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/fiemap.h>
28 #include <linux/rculist_bl.h>
29 #include <linux/atomic.h>
30 #include <linux/shrinker.h>
31 #include <linux/migrate_mode.h>
32 #include <linux/uidgid.h>
33 #include <linux/lockdep.h>
34 #include <linux/percpu-rwsem.h>
35 #include <linux/workqueue.h>
36 #include <linux/delayed_call.h>
37 #include <linux/uuid.h>
38 #include <linux/errseq.h>
39 #include <linux/ioprio.h>
40 #include <linux/fs_types.h>
41 #include <linux/build_bug.h>
42 #include <linux/stddef.h>
44 #include <asm/byteorder.h>
45 #include <uapi/linux/fs.h>
47 struct backing_dev_info;
50 struct export_operations;
55 struct pipe_inode_info;
56 struct poll_table_struct;
58 struct vm_area_struct;
61 struct swap_info_struct;
63 struct workqueue_struct;
66 struct fscrypt_operations;
68 struct fs_parameter_description;
70 extern void __init inode_init(void);
71 extern void __init inode_init_early(void);
72 extern void __init files_init(void);
73 extern void __init files_maxfiles_init(void);
75 extern struct files_stat_struct files_stat;
76 extern unsigned long get_max_files(void);
77 extern unsigned int sysctl_nr_open;
78 extern struct inodes_stat_t inodes_stat;
79 extern int leases_enable, lease_break_time;
80 extern int sysctl_protected_symlinks;
81 extern int sysctl_protected_hardlinks;
82 extern int sysctl_protected_fifos;
83 extern int sysctl_protected_regular;
85 typedef __kernel_rwf_t rwf_t;
88 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
89 struct buffer_head *bh_result, int create);
90 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
91 ssize_t bytes, void *private);
93 #define MAY_EXEC 0x00000001
94 #define MAY_WRITE 0x00000002
95 #define MAY_READ 0x00000004
96 #define MAY_APPEND 0x00000008
97 #define MAY_ACCESS 0x00000010
98 #define MAY_OPEN 0x00000020
99 #define MAY_CHDIR 0x00000040
100 /* called from RCU mode, don't block */
101 #define MAY_NOT_BLOCK 0x00000080
104 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
105 * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
108 /* file is open for reading */
109 #define FMODE_READ ((__force fmode_t)0x1)
110 /* file is open for writing */
111 #define FMODE_WRITE ((__force fmode_t)0x2)
112 /* file is seekable */
113 #define FMODE_LSEEK ((__force fmode_t)0x4)
114 /* file can be accessed using pread */
115 #define FMODE_PREAD ((__force fmode_t)0x8)
116 /* file can be accessed using pwrite */
117 #define FMODE_PWRITE ((__force fmode_t)0x10)
118 /* File is opened for execution with sys_execve / sys_uselib */
119 #define FMODE_EXEC ((__force fmode_t)0x20)
120 /* File is opened with O_NDELAY (only set for block devices) */
121 #define FMODE_NDELAY ((__force fmode_t)0x40)
122 /* File is opened with O_EXCL (only set for block devices) */
123 #define FMODE_EXCL ((__force fmode_t)0x80)
124 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
125 (specialy hack for floppy.c) */
126 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
127 /* 32bit hashes as llseek() offset (for directories) */
128 #define FMODE_32BITHASH ((__force fmode_t)0x200)
129 /* 64bit hashes as llseek() offset (for directories) */
130 #define FMODE_64BITHASH ((__force fmode_t)0x400)
133 * Don't update ctime and mtime.
135 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
136 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
138 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
140 /* Expect random access pattern */
141 #define FMODE_RANDOM ((__force fmode_t)0x1000)
143 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
144 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
146 /* File is opened with O_PATH; almost nothing can be done with it */
147 #define FMODE_PATH ((__force fmode_t)0x4000)
149 /* File needs atomic accesses to f_pos */
150 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
151 /* Write access to underlying fs */
152 #define FMODE_WRITER ((__force fmode_t)0x10000)
153 /* Has read method(s) */
154 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
155 /* Has write method(s) */
156 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
158 #define FMODE_OPENED ((__force fmode_t)0x80000)
159 #define FMODE_CREATED ((__force fmode_t)0x100000)
161 /* File is stream-like */
162 #define FMODE_STREAM ((__force fmode_t)0x200000)
164 /* File was opened by fanotify and shouldn't generate fanotify events */
165 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
167 /* File is capable of returning -EAGAIN if I/O will block */
168 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
170 /* File does not contribute to nr_files count */
171 #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000)
174 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
175 * that indicates that they should check the contents of the iovec are
176 * valid, but not check the memory that the iovec elements
179 #define CHECK_IOVEC_ONLY -1
182 * Attribute flags. These should be or-ed together to figure out what
185 #define ATTR_MODE (1 << 0)
186 #define ATTR_UID (1 << 1)
187 #define ATTR_GID (1 << 2)
188 #define ATTR_SIZE (1 << 3)
189 #define ATTR_ATIME (1 << 4)
190 #define ATTR_MTIME (1 << 5)
191 #define ATTR_CTIME (1 << 6)
192 #define ATTR_ATIME_SET (1 << 7)
193 #define ATTR_MTIME_SET (1 << 8)
194 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
195 #define ATTR_KILL_SUID (1 << 11)
196 #define ATTR_KILL_SGID (1 << 12)
197 #define ATTR_FILE (1 << 13)
198 #define ATTR_KILL_PRIV (1 << 14)
199 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
200 #define ATTR_TIMES_SET (1 << 16)
201 #define ATTR_TOUCH (1 << 17)
204 * Whiteout is represented by a char device. The following constants define the
205 * mode and device number to use.
207 #define WHITEOUT_MODE 0
208 #define WHITEOUT_DEV 0
211 * This is the Inode Attributes structure, used for notify_change(). It
212 * uses the above definitions as flags, to know which values have changed.
213 * Also, in this manner, a Filesystem can look at only the values it cares
214 * about. Basically, these are the attributes that the VFS layer can
215 * request to change from the FS layer.
217 * Derek Atkins <warlord@MIT.EDU> 94-10-20
220 unsigned int ia_valid;
225 struct timespec64 ia_atime;
226 struct timespec64 ia_mtime;
227 struct timespec64 ia_ctime;
230 * Not an attribute, but an auxiliary info for filesystems wanting to
231 * implement an ftruncate() like method. NOTE: filesystem should
232 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
234 struct file *ia_file;
238 * Includes for diskquotas.
240 #include <linux/quota.h>
243 * Maximum number of layers of fs stack. Needs to be limited to
244 * prevent kernel stack overflow
246 #define FILESYSTEM_MAX_STACK_DEPTH 2
249 * enum positive_aop_returns - aop return codes with specific semantics
251 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
252 * completed, that the page is still locked, and
253 * should be considered active. The VM uses this hint
254 * to return the page to the active list -- it won't
255 * be a candidate for writeback again in the near
256 * future. Other callers must be careful to unlock
257 * the page if they get this return. Returned by
260 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
261 * unlocked it and the page might have been truncated.
262 * The caller should back up to acquiring a new page and
263 * trying again. The aop will be taking reasonable
264 * precautions not to livelock. If the caller held a page
265 * reference, it should drop it before retrying. Returned
268 * address_space_operation functions return these large constants to indicate
269 * special semantics to the caller. These are much larger than the bytes in a
270 * page to allow for functions that return the number of bytes operated on in a
274 enum positive_aop_returns {
275 AOP_WRITEPAGE_ACTIVATE = 0x80000,
276 AOP_TRUNCATED_PAGE = 0x80001,
279 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
280 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
281 * helper code (eg buffer layer)
282 * to clear GFP_FS from alloc */
285 * oh the beauties of C type declarations.
288 struct address_space;
289 struct writeback_control;
292 * Write life time hint values.
293 * Stored in struct inode as u8.
296 WRITE_LIFE_NOT_SET = 0,
297 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
298 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
299 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
300 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
301 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
304 #define IOCB_EVENTFD (1 << 0)
305 #define IOCB_APPEND (1 << 1)
306 #define IOCB_DIRECT (1 << 2)
307 #define IOCB_HIPRI (1 << 3)
308 #define IOCB_DSYNC (1 << 4)
309 #define IOCB_SYNC (1 << 5)
310 #define IOCB_WRITE (1 << 6)
311 #define IOCB_NOWAIT (1 << 7)
314 struct file *ki_filp;
316 /* The 'ki_filp' pointer is shared in a union for aio */
317 randomized_struct_fields_start
320 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
324 u16 ki_ioprio; /* See linux/ioprio.h */
325 unsigned int ki_cookie; /* for ->iopoll */
327 randomized_struct_fields_end
330 static inline bool is_sync_kiocb(struct kiocb *kiocb)
332 return kiocb->ki_complete == NULL;
336 * "descriptor" for what we're up to with a read.
337 * This allows us to use the same read code yet
338 * have multiple different users of the data that
339 * we read from a file.
341 * The simplest case just copies the data to user
354 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
355 unsigned long, unsigned long);
357 struct address_space_operations {
358 int (*writepage)(struct page *page, struct writeback_control *wbc);
359 int (*readpage)(struct file *, struct page *);
361 /* Write back some dirty pages from this mapping. */
362 int (*writepages)(struct address_space *, struct writeback_control *);
364 /* Set a page dirty. Return true if this dirtied it */
365 int (*set_page_dirty)(struct page *page);
368 * Reads in the requested pages. Unlike ->readpage(), this is
369 * PURELY used for read-ahead!.
371 int (*readpages)(struct file *filp, struct address_space *mapping,
372 struct list_head *pages, unsigned nr_pages);
374 int (*write_begin)(struct file *, struct address_space *mapping,
375 loff_t pos, unsigned len, unsigned flags,
376 struct page **pagep, void **fsdata);
377 int (*write_end)(struct file *, struct address_space *mapping,
378 loff_t pos, unsigned len, unsigned copied,
379 struct page *page, void *fsdata);
381 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
382 sector_t (*bmap)(struct address_space *, sector_t);
383 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
384 int (*releasepage) (struct page *, gfp_t);
385 void (*freepage)(struct page *);
386 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
388 * migrate the contents of a page to the specified target. If
389 * migrate_mode is MIGRATE_ASYNC, it must not block.
391 int (*migratepage) (struct address_space *,
392 struct page *, struct page *, enum migrate_mode);
393 bool (*isolate_page)(struct page *, isolate_mode_t);
394 void (*putback_page)(struct page *);
395 int (*launder_page) (struct page *);
396 int (*is_partially_uptodate) (struct page *, unsigned long,
398 void (*is_dirty_writeback) (struct page *, bool *, bool *);
399 int (*error_remove_page)(struct address_space *, struct page *);
401 /* swapfile support */
402 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
404 void (*swap_deactivate)(struct file *file);
407 extern const struct address_space_operations empty_aops;
410 * pagecache_write_begin/pagecache_write_end must be used by general code
411 * to write into the pagecache.
413 int pagecache_write_begin(struct file *, struct address_space *mapping,
414 loff_t pos, unsigned len, unsigned flags,
415 struct page **pagep, void **fsdata);
417 int pagecache_write_end(struct file *, struct address_space *mapping,
418 loff_t pos, unsigned len, unsigned copied,
419 struct page *page, void *fsdata);
422 * struct address_space - Contents of a cacheable, mappable object.
423 * @host: Owner, either the inode or the block_device.
424 * @i_pages: Cached pages.
425 * @gfp_mask: Memory allocation flags to use for allocating pages.
426 * @i_mmap_writable: Number of VM_SHARED mappings.
