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 */
698 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
699 void (*free_inode)(struct inode *);
701 struct file_lock_context *i_flctx;
702 struct address_space i_data;
703 struct list_head i_devices;
705 struct pipe_inode_info *i_pipe;
706 struct block_device *i_bdev;
714 #ifdef CONFIG_FSNOTIFY
715 __u32 i_fsnotify_mask; /* all events this inode cares about */
716 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
719 #ifdef CONFIG_FS_ENCRYPTION
720 struct fscrypt_info *i_crypt_info;
723 void *i_private; /* fs or device private pointer */
724 } __randomize_layout;
726 static inline unsigned int i_blocksize(const struct inode *node)
728 return (1 << node->i_blkbits);
731 static inline int inode_unhashed(struct inode *inode)
733 return hlist_unhashed(&inode->i_hash);
737 * __mark_inode_dirty expects inodes to be hashed. Since we don't
738 * want special inodes in the fileset inode space, we make them
739 * appear hashed, but do not put on any lists. hlist_del()
740 * will work fine and require no locking.
742 static inline void inode_fake_hash(struct inode *inode)
744 hlist_add_fake(&inode->i_hash);
748 * inode->i_mutex nesting subclasses for the lock validator:
750 * 0: the object of the current VFS operation
754 * 4: second non-directory
755 * 5: second parent (when locking independent directories in rename)
757 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
758 * non-directories at once.
760 * The locking order between these classes is
761 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
763 enum inode_i_mutex_lock_class
773 static inline void inode_lock(struct inode *inode)
775 down_write(&inode->i_rwsem);
778 static inline void inode_unlock(struct inode *inode)
780 up_write(&inode->i_rwsem);
783 static inline void inode_lock_shared(struct inode *inode)
785 down_read(&inode->i_rwsem);
788 static inline void inode_unlock_shared(struct inode *inode)
790 up_read(&inode->i_rwsem);
793 static inline int inode_trylock(struct inode *inode)
795 return down_write_trylock(&inode->i_rwsem);
798 static inline int inode_trylock_shared(struct inode *inode)
800 return down_read_trylock(&inode->i_rwsem);
803 static inline int inode_is_locked(struct inode *inode)
805 return rwsem_is_locked(&inode->i_rwsem);
808 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
810 down_write_nested(&inode->i_rwsem, subclass);
813 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
815 down_read_nested(&inode->i_rwsem, subclass);
818 void lock_two_nondirectories(struct inode *, struct inode*);
819 void unlock_two_nondirectories(struct inode *, struct inode*);
822 * NOTE: in a 32bit arch with a preemptable kernel and
823 * an UP compile the i_size_read/write must be atomic
824 * with respect to the local cpu (unlike with preempt disabled),
825 * but they don't need to be atomic with respect to other cpus like in
826 * true SMP (so they need either to either locally disable irq around
827 * the read or for example on x86 they can be still implemented as a
828 * cmpxchg8b without the need of the lock prefix). For SMP compiles
829 * and 64bit archs it makes no difference if preempt is enabled or not.
831 static inline loff_t i_size_read(const struct inode *inode)
833 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
838 seq = read_seqcount_begin(&inode->i_size_seqcount);
839 i_size = inode->i_size;
840 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
842 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
846 i_size = inode->i_size;
850 return inode->i_size;
855 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
856 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
857 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
859 static inline void i_size_write(struct inode *inode, loff_t i_size)
861 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
863 write_seqcount_begin(&inode->i_size_seqcount);
864 inode->i_size = i_size;
865 write_seqcount_end(&inode->i_size_seqcount);
867 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
869 inode->i_size = i_size;
872 inode->i_size = i_size;
876 static inline unsigned iminor(const struct inode *inode)
878 return MINOR(inode->i_rdev);
881 static inline unsigned imajor(const struct inode *inode)
883 return MAJOR(inode->i_rdev);
886 extern struct block_device *I_BDEV(struct inode *inode);
889 rwlock_t lock; /* protects pid, uid, euid fields */
890 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
891 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
892 kuid_t uid, euid; /* uid/euid of process setting the owner */
893 int signum; /* posix.1b rt signal to be delivered on IO */
897 * Track a single file's readahead state
899 struct file_ra_state {
900 pgoff_t start; /* where readahead started */
901 unsigned int size; /* # of readahead pages */
902 unsigned int async_size; /* do asynchronous readahead when
903 there are only # of pages ahead */
905 unsigned int ra_pages; /* Maximum readahead window */
906 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
907 loff_t prev_pos; /* Cache last read() position */
911 * Check if @index falls in the readahead windows.
913 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
915 return (index >= ra->start &&
916 index < ra->start + ra->size);
921 struct llist_node fu_llist;
922 struct rcu_head fu_rcuhead;
925 struct inode *f_inode; /* cached value */
926 const struct file_operations *f_op;
929 * Protects f_ep_links, f_flags.
930 * Must not be taken from IRQ context.
933 enum rw_hint f_write_hint;
934 atomic_long_t f_count;
935 unsigned int f_flags;
937 struct mutex f_pos_lock;
939 struct fown_struct f_owner;
940 const struct cred *f_cred;
941 struct file_ra_state f_ra;
944 #ifdef CONFIG_SECURITY
947 /* needed for tty driver, and maybe others */
951 /* Used by fs/eventpoll.c to link all the hooks to this file */
952 struct list_head f_ep_links;
953 struct list_head f_tfile_llink;
954 #endif /* #ifdef CONFIG_EPOLL */
955 struct address_space *f_mapping;
958 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
963 /* file identifier */
964 unsigned char f_handle[0];
967 static inline struct file *get_file(struct file *f)
969 atomic_long_inc(&f->f_count);
972 #define get_file_rcu_many(x, cnt) \
973 atomic_long_add_unless(&(x)->f_count, (cnt), 0)
974 #define get_file_rcu(x) get_file_rcu_many((x), 1)
975 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
976 #define file_count(x) atomic_long_read(&(x)->f_count)
978 #define MAX_NON_LFS ((1UL<<31) - 1)
980 /* Page cache limit. The filesystems should put that into their s_maxbytes
981 limits, otherwise bad things can happen in VM. */
982 #if BITS_PER_LONG==32
983 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
984 #elif BITS_PER_LONG==64
985 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
990 #define FL_DELEG 4 /* NFSv4 delegation */
991 #define FL_ACCESS 8 /* not trying to lock, just looking */
992 #define FL_EXISTS 16 /* when unlocking, test for existence */
993 #define FL_LEASE 32 /* lease held on this file */
994 #define FL_CLOSE 64 /* unlock on close */
995 #define FL_SLEEP 128 /* A blocking lock */
996 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
997 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
998 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
999 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
1001 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
1004 * Special return value from posix_lock_file() and vfs_lock_file() for
1005 * asynchronous locking.
1007 #define FILE_LOCK_DEFERRED 1
1009 /* legacy typedef, should eventually be removed */
1010 typedef void *fl_owner_t;
1014 struct file_lock_operations {
1015 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1016 void (*fl_release_private)(struct file_lock *);
1019 struct lock_manager_operations {
1020 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
1021 unsigned long (*lm_owner_key)(struct file_lock *);
1022 fl_owner_t (*lm_get_owner)(fl_owner_t);
1023 void (*lm_put_owner)(fl_owner_t);
1024 void (*lm_notify)(struct file_lock *); /* unblock callback */
1025 int (*lm_grant)(struct file_lock *, int);
1026 bool (*lm_break)(struct file_lock *);
1027 int (*lm_change)(struct file_lock *, int, struct list_head *);
1028 void (*lm_setup)(struct file_lock *, void **);
1031 struct lock_manager {
1032 struct list_head list;
1034 * NFSv4 and up also want opens blocked during the grace period;
1041 void locks_start_grace(struct net *, struct lock_manager *);
1042 void locks_end_grace(struct lock_manager *);
1043 bool locks_in_grace(struct net *);
1044 bool opens_in_grace(struct net *);
1046 /* that will die - we need it for nfs_lock_info */
1047 #include <linux/nfs_fs_i.h>
1050 * struct file_lock represents a generic "file lock". It's used to represent
1051 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1052 * note that the same struct is used to represent both a request for a lock and
1053 * the lock itself, but the same object is never used for both.
1055 * FIXME: should we create a separate "struct lock_request" to help distinguish
1058 * The varous i_flctx lists are ordered by:
1061 * 2) lock range start
1064 * Obviously, the last two criteria only matter for POSIX locks.
