_writecount) users are writing to the file.
3107 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
3108 * except for the cases where we don't hold i_writecount yet. Then we need to
3109 * use {get,deny}_write_access() - these functions check the sign and refuse
3110 * to do the change if sign is wrong.
3112 static inline int get_write_access(struct inode
*inode
)
3114 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
3116 static inline int deny_write_access(struct file
*file
)
3118 struct inode
*inode
= file_inode(file
);
3119 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
3121 static inline void put_write_access(struct inode
* inode
)
3123 atomic_dec(&inode
->i_writecount
);
3125 static inline void allow_write_access(struct file
*file
)
3128 atomic_inc(&file_inode(file
)->i_writecount
);
3130 static inline bool inode_is_open_for_write(const struct inode
*inode
)
3132 return atomic_read(&inode
->i_writecount
) > 0;
3135 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
3136 static inline void i_readcount_dec(struct inode
*inode
)
3138 BUG_ON(!atomic_read(&inode
->i_readcount
));
3139 atomic_dec(&inode
->i_readcount
);
3141 static inline void i_readcount_inc(struct inode
*inode
)
3143 atomic_inc(&inode
->i_readcount
);
3146 static inline void i_readcount_dec(struct inode
*inode
)
3150 static inline void i_readcount_inc(struct inode
*inode
)
3155 extern int do_pipe_flags(int *, int);
3157 extern ssize_t
kernel_read(struct file
*, void *, size_t, loff_t
*);
3158 ssize_t
__kernel_read(struct file
*file
, void *buf
, size_t count
, loff_t
*pos
);
3159 extern ssize_t
kernel_write(struct file
*, const void *, size_t, loff_t
*);
3160 extern ssize_t
__kernel_write(struct file
*, const void *, size_t, loff_t
*);
3161 extern struct file
* open_exec(const char *);
3163 /* fs/dcache.c -- generic fs support functions */
3164 extern bool is_subdir(struct dentry
*, struct dentry
*);
3165 extern bool path_is_under(const struct path
*, const struct path
*);
3167 extern char *file_path(struct file
*, char *, int);
3169 #include <linux/err.h>
3171 /* needed for stackable file system support */
3172 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int whence
);
3174 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int whence
);
3176 extern int inode_init_always(struct super_block
*, struct inode
*);
3177 extern void inode_init_once(struct inode
*);
3178 extern void address_space_init_once(struct address_space
*mapping
);
3179 extern struct inode
* igrab(struct inode
*);
3180 extern ino_t
iunique(struct super_block
*, ino_t
);
3181 extern int inode_needs_sync(struct inode
*inode
);
3182 extern int generic_delete_inode(struct inode
*inode
);
3183 static inline int generic_drop_inode(struct inode
*inode
)
3185 return !inode
->i_nlink
|| inode_unhashed(inode
);
3187 extern void d_mark_dontcache(struct inode
*inode
);
3189 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
3190 unsigned long hashval
, int (*test
)(struct inode
*, void *),
3192 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
3193 int (*test
)(struct inode
*, void *), void *data
);
3194 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
3196 extern struct inode
*inode_insert5(struct inode
*inode
, unsigned long hashval
,
3197 int (*test
)(struct inode
*, void *),
3198 int (*set
)(struct inode
*, void *),
3200 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
3201 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
3202 extern struct inode
*find_inode_nowait(struct super_block
*,
3204 int (*match
)(struct inode
*,
3205 unsigned long, void *),
3207 extern struct inode
*find_inode_rcu(struct super_block
*, unsigned long,
3208 int (*)(struct inode
*, void *), void *);
3209 extern struct inode
*find_inode_by_ino_rcu(struct super_block
*, unsigned long);
3210 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
3211 extern int insert_inode_locked(struct inode
*);
3212 #ifdef CONFIG_DEBUG_LOCK_ALLOC
3213 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
3215 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
3217 extern void unlock_new_inode(struct inode
*);
3218 extern void discard_new_inode(struct inode
*);
3219 extern unsigned int get_next_ino(void);
3220 extern void evict_inodes(struct super_block
*sb
);
3223 * Userspace may rely on the the inode number being non-zero. For example, glibc
3224 * simply ignores files with zero i_ino in unlink() and other places.
