1 /* SPDX-License-Identifier: GPL-2.0 */
5 #include <linux/linkage.h>
6 #include <linux/wait_bit.h>
7 #include <linux/kdev_t.h>
8 #include <linux/dcache.h>
9 #include <linux/path.h>
10 #include <linux/stat.h>
11 #include <linux/cache.h>
12 #include <linux/list.h>
13 #include <linux/list_lru.h>
14 #include <linux/llist.h>
15 #include <linux/radix-tree.h>
16 #include <linux/xarray.h>
17 #include <linux/rbtree.h>
18 #include <linux/init.h>
19 #include <linux/pid.h>
20 #include <linux/bug.h>
21 #include <linux/mutex.h>
22 #include <linux/rwsem.h>
23 #include <linux/mm_types.h>
24 #include <linux/capability.h>
25 #include <linux/semaphore.h>
26 #include <linux/fcntl.h>
27 #include <linux/rculist_bl.h>
28 #include <linux/atomic.h>
29 #include <linux/shrinker.h>
30 #include <linux/migrate_mode.h>
31 #include <linux/uidgid.h>
32 #include <linux/lockdep.h>
33 #include <linux/percpu-rwsem.h>
34 #include <linux/workqueue.h>
35 #include <linux/delayed_call.h>
36 #include <linux/uuid.h>
37 #include <linux/errseq.h>
38 #include <linux/ioprio.h>
39 #include <linux/fs_types.h>
40 #include <linux/build_bug.h>
41 #include <linux/stddef.h>
43 #include <asm/byteorder.h>
44 #include <uapi/linux/fs.h>
46 struct backing_dev_info;
49 struct export_operations;
50 struct fiemap_extent_info;
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 fsverity_operations;
70 struct fs_parameter_spec;
72 extern void __init inode_init(void);
73 extern void __init inode_init_early(void);
74 extern void __init files_init(void);
75 extern void __init files_maxfiles_init(void);
77 extern struct files_stat_struct files_stat;
78 extern unsigned long get_max_files(void);
79 extern unsigned int sysctl_nr_open;
80 extern struct inodes_stat_t inodes_stat;
81 extern int leases_enable, lease_break_time;
82 extern int sysctl_protected_symlinks;
83 extern int sysctl_protected_hardlinks;
84 extern int sysctl_protected_fifos;
85 extern int sysctl_protected_regular;
87 typedef __kernel_rwf_t rwf_t;
90 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
91 struct buffer_head *bh_result, int create);
92 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
93 ssize_t bytes, void *private);
95 #define MAY_EXEC 0x00000001
96 #define MAY_WRITE 0x00000002
97 #define MAY_READ 0x00000004
98 #define MAY_APPEND 0x00000008
99 #define MAY_ACCESS 0x00000010
100 #define MAY_OPEN 0x00000020
101 #define MAY_CHDIR 0x00000040
102 /* called from RCU mode, don't block */
103 #define MAY_NOT_BLOCK 0x00000080
106 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
107 * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
110 /* file is open for reading */
111 #define FMODE_READ ((__force fmode_t)0x1)
112 /* file is open for writing */
113 #define FMODE_WRITE ((__force fmode_t)0x2)
114 /* file is seekable */
115 #define FMODE_LSEEK ((__force fmode_t)0x4)
116 /* file can be accessed using pread */
117 #define FMODE_PREAD ((__force fmode_t)0x8)
118 /* file can be accessed using pwrite */
119 #define FMODE_PWRITE ((__force fmode_t)0x10)
120 /* File is opened for execution with sys_execve / sys_uselib */
121 #define FMODE_EXEC ((__force fmode_t)0x20)
122 /* File is opened with O_NDELAY (only set for block devices) */
123 #define FMODE_NDELAY ((__force fmode_t)0x40)
124 /* File is opened with O_EXCL (only set for block devices) */
125 #define FMODE_EXCL ((__force fmode_t)0x80)
126 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
127 (specialy hack for floppy.c) */
128 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
129 /* 32bit hashes as llseek() offset (for directories) */
130 #define FMODE_32BITHASH ((__force fmode_t)0x200)
131 /* 64bit hashes as llseek() offset (for directories) */
132 #define FMODE_64BITHASH ((__force fmode_t)0x400)
135 * Don't update ctime and mtime.
137 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
138 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
140 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
142 /* Expect random access pattern */
143 #define FMODE_RANDOM ((__force fmode_t)0x1000)
145 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
146 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
148 /* File is opened with O_PATH; almost nothing can be done with it */
149 #define FMODE_PATH ((__force fmode_t)0x4000)
151 /* File needs atomic accesses to f_pos */
152 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
153 /* Write access to underlying fs */
154 #define FMODE_WRITER ((__force fmode_t)0x10000)
155 /* Has read method(s) */
156 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
157 /* Has write method(s) */
158 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
160 #define FMODE_OPENED ((__force fmode_t)0x80000)
161 #define FMODE_CREATED ((__force fmode_t)0x100000)
163 /* File is stream-like */
164 #define FMODE_STREAM ((__force fmode_t)0x200000)
166 /* File was opened by fanotify and shouldn't generate fanotify events */
167 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
169 /* File is capable of returning -EAGAIN if I/O will block */
170 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
172 /* File represents mount that needs unmounting */
173 #define FMODE_NEED_UNMOUNT ((__force fmode_t)0x10000000)
175 /* File does not contribute to nr_files count */
176 #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000)
178 /* File supports async buffered reads */
179 #define FMODE_BUF_RASYNC ((__force fmode_t)0x40000000)
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;
290 struct readahead_control;
293 * Write life time hint values.
294 * Stored in struct inode as u8.
297 WRITE_LIFE_NOT_SET = 0,
298 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
299 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
300 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
301 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
302 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
305 #define IOCB_EVENTFD (1 << 0)
306 #define IOCB_APPEND (1 << 1)
307 #define IOCB_DIRECT (1 << 2)
308 #define IOCB_HIPRI (1 << 3)
309 #define IOCB_DSYNC (1 << 4)
310 #define IOCB_SYNC (1 << 5)
311 #define IOCB_WRITE (1 << 6)
312 #define IOCB_NOWAIT (1 << 7)
313 /* iocb->ki_waitq is valid */
314 #define IOCB_WAITQ (1 << 8)
315 #define IOCB_NOIO (1 << 9)
318 struct file *ki_filp;
320 /* The 'ki_filp' pointer is shared in a union for aio */
321 randomized_struct_fields_start
324 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
328 u16 ki_ioprio; /* See linux/ioprio.h */
330 unsigned int ki_cookie; /* for ->iopoll */
331 struct wait_page_queue *ki_waitq; /* for async buffered IO */
334 randomized_struct_fields_end
337 static inline bool is_sync_kiocb(struct kiocb *kiocb)
339 return kiocb->ki_complete == NULL;
343 * "descriptor" for what we're up to with a read.
344 * This allows us to use the same read code yet
345 * have multiple different users of the data that
346 * we read from a file.
348 * The simplest case just copies the data to user
361 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
362 unsigned long, unsigned long);
364 struct address_space_operations {
365 int (*writepage)(struct page *page, struct writeback_control *wbc);
366 int (*readpage)(struct file *, struct page *);
368 /* Write back some dirty pages from this mapping. */
369 int (*writepages)(struct address_space *, struct writeback_control *);
371 /* Set a page dirty. Return true if this dirtied it */
372 int (*set_page_dirty)(struct page *page);
375 * Reads in the requested pages. Unlike ->readpage(), this is
376 * PURELY used for read-ahead!.
378 int (*readpages)(struct file *filp, struct address_space *mapping,
379 struct list_head *pages, unsigned nr_pages);
380 void (*readahead)(struct readahead_control *);
382 int (*write_begin)(struct file *, struct address_space *mapping,
383 loff_t pos, unsigned len, unsigned flags,
384 struct page **pagep, void **fsdata);
385 int (*write_end)(struct file *, struct address_space *mapping,
386 loff_t pos, unsigned len, unsigned copied,
387 struct page *page, void *fsdata);
389 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
390 sector_t (*bmap)(struct address_space *, sector_t);
391 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
392 int (*releasepage) (struct page *, gfp_t);
393 void (*freepage)(struct page *);
394 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
396 * migrate the contents of a page to the specified target. If
397 * migrate_mode is MIGRATE_ASYNC, it must not block.
399 int (*migratepage) (struct address_space *,
400 struct page *, struct page *, enum migrate_mode);
401 bool (*isolate_page)(struct page *, isolate_mode_t);
402 void (*putback_page)(struct page *);
403 int (*launder_page) (struct page *);
404 int (*is_partially_uptodate) (struct page *, unsigned long,
406 void (*is_dirty_writeback) (struct page *, bool *, bool *);
407 int (*error_remove_page)(struct address_space *, struct page *);
409 /* swapfile support */
410 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
412 void (*swap_deactivate)(struct file *file);
415 extern const struct address_space_operations empty_aops;
418 * pagecache_write_begin/pagecache_write_end must be used by general code
419 * to write into the pagecache.
421 int pagecache_write_begin(struct file *, struct address_space *mapping,
422 loff_t pos, unsigned len, unsigned flags,
423 struct page **pagep, void **fsdata);
425 int pagecache_write_end(struct file *, struct address_space *mapping,
426 loff_t pos, unsigned len, unsigned copied,
427 struct page *page, void *fsdata);
430 * struct address_space - Contents of a cacheable, mappable object.
431 * @host: Owner, either the inode or the block_device.
432 * @i_pages: Cached pages.
433 * @gfp_mask: Memory allocation flags to use for allocating pages.
434 * @i_mmap_writable: Number of VM_SHARED mappings.
435 * @nr_thps: Number of THPs in the pagecache (non-shmem only).
436 * @i_mmap: Tree of private and shared mappings.
437 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
438 * @nrpages: Number of page entries, protected by the i_pages lock.
439 * @nrexceptional: Shadow or DAX entries, protected by the i_pages lock.
440 * @writeback_index: Writeback starts here.
442 * @flags: Error bits and flags (AS_*).
443 * @wb_err: The most recent error which has occurred.
444 * @private_lock: For use by the owner of the address_space.
445 * @private_list: For use by the owner of the address_space.
446 * @private_data: For use by the owner of the address_space.
448 struct address_space {
450 struct xarray i_pages;
452 atomic_t i_mmap_writable;
453 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
454 /* number of thp, only for non-shmem files */
457 struct rb_root_cached i_mmap;
458 struct rw_semaphore i_mmap_rwsem;
459 unsigned long nrpages;
460 unsigned long nrexceptional;
461 pgoff_t writeback_index;
462 const struct address_space_operations *a_ops;
465 spinlock_t private_lock;
466 struct list_head private_list;
468 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
470 * On most architectures that alignment is already the case; but
471 * must be enforced here for CRIS, to let the least significant bit
472 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
475 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
476 #define PAGECACHE_TAG_DIRTY XA_MARK_0
477 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
478 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
481 * Returns true if any of the pages in the mapping are marked with the tag.
483 static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag)
485 return xa_marked(&mapping->i_pages, tag);
488 static inline void i_mmap_lock_write(struct address_space *mapping)
490 down_write(&mapping->i_mmap_rwsem);
493 static inline int i_mmap_trylock_write(struct address_space *mapping)
495 return down_write_trylock(&mapping->i_mmap_rwsem);
498 static inline void i_mmap_unlock_write(struct address_space *mapping)
500 up_write(&mapping->i_mmap_rwsem);
503 static inline void i_mmap_lock_read(struct address_space *mapping)
505 down_read(&mapping->i_mmap_rwsem);
508 static inline void i_mmap_unlock_read(struct address_space *mapping)
510 up_read(&mapping->i_mmap_rwsem);
513 static inline void i_mmap_assert_locked(struct address_space *mapping)
515 lockdep_assert_held(&mapping->i_mmap_rwsem);
518 static inline void i_mmap_assert_write_locked(struct address_space *mapping)
520 lockdep_assert_held_write(&mapping->i_mmap_rwsem);
524 * Might pages of this file be mapped into userspace?