1067 struct file_lock *fl_blocker; /* The lock, that is blocking us */
1068 struct list_head fl_list; /* link into file_lock_context */
1069 struct hlist_node fl_link; /* node in global lists */
1070 struct list_head fl_blocked_requests; /* list of requests with
1071 * ->fl_blocker pointing here
1073 struct list_head fl_blocked_member; /* node in
1074 * ->fl_blocker->fl_blocked_requests
1076 fl_owner_t fl_owner;
1077 unsigned int fl_flags;
1078 unsigned char fl_type;
1079 unsigned int fl_pid;
1080 int fl_link_cpu; /* what cpu's list is this on? */
1081 wait_queue_head_t fl_wait;
1082 struct file *fl_file;
1086 struct fasync_struct * fl_fasync; /* for lease break notifications */
1087 /* for lease breaks: */
1088 unsigned long fl_break_time;
1089 unsigned long fl_downgrade_time;
1091 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1092 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1094 struct nfs_lock_info nfs_fl;
1095 struct nfs4_lock_info nfs4_fl;
1097 struct list_head link; /* link in AFS vnode's pending_locks list */
1098 int state; /* state of grant or error if -ve */
1101 } __randomize_layout;
1103 struct file_lock_context {
1104 spinlock_t flc_lock;
1105 struct list_head flc_flock;
1106 struct list_head flc_posix;
1107 struct list_head flc_lease;
1110 /* The following constant reflects the upper bound of the file/locking space */
1112 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1113 #define OFFSET_MAX INT_LIMIT(loff_t)
1114 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1117 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1119 #define locks_inode(f) file_inode(f)
1121 #ifdef CONFIG_FILE_LOCKING
1122 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1123 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1126 #if BITS_PER_LONG == 32
1127 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1128 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1132 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1133 extern int fcntl_getlease(struct file *filp);
1136 void locks_free_lock_context(struct inode *inode);
1137 void locks_free_lock(struct file_lock *fl);
1138 extern void locks_init_lock(struct file_lock *);
1139 extern struct file_lock * locks_alloc_lock(void);
1140 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1141 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1142 extern void locks_remove_posix(struct file *, fl_owner_t);
1143 extern void locks_remove_file(struct file *);
1144 extern void locks_release_private(struct file_lock *);
1145 extern void posix_test_lock(struct file *, struct file_lock *);
1146 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1147 extern int locks_delete_block(struct file_lock *);
1148 extern int vfs_test_lock(struct file *, struct file_lock *);
1149 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1150 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1151 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1152 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1153 extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1154 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1155 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1156 extern int lease_modify(struct file_lock *, int, struct list_head *);
1157 struct files_struct;
1158 extern void show_fd_locks(struct seq_file *f,
1159 struct file *filp, struct files_struct *files);
1160 #else /* !CONFIG_FILE_LOCKING */
1161 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1162 struct flock __user *user)
1167 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1168 unsigned int cmd, struct flock __user *user)
1173 #if BITS_PER_LONG == 32
1174 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1175 struct flock64 __user *user)
1180 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1181 unsigned int cmd, struct flock64 __user *user)
1186 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1191 static inline int fcntl_getlease(struct file *filp)
1197 locks_free_lock_context(struct inode *inode)
1201 static inline void locks_init_lock(struct file_lock *fl)
1206 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1211 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1216 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1221 static inline void locks_remove_file(struct file *filp)
1226 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1231 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1232 struct file_lock *conflock)
1237 static inline int locks_delete_block(struct file_lock *waiter)
1242 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1247 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1248 struct file_lock *fl, struct file_lock *conf)
1253 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1258 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1263 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1268 static inline void lease_get_mtime(struct inode *inode,
1269 struct timespec64 *time)
1274 static inline int generic_setlease(struct file *filp, long arg,
1275 struct file_lock **flp, void **priv)
1280 static inline int vfs_setlease(struct file *filp, long arg,
1281 struct file_lock **lease, void **priv)
1286 static inline int lease_modify(struct file_lock *fl, int arg,
1287 struct list_head *dispose)
1292 struct files_struct;
1293 static inline void show_fd_locks(struct seq_file *f,
1294 struct file *filp, struct files_struct *files) {}
1295 #endif /* !CONFIG_FILE_LOCKING */
1297 static inline struct inode *file_inode(const struct file *f)
1302 static inline struct dentry *file_dentry(const struct file *file)
1304 return d_real(file->f_path.dentry, file_inode(file));
1307 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1309 return locks_lock_inode_wait(locks_inode(filp), fl);
1312 struct fasync_struct {
1316 struct fasync_struct *fa_next; /* singly linked list */
1317 struct file *fa_file;
1318 struct rcu_head fa_rcu;
1321 #define FASYNC_MAGIC 0x4601
1323 /* SMP safe fasync helpers: */
1324 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1325 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1326 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1327 extern struct fasync_struct *fasync_alloc(void);
1328 extern void fasync_free(struct fasync_struct *);
1330 /* can be called from interrupts */
1331 extern void kill_fasync(struct fasync_struct **, int, int);
1333 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1334 extern int f_setown(struct file *filp, unsigned long arg, int force);
1335 extern void f_delown(struct file *filp);
1336 extern pid_t f_getown(struct file *filp);
1337 extern int send_sigurg(struct fown_struct *fown);
1340 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1341 * represented in both.
1343 #define SB_RDONLY 1 /* Mount read-only */
1344 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1345 #define SB_NODEV 4 /* Disallow access to device special files */
1346 #define SB_NOEXEC 8 /* Disallow program execution */
1347 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1348 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1349 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1350 #define SB_NOATIME 1024 /* Do not update access times. */
1351 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1352 #define SB_SILENT 32768
1353 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1354 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1355 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1356 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1358 /* These sb flags are internal to the kernel */
1359 #define SB_SUBMOUNT (1<<26)
1360 #define SB_FORCE (1<<27)
1361 #define SB_NOSEC (1<<28)
1362 #define SB_BORN (1<<29)
1363 #define SB_ACTIVE (1<<30)
1364 #define SB_NOUSER (1<<31)
1370 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1371 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1372 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1373 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1374 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1377 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1378 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1379 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1380 #define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */
1382 /* sb->s_iflags to limit user namespace mounts */
1383 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1384 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1385 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1387 /* Possible states of 'frozen' field */
1389 SB_UNFROZEN = 0, /* FS is unfrozen */
1390 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1391 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1392 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1393 * internal threads if needed) */
1394 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1397 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1400 int frozen; /* Is sb frozen? */
1401 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1402 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1405 struct super_block {
1406 struct list_head s_list; /* Keep this first */
1407 dev_t s_dev; /* search index; _not_ kdev_t */
1408 unsigned char s_blocksize_bits;
1409 unsigned long s_blocksize;
1410 loff_t s_maxbytes; /* Max file size */
1411 struct file_system_type *s_type;
1412 const struct super_operations *s_op;
1413 const struct dquot_operations *dq_op;
1414 const struct quotactl_ops *s_qcop;
1415 const struct export_operations *s_export_op;
1416 unsigned long s_flags;
1417 unsigned long s_iflags; /* internal SB_I_* flags */
1418 unsigned long s_magic;
1419 struct dentry *s_root;
1420 struct rw_semaphore s_umount;
1423 #ifdef CONFIG_SECURITY
1426 const struct xattr_handler **s_xattr;
1427 #ifdef CONFIG_FS_ENCRYPTION
1428 const struct fscrypt_operations *s_cop;
1430 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1431 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1432 struct block_device *s_bdev;
1433 struct backing_dev_info *s_bdi;
1434 struct mtd_info *s_mtd;
1435 struct hlist_node s_instances;
1436 unsigned int s_quota_types; /* Bitmask of supported quota types */
1437 struct quota_info s_dquot; /* Diskquota specific options */
1439 struct sb_writers s_writers;
1442 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1443 * s_fsnotify_marks together for cache efficiency. They are frequently
1444 * accessed and rarely modified.
1446 void *s_fs_info; /* Filesystem private info */
1448 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1450 #ifdef CONFIG_FSNOTIFY
1451 __u32 s_fsnotify_mask;
1452 struct fsnotify_mark_connector __rcu *s_fsnotify_marks;
1455 char s_id[32]; /* Informational name */
1456 uuid_t s_uuid; /* UUID */
1458 unsigned int s_max_links;
1462 * The next field is for VFS *only*. No filesystems have any business
1463 * even looking at it. You had been warned.
1465 struct mutex s_vfs_rename_mutex; /* Kludge */
1468 * Filesystem subtype. If non-empty the filesystem type field
1469 * in /proc/mounts will be "type.subtype"
1471 const char *s_subtype;
1473 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1476 * Saved pool identifier for cleancache (-1 means none)
1478 int cleancache_poolid;
1480 struct shrinker s_shrink; /* per-sb shrinker handle */
1482 /* Number of inodes with nlink == 0 but still referenced */
1483 atomic_long_t s_remove_count;
1485 /* Pending fsnotify inode refs */
1486 atomic_long_t s_fsnotify_inode_refs;
1488 /* Being remounted read-only */
1489 int s_readonly_remount;
1491 /* AIO completions deferred from interrupt context */
1492 struct workqueue_struct *s_dio_done_wq;
1493 struct hlist_head s_pins;
1496 * Owning user namespace and default context in which to
1497 * interpret filesystem uids, gids, quotas, device nodes,
1498 * xattrs and security labels.
1500 struct user_namespace *s_user_ns;
1503 * The list_lru structure is essentially just a pointer to a table
1504 * of per-node lru lists, each of which has its own spinlock.
1505 * There is no need to put them into separate cachelines.
1507 struct list_lru s_dentry_lru;
1508 struct list_lru s_inode_lru;
1509 struct rcu_head rcu;
1510 struct work_struct destroy_work;
1512 struct mutex s_sync_lock; /* sync serialisation lock */
1515 * Indicates how deep in a filesystem stack this SB is
1519 /* s_inode_list_lock protects s_inodes */
1520 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1521 struct list_head s_inodes; /* all inodes */
1523 spinlock_t s_inode_wblist_lock;
1524 struct list_head s_inodes_wb; /* writeback inodes */
1525 } __randomize_layout;
1527 /* Helper functions so that in most cases filesystems will
1528 * not need to deal directly with kuid_t and kgid_t and can
1529 * instead deal with the raw numeric values that are stored
1530 * in the filesystem.