3226 * As an additional complication, if userspace was compiled with
3227 * _FILE_OFFSET_BITS=32 on a 64-bit kernel we'll only end up reading out the
3228 * lower 32 bits, so we need to check that those aren't zero explicitly. With
3229 * _FILE_OFFSET_BITS=64, this may cause some harmless false-negatives, but
3230 * better safe than sorry.
3232 static inline bool is_zero_ino(ino_t ino
)
3234 return (u32
)ino
== 0;
3237 extern void __iget(struct inode
* inode
);
3238 extern void iget_failed(struct inode
*);
3239 extern void clear_inode(struct inode
*);
3240 extern void __destroy_inode(struct inode
*);
3241 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
3242 extern struct inode
*new_inode(struct super_block
*sb
);
3243 extern void free_inode_nonrcu(struct inode
*inode
);
3244 extern int should_remove_suid(struct dentry
*);
3245 extern int file_remove_privs(struct file
*);
3247 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
3248 static inline void insert_inode_hash(struct inode
*inode
)
3250 __insert_inode_hash(inode
, inode
->i_ino
);
3253 extern void __remove_inode_hash(struct inode
*);
3254 static inline void remove_inode_hash(struct inode
*inode
)
3256 if (!inode_unhashed(inode
) && !hlist_fake(&inode
->i_hash
))
3257 __remove_inode_hash(inode
);
3260 extern void inode_sb_list_add(struct inode
*inode
);
3262 extern int sb_set_blocksize(struct super_block
*, int);
3263 extern int sb_min_blocksize(struct super_block
*, int);
3265 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
3266 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
3267 extern ssize_t
generic_write_checks(struct kiocb
*, struct iov_iter
*);
3268 extern int generic_write_check_limits(struct file
*file
, loff_t pos
,
3270 extern int generic_file_rw_checks(struct file
*file_in
, struct file
*file_out
);
3271 ssize_t
filemap_read(struct kiocb
*iocb
, struct iov_iter
*to
,
3272 ssize_t already_read
);
3273 extern ssize_t
generic_file_read_iter(struct kiocb
*, struct iov_iter
*);
3274 extern ssize_t
__generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
3275 extern ssize_t
generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
3276 extern ssize_t
generic_file_direct_write(struct kiocb
*, struct iov_iter
*);
3277 extern ssize_t
generic_perform_write(struct file
*, struct iov_iter
*, loff_t
);
3279 ssize_t
vfs_iter_read(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
3281 ssize_t
vfs_iter_write(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
3283 ssize_t
vfs_iocb_iter_read(struct file
*file
, struct kiocb
*iocb
,
3284 struct iov_iter
*iter
);
3285 ssize_t
vfs_iocb_iter_write(struct file
*file
, struct kiocb
*iocb
,
3286 struct iov_iter
*iter
);
3288 /* fs/block_dev.c */
3289 extern int blkdev_fsync(struct file
*filp
, loff_t start
, loff_t end
,
3293 extern ssize_t
generic_file_splice_read(struct file
*, loff_t
*,
3294 struct pipe_inode_info
*, size_t, unsigned int);
3295 extern ssize_t
iter_file_splice_write(struct pipe_inode_info
*,
3296 struct file
*, loff_t
*, size_t, unsigned int);
3297 extern ssize_t
generic_splice_sendpage(struct pipe_inode_info
*pipe
,
3298 struct file
*out
, loff_t
*, size_t len
, unsigned int flags
);
3299 extern long do_splice_direct(struct file
*in
, loff_t
*ppos
, struct file
*out
,
3300 loff_t
*opos
, size_t len
, unsigned int flags
);
3304 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
3305 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int whence
);
3306 extern loff_t
no_llseek(struct file
*file
, loff_t offset
, int whence
);
3307 extern loff_t
vfs_setpos(struct file
*file
, loff_t offset
, loff_t maxsize
);
3308 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int whence
);
3309 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
3310 int whence
, loff_t maxsize
, loff_t eof
);
3311 extern loff_t
fixed_size_llseek(struct file
*file
, loff_t offset
,
3312 int whence
, loff_t size
);
3313 extern loff_t
no_seek_end_llseek_size(struct file