427 * @i_mmap: Tree of private and shared mappings.
428 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
429 * @nrpages: Number of page entries, protected by the i_pages lock.
430 * @nrexceptional: Shadow or DAX entries, protected by the i_pages lock.
431 * @writeback_index: Writeback starts here.
433 * @flags: Error bits and flags (AS_*).
434 * @wb_err: The most recent error which has occurred.
435 * @private_lock: For use by the owner of the address_space.
436 * @private_list: For use by the owner of the address_space.
437 * @private_data: For use by the owner of the address_space.
439 struct address_space {
441 struct xarray i_pages;
443 atomic_t i_mmap_writable;
444 struct rb_root_cached i_mmap;
445 struct rw_semaphore i_mmap_rwsem;
446 unsigned long nrpages;
447 unsigned long nrexceptional;
448 pgoff_t writeback_index;
449 const struct address_space_operations *a_ops;
452 spinlock_t private_lock;
453 struct list_head private_list;
455 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
457 * On most architectures that alignment is already the case; but
458 * must be enforced here for CRIS, to let the least significant bit
459 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
461 struct request_queue;
463 struct block_device {
464 dev_t bd_dev; /* not a kdev_t - it's a search key */
466 struct inode * bd_inode; /* will die */
467 struct super_block * bd_super;
468 struct mutex bd_mutex; /* open/close mutex */
472 bool bd_write_holder;
474 struct list_head bd_holder_disks;
476 struct block_device * bd_contains;
477 unsigned bd_block_size;
479 struct hd_struct * bd_part;
480 /* number of times partitions within this device have been opened. */
481 unsigned bd_part_count;
483 struct gendisk * bd_disk;
484 struct request_queue * bd_queue;
485 struct backing_dev_info *bd_bdi;
486 struct list_head bd_list;
488 * Private data. You must have bd_claim'ed the block_device
489 * to use this. NOTE: bd_claim allows an owner to claim
490 * the same device multiple times, the owner must take special
491 * care to not mess up bd_private for that case.
493 unsigned long bd_private;
495 /* The counter of freeze processes */
496 int bd_fsfreeze_count;
497 /* Mutex for freeze */
498 struct mutex bd_fsfreeze_mutex;
499 } __randomize_layout;
501 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
502 #define PAGECACHE_TAG_DIRTY XA_MARK_0
503 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
504 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
507 * Returns true if any of the pages in the mapping are marked with the tag.
509 static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag)
511 return xa_marked(&mapping->i_pages, tag);
514 static inline void i_mmap_lock_write(struct address_space *mapping)
516 down_write(&mapping->i_mmap_rwsem);
519 static inline void i_mmap_unlock_write(struct address_space *mapping)
521 up_write(&mapping->i_mmap_rwsem);
524 static inline void i_mmap_lock_read(struct address_space *mapping)
526 down_read(&mapping->i_mmap_rwsem);
529 static inline void i_mmap_unlock_read(struct address_space *mapping)
531 up_read(&mapping->i_mmap_rwsem);
535 * Might pages of this file be mapped into userspace?
537 static inline int mapping_mapped(struct address_space *mapping)
539 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
543 * Might pages of this file have been modified in userspace?
544 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
545 * marks vma as VM_SHARED if it is shared, and the file was opened for
546 * writing i.e. vma may be mprotected writable even if now readonly.
548 * If i_mmap_writable is negative, no new writable mappings are allowed. You
549 * can only deny writable mappings, if none exists right now.
551 static inline int mapping_writably_mapped(struct address_space *mapping)
553 return atomic_read(&mapping->i_mmap_writable) > 0;
556 static inline int mapping_map_writable(struct address_space *mapping)
558 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
562 static inline void mapping_unmap_writable(struct address_space *mapping)
564 atomic_dec(&mapping->i_mmap_writable);
567 static inline int mapping_deny_writable(struct address_space *mapping)
569 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
573 static inline void mapping_allow_writable(struct address_space *mapping)
575 atomic_inc(&mapping->i_mmap_writable);
579 * Use sequence counter to get consistent i_size on 32-bit processors.
581 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
582 #include <linux/seqlock.h>
583 #define __NEED_I_SIZE_ORDERED
584 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
586 #define i_size_ordered_init(inode) do { } while (0)
590 #define ACL_NOT_CACHED ((void *)(-1))
591 #define ACL_DONT_CACHE ((void *)(-3))
593 static inline struct posix_acl *
594 uncached_acl_sentinel(struct task_struct *task)
596 return (void *)task + 1;
600 is_uncached_acl(struct posix_acl *acl)
602 return (long)acl & 1;
605 #define IOP_FASTPERM 0x0001
606 #define IOP_LOOKUP 0x0002
607 #define IOP_NOFOLLOW 0x0004
608 #define IOP_XATTR 0x0008
609 #define IOP_DEFAULT_READLINK 0x0010
611 struct fsnotify_mark_connector;
614 * Keep mostly read-only and often accessed (especially for
615 * the RCU path lookup and 'stat' data) fields at the beginning
616 * of the 'struct inode'
620 unsigned short i_opflags;
623 unsigned int i_flags;
625 #ifdef CONFIG_FS_POSIX_ACL
626 struct posix_acl *i_acl;
627 struct posix_acl *i_default_acl;
630 const struct inode_operations *i_op;
631 struct super_block *i_sb;
632 struct address_space *i_mapping;
634 #ifdef CONFIG_SECURITY
638 /* Stat data, not accessed from path walking */
641 * Filesystems may only read i_nlink directly. They shall use the
642 * following functions for modification:
644 * (set|clear|inc|drop)_nlink
645 * inode_(inc|dec)_link_count
648 const unsigned int i_nlink;
649 unsigned int __i_nlink;
653 struct timespec64 i_atime;
654 struct timespec64 i_mtime;
655 struct timespec64 i_ctime;
656 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
657 unsigned short i_bytes;
662 #ifdef __NEED_I_SIZE_ORDERED
663 seqcount_t i_size_seqcount;
667 unsigned long i_state;
668 struct rw_semaphore i_rwsem;
670 unsigned long dirtied_when; /* jiffies of first dirtying */
671 unsigned long dirtied_time_when;
673 struct hlist_node i_hash;
674 struct list_head i_io_list; /* backing dev IO list */
675 #ifdef CONFIG_CGROUP_WRITEBACK
676 struct bdi_writeback *i_wb; /* the associated cgroup wb */
678 /* foreign inode detection, see wbc_detach_inode() */
680 u16 i_wb_frn_avg_time;
681 u16 i_wb_frn_history;
683 struct list_head i_lru; /* inode LRU list */
684 struct list_head i_sb_list;
685 struct list_head i_wb_list; /* backing dev writeback list */
687 struct hlist_head i_dentry;
688 struct rcu_head i_rcu;
690 atomic64_t i_version;
692 atomic_t i_dio_count;
693 atomic_t i_writecount;
695 atomic_t i_readcount; /* struct files open RO */
697 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
698 struct file_lock_context *i_flctx;
699 struct address_space i_data;
700 struct list_head i_devices;
702 struct pipe_inode_info *i_pipe;
703 struct block_device *i_bdev;
711 #ifdef CONFIG_FSNOTIFY
712 __u32 i_fsnotify_mask; /* all events this inode cares about */
713 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
716 #ifdef CONFIG_FS_ENCRYPTION
717 struct fscrypt_info *i_crypt_info;
720 void *i_private; /* fs or device private pointer */
721 } __randomize_layout;
723 static inline unsigned int i_blocksize(const struct inode *node)
725 return (1 << node->i_blkbits);
728 static inline int inode_unhashed(struct inode *inode)
730 return hlist_unhashed(&inode->i_hash);
734 * __mark_inode_dirty expects inodes to be hashed. Since we don't
735 * want special inodes in the fileset inode space, we make them
736 * appear hashed, but do not put on any lists. hlist_del()
737 * will work fine and require no locking.
739 static inline void inode_fake_hash(struct inode *inode)
741 hlist_add_fake(&inode->i_hash);
745 * inode->i_mutex nesting subclasses for the lock validator:
747 * 0: the object of the current VFS operation
751 * 4: second non-directory
752 * 5: second parent (when locking independent directories in rename)
754 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
755 * non-directories at once.
757 * The locking order between these classes is
758 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
760 enum inode_i_mutex_lock_class
770 static inline void inode_lock(struct inode *inode)
772 down_write(&inode->i_rwsem);
775 static inline void inode_unlock(struct inode *inode)
777 up_write(&inode->i_rwsem);
780 static inline void inode_lock_shared(struct inode *inode)
782 down_read(&inode->i_rwsem);
785 static inline void inode_unlock_shared(struct inode *inode)
787 up_read(&inode->i_rwsem);
790 static inline int inode_trylock(struct inode *inode)
792 return down_write_trylock(&inode->i_rwsem);
795 static inline int inode_trylock_shared(struct inode *inode)
797 return down_read_trylock(&inode->i_rwsem);
800 static inline int inode_is_locked(struct inode *inode)
802 return rwsem_is_locked(&inode->i_rwsem);
805 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
807 down_write_nested(&inode->i_rwsem, subclass);
810 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
812 down_read_nested(&inode->i_rwsem, subclass);
815 void lock_two_nondirectories(struct inode *, struct inode*);
816 void unlock_two_nondirectories(struct inode *, struct inode*);
819 * NOTE: in a 32bit arch with a preemptable kernel and
820 * an UP compile the i_size_read/write must be atomic
821 * with respect to the local cpu (unlike with preempt disabled),
822 * but they don't need to be atomic with respect to other cpus like in
823 * true SMP (so they need either to either locally disable irq around
824 * the read or for example on x86 they can be still implemented as a
825 * cmpxchg8b without the need of the lock prefix). For SMP compiles
826 * and 64bit archs it makes no difference if preempt is enabled or not.
828 static inline loff_t i_size_read(const struct inode *inode)
830 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
835 seq = read_seqcount_begin(&inode->i_size_seqcount);
836 i_size = inode->i_size;
837 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
839 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
843 i_size = inode->i_size;
847 return inode->i_size;
852 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
853 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
854 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
856 static inline void i_size_write(struct inode *inode, loff_t i_size)
858 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
860 write_seqcount_begin(&inode->i_size_seqcount);
861 inode->i_size = i_size;
862 write_seqcount_end(&inode->i_size_seqcount);
864 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
866 inode->i_size = i_size;
869 inode->i_size = i_size;
873 static inline unsigned iminor(const struct inode *inode)
875 return MINOR(inode->i_rdev);
878 static inline unsigned imajor(const struct inode *inode)
880 return MAJOR(inode->i_rdev);
883 extern struct block_device *I_BDEV(struct inode *inode);
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 * Track a single file's readahead state
896 struct file_ra_state {
897 pgoff_t start; /* where readahead started */
898 unsigned int size; /* # of readahead pages */
899 unsigned int async_size; /* do asynchronous readahead when
900 there are only # of pages ahead */
902 unsigned int ra_pages; /* Maximum readahead window */
903 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
904 loff_t prev_pos; /* Cache last read() position */
908 * Check if @index falls in the readahead windows.
910 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
912 return (index >= ra->start &&
913 index < ra->start + ra->size);
918 struct llist_node fu_llist;
919 struct rcu_head fu_rcuhead;
922 struct inode *f_inode; /* cached value */
923 const struct file_operations *f_op;
926 * Protects f_ep_links, f_flags.
927 * Must not be taken from IRQ context.
