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 * @gfp_mask: Memory allocation flags to use for allocating pages.
440 * @i_mmap_writable: Number of VM_SHARED mappings.
441 * @nr_thps: Number of THPs in the pagecache (non-shmem only).
442 * @i_mmap: Tree of private and shared mappings.
443 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
444 * @nrpages: Number of page entries, protected by the i_pages lock.
445 * @writeback_index: Writeback starts here.
447 * @flags: Error bits and flags (AS_*).
448 * @wb_err: The most recent error which has occurred.
449 * @private_lock: For use by the owner of the address_space.
450 * @private_list: For use by the owner of the address_space.
451 * @private_data: For use by the owner of the address_space.
453 struct address_space
{
455 struct xarray i_pages
;
457 atomic_t i_mmap_writable
;
458 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
459 /* number of thp, only for non-shmem files */
462 struct rb_root_cached i_mmap
;
463 struct rw_semaphore i_mmap_rwsem
;
464 unsigned long nrpages
;
465 pgoff_t writeback_index
;
466 const struct address_space_operations
*a_ops
;
469 spinlock_t private_lock
;
470 struct list_head private_list
;
472 } __attribute__((aligned(sizeof(long)))) __randomize_layout
;
474 * On most architectures that alignment is already the case; but
475 * must be enforced here for CRIS, to let the least significant bit
476 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
479 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
480 #define PAGECACHE_TAG_DIRTY XA_MARK_0
481 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
482 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
485 * Returns true if any of the pages in the mapping are marked with the tag.
487 static inline bool mapping_tagged(struct address_space
*mapping
, xa_mark_t tag
)
489 return xa_marked(&mapping
->i_pages
, tag
);
492 static inline void i_mmap_lock_write(struct address_space
*mapping
)
494 down_write(&mapping
->i_mmap_rwsem
);
497 static inline int i_mmap_trylock_write(struct address_space
*mapping
)
499 return down_write_trylock(&mapping
->i_mmap_rwsem
);
502 static inline void i_mmap_unlock_write(struct address_space
*mapping
)
504 up_write(&mapping
->i_mmap_rwsem
);
507 static inline void i_mmap_lock_read(struct address_space
*mapping
)
509 down_read(&mapping
->i_mmap_rwsem
);
512 static inline void i_mmap_unlock_read(struct address_space
*mapping
)
514 up_read(&mapping
->i_mmap_rwsem
);
517 static inline void i_mmap_assert_locked(struct address_space
*mapping
)
519 lockdep_assert_held(&mapping
->i_mmap_rwsem
);
522 static inline void i_mmap_assert_write_locked(struct address_space
*mapping
)
524 lockdep_assert_held_write(&mapping
->i_mmap_rwsem
);
528 * Might pages of this file be mapped into userspace?
530 static inline int mapping_mapped(struct address_space
*mapping
)
532 return !RB_EMPTY_ROOT(&mapping
->i_mmap
.rb_root
);
536 * Might pages of this file have been modified in userspace?
537 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap
538 * marks vma as VM_SHARED if it is shared, and the file was opened for
539 * writing i.e. vma may be mprotected writable even if now readonly.
541 * If i_mmap_writable is negative, no new writable mappings are allowed. You
542 * can only deny writable mappings, if none exists right now.
544 static inline int mapping_writably_mapped(struct address_space
*mapping
)
546 return atomic_read(&mapping
->i_mmap_writable
) > 0;
549 static inline int mapping_map_writable(struct address_space
*mapping
)
551 return atomic_inc_unless_negative(&mapping
->i_mmap_writable
) ?
555 static inline void mapping_unmap_writable(struct address_space
*mapping
)
557 atomic_dec(&mapping
->i_mmap_writable
);
560 static inline int mapping_deny_writable(struct address_space
*mapping
)
562 return atomic_dec_unless_positive(&mapping
->i_mmap_writable
) ?
566 static inline void mapping_allow_writable(struct address_space
*mapping
)
568 atomic_inc(&mapping
->i_mmap_writable
);
572 * Use sequence counter to get consistent i_size on 32-bit processors.
574 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
575 #include <linux/seqlock.h>
576 #define __NEED_I_SIZE_ORDERED
577 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
579 #define i_size_ordered_init(inode) do { } while (0)
583 #define ACL_NOT_CACHED ((void *)(-1))
584 #define ACL_DONT_CACHE ((void *)(-3))
586 static inline struct posix_acl
*
587 uncached_acl_sentinel(struct task_struct
*task
)
589 return (void *)task
+ 1;
593 is_uncached_acl(struct posix_acl
*acl
)
595 return (long)acl
& 1;
598 #define IOP_FASTPERM 0x0001
599 #define IOP_LOOKUP 0x0002
600 #define IOP_NOFOLLOW 0x0004
601 #define IOP_XATTR 0x0008
602 #define IOP_DEFAULT_READLINK 0x0010
604 struct fsnotify_mark_connector
;
607 * Keep mostly read-only and often accessed (especially for
608 * the RCU path lookup and 'stat' data) fields at the beginning
609 * of the 'struct inode'
613 unsigned short i_opflags
;
616 unsigned int i_flags
;
618 #ifdef CONFIG_FS_POSIX_ACL
619 struct posix_acl
*i_acl
;
620 struct posix_acl
*i_default_acl
;
623 const struct inode_operations
*i_op
;
624 struct super_block
*i_sb
;
625 struct address_space
*i_mapping
;
627 #ifdef CONFIG_SECURITY
631 /* Stat data, not accessed from path walking */
634 * Filesystems may only read i_nlink directly. They shall use the
635 * following functions for modification:
637 * (set|clear|inc|drop)_nlink
638 * inode_(inc|dec)_link_count
641 const unsigned int i_nlink
;
642 unsigned int __i_nlink
;
646 struct timespec64 i_atime
;
647 struct timespec64 i_mtime
;
648 struct timespec64 i_ctime
;
649 spinlock_t i_lock
; /* i_blocks, i_bytes, maybe i_size */
650 unsigned short i_bytes
;
655 #ifdef __NEED_I_SIZE_ORDERED
656 seqcount_t i_size_seqcount
;
660 unsigned long i_state
;
661 struct rw_semaphore i_rwsem
;
663 unsigned long dirtied_when
; /* jiffies of first dirtying */
664 unsigned long dirtied_time_when
;
666 struct hlist_node i_hash
;
667 struct list_head i_io_list
; /* backing dev IO list */
668 #ifdef CONFIG_CGROUP_WRITEBACK
669 struct bdi_writeback
*i_wb
; /* the associated cgroup wb */
671 /* foreign inode detection, see wbc_detach_inode() */
673 u16 i_wb_frn_avg_time
;
674 u16 i_wb_frn_history
;
676 struct list_head i_lru
; /* inode LRU list */
677 struct list_head i_sb_list
;
678 struct list_head i_wb_list
; /* backing dev writeback list */
680 struct hlist_head i_dentry
;
681 struct rcu_head i_rcu
;
683 atomic64_t i_version
;
684 atomic64_t i_sequence
; /* see futex */
686 atomic_t i_dio_count
;
687 atomic_t i_writecount
;
688 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
689 atomic_t i_readcount
; /* struct files open RO */
692 const struct file_operations
*i_fop
; /* former ->i_op->default_file_ops */
693 void (*free_inode
)(struct inode
*);
695 struct file_lock_context
*i_flctx
;
696 struct address_space i_data
;
697 struct list_head i_devices
;
699 struct pipe_inode_info
*i_pipe
;
707 #ifdef CONFIG_FSNOTIFY
708 __u32 i_fsnotify_mask
; /* all events this inode cares about */
709 struct fsnotify_mark_connector __rcu
*i_fsnotify_marks
;
712 #ifdef CONFIG_FS_ENCRYPTION
713 struct fscrypt_info
*i_crypt_info
;
716 #ifdef CONFIG_FS_VERITY
717 struct fsverity_info
*i_verity_info
;
720 void *i_private
; /* fs or device private pointer */
721 } __randomize_layout
;
723 struct timespec64
timestamp_truncate(struct timespec64 t
, struct inode
*inode
);
725 static inline unsigned int i_blocksize(const struct inode
*node
)
727 return (1 << node
->i_blkbits
);
730 static inline int inode_unhashed(struct inode
*inode
)
732 return hlist_unhashed(&inode
->i_hash
);
736 * __mark_inode_dirty expects inodes to be hashed. Since we don't
737 * want special inodes in the fileset inode space, we make them
738 * appear hashed, but do not put on any lists. hlist_del()
739 * will work fine and require no locking.
741 static inline void inode_fake_hash(struct inode
*inode
)
743 hlist_add_fake(&inode
->i_hash
);
747 * inode->i_mutex nesting subclasses for the lock validator:
749 * 0: the object of the current VFS operation
753 * 4: second non-directory
754 * 5: second parent (when locking independent directories in rename)
756 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
757 * non-directories at once.
759 * The locking order between these classes is
760 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
762 enum inode_i_mutex_lock_class
772 static inline void inode_lock(struct inode
*inode
)
774 down_write(&inode
->i_rwsem
);
777 static inline void inode_unlock(struct inode
*inode
)
779 up_write(&inode
->i_rwsem
);
782 static inline void inode_lock_shared(struct inode
*inode
)
784 down_read(&inode
->i_rwsem
);
787 static inline void inode_unlock_shared(struct inode
*inode
)
789 up_read(&inode
->i_rwsem
);
792 static inline int inode_trylock(struct inode
*inode
)
794 return down_write_trylock(&inode
->i_rwsem
);
797 static inline int inode_trylock_shared(struct inode
*inode
)
799 return down_read_trylock(&inode
->i_rwsem
);
802 static inline int inode_is_locked(struct inode
*inode
)
804 return rwsem_is_locked(&inode
->i_rwsem
);
807 static inline void inode_lock_nested(struct inode
*inode
, unsigned subclass
)
809 down_write_nested(&inode
->i_rwsem
, subclass
);
812 static inline void inode_lock_shared_nested(struct inode
*inode
, unsigned subclass
)
814 down_read_nested(&inode
->i_rwsem
, subclass
);
817 void lock_two_nondirectories(struct inode
*, struct inode
*);
818 void unlock_two_nondirectories(struct inode
*, struct inode
*);
821 * NOTE: in a 32bit arch with a preemptable kernel and
822 * an UP compile the i_size_read/write must be atomic
823 * with respect to the local cpu (unlike with preempt disabled),
824 * but they don't need to be atomic with respect to other cpus like in
825 * true SMP (so they need either to either locally disable irq around
826 * the read or for example on x86 they can be still implemented as a
827 * cmpxchg8b without the need of the lock prefix). For SMP compiles
828 * and 64bit archs it makes no difference if preempt is enabled or not.
830 static inline loff_t
i_size_read(const struct inode
*inode
)
832 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
837 seq
= read_seqcount_begin(&inode
->i_size_seqcount
);
838 i_size
= inode
->i_size
;
839 } while (read_seqcount_retry(&inode
->i_size_seqcount
, seq
));
841 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
845 i_size
= inode
->i_size
;
849 return inode
->i_size
;
854 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
855 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
856 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
858 static inline void i_size_write(struct inode
*inode
, loff_t i_size
)
860 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
862 write_seqcount_begin(&inode
->i_size_seqcount
);
863 inode
->i_size
= i_size
;
864 write_seqcount_end(&inode
->i_size_seqcount
);
866 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
868 inode
->i_size
= i_size
;
871 inode
->i_size
= i_size
;
875 static inline unsigned iminor(const struct inode
*inode
)
877 return MINOR(inode
->i_rdev
);
880 static inline unsigned imajor(const struct inode
*inode
)
882 return MAJOR(inode
->i_rdev
);
886 rwlock_t lock
; /* protects pid, uid, euid fields */
887 struct pid
*pid
; /* pid or -pgrp where SIGIO should be sent */
888 enum pid_type pid_type
; /* Kind of process group SIGIO should be sent to */
889 kuid_t uid
, euid
; /* uid/euid of process setting the owner */
890 int signum
; /* posix.1b rt signal to be delivered on IO */
894 * struct file_ra_state - Track a file's readahead state.
895 * @start: Where the most recent readahead started.
896 * @size: Number of pages read in the most recent readahead.
897 * @async_size: Start next readahead when this many pages are left.
898 * @ra_pages: Maximum size of a readahead request.
899 * @mmap_miss: How many mmap accesses missed in the page cache.
900 * @prev_pos: The last byte in the most recent read request.
902 struct file_ra_state
{
905 unsigned int async_size
;
906 unsigned int ra_pages
;
907 unsigned int mmap_miss
;
912 * Check if @index falls in the readahead windows.
914 static inline int ra_has_index(struct file_ra_state
*ra
, pgoff_t index
)
916 return (index
>= ra
->start
&&
917 index
< ra
->start
+ ra
->size
);
922 struct llist_node fu_llist
;
923 struct rcu_head fu_rcuhead
;
926 struct inode
*f_inode
; /* cached value */
927 const struct file_operations
*f_op
;
930 * Protects f_ep, f_flags.
931 * Must not be taken from IRQ context.
934 enum rw_hint f_write_hint
;
935 atomic_long_t f_count
;
936 unsigned int f_flags
;
938 struct mutex f_pos_lock
;
940 struct fown_struct f_owner
;
941 const struct cred
*f_cred
;
942 struct file_ra_state f_ra
;
945 #ifdef CONFIG_SECURITY
948 /* needed for tty driver, and maybe others */
952 /* Used by fs/eventpoll.c to link all the hooks to this file */
953 struct hlist_head
*f_ep
;
954 #endif /* #ifdef CONFIG_EPOLL */
955 struct address_space
*f_mapping
;
957 errseq_t f_sb_err
; /* for syncfs */
959 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
964 /* file identifier */
965 unsigned char f_handle
[];
968 static inline struct file
*get_file(struct file
*f
)
970 atomic_long_inc(&f
->f_count
);
973 #define get_file_rcu_many(x, cnt) \
974 atomic_long_add_unless(&(x)->f_count, (cnt), 0)
975 #define get_file_rcu(x) get_file_rcu_many((x), 1)
976 #define file_count(x) atomic_long_read(&(x)->f_count)
978 #define MAX_NON_LFS ((1UL<<31) - 1)
980 /* Page cache limit. The filesystems should put that into their s_maxbytes
981 limits, otherwise bad things can happen in VM. */
982 #if BITS_PER_LONG==32
983 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
984 #elif BITS_PER_LONG==64
985 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
990 #define FL_DELEG 4 /* NFSv4 delegation */
991 #define FL_ACCESS 8 /* not trying to lock, just looking */
992 #define FL_EXISTS 16 /* when unlocking, test for existence */
993 #define FL_LEASE 32 /* lease held on this file */
994 #define FL_CLOSE 64 /* unlock on close */
995 #define FL_SLEEP 128 /* A blocking lock */
996 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
997 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
998 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
999 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
1001 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
1004 * Special return value from posix_lock_file() and vfs_lock_file() for
1005 * asynchronous locking.