1532 static inline uid_t i_uid_read(const struct inode *inode)
1534 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1537 static inline gid_t i_gid_read(const struct inode *inode)
1539 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1542 static inline void i_uid_write(struct inode *inode, uid_t uid)
1544 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1547 static inline void i_gid_write(struct inode *inode, gid_t gid)
1549 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1552 extern struct timespec64 timespec64_trunc(struct timespec64 t, unsigned gran);
1553 extern struct timespec64 current_time(struct inode *inode);
1556 * Snapshotting support.
1559 void __sb_end_write(struct super_block *sb, int level);
1560 int __sb_start_write(struct super_block *sb, int level, bool wait);
1562 #define __sb_writers_acquired(sb, lev) \
1563 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1564 #define __sb_writers_release(sb, lev) \
1565 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1568 * sb_end_write - drop write access to a superblock
1569 * @sb: the super we wrote to
1571 * Decrement number of writers to the filesystem. Wake up possible waiters
1572 * wanting to freeze the filesystem.
1574 static inline void sb_end_write(struct super_block *sb)
1576 __sb_end_write(sb, SB_FREEZE_WRITE);
1580 * sb_end_pagefault - drop write access to a superblock from a page fault
1581 * @sb: the super we wrote to
1583 * Decrement number of processes handling write page fault to the filesystem.
1584 * Wake up possible waiters wanting to freeze the filesystem.
1586 static inline void sb_end_pagefault(struct super_block *sb)
1588 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1592 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1593 * @sb: the super we wrote to
1595 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1596 * waiters wanting to freeze the filesystem.
1598 static inline void sb_end_intwrite(struct super_block *sb)
1600 __sb_end_write(sb, SB_FREEZE_FS);
1604 * sb_start_write - get write access to a superblock
1605 * @sb: the super we write to
1607 * When a process wants to write data or metadata to a file system (i.e. dirty
1608 * a page or an inode), it should embed the operation in a sb_start_write() -
1609 * sb_end_write() pair to get exclusion against file system freezing. This
1610 * function increments number of writers preventing freezing. If the file
1611 * system is already frozen, the function waits until the file system is
1614 * Since freeze protection behaves as a lock, users have to preserve
1615 * ordering of freeze protection and other filesystem locks. Generally,
1616 * freeze protection should be the outermost lock. In particular, we have:
1619 * -> i_mutex (write path, truncate, directory ops, ...)
1620 * -> s_umount (freeze_super, thaw_super)
1622 static inline void sb_start_write(struct super_block *sb)
1624 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1627 static inline int sb_start_write_trylock(struct super_block *sb)
1629 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1633 * sb_start_pagefault - get write access to a superblock from a page fault
1634 * @sb: the super we write to
1636 * When a process starts handling write page fault, it should embed the
1637 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1638 * exclusion against file system freezing. This is needed since the page fault
1639 * is going to dirty a page. This function increments number of running page
1640 * faults preventing freezing. If the file system is already frozen, the
1641 * function waits until the file system is thawed.
1643 * Since page fault freeze protection behaves as a lock, users have to preserve
1644 * ordering of freeze protection and other filesystem locks. It is advised to
1645 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1646 * handling code implies lock dependency:
1649 * -> sb_start_pagefault
1651 static inline void sb_start_pagefault(struct super_block *sb)
1653 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1657 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1658 * @sb: the super we write to
1660 * This is the third level of protection against filesystem freezing. It is
1661 * free for use by a filesystem. The only requirement is that it must rank
1662 * below sb_start_pagefault.
1664 * For example filesystem can call sb_start_intwrite() when starting a
1665 * transaction which somewhat eases handling of freezing for internal sources
1666 * of filesystem changes (internal fs threads, discarding preallocation on file
1669 static inline void sb_start_intwrite(struct super_block *sb)
1671 __sb_start_write(sb, SB_FREEZE_FS, true);
1674 static inline int sb_start_intwrite_trylock(struct super_block *sb)
1676 return __sb_start_write(sb, SB_FREEZE_FS, false);
1680 extern bool inode_owner_or_capable(const struct inode *inode);
1683 * VFS helper functions..
1685 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1686 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1687 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1688 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1689 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1690 extern int vfs_rmdir(struct inode *, struct dentry *);
1691 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1692 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1693 extern int vfs_whiteout(struct inode *, struct dentry *);
1695 extern struct dentry *vfs_tmpfile(struct dentry *dentry, umode_t mode,
1698 int vfs_mkobj(struct dentry *, umode_t,
1699 int (*f)(struct dentry *, umode_t, void *),
1702 extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1705 * VFS file helper functions.
1707 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1709 extern bool may_open_dev(const struct path *path);
1711 * VFS FS_IOC_FIEMAP helper definitions.
1713 struct fiemap_extent_info {
1714 unsigned int fi_flags; /* Flags as passed from user */
1715 unsigned int fi_extents_mapped; /* Number of mapped extents */
1716 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1717 struct fiemap_extent __user *fi_extents_start; /* Start of
1718 fiemap_extent array */
1720 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1721 u64 phys, u64 len, u32 flags);
1722 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1725 * This is the "filldir" function type, used by readdir() to let
1726 * the kernel specify what kind of dirent layout it wants to have.
1727 * This allows the kernel to read directories into kernel space or
1728 * to have different dirent layouts depending on the binary type.
1731 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1734 struct dir_context {
1739 struct block_device_operations;
1741 /* These macros are for out of kernel modules to test that
1742 * the kernel supports the unlocked_ioctl and compat_ioctl
1743 * fields in struct file_operations. */
1744 #define HAVE_COMPAT_IOCTL 1
1745 #define HAVE_UNLOCKED_IOCTL 1
1748 * These flags let !MMU mmap() govern direct device mapping vs immediate
1749 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1751 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1752 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1753 * NOMMU_MAP_READ: Can be mapped for reading
1754 * NOMMU_MAP_WRITE: Can be mapped for writing
1755 * NOMMU_MAP_EXEC: Can be mapped for execution
1757 #define NOMMU_MAP_COPY 0x00000001
1758 #define NOMMU_MAP_DIRECT 0x00000008
1759 #define NOMMU_MAP_READ VM_MAYREAD
1760 #define NOMMU_MAP_WRITE VM_MAYWRITE
1761 #define NOMMU_MAP_EXEC VM_MAYEXEC
1763 #define NOMMU_VMFLAGS \
1764 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1767 * These flags control the behavior of the remap_file_range function pointer.
1768 * If it is called with len == 0 that means "remap to end of source file".
1769 * See Documentation/filesystems/vfs.txt for more details about this call.
1771 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
1772 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
1774 #define REMAP_FILE_DEDUP (1 << 0)
1775 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
1778 * These flags signal that the caller is ok with altering various aspects of
1779 * the behavior of the remap operation. The changes must be made by the
1780 * implementation; the vfs remap helper functions can take advantage of them.
1781 * Flags in this category exist to preserve the quirky behavior of the hoisted
1782 * btrfs clone/dedupe ioctls.
