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)
179 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
180 * that indicates that they should check the contents of the iovec are
181 * valid, but not check the memory that the iovec elements
184 #define CHECK_IOVEC_ONLY -1
187 * Attribute flags. These should be or-ed together to figure out what
190 #define ATTR_MODE (1 << 0)
191 #define ATTR_UID (1 << 1)
192 #define ATTR_GID (1 << 2)
193 #define ATTR_SIZE (1 << 3)
194 #define ATTR_ATIME (1 << 4)
195 #define ATTR_MTIME (1 << 5)
196 #define ATTR_CTIME (1 << 6)
197 #define ATTR_ATIME_SET (1 << 7)
198 #define ATTR_MTIME_SET (1 << 8)
199 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
200 #define ATTR_KILL_SUID (1 << 11)
201 #define ATTR_KILL_SGID (1 << 12)
202 #define ATTR_FILE (1 << 13)
203 #define ATTR_KILL_PRIV (1 << 14)
204 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
205 #define ATTR_TIMES_SET (1 << 16)
206 #define ATTR_TOUCH (1 << 17)
209 * Whiteout is represented by a char device. The following constants define the
210 * mode and device number to use.
212 #define WHITEOUT_MODE 0
213 #define WHITEOUT_DEV 0
216 * This is the Inode Attributes structure, used for notify_change(). It
217 * uses the above definitions as flags, to know which values have changed.
218 * Also, in this manner, a Filesystem can look at only the values it cares
219 * about. Basically, these are the attributes that the VFS layer can
220 * request to change from the FS layer.
222 * Derek Atkins <warlord@MIT.EDU> 94-10-20
225 unsigned int ia_valid;
230 struct timespec64 ia_atime;
231 struct timespec64 ia_mtime;
232 struct timespec64 ia_ctime;
235 * Not an attribute, but an auxiliary info for filesystems wanting to
236 * implement an ftruncate() like method. NOTE: filesystem should
237 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
239 struct file *ia_file;
243 * Includes for diskquotas.
245 #include <linux/quota.h>
248 * Maximum number of layers of fs stack. Needs to be limited to
249 * prevent kernel stack overflow
251 #define FILESYSTEM_MAX_STACK_DEPTH 2
254 * enum positive_aop_returns - aop return codes with specific semantics
256 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
257 * completed, that the page is still locked, and
258 * should be considered active. The VM uses this hint
259 * to return the page to the active list -- it won't
260 * be a candidate for writeback again in the near
261 * future. Other callers must be careful to unlock
262 * the page if they get this return. Returned by
265 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
266 * unlocked it and the page might have been truncated.
267 * The caller should back up to acquiring a new page and
268 * trying again. The aop will be taking reasonable
269 * precautions not to livelock. If the caller held a page
270 * reference, it should drop it before retrying. Returned
273 * address_space_operation functions return these large constants to indicate
274 * special semantics to the caller. These are much larger than the bytes in a
275 * page to allow for functions that return the number of bytes operated on in a
279 enum positive_aop_returns {
280 AOP_WRITEPAGE_ACTIVATE = 0x80000,
281 AOP_TRUNCATED_PAGE = 0x80001,
284 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
285 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
286 * helper code (eg buffer layer)
287 * to clear GFP_FS from alloc */
290 * oh the beauties of C type declarations.
293 struct address_space;
294 struct writeback_control;
295 struct readahead_control;
298 * Write life time hint values.
299 * Stored in struct inode as u8.
302 WRITE_LIFE_NOT_SET = 0,
303 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
304 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
305 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
306 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
307 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
310 #define IOCB_EVENTFD (1 << 0)
311 #define IOCB_APPEND (1 << 1)
312 #define IOCB_DIRECT (1 << 2)
313 #define IOCB_HIPRI (1 << 3)
314 #define IOCB_DSYNC (1 << 4)
315 #define IOCB_SYNC (1 << 5)
316 #define IOCB_WRITE (1 << 6)
317 #define IOCB_NOWAIT (1 << 7)
318 #define IOCB_NOIO (1 << 9)
321 struct file *ki_filp;
323 /* The 'ki_filp' pointer is shared in a union for aio */
324 randomized_struct_fields_start
327 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
331 u16 ki_ioprio; /* See linux/ioprio.h */
332 unsigned int ki_cookie; /* for ->iopoll */
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.
474 struct request_queue;
476 struct block_device {
477 dev_t bd_dev; /* not a kdev_t - it's a search key */
479 struct inode * bd_inode; /* will die */
480 struct super_block * bd_super;
481 struct mutex bd_mutex; /* open/close mutex */
485 bool bd_write_holder;
487 struct list_head bd_holder_disks;
489 struct block_device * bd_contains;
490 unsigned bd_block_size;
492 struct hd_struct * bd_part;
493 /* number of times partitions within this device have been opened. */
494 unsigned bd_part_count;
496 struct gendisk * bd_disk;
497 struct request_queue * bd_queue;
498 struct backing_dev_info *bd_bdi;
499 struct list_head bd_list;
501 * Private data. You must have bd_claim'ed the block_device
502 * to use this. NOTE: bd_claim allows an owner to claim
503 * the same device multiple times, the owner must take special
504 * care to not mess up bd_private for that case.
506 unsigned long bd_private;
508 /* The counter of freeze processes */
509 int bd_fsfreeze_count;
510 /* Mutex for freeze */
511 struct mutex bd_fsfreeze_mutex;
512 } __randomize_layout;
514 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
515 #define PAGECACHE_TAG_DIRTY XA_MARK_0
516 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
517 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
520 * Returns true if any of the pages in the mapping are marked with the tag.
522 static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag)
524 return xa_marked(&mapping->i_pages, tag);
527 static inline void i_mmap_lock_write(struct address_space *mapping)
529 down_write(&mapping->i_mmap_rwsem);
532 static inline int i_mmap_trylock_write(struct address_space *mapping)
534 return down_write_trylock(&mapping->i_mmap_rwsem);
537 static inline void i_mmap_unlock_write(struct address_space *mapping)
539 up_write(&mapping->i_mmap_rwsem);
542 static inline void i_mmap_lock_read(struct address_space *mapping)
544 down_read(&mapping->i_mmap_rwsem);
547 static inline void i_mmap_unlock_read(struct address_space *mapping)
549 up_read(&mapping->i_mmap_rwsem);
553 * Might pages of this file be mapped into userspace?
555 static inline int mapping_mapped(struct address_space *mapping)
557 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
561 * Might pages of this file have been modified in userspace?
562 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
563 * marks vma as VM_SHARED if it is shared, and the file was opened for
564 * writing i.e. vma may be mprotected writable even if now readonly.
566 * If i_mmap_writable is negative, no new writable mappings are allowed. You
567 * can only deny writable mappings, if none exists right now.
569 static inline int mapping_writably_mapped(struct address_space *mapping)
571 return atomic_read(&mapping->i_mmap_writable) > 0;
574 static inline int mapping_map_writable(struct address_space *mapping)
576 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
580 static inline void mapping_unmap_writable(struct address_space *mapping)
582 atomic_dec(&mapping->i_mmap_writable);
585 static inline int mapping_deny_writable(struct address_space *mapping)
587 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
591 static inline void mapping_allow_writable(struct address_space *mapping)
593 atomic_inc(&mapping->i_mmap_writable);
597 * Use sequence counter to get consistent i_size on 32-bit processors.