1007 #define FILE_LOCK_DEFERRED 1
1009 /* legacy typedef, should eventually be removed */
1010 typedef void *fl_owner_t
;
1014 struct file_lock_operations
{
1015 void (*fl_copy_lock
)(struct file_lock
*, struct file_lock
*);
1016 void (*fl_release_private
)(struct file_lock
*);
1019 struct lock_manager_operations
{
1020 fl_owner_t (*lm_get_owner
)(fl_owner_t
);
1021 void (*lm_put_owner
)(fl_owner_t
);
1022 void (*lm_notify
)(struct file_lock
*); /* unblock callback */
1023 int (*lm_grant
)(struct file_lock
*, int);
1024 bool (*lm_break
)(struct file_lock
*);
1025 int (*lm_change
)(struct file_lock
*, int, struct list_head
*);
1026 void (*lm_setup
)(struct file_lock
*, void **);
1027 bool (*lm_breaker_owns_lease
)(struct file_lock
*);
1030 struct lock_manager
{
1031 struct list_head list
;
1033 * NFSv4 and up also want opens blocked during the grace period;
1040 void locks_start_grace(struct net
*, struct lock_manager
*);
1041 void locks_end_grace(struct lock_manager
*);
1042 bool locks_in_grace(struct net
*);
1043 bool opens_in_grace(struct net
*);
1045 /* that will die - we need it for nfs_lock_info */
1046 #include <linux/nfs_fs_i.h>
1049 * struct file_lock represents a generic "file lock". It's used to represent
1050 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1051 * note that the same struct is used to represent both a request for a lock and
1052 * the lock itself, but the same object is never used for both.
1054 * FIXME: should we create a separate "struct lock_request" to help distinguish
1057 * The varous i_flctx lists are ordered by:
1060 * 2) lock range start
1063 * Obviously, the last two criteria only matter for POSIX locks.
1066 struct file_lock
*fl_blocker
; /* The lock, that is blocking us */
1067 struct list_head fl_list
; /* link into file_lock_context */
1068 struct hlist_node fl_link
; /* node in global lists */
1069 struct list_head fl_blocked_requests
; /* list of requests with
1070 * ->fl_blocker pointing here
1072 struct list_head fl_blocked_member
; /* node in
1073 * ->fl_blocker->fl_blocked_requests
1075 fl_owner_t fl_owner
;
1076 unsigned int fl_flags
;
1077 unsigned char fl_type
;
1078 unsigned int fl_pid
;
1079 int fl_link_cpu
; /* what cpu's list is this on? */
1080 wait_queue_head_t fl_wait
;
1081 struct file
*fl_file
;
1085 struct fasync_struct
* fl_fasync
; /* for lease break notifications */
1086 /* for lease breaks: */
1087 unsigned long fl_break_time
;
1088 unsigned long fl_downgrade_time
;
1090 const struct file_lock_operations
*fl_ops
; /* Callbacks for filesystems */
1091 const struct lock_manager_operations
*fl_lmops
; /* Callbacks for lockmanagers */
1093 struct nfs_lock_info nfs_fl
;
1094 struct nfs4_lock_info nfs4_fl
;
1096 struct list_head link
; /* link in AFS vnode's pending_locks list */
1097 int state
; /* state of grant or error if -ve */
1098 unsigned int debug_id
;
1101 } __randomize_layout
;
1103 struct file_lock_context
{
1104 spinlock_t flc_lock
;
1105 struct list_head flc_flock
;
1106 struct list_head flc_posix
;
1107 struct list_head flc_lease
;
1110 /* The following constant reflects the upper bound of the file/locking space */
1112 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1113 #define OFFSET_MAX INT_LIMIT(loff_t)
1114 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1117 extern void send_sigio(struct fown_struct
*fown
, int fd
, int band
);
1119 #define locks_inode(f) file_inode(f)
1121 #ifdef CONFIG_FILE_LOCKING
1122 extern int fcntl_getlk(struct file
*, unsigned int, struct flock
*);
1123 extern int fcntl_setlk(unsigned int, struct file
*, unsigned int,
1126 #if BITS_PER_LONG == 32
1127 extern int fcntl_getlk64(struct file
*, unsigned int, struct flock64
*);
1128 extern int fcntl_setlk64(unsigned int, struct file
*, unsigned int,
1132 extern int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
);
1133 extern int fcntl_getlease(struct file
*filp
);
1136 void locks_free_lock_context(struct inode
*inode
);
1137 void locks_free_lock(struct file_lock
*fl
);
1138 extern void locks_init_lock(struct file_lock
*);
1139 extern struct file_lock
* locks_alloc_lock(void);
1140 extern void locks_copy_lock(struct file_lock
*, struct file_lock
*);
1141 extern void locks_copy_conflock(struct file_lock
*, struct file_lock
*);
1142 extern void locks_remove_posix(struct file
*, fl_owner_t
);
1143 extern void locks_remove_file(struct file
*);
1144 extern void locks_release_private(struct file_lock
*);
1145 extern void posix_test_lock(struct file
*, struct file_lock
*);
1146 extern int posix_lock_file(struct file
*, struct file_lock
*, struct file_lock
*);
1147 extern int locks_delete_block(struct file_lock
*);
1148 extern int vfs_test_lock(struct file
*, struct file_lock
*);
1149 extern int vfs_lock_file(struct file
*, unsigned int, struct file_lock
*, struct file_lock
*);
1150 extern int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
);
1151 extern int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
);
1152 extern int __break_lease(struct inode
*inode
, unsigned int flags
, unsigned int type
);
1153 extern void lease_get_mtime(struct inode
*, struct timespec64
*time
);
1154 extern int generic_setlease(struct file
*, long, struct file_lock
**, void **priv
);
1155 extern int vfs_setlease(struct file
*, long, struct file_lock
**, void **);
1156 extern int lease_modify(struct file_lock
*, int, struct list_head
*);
1158 struct notifier_block
;
1159 extern int lease_register_notifier(struct notifier_block
*);
1160 extern void lease_unregister_notifier(struct notifier_block
*);
1162 struct files_struct
;
1163 extern void show_fd_locks(struct seq_file
*f
,
1164 struct file
*filp
, struct files_struct
*files
);
1165 #else /* !CONFIG_FILE_LOCKING */
1166 static inline int fcntl_getlk(struct file
*file
, unsigned int cmd
,
1167 struct flock __user
*user
)
1172 static inline int fcntl_setlk(unsigned int fd
, struct file
*file
,
1173 unsigned int cmd
, struct flock __user
*user
)
1178 #if BITS_PER_LONG == 32
1179 static inline int fcntl_getlk64(struct file
*file
, unsigned int cmd
,
1180 struct flock64 __user
*user
)
1185 static inline int fcntl_setlk64(unsigned int fd
, struct file
*file
,
1186 unsigned int cmd
, struct flock64 __user
*user
)
1191 static inline int fcntl_setlease(unsigned int fd
, struct file
*filp
, long arg
)
1196 static inline int fcntl_getlease(struct file
*filp
)
1202 locks_free_lock_context(struct inode
*inode
)
1206 static inline void locks_init_lock(struct file_lock
*fl
)
1211 static inline void locks_copy_conflock(struct file_lock
*new, struct file_lock
*fl
)
1216 static inline void locks_copy_lock(struct file_lock
*new, struct file_lock
*fl
)
1221 static inline void locks_remove_posix(struct file
*filp
, fl_owner_t owner
)
1226 static inline void locks_remove_file(struct file
*filp
)
1231 static inline void posix_test_lock(struct file
*filp
, struct file_lock
*fl
)
1236 static inline int posix_lock_file(struct file
*filp
, struct file_lock
*fl
,
1237 struct file_lock
*conflock
)
1242 static inline int locks_delete_block(struct file_lock
*waiter
)
1247 static inline int vfs_test_lock(struct file
*filp
, struct file_lock
*fl
)
1252 static inline int vfs_lock_file(struct file
*filp
, unsigned int cmd
,
1253 struct file_lock
*fl
, struct file_lock
*conf
)
1258 static inline int vfs_cancel_lock(struct file
*filp
, struct file_lock
*fl
)
1263 static inline int locks_lock_inode_wait(struct inode
*inode
, struct file_lock
*fl
)
1268 static inline int __break_lease(struct inode
*inode
, unsigned int mode
, unsigned int type
)
1273 static inline void lease_get_mtime(struct inode
*inode
,
1274 struct timespec64
*time
)
1279 static inline int generic_setlease(struct file
*filp
, long arg
,
1280 struct file_lock
**flp
, void **priv
)
1285 static inline int vfs_setlease(struct file
*filp
, long arg
,
1286 struct file_lock
**lease
, void **priv
)
1291 static inline int lease_modify(struct file_lock
*fl
, int arg
,
1292 struct list_head
*dispose
)
1297 struct files_struct
;
1298 static inline void show_fd_locks(struct seq_file
*f
,
1299 struct file
*filp
, struct files_struct
*files
) {}
1300 #endif /* !CONFIG_FILE_LOCKING */
1302 static inline struct inode
*file_inode(const struct file
*f
)
1307 static inline struct dentry
*file_dentry(const struct file
*file
)
1309 return d_real(file
->f_path
.dentry
, file_inode(file
));
1312 static inline int locks_lock_file_wait(struct file
*filp
, struct file_lock
*fl
)
1314 return locks_lock_inode_wait(locks_inode(filp
), fl
);
1317 struct fasync_struct
{
1321 struct fasync_struct
*fa_next
; /* singly linked list */
1322 struct file
*fa_file
;
1323 struct rcu_head fa_rcu
;
1326 #define FASYNC_MAGIC 0x4601
1328 /* SMP safe fasync helpers: */
1329 extern int fasync_helper(int, struct file
*, int, struct fasync_struct
**);
1330 extern struct fasync_struct
*fasync_insert_entry(int, struct file
*, struct fasync_struct
**, struct fasync_struct
*);
1331 extern int fasync_remove_entry(struct file
*, struct fasync_struct
**);
1332 extern struct fasync_struct
*fasync_alloc(void);
1333 extern void fasync_free(struct fasync_struct
*);
1335 /* can be called from interrupts */
1336 extern void kill_fasync(struct fasync_struct
**, int, int);
1338 extern void __f_setown(struct file
*filp
, struct pid
*, enum pid_type
, int force
);
1339 extern int f_setown(struct file
*filp
, unsigned long arg
, int force
);
1340 extern void f_delown(struct file
*filp
);
1341 extern pid_t
f_getown(struct file
*filp
);
1342 extern int send_sigurg(struct fown_struct
*fown
);
1345 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1346 * represented in both.
1348 #define SB_RDONLY 1 /* Mount read-only */
1349 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1350 #define SB_NODEV 4 /* Disallow access to device special files */
1351 #define SB_NOEXEC 8 /* Disallow program execution */
1352 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1353 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1354 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1355 #define SB_NOATIME 1024 /* Do not update access times. */
1356 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1357 #define SB_SILENT 32768
1358 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1359 #define SB_INLINECRYPT (1<<17) /* Use blk-crypto for encrypted files */
1360 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1361 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1362 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1364 /* These sb flags are internal to the kernel */
1365 #define SB_SUBMOUNT (1<<26)
1366 #define SB_FORCE (1<<27)
1367 #define SB_NOSEC (1<<28)
1368 #define SB_BORN (1<<29)
1369 #define SB_ACTIVE (1<<30)
1370 #define SB_NOUSER (1<<31)
1372 /* These flags relate to encoding and casefolding */
1373 #define SB_ENC_STRICT_MODE_FL (1 << 0)
1375 #define sb_has_strict_encoding(sb) \
1376 (sb->s_encoding_flags & SB_ENC_STRICT_MODE_FL)
1382 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1383 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1384 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1385 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1386 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1389 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1390 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1391 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1392 #define SB_I_STABLE_WRITES 0x00000008 /* don't modify blks until WB is done */
1394 /* sb->s_iflags to limit user namespace mounts */
1395 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1396 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1397 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1399 #define SB_I_SKIP_SYNC 0x00000100 /* Skip superblock at global sync */
1401 /* Possible states of 'frozen' field */
1403 SB_UNFROZEN
= 0, /* FS is unfrozen */
1404 SB_FREEZE_WRITE
= 1, /* Writes, dir ops, ioctls frozen */
1405 SB_FREEZE_PAGEFAULT
= 2, /* Page faults stopped as well */
1406 SB_FREEZE_FS
= 3, /* For internal FS use (e.g. to stop
1407 * internal threads if needed) */
1408 SB_FREEZE_COMPLETE
= 4, /* ->freeze_fs finished successfully */
1411 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1414 int frozen
; /* Is sb frozen? */
1415 wait_queue_head_t wait_unfrozen
; /* wait for thaw */
1416 struct percpu_rw_semaphore rw_sem
[SB_FREEZE_LEVELS
];
1419 struct super_block
{
1420 struct list_head s_list
; /* Keep this first */
1421 dev_t s_dev
; /* search index; _not_ kdev_t */
1422 unsigned char s_blocksize_bits
;
1423 unsigned long s_blocksize
;
1424 loff_t s_maxbytes
; /* Max file size */
1425 struct file_system_type
*s_type
;
1426 const struct super_operations
*s_op
;
1427 const struct dquot_operations
*dq_op
;
1428 const struct quotactl_ops
*s_qcop
;
1429 const struct export_operations
*s_export_op
;
1430 unsigned long s_flags
;
1431 unsigned long s_iflags
; /* internal SB_I_* flags */
1432 unsigned long s_magic
;
1433 struct dentry
*s_root
;
1434 struct rw_semaphore s_umount
;
1437 #ifdef CONFIG_SECURITY
1440 const struct xattr_handler
**s_xattr
;
1441 #ifdef CONFIG_FS_ENCRYPTION
1442 const struct fscrypt_operations
*s_cop
;
1443 struct key
*s_master_keys
; /* master crypto keys in use */
1445 #ifdef CONFIG_FS_VERITY
1446 const struct fsverity_operations
*s_vop
;
1448 #ifdef CONFIG_UNICODE
1449 struct unicode_map
*s_encoding
;
1450 __u16 s_encoding_flags
;
1452 struct hlist_bl_head s_roots
; /* alternate root dentries for NFS */
1453 struct list_head s_mounts
; /* list of mounts; _not_ for fs use */
1454 struct block_device
*s_bdev
;
1455 struct backing_dev_info
*s_bdi
;
1456 struct mtd_info
*s_mtd
;
1457 struct hlist_node s_instances
;
1458 unsigned int s_quota_types
; /* Bitmask of supported quota types */
1459 struct quota_info s_dquot
; /* Diskquota specific options */
1461 struct sb_writers s_writers
;
1464 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1465 * s_fsnotify_marks together for cache efficiency. They are frequently
1466 * accessed and rarely modified.