526 static inline int mapping_mapped(struct address_space *mapping)
528 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
532 * Might pages of this file have been modified in userspace?
533 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap
534 * marks vma as VM_SHARED if it is shared, and the file was opened for
535 * writing i.e. vma may be mprotected writable even if now readonly.
537 * If i_mmap_writable is negative, no new writable mappings are allowed. You
538 * can only deny writable mappings, if none exists right now.
540 static inline int mapping_writably_mapped(struct address_space *mapping)
542 return atomic_read(&mapping->i_mmap_writable) > 0;
545 static inline int mapping_map_writable(struct address_space *mapping)
547 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
551 static inline void mapping_unmap_writable(struct address_space *mapping)
553 atomic_dec(&mapping->i_mmap_writable);
556 static inline int mapping_deny_writable(struct address_space *mapping)
558 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
562 static inline void mapping_allow_writable(struct address_space *mapping)
564 atomic_inc(&mapping->i_mmap_writable);
568 * Use sequence counter to get consistent i_size on 32-bit processors.
570 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
571 #include <linux/seqlock.h>
572 #define __NEED_I_SIZE_ORDERED
573 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
575 #define i_size_ordered_init(inode) do { } while (0)
579 #define ACL_NOT_CACHED ((void *)(-1))
580 #define ACL_DONT_CACHE ((void *)(-3))
582 static inline struct posix_acl *
583 uncached_acl_sentinel(struct task_struct *task)
585 return (void *)task + 1;
589 is_uncached_acl(struct posix_acl *acl)
591 return (long)acl & 1;
594 #define IOP_FASTPERM 0x0001
595 #define IOP_LOOKUP 0x0002
596 #define IOP_NOFOLLOW 0x0004
597 #define IOP_XATTR 0x0008
598 #define IOP_DEFAULT_READLINK 0x0010
600 struct fsnotify_mark_connector;
603 * Keep mostly read-only and often accessed (especially for
604 * the RCU path lookup and 'stat' data) fields at the beginning
605 * of the 'struct inode'
609 unsigned short i_opflags;
612 unsigned int i_flags;
614 #ifdef CONFIG_FS_POSIX_ACL
615 struct posix_acl *i_acl;
616 struct posix_acl *i_default_acl;
619 const struct inode_operations *i_op;
620 struct super_block *i_sb;
621 struct address_space *i_mapping;
623 #ifdef CONFIG_SECURITY
627 /* Stat data, not accessed from path walking */
630 * Filesystems may only read i_nlink directly. They shall use the
631 * following functions for modification:
633 * (set|clear|inc|drop)_nlink
634 * inode_(inc|dec)_link_count
637 const unsigned int i_nlink;
638 unsigned int __i_nlink;
642 struct timespec64 i_atime;
643 struct timespec64 i_mtime;
644 struct timespec64 i_ctime;
645 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
646 unsigned short i_bytes;
651 #ifdef __NEED_I_SIZE_ORDERED
652 seqcount_t i_size_seqcount;
656 unsigned long i_state;
657 struct rw_semaphore i_rwsem;
659 unsigned long dirtied_when; /* jiffies of first dirtying */
660 unsigned long dirtied_time_when;
662 struct hlist_node i_hash;
663 struct list_head i_io_list; /* backing dev IO list */
664 #ifdef CONFIG_CGROUP_WRITEBACK
665 struct bdi_writeback *i_wb; /* the associated cgroup wb */
667 /* foreign inode detection, see wbc_detach_inode() */
669 u16 i_wb_frn_avg_time;
670 u16 i_wb_frn_history;
672 struct list_head i_lru; /* inode LRU list */
673 struct list_head i_sb_list;
674 struct list_head i_wb_list; /* backing dev writeback list */
676 struct hlist_head i_dentry;
677 struct rcu_head i_rcu;
679 atomic64_t i_version;
680 atomic64_t i_sequence; /* see futex */
682 atomic_t i_dio_count;
683 atomic_t i_writecount;
684 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
685 atomic_t i_readcount; /* struct files open RO */
688 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
689 void (*free_inode)(struct inode *);
691 struct file_lock_context *i_flctx;
692 struct address_space i_data;
693 struct list_head i_devices;
695 struct pipe_inode_info *i_pipe;
696 struct block_device *i_bdev;
704 #ifdef CONFIG_FSNOTIFY
705 __u32 i_fsnotify_mask; /* all events this inode cares about */
706 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
709 #ifdef CONFIG_FS_ENCRYPTION
710 struct fscrypt_info *i_crypt_info;
713 #ifdef CONFIG_FS_VERITY
714 struct fsverity_info *i_verity_info;
717 void *i_private; /* fs or device private pointer */
718 } __randomize_layout;
720 struct timespec64 timestamp_truncate(struct timespec64 t, struct inode *inode);
722 static inline unsigned int i_blocksize(const struct inode *node)
724 return (1 << node->i_blkbits);
727 static inline int inode_unhashed(struct inode *inode)
729 return hlist_unhashed(&inode->i_hash);
733 * __mark_inode_dirty expects inodes to be hashed. Since we don't
734 * want special inodes in the fileset inode space, we make them
735 * appear hashed, but do not put on any lists. hlist_del()
736 * will work fine and require no locking.
738 static inline void inode_fake_hash(struct inode *inode)
740 hlist_add_fake(&inode->i_hash);
744 * inode->i_mutex nesting subclasses for the lock validator:
746 * 0: the object of the current VFS operation
750 * 4: second non-directory
751 * 5: second parent (when locking independent directories in rename)
753 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
754 * non-directories at once.
756 * The locking order between these classes is
757 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
759 enum inode_i_mutex_lock_class
769 static inline void inode_lock(struct inode *inode)
771 down_write(&inode->i_rwsem);
774 static inline void inode_unlock(struct inode *inode)
776 up_write(&inode->i_rwsem);
779 static inline void inode_lock_shared(struct inode *inode)
781 down_read(&inode->i_rwsem);
784 static inline void inode_unlock_shared(struct inode *inode)
786 up_read(&inode->i_rwsem);
789 static inline int inode_trylock(struct inode *inode)
791 return down_write_trylock(&inode->i_rwsem);
794 static inline int inode_trylock_shared(struct inode *inode)
796 return down_read_trylock(&inode->i_rwsem);
799 static inline int inode_is_locked(struct inode *inode)
801 return rwsem_is_locked(&inode->i_rwsem);
804 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
806 down_write_nested(&inode->i_rwsem, subclass);
809 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
811 down_read_nested(&inode->i_rwsem, subclass);
814 void lock_two_nondirectories(struct inode *, struct inode*);
815 void unlock_two_nondirectories(struct inode *, struct inode*);
818 * NOTE: in a 32bit arch with a preemptable kernel and
819 * an UP compile the i_size_read/write must be atomic
820 * with respect to the local cpu (unlike with preempt disabled),
821 * but they don't need to be atomic with respect to other cpus like in
822 * true SMP (so they need either to either locally disable irq around
823 * the read or for example on x86 they can be still implemented as a
824 * cmpxchg8b without the need of the lock prefix). For SMP compiles
825 * and 64bit archs it makes no difference if preempt is enabled or not.
827 static inline loff_t i_size_read(const struct inode *inode)
829 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
834 seq = read_seqcount_begin(&inode->i_size_seqcount);
835 i_size = inode->i_size;
836 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
838 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
842 i_size = inode->i_size;
846 return inode->i_size;
851 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
852 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
853 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
855 static inline void i_size_write(struct inode *inode, loff_t i_size)
857 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
859 write_seqcount_begin(&inode->i_size_seqcount);
860 inode->i_size = i_size;
861 write_seqcount_end(&inode->i_size_seqcount);
863 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
865 inode->i_size = i_size;
868 inode->i_size = i_size;
872 static inline unsigned iminor(const struct inode *inode)
874 return MINOR(inode->i_rdev);
877 static inline unsigned imajor(const struct inode *inode)
879 return MAJOR(inode->i_rdev);
883 rwlock_t lock; /* protects pid, uid, euid fields */
884 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
885 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
886 kuid_t uid, euid; /* uid/euid of process setting the owner */
887 int signum; /* posix.1b rt signal to be delivered on IO */
891 * Track a single file's readahead state
893 struct file_ra_state {
894 pgoff_t start; /* where readahead started */
895 unsigned int size; /* # of readahead pages */
896 unsigned int async_size; /* do asynchronous readahead when
897 there are only # of pages ahead */
899 unsigned int ra_pages; /* Maximum readahead window */
900 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
901 loff_t prev_pos; /* Cache last read() position */
905 * Check if @index falls in the readahead windows.
907 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
909 return (index >= ra->start &&
910 index < ra->start + ra->size);
915 struct llist_node fu_llist;
916 struct rcu_head fu_rcuhead;
919 struct inode *f_inode; /* cached value */
920 const struct file_operations *f_op;
923 * Protects f_ep_links, f_flags.
924 * Must not be taken from IRQ context.
927 enum rw_hint f_write_hint;
928 atomic_long_t f_count;
929 unsigned int f_flags;
931 struct mutex f_pos_lock;
933 struct fown_struct f_owner;
934 const struct cred *f_cred;
935 struct file_ra_state f_ra;
938 #ifdef CONFIG_SECURITY
941 /* needed for tty driver, and maybe others */
945 /* Used by fs/eventpoll.c to link all the hooks to this file */
946 struct list_head f_ep_links;
947 struct list_head f_tfile_llink;
948 #endif /* #ifdef CONFIG_EPOLL */
949 struct address_space *f_mapping;
951 errseq_t f_sb_err; /* for syncfs */
953 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
958 /* file identifier */
959 unsigned char f_handle[];
962 static inline struct file *get_file(struct file *f)
964 atomic_long_inc(&f->f_count);
967 #define get_file_rcu_many(x, cnt) \
968 atomic_long_add_unless(&(x)->f_count, (cnt), 0)
969 #define get_file_rcu(x) get_file_rcu_many((x), 1)
970 #define file_count(x) atomic_long_read(&(x)->f_count)
972 #define MAX_NON_LFS ((1UL<<31) - 1)
974 /* Page cache limit. The filesystems should put that into their s_maxbytes
975 limits, otherwise bad things can happen in VM. */
976 #if BITS_PER_LONG==32
977 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
978 #elif BITS_PER_LONG==64
979 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
984 #define FL_DELEG 4 /* NFSv4 delegation */
985 #define FL_ACCESS 8 /* not trying to lock, just looking */
986 #define FL_EXISTS 16 /* when unlocking, test for existence */
987 #define FL_LEASE 32 /* lease held on this file */
988 #define FL_CLOSE 64 /* unlock on close */
989 #define FL_SLEEP 128 /* A blocking lock */
990 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
991 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
992 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
993 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
995 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
998 * Special return value from posix_lock_file() and vfs_lock_file() for
999 * asynchronous locking.