*, loff_t
, int, loff_t
);
3314 extern loff_t
no_seek_end_llseek(struct file
*, loff_t
, int);
3315 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
3316 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
3317 extern int stream_open(struct inode
* inode
, struct file
* filp
);
3320 typedef void (dio_submit_t
)(struct bio
*bio
, struct inode
*inode
,
3321 loff_t file_offset
);
3324 /* need locking between buffered and direct access */
3327 /* filesystem does not support filling holes */
3328 DIO_SKIP_HOLES
= 0x02,
3331 ssize_t
__blockdev_direct_IO(struct kiocb
*iocb
, struct inode
*inode
,
3332 struct block_device
*bdev
, struct iov_iter
*iter
,
3333 get_block_t get_block
,
3334 dio_iodone_t end_io
, dio_submit_t submit_io
,
3337 static inline ssize_t
blockdev_direct_IO(struct kiocb
*iocb
,
3338 struct inode
*inode
,
3339 struct iov_iter
*iter
,
3340 get_block_t get_block
)
3342 return __blockdev_direct_IO(iocb
, inode
, inode
->i_sb
->s_bdev
, iter
,
3343 get_block
, NULL
, NULL
, DIO_LOCKING
| DIO_SKIP_HOLES
);
3347 void inode_dio_wait(struct inode
*inode
);
3350 * inode_dio_begin - signal start of a direct I/O requests
3351 * @inode: inode the direct I/O happens on
3353 * This is called once we've finished processing a direct I/O request,
3354 * and is used to wake up callers waiting for direct I/O to be quiesced.
3356 static inline void inode_dio_begin(struct inode
*inode
)
3358 atomic_inc(&inode
->i_dio_count
);
3362 * inode_dio_end - signal finish of a direct I/O requests
3363 * @inode: inode the direct I/O happens on
3365 * This is called once we've finished processing a direct I/O request,
3366 * and is used to wake up callers waiting for direct I/O to be quiesced.
3368 static inline void inode_dio_end(struct inode
*inode
)
3370 if (atomic_dec_and_test(&inode
->i_dio_count
))
3371 wake_up_bit(&inode
->i_state
, __I_DIO_WAKEUP
);
3375 * Warn about a page cache invalidation failure diring a direct I/O write.
3377 void dio_warn_stale_pagecache(struct file
*filp
);
3379 extern void inode_set_flags(struct inode
*inode
, unsigned int flags
,
3382 extern const struct file_operations generic_ro_fops
;
3384 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3386 extern int readlink_copy(char __user
*, int, const char *);
3387 extern int page_readlink(struct dentry
*, char __user
*, int);
3388 extern const char *page_get_link(struct dentry
*, struct inode
*,
3389 struct delayed_call
*);
3390 extern void page_put_link(void *);
3391 extern int __page_symlink(struct inode
*inode
, const char *symname
, int len
,
3393 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
3394 extern const struct inode_operations page_symlink_inode_operations
;
3395 extern void kfree_link(void *);
3396 void generic_fillattr(struct user_namespace
*, struct inode
*, struct kstat
*);
3397 extern int vfs_getattr_nosec(const struct path
*, struct kstat
*, u32
, unsigned int);
3398 extern int vfs_getattr(const struct path
*, struct kstat
*, u32
, unsigned int);
3399 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
3400 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
3401 void __inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
3402 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
3403 static inline loff_t
__inode_get_bytes(struct inode
*inode
)
3405 return (((loff_t
)inode
->i_blocks
) << 9) + inode
->i_bytes
;
3407 loff_t
inode_get_bytes(struct inode
*inode
);
3408 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
3409 const char *simple_get_link(struct dentry
*, struct inode
*,
3410 struct delayed_call
*);
3411 extern const struct inode_operations simple_symlink_inode_operations
;
3413 extern int iterate_dir(struct file
*, struct dir_context
*);
3415 int vfs_fstatat(int dfd
, const char __user
*filename
, struct kstat
*stat
,
3417 int vfs_fstat(int fd
, struct kstat
*stat
);
3419 static inline int vfs_stat(const char __user
*filename
, struct kstat