1468 void *s_fs_info
; /* Filesystem private info */
1470 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1472 /* Time limits for c/m/atime in seconds */
1473 time64_t s_time_min
;
1474 time64_t s_time_max
;
1475 #ifdef CONFIG_FSNOTIFY
1476 __u32 s_fsnotify_mask
;
1477 struct fsnotify_mark_connector __rcu
*s_fsnotify_marks
;
1480 char s_id
[32]; /* Informational name */
1481 uuid_t s_uuid
; /* UUID */
1483 unsigned int s_max_links
;
1487 * The next field is for VFS *only*. No filesystems have any business
1488 * even looking at it. You had been warned.
1490 struct mutex s_vfs_rename_mutex
; /* Kludge */
1493 * Filesystem subtype. If non-empty the filesystem type field
1494 * in /proc/mounts will be "type.subtype"
1496 const char *s_subtype
;
1498 const struct dentry_operations
*s_d_op
; /* default d_op for dentries */
1501 * Saved pool identifier for cleancache (-1 means none)
1503 int cleancache_poolid
;
1505 struct shrinker s_shrink
; /* per-sb shrinker handle */
1507 /* Number of inodes with nlink == 0 but still referenced */
1508 atomic_long_t s_remove_count
;
1510 /* Pending fsnotify inode refs */
1511 atomic_long_t s_fsnotify_inode_refs
;
1513 /* Being remounted read-only */
1514 int s_readonly_remount
;
1516 /* per-sb errseq_t for reporting writeback errors via syncfs */
1519 /* AIO completions deferred from interrupt context */
1520 struct workqueue_struct
*s_dio_done_wq
;
1521 struct hlist_head s_pins
;
1524 * Owning user namespace and default context in which to
1525 * interpret filesystem uids, gids, quotas, device nodes,
1526 * xattrs and security labels.
1528 struct user_namespace
*s_user_ns
;
1531 * The list_lru structure is essentially just a pointer to a table
1532 * of per-node lru lists, each of which has its own spinlock.
1533 * There is no need to put them into separate cachelines.
1535 struct list_lru s_dentry_lru
;
1536 struct list_lru s_inode_lru
;
1537 struct rcu_head rcu
;
1538 struct work_struct destroy_work
;
1540 struct mutex s_sync_lock
; /* sync serialisation lock */
1543 * Indicates how deep in a filesystem stack this SB is
1547 /* s_inode_list_lock protects s_inodes */
1548 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp
;
1549 struct list_head s_inodes
; /* all inodes */
1551 spinlock_t s_inode_wblist_lock
;
1552 struct list_head s_inodes_wb
; /* writeback inodes */
1553 } __randomize_layout
;
1555 /* Helper functions so that in most cases filesystems will
1556 * not need to deal directly with kuid_t and kgid_t and can
1557 * instead deal with the raw numeric values that are stored
1558 * in the filesystem.
1560 static inline uid_t
i_uid_read(const struct inode
*inode
)
1562 return from_kuid(inode
->i_sb
->s_user_ns
, inode
->i_uid
);
1565 static inline gid_t
i_gid_read(const struct inode
*inode
)
1567 return from_kgid(inode
->i_sb
->s_user_ns
, inode
->i_gid
);
1570 static inline void i_uid_write(struct inode
*inode
, uid_t uid
)
1572 inode
->i_uid
= make_kuid(inode
->i_sb
->s_user_ns
, uid
);
1575 static inline void i_gid_write(struct inode
*inode
, gid_t gid
)
1577 inode
->i_gid
= make_kgid(inode
->i_sb
->s_user_ns
, gid
);
1581 * kuid_into_mnt - map a kuid down into a mnt_userns
1582 * @mnt_userns: user namespace of the relevant mount
1583 * @kuid: kuid to be mapped
1585 * Return: @kuid mapped according to @mnt_userns.
1586 * If @kuid has no mapping INVALID_UID is returned.
1588 static inline kuid_t
kuid_into_mnt(struct user_namespace
*mnt_userns
,
1591 return make_kuid(mnt_userns
, __kuid_val(kuid
));
1595 * kgid_into_mnt - map a kgid down into a mnt_userns
1596 * @mnt_userns: user namespace of the relevant mount
1597 * @kgid: kgid to be mapped
1599 * Return: @kgid mapped according to @mnt_userns.
1600 * If @kgid has no mapping INVALID_GID is returned.
1602 static inline kgid_t
kgid_into_mnt(struct user_namespace
*mnt_userns
,
1605 return make_kgid(mnt_userns
, __kgid_val(kgid
));
1609 * i_uid_into_mnt - map an inode's i_uid down into a mnt_userns
1610 * @mnt_userns: user namespace of the mount the inode was found from
1611 * @inode: inode to map
1613 * Return: the inode's i_uid mapped down according to @mnt_userns.
1614 * If the inode's i_uid has no mapping INVALID_UID is returned.
1616 static inline kuid_t
i_uid_into_mnt(struct user_namespace
*mnt_userns
,
1617 const struct inode
*inode
)
1619 return kuid_into_mnt(mnt_userns
, inode
->i_uid
);
1623 * i_gid_into_mnt - map an inode's i_gid down into a mnt_userns
1624 * @mnt_userns: user namespace of the mount the inode was found from
1625 * @inode: inode to map
1627 * Return: the inode's i_gid mapped down according to @mnt_userns.
1628 * If the inode's i_gid has no mapping INVALID_GID is returned.
1630 static inline kgid_t
i_gid_into_mnt(struct user_namespace
*mnt_userns
,
1631 const struct inode
*inode
)
1633 return kgid_into_mnt(mnt_userns
, inode
->i_gid
);
1637 * kuid_from_mnt - map a kuid up into a mnt_userns
1638 * @mnt_userns: user namespace of the relevant mount
1639 * @kuid: kuid to be mapped
1641 * Return: @kuid mapped up according to @mnt_userns.
1642 * If @kuid has no mapping INVALID_UID is returned.
1644 static inline kuid_t
kuid_from_mnt(struct user_namespace
*mnt_userns
,
1647 return KUIDT_INIT(from_kuid(mnt_userns
, kuid
));
1651 * kgid_from_mnt - map a kgid up into a mnt_userns
1652 * @mnt_userns: user namespace of the relevant mount
1653 * @kgid: kgid to be mapped
1655 * Return: @kgid mapped up according to @mnt_userns.
1656 * If @kgid has no mapping INVALID_GID is returned.
1658 static inline kgid_t
kgid_from_mnt(struct user_namespace
*mnt_userns
,
1661 return KGIDT_INIT(from_kgid(mnt_userns
, kgid
));
1665 * mapped_fsuid - return caller's fsuid mapped up into a mnt_userns
1666 * @mnt_userns: user namespace of the relevant mount
1668 * Use this helper to initialize a new vfs or filesystem object based on
1669 * the caller's fsuid. A common example is initializing the i_uid field of
1670 * a newly allocated inode triggered by a creation event such as mkdir or
1671 * O_CREAT. Other examples include the allocation of quotas for a specific
1674 * Return: the caller's current fsuid mapped up according to @mnt_userns.
1676 static inline kuid_t
mapped_fsuid(struct user_namespace
*mnt_userns
)
1678 return kuid_from_mnt(mnt_userns
, current_fsuid());
1682 * mapped_fsgid - return caller's fsgid mapped up into a mnt_userns
1683 * @mnt_userns: user namespace of the relevant mount
1685 * Use this helper to initialize a new vfs or filesystem object based on
1686 * the caller's fsgid. A common example is initializing the i_gid field of
1687 * a newly allocated inode triggered by a creation event such as mkdir or
1688 * O_CREAT. Other examples include the allocation of quotas for a specific
1691 * Return: the caller's current fsgid mapped up according to @mnt_userns.
1693 static inline kgid_t
mapped_fsgid(struct user_namespace
*mnt_userns
)
1695 return kgid_from_mnt(mnt_userns
, current_fsgid());
1699 * inode_fsuid_set - initialize inode's i_uid field with callers fsuid
1700 * @inode: inode to initialize
1701 * @mnt_userns: user namespace of the mount the inode was found from
1703 * Initialize the i_uid field of @inode. If the inode was found/created via
1704 * an idmapped mount map the caller's fsuid according to @mnt_users.
1706 static inline void inode_fsuid_set(struct inode
*inode
,
1707 struct user_namespace
*mnt_userns
)
1709 inode
->i_uid
= mapped_fsuid(mnt_userns
);
1713 * inode_fsgid_set - initialize inode's i_gid field with callers fsgid
1714 * @inode: inode to initialize
1715 * @mnt_userns: user namespace of the mount the inode was found from
1717 * Initialize the i_gid field of @inode. If the inode was found/created via
1718 * an idmapped mount map the caller's fsgid according to @mnt_users.
1720 static inline void inode_fsgid_set(struct inode
*inode
,
1721 struct user_namespace
*mnt_userns
)
1723 inode
->i_gid
= mapped_fsgid(mnt_userns
);
1727 * fsuidgid_has_mapping() - check whether caller's fsuid/fsgid is mapped
1728 * @sb: the superblock we want a mapping in
1729 * @mnt_userns: user namespace of the relevant mount
1731 * Check whether the caller's fsuid and fsgid have a valid mapping in the
1732 * s_user_ns of the superblock @sb. If the caller is on an idmapped mount map
1733 * the caller's fsuid and fsgid according to the @mnt_userns first.
1735 * Return: true if fsuid and fsgid is mapped, false if not.
1737 static inline bool fsuidgid_has_mapping(struct super_block
*sb
,
1738 struct user_namespace
*mnt_userns
)
1740 struct user_namespace
*s_user_ns
= sb
->s_user_ns
;
1742 return kuid_has_mapping(s_user_ns
, mapped_fsuid(mnt_userns
)) &&
1743 kgid_has_mapping(s_user_ns
, mapped_fsgid(mnt_userns
));
1746 extern struct timespec64
current_time(struct inode
*inode
);
1749 * Snapshotting support.
1753 * These are internal functions, please use sb_start_{write,pagefault,intwrite}
1756 static inline void __sb_end_write(struct super_block
*sb
, int level
)
1758 percpu_up_read(sb
->s_writers
.rw_sem
+ level
-1);
1761 static inline void __sb_start_write(struct super_block
*sb
, int level
)
1763 percpu_down_read(sb
->s_writers
.rw_sem
+ level
- 1);
1766 static inline bool __sb_start_write_trylock(struct super_block
*sb
, int level
)
1768 return percpu_down_read_trylock(sb
->s_writers
.rw_sem
+ level
- 1);
1771 #define __sb_writers_acquired(sb, lev) \
1772 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1773 #define __sb_writers_release(sb, lev) \
1774 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1777 * sb_end_write - drop write access to a superblock
1778 * @sb: the super we wrote to
1780 * Decrement number of writers to the filesystem. Wake up possible waiters
1781 * wanting to freeze the filesystem.
1783 static inline void sb_end_write(struct super_block
*sb
)
1785 __sb_end_write(sb
, SB_FREEZE_WRITE
);
1789 * sb_end_pagefault - drop write access to a superblock from a page fault
1790 * @sb: the super we wrote to
1792 * Decrement number of processes handling write page fault to the filesystem.
1793 * Wake up possible waiters wanting to freeze the filesystem.
1795 static inline void sb_end_pagefault(struct super_block
*sb
)
1797 __sb_end_write(sb
, SB_FREEZE_PAGEFAULT
);
1801 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1802 * @sb: the super we wrote to
1804 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1805 * waiters wanting to freeze the filesystem.
1807 static inline void sb_end_intwrite(struct super_block
*sb
)
1809 __sb_end_write(sb
, SB_FREEZE_FS
);
1813 * sb_start_write - get write access to a superblock
1814 * @sb: the super we write to
1816 * When a process wants to write data or metadata to a file system (i.e. dirty
1817 * a page or an inode), it should embed the operation in a sb_start_write() -
1818 * sb_end_write() pair to get exclusion against file system freezing. This
1819 * function increments number of writers preventing freezing. If the file
1820 * system is already frozen, the function waits until the file system is
1823 * Since freeze protection behaves as a lock, users have to preserve
1824 * ordering of freeze protection and other filesystem locks. Generally,
1825 * freeze protection should be the outermost lock. In particular, we have:
1828 * -> i_mutex (write path, truncate, directory ops, ...)
1829 * -> s_umount (freeze_super, thaw_super)
1831 static inline void sb_start_write(struct super_block
*sb
)
1833 __sb_start_write(sb
, SB_FREEZE_WRITE
);
1836 static inline bool sb_start_write_trylock(struct super_block
*sb
)
1838 return __sb_start_write_trylock(sb
, SB_FREEZE_WRITE
);
1842 * sb_start_pagefault - get write access to a superblock from a page fault
1843 * @sb: the super we write to
1845 * When a process starts handling write page fault, it should embed the
1846 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1847 * exclusion against file system freezing. This is needed since the page fault
1848 * is going to dirty a page. This function increments number of running page
1849 * faults preventing freezing. If the file system is already frozen, the
1850 * function waits until the file system is thawed.
1852 * Since page fault freeze protection behaves as a lock, users have to preserve
1853 * ordering of freeze protection and other filesystem locks. It is advised to
1854 * put sb_start_pagefault() close to mmap_lock in lock ordering. Page fault
1855 * handling code implies lock dependency:
1858 * -> sb_start_pagefault
1860 static inline void sb_start_pagefault(struct super_block
*sb
)
1862 __sb_start_write(sb
, SB_FREEZE_PAGEFAULT
);
1866 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1867 * @sb: the super we write to
1869 * This is the third level of protection against filesystem freezing. It is
1870 * free for use by a filesystem. The only requirement is that it must rank
1871 * below sb_start_pagefault.
1873 * For example filesystem can call sb_start_intwrite() when starting a
1874 * transaction which somewhat eases handling of freezing for internal sources
1875 * of filesystem changes (internal fs threads, discarding preallocation on file
1878 static inline void sb_start_intwrite(struct super_block
*sb
)
1880 __sb_start_write(sb
, SB_FREEZE_FS
);
1883 static inline bool sb_start_intwrite_trylock(struct super_block
*sb
)
1885 return __sb_start_write_trylock(sb
, SB_FREEZE_FS
);
1888 bool inode_owner_or_capable(struct user_namespace
*mnt_userns
,
1889 const struct inode
*inode
);
1892 * VFS helper functions..