1784 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
1788 struct file_operations {
1789 struct module *owner;
1790 loff_t (*llseek) (struct file *, loff_t, int);
1791 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1792 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1793 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1794 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1795 int (*iopoll)(struct kiocb *kiocb, bool spin);
1796 int (*iterate) (struct file *, struct dir_context *);
1797 int (*iterate_shared) (struct file *, struct dir_context *);
1798 __poll_t (*poll) (struct file *, struct poll_table_struct *);
1799 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1800 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1801 int (*mmap) (struct file *, struct vm_area_struct *);
1802 unsigned long mmap_supported_flags;
1803 int (*open) (struct inode *, struct file *);
1804 int (*flush) (struct file *, fl_owner_t id);
1805 int (*release) (struct inode *, struct file *);
1806 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1807 int (*fasync) (int, struct file *, int);
1808 int (*lock) (struct file *, int, struct file_lock *);
1809 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1810 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1811 int (*check_flags)(int);
1812 int (*flock) (struct file *, int, struct file_lock *);
1813 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1814 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1815 int (*setlease)(struct file *, long, struct file_lock **, void **);
1816 long (*fallocate)(struct file *file, int mode, loff_t offset,
1818 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1820 unsigned (*mmap_capabilities)(struct file *);
1822 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1823 loff_t, size_t, unsigned int);
1824 loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
1825 struct file *file_out, loff_t pos_out,
1826 loff_t len, unsigned int remap_flags);
1827 int (*fadvise)(struct file *, loff_t, loff_t, int);
1828 } __randomize_layout;
1830 struct inode_operations {
1831 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1832 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1833 int (*permission) (struct inode *, int);
1834 struct posix_acl * (*get_acl)(struct inode *, int);
1836 int (*readlink) (struct dentry *, char __user *,int);
1838 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1839 int (*link) (struct dentry *,struct inode *,struct dentry *);
1840 int (*unlink) (struct inode *,struct dentry *);
1841 int (*symlink) (struct inode *,struct dentry *,const char *);
1842 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1843 int (*rmdir) (struct inode *,struct dentry *);
1844 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1845 int (*rename) (struct inode *, struct dentry *,
1846 struct inode *, struct dentry *, unsigned int);
1847 int (*setattr) (struct dentry *, struct iattr *);
1848 int (*getattr) (const struct path *, struct kstat *, u32, unsigned int);
1849 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1850 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1852 int (*update_time)(struct inode *, struct timespec64 *, int);
1853 int (*atomic_open)(struct inode *, struct dentry *,
1854 struct file *, unsigned open_flag,
1855 umode_t create_mode);
1856 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1857 int (*set_acl)(struct inode *, struct posix_acl *, int);
1858 } ____cacheline_aligned;
1860 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
1861 struct iov_iter *iter)
1863 return file->f_op->read_iter(kio, iter);
1866 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
1867 struct iov_iter *iter)
1869 return file->f_op->write_iter(kio, iter);
1872 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
1874 return file->f_op->mmap(file, vma);
1877 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1878 unsigned long nr_segs, unsigned long fast_segs,
1879 struct iovec *fast_pointer,
1880 struct iovec **ret_pointer);
1882 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1883 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1884 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1885 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1886 unsigned long, loff_t *, rwf_t);
1887 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1888 loff_t, size_t, unsigned int);
1889 extern int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
1890 struct file *file_out, loff_t pos_out,
1892 unsigned int remap_flags);
1893 extern loff_t do_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 loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1897 struct file *file_out, loff_t pos_out,
1898 loff_t len, unsigned int remap_flags);
1899 extern int vfs_dedupe_file_range(struct file *file,
1900 struct file_dedupe_range *same);
1901 extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
1902 struct file *dst_file, loff_t dst_pos,
1903 loff_t len, unsigned int remap_flags);
1906 struct super_operations {
1907 struct inode *(*alloc_inode)(struct super_block *sb);
1908 void (*destroy_inode)(struct inode *);
1909 void (*free_inode)(struct inode *);
1911 void (*dirty_inode) (struct inode *, int flags);
1912 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1913 int (*drop_inode) (struct inode *);
1914 void (*evict_inode) (struct inode *);
1915 void (*put_super) (struct super_block *);
1916 int (*sync_fs)(struct super_block *sb, int wait);
1917 int (*freeze_super) (struct super_block *);
1918 int (*freeze_fs) (struct super_block *);
1919 int (*thaw_super) (struct super_block *);
1920 int (*unfreeze_fs) (struct super_block *);
1921 int (*statfs) (struct dentry *, struct kstatfs *);
1922 int (*remount_fs) (struct super_block *, int *, char *);
1923 void (*umount_begin) (struct super_block *);
1925 int (*show_options)(struct seq_file *, struct dentry *);
1926 int (*show_devname)(struct seq_file *, struct dentry *);
1927 int (*show_path)(struct seq_file *, struct dentry *);
1928 int (*show_stats)(struct seq_file *, struct dentry *);
1930 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1931 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1932 struct dquot **(*get_dquots)(struct inode *);
1934 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1935 long (*nr_cached_objects)(struct super_block *,
1936 struct shrink_control *);
1937 long (*free_cached_objects)(struct super_block *,
1938 struct shrink_control *);
1942 * Inode flags - they have no relation to superblock flags now
1944 #define S_SYNC 1 /* Writes are synced at once */
1945 #define S_NOATIME 2 /* Do not update access times */
1946 #define S_APPEND 4 /* Append-only file */
1947 #define S_IMMUTABLE 8 /* Immutable file */
1948 #define S_DEAD 16 /* removed, but still open directory */
1949 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1950 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1951 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1952 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1953 #define S_PRIVATE 512 /* Inode is fs-internal */
1954 #define S_IMA 1024 /* Inode has an associated IMA struct */
1955 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1956 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1957 #ifdef CONFIG_FS_DAX
1958 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1960 #define S_DAX 0 /* Make all the DAX code disappear */
1962 #define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */
1965 * Note that nosuid etc flags are inode-specific: setting some file-system
1966 * flags just means all the inodes inherit those flags by default. It might be
1967 * possible to override it selectively if you really wanted to with some
1968 * ioctl() that is not currently implemented.
1970 * Exception: SB_RDONLY is always applied to the entire file system.
1972 * Unfortunately, it is possible to change a filesystems flags with it mounted
1973 * with files in use. This means that all of the inodes will not have their
1974 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1975 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1977 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1979 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
1980 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
1981 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
1982 ((inode)->i_flags & S_SYNC))
1983 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
1984 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1985 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
1986 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
1987 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
1989 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1990 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1991 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1992 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
1994 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1995 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1996 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1997 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1998 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1999 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2000 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2001 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2002 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2004 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2005 (inode)->i_rdev == WHITEOUT_DEV)
2007 static inline bool HAS_UNMAPPED_ID(struct inode *inode)
2009 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
2012 static inline enum rw_hint file_write_hint(struct file *file)
2014 if (file->f_write_hint != WRITE_LIFE_NOT_SET)
2015 return file->f_write_hint;
2017 return file_inode(file)->i_write_hint;
2020 static inline int iocb_flags(struct file *file);
2022 static inline u16 ki_hint_validate(enum rw_hint hint)
2024 typeof(((struct kiocb *)0)->ki_hint) max_hint = -1;
2026 if (hint <= max_hint)
2031 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
2033 *kiocb = (struct kiocb) {
2035 .ki_flags = iocb_flags(filp),
2036 .ki_hint = ki_hint_validate(file_write_hint(filp)),
2037 .ki_ioprio = get_current_ioprio(),
2042 * Inode state bits. Protected by inode->i_lock
2044 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
2045 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
2047 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2048 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2049 * various stages of removing an inode.
2051 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2053 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2054 * fdatasync(). i_atime is the usual cause.
2055 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2056 * these changes separately from I_DIRTY_SYNC so that we
2057 * don't have to write inode on fdatasync() when only
2058 * mtime has changed in it.
2059 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2060 * I_NEW Serves as both a mutex and completion notification.
2061 * New inodes set I_NEW. If two processes both create
2062 * the same inode, one of them will release its inode and
2063 * wait for I_NEW to be released before returning.
2064 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2065 * also cause waiting on I_NEW, without I_NEW actually
2066 * being set. find_inode() uses this to prevent returning
2067 * nearly-dead inodes.
2068 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2069 * is zero. I_FREEING must be set when I_WILL_FREE is
2071 * I_FREEING Set when inode is about to be freed but still has dirty
2072 * pages or buffers attached or the inode itself is still
2074 * I_CLEAR Added by clear_inode(). In this state the inode is
2075 * clean and can be destroyed. Inode keeps I_FREEING.
2077 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2078 * prohibited for many purposes. iget() must wait for
2079 * the inode to be completely released, then create it
2080 * anew. Other functions will just ignore such inodes,
2081 * if appropriate. I_NEW is used for waiting.
2083 * I_SYNC Writeback of inode is running. The bit is set during
2084 * data writeback, and cleared with a wakeup on the bit
2085 * address once it is done. The bit is also used to pin
2086 * the inode in memory for flusher thread.
2088 * I_REFERENCED Marks the inode as recently references on the LRU list.
2090 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2092 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2093 * synchronize competing switching instances and to tell
2094 * wb stat updates to grab the i_pages lock. See
2095 * inode_switch_wbs_work_fn() for details.
2097 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2098 * and work dirs among overlayfs mounts.
2100 * I_CREATING New object's inode in the middle of setting up.