1001 #define FILE_LOCK_DEFERRED 1
1003 /* legacy typedef, should eventually be removed */
1004 typedef void *fl_owner_t;
1008 struct file_lock_operations {
1009 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1010 void (*fl_release_private)(struct file_lock *);
1013 struct lock_manager_operations {
1014 fl_owner_t (*lm_get_owner)(fl_owner_t);
1015 void (*lm_put_owner)(fl_owner_t);
1016 void (*lm_notify)(struct file_lock *); /* unblock callback */
1017 int (*lm_grant)(struct file_lock *, int);
1018 bool (*lm_break)(struct file_lock *);
1019 int (*lm_change)(struct file_lock *, int, struct list_head *);
1020 void (*lm_setup)(struct file_lock *, void **);
1021 bool (*lm_breaker_owns_lease)(struct file_lock *);
1024 struct lock_manager {
1025 struct list_head list;
1027 * NFSv4 and up also want opens blocked during the grace period;
1034 void locks_start_grace(struct net *, struct lock_manager *);
1035 void locks_end_grace(struct lock_manager *);
1036 bool locks_in_grace(struct net *);
1037 bool opens_in_grace(struct net *);
1039 /* that will die - we need it for nfs_lock_info */
1040 #include <linux/nfs_fs_i.h>
1043 * struct file_lock represents a generic "file lock". It's used to represent
1044 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1045 * note that the same struct is used to represent both a request for a lock and
1046 * the lock itself, but the same object is never used for both.
1048 * FIXME: should we create a separate "struct lock_request" to help distinguish
1051 * The varous i_flctx lists are ordered by:
1054 * 2) lock range start
1057 * Obviously, the last two criteria only matter for POSIX locks.
1060 struct file_lock *fl_blocker; /* The lock, that is blocking us */
1061 struct list_head fl_list; /* link into file_lock_context */
1062 struct hlist_node fl_link; /* node in global lists */
1063 struct list_head fl_blocked_requests; /* list of requests with
1064 * ->fl_blocker pointing here
1066 struct list_head fl_blocked_member; /* node in
1067 * ->fl_blocker->fl_blocked_requests
1069 fl_owner_t fl_owner;
1070 unsigned int fl_flags;
1071 unsigned char fl_type;
1072 unsigned int fl_pid;
1073 int fl_link_cpu; /* what cpu's list is this on? */
1074 wait_queue_head_t fl_wait;
1075 struct file *fl_file;
1079 struct fasync_struct * fl_fasync; /* for lease break notifications */
1080 /* for lease breaks: */
1081 unsigned long fl_break_time;
1082 unsigned long fl_downgrade_time;
1084 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1085 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1087 struct nfs_lock_info nfs_fl;
1088 struct nfs4_lock_info nfs4_fl;
1090 struct list_head link; /* link in AFS vnode's pending_locks list */
1091 int state; /* state of grant or error if -ve */
1092 unsigned int debug_id;
1095 } __randomize_layout;
1097 struct file_lock_context {
1098 spinlock_t flc_lock;
1099 struct list_head flc_flock;
1100 struct list_head flc_posix;
1101 struct list_head flc_lease;
1104 /* The following constant reflects the upper bound of the file/locking space */
1106 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1107 #define OFFSET_MAX INT_LIMIT(loff_t)
1108 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1111 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1113 #define locks_inode(f) file_inode(f)
1115 #ifdef CONFIG_FILE_LOCKING
1116 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1117 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1120 #if BITS_PER_LONG == 32
1121 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1122 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1126 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1127 extern int fcntl_getlease(struct file *filp);
1130 void locks_free_lock_context(struct inode *inode);
1131 void locks_free_lock(struct file_lock *fl);
1132 extern void locks_init_lock(struct file_lock *);
1133 extern struct file_lock * locks_alloc_lock(void);
1134 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1135 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1136 extern void locks_remove_posix(struct file *, fl_owner_t);
1137 extern void locks_remove_file(struct file *);
1138 extern void locks_release_private(struct file_lock *);
1139 extern void posix_test_lock(struct file *, struct file_lock *);
1140 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1141 extern int locks_delete_block(struct file_lock *);
1142 extern int vfs_test_lock(struct file *, struct file_lock *);
1143 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1144 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1145 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1146 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1147 extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1148 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1149 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1150 extern int lease_modify(struct file_lock *, int, struct list_head *);
1152 struct notifier_block;
1153 extern int lease_register_notifier(struct notifier_block *);
1154 extern void lease_unregister_notifier(struct notifier_block *);
1156 struct files_struct;
1157 extern void show_fd_locks(struct seq_file *f,
1158 struct file *filp, struct files_struct *files);
1159 #else /* !CONFIG_FILE_LOCKING */
1160 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1161 struct flock __user *user)
1166 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1167 unsigned int cmd, struct flock __user *user)
1172 #if BITS_PER_LONG == 32
1173 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1174 struct flock64 __user *user)
1179 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1180 unsigned int cmd, struct flock64 __user *user)
1185 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1190 static inline int fcntl_getlease(struct file *filp)
1196 locks_free_lock_context(struct inode *inode)
1200 static inline void locks_init_lock(struct file_lock *fl)
1205 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1210 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1215 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1220 static inline void locks_remove_file(struct file *filp)
1225 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1230 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1231 struct file_lock *conflock)
1236 static inline int locks_delete_block(struct file_lock *waiter)
1241 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1246 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1247 struct file_lock *fl, struct file_lock *conf)
1252 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1257 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1262 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1267 static inline void lease_get_mtime(struct inode *inode,
1268 struct timespec64 *time)
1273 static inline int generic_setlease(struct file *filp, long arg,
1274 struct file_lock **flp, void **priv)
1279 static inline int vfs_setlease(struct file *filp, long arg,
1280 struct file_lock **lease, void **priv)
1285 static inline int lease_modify(struct file_lock *fl, int arg,
1286 struct list_head *dispose)
1291 struct files_struct;
1292 static inline void show_fd_locks(struct seq_file *f,
1293 struct file *filp, struct files_struct *files) {}
1294 #endif /* !CONFIG_FILE_LOCKING */
1296 static inline struct inode *file_inode(const struct file *f)
1301 static inline struct dentry *file_dentry(const struct file *file)
1303 return d_real(file->f_path.dentry, file_inode(file));
1306 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1308 return locks_lock_inode_wait(locks_inode(filp), fl);
1311 struct fasync_struct {
1315 struct fasync_struct *fa_next; /* singly linked list */
1316 struct file *fa_file;
1317 struct rcu_head fa_rcu;
1320 #define FASYNC_MAGIC 0x4601
1322 /* SMP safe fasync helpers: */
1323 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1324 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1325 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1326 extern struct fasync_struct *fasync_alloc(void);
1327 extern void fasync_free(struct fasync_struct *);
1329 /* can be called from interrupts */
1330 extern void kill_fasync(struct fasync_struct **, int, int);
1332 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1333 extern int f_setown(struct file *filp, unsigned long arg, int force);
1334 extern void f_delown(struct file *filp);
1335 extern pid_t f_getown(struct file *filp);
1336 extern int send_sigurg(struct fown_struct *fown);
1339 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1340 * represented in both.
1342 #define SB_RDONLY 1 /* Mount read-only */
1343 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1344 #define SB_NODEV 4 /* Disallow access to device special files */
1345 #define SB_NOEXEC 8 /* Disallow program execution */
1346 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1347 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1348 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1349 #define SB_NOATIME 1024 /* Do not update access times. */
1350 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1351 #define SB_SILENT 32768
1352 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1353 #define SB_INLINECRYPT (1<<17) /* Use blk-crypto for encrypted files */
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_STABLE_WRITES 0x00000008 /* don't modify blks until WB is done */
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 #define SB_I_SKIP_SYNC 0x00000100 /* Skip superblock at global sync */
1389 /* Possible states of 'frozen' field */
1391 SB_UNFROZEN = 0, /* FS is unfrozen */
1392 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1393 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1394 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1395 * internal threads if needed) */
1396 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1399 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1402 int frozen; /* Is sb frozen? */
1403 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1404 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1407 struct super_block {
1408 struct list_head s_list; /* Keep this first */
1409 dev_t s_dev; /* search index; _not_ kdev_t */
1410 unsigned char s_blocksize_bits;
1411 unsigned long s_blocksize;
1412 loff_t s_maxbytes; /* Max file size */
1413 struct file_system_type *s_type;
1414 const struct super_operations *s_op;
1415 const struct dquot_operations *dq_op;
1416 const struct quotactl_ops *s_qcop;
1417 const struct export_operations *s_export_op;
1418 unsigned long s_flags;
1419 unsigned long s_iflags; /* internal SB_I_* flags */
1420 unsigned long s_magic;
1421 struct dentry *s_root;
1422 struct rw_semaphore s_umount;
1425 #ifdef CONFIG_SECURITY
1428 const struct xattr_handler **s_xattr;
1429 #ifdef CONFIG_FS_ENCRYPTION
1430 const struct fscrypt_operations *s_cop;
1431 struct key *s_master_keys; /* master crypto keys in use */
1433 #ifdef CONFIG_FS_VERITY
1434 const struct fsverity_operations *s_vop;
1436 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1437 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1438 struct block_device *s_bdev;
1439 struct backing_dev_info *s_bdi;
1440 struct mtd_info *s_mtd;
1441 struct hlist_node s_instances;
1442 unsigned int s_quota_types; /* Bitmask of supported quota types */
1443 struct quota_info s_dquot; /* Diskquota specific options */
1445 struct sb_writers s_writers;
1448 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1449 * s_fsnotify_marks together for cache efficiency. They are frequently
1450 * accessed and rarely modified.
1452 void *s_fs_info; /* Filesystem private info */
1454 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1456 /* Time limits for c/m/atime in seconds */
1457 time64_t s_time_min;
1458 time64_t s_time_max;
1459 #ifdef CONFIG_FSNOTIFY
1460 __u32 s_fsnotify_mask;
1461 struct fsnotify_mark_connector __rcu *s_fsnotify_marks;
1464 char s_id[32]; /* Informational name */
1465 uuid_t s_uuid; /* UUID */
1467 unsigned int s_max_links;
1471 * The next field is for VFS *only*. No filesystems have any business
1472 * even looking at it. You had been warned.
1474 struct mutex s_vfs_rename_mutex; /* Kludge */
1477 * Filesystem subtype. If non-empty the filesystem type field
1478 * in /proc/mounts will be "type.subtype"
1480 const char *s_subtype;
1482 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1485 * Saved pool identifier for cleancache (-1 means none)
1487 int cleancache_poolid;
1489 struct shrinker s_shrink; /* per-sb shrinker handle */
1491 /* Number of inodes with nlink == 0 but still referenced */
1492 atomic_long_t s_remove_count;
1494 /* Pending fsnotify inode refs */
1495 atomic_long_t s_fsnotify_inode_refs;
1497 /* Being remounted read-only */
1498 int s_readonly_remount;
1500 /* per-sb errseq_t for reporting writeback errors via syncfs */
1503 /* AIO completions deferred from interrupt context */
1504 struct workqueue_struct *s_dio_done_wq;
1505 struct hlist_head s_pins;
1508 * Owning user namespace and default context in which to
1509 * interpret filesystem uids, gids, quotas, device nodes,
1510 * xattrs and security labels.
1512 struct user_namespace *s_user_ns;
1515 * The list_lru structure is essentially just a pointer to a table
1516 * of per-node lru lists, each of which has its own spinlock.
1517 * There is no need to put them into separate cachelines.