1894 int vfs_create(struct user_namespace
*, struct inode
*,
1895 struct dentry
*, umode_t
, bool);
1896 int vfs_mkdir(struct user_namespace
*, struct inode
*,
1897 struct dentry
*, umode_t
);
1898 int vfs_mknod(struct user_namespace
*, struct inode
*, struct dentry
*,
1900 int vfs_symlink(struct user_namespace
*, struct inode
*,
1901 struct dentry
*, const char *);
1902 int vfs_link(struct dentry
*, struct user_namespace
*, struct inode
*,
1903 struct dentry
*, struct inode
**);
1904 int vfs_rmdir(struct user_namespace
*, struct inode
*, struct dentry
*);
1905 int vfs_unlink(struct user_namespace
*, struct inode
*, struct dentry
*,
1909 * struct renamedata - contains all information required for renaming
1910 * @old_mnt_userns: old user namespace of the mount the inode was found from
1911 * @old_dir: parent of source
1912 * @old_dentry: source
1913 * @new_mnt_userns: new user namespace of the mount the inode was found from
1914 * @new_dir: parent of destination
1915 * @new_dentry: destination
1916 * @delegated_inode: returns an inode needing a delegation break
1917 * @flags: rename flags
1920 struct user_namespace
*old_mnt_userns
;
1921 struct inode
*old_dir
;
1922 struct dentry
*old_dentry
;
1923 struct user_namespace
*new_mnt_userns
;
1924 struct inode
*new_dir
;
1925 struct dentry
*new_dentry
;
1926 struct inode
**delegated_inode
;
1928 } __randomize_layout
;
1930 int vfs_rename(struct renamedata
*);
1932 static inline int vfs_whiteout(struct user_namespace
*mnt_userns
,
1933 struct inode
*dir
, struct dentry
*dentry
)
1935 return vfs_mknod(mnt_userns
, dir
, dentry
, S_IFCHR
| WHITEOUT_MODE
,
1939 struct dentry
*vfs_tmpfile(struct user_namespace
*mnt_userns
,
1940 struct dentry
*dentry
, umode_t mode
, int open_flag
);
1942 int vfs_mkobj(struct dentry
*, umode_t
,
1943 int (*f
)(struct dentry
*, umode_t
, void *),
1946 int vfs_fchown(struct file
*file
, uid_t user
, gid_t group
);
1947 int vfs_fchmod(struct file
*file
, umode_t mode
);
1948 int vfs_utimes(const struct path
*path
, struct timespec64
*times
);
1950 extern long vfs_ioctl(struct file
*file
, unsigned int cmd
, unsigned long arg
);
1952 #ifdef CONFIG_COMPAT
1953 extern long compat_ptr_ioctl(struct file
*file
, unsigned int cmd
,
1956 #define compat_ptr_ioctl NULL
1960 * VFS file helper functions.
1962 void inode_init_owner(struct user_namespace
*mnt_userns
, struct inode
*inode
,
1963 const struct inode
*dir
, umode_t mode
);
1964 extern bool may_open_dev(const struct path
*path
);
1967 * This is the "filldir" function type, used by readdir() to let
1968 * the kernel specify what kind of dirent layout it wants to have.
1969 * This allows the kernel to read directories into kernel space or
1970 * to have different dirent layouts depending on the binary type.
1973 typedef int (*filldir_t
)(struct dir_context
*, const char *, int, loff_t
, u64
,
1976 struct dir_context
{
1982 * These flags let !MMU mmap() govern direct device mapping vs immediate
1983 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1985 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1986 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1987 * NOMMU_MAP_READ: Can be mapped for reading
1988 * NOMMU_MAP_WRITE: Can be mapped for writing
1989 * NOMMU_MAP_EXEC: Can be mapped for execution
1991 #define NOMMU_MAP_COPY 0x00000001
1992 #define NOMMU_MAP_DIRECT 0x00000008
1993 #define NOMMU_MAP_READ VM_MAYREAD
1994 #define NOMMU_MAP_WRITE VM_MAYWRITE
1995 #define NOMMU_MAP_EXEC VM_MAYEXEC
1997 #define NOMMU_VMFLAGS \
1998 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
2001 * These flags control the behavior of the remap_file_range function pointer.
2002 * If it is called with len == 0 that means "remap to end of source file".
2003 * See Documentation/filesystems/vfs.rst for more details about this call.
2005 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
2006 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
2008 #define REMAP_FILE_DEDUP (1 << 0)
2009 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
2012 * These flags signal that the caller is ok with altering various aspects of
2013 * the behavior of the remap operation. The changes must be made by the
2014 * implementation; the vfs remap helper functions can take advantage of them.
2015 * Flags in this category exist to preserve the quirky behavior of the hoisted
2016 * btrfs clone/dedupe ioctls.
2018 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
2022 struct file_operations
{
2023 struct module
*owner
;
2024 loff_t (*llseek
) (struct file
*, loff_t
, int);
2025 ssize_t (*read
) (struct file
*, char __user
*, size_t, loff_t
*);
2026 ssize_t (*write
) (struct file
*, const char __user
*, size_t, loff_t
*);
2027 ssize_t (*read_iter
) (struct kiocb
*, struct iov_iter
*);
2028 ssize_t (*write_iter
) (struct kiocb
*, struct iov_iter
*);
2029 int (*iopoll
)(struct kiocb
*kiocb
, bool spin
);
2030 int (*iterate
) (struct file
*, struct dir_context
*);
2031 int (*iterate_shared
) (struct file
*, struct dir_context
*);
2032 __poll_t (*poll
) (struct file
*, struct poll_table_struct
*);
2033 long (*unlocked_ioctl
) (struct file
*, unsigned int, unsigned long);
2034 long (*compat_ioctl
) (struct file
*, unsigned int, unsigned long);
2035 int (*mmap
) (struct file
*, struct vm_area_struct
*);
2036 unsigned long mmap_supported_flags
;
2037 int (*open
) (struct inode
*, struct file
*);
2038 int (*flush
) (struct file
*, fl_owner_t id
);
2039 int (*release
) (struct inode
*, struct file
*);
2040 int (*fsync
) (struct file
*, loff_t
, loff_t
, int datasync
);
2041 int (*fasync
) (int, struct file
*, int);
2042 int (*lock
) (struct file
*, int, struct file_lock
*);
2043 ssize_t (*sendpage
) (struct file
*, struct page
*, int, size_t, loff_t
*, int);
2044 unsigned long (*get_unmapped_area
)(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
2045 int (*check_flags
)(int);
2046 int (*flock
) (struct file
*, int, struct file_lock
*);
2047 ssize_t (*splice_write
)(struct pipe_inode_info
*, struct file
*, loff_t
*, size_t, unsigned int);
2048 ssize_t (*splice_read
)(struct file
*, loff_t
*, struct pipe_inode_info
*, size_t, unsigned int);
2049 int (*setlease
)(struct file
*, long, struct file_lock
**, void **);
2050 long (*fallocate
)(struct file
*file
, int mode
, loff_t offset
,
2052 void (*show_fdinfo
)(struct seq_file
*m
, struct file
*f
);
2054 unsigned (*mmap_capabilities
)(struct file
*);
2056 ssize_t (*copy_file_range
)(struct file
*, loff_t
, struct file
*,
2057 loff_t
, size_t, unsigned int);
2058 loff_t (*remap_file_range
)(struct file
*file_in
, loff_t pos_in
,
2059 struct file
*file_out
, loff_t pos_out
,
2060 loff_t len
, unsigned int remap_flags
);
2061 int (*fadvise
)(struct file
*, loff_t
, loff_t
, int);
2062 } __randomize_layout
;
2064 struct inode_operations
{
2065 struct dentry
* (*lookup
) (struct inode
*,struct dentry
*, unsigned int);
2066 const char * (*get_link
) (struct dentry
*, struct inode
*, struct delayed_call
*);
2067 int (*permission
) (struct user_namespace
*, struct inode
*, int);
2068 struct posix_acl
* (*get_acl
)(struct inode
*, int);
2070 int (*readlink
) (struct dentry
*, char __user
*,int);
2072 int (*create
) (struct user_namespace
*, struct inode
*,struct dentry
*,
2074 int (*link
) (struct dentry
*,struct inode
*,struct dentry
*);
2075 int (*unlink
) (struct inode
*,struct dentry
*);
2076 int (*symlink
) (struct user_namespace
*, struct inode
*,struct dentry
*,
2078 int (*mkdir
) (struct user_namespace
*, struct inode
*,struct dentry
*,
2080 int (*rmdir
) (struct inode
*,struct dentry
*);
2081 int (*mknod
) (struct user_namespace
*, struct inode
*,struct dentry
*,
2083 int (*rename
) (struct user_namespace
*, struct inode
*, struct dentry
*,
2084 struct inode
*, struct dentry
*, unsigned int);
2085 int (*setattr
) (struct user_namespace
*, struct dentry
*,
2087 int (*getattr
) (struct user_namespace
*, const struct path
*,
2088 struct kstat
*, u32
, unsigned int);
2089 ssize_t (*listxattr
) (struct dentry
*, char *, size_t);
2090 int (*fiemap
)(struct inode
*, struct fiemap_extent_info
*, u64 start
,
2092 int (*update_time
)(struct inode
*, struct timespec64
*, int);
2093 int (*atomic_open
)(struct inode
*, struct dentry
*,
2094 struct file
*, unsigned open_flag
,
2095 umode_t create_mode
);
2096 int (*tmpfile
) (struct user_namespace
*, struct inode
*,
2097 struct dentry
*, umode_t
);
2098 int (*set_acl
)(struct user_namespace
*, struct inode
*,
2099 struct posix_acl
*, int);
2100 int (*fileattr_set
)(struct user_namespace
*mnt_userns
,
2101 struct dentry
*dentry
, struct fileattr
*fa
);
2102 int (*fileattr_get
)(struct dentry
*dentry
, struct fileattr
*fa
);
2103 } ____cacheline_aligned
;
2105 static inline ssize_t
call_read_iter(struct file
*file
, struct kiocb
*kio
,
2106 struct iov_iter
*iter
)
2108 return file
->f_op
->read_iter(kio
, iter
);
2111 static inline ssize_t
call_write_iter(struct file
*file
, struct kiocb
*kio
,
2112 struct iov_iter
*iter
)
2114 return file
->f_op
->write_iter(kio
, iter
);
2117 static inline int call_mmap(struct file
*file
, struct vm_area_struct
*vma
)
2119 return file
->f_op
->mmap(file
, vma
);
2122 extern ssize_t
vfs_read(struct file
*, char __user
*, size_t, loff_t
*);
2123 extern ssize_t
vfs_write(struct file
*, const char __user
*, size_t, loff_t
*);
2124 extern ssize_t
vfs_copy_file_range(struct file
*, loff_t
, struct file
*,
2125 loff_t
, size_t, unsigned int);
2126 extern ssize_t
generic_copy_file_range(struct file
*file_in
, loff_t pos_in
,
2127 struct file
*file_out
, loff_t pos_out
,
2128 size_t len
, unsigned int flags
);
2129 extern int generic_remap_file_range_prep(struct file
*file_in
, loff_t pos_in
,
2130 struct file
*file_out
, loff_t pos_out
,
2132 unsigned int remap_flags
);
2133 extern loff_t
do_clone_file_range(struct file
*file_in
, loff_t pos_in
,
2134 struct file
*file_out
, loff_t pos_out
,
2135 loff_t len
, unsigned int remap_flags
);
2136 extern loff_t
vfs_clone_file_range(struct file
*file_in
, loff_t pos_in
,
2137 struct file
*file_out
, loff_t pos_out
,
2138 loff_t len
, unsigned int remap_flags
);
2139 extern int vfs_dedupe_file_range(struct file
*file
,
2140 struct file_dedupe_range
*same
);
2141 extern loff_t
vfs_dedupe_file_range_one(struct file
*src_file
, loff_t src_pos
,
2142 struct file
*dst_file
, loff_t dst_pos
,
2143 loff_t len
, unsigned int remap_flags
);
2146 struct super_operations
{
2147 struct inode
*(*alloc_inode
)(struct super_block
*sb
);
2148 void (*destroy_inode
)(struct inode
*);
2149 void (*free_inode
)(struct inode
*);
2151 void (*dirty_inode
) (struct inode
*, int flags
);
2152 int (*write_inode
) (struct inode
*, struct writeback_control
*wbc
);
2153 int (*drop_inode
) (struct inode
*);
2154 void (*evict_inode
) (struct inode
*);
2155 void (*put_super
) (struct super_block
*);
2156 int (*sync_fs
)(struct super_block
*sb
, int wait
);
2157 int (*freeze_super
) (struct super_block
*);
2158 int (*freeze_fs
) (struct super_block
*);
2159 int (*thaw_super
) (struct super_block
*);
2160 int (*unfreeze_fs
) (struct super_block
*);
2161 int (*statfs
) (struct dentry
*, struct kstatfs
*);
2162 int (*remount_fs
) (struct super_block
*, int *, char *);
2163 void (*umount_begin
) (struct super_block
*);
2165 int (*show_options
)(struct seq_file
*, struct dentry
*);
2166 int (*show_devname
)(struct seq_file
*, struct dentry
*);
2167 int (*show_path
)(struct seq_file
*, struct dentry
*);
2168 int (*show_stats
)(struct seq_file
*, struct dentry
*);
2170 ssize_t (*quota_read
)(struct super_block
*, int, char *, size_t, loff_t
);
2171 ssize_t (*quota_write
)(struct super_block
*, int, const char *, size_t, loff_t
);
2172 struct dquot
**(*get_dquots
)(struct inode
*);
2174 long (*nr_cached_objects
)(struct super_block
*,
2175 struct shrink_control
*);
2176 long (*free_cached_objects
)(struct super_block
*,
2177 struct shrink_control
*);
2181 * Inode flags - they have no relation to superblock flags now
2183 #define S_SYNC (1 << 0) /* Writes are synced at once */
2184 #define S_NOATIME (1 << 1) /* Do not update access times */
2185 #define S_APPEND (1 << 2) /* Append-only file */
2186 #define S_IMMUTABLE (1 << 3) /* Immutable file */
2187 #define S_DEAD (1 << 4) /* removed, but still open directory */
2188 #define S_NOQUOTA (1 << 5) /* Inode is not counted to quota */
2189 #define S_DIRSYNC (1 << 6) /* Directory modifications are synchronous */
2190 #define S_NOCMTIME (1 << 7) /* Do not update file c/mtime */
2191 #define S_SWAPFILE (1 << 8) /* Do not truncate: swapon got its bmaps */
2192 #define S_PRIVATE (1 << 9) /* Inode is fs-internal */
2193 #define S_IMA (1 << 10) /* Inode has an associated IMA struct */
2194 #define S_AUTOMOUNT (1 << 11) /* Automount/referral quasi-directory */
2195 #define S_NOSEC (1 << 12) /* no suid or xattr security attributes */
2196 #ifdef CONFIG_FS_DAX
2197 #define S_DAX (1 << 13) /* Direct Access, avoiding the page cache */
2199 #define S_DAX 0 /* Make all the DAX code disappear */
2201 #define S_ENCRYPTED (1 << 14) /* Encrypted file (using fs/crypto/) */
2202 #define S_CASEFOLD (1 << 15) /* Casefolded file */
2203 #define S_VERITY (1 << 16) /* Verity file (using fs/verity/) */
2206 * Note that nosuid etc flags are inode-specific: setting some file-system
2207 * flags just means all the inodes inherit those flags by default. It might be
2208 * possible to override it selectively if you really wanted to with some
2209 * ioctl() that is not currently implemented.