2102 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2104 #define I_DIRTY_SYNC (1 << 0)
2105 #define I_DIRTY_DATASYNC (1 << 1)
2106 #define I_DIRTY_PAGES (1 << 2)
2108 #define I_NEW (1 << __I_NEW)
2109 #define I_WILL_FREE (1 << 4)
2110 #define I_FREEING (1 << 5)
2111 #define I_CLEAR (1 << 6)
2113 #define I_SYNC (1 << __I_SYNC)
2114 #define I_REFERENCED (1 << 8)
2115 #define __I_DIO_WAKEUP 9
2116 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2117 #define I_LINKABLE (1 << 10)
2118 #define I_DIRTY_TIME (1 << 11)
2119 #define __I_DIRTY_TIME_EXPIRED 12
2120 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
2121 #define I_WB_SWITCH (1 << 13)
2122 #define I_OVL_INUSE (1 << 14)
2123 #define I_CREATING (1 << 15)
2125 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2126 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2127 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2129 extern void __mark_inode_dirty(struct inode *, int);
2130 static inline void mark_inode_dirty(struct inode *inode)
2132 __mark_inode_dirty(inode, I_DIRTY);
2135 static inline void mark_inode_dirty_sync(struct inode *inode)
2137 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2140 extern void inc_nlink(struct inode *inode);
2141 extern void drop_nlink(struct inode *inode);
2142 extern void clear_nlink(struct inode *inode);
2143 extern void set_nlink(struct inode *inode, unsigned int nlink);
2145 static inline void inode_inc_link_count(struct inode *inode)
2148 mark_inode_dirty(inode);
2151 static inline void inode_dec_link_count(struct inode *inode)
2154 mark_inode_dirty(inode);
2157 enum file_time_flags {
2164 extern bool atime_needs_update(const struct path *, struct inode *);
2165 extern void touch_atime(const struct path *);
2166 static inline void file_accessed(struct file *file)
2168 if (!(file->f_flags & O_NOATIME))
2169 touch_atime(&file->f_path);
2172 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2173 int sync_inode_metadata(struct inode *inode, int wait);
2175 struct file_system_type {
2178 #define FS_REQUIRES_DEV 1
2179 #define FS_BINARY_MOUNTDATA 2
2180 #define FS_HAS_SUBTYPE 4
2181 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2182 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2183 int (*init_fs_context)(struct fs_context *);
2184 const struct fs_parameter_description *parameters;
2185 struct dentry *(*mount) (struct file_system_type *, int,
2186 const char *, void *);
2187 void (*kill_sb) (struct super_block *);
2188 struct module *owner;
2189 struct file_system_type * next;
2190 struct hlist_head fs_supers;
2192 struct lock_class_key s_lock_key;
2193 struct lock_class_key s_umount_key;
2194 struct lock_class_key s_vfs_rename_key;
2195 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2197 struct lock_class_key i_lock_key;
2198 struct lock_class_key i_mutex_key;
2199 struct lock_class_key i_mutex_dir_key;
2202 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2204 extern struct dentry *mount_ns(struct file_system_type *fs_type,
2205 int flags, void *data, void *ns, struct user_namespace *user_ns,
2206 int (*fill_super)(struct super_block *, void *, int));
2208 extern 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 static inline struct dentry *mount_bdev(struct file_system_type *fs_type,
2213 int flags, const char *dev_name, void *data,
2214 int (*fill_super)(struct super_block *, void *, int))
2216 return ERR_PTR(-ENODEV);
2219 extern struct dentry *mount_single(struct file_system_type *fs_type,
2220 int flags, void *data,
2221 int (*fill_super)(struct super_block *, void *, int));
2222 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2223 int flags, void *data,
2224 int (*fill_super)(struct super_block *, void *, int));
2225 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2226 void generic_shutdown_super(struct super_block *sb);
2228 void kill_block_super(struct super_block *sb);
2230 static inline void kill_block_super(struct super_block *sb)
2235 void kill_anon_super(struct super_block *sb);
2236 void kill_litter_super(struct super_block *sb);
2237 void deactivate_super(struct super_block *sb);
2238 void deactivate_locked_super(struct super_block *sb);
2239 int set_anon_super(struct super_block *s, void *data);
2240 int set_anon_super_fc(struct super_block *s, struct fs_context *fc);
2241 int get_anon_bdev(dev_t *);
2242 void free_anon_bdev(dev_t);
2243 struct super_block *sget_fc(struct fs_context *fc,
2244 int (*test)(struct super_block *, struct fs_context *),
2245 int (*set)(struct super_block *, struct fs_context *));
2246 struct super_block *sget_userns(struct file_system_type *type,
2247 int (*test)(struct super_block *,void *),
2248 int (*set)(struct super_block *,void *),
2249 int flags, struct user_namespace *user_ns,
2251 struct super_block *sget(struct file_system_type *type,
2252 int (*test)(struct super_block *,void *),
2253 int (*set)(struct super_block *,void *),
2254 int flags, void *data);
2255 extern struct dentry *mount_pseudo_xattr(struct file_system_type *, char *,
2256 const struct super_operations *ops,
2257 const struct xattr_handler **xattr,
2258 const struct dentry_operations *dops,
2261 static inline struct dentry *
2262 mount_pseudo(struct file_system_type *fs_type, char *name,
2263 const struct super_operations *ops,
2264 const struct dentry_operations *dops, unsigned long magic)
2266 return mount_pseudo_xattr(fs_type, name, ops, NULL, dops, magic);
2269 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2270 #define fops_get(fops) \
2271 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2272 #define fops_put(fops) \
2273 do { if (fops) module_put((fops)->owner); } while(0)
2275 * This one is to be used *ONLY* from ->open() instances.
2276 * fops must be non-NULL, pinned down *and* module dependencies
2277 * should be sufficient to pin the caller down as well.
2279 #define replace_fops(f, fops) \
2281 struct file *__file = (f); \
2282 fops_put(__file->f_op); \
2283 BUG_ON(!(__file->f_op = (fops))); \
2286 extern int register_filesystem(struct file_system_type *);
2287 extern int unregister_filesystem(struct file_system_type *);
2288 extern struct vfsmount *kern_mount(struct file_system_type *);
2289 extern void kern_unmount(struct vfsmount *mnt);
2290 extern int may_umount_tree(struct vfsmount *);
2291 extern int may_umount(struct vfsmount *);
2292 extern long do_mount(const char *, const char __user *,
2293 const char *, unsigned long, void *);
2294 extern struct vfsmount *collect_mounts(const struct path *);
2295 extern void drop_collected_mounts(struct vfsmount *);
2296 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2298 extern int vfs_statfs(const struct path *, struct kstatfs *);
2299 extern int user_statfs(const char __user *, struct kstatfs *);
2300 extern int fd_statfs(int, struct kstatfs *);
2301 extern int freeze_super(struct super_block *super);
2302 extern int thaw_super(struct super_block *super);
2303 extern bool our_mnt(struct vfsmount *mnt);
2304 extern __printf(2, 3)
2305 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2306 extern int super_setup_bdi(struct super_block *sb);
2308 extern int current_umask(void);
2310 extern void ihold(struct inode * inode);
2311 extern void iput(struct inode *);
2312 extern int generic_update_time(struct inode *, struct timespec64 *, int);
2315 extern struct kobject *fs_kobj;
2317 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2319 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2320 extern int locks_mandatory_locked(struct file *);
2321 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2324 * Candidates for mandatory locking have the setgid bit set
2325 * but no group execute bit - an otherwise meaningless combination.
2328 static inline int __mandatory_lock(struct inode *ino)
2330 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2334 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2335 * otherwise these will be advisory locks
2338 static inline int mandatory_lock(struct inode *ino)
2340 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2343 static inline int locks_verify_locked(struct file *file)
2345 if (mandatory_lock(locks_inode(file)))
2346 return locks_mandatory_locked(file);
2350 static inline int locks_verify_truncate(struct inode *inode,
2354 if (!inode->i_flctx || !mandatory_lock(inode))
2357 if (size < inode->i_size) {
2358 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2361 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2366 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2368 static inline int locks_mandatory_locked(struct file *file)
2373 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2374 loff_t start, loff_t end, unsigned char type)
2379 static inline int __mandatory_lock(struct inode *inode)
2384 static inline int mandatory_lock(struct inode *inode)
2389 static inline int locks_verify_locked(struct file *file)
2394 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2400 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2403 #ifdef CONFIG_FILE_LOCKING
2404 static inline int break_lease(struct inode *inode, unsigned int mode)
2407 * Since this check is lockless, we must ensure that any refcounts
2408 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2409 * could end up racing with tasks trying to set a new lease on this
2413 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2414 return __break_lease(inode, mode, FL_LEASE);
2418 static inline int break_deleg(struct inode *inode, unsigned int mode)
2421 * Since this check is lockless, we must ensure that any refcounts
2422 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2423 * could end up racing with tasks trying to set a new lease on this
2427 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2428 return __break_lease(inode, mode, FL_DELEG);
2432 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2436 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2437 if (ret == -EWOULDBLOCK && delegated_inode) {
2438 *delegated_inode = inode;
2444 static inline int break_deleg_wait(struct inode **delegated_inode)
2448 ret = break_deleg(*delegated_inode, O_WRONLY);
2449 iput(*delegated_inode);
2450 *delegated_inode = NULL;
2454 static inline int break_layout(struct inode *inode, bool wait)
2457 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2458 return __break_lease(inode,
2459 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2464 #else /* !CONFIG_FILE_LOCKING */
2465 static inline int break_lease(struct inode *inode, unsigned int mode)
2470 static inline int break_deleg(struct inode *inode, unsigned int mode)
2475 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2480 static inline int break_deleg_wait(struct inode **delegated_inode)
2486 static inline int break_layout(struct inode *inode, bool wait)
2491 #endif /* CONFIG_FILE_LOCKING */
2496 const char *name; /* pointer to actual string */
2497 const __user char *uptr; /* original userland pointer */
2499 struct audit_names *aname;
2502 static_assert(offsetof(struct filename, iname) % sizeof(long) == 0);
2504 extern long vfs_truncate(const struct path *, loff_t);
2505 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2507 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2509 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2511 extern struct file *file_open_name(struct filename *, int, umode_t);
2512 extern struct file *filp_open(const char *, int, umode_t);
2513 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2514 const char *, int, umode_t);
2515 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2516 extern struct file * open_with_fake_path(const struct path *, int,
2517 struct inode*, const struct cred *);