*stat
)
3421 return vfs_fstatat(AT_FDCWD
, filename
, stat
, 0);
3423 static inline int vfs_lstat(const char __user
*name
, struct kstat
*stat
)
3425 return vfs_fstatat(AT_FDCWD
, name
, stat
, AT_SYMLINK_NOFOLLOW
);
3428 extern const char *vfs_get_link(struct dentry
*, struct delayed_call
*);
3429 extern int vfs_readlink(struct dentry
*, char __user
*, int);
3431 extern struct file_system_type
*get_filesystem(struct file_system_type
*fs
);
3432 extern void put_filesystem(struct file_system_type
*fs
);
3433 extern struct file_system_type
*get_fs_type(const char *name
);
3434 extern struct super_block
*get_super(struct block_device
*);
3435 extern struct super_block
*get_active_super(struct block_device
*bdev
);
3436 extern void drop_super(struct super_block
*sb
);
3437 extern void drop_super_exclusive(struct super_block
*sb
);
3438 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
3439 extern void iterate_supers_type(struct file_system_type
*,
3440 void (*)(struct super_block
*, void *), void *);
3442 extern int dcache_dir_open(struct inode
*, struct file
*);
3443 extern int dcache_dir_close(struct inode
*, struct file
*);
3444 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
3445 extern int dcache_readdir(struct file
*, struct dir_context
*);
3446 extern int simple_setattr(struct user_namespace
*, struct dentry
*,
3448 extern int simple_getattr(struct user_namespace
*, const struct path
*,
3449 struct kstat
*, u32
, unsigned int);
3450 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
3451 extern int simple_open(struct inode
*inode
, struct file
*file
);
3452 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
3453 extern int simple_unlink(struct inode
*, struct dentry
*);
3454 extern int simple_rmdir(struct inode
*, struct dentry
*);
3455 extern int simple_rename(struct user_namespace
*, struct inode
*,
3456 struct dentry
*, struct inode
*, struct dentry
*,
3458 extern void simple_recursive_removal(struct dentry
*,
3459 void (*callback
)(struct dentry
*));
3460 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
3461 extern void noop_invalidatepage(struct page
*page
, unsigned int offset
,
3462 unsigned int length
);
3463 extern ssize_t
noop_direct_IO(struct kiocb
*iocb
, struct iov_iter
*iter
);
3464 extern int simple_empty(struct dentry
*);
3465 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
3466 loff_t pos
, unsigned len
, unsigned flags
,
3467 struct page
**pagep
, void **fsdata
);
3468 extern const struct address_space_operations ram_aops
;
3469 extern int always_delete_dentry(const struct dentry
*);
3470 extern struct inode
*alloc_anon_inode(struct super_block
*);
3471 extern int simple_nosetlease(struct file
*, long, struct file_lock
**, void **);
3472 extern const struct dentry_operations simple_dentry_operations
;
3474 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
3475 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
3476 extern const struct file_operations simple_dir_operations
;
3477 extern const struct inode_operations simple_dir_inode_operations
;
3478 extern void make_empty_dir_inode(struct inode
*inode
);
3479 extern bool is_empty_dir_inode(struct inode
*inode
);
3480 struct tree_descr
{ const char *name
; const struct file_operations
*ops
; int mode
; };
3481 struct dentry
*d_alloc_name(struct dentry
*, const char *);
3482 extern int simple_fill_super(struct super_block
*, unsigned long,
3483 const struct tree_descr
*);
3484 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
3485 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
3487 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
3488 loff_t
*ppos
, const void *from
, size_t available
);
3489 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
3490 const void __user
*from
, size_t count
);
3492 extern int __generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3493 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3495 extern int generic_check_addressable(unsigned, u64