2211 * Exception: SB_RDONLY is always applied to the entire file system.
2213 * Unfortunately, it is possible to change a filesystems flags with it mounted
2214 * with files in use. This means that all of the inodes will not have their
2215 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
2216 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
2218 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
2220 static inline bool sb_rdonly(const struct super_block
*sb
) { return sb
->s_flags
& SB_RDONLY
; }
2221 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
2222 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
2223 ((inode)->i_flags & S_SYNC))
2224 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
2225 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
2226 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
2227 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
2228 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
2230 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
2231 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
2232 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
2233 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
2235 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
2236 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
2237 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
2238 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
2239 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
2240 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2241 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2242 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2243 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2244 #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD)
2245 #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY)
2247 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2248 (inode)->i_rdev == WHITEOUT_DEV)
2250 static inline bool HAS_UNMAPPED_ID(struct user_namespace
*mnt_userns
,
2251 struct inode
*inode
)
2253 return !uid_valid(i_uid_into_mnt(mnt_userns
, inode
)) ||
2254 !gid_valid(i_gid_into_mnt(mnt_userns
, inode
));
2257 static inline enum rw_hint
file_write_hint(struct file
*file
)
2259 if (file
->f_write_hint
!= WRITE_LIFE_NOT_SET
)
2260 return file
->f_write_hint
;
2262 return file_inode(file
)->i_write_hint
;
2265 static inline int iocb_flags(struct file
*file
);
2267 static inline u16
ki_hint_validate(enum rw_hint hint
)
2269 typeof(((struct kiocb
*)0)->ki_hint
) max_hint
= -1;
2271 if (hint
<= max_hint
)
2276 static inline void init_sync_kiocb(struct kiocb
*kiocb
, struct file
*filp
)
2278 *kiocb
= (struct kiocb
) {
2280 .ki_flags
= iocb_flags(filp
),
2281 .ki_hint
= ki_hint_validate(file_write_hint(filp
)),
2282 .ki_ioprio
= get_current_ioprio(),
2286 static inline void kiocb_clone(struct kiocb
*kiocb
, struct kiocb
*kiocb_src
,
2289 *kiocb
= (struct kiocb
) {
2291 .ki_flags
= kiocb_src
->ki_flags
,
2292 .ki_hint
= kiocb_src
->ki_hint
,
2293 .ki_ioprio
= kiocb_src
->ki_ioprio
,
2294 .ki_pos
= kiocb_src
->ki_pos
,
2299 * Inode state bits. Protected by inode->i_lock
2301 * Four bits determine the dirty state of the inode: I_DIRTY_SYNC,
2302 * I_DIRTY_DATASYNC, I_DIRTY_PAGES, and I_DIRTY_TIME.
2304 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2305 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2306 * various stages of removing an inode.
2308 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2310 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2311 * fdatasync() (unless I_DIRTY_DATASYNC is also set).
2312 * Timestamp updates are the usual cause.
2313 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2314 * these changes separately from I_DIRTY_SYNC so that we
2315 * don't have to write inode on fdatasync() when only
2316 * e.g. the timestamps have changed.
2317 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2318 * I_DIRTY_TIME The inode itself only has dirty timestamps, and the
2319 * lazytime mount option is enabled. We keep track of this
2320 * separately from I_DIRTY_SYNC in order to implement
2321 * lazytime. This gets cleared if I_DIRTY_INODE
2322 * (I_DIRTY_SYNC and/or I_DIRTY_DATASYNC) gets set. I.e.
2323 * either I_DIRTY_TIME *or* I_DIRTY_INODE can be set in
2324 * i_state, but not both. I_DIRTY_PAGES may still be set.
2325 * I_NEW Serves as both a mutex and completion notification.
2326 * New inodes set I_NEW. If two processes both create
2327 * the same inode, one of them will release its inode and
2328 * wait for I_NEW to be released before returning.
2329 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2330 * also cause waiting on I_NEW, without I_NEW actually
2331 * being set. find_inode() uses this to prevent returning
2332 * nearly-dead inodes.
2333 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2334 * is zero. I_FREEING must be set when I_WILL_FREE is
2336 * I_FREEING Set when inode is about to be freed but still has dirty
2337 * pages or buffers attached or the inode itself is still
2339 * I_CLEAR Added by clear_inode(). In this state the inode is
2340 * clean and can be destroyed. Inode keeps I_FREEING.
2342 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2343 * prohibited for many purposes. iget() must wait for
2344 * the inode to be completely released, then create it
2345 * anew. Other functions will just ignore such inodes,
2346 * if appropriate. I_NEW is used for waiting.
2348 * I_SYNC Writeback of inode is running. The bit is set during
2349 * data writeback, and cleared with a wakeup on the bit
2350 * address once it is done. The bit is also used to pin
2351 * the inode in memory for flusher thread.
2353 * I_REFERENCED Marks the inode as recently references on the LRU list.
2355 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2357 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2358 * synchronize competing switching instances and to tell
2359 * wb stat updates to grab the i_pages lock. See
2360 * inode_switch_wbs_work_fn() for details.
2362 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2363 * and work dirs among overlayfs mounts.
2365 * I_CREATING New object's inode in the middle of setting up.
2367 * I_DONTCACHE Evict inode as soon as it is not used anymore.
2369 * I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists.
2370 * Used to detect that mark_inode_dirty() should not move
2371 * inode between dirty lists.
2373 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2375 #define I_DIRTY_SYNC (1 << 0)
2376 #define I_DIRTY_DATASYNC (1 << 1)
2377 #define I_DIRTY_PAGES (1 << 2)
2379 #define I_NEW (1 << __I_NEW)
2380 #define I_WILL_FREE (1 << 4)
2381 #define I_FREEING (1 << 5)
2382 #define I_CLEAR (1 << 6)
2384 #define I_SYNC (1 << __I_SYNC)
2385 #define I_REFERENCED (1 << 8)
2386 #define __I_DIO_WAKEUP 9
2387 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2388 #define I_LINKABLE (1 << 10)
2389 #define I_DIRTY_TIME (1 << 11)
2390 #define I_WB_SWITCH (1 << 13)
2391 #define I_OVL_INUSE (1 << 14)
2392 #define I_CREATING (1 << 15)
2393 #define I_DONTCACHE (1 << 16)
2394 #define I_SYNC_QUEUED (1 << 17)
2396 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2397 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2398 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2400 extern void __mark_inode_dirty(struct inode
*, int);
2401 static inline void mark_inode_dirty(struct inode
*inode
)
2403 __mark_inode_dirty(inode
, I_DIRTY
);
2406 static inline void mark_inode_dirty_sync(struct inode
*inode
)
2408 __mark_inode_dirty(inode
, I_DIRTY_SYNC
);
2412 * Returns true if the given inode itself only has dirty timestamps (its pages
2413 * may still be dirty) and isn't currently being allocated or freed.
2414 * Filesystems should call this if when writing an inode when lazytime is
2415 * enabled, they want to opportunistically write the timestamps of other inodes
2416 * located very nearby on-disk, e.g. in the same inode block. This returns true
2417 * if the given inode is in need of such an opportunistic update. Requires
2418 * i_lock, or at least later re-checking under i_lock.
2420 static inline bool inode_is_dirtytime_only(struct inode
*inode
)
2422 return (inode
->i_state
& (I_DIRTY_TIME
| I_NEW
|
2423 I_FREEING
| I_WILL_FREE
)) == I_DIRTY_TIME
;
2426 extern void inc_nlink(struct inode
*inode
);
2427 extern void drop_nlink(struct inode
*inode
);
2428 extern void clear_nlink(struct inode
*inode
);
2429 extern void set_nlink(struct inode
*inode
, unsigned int nlink
);
2431 static inline void inode_inc_link_count(struct inode
*inode
)
2434 mark_inode_dirty(inode
);
2437 static inline void inode_dec_link_count(struct inode
*inode
)
2440 mark_inode_dirty(inode
);
2443 enum file_time_flags
{
2450 extern bool atime_needs_update(const struct path
*, struct inode
*);
2451 extern void touch_atime(const struct path
*);
2452 static inline void file_accessed(struct file
*file
)
2454 if (!(file
->f_flags
& O_NOATIME
))
2455 touch_atime(&file
->f_path
);
2458 extern int file_modified(struct file
*file
);
2460 int sync_inode(struct inode
*inode
, struct writeback_control
*wbc
);
2461 int sync_inode_metadata(struct inode
*inode
, int wait
);
2463 struct file_system_type
{
2466 #define FS_REQUIRES_DEV 1
2467 #define FS_BINARY_MOUNTDATA 2
2468 #define FS_HAS_SUBTYPE 4
2469 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2470 #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */
2471 #define FS_ALLOW_IDMAP 32 /* FS has been updated to handle vfs idmappings. */
2472 #define FS_THP_SUPPORT 8192 /* Remove once all fs converted */
2473 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2474 int (*init_fs_context
)(struct fs_context
*);
2475 const struct fs_parameter_spec
*parameters
;
2476 struct dentry
*(*mount
) (struct file_system_type
*, int,
2477 const char *, void *);
2478 void (*kill_sb
) (struct super_block
*);
2479 struct module
*owner
;
2480 struct file_system_type
* next
;
2481 struct hlist_head fs_supers
;
2483 struct lock_class_key s_lock_key
;
2484 struct lock_class_key s_umount_key
;
2485 struct lock_class_key s_vfs_rename_key
;
2486 struct lock_class_key s_writers_key
[SB_FREEZE_LEVELS
];
2488 struct lock_class_key i_lock_key
;
2489 struct lock_class_key i_mutex_key
;
2490 struct lock_class_key i_mutex_dir_key
;
2493 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2495 extern struct dentry
*mount_bdev(struct file_system_type
*fs_type
,
2496 int flags
, const char *dev_name
, void *data
,
2497 int (*fill_super
)(struct super_block
*, void *, int));
2498 extern struct dentry
*mount_single(struct file_system_type
*fs_type
,
2499 int flags
, void *data
,
2500 int (*fill_super
)(struct super_block
*, void *, int));
2501 extern struct dentry
*mount_nodev(struct file_system_type
*fs_type
,
2502 int flags
, void *data
,
2503 int (*fill_super
)(struct super_block
*, void *, int));
2504 extern struct dentry
*mount_subtree(struct vfsmount
*mnt
, const char *path
);
2505 void generic_shutdown_super(struct super_block
*sb
);
2506 void kill_block_super(struct super_block
*sb
);
2507 void kill_anon_super(struct super_block
*sb
);
2508 void kill_litter_super(struct super_block
*sb
);
2509 void deactivate_super(struct super_block
*sb
);
2510 void deactivate_locked_super(struct super_block
*sb
);
2511 int set_anon_super(struct super_block
*s
, void *data
);
2512 int set_anon_super_fc(struct super_block
*s
, struct fs_context
*fc
);
2513 int get_anon_bdev(dev_t
*);
2514 void free_anon_bdev(dev_t
);
2515 struct super_block
*sget_fc(struct fs_context
*fc
,
2516 int (*test
)(struct super_block
*, struct fs_context
*),
2517 int (*set
)(struct super_block
*, struct fs_context
*));
2518 struct super_block
*sget(struct file_system_type
*type
,
2519 int (*test
)(struct super_block
*,void *),
2520 int (*set
)(struct super_block
*,void *),
2521 int flags
, void *data
);
2523 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2524 #define fops_get(fops) \
2525 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2526 #define fops_put(fops) \
2527 do { if (fops) module_put((fops)->owner); } while(0)
2529 * This one is to be used *ONLY* from ->open() instances.
2530 * fops must be non-NULL, pinned down *and* module dependencies
2531 * should be sufficient to pin the caller down as well.
2533 #define replace_fops(f, fops) \
2535 struct file *__file = (f); \
2536 fops_put(__file->f_op); \
2537 BUG_ON(!(__file->f_op = (fops))); \
2540 extern int register_filesystem(struct file_system_type
*);
2541 extern int unregister_filesystem(struct file_system_type
*);
2542 extern struct vfsmount
*kern_mount(struct file_system_type
*);
2543 extern void kern_unmount(struct vfsmount
*mnt
);
2544 extern int may_umount_tree(struct vfsmount
*);
2545 extern int may_umount(struct vfsmount
*);
2546 extern long do_mount(const char *, const char __user
*,
2547 const char *, unsigned long, void *);
2548 extern struct vfsmount
*collect_mounts(const struct path
*);
2549 extern void drop_collected_mounts(struct vfsmount
*);
2550 extern int iterate_mounts(int (*)(struct vfsmount
*, void *), void *,
2552 extern int vfs_statfs(const struct path
*, struct kstatfs
*);
2553 extern int user_statfs(const char __user
*, struct kstatfs
*);
2554 extern int fd_statfs(int, struct kstatfs
*);
2555 extern int freeze_super(struct super_block
*super
);
2556 extern int thaw_super(struct super_block
*super
);
2557 extern bool our_mnt(struct vfsmount
*mnt
);
2558 extern __printf(2, 3)
2559 int super_setup_bdi_name(struct super_block
*sb
, char *fmt
, ...);
2560 extern int super_setup_bdi(struct super_block
*sb
);
2562 extern int current_umask(void);
2564 extern void ihold(struct inode
* inode
);
2565 extern void iput(struct inode
*);
2566 extern int generic_update_time(struct inode
*, struct timespec64
*, int);
2569 extern struct kobject
*fs_kobj
;
2571 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2573 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2574 extern int locks_mandatory_locked(struct file
*);
2575 extern int locks_mandatory_area(struct inode
*, struct file
*, loff_t
, loff_t
, unsigned char);
2578 * Candidates for mandatory locking have the setgid bit set
2579 * but no group execute bit - an otherwise meaningless combination.