2518 static inline struct file *file_clone_open(struct file *file)
2520 return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2522 extern int filp_close(struct file *, fl_owner_t id);
2524 extern struct filename *getname_flags(const char __user *, int, int *);
2525 extern struct filename *getname(const char __user *);
2526 extern struct filename *getname_kernel(const char *);
2527 extern void putname(struct filename *name);
2529 extern int finish_open(struct file *file, struct dentry *dentry,
2530 int (*open)(struct inode *, struct file *));
2531 extern int finish_no_open(struct file *file, struct dentry *dentry);
2535 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2538 extern void __init vfs_caches_init_early(void);
2539 extern void __init vfs_caches_init(void);
2541 extern struct kmem_cache *names_cachep;
2543 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2544 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2547 extern int register_blkdev(unsigned int, const char *);
2548 extern void unregister_blkdev(unsigned int, const char *);
2549 extern void bdev_unhash_inode(dev_t dev);
2550 extern struct block_device *bdget(dev_t);
2551 extern struct block_device *bdgrab(struct block_device *bdev);
2552 extern void bd_set_size(struct block_device *, loff_t size);
2553 extern void bd_forget(struct inode *inode);
2554 extern void bdput(struct block_device *);
2555 extern void invalidate_bdev(struct block_device *);
2556 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2557 extern int sync_blockdev(struct block_device *bdev);
2558 extern void kill_bdev(struct block_device *);
2559 extern struct super_block *freeze_bdev(struct block_device *);
2560 extern void emergency_thaw_all(void);
2561 extern void emergency_thaw_bdev(struct super_block *sb);
2562 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2563 extern int fsync_bdev(struct block_device *);
2565 extern struct super_block *blockdev_superblock;
2567 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2569 return sb == blockdev_superblock;
2572 static inline void bd_forget(struct inode *inode) {}
2573 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2574 static inline void kill_bdev(struct block_device *bdev) {}
2575 static inline void invalidate_bdev(struct block_device *bdev) {}
2577 static inline struct super_block *freeze_bdev(struct block_device *sb)
2582 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2587 static inline int emergency_thaw_bdev(struct super_block *sb)
2592 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2596 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2601 extern int sync_filesystem(struct super_block *);
2602 extern const struct file_operations def_blk_fops;
2603 extern const struct file_operations def_chr_fops;
2605 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2606 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2607 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2608 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2609 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2611 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2613 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2614 extern int __blkdev_reread_part(struct block_device *bdev);
2615 extern int blkdev_reread_part(struct block_device *bdev);
2618 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2619 extern void bd_unlink_disk_holder(struct block_device *bdev,
2620 struct gendisk *disk);
2622 static inline int bd_link_disk_holder(struct block_device *bdev,
2623 struct gendisk *disk)
2627 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2628 struct gendisk *disk)
2635 #define CHRDEV_MAJOR_MAX 512
2636 /* Marks the bottom of the first segment of free char majors */
2637 #define CHRDEV_MAJOR_DYN_END 234
2638 /* Marks the top and bottom of the second segment of free char majors */
2639 #define CHRDEV_MAJOR_DYN_EXT_START 511
2640 #define CHRDEV_MAJOR_DYN_EXT_END 384
2642 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2643 extern int register_chrdev_region(dev_t, unsigned, const char *);
2644 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2645 unsigned int count, const char *name,
2646 const struct file_operations *fops);
2647 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2648 unsigned int count, const char *name);
2649 extern void unregister_chrdev_region(dev_t, unsigned);
2650 extern void chrdev_show(struct seq_file *,off_t);
2652 static inline int register_chrdev(unsigned int major, const char *name,
2653 const struct file_operations *fops)
2655 return __register_chrdev(major, 0, 256, name, fops);
2658 static inline void unregister_chrdev(unsigned int major, const char *name)
2660 __unregister_chrdev(major, 0, 256, name);
2663 /* fs/block_dev.c */
2664 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2665 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2668 #define BLKDEV_MAJOR_MAX 512
2669 extern const char *__bdevname(dev_t, char *buffer);
2670 extern const char *bdevname(struct block_device *bdev, char *buffer);
2671 extern struct block_device *lookup_bdev(const char *);
2672 extern void blkdev_show(struct seq_file *,off_t);
2675 #define BLKDEV_MAJOR_MAX 0
2678 extern void init_special_inode(struct inode *, umode_t, dev_t);
2680 /* Invalid inode operations -- fs/bad_inode.c */
2681 extern void make_bad_inode(struct inode *);
2682 extern bool is_bad_inode(struct inode *);
2685 extern void check_disk_size_change(struct gendisk *disk,
2686 struct block_device *bdev, bool verbose);
2687 extern int revalidate_disk(struct gendisk *);
2688 extern int check_disk_change(struct block_device *);
2689 extern int __invalidate_device(struct block_device *, bool);
2690 extern int invalidate_partition(struct gendisk *, int);
2692 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2693 pgoff_t start, pgoff_t end);
2695 static inline void invalidate_remote_inode(struct inode *inode)
2697 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2698 S_ISLNK(inode->i_mode))
2699 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2701 extern int invalidate_inode_pages2(struct address_space *mapping);
2702 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2703 pgoff_t start, pgoff_t end);
2704 extern int write_inode_now(struct inode *, int);
2705 extern int filemap_fdatawrite(struct address_space *);
2706 extern int filemap_flush(struct address_space *);
2707 extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2708 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2711 static inline int filemap_fdatawait(struct address_space *mapping)
2713 return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2716 extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2718 extern int filemap_write_and_wait(struct address_space *mapping);
2719 extern int filemap_write_and_wait_range(struct address_space *mapping,
2720 loff_t lstart, loff_t lend);
2721 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2722 loff_t start, loff_t end, int sync_mode);
2723 extern int filemap_fdatawrite_range(struct address_space *mapping,
2724 loff_t start, loff_t end);
2725 extern int filemap_check_errors(struct address_space *mapping);
2726 extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2728 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2730 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2731 extern int __must_check file_write_and_wait_range(struct file *file,
2732 loff_t start, loff_t end);
2734 static inline int file_write_and_wait(struct file *file)
2736 return file_write_and_wait_range(file, 0, LLONG_MAX);
2740 * filemap_set_wb_err - set a writeback error on an address_space
2741 * @mapping: mapping in which to set writeback error
2742 * @err: error to be set in mapping
2744 * When writeback fails in some way, we must record that error so that
2745 * userspace can be informed when fsync and the like are called. We endeavor
2746 * to report errors on any file that was open at the time of the error. Some
2747 * internal callers also need to know when writeback errors have occurred.
2749 * When a writeback error occurs, most filesystems will want to call
2750 * filemap_set_wb_err to record the error in the mapping so that it will be
2751 * automatically reported whenever fsync is called on the file.
2753 static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2755 /* Fastpath for common case of no error */
2757 __filemap_set_wb_err(mapping, err);
2761 * filemap_check_wb_error - has an error occurred since the mark was sampled?
2762 * @mapping: mapping to check for writeback errors
2763 * @since: previously-sampled errseq_t
2765 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2766 * the given value was sampled.
2768 * If it has then report the latest error set, otherwise return 0.
2770 static inline int filemap_check_wb_err(struct address_space *mapping,
2773 return errseq_check(&mapping->wb_err, since);
2777 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2778 * @mapping: mapping to be sampled
2780 * Writeback errors are always reported relative to a particular sample point
2781 * in the past. This function provides those sample points.
2783 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2785 return errseq_sample(&mapping->wb_err);
2788 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2790 extern int vfs_fsync(struct file *file, int datasync);
2792 extern int sync_file_range(struct file *file, loff_t offset, loff_t nbytes,
2793 unsigned int flags);
2796 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2797 * to already be updated for the write, and will return either the amount
2798 * of bytes passed in, or an error if syncing the file failed.
2800 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2802 if (iocb->ki_flags & IOCB_DSYNC) {
2803 int ret = vfs_fsync_range(iocb->ki_filp,
2804 iocb->ki_pos - count, iocb->ki_pos - 1,
2805 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2813 extern void emergency_sync(void);
2814 extern void emergency_remount(void);
2816 extern sector_t bmap(struct inode *, sector_t);
2818 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2819 extern int inode_permission(struct inode *, int);
2820 extern int generic_permission(struct inode *, int);
2821 extern int __check_sticky(struct inode *dir, struct inode *inode);
2823 static inline bool execute_ok(struct inode *inode)
2825 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2828 static inline void file_start_write(struct file *file)
2830 if (!S_ISREG(file_inode(file)->i_mode))
2832 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2835 static inline bool file_start_write_trylock(struct file *file)
2837 if (!S_ISREG(file_inode(file)->i_mode))
2839 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2842 static inline void file_end_write(struct file *file)
2844 if (!S_ISREG(file_inode(file)->i_mode))
2846 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2850 * get_write_access() gets write permission for a file.
2851 * put_write_access() releases this write permission.
2852 * This is used for regular files.
2853 * We cannot support write (and maybe mmap read-write shared) accesses and
2854 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2855 * can have the following values:
2856 * 0: no writers, no VM_DENYWRITE mappings
2857 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2858 * > 0: (i_writecount) users are writing to the file.
2860 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2861 * except for the cases where we don't hold i_writecount yet. Then we need to
2862 * use {get,deny}_write_access() - these functions check the sign and refuse
2863 * to do the change if sign is wrong.