1519 struct list_lru s_dentry_lru;
1520 struct list_lru s_inode_lru;
1521 struct rcu_head rcu;
1522 struct work_struct destroy_work;
1524 struct mutex s_sync_lock; /* sync serialisation lock */
1527 * Indicates how deep in a filesystem stack this SB is
1531 /* s_inode_list_lock protects s_inodes */
1532 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1533 struct list_head s_inodes; /* all inodes */
1535 spinlock_t s_inode_wblist_lock;
1536 struct list_head s_inodes_wb; /* writeback inodes */
1537 } __randomize_layout;
1539 /* Helper functions so that in most cases filesystems will
1540 * not need to deal directly with kuid_t and kgid_t and can
1541 * instead deal with the raw numeric values that are stored
1542 * in the filesystem.
1544 static inline uid_t i_uid_read(const struct inode *inode)
1546 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1549 static inline gid_t i_gid_read(const struct inode *inode)
1551 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1554 static inline void i_uid_write(struct inode *inode, uid_t uid)
1556 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1559 static inline void i_gid_write(struct inode *inode, gid_t gid)
1561 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1564 extern struct timespec64 current_time(struct inode *inode);
1567 * Snapshotting support.
1570 void __sb_end_write(struct super_block *sb, int level);
1571 int __sb_start_write(struct super_block *sb, int level, bool wait);
1573 #define __sb_writers_acquired(sb, lev) \
1574 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1575 #define __sb_writers_release(sb, lev) \
1576 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1579 * sb_end_write - drop write access to a superblock
1580 * @sb: the super we wrote to
1582 * Decrement number of writers to the filesystem. Wake up possible waiters
1583 * wanting to freeze the filesystem.
1585 static inline void sb_end_write(struct super_block *sb)
1587 __sb_end_write(sb, SB_FREEZE_WRITE);
1591 * sb_end_pagefault - drop write access to a superblock from a page fault
1592 * @sb: the super we wrote to
1594 * Decrement number of processes handling write page fault to the filesystem.
1595 * Wake up possible waiters wanting to freeze the filesystem.
1597 static inline void sb_end_pagefault(struct super_block *sb)
1599 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1603 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1604 * @sb: the super we wrote to
1606 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1607 * waiters wanting to freeze the filesystem.
1609 static inline void sb_end_intwrite(struct super_block *sb)
1611 __sb_end_write(sb, SB_FREEZE_FS);
1615 * sb_start_write - get write access to a superblock
1616 * @sb: the super we write to
1618 * When a process wants to write data or metadata to a file system (i.e. dirty
1619 * a page or an inode), it should embed the operation in a sb_start_write() -
1620 * sb_end_write() pair to get exclusion against file system freezing. This
1621 * function increments number of writers preventing freezing. If the file
1622 * system is already frozen, the function waits until the file system is
1625 * Since freeze protection behaves as a lock, users have to preserve
1626 * ordering of freeze protection and other filesystem locks. Generally,
1627 * freeze protection should be the outermost lock. In particular, we have:
1630 * -> i_mutex (write path, truncate, directory ops, ...)
1631 * -> s_umount (freeze_super, thaw_super)
1633 static inline void sb_start_write(struct super_block *sb)
1635 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1638 static inline int sb_start_write_trylock(struct super_block *sb)
1640 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1644 * sb_start_pagefault - get write access to a superblock from a page fault
1645 * @sb: the super we write to
1647 * When a process starts handling write page fault, it should embed the
1648 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1649 * exclusion against file system freezing. This is needed since the page fault
1650 * is going to dirty a page. This function increments number of running page
1651 * faults preventing freezing. If the file system is already frozen, the
1652 * function waits until the file system is thawed.
1654 * Since page fault freeze protection behaves as a lock, users have to preserve
1655 * ordering of freeze protection and other filesystem locks. It is advised to
1656 * put sb_start_pagefault() close to mmap_lock in lock ordering. Page fault
1657 * handling code implies lock dependency:
1660 * -> sb_start_pagefault
1662 static inline void sb_start_pagefault(struct super_block *sb)
1664 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1668 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1669 * @sb: the super we write to
1671 * This is the third level of protection against filesystem freezing. It is
1672 * free for use by a filesystem. The only requirement is that it must rank
1673 * below sb_start_pagefault.
1675 * For example filesystem can call sb_start_intwrite() when starting a
1676 * transaction which somewhat eases handling of freezing for internal sources
1677 * of filesystem changes (internal fs threads, discarding preallocation on file
1680 static inline void sb_start_intwrite(struct super_block *sb)
1682 __sb_start_write(sb, SB_FREEZE_FS, true);
1685 static inline int sb_start_intwrite_trylock(struct super_block *sb)
1687 return __sb_start_write(sb, SB_FREEZE_FS, false);
1691 extern bool inode_owner_or_capable(const struct inode *inode);
1694 * VFS helper functions..
1696 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1697 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1698 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1699 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1700 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1701 extern int vfs_rmdir(struct inode *, struct dentry *);
1702 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1703 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1705 static inline int vfs_whiteout(struct inode *dir, struct dentry *dentry)
1707 return vfs_mknod(dir, dentry, S_IFCHR | WHITEOUT_MODE, WHITEOUT_DEV);
1710 extern struct dentry *vfs_tmpfile(struct dentry *dentry, umode_t mode,
1713 int vfs_mkobj(struct dentry *, umode_t,
1714 int (*f)(struct dentry *, umode_t, void *),
1717 int vfs_fchown(struct file *file, uid_t user, gid_t group);
1718 int vfs_fchmod(struct file *file, umode_t mode);
1719 int vfs_utimes(const struct path *path, struct timespec64 *times);
1721 extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1723 #ifdef CONFIG_COMPAT
1724 extern long compat_ptr_ioctl(struct file *file, unsigned int cmd,
1727 #define compat_ptr_ioctl NULL
1731 * VFS file helper functions.
1733 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1735 extern bool may_open_dev(const struct path *path);
1738 * This is the "filldir" function type, used by readdir() to let
1739 * the kernel specify what kind of dirent layout it wants to have.
1740 * This allows the kernel to read directories into kernel space or
1741 * to have different dirent layouts depending on the binary type.
1744 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1747 struct dir_context {
1753 * These flags let !MMU mmap() govern direct device mapping vs immediate
1754 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1756 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1757 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1758 * NOMMU_MAP_READ: Can be mapped for reading
1759 * NOMMU_MAP_WRITE: Can be mapped for writing
1760 * NOMMU_MAP_EXEC: Can be mapped for execution
1762 #define NOMMU_MAP_COPY 0x00000001
1763 #define NOMMU_MAP_DIRECT 0x00000008
1764 #define NOMMU_MAP_READ VM_MAYREAD
1765 #define NOMMU_MAP_WRITE VM_MAYWRITE
1766 #define NOMMU_MAP_EXEC VM_MAYEXEC
1768 #define NOMMU_VMFLAGS \
1769 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1772 * These flags control the behavior of the remap_file_range function pointer.
1773 * If it is called with len == 0 that means "remap to end of source file".
1774 * See Documentation/filesystems/vfs.rst for more details about this call.
1776 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
1777 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
1779 #define REMAP_FILE_DEDUP (1 << 0)
1780 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
1783 * These flags signal that the caller is ok with altering various aspects of
1784 * the behavior of the remap operation. The changes must be made by the
1785 * implementation; the vfs remap helper functions can take advantage of them.
1786 * Flags in this category exist to preserve the quirky behavior of the hoisted
1787 * btrfs clone/dedupe ioctls.
1789 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
1793 struct file_operations {
1794 struct module *owner;
1795 loff_t (*llseek) (struct file *, loff_t, int);
1796 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1797 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1798 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1799 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1800 int (*iopoll)(struct kiocb *kiocb, bool spin);
1801 int (*iterate) (struct file *, struct dir_context *);
1802 int (*iterate_shared) (struct file *, struct dir_context *);
1803 __poll_t (*poll) (struct file *, struct poll_table_struct *);
1804 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1805 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1806 int (*mmap) (struct file *, struct vm_area_struct *);
1807 unsigned long mmap_supported_flags;
1808 int (*open) (struct inode *, struct file *);
1809 int (*flush) (struct file *, fl_owner_t id);
1810 int (*release) (struct inode *, struct file *);
1811 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1812 int (*fasync) (int, struct file *, int);
1813 int (*lock) (struct file *, int, struct file_lock *);
1814 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1815 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1816 int (*check_flags)(int);
1817 int (*flock) (struct file *, int, struct file_lock *);
1818 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1819 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1820 int (*setlease)(struct file *, long, struct file_lock **, void **);
1821 long (*fallocate)(struct file *file, int mode, loff_t offset,
1823 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1825 unsigned (*mmap_capabilities)(struct file *);
1827 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1828 loff_t, size_t, unsigned int);
1829 loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
1830 struct file *file_out, loff_t pos_out,
1831 loff_t len, unsigned int remap_flags);
1832 int (*fadvise)(struct file *, loff_t, loff_t, int);
1833 } __randomize_layout;
1835 struct inode_operations {
1836 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1837 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1838 int (*permission) (struct inode *, int);
1839 struct posix_acl * (*get_acl)(struct inode *, int);
1841 int (*readlink) (struct dentry *, char __user *,int);
1843 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1844 int (*link) (struct dentry *,struct inode *,struct dentry *);
1845 int (*unlink) (struct inode *,struct dentry *);
1846 int (*symlink) (struct inode *,struct dentry *,const char *);
1847 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1848 int (*rmdir) (struct inode *,struct dentry *);
1849 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1850 int (*rename) (struct inode *, struct dentry *,
1851 struct inode *, struct dentry *, unsigned int);
1852 int (*setattr) (struct dentry *, struct iattr *);
1853 int (*getattr) (const struct path *, struct kstat *, u32, unsigned int);
1854 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1855 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1857 int (*update_time)(struct inode *, struct timespec64 *, int);
1858 int (*atomic_open)(struct inode *, struct dentry *,
1859 struct file *, unsigned open_flag,
1860 umode_t create_mode);
1861 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1862 int (*set_acl)(struct inode *, struct posix_acl *, int);
1863 } ____cacheline_aligned;
1865 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
1866 struct iov_iter *iter)
1868 return file->f_op->read_iter(kio, iter);
1871 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
1872 struct iov_iter *iter)
1874 return file->f_op->write_iter(kio, iter);
1877 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
1879 return file->f_op->mmap(file, vma);
1882 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1883 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1884 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1885 unsigned long, loff_t *, rwf_t);
1886 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1887 loff_t, size_t, unsigned int);
1888 extern ssize_t generic_copy_file_range(struct file *file_in, loff_t pos_in,
1889 struct file *file_out, loff_t pos_out,
1890 size_t len, unsigned int flags);
1891 extern int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
1892 struct file *file_out, loff_t pos_out,
1894 unsigned int remap_flags);
1895 extern loff_t do_clone_file_range(struct file *file_in, loff_t pos_in,
1896 struct file *file_out, loff_t pos_out,
1897 loff_t len, unsigned int remap_flags);
1898 extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1899 struct file *file_out, loff_t pos_out,
1900 loff_t len, unsigned int remap_flags);
1901 extern int vfs_dedupe_file_range(struct file *file,
1902 struct file_dedupe_range *same);
1903 extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
1904 struct file *dst_file, loff_t dst_pos,
1905 loff_t len, unsigned int remap_flags);
1908 struct super_operations {
1909 struct inode *(*alloc_inode)(struct super_block *sb);
1910 void (*destroy_inode)(struct inode *);
1911 void (*free_inode)(struct inode *);
1913 void (*dirty_inode) (struct inode *, int flags);
1914 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1915 int (*drop_inode) (struct inode *);
1916 void (*evict_inode) (struct inode *);
1917 void (*put_super) (struct super_block *);
1918 int (*sync_fs)(struct super_block *sb, int wait);
1919 int (*freeze_super) (struct super_block *);
1920 int (*freeze_fs) (struct super_block *);
1921 int (*thaw_super) (struct super_block *);
1922 int (*unfreeze_fs) (struct super_block *);
1923 int (*statfs) (struct dentry *, struct kstatfs *);
1924 int (*remount_fs) (struct super_block *, int *, char *);
1925 void (*umount_begin) (struct super_block *);
1927 int (*show_options)(struct seq_file *, struct dentry *);
1928 int (*show_devname)(struct seq_file *, struct dentry *);
1929 int (*show_path)(struct seq_file *, struct dentry *);
1930 int (*show_stats)(struct seq_file *, struct dentry *);
1932 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1933 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1934 struct dquot **(*get_dquots)(struct inode *);
1936 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1937 long (*nr_cached_objects)(struct super_block *,
1938 struct shrink_control *);
1939 long (*free_cached_objects)(struct super_block *,
1940 struct shrink_control *);
1944 * Inode flags - they have no relation to superblock flags now
1946 #define S_SYNC (1 << 0) /* Writes are synced at once */
1947 #define S_NOATIME (1 << 1) /* Do not update access times */
1948 #define S_APPEND (1 << 2) /* Append-only file */
1949 #define S_IMMUTABLE (1 << 3) /* Immutable file */
1950 #define S_DEAD (1 << 4) /* removed, but still open directory */
1951 #define S_NOQUOTA (1 << 5) /* Inode is not counted to quota */
1952 #define S_DIRSYNC (1 << 6) /* Directory modifications are synchronous */
1953 #define S_NOCMTIME (1 << 7) /* Do not update file c/mtime */
1954 #define S_SWAPFILE (1 << 8) /* Do not truncate: swapon got its bmaps */
1955 #define S_PRIVATE (1 << 9) /* Inode is fs-internal */
1956 #define S_IMA (1 << 10) /* Inode has an associated IMA struct */
1957 #define S_AUTOMOUNT (1 << 11) /* Automount/referral quasi-directory */
1958 #define S_NOSEC (1 << 12) /* no suid or xattr security attributes */
1959 #ifdef CONFIG_FS_DAX
1960 #define S_DAX (1 << 13) /* Direct Access, avoiding the page cache */
1962 #define S_DAX 0 /* Make all the DAX code disappear */
1964 #define S_ENCRYPTED (1 << 14) /* Encrypted file (using fs/crypto/) */
1965 #define S_CASEFOLD (1 << 15) /* Casefolded file */
1966 #define S_VERITY (1 << 16) /* Verity file (using fs/verity/) */
1969 * Note that nosuid etc flags are inode-specific: setting some file-system
1970 * flags just means all the inodes inherit those flags by default. It might be
1971 * possible to override it selectively if you really wanted to with some
1972 * ioctl() that is not currently implemented.