2582 static inline int __mandatory_lock(struct inode
*ino
)
2584 return (ino
->i_mode
& (S_ISGID
| S_IXGRP
)) == S_ISGID
;
2588 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2589 * otherwise these will be advisory locks
2592 static inline int mandatory_lock(struct inode
*ino
)
2594 return IS_MANDLOCK(ino
) && __mandatory_lock(ino
);
2597 static inline int locks_verify_locked(struct file
*file
)
2599 if (mandatory_lock(locks_inode(file
)))
2600 return locks_mandatory_locked(file
);
2604 static inline int locks_verify_truncate(struct inode
*inode
,
2608 if (!inode
->i_flctx
|| !mandatory_lock(inode
))
2611 if (size
< inode
->i_size
) {
2612 return locks_mandatory_area(inode
, f
, size
, inode
->i_size
- 1,
2615 return locks_mandatory_area(inode
, f
, inode
->i_size
, size
- 1,
2620 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2622 static inline int locks_mandatory_locked(struct file
*file
)
2627 static inline int locks_mandatory_area(struct inode
*inode
, struct file
*filp
,
2628 loff_t start
, loff_t end
, unsigned char type
)
2633 static inline int __mandatory_lock(struct inode
*inode
)
2638 static inline int mandatory_lock(struct inode
*inode
)
2643 static inline int locks_verify_locked(struct file
*file
)
2648 static inline int locks_verify_truncate(struct inode
*inode
, struct file
*filp
,
2654 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2657 #ifdef CONFIG_FILE_LOCKING
2658 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2661 * Since this check is lockless, we must ensure that any refcounts
2662 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2663 * could end up racing with tasks trying to set a new lease on this
2667 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2668 return __break_lease(inode
, mode
, FL_LEASE
);
2672 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2675 * Since this check is lockless, we must ensure that any refcounts
2676 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2677 * could end up racing with tasks trying to set a new lease on this
2681 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2682 return __break_lease(inode
, mode
, FL_DELEG
);
2686 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2690 ret
= break_deleg(inode
, O_WRONLY
|O_NONBLOCK
);
2691 if (ret
== -EWOULDBLOCK
&& delegated_inode
) {
2692 *delegated_inode
= inode
;
2698 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2702 ret
= break_deleg(*delegated_inode
, O_WRONLY
);
2703 iput(*delegated_inode
);
2704 *delegated_inode
= NULL
;
2708 static inline int break_layout(struct inode
*inode
, bool wait
)
2711 if (inode
->i_flctx
&& !list_empty_careful(&inode
->i_flctx
->flc_lease
))
2712 return __break_lease(inode
,
2713 wait
? O_WRONLY
: O_WRONLY
| O_NONBLOCK
,
2718 #else /* !CONFIG_FILE_LOCKING */
2719 static inline int break_lease(struct inode
*inode
, unsigned int mode
)
2724 static inline int break_deleg(struct inode
*inode
, unsigned int mode
)
2729 static inline int try_break_deleg(struct inode
*inode
, struct inode
**delegated_inode
)
2734 static inline int break_deleg_wait(struct inode
**delegated_inode
)
2740 static inline int break_layout(struct inode
*inode
, bool wait
)
2745 #endif /* CONFIG_FILE_LOCKING */
2750 const char *name
; /* pointer to actual string */
2751 const __user
char *uptr
; /* original userland pointer */
2753 struct audit_names
*aname
;
2756 static_assert(offsetof(struct filename
, iname
) % sizeof(long) == 0);
2758 static inline struct user_namespace
*file_mnt_user_ns(struct file
*file
)
2760 return mnt_user_ns(file
->f_path
.mnt
);
2762 extern long vfs_truncate(const struct path
*, loff_t
);
2763 int do_truncate(struct user_namespace
*, struct dentry
*, loff_t start
,
2764 unsigned int time_attrs
, struct file
*filp
);
2765 extern int vfs_fallocate(struct file
*file
, int mode
, loff_t offset
,
2767 extern long do_sys_open(int dfd
, const char __user
*filename
, int flags
,
2769 extern struct file
*file_open_name(struct filename
*, int, umode_t
);
2770 extern struct file
*filp_open(const char *, int, umode_t
);
2771 extern struct file
*file_open_root(const struct path
*,
2772 const char *, int, umode_t
);
2773 static inline struct file
*file_open_root_mnt(struct vfsmount
*mnt
,
2774 const char *name
, int flags
, umode_t mode
)
2776 return file_open_root(&(struct path
){.mnt
= mnt
, .dentry
= mnt
->mnt_root
},
2779 extern struct file
* dentry_open(const struct path
*, int, const struct cred
*);
2780 extern struct file
* open_with_fake_path(const struct path
*, int,
2781 struct inode
*, const struct cred
*);
2782 static inline struct file
*file_clone_open(struct file
*file
)
2784 return dentry_open(&file
->f_path
, file
->f_flags
, file
->f_cred
);
2786 extern int filp_close(struct file
*, fl_owner_t id
);
2788 extern struct filename
*getname_flags(const char __user
*, int, int *);
2789 extern struct filename
*getname(const char __user
*);
2790 extern struct filename
*getname_kernel(const char *);
2791 extern void putname(struct filename
*name
);
2793 extern int finish_open(struct file
*file
, struct dentry
*dentry
,
2794 int (*open
)(struct inode
*, struct file
*));
2795 extern int finish_no_open(struct file
*file
, struct dentry
*dentry
);
2798 extern void __init
vfs_caches_init_early(void);
2799 extern void __init
vfs_caches_init(void);
2801 extern struct kmem_cache
*names_cachep
;
2803 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2804 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2806 extern struct super_block
*blockdev_superblock
;
2807 static inline bool sb_is_blkdev_sb(struct super_block
*sb
)
2809 return IS_ENABLED(CONFIG_BLOCK
) && sb
== blockdev_superblock
;
2812 void emergency_thaw_all(void);
2813 extern int sync_filesystem(struct super_block
*);
2814 extern const struct file_operations def_blk_fops
;
2815 extern const struct file_operations def_chr_fops
;
2818 #define CHRDEV_MAJOR_MAX 512
2819 /* Marks the bottom of the first segment of free char majors */
2820 #define CHRDEV_MAJOR_DYN_END 234
2821 /* Marks the top and bottom of the second segment of free char majors */
2822 #define CHRDEV_MAJOR_DYN_EXT_START 511
2823 #define CHRDEV_MAJOR_DYN_EXT_END 384
2825 extern int alloc_chrdev_region(dev_t
*, unsigned, unsigned, const char *);
2826 extern int register_chrdev_region(dev_t
, unsigned, const char *);
2827 extern int __register_chrdev(unsigned int major
, unsigned int baseminor
,
2828 unsigned int count
, const char *name
,
2829 const struct file_operations
*fops
);
2830 extern void __unregister_chrdev(unsigned int major
, unsigned int baseminor
,
2831 unsigned int count
, const char *name
);
2832 extern void unregister_chrdev_region(dev_t
, unsigned);
2833 extern void chrdev_show(struct seq_file
*,off_t
);
2835 static inline int register_chrdev(unsigned int major
, const char *name
,
2836 const struct file_operations
*fops
)
2838 return __register_chrdev(major
, 0, 256, name
, fops
);
2841 static inline void unregister_chrdev(unsigned int major
, const char *name
)
2843 __unregister_chrdev(major
, 0, 256, name
);
2846 extern void init_special_inode(struct inode
*, umode_t
, dev_t
);
2848 /* Invalid inode operations -- fs/bad_inode.c */
2849 extern void make_bad_inode(struct inode
*);
2850 extern bool is_bad_inode(struct inode
*);
2852 unsigned long invalidate_mapping_pages(struct address_space
*mapping
,
2853 pgoff_t start
, pgoff_t end
);
2855 void invalidate_mapping_pagevec(struct address_space
*mapping
,
2856 pgoff_t start
, pgoff_t end
,
2857 unsigned long *nr_pagevec
);
2859 static inline void invalidate_remote_inode(struct inode
*inode
)
2861 if (S_ISREG(inode
->i_mode
) || S_ISDIR(inode
->i_mode
) ||
2862 S_ISLNK(inode
->i_mode
))
2863 invalidate_mapping_pages(inode
->i_mapping
, 0, -1);
2865 extern int invalidate_inode_pages2(struct address_space
*mapping
);
2866 extern int invalidate_inode_pages2_range(struct address_space
*mapping
,
2867 pgoff_t start
, pgoff_t end
);
2868 extern int write_inode_now(struct inode
*, int);
2869 extern int filemap_fdatawrite(struct address_space
*);
2870 extern int filemap_flush(struct address_space
*);
2871 extern int filemap_fdatawait_keep_errors(struct address_space
*mapping
);
2872 extern int filemap_fdatawait_range(struct address_space
*, loff_t lstart
,
2874 extern int filemap_fdatawait_range_keep_errors(struct address_space
*mapping
,
2875 loff_t start_byte
, loff_t end_byte
);
2877 static inline int filemap_fdatawait(struct address_space
*mapping
)
2879 return filemap_fdatawait_range(mapping
, 0, LLONG_MAX
);
2882 extern bool filemap_range_has_page(struct address_space
*, loff_t lstart
,
2884 extern bool filemap_range_needs_writeback(struct address_space
*,
2885 loff_t lstart
, loff_t lend
);
2886 extern int filemap_write_and_wait_range(struct address_space
*mapping
,
2887 loff_t lstart
, loff_t lend
);
2888 extern int __filemap_fdatawrite_range(struct address_space
*mapping
,
2889 loff_t start
, loff_t end
, int sync_mode
);
2890 extern int filemap_fdatawrite_range(struct address_space
*mapping
,
2891 loff_t start
, loff_t end
);
2892 extern int filemap_check_errors(struct address_space
*mapping
);
2893 extern void __filemap_set_wb_err(struct address_space
*mapping
, int err
);
2895 static inline int filemap_write_and_wait(struct address_space
*mapping
)
2897 return filemap_write_and_wait_range(mapping
, 0, LLONG_MAX
);
2900 extern int __must_check
file_fdatawait_range(struct file
*file
, loff_t lstart
,
2902 extern int __must_check
file_check_and_advance_wb_err(struct file
*file
);
2903 extern int __must_check
file_write_and_wait_range(struct file
*file
,
2904 loff_t start
, loff_t end
);
2906 static inline int file_write_and_wait(struct file
*file
)
2908 return file_write_and_wait_range(file
, 0, LLONG_MAX
);
2912 * filemap_set_wb_err - set a writeback error on an address_space
2913 * @mapping: mapping in which to set writeback error
2914 * @err: error to be set in mapping
2916 * When writeback fails in some way, we must record that error so that
2917 * userspace can be informed when fsync and the like are called. We endeavor
2918 * to report errors on any file that was open at the time of the error. Some
2919 * internal callers also need to know when writeback errors have occurred.
2921 * When a writeback error occurs, most filesystems will want to call
2922 * filemap_set_wb_err to record the error in the mapping so that it will be
2923 * automatically reported whenever fsync is called on the file.
2925 static inline void filemap_set_wb_err(struct address_space
*mapping
, int err
)
2927 /* Fastpath for common case of no error */
2929 __filemap_set_wb_err(mapping
, err
);
2933 * filemap_check_wb_err - has an error occurred since the mark was sampled?
2934 * @mapping: mapping to check for writeback errors
2935 * @since: previously-sampled errseq_t
2937 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2938 * the given value was sampled.
2940 * If it has then report the latest error set, otherwise return 0.
2942 static inline int filemap_check_wb_err(struct address_space
*mapping
,
2945 return errseq_check(&mapping
->wb_err
, since
);
2949 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2950 * @mapping: mapping to be sampled
2952 * Writeback errors are always reported relative to a particular sample point
2953 * in the past. This function provides those sample points.
2955 static inline errseq_t
filemap_sample_wb_err(struct address_space
*mapping
)
2957 return errseq_sample(&mapping
->wb_err
);
2961 * file_sample_sb_err - sample the current errseq_t to test for later errors
2962 * @file: file pointer to be sampled
2964 * Grab the most current superblock-level errseq_t value for the given
2967 static inline errseq_t
file_sample_sb_err(struct file
*file
)
2969 return errseq_sample(&file
->f_path
.dentry
->d_sb
->s_wb_err
);
2972 extern int vfs_fsync_range(struct file
*file
, loff_t start
, loff_t end
,
2974 extern int vfs_fsync(struct file
*file
, int datasync
);
2976 extern int sync_file_range(struct file
*file
, loff_t offset
, loff_t nbytes
,
2977 unsigned int flags
);
2980 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2981 * to already be updated for the write, and will return either the amount
2982 * of bytes passed in, or an error if syncing the file failed.
2984 static inline ssize_t
generic_write_sync(struct kiocb
*iocb
, ssize_t count
)
2986 if (iocb
->ki_flags
& IOCB_DSYNC
) {
2987 int ret
= vfs_fsync_range(iocb
->ki_filp
,
2988 iocb
->ki_pos
- count
, iocb
->ki_pos
- 1,
2989 (iocb
->ki_flags
& IOCB_SYNC
) ? 0 : 1);
2997 extern void emergency_sync(void);
2998 extern void emergency_remount(void);
3001 extern int bmap(struct inode
*inode
, sector_t
*block
);
3003 static inline int bmap(struct inode
*inode
, sector_t
*block
)
3009 int notify_change(struct user_namespace
*, struct dentry
*,
3010 struct iattr
*, struct inode
**);
3011 int inode_permission(struct user_namespace
*, struct inode
*, int);
3012 int generic_permission(struct user_namespace
*, struct inode
*, int);
3013 static inline int file_permission(struct file
*file
, int mask
)
3015 return inode_permission(file_mnt_user_ns(file
),
3016 file_inode(file
), mask
);
3018 static inline int path_permission(const struct path
*path
, int mask
)
3020 return inode_permission(mnt_user_ns(path
->mnt
),
3021 d_inode(path
->dentry
), mask
);
3023 int __check_sticky(struct user_namespace
*mnt_userns
, struct inode
*dir
,
3024 struct inode
*inode
);
3026 static inline bool execute_ok(struct inode
*inode
)
3028 return (inode
->i_mode
& S_IXUGO
) || S_ISDIR(inode
->i_mode
);
3031 static inline bool inode_wrong_type(const struct inode
*inode
, umode_t mode
)
3033 return (inode
->i_mode
^ mode
) & S_IFMT
;
3036 static inline void file_start_write(struct file
*file
)
3038 if (!S_ISREG(file_inode(file
)->i_mode
))
3040 sb_start_write(file_inode(file
)->i_sb
);
3043 static inline bool file_start_write_trylock(struct file
*file
)
3045 if (!S_ISREG(file_inode(file
)->i_mode
))
3047 return sb_start_write_trylock(file_inode(file
)->i_sb
);
3050 static inline void file_end_write(struct file
*file
)
3052 if (!S_ISREG(file_inode(file
)->i_mode
))
3054 __sb_end_write(file_inode(file
)->i_sb
, SB_FREEZE_WRITE
);
3058 * get_write_access() gets write permission for a file.