599 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
600 #include <linux/seqlock.h>
601 #define __NEED_I_SIZE_ORDERED
602 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
604 #define i_size_ordered_init(inode) do { } while (0)
608 #define ACL_NOT_CACHED ((void *)(-1))
609 #define ACL_DONT_CACHE ((void *)(-3))
611 static inline struct posix_acl *
612 uncached_acl_sentinel(struct task_struct *task)
614 return (void *)task + 1;
618 is_uncached_acl(struct posix_acl *acl)
620 return (long)acl & 1;
623 #define IOP_FASTPERM 0x0001
624 #define IOP_LOOKUP 0x0002
625 #define IOP_NOFOLLOW 0x0004
626 #define IOP_XATTR 0x0008
627 #define IOP_DEFAULT_READLINK 0x0010
629 struct fsnotify_mark_connector;
632 * Keep mostly read-only and often accessed (especially for
633 * the RCU path lookup and 'stat' data) fields at the beginning
634 * of the 'struct inode'
638 unsigned short i_opflags;
641 unsigned int i_flags;
643 #ifdef CONFIG_FS_POSIX_ACL
644 struct posix_acl *i_acl;
645 struct posix_acl *i_default_acl;
648 const struct inode_operations *i_op;
649 struct super_block *i_sb;
650 struct address_space *i_mapping;
652 #ifdef CONFIG_SECURITY
656 /* Stat data, not accessed from path walking */
659 * Filesystems may only read i_nlink directly. They shall use the
660 * following functions for modification:
662 * (set|clear|inc|drop)_nlink
663 * inode_(inc|dec)_link_count
666 const unsigned int i_nlink;
667 unsigned int __i_nlink;
671 struct timespec64 i_atime;
672 struct timespec64 i_mtime;
673 struct timespec64 i_ctime;
674 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
675 unsigned short i_bytes;
680 #ifdef __NEED_I_SIZE_ORDERED
681 seqcount_t i_size_seqcount;
685 unsigned long i_state;
686 struct rw_semaphore i_rwsem;
688 unsigned long dirtied_when; /* jiffies of first dirtying */
689 unsigned long dirtied_time_when;
691 struct hlist_node i_hash;
692 struct list_head i_io_list; /* backing dev IO list */
693 #ifdef CONFIG_CGROUP_WRITEBACK
694 struct bdi_writeback *i_wb; /* the associated cgroup wb */
696 /* foreign inode detection, see wbc_detach_inode() */
698 u16 i_wb_frn_avg_time;
699 u16 i_wb_frn_history;
701 struct list_head i_lru; /* inode LRU list */
702 struct list_head i_sb_list;
703 struct list_head i_wb_list; /* backing dev writeback list */
705 struct hlist_head i_dentry;
706 struct rcu_head i_rcu;
708 atomic64_t i_version;
709 atomic64_t i_sequence; /* see futex */
711 atomic_t i_dio_count;
712 atomic_t i_writecount;
713 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
714 atomic_t i_readcount; /* struct files open RO */
717 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
718 void (*free_inode)(struct inode *);
720 struct file_lock_context *i_flctx;
721 struct address_space i_data;
722 struct list_head i_devices;
724 struct pipe_inode_info *i_pipe;
725 struct block_device *i_bdev;
733 #ifdef CONFIG_FSNOTIFY
734 __u32 i_fsnotify_mask; /* all events this inode cares about */
735 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
738 #ifdef CONFIG_FS_ENCRYPTION
739 struct fscrypt_info *i_crypt_info;
742 #ifdef CONFIG_FS_VERITY
743 struct fsverity_info *i_verity_info;
746 void *i_private; /* fs or device private pointer */
747 } __randomize_layout;
749 struct timespec64 timestamp_truncate(struct timespec64 t, struct inode *inode);
751 static inline unsigned int i_blocksize(const struct inode *node)
753 return (1 << node->i_blkbits);
756 static inline int inode_unhashed(struct inode *inode)
758 return hlist_unhashed(&inode->i_hash);
762 * __mark_inode_dirty expects inodes to be hashed. Since we don't
763 * want special inodes in the fileset inode space, we make them
764 * appear hashed, but do not put on any lists. hlist_del()
765 * will work fine and require no locking.
767 static inline void inode_fake_hash(struct inode *inode)
769 hlist_add_fake(&inode->i_hash);
773 * inode->i_mutex nesting subclasses for the lock validator:
775 * 0: the object of the current VFS operation
779 * 4: second non-directory
780 * 5: second parent (when locking independent directories in rename)
782 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
783 * non-directories at once.
785 * The locking order between these classes is
786 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
788 enum inode_i_mutex_lock_class
798 static inline void inode_lock(struct inode *inode)
800 down_write(&inode->i_rwsem);
803 static inline void inode_unlock(struct inode *inode)
805 up_write(&inode->i_rwsem);
808 static inline void inode_lock_shared(struct inode *inode)
810 down_read(&inode->i_rwsem);
813 static inline void inode_unlock_shared(struct inode *inode)
815 up_read(&inode->i_rwsem);
818 static inline int inode_trylock(struct inode *inode)
820 return down_write_trylock(&inode->i_rwsem);
823 static inline int inode_trylock_shared(struct inode *inode)
825 return down_read_trylock(&inode->i_rwsem);
828 static inline int inode_is_locked(struct inode *inode)
830 return rwsem_is_locked(&inode->i_rwsem);
833 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
835 down_write_nested(&inode->i_rwsem, subclass);
838 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
840 down_read_nested(&inode->i_rwsem, subclass);
843 void lock_two_nondirectories(struct inode *, struct inode*);
844 void unlock_two_nondirectories(struct inode *, struct inode*);
847 * NOTE: in a 32bit arch with a preemptable kernel and
848 * an UP compile the i_size_read/write must be atomic
849 * with respect to the local cpu (unlike with preempt disabled),
850 * but they don't need to be atomic with respect to other cpus like in
851 * true SMP (so they need either to either locally disable irq around
852 * the read or for example on x86 they can be still implemented as a
853 * cmpxchg8b without the need of the lock prefix). For SMP compiles
854 * and 64bit archs it makes no difference if preempt is enabled or not.
856 static inline loff_t i_size_read(const struct inode *inode)
858 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
863 seq = read_seqcount_begin(&inode->i_size_seqcount);
864 i_size = inode->i_size;
865 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
867 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
871 i_size = inode->i_size;
875 return inode->i_size;
880 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
881 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
882 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
884 static inline void i_size_write(struct inode *inode, loff_t i_size)
886 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
888 write_seqcount_begin(&inode->i_size_seqcount);
889 inode->i_size = i_size;
890 write_seqcount_end(&inode->i_size_seqcount);
892 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
894 inode->i_size = i_size;
897 inode->i_size = i_size;
901 static inline unsigned iminor(const struct inode *inode)
903 return MINOR(inode->i_rdev);
906 static inline unsigned imajor(const struct inode *inode)
908 return MAJOR(inode->i_rdev);
911 extern struct block_device *I_BDEV(struct inode *inode);
914 rwlock_t lock; /* protects pid, uid, euid fields */
915 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
916 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
917 kuid_t uid, euid; /* uid/euid of process setting the owner */
918 int signum; /* posix.1b rt signal to be delivered on IO */
922 * Track a single file's readahead state
924 struct file_ra_state {
925 pgoff_t start; /* where readahead started */
926 unsigned int size; /* # of readahead pages */
927 unsigned int async_size; /* do asynchronous readahead when
928 there are only # of pages ahead */
930 unsigned int ra_pages; /* Maximum readahead window */
931 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
932 loff_t prev_pos; /* Cache last read() position */
936 * Check if @index falls in the readahead windows.
938 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
940 return (index >= ra->start &&
941 index < ra->start + ra->size);
946 struct llist_node fu_llist;
947 struct rcu_head fu_rcuhead;
950 struct inode *f_inode; /* cached value */
951 const struct file_operations *f_op;
954 * Protects f_ep_links, f_flags.
955 * Must not be taken from IRQ context.
958 enum rw_hint f_write_hint;
959 atomic_long_t f_count;
960 unsigned int f_flags;
962 struct mutex f_pos_lock;
964 struct fown_struct f_owner;
965 const struct cred *f_cred;
966 struct file_ra_state f_ra;
969 #ifdef CONFIG_SECURITY
972 /* needed for tty driver, and maybe others */
976 /* Used by fs/eventpoll.c to link all the hooks to this file */
977 struct list_head f_ep_links;
978 struct list_head f_tfile_llink;
979 #endif /* #ifdef CONFIG_EPOLL */
980 struct address_space *f_mapping;
982 errseq_t f_sb_err; /* for syncfs */
984 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
989 /* file identifier */
990 unsigned char f_handle[];
993 static inline struct file *get_file(struct file *f)
995 atomic_long_inc(&f->f_count);
998 #define get_file_rcu_many(x, cnt) \
999 atomic_long_add_unless(&(x)->f_count, (cnt), 0)
1000 #define get_file_rcu(x) get_file_rcu_many((x), 1)
1001 #define file_count(x) atomic_long_read(&(x)->f_count)
1003 #define MAX_NON_LFS ((1UL<<31) - 1)
1005 /* Page cache limit. The filesystems should put that into their s_maxbytes
1006 limits, otherwise bad things can happen in VM. */
1007 #if BITS_PER_LONG==32
1008 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
1009 #elif BITS_PER_LONG==64
1010 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
1015 #define FL_DELEG 4 /* NFSv4 delegation */
1016 #define FL_ACCESS 8 /* not trying to lock, just looking */
1017 #define FL_EXISTS 16 /* when unlocking, test for existence */
1018 #define FL_LEASE 32 /* lease held on this file */
1019 #define FL_CLOSE 64 /* unlock on close */
1020 #define FL_SLEEP 128 /* A blocking lock */
1021 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
1022 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
1023 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
1024 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
1026 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
1029 * Special return value from posix_lock_file() and vfs_lock_file() for
1030 * asynchronous locking.