1974 * Exception: SB_RDONLY is always applied to the entire file system.
1976 * Unfortunately, it is possible to change a filesystems flags with it mounted
1977 * with files in use. This means that all of the inodes will not have their
1978 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1979 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1981 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1983 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
1984 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
1985 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
1986 ((inode)->i_flags & S_SYNC))
1987 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
1988 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1989 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
1990 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
1991 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
1993 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1994 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1995 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1996 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
1998 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1999 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
2000 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
2001 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
2002 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
2003 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2004 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2005 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2006 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2007 #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD)
2008 #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY)
2010 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2011 (inode)->i_rdev == WHITEOUT_DEV)
2013 static inline bool HAS_UNMAPPED_ID(struct inode *inode)
2015 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
2018 static inline enum rw_hint file_write_hint(struct file *file)
2020 if (file->f_write_hint != WRITE_LIFE_NOT_SET)
2021 return file->f_write_hint;
2023 return file_inode(file)->i_write_hint;
2026 static inline int iocb_flags(struct file *file);
2028 static inline u16 ki_hint_validate(enum rw_hint hint)
2030 typeof(((struct kiocb *)0)->ki_hint) max_hint = -1;
2032 if (hint <= max_hint)
2037 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
2039 *kiocb = (struct kiocb) {
2041 .ki_flags = iocb_flags(filp),
2042 .ki_hint = ki_hint_validate(file_write_hint(filp)),
2043 .ki_ioprio = get_current_ioprio(),
2047 static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src,
2050 *kiocb = (struct kiocb) {
2052 .ki_flags = kiocb_src->ki_flags,
2053 .ki_hint = kiocb_src->ki_hint,
2054 .ki_ioprio = kiocb_src->ki_ioprio,
2055 .ki_pos = kiocb_src->ki_pos,
2060 * Inode state bits. Protected by inode->i_lock
2062 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
2063 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
2065 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2066 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2067 * various stages of removing an inode.
2069 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2071 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2072 * fdatasync(). i_atime is the usual cause.
2073 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2074 * these changes separately from I_DIRTY_SYNC so that we
2075 * don't have to write inode on fdatasync() when only
2076 * mtime has changed in it.
2077 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2078 * I_NEW Serves as both a mutex and completion notification.
2079 * New inodes set I_NEW. If two processes both create
2080 * the same inode, one of them will release its inode and
2081 * wait for I_NEW to be released before returning.
2082 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2083 * also cause waiting on I_NEW, without I_NEW actually
2084 * being set. find_inode() uses this to prevent returning
2085 * nearly-dead inodes.
2086 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2087 * is zero. I_FREEING must be set when I_WILL_FREE is
2089 * I_FREEING Set when inode is about to be freed but still has dirty
2090 * pages or buffers attached or the inode itself is still
2092 * I_CLEAR Added by clear_inode(). In this state the inode is
2093 * clean and can be destroyed. Inode keeps I_FREEING.
2095 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2096 * prohibited for many purposes. iget() must wait for
2097 * the inode to be completely released, then create it
2098 * anew. Other functions will just ignore such inodes,
2099 * if appropriate. I_NEW is used for waiting.
2101 * I_SYNC Writeback of inode is running. The bit is set during
2102 * data writeback, and cleared with a wakeup on the bit
2103 * address once it is done. The bit is also used to pin
2104 * the inode in memory for flusher thread.
2106 * I_REFERENCED Marks the inode as recently references on the LRU list.
2108 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2110 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2111 * synchronize competing switching instances and to tell
2112 * wb stat updates to grab the i_pages lock. See
2113 * inode_switch_wbs_work_fn() for details.
2115 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2116 * and work dirs among overlayfs mounts.
2118 * I_CREATING New object's inode in the middle of setting up.
2120 * I_DONTCACHE Evict inode as soon as it is not used anymore.
2122 * I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists.
2123 * Used to detect that mark_inode_dirty() should not move
2124 * inode between dirty lists.
2126 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2128 #define I_DIRTY_SYNC (1 << 0)
2129 #define I_DIRTY_DATASYNC (1 << 1)
2130 #define I_DIRTY_PAGES (1 << 2)
2132 #define I_NEW (1 << __I_NEW)
2133 #define I_WILL_FREE (1 << 4)
2134 #define I_FREEING (1 << 5)
2135 #define I_CLEAR (1 << 6)
2137 #define I_SYNC (1 << __I_SYNC)
2138 #define I_REFERENCED (1 << 8)
2139 #define __I_DIO_WAKEUP 9
2140 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2141 #define I_LINKABLE (1 << 10)
2142 #define I_DIRTY_TIME (1 << 11)
2143 #define I_WB_SWITCH (1 << 13)
2144 #define I_OVL_INUSE (1 << 14)
2145 #define I_CREATING (1 << 15)
2146 #define I_DONTCACHE (1 << 16)
2147 #define I_SYNC_QUEUED (1 << 17)
2149 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2150 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2151 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2153 extern void __mark_inode_dirty(struct inode *, int);
2154 static inline void mark_inode_dirty(struct inode *inode)
2156 __mark_inode_dirty(inode, I_DIRTY);
2159 static inline void mark_inode_dirty_sync(struct inode *inode)
2161 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2164 extern void inc_nlink(struct inode *inode);
2165 extern void drop_nlink(struct inode *inode);
2166 extern void clear_nlink(struct inode *inode);
2167 extern void set_nlink(struct inode *inode, unsigned int nlink);
2169 static inline void inode_inc_link_count(struct inode *inode)
2172 mark_inode_dirty(inode);
2175 static inline void inode_dec_link_count(struct inode *inode)
2178 mark_inode_dirty(inode);
2181 enum file_time_flags {
2188 extern bool atime_needs_update(const struct path *, struct inode *);
2189 extern void touch_atime(const struct path *);
2190 static inline void file_accessed(struct file *file)
2192 if (!(file->f_flags & O_NOATIME))
2193 touch_atime(&file->f_path);
2196 extern int file_modified(struct file *file);
2198 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2199 int sync_inode_metadata(struct inode *inode, int wait);
2201 struct file_system_type {
2204 #define FS_REQUIRES_DEV 1
2205 #define FS_BINARY_MOUNTDATA 2
2206 #define FS_HAS_SUBTYPE 4
2207 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2208 #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */
2209 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2210 int (*init_fs_context)(struct fs_context *);
2211 const struct fs_parameter_spec *parameters;
2212 struct dentry *(*mount) (struct file_system_type *, int,
2213 const char *, void *);
2214 void (*kill_sb) (struct super_block *);
2215 struct module *owner;
2216 struct file_system_type * next;
2217 struct hlist_head fs_supers;
2219 struct lock_class_key s_lock_key;
2220 struct lock_class_key s_umount_key;
2221 struct lock_class_key s_vfs_rename_key;
2222 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2224 struct lock_class_key i_lock_key;
2225 struct lock_class_key i_mutex_key;
2226 struct lock_class_key i_mutex_dir_key;
2229 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2231 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2232 int flags, const char *dev_name, void *data,
2233 int (*fill_super)(struct super_block *, void *, int));
2234 extern struct dentry *mount_single(struct file_system_type *fs_type,
2235 int flags, void *data,
2236 int (*fill_super)(struct super_block *, void *, int));
2237 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2238 int flags, void *data,
2239 int (*fill_super)(struct super_block *, void *, int));
2240 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2241 void generic_shutdown_super(struct super_block *sb);
2242 void kill_block_super(struct super_block *sb);
2243 void kill_anon_super(struct super_block *sb);
2244 void kill_litter_super(struct super_block *sb);
2245 void deactivate_super(struct super_block *sb);
2246 void deactivate_locked_super(struct super_block *sb);
2247 int set_anon_super(struct super_block *s, void *data);
2248 int set_anon_super_fc(struct super_block *s, struct fs_context *fc);
2249 int get_anon_bdev(dev_t *);
2250 void free_anon_bdev(dev_t);
2251 struct super_block *sget_fc(struct fs_context *fc,
2252 int (*test)(struct super_block *, struct fs_context *),
2253 int (*set)(struct super_block *, struct fs_context *));
2254 struct super_block *sget(struct file_system_type *type,
2255 int (*test)(struct super_block *,void *),
2256 int (*set)(struct super_block *,void *),
2257 int flags, void *data);
2259 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2260 #define fops_get(fops) \
2261 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2262 #define fops_put(fops) \
2263 do { if (fops) module_put((fops)->owner); } while(0)
2265 * This one is to be used *ONLY* from ->open() instances.
2266 * fops must be non-NULL, pinned down *and* module dependencies
2267 * should be sufficient to pin the caller down as well.