930 enum rw_hint f_write_hint;
931 atomic_long_t f_count;
932 unsigned int f_flags;
934 struct mutex f_pos_lock;
936 struct fown_struct f_owner;
937 const struct cred *f_cred;
938 struct file_ra_state f_ra;
941 #ifdef CONFIG_SECURITY
944 /* needed for tty driver, and maybe others */
948 /* Used by fs/eventpoll.c to link all the hooks to this file */
949 struct list_head f_ep_links;
950 struct list_head f_tfile_llink;
951 #endif /* #ifdef CONFIG_EPOLL */
952 struct address_space *f_mapping;
955 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
960 /* file identifier */
961 unsigned char f_handle[0];
964 static inline struct file *get_file(struct file *f)
966 atomic_long_inc(&f->f_count);
969 #define get_file_rcu_many(x, cnt) \
970 atomic_long_add_unless(&(x)->f_count, (cnt), 0)
971 #define get_file_rcu(x) get_file_rcu_many((x), 1)
972 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
973 #define file_count(x) atomic_long_read(&(x)->f_count)
975 #define MAX_NON_LFS ((1UL<<31) - 1)
977 /* Page cache limit. The filesystems should put that into their s_maxbytes
978 limits, otherwise bad things can happen in VM. */
979 #if BITS_PER_LONG==32
980 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
981 #elif BITS_PER_LONG==64
982 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
987 #define FL_DELEG 4 /* NFSv4 delegation */
988 #define FL_ACCESS 8 /* not trying to lock, just looking */
989 #define FL_EXISTS 16 /* when unlocking, test for existence */
990 #define FL_LEASE 32 /* lease held on this file */
991 #define FL_CLOSE 64 /* unlock on close */
992 #define FL_SLEEP 128 /* A blocking lock */
993 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
994 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
995 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
996 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
998 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
1001 * Special return value from posix_lock_file() and vfs_lock_file() for
1002 * asynchronous locking.
1004 #define FILE_LOCK_DEFERRED 1
1006 /* legacy typedef, should eventually be removed */
1007 typedef void *fl_owner_t;
1011 struct file_lock_operations {
1012 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1013 void (*fl_release_private)(struct file_lock *);
1016 struct lock_manager_operations {
1017 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
1018 unsigned long (*lm_owner_key)(struct file_lock *);
1019 fl_owner_t (*lm_get_owner)(fl_owner_t);
1020 void (*lm_put_owner)(fl_owner_t);
1021 void (*lm_notify)(struct file_lock *); /* unblock callback */
1022 int (*lm_grant)(struct file_lock *, int);
1023 bool (*lm_break)(struct file_lock *);
1024 int (*lm_change)(struct file_lock *, int, struct list_head *);
1025 void (*lm_setup)(struct file_lock *, void **);
1028 struct lock_manager {
1029 struct list_head list;
1031 * NFSv4 and up also want opens blocked during the grace period;
1038 void locks_start_grace(struct net *, struct lock_manager *);
1039 void locks_end_grace(struct lock_manager *);
1040 bool locks_in_grace(struct net *);
1041 bool opens_in_grace(struct net *);
1043 /* that will die - we need it for nfs_lock_info */
1044 #include <linux/nfs_fs_i.h>
1047 * struct file_lock represents a generic "file lock". It's used to represent
1048 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1049 * note that the same struct is used to represent both a request for a lock and
1050 * the lock itself, but the same object is never used for both.
1052 * FIXME: should we create a separate "struct lock_request" to help distinguish
1055 * The varous i_flctx lists are ordered by:
1058 * 2) lock range start
1061 * Obviously, the last two criteria only matter for POSIX locks.
1064 struct file_lock *fl_blocker; /* The lock, that is blocking us */
1065 struct list_head fl_list; /* link into file_lock_context */
1066 struct hlist_node fl_link; /* node in global lists */
1067 struct list_head fl_blocked_requests; /* list of requests with
1068 * ->fl_blocker pointing here
1070 struct list_head fl_blocked_member; /* node in
1071 * ->fl_blocker->fl_blocked_requests
1073 fl_owner_t fl_owner;
1074 unsigned int fl_flags;
1075 unsigned char fl_type;
1076 unsigned int fl_pid;
1077 int fl_link_cpu; /* what cpu's list is this on? */
1078 wait_queue_head_t fl_wait;
1079 struct file *fl_file;
1083 struct fasync_struct * fl_fasync; /* for lease break notifications */
1084 /* for lease breaks: */
1085 unsigned long fl_break_time;
1086 unsigned long fl_downgrade_time;
1088 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1089 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1091 struct nfs_lock_info nfs_fl;
1092 struct nfs4_lock_info nfs4_fl;
1094 struct list_head link; /* link in AFS vnode's pending_locks list */
1095 int state; /* state of grant or error if -ve */
1098 } __randomize_layout;
1100 struct file_lock_context {
1101 spinlock_t flc_lock;
1102 struct list_head flc_flock;
1103 struct list_head flc_posix;
1104 struct list_head flc_lease;
1107 /* The following constant reflects the upper bound of the file/locking space */
1109 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1110 #define OFFSET_MAX INT_LIMIT(loff_t)
1111 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1114 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1116 #define locks_inode(f) file_inode(f)
1118 #ifdef CONFIG_FILE_LOCKING
1119 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1120 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1123 #if BITS_PER_LONG == 32
1124 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1125 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1129 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1130 extern int fcntl_getlease(struct file *filp);
1133 void locks_free_lock_context(struct inode *inode);
1134 void locks_free_lock(struct file_lock *fl);
1135 extern void locks_init_lock(struct file_lock *);
1136 extern struct file_lock * locks_alloc_lock(void);
1137 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1138 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1139 extern void locks_remove_posix(struct file *, fl_owner_t);
1140 extern void locks_remove_file(struct file *);
1141 extern void locks_release_private(struct file_lock *);
1142 extern void posix_test_lock(struct file *, struct file_lock *);
1143 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1144 extern int locks_delete_block(struct file_lock *);
1145 extern int vfs_test_lock(struct file *, struct file_lock *);
1146 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1147 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1148 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1149 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1150 extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1151 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1152 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1153 extern int lease_modify(struct file_lock *, int, struct list_head *);
1154 struct files_struct;
1155 extern void show_fd_locks(struct seq_file *f,
1156 struct file *filp, struct files_struct *files);
1157 #else /* !CONFIG_FILE_LOCKING */
1158 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1159 struct flock __user *user)
1164 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1165 unsigned int cmd, struct flock __user *user)
1170 #if BITS_PER_LONG == 32
1171 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1172 struct flock64 __user *user)
1177 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1178 unsigned int cmd, struct flock64 __user *user)
1183 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1188 static inline int fcntl_getlease(struct file *filp)
1194 locks_free_lock_context(struct inode *inode)
1198 static inline void locks_init_lock(struct file_lock *fl)
1203 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1208 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1213 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1218 static inline void locks_remove_file(struct file *filp)
1223 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1228 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1229 struct file_lock *conflock)
1234 static inline int locks_delete_block(struct file_lock *waiter)
1239 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1244 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1245 struct file_lock *fl, struct file_lock *conf)
1250 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1255 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1260 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1265 static inline void lease_get_mtime(struct inode *inode,
1266 struct timespec64 *time)
1271 static inline int generic_setlease(struct file *filp, long arg,
1272 struct file_lock **flp, void **priv)
1277 static inline int vfs_setlease(struct file *filp, long arg,
1278 struct file_lock **lease, void **priv)
1283 static inline int lease_modify(struct file_lock *fl, int arg,
1284 struct list_head *dispose)
1289 struct files_struct;
1290 static inline void show_fd_locks(struct seq_file *f,
1291 struct file *filp, struct files_struct *files) {}
1292 #endif /* !CONFIG_FILE_LOCKING */
1294 static inline struct inode *file_inode(const struct file *f)
1299 static inline struct dentry *file_dentry(const struct file *file)
1301 return d_real(file->f_path.dentry, file_inode(file));
1304 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1306 return locks_lock_inode_wait(locks_inode(filp), fl);
1309 struct fasync_struct {
1313 struct fasync_struct *fa_next; /* singly linked list */
1314 struct file *fa_file;
1315 struct rcu_head fa_rcu;
1318 #define FASYNC_MAGIC 0x4601
1320 /* SMP safe fasync helpers: */
1321 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1322 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1323 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1324 extern struct fasync_struct *fasync_alloc(void);
1325 extern void fasync_free(struct fasync_struct *);
1327 /* can be called from interrupts */
1328 extern void kill_fasync(struct fasync_struct **, int, int);
1330 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1331 extern int f_setown(struct file *filp, unsigned long arg, int force);
1332 extern void f_delown(struct file *filp);
1333 extern pid_t f_getown(struct file *filp);
1334 extern int send_sigurg(struct fown_struct *fown);
1337 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1338 * represented in both.
1340 #define SB_RDONLY 1 /* Mount read-only */
1341 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1342 #define SB_NODEV 4 /* Disallow access to device special files */
1343 #define SB_NOEXEC 8 /* Disallow program execution */
1344 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1345 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1346 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1347 #define SB_NOATIME 1024 /* Do not update access times. */
1348 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1349 #define SB_SILENT 32768
1350 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1351 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1352 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1353 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1355 /* These sb flags are internal to the kernel */
1356 #define SB_SUBMOUNT (1<<26)
1357 #define SB_FORCE (1<<27)
1358 #define SB_NOSEC (1<<28)
1359 #define SB_BORN (1<<29)
1360 #define SB_ACTIVE (1<<30)
1361 #define SB_NOUSER (1<<31)
1367 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1368 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1369 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1370 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1371 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1374 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1375 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1376 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1377 #define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */
1379 /* sb->s_iflags to limit user namespace mounts */
1380 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1381 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1382 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1384 /* Possible states of 'frozen' field */
1386 SB_UNFROZEN = 0, /* FS is unfrozen */
1387 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1388 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1389 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1390 * internal threads if needed) */
1391 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1394 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1397 int frozen; /* Is sb frozen? */
1398 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1399 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1402 struct super_block {
1403 struct list_head s_list; /* Keep this first */
1404 dev_t s_dev; /* search index; _not_ kdev_t */
1405 unsigned char s_blocksize_bits;
1406 unsigned long s_blocksize;
1407 loff_t s_maxbytes; /* Max file size */
1408 struct file_system_type *s_type;
1409 const struct super_operations *s_op;
1410 const struct dquot_operations *dq_op;
1411 const struct quotactl_ops *s_qcop;
1412 const struct export_operations *s_export_op;
1413 unsigned long s_flags;
1414 unsigned long s_iflags; /* internal SB_I_* flags */
1415 unsigned long s_magic;
1416 struct dentry *s_root;
1417 struct rw_semaphore s_umount;
1420 #ifdef CONFIG_SECURITY
1423 const struct xattr_handler **s_xattr;
1424 #ifdef CONFIG_FS_ENCRYPTION
1425 const struct fscrypt_operations *s_cop;
1427 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1428 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1429 struct block_device *s_bdev;
1430 struct backing_dev_info *s_bdi;
1431 struct mtd_info *s_mtd;
1432 struct hlist_node s_instances;
1433 unsigned int s_quota_types; /* Bitmask of supported quota types */
1434 struct quota_info s_dquot; /* Diskquota specific options */
1436 struct sb_writers s_writers;
1439 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1440 * s_fsnotify_marks together for cache efficiency. They are frequently
1441 * accessed and rarely modified.
1443 void *s_fs_info; /* Filesystem private info */
1445 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1447 #ifdef CONFIG_FSNOTIFY
1448 __u32 s_fsnotify_mask;
1449 struct fsnotify_mark_connector __rcu *s_fsnotify_marks;
1452 char s_id[32]; /* Informational name */
1453 uuid_t s_uuid; /* UUID */
1455 unsigned int s_max_links;
1459 * The next field is for VFS *only*. No filesystems have any business
1460 * even looking at it. You had been warned.