);
3497 extern void generic_set_encrypted_ci_d_ops(struct dentry
*dentry
);
3499 #ifdef CONFIG_MIGRATION
3500 extern int buffer_migrate_page(struct address_space
*,
3501 struct page
*, struct page
*,
3503 extern int buffer_migrate_page_norefs(struct address_space
*,
3504 struct page
*, struct page
*,
3507 #define buffer_migrate_page NULL
3508 #define buffer_migrate_page_norefs NULL
3511 int setattr_prepare(struct user_namespace
*, struct dentry
*, struct iattr
*);
3512 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
3513 void setattr_copy(struct user_namespace
*, struct inode
*inode
,
3514 const struct iattr
*attr
);
3516 extern int file_update_time(struct file
*file
);
3518 static inline bool vma_is_dax(const struct vm_area_struct
*vma
)
3520 return vma
->vm_file
&& IS_DAX(vma
->vm_file
->f_mapping
->host
);
3523 static inline bool vma_is_fsdax(struct vm_area_struct
*vma
)
3525 struct inode
*inode
;
3527 if (!IS_ENABLED(CONFIG_FS_DAX
) || !vma
->vm_file
)
3529 if (!vma_is_dax(vma
))
3531 inode
= file_inode(vma
->vm_file
);
3532 if (S_ISCHR(inode
->i_mode
))
3533 return false; /* device-dax */
3537 static inline int iocb_flags(struct file
*file
)
3540 if (file
->f_flags
& O_APPEND
)
3542 if (file
->f_flags
& O_DIRECT
)
3544 if ((file
->f_flags
& O_DSYNC
) || IS_SYNC(file
->f_mapping
->host
))
3546 if (file
->f_flags
& __O_SYNC
)
3551 static inline int kiocb_set_rw_flags(struct kiocb
*ki
, rwf_t flags
)
3553 int kiocb_flags
= 0;
3555 /* make sure there's no overlap between RWF and private IOCB flags */
3556 BUILD_BUG_ON((__force
int) RWF_SUPPORTED
& IOCB_EVENTFD
);
3560 if (unlikely(flags
& ~RWF_SUPPORTED
))
3563 if (flags
& RWF_NOWAIT
) {
3564 if (!(ki
->ki_filp
->f_mode
& FMODE_NOWAIT
))
3566 kiocb_flags
|= IOCB_NOIO
;
3568 kiocb_flags
|= (__force
int) (flags
& RWF_SUPPORTED
);
3569 if (flags
& RWF_SYNC
)
3570 kiocb_flags
|= IOCB_DSYNC
;
3572 ki
->ki_flags
|= kiocb_flags
;
3576 static inline ino_t
parent_ino(struct dentry
*dentry
)
3581 * Don't strictly need d_lock here? If the parent ino could change
3582 * then surely we'd have a deeper race in the caller?
3584 spin_lock(&dentry
->d_lock
);
3585 res
= dentry
->d_parent
->d_inode
->i_ino
;
3586 spin_unlock(&dentry
->d_lock
);
3590 /* Transaction based IO helpers */
3593 * An argresp is stored in an allocated page and holds the
3594 * size of the argument or response, along with its content
3596 struct simple_transaction_argresp
{
3601 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3603 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
3605 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
3606 size_t size
, loff_t
*pos
);
3607 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
3609 void simple_transaction_set(struct file
*file
, size_t n
);
3612 * simple attribute files
3614 * These attributes behave similar to those in sysfs:
3616 * Writing to an attribute immediately sets a value, an open file can be
3617 * written to multiple times.
3619 * Reading from an attribute creates a buffer from the value that might get
3620 * read with multiple read calls. When the attribute has been read
3621 * completely, no further read calls are possible until the file is opened
3624 * All attributes contain a text representation of a numeric value
3625 * that are accessed with the get() and set() functions.
3627 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3628 static int __fops ## _open(struct inode *inode, struct file *file) \
3630 __simple_attr_check_format(__fmt, 0ull); \
3631 return simple_attr_open(inode, file, __get, __set, __fmt); \
3633 static const struct file_operations __fops = { \
3634 .owner = THIS_MODULE, \
3635 .open = __fops ## _open, \
3636 .release = simple_attr_release, \
3637 .read = simple_attr_read, \
3638 .write = simple_attr_write, \
3639 .llseek = generic_file_llseek, \
3642 static inline __printf(1, 2)
3643 void __simple_attr_check_format(const char *fmt
, ...)