1032 #define FILE_LOCK_DEFERRED 1
1034 /* legacy typedef, should eventually be removed */
1035 typedef void *fl_owner_t;
1039 struct file_lock_operations {
1040 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1041 void (*fl_release_private)(struct file_lock *);
1044 struct lock_manager_operations {
1045 fl_owner_t (*lm_get_owner)(fl_owner_t);
1046 void (*lm_put_owner)(fl_owner_t);
1047 void (*lm_notify)(struct file_lock *); /* unblock callback */
1048 int (*lm_grant)(struct file_lock *, int);
1049 bool (*lm_break)(struct file_lock *);
1050 int (*lm_change)(struct file_lock *, int, struct list_head *);
1051 void (*lm_setup)(struct file_lock *, void **);
1052 bool (*lm_breaker_owns_lease)(struct file_lock *);
1055 struct lock_manager {
1056 struct list_head list;
1058 * NFSv4 and up also want opens blocked during the grace period;
1065 void locks_start_grace(struct net *, struct lock_manager *);
1066 void locks_end_grace(struct lock_manager *);
1067 bool locks_in_grace(struct net *);
1068 bool opens_in_grace(struct net *);
1070 /* that will die - we need it for nfs_lock_info */
1071 #include <linux/nfs_fs_i.h>
1074 * struct file_lock represents a generic "file lock". It's used to represent
1075 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1076 * note that the same struct is used to represent both a request for a lock and
1077 * the lock itself, but the same object is never used for both.
1079 * FIXME: should we create a separate "struct lock_request" to help distinguish
1082 * The varous i_flctx lists are ordered by:
1085 * 2) lock range start
1088 * Obviously, the last two criteria only matter for POSIX locks.
1091 struct file_lock *fl_blocker; /* The lock, that is blocking us */
1092 struct list_head fl_list; /* link into file_lock_context */
1093 struct hlist_node fl_link; /* node in global lists */
1094 struct list_head fl_blocked_requests; /* list of requests with
1095 * ->fl_blocker pointing here
1097 struct list_head fl_blocked_member; /* node in
1098 * ->fl_blocker->fl_blocked_requests
1100 fl_owner_t fl_owner;
1101 unsigned int fl_flags;
1102 unsigned char fl_type;
1103 unsigned int fl_pid;
1104 int fl_link_cpu; /* what cpu's list is this on? */
1105 wait_queue_head_t fl_wait;
1106 struct file *fl_file;
1110 struct fasync_struct * fl_fasync; /* for lease break notifications */
1111 /* for lease breaks: */
1112 unsigned long fl_break_time;
1113 unsigned long fl_downgrade_time;
1115 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1116 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1118 struct nfs_lock_info nfs_fl;
1119 struct nfs4_lock_info nfs4_fl;
1121 struct list_head link; /* link in AFS vnode's pending_locks list */
1122 int state; /* state of grant or error if -ve */
1123 unsigned int debug_id;
1126 } __randomize_layout;
1128 struct file_lock_context {
1129 spinlock_t flc_lock;
1130 struct list_head flc_flock;
1131 struct list_head flc_posix;
1132 struct list_head flc_lease;
1135 /* The following constant reflects the upper bound of the file/locking space */
1137 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1138 #define OFFSET_MAX INT_LIMIT(loff_t)
1139 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1142 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1144 #define locks_inode(f) file_inode(f)
1146 #ifdef CONFIG_FILE_LOCKING
1147 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1148 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1151 #if BITS_PER_LONG == 32
1152 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1153 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1157 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1158 extern int fcntl_getlease(struct file *filp);
1161 void locks_free_lock_context(struct inode *inode);
1162 void locks_free_lock(struct file_lock *fl);
1163 extern void locks_init_lock(struct file_lock *);
1164 extern struct file_lock * locks_alloc_lock(void);
1165 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1166 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1167 extern void locks_remove_posix(struct file *, fl_owner_t);
1168 extern void locks_remove_file(struct file *);
1169 extern void locks_release_private(struct file_lock *);
1170 extern void posix_test_lock(struct file *, struct file_lock *);
1171 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1172 extern int locks_delete_block(struct file_lock *);
1173 extern int vfs_test_lock(struct file *, struct file_lock *);
1174 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1175 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1176 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1177 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1178 extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1179 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1180 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1181 extern int lease_modify(struct file_lock *, int, struct list_head *);
1183 struct notifier_block;
1184 extern int lease_register_notifier(struct notifier_block *);
1185 extern void lease_unregister_notifier(struct notifier_block *);
1187 struct files_struct;
1188 extern void show_fd_locks(struct seq_file *f,
1189 struct file *filp, struct files_struct *files);
1190 #else /* !CONFIG_FILE_LOCKING */
1191 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1192 struct flock __user *user)
1197 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1198 unsigned int cmd, struct flock __user *user)
1203 #if BITS_PER_LONG == 32
1204 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1205 struct flock64 __user *user)
1210 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1211 unsigned int cmd, struct flock64 __user *user)
1216 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1221 static inline int fcntl_getlease(struct file *filp)
1227 locks_free_lock_context(struct inode *inode)
1231 static inline void locks_init_lock(struct file_lock *fl)
1236 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1241 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1246 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1251 static inline void locks_remove_file(struct file *filp)
1256 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1261 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1262 struct file_lock *conflock)
1267 static inline int locks_delete_block(struct file_lock *waiter)
1272 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1277 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1278 struct file_lock *fl, struct file_lock *conf)
1283 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1288 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1293 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1298 static inline void lease_get_mtime(struct inode *inode,
1299 struct timespec64 *time)
1304 static inline int generic_setlease(struct file *filp, long arg,
1305 struct file_lock **flp, void **priv)
1310 static inline int vfs_setlease(struct file *filp, long arg,
1311 struct file_lock **lease, void **priv)
1316 static inline int lease_modify(struct file_lock *fl, int arg,
1317 struct list_head *dispose)
1322 struct files_struct;
1323 static inline void show_fd_locks(struct seq_file *f,
1324 struct file *filp, struct files_struct *files) {}
1325 #endif /* !CONFIG_FILE_LOCKING */
1327 static inline struct inode *file_inode(const struct file *f)
1332 static inline struct dentry *file_dentry(const struct file *file)
1334 return d_real(file->f_path.dentry, file_inode(file));
1337 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1339 return locks_lock_inode_wait(locks_inode(filp), fl);
1342 struct fasync_struct {
1346 struct fasync_struct *fa_next; /* singly linked list */
1347 struct file *fa_file;
1348 struct rcu_head fa_rcu;
1351 #define FASYNC_MAGIC 0x4601
1353 /* SMP safe fasync helpers: */
1354 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1355 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1356 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1357 extern struct fasync_struct *fasync_alloc(void);
1358 extern void fasync_free(struct fasync_struct *);
1360 /* can be called from interrupts */
1361 extern void kill_fasync(struct fasync_struct **, int, int);
1363 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1364 extern int f_setown(struct file *filp, unsigned long arg, int force);
1365 extern void f_delown(struct file *filp);
1366 extern pid_t f_getown(struct file *filp);
1367 extern int send_sigurg(struct fown_struct *fown);
1370 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1371 * represented in both.