1462 struct mutex s_vfs_rename_mutex; /* Kludge */
1465 * Filesystem subtype. If non-empty the filesystem type field
1466 * in /proc/mounts will be "type.subtype"
1468 const char *s_subtype;
1470 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1473 * Saved pool identifier for cleancache (-1 means none)
1475 int cleancache_poolid;
1477 struct shrinker s_shrink; /* per-sb shrinker handle */
1479 /* Number of inodes with nlink == 0 but still referenced */
1480 atomic_long_t s_remove_count;
1482 /* Pending fsnotify inode refs */
1483 atomic_long_t s_fsnotify_inode_refs;
1485 /* Being remounted read-only */
1486 int s_readonly_remount;
1488 /* AIO completions deferred from interrupt context */
1489 struct workqueue_struct *s_dio_done_wq;
1490 struct hlist_head s_pins;
1493 * Owning user namespace and default context in which to
1494 * interpret filesystem uids, gids, quotas, device nodes,
1495 * xattrs and security labels.
1497 struct user_namespace *s_user_ns;
1500 * The list_lru structure is essentially just a pointer to a table
1501 * of per-node lru lists, each of which has its own spinlock.
1502 * There is no need to put them into separate cachelines.
1504 struct list_lru s_dentry_lru;
1505 struct list_lru s_inode_lru;
1506 struct rcu_head rcu;
1507 struct work_struct destroy_work;
1509 struct mutex s_sync_lock; /* sync serialisation lock */
1512 * Indicates how deep in a filesystem stack this SB is
1516 /* s_inode_list_lock protects s_inodes */
1517 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1518 struct list_head s_inodes; /* all inodes */
1520 spinlock_t s_inode_wblist_lock;
1521 struct list_head s_inodes_wb; /* writeback inodes */
1522 } __randomize_layout;
1524 /* Helper functions so that in most cases filesystems will
1525 * not need to deal directly with kuid_t and kgid_t and can
1526 * instead deal with the raw numeric values that are stored
1527 * in the filesystem.
1529 static inline uid_t i_uid_read(const struct inode *inode)
1531 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1534 static inline gid_t i_gid_read(const struct inode *inode)
1536 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1539 static inline void i_uid_write(struct inode *inode, uid_t uid)
1541 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1544 static inline void i_gid_write(struct inode *inode, gid_t gid)
1546 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1549 extern struct timespec64 timespec64_trunc(struct timespec64 t, unsigned gran);
1550 extern struct timespec64 current_time(struct inode *inode);
1553 * Snapshotting support.
1556 void __sb_end_write(struct super_block *sb, int level);
1557 int __sb_start_write(struct super_block *sb, int level, bool wait);
1559 #define __sb_writers_acquired(sb, lev) \
1560 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1561 #define __sb_writers_release(sb, lev) \
1562 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1565 * sb_end_write - drop write access to a superblock
1566 * @sb: the super we wrote to
1568 * Decrement number of writers to the filesystem. Wake up possible waiters
1569 * wanting to freeze the filesystem.
1571 static inline void sb_end_write(struct super_block *sb)
1573 __sb_end_write(sb, SB_FREEZE_WRITE);
1577 * sb_end_pagefault - drop write access to a superblock from a page fault
1578 * @sb: the super we wrote to
1580 * Decrement number of processes handling write page fault to the filesystem.
1581 * Wake up possible waiters wanting to freeze the filesystem.
1583 static inline void sb_end_pagefault(struct super_block *sb)
1585 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1589 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1590 * @sb: the super we wrote to
1592 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1593 * waiters wanting to freeze the filesystem.
1595 static inline void sb_end_intwrite(struct super_block *sb)
1597 __sb_end_write(sb, SB_FREEZE_FS);
1601 * sb_start_write - get write access to a superblock
1602 * @sb: the super we write to
1604 * When a process wants to write data or metadata to a file system (i.e. dirty
1605 * a page or an inode), it should embed the operation in a sb_start_write() -
1606 * sb_end_write() pair to get exclusion against file system freezing. This
1607 * function increments number of writers preventing freezing. If the file
1608 * system is already frozen, the function waits until the file system is
1611 * Since freeze protection behaves as a lock, users have to preserve
1612 * ordering of freeze protection and other filesystem locks. Generally,
1613 * freeze protection should be the outermost lock. In particular, we have:
1616 * -> i_mutex (write path, truncate, directory ops, ...)
1617 * -> s_umount (freeze_super, thaw_super)
1619 static inline void sb_start_write(struct super_block *sb)
1621 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1624 static inline int sb_start_write_trylock(struct super_block *sb)
1626 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1630 * sb_start_pagefault - get write access to a superblock from a page fault
1631 * @sb: the super we write to
1633 * When a process starts handling write page fault, it should embed the
1634 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1635 * exclusion against file system freezing. This is needed since the page fault
1636 * is going to dirty a page. This function increments number of running page
1637 * faults preventing freezing. If the file system is already frozen, the
1638 * function waits until the file system is thawed.
1640 * Since page fault freeze protection behaves as a lock, users have to preserve
1641 * ordering of freeze protection and other filesystem locks. It is advised to
1642 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1643 * handling code implies lock dependency:
1646 * -> sb_start_pagefault
1648 static inline void sb_start_pagefault(struct super_block *sb)
1650 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1654 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1655 * @sb: the super we write to
1657 * This is the third level of protection against filesystem freezing. It is
1658 * free for use by a filesystem. The only requirement is that it must rank
1659 * below sb_start_pagefault.
1661 * For example filesystem can call sb_start_intwrite() when starting a
1662 * transaction which somewhat eases handling of freezing for internal sources
1663 * of filesystem changes (internal fs threads, discarding preallocation on file
1666 static inline void sb_start_intwrite(struct super_block *sb)
1668 __sb_start_write(sb, SB_FREEZE_FS, true);
1671 static inline int sb_start_intwrite_trylock(struct super_block *sb)
1673 return __sb_start_write(sb, SB_FREEZE_FS, false);
1677 extern bool inode_owner_or_capable(const struct inode *inode);
1680 * VFS helper functions..
1682 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1683 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1684 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1685 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1686 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1687 extern int vfs_rmdir(struct inode *, struct dentry *);
1688 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1689 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1690 extern int vfs_whiteout(struct inode *, struct dentry *);
1692 extern struct dentry *vfs_tmpfile(struct dentry *dentry, umode_t mode,
1695 int vfs_mkobj(struct dentry *, umode_t,
1696 int (*f)(struct dentry *, umode_t, void *),
1699 extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1702 * VFS file helper functions.
1704 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1706 extern bool may_open_dev(const struct path *path);
1708 * VFS FS_IOC_FIEMAP helper definitions.
1710 struct fiemap_extent_info {
1711 unsigned int fi_flags; /* Flags as passed from user */
1712 unsigned int fi_extents_mapped; /* Number of mapped extents */
1713 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1714 struct fiemap_extent __user *fi_extents_start; /* Start of
1715 fiemap_extent array */
1717 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1718 u64 phys, u64 len, u32 flags);
1719 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1722 * This is the "filldir" function type, used by readdir() to let
1723 * the kernel specify what kind of dirent layout it wants to have.
1724 * This allows the kernel to read directories into kernel space or
1725 * to have different dirent layouts depending on the binary type.
1728 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1731 struct dir_context {
1736 struct block_device_operations;
1738 /* These macros are for out of kernel modules to test that
1739 * the kernel supports the unlocked_ioctl and compat_ioctl
1740 * fields in struct file_operations. */
1741 #define HAVE_COMPAT_IOCTL 1
1742 #define HAVE_UNLOCKED_IOCTL 1
1745 * These flags let !MMU mmap() govern direct device mapping vs immediate
1746 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1748 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1749 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1750 * NOMMU_MAP_READ: Can be mapped for reading
1751 * NOMMU_MAP_WRITE: Can be mapped for writing
1752 * NOMMU_MAP_EXEC: Can be mapped for execution
1754 #define NOMMU_MAP_COPY 0x00000001
1755 #define NOMMU_MAP_DIRECT 0x00000008
1756 #define NOMMU_MAP_READ VM_MAYREAD
1757 #define NOMMU_MAP_WRITE VM_MAYWRITE
1758 #define NOMMU_MAP_EXEC VM_MAYEXEC
1760 #define NOMMU_VMFLAGS \
1761 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1764 * These flags control the behavior of the remap_file_range function pointer.
1765 * If it is called with len == 0 that means "remap to end of source file".
1766 * See Documentation/filesystems/vfs.txt for more details about this call.
1768 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
1769 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
1771 #define REMAP_FILE_DEDUP (1 << 0)
1772 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
1775 * These flags signal that the caller is ok with altering various aspects of
1776 * the behavior of the remap operation. The changes must be made by the
1777 * implementation; the vfs remap helper functions can take advantage of them.
1778 * Flags in this category exist to preserve the quirky behavior of the hoisted
1779 * btrfs clone/dedupe ioctls.