2269 #define replace_fops(f, fops) \
2271 struct file *__file = (f); \
2272 fops_put(__file->f_op); \
2273 BUG_ON(!(__file->f_op = (fops))); \
2276 extern int register_filesystem(struct file_system_type *);
2277 extern int unregister_filesystem(struct file_system_type *);
2278 extern struct vfsmount *kern_mount(struct file_system_type *);
2279 extern void kern_unmount(struct vfsmount *mnt);
2280 extern int may_umount_tree(struct vfsmount *);
2281 extern int may_umount(struct vfsmount *);
2282 extern long do_mount(const char *, const char __user *,
2283 const char *, unsigned long, void *);
2284 extern struct vfsmount *collect_mounts(const struct path *);
2285 extern void drop_collected_mounts(struct vfsmount *);
2286 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2288 extern int vfs_statfs(const struct path *, struct kstatfs *);
2289 extern int user_statfs(const char __user *, struct kstatfs *);
2290 extern int fd_statfs(int, struct kstatfs *);
2291 extern int freeze_super(struct super_block *super);
2292 extern int thaw_super(struct super_block *super);
2293 extern bool our_mnt(struct vfsmount *mnt);
2294 extern __printf(2, 3)
2295 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2296 extern int super_setup_bdi(struct super_block *sb);
2298 extern int current_umask(void);
2300 extern void ihold(struct inode * inode);
2301 extern void iput(struct inode *);
2302 extern int generic_update_time(struct inode *, struct timespec64 *, int);
2305 extern struct kobject *fs_kobj;
2307 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2309 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2310 extern int locks_mandatory_locked(struct file *);
2311 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2314 * Candidates for mandatory locking have the setgid bit set
2315 * but no group execute bit - an otherwise meaningless combination.
2318 static inline int __mandatory_lock(struct inode *ino)
2320 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2324 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2325 * otherwise these will be advisory locks
2328 static inline int mandatory_lock(struct inode *ino)
2330 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2333 static inline int locks_verify_locked(struct file *file)
2335 if (mandatory_lock(locks_inode(file)))
2336 return locks_mandatory_locked(file);
2340 static inline int locks_verify_truncate(struct inode *inode,
2344 if (!inode->i_flctx || !mandatory_lock(inode))
2347 if (size < inode->i_size) {
2348 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2351 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2356 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2358 static inline int locks_mandatory_locked(struct file *file)
2363 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2364 loff_t start, loff_t end, unsigned char type)
2369 static inline int __mandatory_lock(struct inode *inode)
2374 static inline int mandatory_lock(struct inode *inode)
2379 static inline int locks_verify_locked(struct file *file)
2384 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2390 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2393 #ifdef CONFIG_FILE_LOCKING
2394 static inline int break_lease(struct inode *inode, unsigned int mode)
2397 * Since this check is lockless, we must ensure that any refcounts
2398 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2399 * could end up racing with tasks trying to set a new lease on this
2403 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2404 return __break_lease(inode, mode, FL_LEASE);
2408 static inline int break_deleg(struct inode *inode, unsigned int mode)
2411 * Since this check is lockless, we must ensure that any refcounts
2412 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2413 * could end up racing with tasks trying to set a new lease on this
2417 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2418 return __break_lease(inode, mode, FL_DELEG);
2422 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2426 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2427 if (ret == -EWOULDBLOCK && delegated_inode) {
2428 *delegated_inode = inode;
2434 static inline int break_deleg_wait(struct inode **delegated_inode)
2438 ret = break_deleg(*delegated_inode, O_WRONLY);
2439 iput(*delegated_inode);
2440 *delegated_inode = NULL;
2444 static inline int break_layout(struct inode *inode, bool wait)
2447 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2448 return __break_lease(inode,
2449 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2454 #else /* !CONFIG_FILE_LOCKING */
2455 static inline int break_lease(struct inode *inode, unsigned int mode)
2460 static inline int break_deleg(struct inode *inode, unsigned int mode)
2465 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2470 static inline int break_deleg_wait(struct inode **delegated_inode)
2476 static inline int break_layout(struct inode *inode, bool wait)
2481 #endif /* CONFIG_FILE_LOCKING */
2486 const char *name; /* pointer to actual string */
2487 const __user char *uptr; /* original userland pointer */
2489 struct audit_names *aname;
2492 static_assert(offsetof(struct filename, iname) % sizeof(long) == 0);
2494 extern long vfs_truncate(const struct path *, loff_t);
2495 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2497 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2499 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2501 extern struct file *file_open_name(struct filename *, int, umode_t);
2502 extern struct file *filp_open(const char *, int, umode_t);
2503 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2504 const char *, int, umode_t);
2505 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2506 extern struct file * open_with_fake_path(const struct path *, int,
2507 struct inode*, const struct cred *);
2508 static inline struct file *file_clone_open(struct file *file)
2510 return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2512 extern int filp_close(struct file *, fl_owner_t id);
2514 extern struct filename *getname_flags(const char __user *, int, int *);
2515 extern struct filename *getname(const char __user *);
2516 extern struct filename *getname_kernel(const char *);
2517 extern void putname(struct filename *name);
2519 extern int finish_open(struct file *file, struct dentry *dentry,
2520 int (*open)(struct inode *, struct file *));
2521 extern int finish_no_open(struct file *file, struct dentry *dentry);
2524 extern void __init vfs_caches_init_early(void);
2525 extern void __init vfs_caches_init(void);
2527 extern struct kmem_cache *names_cachep;
2529 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2530 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2532 extern struct super_block *blockdev_superblock;
2533 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2535 return IS_ENABLED(CONFIG_BLOCK) && sb == blockdev_superblock;
2538 void emergency_thaw_all(void);
2539 extern int sync_filesystem(struct super_block *);
2540 extern const struct file_operations def_blk_fops;
2541 extern const struct file_operations def_chr_fops;
2544 #define CHRDEV_MAJOR_MAX 512
2545 /* Marks the bottom of the first segment of free char majors */
2546 #define CHRDEV_MAJOR_DYN_END 234
2547 /* Marks the top and bottom of the second segment of free char majors */
2548 #define CHRDEV_MAJOR_DYN_EXT_START 511
2549 #define CHRDEV_MAJOR_DYN_EXT_END 384
2551 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2552 extern int register_chrdev_region(dev_t, unsigned, const char *);
2553 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2554 unsigned int count, const char *name,
2555 const struct file_operations *fops);
2556 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2557 unsigned int count, const char *name);
2558 extern void unregister_chrdev_region(dev_t, unsigned);
2559 extern void chrdev_show(struct seq_file *,off_t);
2561 static inline int register_chrdev(unsigned int major, const char *name,
2562 const struct file_operations *fops)
2564 return __register_chrdev(major, 0, 256, name, fops);
2567 static inline void unregister_chrdev(unsigned int major, const char *name)
2569 __unregister_chrdev(major, 0, 256, name);
2572 extern void init_special_inode(struct inode *, umode_t, dev_t);
2574 /* Invalid inode operations -- fs/bad_inode.c */
2575 extern void make_bad_inode(struct inode *);
2576 extern bool is_bad_inode(struct inode *);
2578 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2579 pgoff_t start, pgoff_t end);
2581 static inline void invalidate_remote_inode(struct inode *inode)
2583 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2584 S_ISLNK(inode->i_mode))
2585 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2587 extern int invalidate_inode_pages2(struct address_space *mapping);
2588 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2589 pgoff_t start, pgoff_t end);
2590 extern int write_inode_now(struct inode *, int);
2591 extern int filemap_fdatawrite(struct address_space *);
2592 extern int filemap_flush(struct address_space *);
2593 extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2594 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2596 extern int filemap_fdatawait_range_keep_errors(struct address_space *mapping,
2597 loff_t start_byte, loff_t end_byte);
2599 static inline int filemap_fdatawait(struct address_space *mapping)
2601 return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2604 extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2606 extern int filemap_write_and_wait_range(struct address_space *mapping,
2607 loff_t lstart, loff_t lend);
2608 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2609 loff_t start, loff_t end, int sync_mode);
2610 extern int filemap_fdatawrite_range(struct address_space *mapping,
2611 loff_t start, loff_t end);
2612 extern int filemap_check_errors(struct address_space *mapping);
2613 extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2615 static inline int filemap_write_and_wait(struct address_space *mapping)
2617 return filemap_write_and_wait_range(mapping, 0, LLONG_MAX);
2620 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2622 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2623 extern int __must_check file_write_and_wait_range(struct file *file,
2624 loff_t start, loff_t end);
2626 static inline int file_write_and_wait(struct file *file)
2628 return file_write_and_wait_range(file, 0, LLONG_MAX);
2632 * filemap_set_wb_err - set a writeback error on an address_space
2633 * @mapping: mapping in which to set writeback error
2634 * @err: error to be set in mapping
2636 * When writeback fails in some way, we must record that error so that
2637 * userspace can be informed when fsync and the like are called. We endeavor
2638 * to report errors on any file that was open at the time of the error. Some
2639 * internal callers also need to know when writeback errors have occurred.
2641 * When a writeback error occurs, most filesystems will want to call
2642 * filemap_set_wb_err to record the error in the mapping so that it will be
2643 * automatically reported whenever fsync is called on the file.
2645 static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2647 /* Fastpath for common case of no error */
2649 __filemap_set_wb_err(mapping, err);
2653 * filemap_check_wb_err - has an error occurred since the mark was sampled?
2654 * @mapping: mapping to check for writeback errors
2655 * @since: previously-sampled errseq_t
2657 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2658 * the given value was sampled.
2660 * If it has then report the latest error set, otherwise return 0.
2662 static inline int filemap_check_wb_err(struct address_space *mapping,
2665 return errseq_check(&mapping->wb_err, since);
2669 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2670 * @mapping: mapping to be sampled
2672 * Writeback errors are always reported relative to a particular sample point
2673 * in the past. This function provides those sample points.
2675 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2677 return errseq_sample(&mapping->wb_err);
2681 * file_sample_sb_err - sample the current errseq_t to test for later errors
2682 * @file: file pointer to be sampled
2684 * Grab the most current superblock-level errseq_t value for the given
2687 static inline errseq_t file_sample_sb_err(struct file *file)
2689 return errseq_sample(&file->f_path.dentry->d_sb->s_wb_err);
2692 static inline int filemap_nr_thps(struct address_space *mapping)
2694 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
2695 return atomic_read(&mapping->nr_thps);
2701 static inline void filemap_nr_thps_inc(struct address_space *mapping)
2703 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
2704 atomic_inc(&mapping->nr_thps);
2710 static inline void filemap_nr_thps_dec(struct address_space *mapping)
2712 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
2713 atomic_dec(&mapping->nr_thps);
2719 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2721 extern int vfs_fsync(struct file *file, int datasync);
2723 extern int sync_file_range(struct file *file, loff_t offset, loff_t nbytes,
2724 unsigned int flags);
2727 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2728 * to already be updated for the write, and will return either the amount
2729 * of bytes passed in, or an error if syncing the file failed.