1373 #define SB_RDONLY 1 /* Mount read-only */
1374 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1375 #define SB_NODEV 4 /* Disallow access to device special files */
1376 #define SB_NOEXEC 8 /* Disallow program execution */
1377 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1378 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1379 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1380 #define SB_NOATIME 1024 /* Do not update access times. */
1381 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1382 #define SB_SILENT 32768
1383 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1384 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1385 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1386 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1388 /* These sb flags are internal to the kernel */
1389 #define SB_SUBMOUNT (1<<26)
1390 #define SB_FORCE (1<<27)
1391 #define SB_NOSEC (1<<28)
1392 #define SB_BORN (1<<29)
1393 #define SB_ACTIVE (1<<30)
1394 #define SB_NOUSER (1<<31)
1400 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1401 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1402 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1403 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1404 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1407 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1408 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1409 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1410 #define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */
1412 /* sb->s_iflags to limit user namespace mounts */
1413 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1414 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1415 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1417 #define SB_I_SKIP_SYNC 0x00000100 /* Skip superblock at global sync */
1419 /* Possible states of 'frozen' field */
1421 SB_UNFROZEN = 0, /* FS is unfrozen */
1422 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1423 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1424 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1425 * internal threads if needed) */
1426 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1429 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1432 int frozen; /* Is sb frozen? */
1433 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1434 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1437 struct super_block {
1438 struct list_head s_list; /* Keep this first */
1439 dev_t s_dev; /* search index; _not_ kdev_t */
1440 unsigned char s_blocksize_bits;
1441 unsigned long s_blocksize;
1442 loff_t s_maxbytes; /* Max file size */
1443 struct file_system_type *s_type;
1444 const struct super_operations *s_op;
1445 const struct dquot_operations *dq_op;
1446 const struct quotactl_ops *s_qcop;
1447 const struct export_operations *s_export_op;
1448 unsigned long s_flags;
1449 unsigned long s_iflags; /* internal SB_I_* flags */
1450 unsigned long s_magic;
1451 struct dentry *s_root;
1452 struct rw_semaphore s_umount;
1455 #ifdef CONFIG_SECURITY
1458 const struct xattr_handler **s_xattr;
1459 #ifdef CONFIG_FS_ENCRYPTION
1460 const struct fscrypt_operations *s_cop;
1461 struct key *s_master_keys; /* master crypto keys in use */
1463 #ifdef CONFIG_FS_VERITY
1464 const struct fsverity_operations *s_vop;
1466 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1467 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1468 struct block_device *s_bdev;
1469 struct backing_dev_info *s_bdi;
1470 struct mtd_info *s_mtd;
1471 struct hlist_node s_instances;
1472 unsigned int s_quota_types; /* Bitmask of supported quota types */
1473 struct quota_info s_dquot; /* Diskquota specific options */
1475 struct sb_writers s_writers;
1478 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1479 * s_fsnotify_marks together for cache efficiency. They are frequently
1480 * accessed and rarely modified.
1482 void *s_fs_info; /* Filesystem private info */
1484 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1486 /* Time limits for c/m/atime in seconds */
1487 time64_t s_time_min;
1488 time64_t s_time_max;
1489 #ifdef CONFIG_FSNOTIFY
1490 __u32 s_fsnotify_mask;
1491 struct fsnotify_mark_connector __rcu *s_fsnotify_marks;
1494 char s_id[32]; /* Informational name */
1495 uuid_t s_uuid; /* UUID */
1497 unsigned int s_max_links;
1501 * The next field is for VFS *only*. No filesystems have any business
1502 * even looking at it. You had been warned.
1504 struct mutex s_vfs_rename_mutex; /* Kludge */
1507 * Filesystem subtype. If non-empty the filesystem type field
1508 * in /proc/mounts will be "type.subtype"
1510 const char *s_subtype;
1512 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1515 * Saved pool identifier for cleancache (-1 means none)
1517 int cleancache_poolid;
1519 struct shrinker s_shrink; /* per-sb shrinker handle */
1521 /* Number of inodes with nlink == 0 but still referenced */
1522 atomic_long_t s_remove_count;
1524 /* Pending fsnotify inode refs */
1525 atomic_long_t s_fsnotify_inode_refs;
1527 /* Being remounted read-only */
1528 int s_readonly_remount;
1530 /* per-sb errseq_t for reporting writeback errors via syncfs */
1533 /* AIO completions deferred from interrupt context */
1534 struct workqueue_struct *s_dio_done_wq;
1535 struct hlist_head s_pins;
1538 * Owning user namespace and default context in which to
1539 * interpret filesystem uids, gids, quotas, device nodes,
1540 * xattrs and security labels.
1542 struct user_namespace *s_user_ns;
1545 * The list_lru structure is essentially just a pointer to a table
1546 * of per-node lru lists, each of which has its own spinlock.
1547 * There is no need to put them into separate cachelines.
1549 struct list_lru s_dentry_lru;
1550 struct list_lru s_inode_lru;
1551 struct rcu_head rcu;
1552 struct work_struct destroy_work;
1554 struct mutex s_sync_lock; /* sync serialisation lock */
1557 * Indicates how deep in a filesystem stack this SB is
1561 /* s_inode_list_lock protects s_inodes */
1562 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1563 struct list_head s_inodes; /* all inodes */
1565 spinlock_t s_inode_wblist_lock;
1566 struct list_head s_inodes_wb; /* writeback inodes */
1567 } __randomize_layout;
1569 /* Helper functions so that in most cases filesystems will
1570 * not need to deal directly with kuid_t and kgid_t and can
1571 * instead deal with the raw numeric values that are stored
1572 * in the filesystem.
1574 static inline uid_t i_uid_read(const struct inode *inode)
1576 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1579 static inline gid_t i_gid_read(const struct inode *inode)
1581 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1584 static inline void i_uid_write(struct inode *inode, uid_t uid)
1586 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1589 static inline void i_gid_write(struct inode *inode, gid_t gid)
1591 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1594 extern struct timespec64 current_time(struct inode *inode);
1597 * Snapshotting support.
1600 void __sb_end_write(struct super_block *sb, int level);
1601 int __sb_start_write(struct super_block *sb, int level, bool wait);
1603 #define __sb_writers_acquired(sb, lev) \
1604 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1605 #define __sb_writers_release(sb, lev) \
1606 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1609 * sb_end_write - drop write access to a superblock
1610 * @sb: the super we wrote to
1612 * Decrement number of writers to the filesystem. Wake up possible waiters
1613 * wanting to freeze the filesystem.
1615 static inline void sb_end_write(struct super_block *sb)
1617 __sb_end_write(sb, SB_FREEZE_WRITE);
1621 * sb_end_pagefault - drop write access to a superblock from a page fault
1622 * @sb: the super we wrote to
1624 * Decrement number of processes handling write page fault to the filesystem.
1625 * Wake up possible waiters wanting to freeze the filesystem.
1627 static inline void sb_end_pagefault(struct super_block *sb)
1629 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1633 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1634 * @sb: the super we wrote to
1636 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1637 * waiters wanting to freeze the filesystem.
1639 static inline void sb_end_intwrite(struct super_block *sb)
1641 __sb_end_write(sb, SB_FREEZE_FS);
1645 * sb_start_write - get write access to a superblock
1646 * @sb: the super we write to
1648 * When a process wants to write data or metadata to a file system (i.e. dirty
1649 * a page or an inode), it should embed the operation in a sb_start_write() -
1650 * sb_end_write() pair to get exclusion against file system freezing. This
1651 * function increments number of writers preventing freezing. If the file
1652 * system is already frozen, the function waits until the file system is
1655 * Since freeze protection behaves as a lock, users have to preserve
1656 * ordering of freeze protection and other filesystem locks. Generally,
1657 * freeze protection should be the outermost lock. In particular, we have:
1660 * -> i_mutex (write path, truncate, directory ops, ...)
1661 * -> s_umount (freeze_super, thaw_super)
1663 static inline void sb_start_write(struct super_block *sb)
1665 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1668 static inline int sb_start_write_trylock(struct super_block *sb)
1670 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1674 * sb_start_pagefault - get write access to a superblock from a page fault
1675 * @sb: the super we write to
1677 * When a process starts handling write page fault, it should embed the
1678 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1679 * exclusion against file system freezing. This is needed since the page fault
1680 * is going to dirty a page. This function increments number of running page
1681 * faults preventing freezing. If the file system is already frozen, the
1682 * function waits until the file system is thawed.
1684 * Since page fault freeze protection behaves as a lock, users have to preserve
1685 * ordering of freeze protection and other filesystem locks. It is advised to
1686 * put sb_start_pagefault() close to mmap_lock in lock ordering. Page fault
1687 * handling code implies lock dependency:
1690 * -> sb_start_pagefault
1692 static inline void sb_start_pagefault(struct super_block *sb)
1694 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1698 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1699 * @sb: the super we write to
1701 * This is the third level of protection against filesystem freezing. It is
1702 * free for use by a filesystem. The only requirement is that it must rank
1703 * below sb_start_pagefault.