1781 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
1785 struct file_operations {
1786 struct module *owner;
1787 loff_t (*llseek) (struct file *, loff_t, int);
1788 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1789 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1790 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1791 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1792 int (*iopoll)(struct kiocb *kiocb, bool spin);
1793 int (*iterate) (struct file *, struct dir_context *);
1794 int (*iterate_shared) (struct file *, struct dir_context *);
1795 __poll_t (*poll) (struct file *, struct poll_table_struct *);
1796 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1797 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1798 int (*mmap) (struct file *, struct vm_area_struct *);
1799 unsigned long mmap_supported_flags;
1800 int (*open) (struct inode *, struct file *);
1801 int (*flush) (struct file *, fl_owner_t id);
1802 int (*release) (struct inode *, struct file *);
1803 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1804 int (*fasync) (int, struct file *, int);
1805 int (*lock) (struct file *, int, struct file_lock *);
1806 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1807 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1808 int (*check_flags)(int);
1809 int (*flock) (struct file *, int, struct file_lock *);
1810 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1811 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1812 int (*setlease)(struct file *, long, struct file_lock **, void **);
1813 long (*fallocate)(struct file *file, int mode, loff_t offset,
1815 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1817 unsigned (*mmap_capabilities)(struct file *);
1819 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1820 loff_t, size_t, unsigned int);
1821 loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
1822 struct file *file_out, loff_t pos_out,
1823 loff_t len, unsigned int remap_flags);
1824 int (*fadvise)(struct file *, loff_t, loff_t, int);
1825 } __randomize_layout;
1827 struct inode_operations {
1828 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1829 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1830 int (*permission) (struct inode *, int);
1831 struct posix_acl * (*get_acl)(struct inode *, int);
1833 int (*readlink) (struct dentry *, char __user *,int);
1835 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1836 int (*link) (struct dentry *,struct inode *,struct dentry *);
1837 int (*unlink) (struct inode *,struct dentry *);
1838 int (*symlink) (struct inode *,struct dentry *,const char *);
1839 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1840 int (*rmdir) (struct inode *,struct dentry *);
1841 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1842 int (*rename) (struct inode *, struct dentry *,
1843 struct inode *, struct dentry *, unsigned int);
1844 int (*setattr) (struct dentry *, struct iattr *);
1845 int (*getattr) (const struct path *, struct kstat *, u32, unsigned int);
1846 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1847 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1849 int (*update_time)(struct inode *, struct timespec64 *, int);
1850 int (*atomic_open)(struct inode *, struct dentry *,
1851 struct file *, unsigned open_flag,
1852 umode_t create_mode);
1853 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1854 int (*set_acl)(struct inode *, struct posix_acl *, int);
1855 } ____cacheline_aligned;
1857 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
1858 struct iov_iter *iter)
1860 return file->f_op->read_iter(kio, iter);
1863 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
1864 struct iov_iter *iter)
1866 return file->f_op->write_iter(kio, iter);
1869 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
1871 return file->f_op->mmap(file, vma);
1874 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1875 unsigned long nr_segs, unsigned long fast_segs,
1876 struct iovec *fast_pointer,
1877 struct iovec **ret_pointer);
1879 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1880 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1881 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1882 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1883 unsigned long, loff_t *, rwf_t);
1884 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1885 loff_t, size_t, unsigned int);
1886 extern int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
1887 struct file *file_out, loff_t pos_out,
1889 unsigned int remap_flags);
1890 extern loff_t do_clone_file_range(struct file *file_in, loff_t pos_in,
1891 struct file *file_out, loff_t pos_out,
1892 loff_t len, unsigned int remap_flags);
1893 extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1894 struct file *file_out, loff_t pos_out,
1895 loff_t len, unsigned int remap_flags);
1896 extern int vfs_dedupe_file_range(struct file *file,
1897 struct file_dedupe_range *same);
1898 extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
1899 struct file *dst_file, loff_t dst_pos,
1900 loff_t len, unsigned int remap_flags);
1903 struct super_operations {
1904 struct inode *(*alloc_inode)(struct super_block *sb);
1905 void (*destroy_inode)(struct inode *);
1907 void (*dirty_inode) (struct inode *, int flags);
1908 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1909 int (*drop_inode) (struct inode *);
1910 void (*evict_inode) (struct inode *);
1911 void (*put_super) (struct super_block *);
1912 int (*sync_fs)(struct super_block *sb, int wait);
1913 int (*freeze_super) (struct super_block *);
1914 int (*freeze_fs) (struct super_block *);
1915 int (*thaw_super) (struct super_block *);
1916 int (*unfreeze_fs) (struct super_block *);
1917 int (*statfs) (struct dentry *, struct kstatfs *);
1918 int (*remount_fs) (struct super_block *, int *, char *);
1919 void (*umount_begin) (struct super_block *);
1921 int (*show_options)(struct seq_file *, struct dentry *);
1922 int (*show_devname)(struct seq_file *, struct dentry *);
1923 int (*show_path)(struct seq_file *, struct dentry *);
1924 int (*show_stats)(struct seq_file *, struct dentry *);
1926 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1927 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1928 struct dquot **(*get_dquots)(struct inode *);
1930 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1931 long (*nr_cached_objects)(struct super_block *,
1932 struct shrink_control *);
1933 long (*free_cached_objects)(struct super_block *,
1934 struct shrink_control *);
1938 * Inode flags - they have no relation to superblock flags now
1940 #define S_SYNC 1 /* Writes are synced at once */
1941 #define S_NOATIME 2 /* Do not update access times */
1942 #define S_APPEND 4 /* Append-only file */
1943 #define S_IMMUTABLE 8 /* Immutable file */
1944 #define S_DEAD 16 /* removed, but still open directory */
1945 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1946 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1947 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1948 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1949 #define S_PRIVATE 512 /* Inode is fs-internal */
1950 #define S_IMA 1024 /* Inode has an associated IMA struct */
1951 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1952 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1953 #ifdef CONFIG_FS_DAX
1954 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1956 #define S_DAX 0 /* Make all the DAX code disappear */
1958 #define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */
1961 * Note that nosuid etc flags are inode-specific: setting some file-system
1962 * flags just means all the inodes inherit those flags by default. It might be
1963 * possible to override it selectively if you really wanted to with some
1964 * ioctl() that is not currently implemented.
1966 * Exception: SB_RDONLY is always applied to the entire file system.
1968 * Unfortunately, it is possible to change a filesystems flags with it mounted
1969 * with files in use. This means that all of the inodes will not have their
1970 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1971 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1973 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1975 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
1976 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
1977 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
1978 ((inode)->i_flags & S_SYNC))
1979 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
1980 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1981 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
1982 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
1983 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
1985 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1986 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1987 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1988 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
1990 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1991 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1992 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1993 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1994 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1995 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1996 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1997 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1998 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2000 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2001 (inode)->i_rdev == WHITEOUT_DEV)
2003 static inline bool HAS_UNMAPPED_ID(struct inode *inode)
2005 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
2008 static inline enum rw_hint file_write_hint(struct file *file)
2010 if (file->f_write_hint != WRITE_LIFE_NOT_SET)
2011 return file->f_write_hint;
2013 return file_inode(file)->i_write_hint;
2016 static inline int iocb_flags(struct file *file);
2018 static inline u16 ki_hint_validate(enum rw_hint hint)
2020 typeof(((struct kiocb *)0)->ki_hint) max_hint = -1;
2022 if (hint <= max_hint)
2027 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
2029 *kiocb = (struct kiocb) {
2031 .ki_flags = iocb_flags(filp),
2032 .ki_hint = ki_hint_validate(file_write_hint(filp)),
2033 .ki_ioprio = get_current_ioprio(),
2038 * Inode state bits. Protected by inode->i_lock
2040 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
2041 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
2043 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2044 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2045 * various stages of removing an inode.
2047 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2049 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2050 * fdatasync(). i_atime is the usual cause.
2051 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2052 * these changes separately from I_DIRTY_SYNC so that we
2053 * don't have to write inode on fdatasync() when only
2054 * mtime has changed in it.
2055 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2056 * I_NEW Serves as both a mutex and completion notification.
2057 * New inodes set I_NEW. If two processes both create
2058 * the same inode, one of them will release its inode and
2059 * wait for I_NEW to be released before returning.
2060 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2061 * also cause waiting on I_NEW, without I_NEW actually
2062 * being set. find_inode() uses this to prevent returning
2063 * nearly-dead inodes.
2064 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2065 * is zero. I_FREEING must be set when I_WILL_FREE is
2067 * I_FREEING Set when inode is about to be freed but still has dirty
2068 * pages or buffers attached or the inode itself is still
2070 * I_CLEAR Added by clear_inode(). In this state the inode is
2071 * clean and can be destroyed. Inode keeps I_FREEING.
2073 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2074 * prohibited for many purposes. iget() must wait for
2075 * the inode to be completely released, then create it
2076 * anew. Other functions will just ignore such inodes,
2077 * if appropriate. I_NEW is used for waiting.
2079 * I_SYNC Writeback of inode is running. The bit is set during
2080 * data writeback, and cleared with a wakeup on the bit
2081 * address once it is done. The bit is also used to pin
2082 * the inode in memory for flusher thread.
2084 * I_REFERENCED Marks the inode as recently references on the LRU list.
2086 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2088 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2089 * synchronize competing switching instances and to tell
2090 * wb stat updates to grab the i_pages lock. See
2091 * inode_switch_wbs_work_fn() for details.
2093 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2094 * and work dirs among overlayfs mounts.
2096 * I_CREATING New object's inode in the middle of setting up.
2098 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2100 #define I_DIRTY_SYNC (1 << 0)
2101 #define I_DIRTY_DATASYNC (1 << 1)
2102 #define I_DIRTY_PAGES (1 << 2)
2104 #define I_NEW (1 << __I_NEW)
2105 #define I_WILL_FREE (1 << 4)
2106 #define I_FREEING (1 << 5)
2107 #define I_CLEAR (1 << 6)
2109 #define I_SYNC (1 << __I_SYNC)
2110 #define I_REFERENCED (1 << 8)
2111 #define __I_DIO_WAKEUP 9
2112 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2113 #define I_LINKABLE (1 << 10)
2114 #define I_DIRTY_TIME (1 << 11)
2115 #define __I_DIRTY_TIME_EXPIRED 12
2116 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
2117 #define I_WB_SWITCH (1 << 13)
2118 #define I_OVL_INUSE (1 << 14)
2119 #define I_CREATING (1 << 15)
2121 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2122 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2123 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2125 extern void __mark_inode_dirty(struct inode *, int);
2126 static inline void mark_inode_dirty(struct inode *inode)
2128 __mark_inode_dirty(inode, I_DIRTY);
2131 static inline void mark_inode_dirty_sync(struct inode *inode)
2133 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2136 extern void inc_nlink(struct inode *inode);
2137 extern void drop_nlink(struct inode *inode);
2138 extern void clear_nlink(struct inode *inode);
2139 extern void set_nlink(struct inode *inode, unsigned int nlink);
2141 static inline void inode_inc_link_count(struct inode *inode)
2144 mark_inode_dirty(inode);
2147 static inline void inode_dec_link_count(struct inode *inode)
2150 mark_inode_dirty(inode);
2153 enum file_time_flags {
2160 extern bool atime_needs_update(const struct path *, struct inode *);
2161 extern void touch_atime(const struct path *);
2162 static inline void file_accessed(struct file *file)
2164 if (!(file->f_flags & O_NOATIME))
2165 touch_atime(&file->f_path);
2168 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2169 int sync_inode_metadata(struct inode *inode, int wait);
2171 struct file_system_type {
2174 #define FS_REQUIRES_DEV 1
2175 #define FS_BINARY_MOUNTDATA 2
2176 #define FS_HAS_SUBTYPE 4
2177 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2178 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2179 int (*init_fs_context)(struct fs_context *);
2180 const struct fs_parameter_description *parameters;
2181 struct dentry *(*mount) (struct file_system_type *, int,
2182 const char *, void *);
2183 void (*kill_sb) (struct super_block *);
2184 struct module *owner;
2185 struct file_system_type * next;
2186 struct hlist_head fs_supers;
2188 struct lock_class_key s_lock_key;
2189 struct lock_class_key s_umount_key;
2190 struct lock_class_key s_vfs_rename_key;
2191 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2193 struct lock_class_key i_lock_key;
2194 struct lock_class_key i_mutex_key;
2195 struct lock_class_key i_mutex_dir_key;
2198 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2200 extern struct dentry *mount_ns(struct file_system_type *fs_type,
2201 int flags, void *data, void *ns, struct user_namespace *user_ns,
2202 int (*fill_super)(struct super_block *, void *, int));
2204 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2205 int flags, const char *dev_name, void *data,
2206 int (*fill_super)(struct super_block *, void *, int));
2208 static inline struct dentry *mount_bdev(struct file_system_type *fs_type,
2209 int flags, const char *dev_name, void *data,
2210 int (*fill_super)(struct super_block *, void *, int))
2212 return ERR_PTR(-ENODEV);
2215 extern struct dentry *mount_single(struct file_system_type *fs_type,
2216 int flags, void *data,
2217 int (*fill_super)(struct super_block *, void *, int));
2218 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2219 int flags, void *data,
2220 int (*fill_super)(struct super_block *, void *, int));
2221 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2222 void generic_shutdown_super(struct super_block *sb);
2224 void kill_block_super(struct super_block *sb);
2226 static inline void kill_block_super(struct super_block *sb)
2231 void kill_anon_super(struct super_block *sb);
2232 void kill_litter_super(struct super_block *sb);
2233 void deactivate_super(struct super_block *sb);
2234 void deactivate_locked_super(struct super_block *sb);
2235 int set_anon_super(struct super_block *s, void *data);
2236 int set_anon_super_fc(struct super_block *s, struct fs_context *fc);
2237 int get_anon_bdev(dev_t *);
2238 void free_anon_bdev(dev_t);
2239 struct super_block *sget_fc(struct fs_context *fc,
2240 int (*test)(struct super_block *, struct fs_context *),
2241 int (*set)(struct super_block *, struct fs_context *));
2242 struct super_block *sget_userns(struct file_system_type *type,
2243 int (*test)(struct super_block *,void *),
2244 int (*set)(struct super_block *,void *),
2245 int flags, struct user_namespace *user_ns,
2247 struct super_block *sget(struct file_system_type *type,
2248 int (*test)(struct super_block *,void *),
2249 int (*set)(struct super_block *,void *),
2250 int flags, void *data);
2251 extern struct dentry *mount_pseudo_xattr(struct file_system_type *, char *,
2252 const struct super_operations *ops,
2253 const struct xattr_handler **xattr,
2254 const struct dentry_operations *dops,
2257 static inline struct dentry *
2258 mount_pseudo(struct file_system_type *fs_type, char *name,
2259 const struct super_operations *ops,
2260 const struct dentry_operations *dops, unsigned long magic)
2262 return mount_pseudo_xattr(fs_type, name, ops, NULL, dops, magic);
2265 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2266 #define fops_get(fops) \
2267 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2268 #define fops_put(fops) \
2269 do { if (fops) module_put((fops)->owner); } while(0)
2271 * This one is to be used *ONLY* from ->open() instances.