2731 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2733 if (iocb->ki_flags & IOCB_DSYNC) {
2734 int ret = vfs_fsync_range(iocb->ki_filp,
2735 iocb->ki_pos - count, iocb->ki_pos - 1,
2736 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2744 extern void emergency_sync(void);
2745 extern void emergency_remount(void);
2748 extern int bmap(struct inode *inode, sector_t *block);
2750 static inline int bmap(struct inode *inode, sector_t *block)
2756 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2757 extern int inode_permission(struct inode *, int);
2758 extern int generic_permission(struct inode *, int);
2759 extern int __check_sticky(struct inode *dir, struct inode *inode);
2761 static inline bool execute_ok(struct inode *inode)
2763 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2766 static inline void file_start_write(struct file *file)
2768 if (!S_ISREG(file_inode(file)->i_mode))
2770 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2773 static inline bool file_start_write_trylock(struct file *file)
2775 if (!S_ISREG(file_inode(file)->i_mode))
2777 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2780 static inline void file_end_write(struct file *file)
2782 if (!S_ISREG(file_inode(file)->i_mode))
2784 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2788 * get_write_access() gets write permission for a file.
2789 * put_write_access() releases this write permission.
2790 * This is used for regular files.
2791 * We cannot support write (and maybe mmap read-write shared) accesses and
2792 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2793 * can have the following values:
2794 * 0: no writers, no VM_DENYWRITE mappings
2795 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2796 * > 0: (i_writecount) users are writing to the file.
2798 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2799 * except for the cases where we don't hold i_writecount yet. Then we need to
2800 * use {get,deny}_write_access() - these functions check the sign and refuse
2801 * to do the change if sign is wrong.
2803 static inline int get_write_access(struct inode *inode)
2805 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2807 static inline int deny_write_access(struct file *file)
2809 struct inode *inode = file_inode(file);
2810 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2812 static inline void put_write_access(struct inode * inode)
2814 atomic_dec(&inode->i_writecount);
2816 static inline void allow_write_access(struct file *file)
2819 atomic_inc(&file_inode(file)->i_writecount);
2821 static inline bool inode_is_open_for_write(const struct inode *inode)
2823 return atomic_read(&inode->i_writecount) > 0;
2826 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
2827 static inline void i_readcount_dec(struct inode *inode)
2829 BUG_ON(!atomic_read(&inode->i_readcount));
2830 atomic_dec(&inode->i_readcount);
2832 static inline void i_readcount_inc(struct inode *inode)
2834 atomic_inc(&inode->i_readcount);
2837 static inline void i_readcount_dec(struct inode *inode)
2841 static inline void i_readcount_inc(struct inode *inode)
2846 extern int do_pipe_flags(int *, int);
2848 #define __kernel_read_file_id(id) \
2849 id(UNKNOWN, unknown) \
2850 id(FIRMWARE, firmware) \
2851 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2852 id(FIRMWARE_EFI_EMBEDDED, firmware) \
2853 id(MODULE, kernel-module) \
2854 id(KEXEC_IMAGE, kexec-image) \
2855 id(KEXEC_INITRAMFS, kexec-initramfs) \
2856 id(POLICY, security-policy) \
2857 id(X509_CERTIFICATE, x509-certificate) \
2860 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2861 #define __fid_stringify(dummy, str) #str,
2863 enum kernel_read_file_id {
2864 __kernel_read_file_id(__fid_enumify)
2867 static const char * const kernel_read_file_str[] = {
2868 __kernel_read_file_id(__fid_stringify)
2871 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
2873 if ((unsigned)id >= READING_MAX_ID)
2874 return kernel_read_file_str[READING_UNKNOWN];
2876 return kernel_read_file_str[id];
2879 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2880 enum kernel_read_file_id);
2881 extern int kernel_read_file_from_path(const char *, void **, loff_t *, loff_t,
2882 enum kernel_read_file_id);
2883 extern int kernel_read_file_from_path_initns(const char *, void **, loff_t *, loff_t,
2884 enum kernel_read_file_id);
2885 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2886 enum kernel_read_file_id);
2887 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
2888 ssize_t __kernel_read(struct file *file, void *buf, size_t count, loff_t *pos);
2889 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
2890 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
2891 extern struct file * open_exec(const char *);
2893 /* fs/dcache.c -- generic fs support functions */
2894 extern bool is_subdir(struct dentry *, struct dentry *);
2895 extern bool path_is_under(const struct path *, const struct path *);
2897 extern char *file_path(struct file *, char *, int);
2899 #include <linux/err.h>
2901 /* needed for stackable file system support */
2902 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2904 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2906 extern int inode_init_always(struct super_block *, struct inode *);
2907 extern void inode_init_once(struct inode *);
2908 extern void address_space_init_once(struct address_space *mapping);
2909 extern struct inode * igrab(struct inode *);
2910 extern ino_t iunique(struct super_block *, ino_t);
2911 extern int inode_needs_sync(struct inode *inode);
2912 extern int generic_delete_inode(struct inode *inode);
2913 static inline int generic_drop_inode(struct inode *inode)
2915 return !inode->i_nlink || inode_unhashed(inode) ||
2916 (inode->i_state & I_DONTCACHE);
2918 extern void d_mark_dontcache(struct inode *inode);
2920 extern struct inode *ilookup5_nowait(struct super_block *sb,
2921 unsigned long hashval, int (*test)(struct inode *, void *),
2923 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2924 int (*test)(struct inode *, void *), void *data);
2925 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2927 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
2928 int (*test)(struct inode *, void *),
2929 int (*set)(struct inode *, void *),
2931 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2932 extern struct inode * iget_locked(struct super_block *, unsigned long);
2933 extern struct inode *find_inode_nowait(struct super_block *,
2935 int (*match)(struct inode *,
2936 unsigned long, void *),
2938 extern struct inode *find_inode_rcu(struct super_block *, unsigned long,
2939 int (*)(struct inode *, void *), void *);
2940 extern struct inode *find_inode_by_ino_rcu(struct super_block *, unsigned long);
2941 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2942 extern int insert_inode_locked(struct inode *);
2943 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2944 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2946 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2948 extern void unlock_new_inode(struct inode *);
2949 extern void discard_new_inode(struct inode *);
2950 extern unsigned int get_next_ino(void);
2951 extern void evict_inodes(struct super_block *sb);
2954 * Userspace may rely on the the inode number being non-zero. For example, glibc
2955 * simply ignores files with zero i_ino in unlink() and other places.
2957 * As an additional complication, if userspace was compiled with
2958 * _FILE_OFFSET_BITS=32 on a 64-bit kernel we'll only end up reading out the
2959 * lower 32 bits, so we need to check that those aren't zero explicitly. With
2960 * _FILE_OFFSET_BITS=64, this may cause some harmless false-negatives, but
2961 * better safe than sorry.
2963 static inline bool is_zero_ino(ino_t ino)
2965 return (u32)ino == 0;
2968 extern void __iget(struct inode * inode);
2969 extern void iget_failed(struct inode *);
2970 extern void clear_inode(struct inode *);
2971 extern void __destroy_inode(struct inode *);
2972 extern struct inode *new_inode_pseudo(struct super_block *sb);
2973 extern struct inode *new_inode(struct super_block *sb);
2974 extern void free_inode_nonrcu(struct inode *inode);
2975 extern int should_remove_suid(struct dentry *);
2976 extern int file_remove_privs(struct file *);
2978 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2979 static inline void insert_inode_hash(struct inode *inode)
2981 __insert_inode_hash(inode, inode->i_ino);
2984 extern void __remove_inode_hash(struct inode *);
2985 static inline void remove_inode_hash(struct inode *inode)
2987 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2988 __remove_inode_hash(inode);
2991 extern void inode_sb_list_add(struct inode *inode);
2993 extern int sb_set_blocksize(struct super_block *, int);
2994 extern int sb_min_blocksize(struct super_block *, int);
2996 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2997 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2998 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2999 extern int generic_remap_checks(struct file *file_in, loff_t pos_in,
3000 struct file *file_out, loff_t pos_out,
3001 loff_t *count, unsigned int remap_flags);
3002 extern int generic_file_rw_checks(struct file *file_in, struct file *file_out);
3003 extern int generic_copy_file_checks(struct file *file_in, loff_t pos_in,
3004 struct file *file_out, loff_t pos_out,
3005 size_t *count, unsigned int flags);
3006 extern ssize_t generic_file_buffered_read(struct kiocb *iocb,
3007 struct iov_iter *to, ssize_t already_read);
3008 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
3009 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
3010 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
3011 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
3012 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
3014 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
3016 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
3018 ssize_t vfs_iocb_iter_read(struct file *file, struct kiocb *iocb,
3019 struct iov_iter *iter);
3020 ssize_t vfs_iocb_iter_write(struct file *file, struct kiocb *iocb,
3021 struct iov_iter *iter);
3023 /* fs/block_dev.c */
3024 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
3025 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
3026 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
3028 extern void block_sync_page(struct page *page);
3031 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
3032 struct pipe_inode_info *, size_t, unsigned int);
3033 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
3034 struct file *, loff_t *, size_t, unsigned int);
3035 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
3036 struct file *out, loff_t *, size_t len, unsigned int flags);
3037 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
3038 loff_t *opos, size_t len, unsigned int flags);
3042 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3043 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3044 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
3045 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3046 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3047 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3048 int whence, loff_t maxsize, loff_t eof);
3049 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3050 int whence, loff_t size);
3051 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3052 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3053 extern int generic_file_open(struct inode * inode, struct file * filp);
3054 extern int nonseekable_open(struct inode * inode, struct file * filp);
3055 extern int stream_open(struct inode * inode, struct file * filp);
3058 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3059 loff_t file_offset);
3062 /* need locking between buffered and direct access */
3065 /* filesystem does not support filling holes */
3066 DIO_SKIP_HOLES = 0x02,
3069 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3070 struct block_device *bdev, struct iov_iter *iter,
3071 get_block_t get_block,
3072 dio_iodone_t end_io, dio_submit_t submit_io,
3075 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3076 struct inode *inode,
3077 struct iov_iter *iter,
3078 get_block_t get_block)
3080 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3081 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3085 void inode_dio_wait(struct inode *inode);
3088 * inode_dio_begin - signal start of a direct I/O requests
3089 * @inode: inode the direct I/O happens on
3091 * This is called once we've finished processing a direct I/O request,
3092 * and is used to wake up callers waiting for direct I/O to be quiesced.
3094 static inline void inode_dio_begin(struct inode *inode)
3096 atomic_inc(&inode->i_dio_count);
3100 * inode_dio_end - signal finish of a direct I/O requests
3101 * @inode: inode the direct I/O happens on
3103 * This is called once we've finished processing a direct I/O request,
3104 * and is used to wake up callers waiting for direct I/O to be quiesced.
3106 static inline void inode_dio_end(struct inode *inode)
3108 if (atomic_dec_and_test(&inode->i_dio_count))
3109 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3113 * Warn about a page cache invalidation failure diring a direct I/O write.