1705 * For example filesystem can call sb_start_intwrite() when starting a
1706 * transaction which somewhat eases handling of freezing for internal sources
1707 * of filesystem changes (internal fs threads, discarding preallocation on file
1710 static inline void sb_start_intwrite(struct super_block *sb)
1712 __sb_start_write(sb, SB_FREEZE_FS, true);
1715 static inline int sb_start_intwrite_trylock(struct super_block *sb)
1717 return __sb_start_write(sb, SB_FREEZE_FS, false);
1721 extern bool inode_owner_or_capable(const struct inode *inode);
1724 * VFS helper functions..
1726 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1727 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1728 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1729 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1730 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1731 extern int vfs_rmdir(struct inode *, struct dentry *);
1732 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1733 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1735 static inline int vfs_whiteout(struct inode *dir, struct dentry *dentry)
1737 return vfs_mknod(dir, dentry, S_IFCHR | WHITEOUT_MODE, WHITEOUT_DEV);
1740 extern struct dentry *vfs_tmpfile(struct dentry *dentry, umode_t mode,
1743 int vfs_mkobj(struct dentry *, umode_t,
1744 int (*f)(struct dentry *, umode_t, void *),
1747 int vfs_fchown(struct file *file, uid_t user, gid_t group);
1748 int vfs_fchmod(struct file *file, umode_t mode);
1750 extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1752 #ifdef CONFIG_COMPAT
1753 extern long compat_ptr_ioctl(struct file *file, unsigned int cmd,
1756 #define compat_ptr_ioctl NULL
1760 * VFS file helper functions.
1762 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1764 extern bool may_open_dev(const struct path *path);
1767 * This is the "filldir" function type, used by readdir() to let
1768 * the kernel specify what kind of dirent layout it wants to have.
1769 * This allows the kernel to read directories into kernel space or
1770 * to have different dirent layouts depending on the binary type.
1773 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1776 struct dir_context {
1781 struct block_device_operations;
1783 /* These macros are for out of kernel modules to test that
1784 * the kernel supports the unlocked_ioctl and compat_ioctl
1785 * fields in struct file_operations. */
1786 #define HAVE_COMPAT_IOCTL 1
1787 #define HAVE_UNLOCKED_IOCTL 1
1790 * These flags let !MMU mmap() govern direct device mapping vs immediate
1791 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1793 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1794 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1795 * NOMMU_MAP_READ: Can be mapped for reading
1796 * NOMMU_MAP_WRITE: Can be mapped for writing
1797 * NOMMU_MAP_EXEC: Can be mapped for execution
1799 #define NOMMU_MAP_COPY 0x00000001
1800 #define NOMMU_MAP_DIRECT 0x00000008
1801 #define NOMMU_MAP_READ VM_MAYREAD
1802 #define NOMMU_MAP_WRITE VM_MAYWRITE
1803 #define NOMMU_MAP_EXEC VM_MAYEXEC
1805 #define NOMMU_VMFLAGS \
1806 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1809 * These flags control the behavior of the remap_file_range function pointer.
1810 * If it is called with len == 0 that means "remap to end of source file".
1811 * See Documentation/filesystems/vfs.rst for more details about this call.
1813 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
1814 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
1816 #define REMAP_FILE_DEDUP (1 << 0)
1817 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
1820 * These flags signal that the caller is ok with altering various aspects of
1821 * the behavior of the remap operation. The changes must be made by the
1822 * implementation; the vfs remap helper functions can take advantage of them.
1823 * Flags in this category exist to preserve the quirky behavior of the hoisted
1824 * btrfs clone/dedupe ioctls.
1826 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
1830 struct file_operations {
1831 struct module *owner;
1832 loff_t (*llseek) (struct file *, loff_t, int);
1833 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1834 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1835 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1836 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1837 int (*iopoll)(struct kiocb *kiocb, bool spin);
1838 int (*iterate) (struct file *, struct dir_context *);
1839 int (*iterate_shared) (struct file *, struct dir_context *);
1840 __poll_t (*poll) (struct file *, struct poll_table_struct *);
1841 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1842 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1843 int (*mmap) (struct file *, struct vm_area_struct *);
1844 unsigned long mmap_supported_flags;
1845 int (*open) (struct inode *, struct file *);
1846 int (*flush) (struct file *, fl_owner_t id);
1847 int (*release) (struct inode *, struct file *);
1848 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1849 int (*fasync) (int, struct file *, int);
1850 int (*lock) (struct file *, int, struct file_lock *);
1851 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1852 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1853 int (*check_flags)(int);
1854 int (*flock) (struct file *, int, struct file_lock *);
1855 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1856 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1857 int (*setlease)(struct file *, long, struct file_lock **, void **);
1858 long (*fallocate)(struct file *file, int mode, loff_t offset,
1860 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1862 unsigned (*mmap_capabilities)(struct file *);
1864 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1865 loff_t, size_t, unsigned int);
1866 loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
1867 struct file *file_out, loff_t pos_out,
1868 loff_t len, unsigned int remap_flags);
1869 int (*fadvise)(struct file *, loff_t, loff_t, int);
1870 } __randomize_layout;
1872 struct inode_operations {
1873 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1874 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1875 int (*permission) (struct inode *, int);
1876 struct posix_acl * (*get_acl)(struct inode *, int);
1878 int (*readlink) (struct dentry *, char __user *,int);
1880 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1881 int (*link) (struct dentry *,struct inode *,struct dentry *);
1882 int (*unlink) (struct inode *,struct dentry *);
1883 int (*symlink) (struct inode *,struct dentry *,const char *);
1884 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1885 int (*rmdir) (struct inode *,struct dentry *);
1886 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1887 int (*rename) (struct inode *, struct dentry *,
1888 struct inode *, struct dentry *, unsigned int);
1889 int (*setattr) (struct dentry *, struct iattr *);
1890 int (*getattr) (const struct path *, struct kstat *, u32, unsigned int);
1891 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1892 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1894 int (*update_time)(struct inode *, struct timespec64 *, int);
1895 int (*atomic_open)(struct inode *, struct dentry *,
1896 struct file *, unsigned open_flag,
1897 umode_t create_mode);
1898 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1899 int (*set_acl)(struct inode *, struct posix_acl *, int);
1900 } ____cacheline_aligned;
1902 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
1903 struct iov_iter *iter)
1905 return file->f_op->read_iter(kio, iter);
1908 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
1909 struct iov_iter *iter)
1911 return file->f_op->write_iter(kio, iter);
1914 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
1916 return file->f_op->mmap(file, vma);
1919 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1920 unsigned long nr_segs, unsigned long fast_segs,
1921 struct iovec *fast_pointer,
1922 struct iovec **ret_pointer);
1924 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1925 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1926 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1927 unsigned long, loff_t *, rwf_t);
1928 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1929 loff_t, size_t, unsigned int);
1930 extern ssize_t generic_copy_file_range(struct file *file_in, loff_t pos_in,
1931 struct file *file_out, loff_t pos_out,
1932 size_t len, unsigned int flags);
1933 extern int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
1934 struct file *file_out, loff_t pos_out,
1936 unsigned int remap_flags);
1937 extern loff_t do_clone_file_range(struct file *file_in, loff_t pos_in,
1938 struct file *file_out, loff_t pos_out,
1939 loff_t len, unsigned int remap_flags);
1940 extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1941 struct file *file_out, loff_t pos_out,
1942 loff_t len, unsigned int remap_flags);
1943 extern int vfs_dedupe_file_range(struct file *file,
1944 struct file_dedupe_range *same);
1945 extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
1946 struct file *dst_file, loff_t dst_pos,
1947 loff_t len, unsigned int remap_flags);
1950 struct super_operations {
1951 struct inode *(*alloc_inode)(struct super_block *sb);
1952 void (*destroy_inode)(struct inode *);
1953 void (*free_inode)(struct inode *);
1955 void (*dirty_inode) (struct inode *, int flags);
1956 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1957 int (*drop_inode) (struct inode *);
1958 void (*evict_inode) (struct inode *);
1959 void (*put_super) (struct super_block *);
1960 int (*sync_fs)(struct super_block *sb, int wait);
1961 int (*freeze_super) (struct super_block *);
1962 int (*freeze_fs) (struct super_block *);
1963 int (*thaw_super) (struct super_block *);
1964 int (*unfreeze_fs) (struct super_block *);
1965 int (*statfs) (struct dentry *, struct kstatfs *);
1966 int (*remount_fs) (struct super_block *, int *, char *);
1967 void (*umount_begin) (struct super_block *);
1969 int (*show_options)(struct seq_file *, struct dentry *);
1970 int (*show_devname)(struct seq_file *, struct dentry *);
1971 int (*show_path)(struct seq_file *, struct dentry *);
1972 int (*show_stats)(struct seq_file *, struct dentry *);
1974 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1975 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1976 struct dquot **(*get_dquots)(struct inode *);
1978 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1979 long (*nr_cached_objects)(struct super_block *,
1980 struct shrink_control *);
1981 long (*free_cached_objects)(struct super_block *,
1982 struct shrink_control *);
1986 * Inode flags - they have no relation to superblock flags now
1988 #define S_SYNC 1 /* Writes are synced at once */
1989 #define S_NOATIME 2 /* Do not update access times */
1990 #define S_APPEND 4 /* Append-only file */
1991 #define S_IMMUTABLE 8 /* Immutable file */
1992 #define S_DEAD 16 /* removed, but still open directory */
1993 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1994 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1995 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1996 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1997 #define S_PRIVATE 512 /* Inode is fs-internal */
1998 #define S_IMA 1024 /* Inode has an associated IMA struct */
1999 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
2000 #define S_NOSEC 4096 /* no suid or xattr security attributes */
2001 #ifdef CONFIG_FS_DAX
2002 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
2004 #define S_DAX 0 /* Make all the DAX code disappear */
2006 #define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */
2007 #define S_CASEFOLD 32768 /* Casefolded file */
2008 #define S_VERITY 65536 /* Verity file (using fs/verity/) */
2011 * Note that nosuid etc flags are inode-specific: setting some file-system
2012 * flags just means all the inodes inherit those flags by default. It might be
2013 * possible to override it selectively if you really wanted to with some
2014 * ioctl() that is not currently implemented.