2272 * fops must be non-NULL, pinned down *and* module dependencies
2273 * should be sufficient to pin the caller down as well.
2275 #define replace_fops(f, fops) \
2277 struct file *__file = (f); \
2278 fops_put(__file->f_op); \
2279 BUG_ON(!(__file->f_op = (fops))); \
2282 extern int register_filesystem(struct file_system_type *);
2283 extern int unregister_filesystem(struct file_system_type *);
2284 extern struct vfsmount *kern_mount(struct file_system_type *);
2285 extern void kern_unmount(struct vfsmount *mnt);
2286 extern int may_umount_tree(struct vfsmount *);
2287 extern int may_umount(struct vfsmount *);
2288 extern long do_mount(const char *, const char __user *,
2289 const char *, unsigned long, void *);
2290 extern struct vfsmount *collect_mounts(const struct path *);
2291 extern void drop_collected_mounts(struct vfsmount *);
2292 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2294 extern int vfs_statfs(const struct path *, struct kstatfs *);
2295 extern int user_statfs(const char __user *, struct kstatfs *);
2296 extern int fd_statfs(int, struct kstatfs *);
2297 extern int freeze_super(struct super_block *super);
2298 extern int thaw_super(struct super_block *super);
2299 extern bool our_mnt(struct vfsmount *mnt);
2300 extern __printf(2, 3)
2301 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2302 extern int super_setup_bdi(struct super_block *sb);
2304 extern int current_umask(void);
2306 extern void ihold(struct inode * inode);
2307 extern void iput(struct inode *);
2308 extern int generic_update_time(struct inode *, struct timespec64 *, int);
2311 extern struct kobject *fs_kobj;
2313 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2315 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2316 extern int locks_mandatory_locked(struct file *);
2317 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2320 * Candidates for mandatory locking have the setgid bit set
2321 * but no group execute bit - an otherwise meaningless combination.
2324 static inline int __mandatory_lock(struct inode *ino)
2326 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2330 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2331 * otherwise these will be advisory locks
2334 static inline int mandatory_lock(struct inode *ino)
2336 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2339 static inline int locks_verify_locked(struct file *file)
2341 if (mandatory_lock(locks_inode(file)))
2342 return locks_mandatory_locked(file);
2346 static inline int locks_verify_truncate(struct inode *inode,
2350 if (!inode->i_flctx || !mandatory_lock(inode))
2353 if (size < inode->i_size) {
2354 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2357 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2362 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2364 static inline int locks_mandatory_locked(struct file *file)
2369 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2370 loff_t start, loff_t end, unsigned char type)
2375 static inline int __mandatory_lock(struct inode *inode)
2380 static inline int mandatory_lock(struct inode *inode)
2385 static inline int locks_verify_locked(struct file *file)
2390 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2396 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2399 #ifdef CONFIG_FILE_LOCKING
2400 static inline int break_lease(struct inode *inode, unsigned int mode)
2403 * Since this check is lockless, we must ensure that any refcounts
2404 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2405 * could end up racing with tasks trying to set a new lease on this
2409 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2410 return __break_lease(inode, mode, FL_LEASE);
2414 static inline int break_deleg(struct inode *inode, unsigned int mode)
2417 * Since this check is lockless, we must ensure that any refcounts
2418 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2419 * could end up racing with tasks trying to set a new lease on this
2423 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2424 return __break_lease(inode, mode, FL_DELEG);
2428 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2432 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2433 if (ret == -EWOULDBLOCK && delegated_inode) {
2434 *delegated_inode = inode;
2440 static inline int break_deleg_wait(struct inode **delegated_inode)
2444 ret = break_deleg(*delegated_inode, O_WRONLY);
2445 iput(*delegated_inode);
2446 *delegated_inode = NULL;
2450 static inline int break_layout(struct inode *inode, bool wait)
2453 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2454 return __break_lease(inode,
2455 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2460 #else /* !CONFIG_FILE_LOCKING */
2461 static inline int break_lease(struct inode *inode, unsigned int mode)
2466 static inline int break_deleg(struct inode *inode, unsigned int mode)
2471 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2476 static inline int break_deleg_wait(struct inode **delegated_inode)
2482 static inline int break_layout(struct inode *inode, bool wait)
2487 #endif /* CONFIG_FILE_LOCKING */
2492 const char *name; /* pointer to actual string */
2493 const __user char *uptr; /* original userland pointer */
2495 struct audit_names *aname;
2498 static_assert(offsetof(struct filename, iname) % sizeof(long) == 0);
2500 extern long vfs_truncate(const struct path *, loff_t);
2501 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2503 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2505 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2507 extern struct file *file_open_name(struct filename *, int, umode_t);
2508 extern struct file *filp_open(const char *, int, umode_t);
2509 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2510 const char *, int, umode_t);
2511 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2512 extern struct file * open_with_fake_path(const struct path *, int,
2513 struct inode*, const struct cred *);
2514 static inline struct file *file_clone_open(struct file *file)
2516 return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2518 extern int filp_close(struct file *, fl_owner_t id);
2520 extern struct filename *getname_flags(const char __user *, int, int *);
2521 extern struct filename *getname(const char __user *);
2522 extern struct filename *getname_kernel(const char *);
2523 extern void putname(struct filename *name);
2525 extern int finish_open(struct file *file, struct dentry *dentry,
2526 int (*open)(struct inode *, struct file *));
2527 extern int finish_no_open(struct file *file, struct dentry *dentry);
2531 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2534 extern void __init vfs_caches_init_early(void);
2535 extern void __init vfs_caches_init(void);
2537 extern struct kmem_cache *names_cachep;
2539 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2540 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2543 extern int register_blkdev(unsigned int, const char *);
2544 extern void unregister_blkdev(unsigned int, const char *);
2545 extern void bdev_unhash_inode(dev_t dev);
2546 extern struct block_device *bdget(dev_t);
2547 extern struct block_device *bdgrab(struct block_device *bdev);
2548 extern void bd_set_size(struct block_device *, loff_t size);
2549 extern void bd_forget(struct inode *inode);
2550 extern void bdput(struct block_device *);
2551 extern void invalidate_bdev(struct block_device *);
2552 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2553 extern int sync_blockdev(struct block_device *bdev);
2554 extern void kill_bdev(struct block_device *);
2555 extern struct super_block *freeze_bdev(struct block_device *);
2556 extern void emergency_thaw_all(void);
2557 extern void emergency_thaw_bdev(struct super_block *sb);
2558 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2559 extern int fsync_bdev(struct block_device *);
2561 extern struct super_block *blockdev_superblock;
2563 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2565 return sb == blockdev_superblock;
2568 static inline void bd_forget(struct inode *inode) {}
2569 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2570 static inline void kill_bdev(struct block_device *bdev) {}
2571 static inline void invalidate_bdev(struct block_device *bdev) {}
2573 static inline struct super_block *freeze_bdev(struct block_device *sb)
2578 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2583 static inline int emergency_thaw_bdev(struct super_block *sb)
2588 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2592 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2597 extern int sync_filesystem(struct super_block *);
2598 extern const struct file_operations def_blk_fops;
2599 extern const struct file_operations def_chr_fops;
2601 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2602 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2603 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2604 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2605 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2607 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2609 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2610 extern int __blkdev_reread_part(struct block_device *bdev);
2611 extern int blkdev_reread_part(struct block_device *bdev);
2614 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2615 extern void bd_unlink_disk_holder(struct block_device *bdev,
2616 struct gendisk *disk);
2618 static inline int bd_link_disk_holder(struct block_device *bdev,
2619 struct gendisk *disk)
2623 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2624 struct gendisk *disk)
2631 #define CHRDEV_MAJOR_MAX 512
2632 /* Marks the bottom of the first segment of free char majors */
2633 #define CHRDEV_MAJOR_DYN_END 234
2634 /* Marks the top and bottom of the second segment of free char majors */
2635 #define CHRDEV_MAJOR_DYN_EXT_START 511
2636 #define CHRDEV_MAJOR_DYN_EXT_END 384
2638 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2639 extern int register_chrdev_region(dev_t, unsigned, const char *);
2640 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2641 unsigned int count, const char *name,
2642 const struct file_operations *fops);
2643 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2644 unsigned int count, const char *name);
2645 extern void unregister_chrdev_region(dev_t, unsigned);
2646 extern void chrdev_show(struct seq_file *,off_t);
2648 static inline int register_chrdev(unsigned int major, const char *name,
2649 const struct file_operations *fops)
2651 return __register_chrdev(major, 0, 256, name, fops);
2654 static inline void unregister_chrdev(unsigned int major, const char *name)
2656 __unregister_chrdev(major, 0, 256, name);
2659 /* fs/block_dev.c */
2660 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2661 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2664 #define BLKDEV_MAJOR_MAX 512
2665 extern const char *__bdevname(dev_t, char *buffer);
2666 extern const char *bdevname(struct block_device *bdev, char *buffer);
2667 extern struct block_device *lookup_bdev(const char *);
2668 extern void blkdev_show(struct seq_file *,off_t);
2671 #define BLKDEV_MAJOR_MAX 0
2674 extern void init_special_inode(struct inode *, umode_t, dev_t);
2676 /* Invalid inode operations -- fs/bad_inode.c */
2677 extern void make_bad_inode(struct inode *);
2678 extern bool is_bad_inode(struct inode *);
2681 extern void check_disk_size_change(struct gendisk *disk,
2682 struct block_device *bdev, bool verbose);
2683 extern int revalidate_disk(struct gendisk *);
2684 extern int check_disk_change(struct block_device *);
2685 extern int __invalidate_device(struct block_device *, bool);
2686 extern int invalidate_partition(struct gendisk *, int);
2688 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2689 pgoff_t start, pgoff_t end);
2691 static inline void invalidate_remote_inode(struct inode *inode)
2693 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2694 S_ISLNK(inode->i_mode))
2695 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2697 extern int invalidate_inode_pages2(struct address_space *mapping);
2698 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2699 pgoff_t start, pgoff_t end);
2700 extern int write_inode_now(struct inode *, int);
2701 extern int filemap_fdatawrite(struct address_space *);
2702 extern int filemap_flush(struct address_space *);
2703 extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2704 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2707 static inline int filemap_fdatawait(struct address_space *mapping)
2709 return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2712 extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2714 extern int filemap_write_and_wait(struct address_space *mapping);
2715 extern int filemap_write_and_wait_range(struct address_space *mapping,
2716 loff_t lstart, loff_t lend);
2717 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2718 loff_t start, loff_t end, int sync_mode);
2719 extern int filemap_fdatawrite_range(struct address_space *mapping,
2720 loff_t start, loff_t end);
2721 extern int filemap_check_errors(struct address_space *mapping);
2722 extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2724 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2726 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2727 extern int __must_check file_write_and_wait_range(struct file *file,
2728 loff_t start, loff_t end);
2730 static inline int file_write_and_wait(struct file *file)
2732 return file_write_and_wait_range(file, 0, LLONG_MAX);
2736 * filemap_set_wb_err - set a writeback error on an address_space
2737 * @mapping: mapping in which to set writeback error
2738 * @err: error to be set in mapping
2740 * When writeback fails in some way, we must record that error so that
2741 * userspace can be informed when fsync and the like are called. We endeavor
2742 * to report errors on any file that was open at the time of the error. Some
2743 * internal callers also need to know when writeback errors have occurred.