3059 * put_write_access() releases this write permission.
3060 * This is used for regular files.
3061 * We cannot support write (and maybe mmap read-write shared) accesses and
3062 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
3063 * can have the following values:
3064 * 0: no writers, no VM_DENYWRITE mappings
3065 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
3066 * > 0: (i_writecount) users are writing to the file.
3068 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
3069 * except for the cases where we don't hold i_writecount yet. Then we need to
3070 * use {get,deny}_write_access() - these functions check the sign and refuse
3071 * to do the change if sign is wrong.
3073 static inline int get_write_access(struct inode
*inode
)
3075 return atomic_inc_unless_negative(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
3077 static inline int deny_write_access(struct file
*file
)
3079 struct inode
*inode
= file_inode(file
);
3080 return atomic_dec_unless_positive(&inode
->i_writecount
) ? 0 : -ETXTBSY
;
3082 static inline void put_write_access(struct inode
* inode
)
3084 atomic_dec(&inode
->i_writecount
);
3086 static inline void allow_write_access(struct file
*file
)
3089 atomic_inc(&file_inode(file
)->i_writecount
);
3091 static inline bool inode_is_open_for_write(const struct inode
*inode
)
3093 return atomic_read(&inode
->i_writecount
) > 0;
3096 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
3097 static inline void i_readcount_dec(struct inode
*inode
)
3099 BUG_ON(!atomic_read(&inode
->i_readcount
));
3100 atomic_dec(&inode
->i_readcount
);
3102 static inline void i_readcount_inc(struct inode
*inode
)
3104 atomic_inc(&inode
->i_readcount
);
3107 static inline void i_readcount_dec(struct inode
*inode
)
3111 static inline void i_readcount_inc(struct inode
*inode
)
3116 extern int do_pipe_flags(int *, int);
3118 extern ssize_t
kernel_read(struct file
*, void *, size_t, loff_t
*);
3119 ssize_t
__kernel_read(struct file
*file
, void *buf
, size_t count
, loff_t
*pos
);
3120 extern ssize_t
kernel_write(struct file
*, const void *, size_t, loff_t
*);
3121 extern ssize_t
__kernel_write(struct file
*, const void *, size_t, loff_t
*);
3122 extern struct file
* open_exec(const char *);
3124 /* fs/dcache.c -- generic fs support functions */
3125 extern bool is_subdir(struct dentry
*, struct dentry
*);
3126 extern bool path_is_under(const struct path
*, const struct path
*);
3128 extern char *file_path(struct file
*, char *, int);
3130 #include <linux/err.h>
3132 /* needed for stackable file system support */
3133 extern loff_t
default_llseek(struct file
*file
, loff_t offset
, int whence
);
3135 extern loff_t
vfs_llseek(struct file
*file
, loff_t offset
, int whence
);
3137 extern int inode_init_always(struct super_block
*, struct inode
*);
3138 extern void inode_init_once(struct inode
*);
3139 extern void address_space_init_once(struct address_space
*mapping
);
3140 extern struct inode
* igrab(struct inode
*);
3141 extern ino_t
iunique(struct super_block
*, ino_t
);
3142 extern int inode_needs_sync(struct inode
*inode
);
3143 extern int generic_delete_inode(struct inode
*inode
);
3144 static inline int generic_drop_inode(struct inode
*inode
)
3146 return !inode
->i_nlink
|| inode_unhashed(inode
);
3148 extern void d_mark_dontcache(struct inode
*inode
);
3150 extern struct inode
*ilookup5_nowait(struct super_block
*sb
,
3151 unsigned long hashval
, int (*test
)(struct inode
*, void *),
3153 extern struct inode
*ilookup5(struct super_block
*sb
, unsigned long hashval
,
3154 int (*test
)(struct inode
*, void *), void *data
);
3155 extern struct inode
*ilookup(struct super_block
*sb
, unsigned long ino
);
3157 extern struct inode
*inode_insert5(struct inode
*inode
, unsigned long hashval
,
3158 int (*test
)(struct inode
*, void *),
3159 int (*set
)(struct inode
*, void *),
3161 extern struct inode
* iget5_locked(struct super_block
*, unsigned long, int (*test
)(struct inode
*, void *), int (*set
)(struct inode
*, void *), void *);
3162 extern struct inode
* iget_locked(struct super_block
*, unsigned long);
3163 extern struct inode
*find_inode_nowait(struct super_block
*,
3165 int (*match
)(struct inode
*,
3166 unsigned long, void *),
3168 extern struct inode
*find_inode_rcu(struct super_block
*, unsigned long,
3169 int (*)(struct inode
*, void *), void *);
3170 extern struct inode
*find_inode_by_ino_rcu(struct super_block
*, unsigned long);
3171 extern int insert_inode_locked4(struct inode
*, unsigned long, int (*test
)(struct inode
*, void *), void *);
3172 extern int insert_inode_locked(struct inode
*);
3173 #ifdef CONFIG_DEBUG_LOCK_ALLOC
3174 extern void lockdep_annotate_inode_mutex_key(struct inode
*inode
);
3176 static inline void lockdep_annotate_inode_mutex_key(struct inode
*inode
) { };
3178 extern void unlock_new_inode(struct inode
*);
3179 extern void discard_new_inode(struct inode
*);
3180 extern unsigned int get_next_ino(void);
3181 extern void evict_inodes(struct super_block
*sb
);
3184 * Userspace may rely on the the inode number being non-zero. For example, glibc
3185 * simply ignores files with zero i_ino in unlink() and other places.
3187 * As an additional complication, if userspace was compiled with
3188 * _FILE_OFFSET_BITS=32 on a 64-bit kernel we'll only end up reading out the
3189 * lower 32 bits, so we need to check that those aren't zero explicitly. With
3190 * _FILE_OFFSET_BITS=64, this may cause some harmless false-negatives, but
3191 * better safe than sorry.
3193 static inline bool is_zero_ino(ino_t ino
)
3195 return (u32
)ino
== 0;
3198 extern void __iget(struct inode
* inode
);
3199 extern void iget_failed(struct inode
*);
3200 extern void clear_inode(struct inode
*);
3201 extern void __destroy_inode(struct inode
*);
3202 extern struct inode
*new_inode_pseudo(struct super_block
*sb
);
3203 extern struct inode
*new_inode(struct super_block
*sb
);
3204 extern void free_inode_nonrcu(struct inode
*inode
);
3205 extern int should_remove_suid(struct dentry
*);
3206 extern int file_remove_privs(struct file
*);
3208 extern void __insert_inode_hash(struct inode
*, unsigned long hashval
);
3209 static inline void insert_inode_hash(struct inode
*inode
)
3211 __insert_inode_hash(inode
, inode
->i_ino
);
3214 extern void __remove_inode_hash(struct inode
*);
3215 static inline void remove_inode_hash(struct inode
*inode
)
3217 if (!inode_unhashed(inode
) && !hlist_fake(&inode
->i_hash
))
3218 __remove_inode_hash(inode
);
3221 extern void inode_sb_list_add(struct inode
*inode
);
3223 extern int sb_set_blocksize(struct super_block
*, int);
3224 extern int sb_min_blocksize(struct super_block
*, int);
3226 extern int generic_file_mmap(struct file
*, struct vm_area_struct
*);
3227 extern int generic_file_readonly_mmap(struct file
*, struct vm_area_struct
*);
3228 extern ssize_t
generic_write_checks(struct kiocb
*, struct iov_iter
*);
3229 extern int generic_write_check_limits(struct file
*file
, loff_t pos
,
3231 extern int generic_file_rw_checks(struct file
*file_in
, struct file
*file_out
);
3232 ssize_t
filemap_read(struct kiocb
*iocb
, struct iov_iter
*to
,
3233 ssize_t already_read
);
3234 extern ssize_t
generic_file_read_iter(struct kiocb
*, struct iov_iter
*);
3235 extern ssize_t
__generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
3236 extern ssize_t
generic_file_write_iter(struct kiocb
*, struct iov_iter
*);
3237 extern ssize_t
generic_file_direct_write(struct kiocb
*, struct iov_iter
*);
3238 extern ssize_t
generic_perform_write(struct file
*, struct iov_iter
*, loff_t
);
3240 ssize_t
vfs_iter_read(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
3242 ssize_t
vfs_iter_write(struct file
*file
, struct iov_iter
*iter
, loff_t
*ppos
,
3244 ssize_t
vfs_iocb_iter_read(struct file
*file
, struct kiocb
*iocb
,
3245 struct iov_iter
*iter
);
3246 ssize_t
vfs_iocb_iter_write(struct file
*file
, struct kiocb
*iocb
,
3247 struct iov_iter
*iter
);
3249 /* fs/block_dev.c */
3250 extern ssize_t
blkdev_read_iter(struct kiocb
*iocb
, struct iov_iter
*to
);
3251 extern ssize_t
blkdev_write_iter(struct kiocb
*iocb
, struct iov_iter
*from
);
3252 extern int blkdev_fsync(struct file
*filp
, loff_t start
, loff_t end
,
3254 extern void block_sync_page(struct page
*page
);
3257 extern ssize_t
generic_file_splice_read(struct file
*, loff_t
*,
3258 struct pipe_inode_info
*, size_t, unsigned int);
3259 extern ssize_t
iter_file_splice_write(struct pipe_inode_info
*,
3260 struct file
*, loff_t
*, size_t, unsigned int);
3261 extern ssize_t
generic_splice_sendpage(struct pipe_inode_info
*pipe
,
3262 struct file
*out
, loff_t
*, size_t len
, unsigned int flags
);
3263 extern long do_splice_direct(struct file
*in
, loff_t
*ppos
, struct file
*out
,
3264 loff_t
*opos
, size_t len
, unsigned int flags
);
3268 file_ra_state_init(struct file_ra_state
*ra
, struct address_space
*mapping
);
3269 extern loff_t
noop_llseek(struct file
*file
, loff_t offset
, int whence
);
3270 extern loff_t
no_llseek(struct file
*file
, loff_t offset
, int whence
);
3271 extern loff_t
vfs_setpos(struct file
*file
, loff_t offset
, loff_t maxsize
);
3272 extern loff_t
generic_file_llseek(struct file
*file
, loff_t offset
, int whence
);
3273 extern loff_t
generic_file_llseek_size(struct file
*file
, loff_t offset
,
3274 int whence
, loff_t maxsize
, loff_t eof
);
3275 extern loff_t
fixed_size_llseek(struct file
*file
, loff_t offset
,
3276 int whence
, loff_t size
);
3277 extern loff_t
no_seek_end_llseek_size(struct file
*, loff_t
, int, loff_t
);
3278 extern loff_t
no_seek_end_llseek(struct file
*, loff_t
, int);
3279 extern int generic_file_open(struct inode
* inode
, struct file
* filp
);
3280 extern int nonseekable_open(struct inode
* inode
, struct file
* filp
);
3281 extern int stream_open(struct inode
* inode
, struct file
* filp
);
3284 typedef void (dio_submit_t
)(struct bio
*bio
, struct inode
*inode
,
3285 loff_t file_offset
);
3288 /* need locking between buffered and direct access */
3291 /* filesystem does not support filling holes */
3292 DIO_SKIP_HOLES
= 0x02,
3295 ssize_t
__blockdev_direct_IO(struct kiocb
*iocb
, struct inode
*inode
,
3296 struct block_device
*bdev
, struct iov_iter
*iter
,
3297 get_block_t get_block
,
3298 dio_iodone_t end_io
, dio_submit_t submit_io
,
3301 static inline ssize_t
blockdev_direct_IO(struct kiocb
*iocb
,
3302 struct inode
*inode
,
3303 struct iov_iter
*iter
,
3304 get_block_t get_block
)
3306 return __blockdev_direct_IO(iocb
, inode
, inode
->i_sb
->s_bdev
, iter
,
3307 get_block
, NULL
, NULL
, DIO_LOCKING
| DIO_SKIP_HOLES
);
3311 void inode_dio_wait(struct inode
*inode
);
3314 * inode_dio_begin - signal start of a direct I/O requests
3315 * @inode: inode the direct I/O happens on
3317 * This is called once we've finished processing a direct I/O request,
3318 * and is used to wake up callers waiting for direct I/O to be quiesced.
3320 static inline void inode_dio_begin(struct inode
*inode
)
3322 atomic_inc(&inode
->i_dio_count
);
3326 * inode_dio_end - signal finish of a direct I/O requests
3327 * @inode: inode the direct I/O happens on
3329 * This is called once we've finished processing a direct I/O request,
3330 * and is used to wake up callers waiting for direct I/O to be quiesced.
3332 static inline void inode_dio_end(struct inode
*inode
)
3334 if (atomic_dec_and_test(&inode
->i_dio_count
))
3335 wake_up_bit(&inode
->i_state
, __I_DIO_WAKEUP
);
3339 * Warn about a page cache invalidation failure diring a direct I/O write.