2865 static inline int get_write_access(struct inode *inode)
2867 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2869 static inline int deny_write_access(struct file *file)
2871 struct inode *inode = file_inode(file);
2872 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2874 static inline void put_write_access(struct inode * inode)
2876 atomic_dec(&inode->i_writecount);
2878 static inline void allow_write_access(struct file *file)
2881 atomic_inc(&file_inode(file)->i_writecount);
2883 static inline bool inode_is_open_for_write(const struct inode *inode)
2885 return atomic_read(&inode->i_writecount) > 0;
2889 static inline void i_readcount_dec(struct inode *inode)
2891 BUG_ON(!atomic_read(&inode->i_readcount));
2892 atomic_dec(&inode->i_readcount);
2894 static inline void i_readcount_inc(struct inode *inode)
2896 atomic_inc(&inode->i_readcount);
2899 static inline void i_readcount_dec(struct inode *inode)
2903 static inline void i_readcount_inc(struct inode *inode)
2908 extern int do_pipe_flags(int *, int);
2910 #define __kernel_read_file_id(id) \
2911 id(UNKNOWN, unknown) \
2912 id(FIRMWARE, firmware) \
2913 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2914 id(MODULE, kernel-module) \
2915 id(KEXEC_IMAGE, kexec-image) \
2916 id(KEXEC_INITRAMFS, kexec-initramfs) \
2917 id(POLICY, security-policy) \
2918 id(X509_CERTIFICATE, x509-certificate) \
2921 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2922 #define __fid_stringify(dummy, str) #str,
2924 enum kernel_read_file_id {
2925 __kernel_read_file_id(__fid_enumify)
2928 static const char * const kernel_read_file_str[] = {
2929 __kernel_read_file_id(__fid_stringify)
2932 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
2934 if ((unsigned)id >= READING_MAX_ID)
2935 return kernel_read_file_str[READING_UNKNOWN];
2937 return kernel_read_file_str[id];
2940 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2941 enum kernel_read_file_id);
2942 extern int kernel_read_file_from_path(const char *, void **, loff_t *, loff_t,
2943 enum kernel_read_file_id);
2944 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2945 enum kernel_read_file_id);
2946 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
2947 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
2948 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
2949 extern struct file * open_exec(const char *);
2951 /* fs/dcache.c -- generic fs support functions */
2952 extern bool is_subdir(struct dentry *, struct dentry *);
2953 extern bool path_is_under(const struct path *, const struct path *);
2955 extern char *file_path(struct file *, char *, int);
2957 #include <linux/err.h>
2959 /* needed for stackable file system support */
2960 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2962 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2964 extern int inode_init_always(struct super_block *, struct inode *);
2965 extern void inode_init_once(struct inode *);
2966 extern void address_space_init_once(struct address_space *mapping);
2967 extern struct inode * igrab(struct inode *);
2968 extern ino_t iunique(struct super_block *, ino_t);
2969 extern int inode_needs_sync(struct inode *inode);
2970 extern int generic_delete_inode(struct inode *inode);
2971 static inline int generic_drop_inode(struct inode *inode)
2973 return !inode->i_nlink || inode_unhashed(inode);
2976 extern struct inode *ilookup5_nowait(struct super_block *sb,
2977 unsigned long hashval, int (*test)(struct inode *, void *),
2979 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2980 int (*test)(struct inode *, void *), void *data);
2981 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2983 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
2984 int (*test)(struct inode *, void *),
2985 int (*set)(struct inode *, void *),
2987 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2988 extern struct inode * iget_locked(struct super_block *, unsigned long);
2989 extern struct inode *find_inode_nowait(struct super_block *,
2991 int (*match)(struct inode *,
2992 unsigned long, void *),
2994 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2995 extern int insert_inode_locked(struct inode *);
2996 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2997 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2999 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
3001 extern void unlock_new_inode(struct inode *);
3002 extern void discard_new_inode(struct inode *);
3003 extern unsigned int get_next_ino(void);
3004 extern void evict_inodes(struct super_block *sb);
3006 extern void __iget(struct inode * inode);
3007 extern void iget_failed(struct inode *);
3008 extern void clear_inode(struct inode *);
3009 extern void __destroy_inode(struct inode *);
3010 extern struct inode *new_inode_pseudo(struct super_block *sb);
3011 extern struct inode *new_inode(struct super_block *sb);
3012 extern void free_inode_nonrcu(struct inode *inode);
3013 extern int should_remove_suid(struct dentry *);
3014 extern int file_remove_privs(struct file *);
3016 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
3017 static inline void insert_inode_hash(struct inode *inode)
3019 __insert_inode_hash(inode, inode->i_ino);
3022 extern void __remove_inode_hash(struct inode *);
3023 static inline void remove_inode_hash(struct inode *inode)
3025 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
3026 __remove_inode_hash(inode);
3029 extern void inode_sb_list_add(struct inode *inode);
3032 extern int bdev_read_only(struct block_device *);
3034 extern int set_blocksize(struct block_device *, int);
3035 extern int sb_set_blocksize(struct super_block *, int);
3036 extern int sb_min_blocksize(struct super_block *, int);
3038 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
3039 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
3040 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
3041 extern int generic_remap_checks(struct file *file_in, loff_t pos_in,
3042 struct file *file_out, loff_t pos_out,
3043 loff_t *count, unsigned int remap_flags);
3044 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
3045 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
3046 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
3047 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
3048 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
3050 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
3052 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
3055 /* fs/block_dev.c */
3056 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
3057 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
3058 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
3060 extern void block_sync_page(struct page *page);
3063 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
3064 struct pipe_inode_info *, size_t, unsigned int);
3065 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
3066 struct file *, loff_t *, size_t, unsigned int);
3067 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
3068 struct file *out, loff_t *, size_t len, unsigned int flags);
3069 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
3070 loff_t *opos, size_t len, unsigned int flags);
3074 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3075 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3076 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
3077 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3078 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3079 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3080 int whence, loff_t maxsize, loff_t eof);
3081 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3082 int whence, loff_t size);
3083 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3084 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3085 extern int generic_file_open(struct inode * inode, struct file * filp);
3086 extern int nonseekable_open(struct inode * inode, struct file * filp);
3087 extern int stream_open(struct inode * inode, struct file * filp);
3090 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3091 loff_t file_offset);
3094 /* need locking between buffered and direct access */
3097 /* filesystem does not support filling holes */
3098 DIO_SKIP_HOLES = 0x02,
3101 void dio_end_io(struct bio *bio);
3102 void dio_warn_stale_pagecache(struct file *filp);
3104 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3105 struct block_device *bdev, struct iov_iter *iter,
3106 get_block_t get_block,
3107 dio_iodone_t end_io, dio_submit_t submit_io,
3110 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3111 struct inode *inode,
3112 struct iov_iter *iter,
3113 get_block_t get_block)
3115 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3116 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3120 void inode_dio_wait(struct inode *inode);
3123 * inode_dio_begin - signal start of a direct I/O requests
3124 * @inode: inode the direct I/O happens on
3126 * This is called once we've finished processing a direct I/O request,
3127 * and is used to wake up callers waiting for direct I/O to be quiesced.
3129 static inline void inode_dio_begin(struct inode *inode)
3131 atomic_inc(&inode->i_dio_count);
3135 * inode_dio_end - signal finish of a direct I/O requests
3136 * @inode: inode the direct I/O happens on
3138 * This is called once we've finished processing a direct I/O request,
3139 * and is used to wake up callers waiting for direct I/O to be quiesced.