2745 * When a writeback error occurs, most filesystems will want to call
2746 * filemap_set_wb_err to record the error in the mapping so that it will be
2747 * automatically reported whenever fsync is called on the file.
2749 static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2751 /* Fastpath for common case of no error */
2753 __filemap_set_wb_err(mapping, err);
2757 * filemap_check_wb_error - has an error occurred since the mark was sampled?
2758 * @mapping: mapping to check for writeback errors
2759 * @since: previously-sampled errseq_t
2761 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2762 * the given value was sampled.
2764 * If it has then report the latest error set, otherwise return 0.
2766 static inline int filemap_check_wb_err(struct address_space *mapping,
2769 return errseq_check(&mapping->wb_err, since);
2773 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2774 * @mapping: mapping to be sampled
2776 * Writeback errors are always reported relative to a particular sample point
2777 * in the past. This function provides those sample points.
2779 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2781 return errseq_sample(&mapping->wb_err);
2784 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2786 extern int vfs_fsync(struct file *file, int datasync);
2789 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2790 * to already be updated for the write, and will return either the amount
2791 * of bytes passed in, or an error if syncing the file failed.
2793 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2795 if (iocb->ki_flags & IOCB_DSYNC) {
2796 int ret = vfs_fsync_range(iocb->ki_filp,
2797 iocb->ki_pos - count, iocb->ki_pos - 1,
2798 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2806 extern void emergency_sync(void);
2807 extern void emergency_remount(void);
2809 extern sector_t bmap(struct inode *, sector_t);
2811 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2812 extern int inode_permission(struct inode *, int);
2813 extern int generic_permission(struct inode *, int);
2814 extern int __check_sticky(struct inode *dir, struct inode *inode);
2816 static inline bool execute_ok(struct inode *inode)
2818 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2821 static inline void file_start_write(struct file *file)
2823 if (!S_ISREG(file_inode(file)->i_mode))
2825 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2828 static inline bool file_start_write_trylock(struct file *file)
2830 if (!S_ISREG(file_inode(file)->i_mode))
2832 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2835 static inline void file_end_write(struct file *file)
2837 if (!S_ISREG(file_inode(file)->i_mode))
2839 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2843 * get_write_access() gets write permission for a file.
2844 * put_write_access() releases this write permission.
2845 * This is used for regular files.
2846 * We cannot support write (and maybe mmap read-write shared) accesses and
2847 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2848 * can have the following values:
2849 * 0: no writers, no VM_DENYWRITE mappings
2850 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2851 * > 0: (i_writecount) users are writing to the file.
2853 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2854 * except for the cases where we don't hold i_writecount yet. Then we need to
2855 * use {get,deny}_write_access() - these functions check the sign and refuse
2856 * to do the change if sign is wrong.
2858 static inline int get_write_access(struct inode *inode)
2860 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2862 static inline int deny_write_access(struct file *file)
2864 struct inode *inode = file_inode(file);
2865 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2867 static inline void put_write_access(struct inode * inode)
2869 atomic_dec(&inode->i_writecount);
2871 static inline void allow_write_access(struct file *file)
2874 atomic_inc(&file_inode(file)->i_writecount);
2876 static inline bool inode_is_open_for_write(const struct inode *inode)
2878 return atomic_read(&inode->i_writecount) > 0;
2882 static inline void i_readcount_dec(struct inode *inode)
2884 BUG_ON(!atomic_read(&inode->i_readcount));
2885 atomic_dec(&inode->i_readcount);
2887 static inline void i_readcount_inc(struct inode *inode)
2889 atomic_inc(&inode->i_readcount);
2892 static inline void i_readcount_dec(struct inode *inode)
2896 static inline void i_readcount_inc(struct inode *inode)
2901 extern int do_pipe_flags(int *, int);
2903 #define __kernel_read_file_id(id) \
2904 id(UNKNOWN, unknown) \
2905 id(FIRMWARE, firmware) \
2906 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2907 id(MODULE, kernel-module) \
2908 id(KEXEC_IMAGE, kexec-image) \
2909 id(KEXEC_INITRAMFS, kexec-initramfs) \
2910 id(POLICY, security-policy) \
2911 id(X509_CERTIFICATE, x509-certificate) \
2914 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2915 #define __fid_stringify(dummy, str) #str,
2917 enum kernel_read_file_id {
2918 __kernel_read_file_id(__fid_enumify)
2921 static const char * const kernel_read_file_str[] = {
2922 __kernel_read_file_id(__fid_stringify)
2925 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
2927 if ((unsigned)id >= READING_MAX_ID)
2928 return kernel_read_file_str[READING_UNKNOWN];
2930 return kernel_read_file_str[id];
2933 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2934 enum kernel_read_file_id);
2935 extern int kernel_read_file_from_path(const char *, void **, loff_t *, loff_t,
2936 enum kernel_read_file_id);
2937 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2938 enum kernel_read_file_id);
2939 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
2940 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
2941 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
2942 extern struct file * open_exec(const char *);
2944 /* fs/dcache.c -- generic fs support functions */
2945 extern bool is_subdir(struct dentry *, struct dentry *);
2946 extern bool path_is_under(const struct path *, const struct path *);
2948 extern char *file_path(struct file *, char *, int);
2950 #include <linux/err.h>
2952 /* needed for stackable file system support */
2953 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2955 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2957 extern int inode_init_always(struct super_block *, struct inode *);
2958 extern void inode_init_once(struct inode *);
2959 extern void address_space_init_once(struct address_space *mapping);
2960 extern struct inode * igrab(struct inode *);
2961 extern ino_t iunique(struct super_block *, ino_t);
2962 extern int inode_needs_sync(struct inode *inode);
2963 extern int generic_delete_inode(struct inode *inode);
2964 static inline int generic_drop_inode(struct inode *inode)
2966 return !inode->i_nlink || inode_unhashed(inode);
2969 extern struct inode *ilookup5_nowait(struct super_block *sb,
2970 unsigned long hashval, int (*test)(struct inode *, void *),
2972 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2973 int (*test)(struct inode *, void *), void *data);
2974 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2976 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
2977 int (*test)(struct inode *, void *),
2978 int (*set)(struct inode *, void *),
2980 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2981 extern struct inode * iget_locked(struct super_block *, unsigned long);
2982 extern struct inode *find_inode_nowait(struct super_block *,
2984 int (*match)(struct inode *,
2985 unsigned long, void *),
2987 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2988 extern int insert_inode_locked(struct inode *);
2989 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2990 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2992 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2994 extern void unlock_new_inode(struct inode *);
2995 extern void discard_new_inode(struct inode *);
2996 extern unsigned int get_next_ino(void);
2997 extern void evict_inodes(struct super_block *sb);
2999 extern void __iget(struct inode * inode);
3000 extern void iget_failed(struct inode *);
3001 extern void clear_inode(struct inode *);
3002 extern void __destroy_inode(struct inode *);
3003 extern struct inode *new_inode_pseudo(struct super_block *sb);
3004 extern struct inode *new_inode(struct super_block *sb);
3005 extern void free_inode_nonrcu(struct inode *inode);
3006 extern int should_remove_suid(struct dentry *);
3007 extern int file_remove_privs(struct file *);
3009 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
3010 static inline void insert_inode_hash(struct inode *inode)
3012 __insert_inode_hash(inode, inode->i_ino);
3015 extern void __remove_inode_hash(struct inode *);
3016 static inline void remove_inode_hash(struct inode *inode)
3018 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
3019 __remove_inode_hash(inode);
3022 extern void inode_sb_list_add(struct inode *inode);
3025 extern int bdev_read_only(struct block_device *);
3027 extern int set_blocksize(struct block_device *, int);
3028 extern int sb_set_blocksize(struct super_block *, int);
3029 extern int sb_min_blocksize(struct super_block *, int);
3031 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
3032 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
3033 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
3034 extern int generic_remap_checks(struct file *file_in, loff_t pos_in,
3035 struct file *file_out, loff_t pos_out,
3036 loff_t *count, unsigned int remap_flags);
3037 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
3038 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
3039 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
3040 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
3041 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
3043 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
3045 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
3048 /* fs/block_dev.c */
3049 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
3050 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
3051 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
3053 extern void block_sync_page(struct page *page);
3056 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
3057 struct pipe_inode_info *, size_t, unsigned int);
3058 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
3059 struct file *, loff_t *, size_t, unsigned int);
3060 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
3061 struct file *out, loff_t *, size_t len, unsigned int flags);
3062 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
3063 loff_t *opos, size_t len, unsigned int flags);
3067 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3068 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3069 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
3070 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3071 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3072 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3073 int whence, loff_t maxsize, loff_t eof);
3074 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3075 int whence, loff_t size);
3076 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3077 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3078 extern int generic_file_open(struct inode * inode, struct file * filp);
3079 extern int nonseekable_open(struct inode * inode, struct file * filp);
3080 extern int stream_open(struct inode * inode, struct file * filp);
3083 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3084 loff_t file_offset);
3087 /* need locking between buffered and direct access */
3090 /* filesystem does not support filling holes */
3091 DIO_SKIP_HOLES = 0x02,
3094 void dio_end_io(struct bio *bio);
3095 void dio_warn_stale_pagecache(struct file *filp);
3097 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3098 struct block_device *bdev, struct iov_iter *iter,
3099 get_block_t get_block,
3100 dio_iodone_t end_io, dio_submit_t submit_io,
3103 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3104 struct inode *inode,
3105 struct iov_iter *iter,
3106 get_block_t get_block)
3108 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3109 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3113 void inode_dio_wait(struct inode *inode);
3116 * inode_dio_begin - signal start of a direct I/O requests
3117 * @inode: inode the direct I/O happens on
3119 * This is called once we've finished processing a direct I/O request,
3120 * and is used to wake up callers waiting for direct I/O to be quiesced.
3122 static inline void inode_dio_begin(struct inode *inode)
3124 atomic_inc(&inode->i_dio_count);
3128 * inode_dio_end - signal finish of a direct I/O requests
3129 * @inode: inode the direct I/O happens on
3131 * This is called once we've finished processing a direct I/O request,
3132 * and is used to wake up callers waiting for direct I/O to be quiesced.