3115 void dio_warn_stale_pagecache(struct file *filp);
3117 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3120 extern const struct file_operations generic_ro_fops;
3122 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3124 extern int readlink_copy(char __user *, int, const char *);
3125 extern int page_readlink(struct dentry *, char __user *, int);
3126 extern const char *page_get_link(struct dentry *, struct inode *,
3127 struct delayed_call *);
3128 extern void page_put_link(void *);
3129 extern int __page_symlink(struct inode *inode, const char *symname, int len,
3131 extern int page_symlink(struct inode *inode, const char *symname, int len);
3132 extern const struct inode_operations page_symlink_inode_operations;
3133 extern void kfree_link(void *);
3134 extern void generic_fillattr(struct inode *, struct kstat *);
3135 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3136 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3137 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3138 void inode_add_bytes(struct inode *inode, loff_t bytes);
3139 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3140 void inode_sub_bytes(struct inode *inode, loff_t bytes);
3141 static inline loff_t __inode_get_bytes(struct inode *inode)
3143 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3145 loff_t inode_get_bytes(struct inode *inode);
3146 void inode_set_bytes(struct inode *inode, loff_t bytes);
3147 const char *simple_get_link(struct dentry *, struct inode *,
3148 struct delayed_call *);
3149 extern const struct inode_operations simple_symlink_inode_operations;
3151 extern int iterate_dir(struct file *, struct dir_context *);
3153 extern int vfs_statx(int, const char __user *, int, struct kstat *, u32);
3154 extern int vfs_statx_fd(unsigned int, struct kstat *, u32, unsigned int);
3156 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3158 return vfs_statx(AT_FDCWD, filename, AT_NO_AUTOMOUNT,
3159 stat, STATX_BASIC_STATS);
3161 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3163 return vfs_statx(AT_FDCWD, name, AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT,
3164 stat, STATX_BASIC_STATS);
3166 static inline int vfs_fstatat(int dfd, const char __user *filename,
3167 struct kstat *stat, int flags)
3169 return vfs_statx(dfd, filename, flags | AT_NO_AUTOMOUNT,
3170 stat, STATX_BASIC_STATS);
3172 static inline int vfs_fstat(int fd, struct kstat *stat)
3174 return vfs_statx_fd(fd, stat, STATX_BASIC_STATS, 0);
3178 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3179 extern int vfs_readlink(struct dentry *, char __user *, int);
3181 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3182 extern void put_filesystem(struct file_system_type *fs);
3183 extern struct file_system_type *get_fs_type(const char *name);
3184 extern struct super_block *get_super(struct block_device *);
3185 extern struct super_block *get_super_thawed(struct block_device *);
3186 extern struct super_block *get_super_exclusive_thawed(struct block_device *bdev);
3187 extern struct super_block *get_active_super(struct block_device *bdev);
3188 extern void drop_super(struct super_block *sb);
3189 extern void drop_super_exclusive(struct super_block *sb);
3190 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3191 extern void iterate_supers_type(struct file_system_type *,
3192 void (*)(struct super_block *, void *), void *);
3194 extern int dcache_dir_open(struct inode *, struct file *);
3195 extern int dcache_dir_close(struct inode *, struct file *);
3196 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3197 extern int dcache_readdir(struct file *, struct dir_context *);
3198 extern int simple_setattr(struct dentry *, struct iattr *);
3199 extern int simple_getattr(const struct path *, struct kstat *, u32, unsigned int);
3200 extern int simple_statfs(struct dentry *, struct kstatfs *);
3201 extern int simple_open(struct inode *inode, struct file *file);
3202 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3203 extern int simple_unlink(struct inode *, struct dentry *);
3204 extern int simple_rmdir(struct inode *, struct dentry *);
3205 extern int simple_rename(struct inode *, struct dentry *,
3206 struct inode *, struct dentry *, unsigned int);
3207 extern void simple_recursive_removal(struct dentry *,
3208 void (*callback)(struct dentry *));
3209 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3210 extern int noop_set_page_dirty(struct page *page);
3211 extern void noop_invalidatepage(struct page *page, unsigned int offset,
3212 unsigned int length);
3213 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3214 extern int simple_empty(struct dentry *);
3215 extern int simple_readpage(struct file *file, struct page *page);
3216 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3217 loff_t pos, unsigned len, unsigned flags,
3218 struct page **pagep, void **fsdata);
3219 extern int simple_write_end(struct file *file, struct address_space *mapping,
3220 loff_t pos, unsigned len, unsigned copied,
3221 struct page *page, void *fsdata);
3222 extern int always_delete_dentry(const struct dentry *);
3223 extern struct inode *alloc_anon_inode(struct super_block *);
3224 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3225 extern const struct dentry_operations simple_dentry_operations;
3227 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3228 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3229 extern const struct file_operations simple_dir_operations;
3230 extern const struct inode_operations simple_dir_inode_operations;
3231 extern void make_empty_dir_inode(struct inode *inode);
3232 extern bool is_empty_dir_inode(struct inode *inode);
3233 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3234 struct dentry *d_alloc_name(struct dentry *, const char *);
3235 extern int simple_fill_super(struct super_block *, unsigned long,
3236 const struct tree_descr *);
3237 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3238 extern void simple_release_fs(struct vfsmount **mount, int *count);
3240 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3241 loff_t *ppos, const void *from, size_t available);
3242 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3243 const void __user *from, size_t count);
3245 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3246 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3248 extern int generic_check_addressable(unsigned, u64);
3250 #ifdef CONFIG_MIGRATION
3251 extern int buffer_migrate_page(struct address_space *,
3252 struct page *, struct page *,
3254 extern int buffer_migrate_page_norefs(struct address_space *,
3255 struct page *, struct page *,
3258 #define buffer_migrate_page NULL
3259 #define buffer_migrate_page_norefs NULL
3262 extern int setattr_prepare(struct dentry *, struct iattr *);
3263 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3264 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
3266 extern int file_update_time(struct file *file);
3268 static inline bool vma_is_dax(const struct vm_area_struct *vma)
3270 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3273 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3275 struct inode *inode;
3279 if (!vma_is_dax(vma))
3281 inode = file_inode(vma->vm_file);
3282 if (S_ISCHR(inode->i_mode))
3283 return false; /* device-dax */
3287 static inline int iocb_flags(struct file *file)
3290 if (file->f_flags & O_APPEND)
3292 if (file->f_flags & O_DIRECT)
3294 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3296 if (file->f_flags & __O_SYNC)
3301 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3303 int kiocb_flags = 0;
3307 if (unlikely(flags & ~RWF_SUPPORTED))
3310 if (flags & RWF_NOWAIT) {
3311 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3313 kiocb_flags |= IOCB_NOWAIT | IOCB_NOIO;
3315 if (flags & RWF_HIPRI)
3316 kiocb_flags |= IOCB_HIPRI;
3317 if (flags & RWF_DSYNC)
3318 kiocb_flags |= IOCB_DSYNC;
3319 if (flags & RWF_SYNC)
3320 kiocb_flags |= (IOCB_DSYNC | IOCB_SYNC);
3321 if (flags & RWF_APPEND)
3322 kiocb_flags |= IOCB_APPEND;
3324 ki->ki_flags |= kiocb_flags;
3328 static inline ino_t parent_ino(struct dentry *dentry)
3333 * Don't strictly need d_lock here? If the parent ino could change
3334 * then surely we'd have a deeper race in the caller?
3336 spin_lock(&dentry->d_lock);
3337 res = dentry->d_parent->d_inode->i_ino;
3338 spin_unlock(&dentry->d_lock);
3342 /* Transaction based IO helpers */
3345 * An argresp is stored in an allocated page and holds the
3346 * size of the argument or response, along with its content
3348 struct simple_transaction_argresp {
3353 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3355 char *simple_transaction_get(struct file *file, const char __user *buf,
3357 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3358 size_t size, loff_t *pos);
3359 int simple_transaction_release(struct inode *inode, struct file *file);
3361 void simple_transaction_set(struct file *file, size_t n);
3364 * simple attribute files
3366 * These attributes behave similar to those in sysfs:
3368 * Writing to an attribute immediately sets a value, an open file can be
3369 * written to multiple times.
3371 * Reading from an attribute creates a buffer from the value that might get
3372 * read with multiple read calls. When the attribute has been read
3373 * completely, no further read calls are possible until the file is opened
3376 * All attributes contain a text representation of a numeric value
3377 * that are accessed with the get() and set() functions.
3379 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3380 static int __fops ## _open(struct inode *inode, struct file *file) \
3382 __simple_attr_check_format(__fmt, 0ull); \
3383 return simple_attr_open(inode, file, __get, __set, __fmt); \
3385 static const struct file_operations __fops = { \
3386 .owner = THIS_MODULE, \
3387 .open = __fops ## _open, \
3388 .release = simple_attr_release, \
3389 .read = simple_attr_read, \
3390 .write = simple_attr_write, \
3391 .llseek = generic_file_llseek, \
3394 static inline __printf(1, 2)
3395 void __simple_attr_check_format(const char *fmt, ...)
3397 /* don't do anything, just let the compiler check the arguments; */
3400 int simple_attr_open(struct inode *inode, struct file *file,
3401 int (*get)(void *, u64 *), int (*set)(void *, u64),
3403 int simple_attr_release(struct inode *inode, struct file *file);
3404 ssize_t simple_attr_read(struct file *file, char __user *buf,
3405 size_t len, loff_t *ppos);
3406 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3407 size_t len, loff_t *ppos);
3410 int proc_nr_files(struct ctl_table *table, int write,
3411 void *buffer, size_t *lenp, loff_t *ppos);
3412 int proc_nr_dentry(struct ctl_table *table, int write,
3413 void *buffer, size_t *lenp, loff_t *ppos);
3414 int proc_nr_inodes(struct ctl_table *table, int write,
3415 void *buffer, size_t *lenp, loff_t *ppos);
3416 int __init get_filesystem_list(char *buf);
3418 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3419 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3421 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3422 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3423 (flag & __FMODE_NONOTIFY)))
3425 static inline bool is_sxid(umode_t mode)
3427 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3430 static inline int check_sticky(struct inode *dir, struct inode *inode)
3432 if (!(dir->i_mode & S_ISVTX))
3435 return __check_sticky(dir, inode);
3438 static inline void inode_has_no_xattr(struct inode *inode)
3440 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3441 inode->i_flags |= S_NOSEC;
3444 static inline bool is_root_inode(struct inode *inode)
3446 return inode == inode->i_sb->s_root->d_inode;
3449 static inline bool dir_emit(struct dir_context *ctx,
3450 const char *name, int namelen,
3451 u64 ino, unsigned type)
3453 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3455 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3457 return ctx->actor(ctx, ".", 1, ctx->pos,
3458 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3460 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3462 return ctx->actor(ctx, "..", 2, ctx->pos,
3463 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3465 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3467 if (ctx->pos == 0) {
3468 if (!dir_emit_dot(file, ctx))
3472 if (ctx->pos == 1) {
3473 if (!dir_emit_dotdot(file, ctx))
3479 static inline bool dir_relax(struct inode *inode)
3481 inode_unlock(inode);
3483 return !IS_DEADDIR(inode);
3486 static inline bool dir_relax_shared(struct inode *inode)
3488 inode_unlock_shared(inode);
3489 inode_lock_shared(inode);
3490 return !IS_DEADDIR(inode);
3493 extern bool path_noexec(const struct path *path);
3494 extern void inode_nohighmem(struct inode *inode);
3497 extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len,
3499 extern int generic_fadvise(struct file *file, loff_t offset, loff_t len,
3502 #if defined(CONFIG_IO_URING)
3503 extern struct sock *io_uring_get_socket(struct file *file);
3505 static inline struct sock *io_uring_get_socket(struct file *file)
3511 int vfs_ioc_setflags_prepare(struct inode *inode, unsigned int oldflags,
3512 unsigned int flags);
3514 int vfs_ioc_fssetxattr_check(struct inode *inode, const struct fsxattr *old_fa,
3515 struct fsxattr *fa);
3517 static inline void simple_fill_fsxattr(struct fsxattr *fa, __u32 xflags)
3519 memset(fa, 0, sizeof(*fa));
3520 fa->fsx_xflags = xflags;
3524 * Flush file data before changing attributes. Caller must hold any locks
3525 * required to prevent further writes to this file until we're done setting
3528 static inline int inode_drain_writes(struct inode *inode)
3530 inode_dio_wait(inode);
3531 return filemap_write_and_wait(inode->i_mapping);
3534 #endif /* _LINUX_FS_H */