2016 * Exception: SB_RDONLY is always applied to the entire file system.
2018 * Unfortunately, it is possible to change a filesystems flags with it mounted
2019 * with files in use. This means that all of the inodes will not have their
2020 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
2021 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
2023 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
2025 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
2026 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
2027 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
2028 ((inode)->i_flags & S_SYNC))
2029 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
2030 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
2031 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
2032 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
2033 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
2035 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
2036 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
2037 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
2038 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
2040 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
2041 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
2042 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
2043 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
2044 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
2045 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2046 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2047 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2048 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2049 #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD)
2050 #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY)
2052 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2053 (inode)->i_rdev == WHITEOUT_DEV)
2055 static inline bool HAS_UNMAPPED_ID(struct inode *inode)
2057 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
2060 static inline enum rw_hint file_write_hint(struct file *file)
2062 if (file->f_write_hint != WRITE_LIFE_NOT_SET)
2063 return file->f_write_hint;
2065 return file_inode(file)->i_write_hint;
2068 static inline int iocb_flags(struct file *file);
2070 static inline u16 ki_hint_validate(enum rw_hint hint)
2072 typeof(((struct kiocb *)0)->ki_hint) max_hint = -1;
2074 if (hint <= max_hint)
2079 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
2081 *kiocb = (struct kiocb) {
2083 .ki_flags = iocb_flags(filp),
2084 .ki_hint = ki_hint_validate(file_write_hint(filp)),
2085 .ki_ioprio = get_current_ioprio(),
2089 static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src,
2092 *kiocb = (struct kiocb) {
2094 .ki_flags = kiocb_src->ki_flags,
2095 .ki_hint = kiocb_src->ki_hint,
2096 .ki_ioprio = kiocb_src->ki_ioprio,
2097 .ki_pos = kiocb_src->ki_pos,
2102 * Inode state bits. Protected by inode->i_lock
2104 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
2105 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
2107 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2108 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2109 * various stages of removing an inode.
2111 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2113 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2114 * fdatasync(). i_atime is the usual cause.
2115 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2116 * these changes separately from I_DIRTY_SYNC so that we
2117 * don't have to write inode on fdatasync() when only
2118 * mtime has changed in it.
2119 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2120 * I_NEW Serves as both a mutex and completion notification.
2121 * New inodes set I_NEW. If two processes both create
2122 * the same inode, one of them will release its inode and
2123 * wait for I_NEW to be released before returning.
2124 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2125 * also cause waiting on I_NEW, without I_NEW actually
2126 * being set. find_inode() uses this to prevent returning
2127 * nearly-dead inodes.
2128 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2129 * is zero. I_FREEING must be set when I_WILL_FREE is
2131 * I_FREEING Set when inode is about to be freed but still has dirty
2132 * pages or buffers attached or the inode itself is still
2134 * I_CLEAR Added by clear_inode(). In this state the inode is
2135 * clean and can be destroyed. Inode keeps I_FREEING.
2137 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2138 * prohibited for many purposes. iget() must wait for
2139 * the inode to be completely released, then create it
2140 * anew. Other functions will just ignore such inodes,
2141 * if appropriate. I_NEW is used for waiting.
2143 * I_SYNC Writeback of inode is running. The bit is set during
2144 * data writeback, and cleared with a wakeup on the bit
2145 * address once it is done. The bit is also used to pin
2146 * the inode in memory for flusher thread.
2148 * I_REFERENCED Marks the inode as recently references on the LRU list.
2150 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2152 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2153 * synchronize competing switching instances and to tell
2154 * wb stat updates to grab the i_pages lock. See
2155 * inode_switch_wbs_work_fn() for details.
2157 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2158 * and work dirs among overlayfs mounts.
2160 * I_CREATING New object's inode in the middle of setting up.
2162 * I_DONTCACHE Evict inode as soon as it is not used anymore.
2164 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2166 #define I_DIRTY_SYNC (1 << 0)
2167 #define I_DIRTY_DATASYNC (1 << 1)
2168 #define I_DIRTY_PAGES (1 << 2)
2170 #define I_NEW (1 << __I_NEW)
2171 #define I_WILL_FREE (1 << 4)
2172 #define I_FREEING (1 << 5)
2173 #define I_CLEAR (1 << 6)
2175 #define I_SYNC (1 << __I_SYNC)
2176 #define I_REFERENCED (1 << 8)
2177 #define __I_DIO_WAKEUP 9
2178 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2179 #define I_LINKABLE (1 << 10)
2180 #define I_DIRTY_TIME (1 << 11)
2181 #define __I_DIRTY_TIME_EXPIRED 12
2182 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
2183 #define I_WB_SWITCH (1 << 13)
2184 #define I_OVL_INUSE (1 << 14)
2185 #define I_CREATING (1 << 15)
2186 #define I_DONTCACHE (1 << 16)
2188 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2189 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2190 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2192 extern void __mark_inode_dirty(struct inode *, int);
2193 static inline void mark_inode_dirty(struct inode *inode)
2195 __mark_inode_dirty(inode, I_DIRTY);
2198 static inline void mark_inode_dirty_sync(struct inode *inode)
2200 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2203 extern void inc_nlink(struct inode *inode);
2204 extern void drop_nlink(struct inode *inode);
2205 extern void clear_nlink(struct inode *inode);
2206 extern void set_nlink(struct inode *inode, unsigned int nlink);
2208 static inline void inode_inc_link_count(struct inode *inode)
2211 mark_inode_dirty(inode);
2214 static inline void inode_dec_link_count(struct inode *inode)
2217 mark_inode_dirty(inode);
2220 enum file_time_flags {
2227 extern bool atime_needs_update(const struct path *, struct inode *);
2228 extern void touch_atime(const struct path *);
2229 static inline void file_accessed(struct file *file)
2231 if (!(file->f_flags & O_NOATIME))
2232 touch_atime(&file->f_path);
2235 extern int file_modified(struct file *file);
2237 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2238 int sync_inode_metadata(struct inode *inode, int wait);
2240 struct file_system_type {
2243 #define FS_REQUIRES_DEV 1
2244 #define FS_BINARY_MOUNTDATA 2
2245 #define FS_HAS_SUBTYPE 4
2246 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2247 #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */
2248 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2249 int (*init_fs_context)(struct fs_context *);
2250 const struct fs_parameter_spec *parameters;
2251 struct dentry *(*mount) (struct file_system_type *, int,
2252 const char *, void *);
2253 void (*kill_sb) (struct super_block *);
2254 struct module *owner;
2255 struct file_system_type * next;
2256 struct hlist_head fs_supers;
2258 struct lock_class_key s_lock_key;
2259 struct lock_class_key s_umount_key;
2260 struct lock_class_key s_vfs_rename_key;
2261 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2263 struct lock_class_key i_lock_key;
2264 struct lock_class_key i_mutex_key;
2265 struct lock_class_key i_mutex_dir_key;
2268 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2271 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2272 int flags, const char *dev_name, void *data,
2273 int (*fill_super)(struct super_block *, void *, int));
2275 static inline struct dentry *mount_bdev(struct file_system_type *fs_type,
2276 int flags, const char *dev_name, void *data,
2277 int (*fill_super)(struct super_block *, void *, int))
2279 return ERR_PTR(-ENODEV);
2282 extern struct dentry *mount_single(struct file_system_type *fs_type,
2283 int flags, void *data,
2284 int (*fill_super)(struct super_block *, void *, int));
2285 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2286 int flags, void *data,
2287 int (*fill_super)(struct super_block *, void *, int));
2288 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2289 void generic_shutdown_super(struct super_block *sb);
2291 void kill_block_super(struct super_block *sb);
2293 static inline void kill_block_super(struct super_block *sb)
2298 void kill_anon_super(struct super_block *sb);
2299 void kill_litter_super(struct super_block *sb);
2300 void deactivate_super(struct super_block *sb);
2301 void deactivate_locked_super(struct super_block *sb);
2302 int set_anon_super(struct super_block *s, void *data);
2303 int set_anon_super_fc(struct super_block *s, struct fs_context *fc);
2304 int get_anon_bdev(dev_t *);
2305 void free_anon_bdev(dev_t);
2306 struct super_block *sget_fc(struct fs_context *fc,
2307 int (*test)(struct super_block *, struct fs_context *),
2308 int (*set)(struct super_block *, struct fs_context *));
2309 struct super_block *sget(struct file_system_type *type,
2310 int (*test)(struct super_block *,void *),
2311 int (*set)(struct super_block *,void *),
2312 int flags, void *data);
2314 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2315 #define fops_get(fops) \
2316 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2317 #define fops_put(fops) \
2318 do { if (fops) module_put((fops)->owner); } while(0)
2320 * This one is to be used *ONLY* from ->open() instances.