3134 static inline void inode_dio_end(struct inode *inode)
3136 if (atomic_dec_and_test(&inode->i_dio_count))
3137 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3140 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3143 extern const struct file_operations generic_ro_fops;
3145 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3147 extern int readlink_copy(char __user *, int, const char *);
3148 extern int page_readlink(struct dentry *, char __user *, int);
3149 extern const char *page_get_link(struct dentry *, struct inode *,
3150 struct delayed_call *);
3151 extern void page_put_link(void *);
3152 extern int __page_symlink(struct inode *inode, const char *symname, int len,
3154 extern int page_symlink(struct inode *inode, const char *symname, int len);
3155 extern const struct inode_operations page_symlink_inode_operations;
3156 extern void kfree_link(void *);
3157 extern void generic_fillattr(struct inode *, struct kstat *);
3158 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3159 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3160 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3161 void inode_add_bytes(struct inode *inode, loff_t bytes);
3162 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3163 void inode_sub_bytes(struct inode *inode, loff_t bytes);
3164 static inline loff_t __inode_get_bytes(struct inode *inode)
3166 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3168 loff_t inode_get_bytes(struct inode *inode);
3169 void inode_set_bytes(struct inode *inode, loff_t bytes);
3170 const char *simple_get_link(struct dentry *, struct inode *,
3171 struct delayed_call *);
3172 extern const struct inode_operations simple_symlink_inode_operations;
3174 extern int iterate_dir(struct file *, struct dir_context *);
3176 extern int vfs_statx(int, const char __user *, int, struct kstat *, u32);
3177 extern int vfs_statx_fd(unsigned int, struct kstat *, u32, unsigned int);
3179 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3181 return vfs_statx(AT_FDCWD, filename, AT_NO_AUTOMOUNT,
3182 stat, STATX_BASIC_STATS);
3184 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3186 return vfs_statx(AT_FDCWD, name, AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT,
3187 stat, STATX_BASIC_STATS);
3189 static inline int vfs_fstatat(int dfd, const char __user *filename,
3190 struct kstat *stat, int flags)
3192 return vfs_statx(dfd, filename, flags | AT_NO_AUTOMOUNT,
3193 stat, STATX_BASIC_STATS);
3195 static inline int vfs_fstat(int fd, struct kstat *stat)
3197 return vfs_statx_fd(fd, stat, STATX_BASIC_STATS, 0);
3201 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3202 extern int vfs_readlink(struct dentry *, char __user *, int);
3204 extern int __generic_block_fiemap(struct inode *inode,
3205 struct fiemap_extent_info *fieinfo,
3206 loff_t start, loff_t len,
3207 get_block_t *get_block);
3208 extern int generic_block_fiemap(struct inode *inode,
3209 struct fiemap_extent_info *fieinfo, u64 start,
3210 u64 len, get_block_t *get_block);
3212 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3213 extern void put_filesystem(struct file_system_type *fs);
3214 extern struct file_system_type *get_fs_type(const char *name);
3215 extern struct super_block *get_super(struct block_device *);
3216 extern struct super_block *get_super_thawed(struct block_device *);
3217 extern struct super_block *get_super_exclusive_thawed(struct block_device *bdev);
3218 extern struct super_block *get_active_super(struct block_device *bdev);
3219 extern void drop_super(struct super_block *sb);
3220 extern void drop_super_exclusive(struct super_block *sb);
3221 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3222 extern void iterate_supers_type(struct file_system_type *,
3223 void (*)(struct super_block *, void *), void *);
3225 extern int dcache_dir_open(struct inode *, struct file *);
3226 extern int dcache_dir_close(struct inode *, struct file *);
3227 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3228 extern int dcache_readdir(struct file *, struct dir_context *);
3229 extern int simple_setattr(struct dentry *, struct iattr *);
3230 extern int simple_getattr(const struct path *, struct kstat *, u32, unsigned int);
3231 extern int simple_statfs(struct dentry *, struct kstatfs *);
3232 extern int simple_open(struct inode *inode, struct file *file);
3233 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3234 extern int simple_unlink(struct inode *, struct dentry *);
3235 extern int simple_rmdir(struct inode *, struct dentry *);
3236 extern int simple_rename(struct inode *, struct dentry *,
3237 struct inode *, struct dentry *, unsigned int);
3238 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3239 extern int noop_set_page_dirty(struct page *page);
3240 extern void noop_invalidatepage(struct page *page, unsigned int offset,
3241 unsigned int length);
3242 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3243 extern int simple_empty(struct dentry *);
3244 extern int simple_readpage(struct file *file, struct page *page);
3245 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3246 loff_t pos, unsigned len, unsigned flags,
3247 struct page **pagep, void **fsdata);
3248 extern int simple_write_end(struct file *file, struct address_space *mapping,
3249 loff_t pos, unsigned len, unsigned copied,
3250 struct page *page, void *fsdata);
3251 extern int always_delete_dentry(const struct dentry *);
3252 extern struct inode *alloc_anon_inode(struct super_block *);
3253 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3254 extern const struct dentry_operations simple_dentry_operations;
3256 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3257 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3258 extern const struct file_operations simple_dir_operations;
3259 extern const struct inode_operations simple_dir_inode_operations;
3260 extern void make_empty_dir_inode(struct inode *inode);
3261 extern bool is_empty_dir_inode(struct inode *inode);
3262 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3263 struct dentry *d_alloc_name(struct dentry *, const char *);
3264 extern int simple_fill_super(struct super_block *, unsigned long,
3265 const struct tree_descr *);
3266 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3267 extern void simple_release_fs(struct vfsmount **mount, int *count);
3269 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3270 loff_t *ppos, const void *from, size_t available);
3271 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3272 const void __user *from, size_t count);
3274 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3275 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3277 extern int generic_check_addressable(unsigned, u64);
3279 #ifdef CONFIG_MIGRATION
3280 extern int buffer_migrate_page(struct address_space *,
3281 struct page *, struct page *,
3283 extern int buffer_migrate_page_norefs(struct address_space *,
3284 struct page *, struct page *,
3287 #define buffer_migrate_page NULL
3288 #define buffer_migrate_page_norefs NULL
3291 extern int setattr_prepare(struct dentry *, struct iattr *);
3292 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3293 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
3295 extern int file_update_time(struct file *file);
3297 static inline bool io_is_direct(struct file *filp)
3299 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
3302 static inline bool vma_is_dax(struct vm_area_struct *vma)
3304 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3307 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3309 struct inode *inode;
3313 if (!vma_is_dax(vma))
3315 inode = file_inode(vma->vm_file);
3316 if (S_ISCHR(inode->i_mode))
3317 return false; /* device-dax */
3321 static inline int iocb_flags(struct file *file)
3324 if (file->f_flags & O_APPEND)
3326 if (io_is_direct(file))
3328 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3330 if (file->f_flags & __O_SYNC)
3335 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3337 if (unlikely(flags & ~RWF_SUPPORTED))
3340 if (flags & RWF_NOWAIT) {
3341 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3343 ki->ki_flags |= IOCB_NOWAIT;
3345 if (flags & RWF_HIPRI)
3346 ki->ki_flags |= IOCB_HIPRI;
3347 if (flags & RWF_DSYNC)
3348 ki->ki_flags |= IOCB_DSYNC;
3349 if (flags & RWF_SYNC)
3350 ki->ki_flags |= (IOCB_DSYNC | IOCB_SYNC);
3351 if (flags & RWF_APPEND)
3352 ki->ki_flags |= IOCB_APPEND;
3356 static inline ino_t parent_ino(struct dentry *dentry)
3361 * Don't strictly need d_lock here? If the parent ino could change
3362 * then surely we'd have a deeper race in the caller?
3364 spin_lock(&dentry->d_lock);
3365 res = dentry->d_parent->d_inode->i_ino;
3366 spin_unlock(&dentry->d_lock);
3370 /* Transaction based IO helpers */
3373 * An argresp is stored in an allocated page and holds the
3374 * size of the argument or response, along with its content
3376 struct simple_transaction_argresp {
3381 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3383 char *simple_transaction_get(struct file *file, const char __user *buf,
3385 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3386 size_t size, loff_t *pos);
3387 int simple_transaction_release(struct inode *inode, struct file *file);
3389 void simple_transaction_set(struct file *file, size_t n);
3392 * simple attribute files
3394 * These attributes behave similar to those in sysfs:
3396 * Writing to an attribute immediately sets a value, an open file can be
3397 * written to multiple times.
3399 * Reading from an attribute creates a buffer from the value that might get
3400 * read with multiple read calls. When the attribute has been read
3401 * completely, no further read calls are possible until the file is opened
3404 * All attributes contain a text representation of a numeric value
3405 * that are accessed with the get() and set() functions.
3407 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3408 static int __fops ## _open(struct inode *inode, struct file *file) \
3410 __simple_attr_check_format(__fmt, 0ull); \
3411 return simple_attr_open(inode, file, __get, __set, __fmt); \
3413 static const struct file_operations __fops = { \
3414 .owner = THIS_MODULE, \
3415 .open = __fops ## _open, \
3416 .release = simple_attr_release, \
3417 .read = simple_attr_read, \
3418 .write = simple_attr_write, \
3419 .llseek = generic_file_llseek, \
3422 static inline __printf(1, 2)
3423 void __simple_attr_check_format(const char *fmt, ...)
3425 /* don't do anything, just let the compiler check the arguments; */
3428 int simple_attr_open(struct inode *inode, struct file *file,
3429 int (*get)(void *, u64 *), int (*set)(void *, u64),
3431 int simple_attr_release(struct inode *inode, struct file *file);
3432 ssize_t simple_attr_read(struct file *file, char __user *buf,
3433 size_t len, loff_t *ppos);
3434 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3435 size_t len, loff_t *ppos);
3438 int proc_nr_files(struct ctl_table *table, int write,
3439 void __user *buffer, size_t *lenp, loff_t *ppos);
3440 int proc_nr_dentry(struct ctl_table *table, int write,
3441 void __user *buffer, size_t *lenp, loff_t *ppos);
3442 int proc_nr_inodes(struct ctl_table *table, int write,
3443 void __user *buffer, size_t *lenp, loff_t *ppos);
3444 int __init get_filesystem_list(char *buf);
3446 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3447 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3449 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3450 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3451 (flag & __FMODE_NONOTIFY)))
3453 static inline bool is_sxid(umode_t mode)
3455 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3458 static inline int check_sticky(struct inode *dir, struct inode *inode)
3460 if (!(dir->i_mode & S_ISVTX))
3463 return __check_sticky(dir, inode);
3466 static inline void inode_has_no_xattr(struct inode *inode)
3468 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3469 inode->i_flags |= S_NOSEC;
3472 static inline bool is_root_inode(struct inode *inode)
3474 return inode == inode->i_sb->s_root->d_inode;
3477 static inline bool dir_emit(struct dir_context *ctx,
3478 const char *name, int namelen,
3479 u64 ino, unsigned type)
3481 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3483 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3485 return ctx->actor(ctx, ".", 1, ctx->pos,
3486 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3488 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3490 return ctx->actor(ctx, "..", 2, ctx->pos,
3491 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3493 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3495 if (ctx->pos == 0) {
3496 if (!dir_emit_dot(file, ctx))
3500 if (ctx->pos == 1) {
3501 if (!dir_emit_dotdot(file, ctx))
3507 static inline bool dir_relax(struct inode *inode)
3509 inode_unlock(inode);
3511 return !IS_DEADDIR(inode);
3514 static inline bool dir_relax_shared(struct inode *inode)
3516 inode_unlock_shared(inode);
3517 inode_lock_shared(inode);
3518 return !IS_DEADDIR(inode);
3521 extern bool path_noexec(const struct path *path);
3522 extern void inode_nohighmem(struct inode *inode);
3525 extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len,
3528 #if defined(CONFIG_IO_URING)
3529 extern struct sock *io_uring_get_socket(struct file *file);
3531 static inline struct sock *io_uring_get_socket(struct file *file)
3537 #endif /* _LINUX_FS_H */