3341 void dio_warn_stale_pagecache(struct file
*filp
);
3343 extern void inode_set_flags(struct inode
*inode
, unsigned int flags
,
3346 extern const struct file_operations generic_ro_fops
;
3348 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3350 extern int readlink_copy(char __user
*, int, const char *);
3351 extern int page_readlink(struct dentry
*, char __user
*, int);
3352 extern const char *page_get_link(struct dentry
*, struct inode
*,
3353 struct delayed_call
*);
3354 extern void page_put_link(void *);
3355 extern int __page_symlink(struct inode
*inode
, const char *symname
, int len
,
3357 extern int page_symlink(struct inode
*inode
, const char *symname
, int len
);
3358 extern const struct inode_operations page_symlink_inode_operations
;
3359 extern void kfree_link(void *);
3360 void generic_fillattr(struct user_namespace
*, struct inode
*, struct kstat
*);
3361 extern int vfs_getattr_nosec(const struct path
*, struct kstat
*, u32
, unsigned int);
3362 extern int vfs_getattr(const struct path
*, struct kstat
*, u32
, unsigned int);
3363 void __inode_add_bytes(struct inode
*inode
, loff_t bytes
);
3364 void inode_add_bytes(struct inode
*inode
, loff_t bytes
);
3365 void __inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
3366 void inode_sub_bytes(struct inode
*inode
, loff_t bytes
);
3367 static inline loff_t
__inode_get_bytes(struct inode
*inode
)
3369 return (((loff_t
)inode
->i_blocks
) << 9) + inode
->i_bytes
;
3371 loff_t
inode_get_bytes(struct inode
*inode
);
3372 void inode_set_bytes(struct inode
*inode
, loff_t bytes
);
3373 const char *simple_get_link(struct dentry
*, struct inode
*,
3374 struct delayed_call
*);
3375 extern const struct inode_operations simple_symlink_inode_operations
;
3377 extern int iterate_dir(struct file
*, struct dir_context
*);
3379 int vfs_fstatat(int dfd
, const char __user
*filename
, struct kstat
*stat
,
3381 int vfs_fstat(int fd
, struct kstat
*stat
);
3383 static inline int vfs_stat(const char __user
*filename
, struct kstat
*stat
)
3385 return vfs_fstatat(AT_FDCWD
, filename
, stat
, 0);
3387 static inline int vfs_lstat(const char __user
*name
, struct kstat
*stat
)
3389 return vfs_fstatat(AT_FDCWD
, name
, stat
, AT_SYMLINK_NOFOLLOW
);
3392 extern const char *vfs_get_link(struct dentry
*, struct delayed_call
*);
3393 extern int vfs_readlink(struct dentry
*, char __user
*, int);
3395 extern struct file_system_type
*get_filesystem(struct file_system_type
*fs
);
3396 extern void put_filesystem(struct file_system_type
*fs
);
3397 extern struct file_system_type
*get_fs_type(const char *name
);
3398 extern struct super_block
*get_super(struct block_device
*);
3399 extern struct super_block
*get_active_super(struct block_device
*bdev
);
3400 extern void drop_super(struct super_block
*sb
);
3401 extern void drop_super_exclusive(struct super_block
*sb
);
3402 extern void iterate_supers(void (*)(struct super_block
*, void *), void *);
3403 extern void iterate_supers_type(struct file_system_type
*,
3404 void (*)(struct super_block
*, void *), void *);
3406 extern int dcache_dir_open(struct inode
*, struct file
*);
3407 extern int dcache_dir_close(struct inode
*, struct file
*);
3408 extern loff_t
dcache_dir_lseek(struct file
*, loff_t
, int);
3409 extern int dcache_readdir(struct file
*, struct dir_context
*);
3410 extern int simple_setattr(struct user_namespace
*, struct dentry
*,
3412 extern int simple_getattr(struct user_namespace
*, const struct path
*,
3413 struct kstat
*, u32
, unsigned int);
3414 extern int simple_statfs(struct dentry
*, struct kstatfs
*);
3415 extern int simple_open(struct inode
*inode
, struct file
*file
);
3416 extern int simple_link(struct dentry
*, struct inode
*, struct dentry
*);
3417 extern int simple_unlink(struct inode
*, struct dentry
*);
3418 extern int simple_rmdir(struct inode
*, struct dentry
*);
3419 extern int simple_rename(struct user_namespace
*, struct inode
*,
3420 struct dentry
*, struct inode
*, struct dentry
*,
3422 extern void simple_recursive_removal(struct dentry
*,
3423 void (*callback
)(struct dentry
*));
3424 extern int noop_fsync(struct file
*, loff_t
, loff_t
, int);
3425 extern void noop_invalidatepage(struct page
*page
, unsigned int offset
,
3426 unsigned int length
);
3427 extern ssize_t
noop_direct_IO(struct kiocb
*iocb
, struct iov_iter
*iter
);
3428 extern int simple_empty(struct dentry
*);
3429 extern int simple_write_begin(struct file
*file
, struct address_space
*mapping
,
3430 loff_t pos
, unsigned len
, unsigned flags
,
3431 struct page
**pagep
, void **fsdata
);
3432 extern const struct address_space_operations ram_aops
;
3433 extern int always_delete_dentry(const struct dentry
*);
3434 extern struct inode
*alloc_anon_inode(struct super_block
*);
3435 extern int simple_nosetlease(struct file
*, long, struct file_lock
**, void **);
3436 extern const struct dentry_operations simple_dentry_operations
;
3438 extern struct dentry
*simple_lookup(struct inode
*, struct dentry
*, unsigned int flags
);
3439 extern ssize_t
generic_read_dir(struct file
*, char __user
*, size_t, loff_t
*);
3440 extern const struct file_operations simple_dir_operations
;
3441 extern const struct inode_operations simple_dir_inode_operations
;
3442 extern void make_empty_dir_inode(struct inode
*inode
);
3443 extern bool is_empty_dir_inode(struct inode
*inode
);
3444 struct tree_descr
{ const char *name
; const struct file_operations
*ops
; int mode
; };
3445 struct dentry
*d_alloc_name(struct dentry
*, const char *);
3446 extern int simple_fill_super(struct super_block
*, unsigned long,
3447 const struct tree_descr
*);
3448 extern int simple_pin_fs(struct file_system_type
*, struct vfsmount
**mount
, int *count
);
3449 extern void simple_release_fs(struct vfsmount
**mount
, int *count
);
3451 extern ssize_t
simple_read_from_buffer(void __user
*to
, size_t count
,
3452 loff_t
*ppos
, const void *from
, size_t available
);
3453 extern ssize_t
simple_write_to_buffer(void *to
, size_t available
, loff_t
*ppos
,
3454 const void __user
*from
, size_t count
);
3456 extern int __generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3457 extern int generic_file_fsync(struct file
*, loff_t
, loff_t
, int);
3459 extern int generic_check_addressable(unsigned, u64
);
3461 extern void generic_set_encrypted_ci_d_ops(struct dentry
*dentry
);
3463 #ifdef CONFIG_MIGRATION
3464 extern int buffer_migrate_page(struct address_space
*,
3465 struct page
*, struct page
*,
3467 extern int buffer_migrate_page_norefs(struct address_space
*,
3468 struct page
*, struct page
*,
3471 #define buffer_migrate_page NULL
3472 #define buffer_migrate_page_norefs NULL
3475 int setattr_prepare(struct user_namespace
*, struct dentry
*, struct iattr
*);
3476 extern int inode_newsize_ok(const struct inode
*, loff_t offset
);
3477 void setattr_copy(struct user_namespace
*, struct inode
*inode
,
3478 const struct iattr
*attr
);
3480 extern int file_update_time(struct file
*file
);
3482 static inline bool vma_is_dax(const struct vm_area_struct
*vma
)
3484 return vma
->vm_file
&& IS_DAX(vma
->vm_file
->f_mapping
->host
);
3487 static inline bool vma_is_fsdax(struct vm_area_struct
*vma
)
3489 struct inode
*inode
;
3491 if (!IS_ENABLED(CONFIG_FS_DAX
) || !vma
->vm_file
)
3493 if (!vma_is_dax(vma
))
3495 inode
= file_inode(vma
->vm_file
);
3496 if (S_ISCHR(inode
->i_mode
))
3497 return false; /* device-dax */
3501 static inline int iocb_flags(struct file
*file
)
3504 if (file
->f_flags
& O_APPEND
)
3506 if (file
->f_flags
& O_DIRECT
)
3508 if ((file
->f_flags
& O_DSYNC
) || IS_SYNC(file
->f_mapping
->host
))
3510 if (file
->f_flags
& __O_SYNC
)
3515 static inline int kiocb_set_rw_flags(struct kiocb
*ki
, rwf_t flags
)
3517 int kiocb_flags
= 0;
3519 /* make sure there's no overlap between RWF and private IOCB flags */
3520 BUILD_BUG_ON((__force
int) RWF_SUPPORTED
& IOCB_EVENTFD
);
3524 if (unlikely(flags
& ~RWF_SUPPORTED
))
3527 if (flags
& RWF_NOWAIT
) {
3528 if (!(ki
->ki_filp
->f_mode
& FMODE_NOWAIT
))
3530 kiocb_flags
|= IOCB_NOIO
;
3532 kiocb_flags
|= (__force
int) (flags
& RWF_SUPPORTED
);
3533 if (flags
& RWF_SYNC
)
3534 kiocb_flags
|= IOCB_DSYNC
;
3536 ki
->ki_flags
|= kiocb_flags
;
3540 static inline ino_t
parent_ino(struct dentry
*dentry
)
3545 * Don't strictly need d_lock here? If the parent ino could change
3546 * then surely we'd have a deeper race in the caller?
3548 spin_lock(&dentry
->d_lock
);
3549 res
= dentry
->d_parent
->d_inode
->i_ino
;
3550 spin_unlock(&dentry
->d_lock
);
3554 /* Transaction based IO helpers */
3557 * An argresp is stored in an allocated page and holds the
3558 * size of the argument or response, along with its content
3560 struct simple_transaction_argresp
{
3565 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3567 char *simple_transaction_get(struct file
*file
, const char __user
*buf
,
3569 ssize_t
simple_transaction_read(struct file
*file
, char __user
*buf
,
3570 size_t size
, loff_t
*pos
);
3571 int simple_transaction_release(struct inode
*inode
, struct file
*file
);
3573 void simple_transaction_set(struct file
*file
, size_t n
);
3576 * simple attribute files
3578 * These attributes behave similar to those in sysfs:
3580 * Writing to an attribute immediately sets a value, an open file can be
3581 * written to multiple times.
3583 * Reading from an attribute creates a buffer from the value that might get
3584 * read with multiple read calls. When the attribute has been read
3585 * completely, no further read calls are possible until the file is opened
3588 * All attributes contain a text representation of a numeric value
3589 * that are accessed with the get() and set() functions.
3591 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3592 static int __fops ## _open(struct inode *inode, struct file *file) \
3594 __simple_attr_check_format(__fmt, 0ull); \
3595 return simple_attr_open(inode, file, __get, __set, __fmt); \
3597 static const struct file_operations __fops = { \
3598 .owner = THIS_MODULE, \
3599 .open = __fops ## _open, \
3600 .release = simple_attr_release, \
3601 .read = simple_attr_read, \
3602 .write = simple_attr_write, \
3603 .llseek = generic_file_llseek, \
3606 static inline __printf(1, 2)
3607 void __simple_attr_check_format(const char *fmt
, ...)
3609 /* don't do anything, just let the compiler check the arguments; */
3612 int simple_attr_open(struct inode
*inode
, struct file
*file
,
3613 int (*get
)(void *, u64
*), int (*set
)(void *, u64
),
3615 int simple_attr_release(struct inode
*inode
, struct file
*file
);
3616 ssize_t
simple_attr_read(struct file
*file
, char __user
*buf
,
3617 size_t len
, loff_t
*ppos
);
3618 ssize_t
simple_attr_write(struct file
*file
, const char __user
*buf
,
3619 size_t len
, loff_t
*ppos
);
3622 int proc_nr_files(struct ctl_table
*table
, int write
,
3623 void *buffer
, size_t *lenp
, loff_t
*ppos
);
3624 int proc_nr_dentry(struct ctl_table
*table
, int write
,
3625 void *buffer
, size_t *lenp
, loff_t
*ppos
);
3626 int proc_nr_inodes(struct ctl_table
*table
, int write
,
3627 void *buffer
, size_t *lenp
, loff_t
*ppos
);
3628 int __init
get_filesystem_list(char *buf
);
3630 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3631 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3633 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3634 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3635 (flag & __FMODE_NONOTIFY)))
3637 static inline bool is_sxid(umode_t mode
)
3639 return (mode
& S_ISUID
) || ((mode
& S_ISGID
) && (mode
& S_IXGRP
));
3642 static inline int check_sticky(struct user_namespace
*mnt_userns
,
3643 struct inode
*dir
, struct inode
*inode
)
3645 if (!(dir
->i_mode
& S_ISVTX
))
3648 return __check_sticky(mnt_userns
, dir
, inode
);
3651 static inline void inode_has_no_xattr(struct inode
*inode
)
3653 if (!is_sxid(inode
->i_mode
) && (inode
->i_sb
->s_flags
& SB_NOSEC
))
3654 inode
->i_flags
|= S_NOSEC
;
3657 static inline bool is_root_inode(struct inode
*inode
)
3659 return inode
== inode
->i_sb
->s_root
->d_inode
;
3662 static inline bool dir_emit(struct dir_context
*ctx
,
3663 const char *name
, int namelen
,
3664 u64 ino
, unsigned type
)
3666 return ctx
->actor(ctx
, name
, namelen
, ctx
->pos
, ino
, type
) == 0;
3668 static inline bool dir_emit_dot(struct file
*file
, struct dir_context
*ctx
)
3670 return ctx
->actor(ctx
, ".", 1, ctx
->pos
,
3671 file
->f_path
.dentry
->d_inode
->i_ino
, DT_DIR
) == 0;
3673 static inline bool dir_emit_dotdot(struct file
*file
, struct dir_context
*ctx
)
3675 return ctx
->actor(ctx
, "..", 2, ctx
->pos
,
3676 parent_ino(file
->f_path
.dentry
), DT_DIR
) == 0;
3678 static inline bool dir_emit_dots(struct file
*file
, struct dir_context
*ctx
)
3680 if (ctx
->pos
== 0) {
3681 if (!dir_emit_dot(file
, ctx
))
3685 if (ctx
->pos
== 1) {
3686 if (!dir_emit_dotdot(file
, ctx
))
3692 static inline bool dir_relax(struct inode
*inode
)
3694 inode_unlock(inode
);
3696 return !IS_DEADDIR(inode
);
3699 static inline bool dir_relax_shared(struct inode
*inode
)
3701 inode_unlock_shared(inode
);
3702 inode_lock_shared(inode
);
3703 return !IS_DEADDIR(inode
);
3706 extern bool path_noexec(const struct path
*path
);
3707 extern void inode_nohighmem(struct inode
*inode
);
3710 extern int vfs_fadvise(struct file
*file
, loff_t offset
, loff_t len
,
3712 extern int generic_fadvise(struct file
*file
, loff_t offset
, loff_t len
,
3716 * Flush file data before changing attributes. Caller must hold any locks
3717 * required to prevent further writes to this file until we're done setting
3720 static inline int inode_drain_writes(struct inode
*inode
)
3722 inode_dio_wait(inode
);
3723 return filemap_write_and_wait(inode
->i_mapping
);
3726 #endif /* _LINUX_FS_H */