2321 * fops must be non-NULL, pinned down *and* module dependencies
2322 * should be sufficient to pin the caller down as well.
2324 #define replace_fops(f, fops) \
2326 struct file *__file = (f); \
2327 fops_put(__file->f_op); \
2328 BUG_ON(!(__file->f_op = (fops))); \
2331 extern int register_filesystem(struct file_system_type *);
2332 extern int unregister_filesystem(struct file_system_type *);
2333 extern struct vfsmount *kern_mount(struct file_system_type *);
2334 extern void kern_unmount(struct vfsmount *mnt);
2335 extern int may_umount_tree(struct vfsmount *);
2336 extern int may_umount(struct vfsmount *);
2337 extern long do_mount(const char *, const char __user *,
2338 const char *, unsigned long, void *);
2339 extern struct vfsmount *collect_mounts(const struct path *);
2340 extern void drop_collected_mounts(struct vfsmount *);
2341 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2343 extern int vfs_statfs(const struct path *, struct kstatfs *);
2344 extern int user_statfs(const char __user *, struct kstatfs *);
2345 extern int fd_statfs(int, struct kstatfs *);
2346 extern int freeze_super(struct super_block *super);
2347 extern int thaw_super(struct super_block *super);
2348 extern bool our_mnt(struct vfsmount *mnt);
2349 extern __printf(2, 3)
2350 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2351 extern int super_setup_bdi(struct super_block *sb);
2353 extern int current_umask(void);
2355 extern void ihold(struct inode * inode);
2356 extern void iput(struct inode *);
2357 extern int generic_update_time(struct inode *, struct timespec64 *, int);
2360 extern struct kobject *fs_kobj;
2362 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2364 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2365 extern int locks_mandatory_locked(struct file *);
2366 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2369 * Candidates for mandatory locking have the setgid bit set
2370 * but no group execute bit - an otherwise meaningless combination.
2373 static inline int __mandatory_lock(struct inode *ino)
2375 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2379 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2380 * otherwise these will be advisory locks
2383 static inline int mandatory_lock(struct inode *ino)
2385 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2388 static inline int locks_verify_locked(struct file *file)
2390 if (mandatory_lock(locks_inode(file)))
2391 return locks_mandatory_locked(file);
2395 static inline int locks_verify_truncate(struct inode *inode,
2399 if (!inode->i_flctx || !mandatory_lock(inode))
2402 if (size < inode->i_size) {
2403 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2406 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2411 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2413 static inline int locks_mandatory_locked(struct file *file)
2418 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2419 loff_t start, loff_t end, unsigned char type)
2424 static inline int __mandatory_lock(struct inode *inode)
2429 static inline int mandatory_lock(struct inode *inode)
2434 static inline int locks_verify_locked(struct file *file)
2439 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2445 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2448 #ifdef CONFIG_FILE_LOCKING
2449 static inline int break_lease(struct inode *inode, unsigned int mode)
2452 * Since this check is lockless, we must ensure that any refcounts
2453 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2454 * could end up racing with tasks trying to set a new lease on this
2458 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2459 return __break_lease(inode, mode, FL_LEASE);
2463 static inline int break_deleg(struct inode *inode, unsigned int mode)
2466 * Since this check is lockless, we must ensure that any refcounts
2467 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2468 * could end up racing with tasks trying to set a new lease on this
2472 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2473 return __break_lease(inode, mode, FL_DELEG);
2477 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2481 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2482 if (ret == -EWOULDBLOCK && delegated_inode) {
2483 *delegated_inode = inode;
2489 static inline int break_deleg_wait(struct inode **delegated_inode)
2493 ret = break_deleg(*delegated_inode, O_WRONLY);
2494 iput(*delegated_inode);
2495 *delegated_inode = NULL;
2499 static inline int break_layout(struct inode *inode, bool wait)
2502 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2503 return __break_lease(inode,
2504 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2509 #else /* !CONFIG_FILE_LOCKING */
2510 static inline int break_lease(struct inode *inode, unsigned int mode)
2515 static inline int break_deleg(struct inode *inode, unsigned int mode)
2520 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2525 static inline int break_deleg_wait(struct inode **delegated_inode)
2531 static inline int break_layout(struct inode *inode, bool wait)
2536 #endif /* CONFIG_FILE_LOCKING */
2541 const char *name; /* pointer to actual string */
2542 const __user char *uptr; /* original userland pointer */
2544 struct audit_names *aname;
2547 static_assert(offsetof(struct filename, iname) % sizeof(long) == 0);
2549 extern long vfs_truncate(const struct path *, loff_t);
2550 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2552 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2554 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2556 extern struct file *file_open_name(struct filename *, int, umode_t);
2557 extern struct file *filp_open(const char *, int, umode_t);
2558 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2559 const char *, int, umode_t);
2560 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2561 extern struct file * open_with_fake_path(const struct path *, int,
2562 struct inode*, const struct cred *);
2563 static inline struct file *file_clone_open(struct file *file)
2565 return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2567 extern int filp_close(struct file *, fl_owner_t id);
2569 extern struct filename *getname_flags(const char __user *, int, int *);
2570 extern struct filename *getname(const char __user *);
2571 extern struct filename *getname_kernel(const char *);
2572 extern void putname(struct filename *name);
2574 extern int finish_open(struct file *file, struct dentry *dentry,
2575 int (*open)(struct inode *, struct file *));
2576 extern int finish_no_open(struct file *file, struct dentry *dentry);
2579 extern void __init vfs_caches_init_early(void);
2580 extern void __init vfs_caches_init(void);
2582 extern struct kmem_cache *names_cachep;
2584 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2585 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2588 extern int register_blkdev(unsigned int, const char *);
2589 extern void unregister_blkdev(unsigned int, const char *);
2590 extern struct block_device *bdget(dev_t);
2591 extern struct block_device *bdgrab(struct block_device *bdev);
2592 extern void bd_set_size(struct block_device *, loff_t size);
2593 extern void bd_forget(struct inode *inode);
2594 extern void bdput(struct block_device *);
2595 extern void invalidate_bdev(struct block_device *);
2596 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2597 extern int sync_blockdev(struct block_device *bdev);
2598 extern struct super_block *freeze_bdev(struct block_device *);
2599 extern void emergency_thaw_all(void);
2600 extern void emergency_thaw_bdev(struct super_block *sb);
2601 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2602 extern int fsync_bdev(struct block_device *);
2604 extern struct super_block *blockdev_superblock;
2606 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2608 return sb == blockdev_superblock;
2611 static inline void bd_forget(struct inode *inode) {}
2612 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2613 static inline void invalidate_bdev(struct block_device *bdev) {}
2615 static inline struct super_block *freeze_bdev(struct block_device *sb)
2620 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2625 static inline int emergency_thaw_bdev(struct super_block *sb)
2630 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2634 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2639 extern int sync_filesystem(struct super_block *);
2640 extern const struct file_operations def_blk_fops;
2641 extern const struct file_operations def_chr_fops;
2643 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2644 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2645 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2646 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2648 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2650 extern struct block_device *bd_start_claiming(struct block_device *bdev,
2652 extern void bd_finish_claiming(struct block_device *bdev,
2653 struct block_device *whole, void *holder);
2654 extern void bd_abort_claiming(struct block_device *bdev,
2655 struct block_device *whole, void *holder);
2656 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2659 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2660 extern void bd_unlink_disk_holder(struct block_device *bdev,