3645 /* don't do anything, just let the compiler check the arguments; */
3648 int simple_attr_open(struct inode
*inode
, struct file
*file
,
3649 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
3651 int simple_attr_release(struct inode
*inode
, struct file
*file
);
3652 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
3653 size_t len
, loff_t
*ppos
);
3654 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
3655 size_t len
, loff_t
*ppos
);
3658 int proc_nr_files(struct ctl_table
*table
, int write
,
3659 void *buffer
, size_t *lenp
, loff_t
*ppos
);
3660 int proc_nr_dentry(struct ctl_table
*table
, int write
,
3661 void *buffer
, size_t *lenp
, loff_t
*ppos
);
3662 int proc_nr_inodes(struct ctl_table
*table
, int write
,
3663 void *buffer
, size_t *lenp
, loff_t
*ppos
);
3664 int __init
get_filesystem_list(char *buf
);
3666 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3667 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3669 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3670 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3671 (flag & __FMODE_NONOTIFY)))
3673 static inline bool is_sxid(umode_t mode
)
3675 return (mode
& S_ISUID
) || ((mode
& S_ISGID
) && (mode
& S_IXGRP
));
3678 static inline int check_sticky(struct user_namespace
*mnt_userns
,
3679 struct inode
*dir
, struct inode
*inode
)
3681 if (!(dir
->i_mode
& S_ISVTX
))
3684 return __check_sticky(mnt_userns
, dir
, inode
);
3687 static inline void inode_has_no_xattr(struct inode
*inode
)
3689 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& SB_NOSEC
))
3690 inode
->i_flags
|= S_NOSEC
;
3693 static inline bool is_root_inode(struct inode
*inode
)
3695 return inode
== inode
->i_sb
->s_root
->d_inode
;
3698 static inline bool dir_emit(struct dir_context
*ctx
,
3699 const char *name
, int namelen
,
3700 u64 ino
, unsigned type
)
3702 return ctx
->actor(ctx
, name
, namelen
, ctx
->pos
, ino
, type
) == 0;
3704 static inline bool dir_emit_dot(struct file
*file
, struct dir_context
*ctx
)
3706 return ctx
->actor(ctx
, ".", 1, ctx
->pos
,
3707 file
->f_path
.dentry
->d_inode
->i_ino
, DT_DIR
) == 0;
3709 static inline bool dir_emit_dotdot(struct file
*file
, struct dir_context
*ctx
)
3711 return ctx
->actor(ctx
, "..", 2, ctx
->pos
,
3712 parent_ino(file
->f_path
.dentry
), DT_DIR
) == 0;
3714 static inline bool dir_emit_dots(struct file
*file
, struct dir_context
*ctx
)
3716 if (ctx
->pos
== 0) {
3717 if (!dir_emit_dot(file
, ctx
))
3721 if (ctx
->pos
== 1) {
3722 if (!dir_emit_dotdot(file
, ctx
))
3728 static inline bool dir_relax(struct inode
*inode
)
3730 inode_unlock(inode
);
3732 return !IS_DEADDIR(inode
);
3735 static inline bool dir_relax_shared(struct inode
*inode
)
3737 inode_unlock_shared(inode
);
3738 inode_lock_shared(inode
);
3739 return !IS_DEADDIR(inode
);
3742 extern bool path_noexec(const struct path
*path
);
3743 extern void inode_nohighmem(struct inode
*inode
);
3746 extern int vfs_fadvise(struct file
*file
, loff_t offset
, loff_t len
,
3748 extern int generic_fadvise(struct file
*file
, loff_t offset
, loff_t len
,
3752 * Flush file data before changing attributes. Caller must hold any locks
3753 * required to prevent further writes to this file until we're done setting
3756 static inline int inode_drain_writes(struct inode
*inode
)
3758 inode_dio_wait(inode
);
3759 return filemap_write_and_wait(inode
->i_mapping
);
3762 #endif /* _LINUX_FS_H */