3141 static inline void inode_dio_end(struct inode *inode)
3143 if (atomic_dec_and_test(&inode->i_dio_count))
3144 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3147 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3150 extern const struct file_operations generic_ro_fops;
3152 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3154 extern int readlink_copy(char __user *, int, const char *);
3155 extern int page_readlink(struct dentry *, char __user *, int);
3156 extern const char *page_get_link(struct dentry *, struct inode *,
3157 struct delayed_call *);
3158 extern void page_put_link(void *);
3159 extern int __page_symlink(struct inode *inode, const char *symname, int len,
3161 extern int page_symlink(struct inode *inode, const char *symname, int len);
3162 extern const struct inode_operations page_symlink_inode_operations;
3163 extern void kfree_link(void *);
3164 extern void generic_fillattr(struct inode *, struct kstat *);
3165 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3166 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3167 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3168 void inode_add_bytes(struct inode *inode, loff_t bytes);
3169 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3170 void inode_sub_bytes(struct inode *inode, loff_t bytes);
3171 static inline loff_t __inode_get_bytes(struct inode *inode)
3173 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3175 loff_t inode_get_bytes(struct inode *inode);
3176 void inode_set_bytes(struct inode *inode, loff_t bytes);
3177 const char *simple_get_link(struct dentry *, struct inode *,
3178 struct delayed_call *);
3179 extern const struct inode_operations simple_symlink_inode_operations;
3181 extern int iterate_dir(struct file *, struct dir_context *);
3183 extern int vfs_statx(int, const char __user *, int, struct kstat *, u32);
3184 extern int vfs_statx_fd(unsigned int, struct kstat *, u32, unsigned int);
3186 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3188 return vfs_statx(AT_FDCWD, filename, AT_NO_AUTOMOUNT,
3189 stat, STATX_BASIC_STATS);
3191 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3193 return vfs_statx(AT_FDCWD, name, AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT,
3194 stat, STATX_BASIC_STATS);
3196 static inline int vfs_fstatat(int dfd, const char __user *filename,
3197 struct kstat *stat, int flags)
3199 return vfs_statx(dfd, filename, flags | AT_NO_AUTOMOUNT,
3200 stat, STATX_BASIC_STATS);
3202 static inline int vfs_fstat(int fd, struct kstat *stat)
3204 return vfs_statx_fd(fd, stat, STATX_BASIC_STATS, 0);
3208 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3209 extern int vfs_readlink(struct dentry *, char __user *, int);
3211 extern int __generic_block_fiemap(struct inode *inode,
3212 struct fiemap_extent_info *fieinfo,
3213 loff_t start, loff_t len,
3214 get_block_t *get_block);
3215 extern int generic_block_fiemap(struct inode *inode,
3216 struct fiemap_extent_info *fieinfo, u64 start,
3217 u64 len, get_block_t *get_block);
3219 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3220 extern void put_filesystem(struct file_system_type *fs);
3221 extern struct file_system_type *get_fs_type(const char *name);
3222 extern struct super_block *get_super(struct block_device *);
3223 extern struct super_block *get_super_thawed(struct block_device *);
3224 extern struct super_block *get_super_exclusive_thawed(struct block_device *bdev);
3225 extern struct super_block *get_active_super(struct block_device *bdev);
3226 extern void drop_super(struct super_block *sb);
3227 extern void drop_super_exclusive(struct super_block *sb);
3228 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3229 extern void iterate_supers_type(struct file_system_type *,
3230 void (*)(struct super_block *, void *), void *);
3232 extern int dcache_dir_open(struct inode *, struct file *);
3233 extern int dcache_dir_close(struct inode *, struct file *);
3234 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3235 extern int dcache_readdir(struct file *, struct dir_context *);
3236 extern int simple_setattr(struct dentry *, struct iattr *);
3237 extern int simple_getattr(const struct path *, struct kstat *, u32, unsigned int);
3238 extern int simple_statfs(struct dentry *, struct kstatfs *);
3239 extern int simple_open(struct inode *inode, struct file *file);
3240 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3241 extern int simple_unlink(struct inode *, struct dentry *);
3242 extern int simple_rmdir(struct inode *, struct dentry *);
3243 extern int simple_rename(struct inode *, struct dentry *,
3244 struct inode *, struct dentry *, unsigned int);
3245 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3246 extern int noop_set_page_dirty(struct page *page);
3247 extern void noop_invalidatepage(struct page *page, unsigned int offset,
3248 unsigned int length);
3249 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3250 extern int simple_empty(struct dentry *);
3251 extern int simple_readpage(struct file *file, struct page *page);
3252 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3253 loff_t pos, unsigned len, unsigned flags,
3254 struct page **pagep, void **fsdata);
3255 extern int simple_write_end(struct file *file, struct address_space *mapping,
3256 loff_t pos, unsigned len, unsigned copied,
3257 struct page *page, void *fsdata);
3258 extern int always_delete_dentry(const struct dentry *);
3259 extern struct inode *alloc_anon_inode(struct super_block *);
3260 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3261 extern const struct dentry_operations simple_dentry_operations;
3263 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3264 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3265 extern const struct file_operations simple_dir_operations;
3266 extern const struct inode_operations simple_dir_inode_operations;
3267 extern void make_empty_dir_inode(struct inode *inode);
3268 extern bool is_empty_dir_inode(struct inode *inode);
3269 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3270 struct dentry *d_alloc_name(struct dentry *, const char *);
3271 extern int simple_fill_super(struct super_block *, unsigned long,
3272 const struct tree_descr *);
3273 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3274 extern void simple_release_fs(struct vfsmount **mount, int *count);
3276 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3277 loff_t *ppos, const void *from, size_t available);
3278 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3279 const void __user *from, size_t count);
3281 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3282 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3284 extern int generic_check_addressable(unsigned, u64);
3286 #ifdef CONFIG_MIGRATION
3287 extern int buffer_migrate_page(struct address_space *,
3288 struct page *, struct page *,
3290 extern int buffer_migrate_page_norefs(struct address_space *,
3291 struct page *, struct page *,
3294 #define buffer_migrate_page NULL
3295 #define buffer_migrate_page_norefs NULL
3298 extern int setattr_prepare(struct dentry *, struct iattr *);
3299 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3300 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
3302 extern int file_update_time(struct file *file);
3304 static inline bool io_is_direct(struct file *filp)
3306 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
3309 static inline bool vma_is_dax(struct vm_area_struct *vma)
3311 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3314 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3316 struct inode *inode;
3320 if (!vma_is_dax(vma))
3322 inode = file_inode(vma->vm_file);
3323 if (S_ISCHR(inode->i_mode))
3324 return false; /* device-dax */
3328 static inline int iocb_flags(struct file *file)
3331 if (file->f_flags & O_APPEND)
3333 if (io_is_direct(file))
3335 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3337 if (file->f_flags & __O_SYNC)
3342 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3344 if (unlikely(flags & ~RWF_SUPPORTED))
3347 if (flags & RWF_NOWAIT) {
3348 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3350 ki->ki_flags |= IOCB_NOWAIT;
3352 if (flags & RWF_HIPRI)
3353 ki->ki_flags |= IOCB_HIPRI;
3354 if (flags & RWF_DSYNC)
3355 ki->ki_flags |= IOCB_DSYNC;
3356 if (flags & RWF_SYNC)
3357 ki->ki_flags |= (IOCB_DSYNC | IOCB_SYNC);
3358 if (flags & RWF_APPEND)
3359 ki->ki_flags |= IOCB_APPEND;
3363 static inline ino_t parent_ino(struct dentry *dentry)
3368 * Don't strictly need d_lock here? If the parent ino could change
3369 * then surely we'd have a deeper race in the caller?
3371 spin_lock(&dentry->d_lock);
3372 res = dentry->d_parent->d_inode->i_ino;
3373 spin_unlock(&dentry->d_lock);
3377 /* Transaction based IO helpers */
3380 * An argresp is stored in an allocated page and holds the
3381 * size of the argument or response, along with its content
3383 struct simple_transaction_argresp {
3388 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3390 char *simple_transaction_get(struct file *file, const char __user *buf,
3392 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3393 size_t size, loff_t *pos);
3394 int simple_transaction_release(struct inode *inode, struct file *file);
3396 void simple_transaction_set(struct file *file, size_t n);
3399 * simple attribute files
3401 * These attributes behave similar to those in sysfs:
3403 * Writing to an attribute immediately sets a value, an open file can be
3404 * written to multiple times.
3406 * Reading from an attribute creates a buffer from the value that might get
3407 * read with multiple read calls. When the attribute has been read
3408 * completely, no further read calls are possible until the file is opened
3411 * All attributes contain a text representation of a numeric value
3412 * that are accessed with the get() and set() functions.
3414 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3415 static int __fops ## _open(struct inode *inode, struct file *file) \
3417 __simple_attr_check_format(__fmt, 0ull); \
3418 return simple_attr_open(inode, file, __get, __set, __fmt); \
3420 static const struct file_operations __fops = { \
3421 .owner = THIS_MODULE, \
3422 .open = __fops ## _open, \
3423 .release = simple_attr_release, \
3424 .read = simple_attr_read, \
3425 .write = simple_attr_write, \
3426 .llseek = generic_file_llseek, \
3429 static inline __printf(1, 2)
3430 void __simple_attr_check_format(const char *fmt, ...)
3432 /* don't do anything, just let the compiler check the arguments; */
3435 int simple_attr_open(struct inode *inode, struct file *file,
3436 int (*get)(void *, u64 *), int (*set)(void *, u64),
3438 int simple_attr_release(struct inode *inode, struct file *file);
3439 ssize_t simple_attr_read(struct file *file, char __user *buf,
3440 size_t len, loff_t *ppos);
3441 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3442 size_t len, loff_t *ppos);
3445 int proc_nr_files(struct ctl_table *table, int write,
3446 void __user *buffer, size_t *lenp, loff_t *ppos);
3447 int proc_nr_dentry(struct ctl_table *table, int write,
3448 void __user *buffer, size_t *lenp, loff_t *ppos);
3449 int proc_nr_inodes(struct ctl_table *table, int write,
3450 void __user *buffer, size_t *lenp, loff_t *ppos);
3451 int __init get_filesystem_list(char *buf);
3453 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3454 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3456 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3457 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3458 (flag & __FMODE_NONOTIFY)))
3460 static inline bool is_sxid(umode_t mode)
3462 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3465 static inline int check_sticky(struct inode *dir, struct inode *inode)
3467 if (!(dir->i_mode & S_ISVTX))
3470 return __check_sticky(dir, inode);
3473 static inline void inode_has_no_xattr(struct inode *inode)
3475 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3476 inode->i_flags |= S_NOSEC;
3479 static inline bool is_root_inode(struct inode *inode)
3481 return inode == inode->i_sb->s_root->d_inode;
3484 static inline bool dir_emit(struct dir_context *ctx,
3485 const char *name, int namelen,
3486 u64 ino, unsigned type)
3488 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3490 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3492 return ctx->actor(ctx, ".", 1, ctx->pos,
3493 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3495 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3497 return ctx->actor(ctx, "..", 2, ctx->pos,
3498 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3500 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3502 if (ctx->pos == 0) {
3503 if (!dir_emit_dot(file, ctx))
3507 if (ctx->pos == 1) {
3508 if (!dir_emit_dotdot(file, ctx))
3514 static inline bool dir_relax(struct inode *inode)
3516 inode_unlock(inode);
3518 return !IS_DEADDIR(inode);
3521 static inline bool dir_relax_shared(struct inode *inode)
3523 inode_unlock_shared(inode);
3524 inode_lock_shared(inode);
3525 return !IS_DEADDIR(inode);
3528 extern bool path_noexec(const struct path *path);
3529 extern void inode_nohighmem(struct inode *inode);
3532 extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len,
3535 #if defined(CONFIG_IO_URING)
3536 extern struct sock *io_uring_get_socket(struct file *file);
3538 static inline struct sock *io_uring_get_socket(struct file *file)
3544 #endif /* _LINUX_FS_H */