2661 struct gendisk *disk);
2663 static inline int bd_link_disk_holder(struct block_device *bdev,
2664 struct gendisk *disk)
2668 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2669 struct gendisk *disk)
2676 #define CHRDEV_MAJOR_MAX 512
2677 /* Marks the bottom of the first segment of free char majors */
2678 #define CHRDEV_MAJOR_DYN_END 234
2679 /* Marks the top and bottom of the second segment of free char majors */
2680 #define CHRDEV_MAJOR_DYN_EXT_START 511
2681 #define CHRDEV_MAJOR_DYN_EXT_END 384
2683 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2684 extern int register_chrdev_region(dev_t, unsigned, const char *);
2685 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2686 unsigned int count, const char *name,
2687 const struct file_operations *fops);
2688 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2689 unsigned int count, const char *name);
2690 extern void unregister_chrdev_region(dev_t, unsigned);
2691 extern void chrdev_show(struct seq_file *,off_t);
2693 static inline int register_chrdev(unsigned int major, const char *name,
2694 const struct file_operations *fops)
2696 return __register_chrdev(major, 0, 256, name, fops);
2699 static inline void unregister_chrdev(unsigned int major, const char *name)
2701 __unregister_chrdev(major, 0, 256, name);
2704 /* fs/block_dev.c */
2705 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2706 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2709 #define BLKDEV_MAJOR_MAX 512
2710 extern const char *bdevname(struct block_device *bdev, char *buffer);
2711 extern struct block_device *lookup_bdev(const char *);
2712 extern void blkdev_show(struct seq_file *,off_t);
2715 #define BLKDEV_MAJOR_MAX 0
2718 extern void init_special_inode(struct inode *, umode_t, dev_t);
2720 /* Invalid inode operations -- fs/bad_inode.c */
2721 extern void make_bad_inode(struct inode *);
2722 extern bool is_bad_inode(struct inode *);
2725 extern int revalidate_disk(struct gendisk *);
2726 extern int check_disk_change(struct block_device *);
2727 extern int __invalidate_device(struct block_device *, bool);
2729 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2730 pgoff_t start, pgoff_t end);
2732 static inline void invalidate_remote_inode(struct inode *inode)
2734 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2735 S_ISLNK(inode->i_mode))
2736 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2738 extern int invalidate_inode_pages2(struct address_space *mapping);
2739 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2740 pgoff_t start, pgoff_t end);
2741 extern int write_inode_now(struct inode *, int);
2742 extern int filemap_fdatawrite(struct address_space *);
2743 extern int filemap_flush(struct address_space *);
2744 extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2745 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2747 extern int filemap_fdatawait_range_keep_errors(struct address_space *mapping,
2748 loff_t start_byte, loff_t end_byte);
2750 static inline int filemap_fdatawait(struct address_space *mapping)
2752 return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2755 extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2757 extern int filemap_write_and_wait_range(struct address_space *mapping,
2758 loff_t lstart, loff_t lend);
2759 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2760 loff_t start, loff_t end, int sync_mode);
2761 extern int filemap_fdatawrite_range(struct address_space *mapping,
2762 loff_t start, loff_t end);
2763 extern int filemap_check_errors(struct address_space *mapping);
2764 extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2766 static inline int filemap_write_and_wait(struct address_space *mapping)
2768 return filemap_write_and_wait_range(mapping, 0, LLONG_MAX);
2771 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2773 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2774 extern int __must_check file_write_and_wait_range(struct file *file,
2775 loff_t start, loff_t end);
2777 static inline int file_write_and_wait(struct file *file)
2779 return file_write_and_wait_range(file, 0, LLONG_MAX);
2783 * filemap_set_wb_err - set a writeback error on an address_space
2784 * @mapping: mapping in which to set writeback error
2785 * @err: error to be set in mapping
2787 * When writeback fails in some way, we must record that error so that
2788 * userspace can be informed when fsync and the like are called. We endeavor
2789 * to report errors on any file that was open at the time of the error. Some
2790 * internal callers also need to know when writeback errors have occurred.
2792 * When a writeback error occurs, most filesystems will want to call
2793 * filemap_set_wb_err to record the error in the mapping so that it will be
2794 * automatically reported whenever fsync is called on the file.
2796 static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2798 /* Fastpath for common case of no error */
2800 __filemap_set_wb_err(mapping, err);
2804 * filemap_check_wb_error - has an error occurred since the mark was sampled?
2805 * @mapping: mapping to check for writeback errors
2806 * @since: previously-sampled errseq_t
2808 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2809 * the given value was sampled.
2811 * If it has then report the latest error set, otherwise return 0.
2813 static inline int filemap_check_wb_err(struct address_space *mapping,
2816 return errseq_check(&mapping->wb_err, since);
2820 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2821 * @mapping: mapping to be sampled
2823 * Writeback errors are always reported relative to a particular sample point
2824 * in the past. This function provides those sample points.
2826 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2828 return errseq_sample(&mapping->wb_err);
2832 * file_sample_sb_err - sample the current errseq_t to test for later errors
2833 * @mapping: mapping to be sampled
2835 * Grab the most current superblock-level errseq_t value for the given
2838 static inline errseq_t file_sample_sb_err(struct file *file)
2840 return errseq_sample(&file->f_path.dentry->d_sb->s_wb_err);
2843 static inline int filemap_nr_thps(struct address_space *mapping)
2845 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
2846 return atomic_read(&mapping->nr_thps);
2852 static inline void filemap_nr_thps_inc(struct address_space *mapping)
2854 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
2855 atomic_inc(&mapping->nr_thps);
2861 static inline void filemap_nr_thps_dec(struct address_space *mapping)
2863 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
2864 atomic_dec(&mapping->nr_thps);
2870 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2872 extern int vfs_fsync(struct file *file, int datasync);
2874 extern int sync_file_range(struct file *file, loff_t offset, loff_t nbytes,
2875 unsigned int flags);
2878 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2879 * to already be updated for the write, and will return either the amount
2880 * of bytes passed in, or an error if syncing the file failed.
2882 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2884 if (iocb->ki_flags & IOCB_DSYNC) {
2885 int ret = vfs_fsync_range(iocb->ki_filp,
2886 iocb->ki_pos - count, iocb->ki_pos - 1,
2887 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2895 extern void emergency_sync(void);
2896 extern void emergency_remount(void);
2899 extern int bmap(struct inode *inode, sector_t *block);
2901 static inline int bmap(struct inode *inode, sector_t *block)
2907 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2908 extern int inode_permission(struct inode *, int);
2909 extern int generic_permission(struct inode *, int);
2910 extern int __check_sticky(struct inode *dir, struct inode *inode);
2912 static inline bool execute_ok(struct inode *inode)
2914 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2917 static inline void file_start_write(struct file *file)
2919 if (!S_ISREG(file_inode(file)->i_mode))
2921 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2924 static inline bool file_start_write_trylock(struct file *file)
2926 if (!S_ISREG(file_inode(file)->i_mode))
2928 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2931 static inline void file_end_write(struct file *file)
2933 if (!S_ISREG(file_inode(file)->i_mode))
2935 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2939 * get_write_access() gets write permission for a file.
2940 * put_write_access() releases this write permission.
2941 * This is used for regular files.
2942 * We cannot support write (and maybe mmap read-write shared) accesses and
2943 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2944 * can have the following values:
2945 * 0: no writers, no VM_DENYWRITE mappings
2946 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2947 * > 0: (i_writecount) users are writing to the file.
2949 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2950 * except for the cases where we don't hold i_writecount yet. Then we need to
2951 * use {get,deny}_write_access() - these functions check the sign and refuse
2952 * to do the change if sign is wrong.
2954 static inline int get_write_access(struct inode *inode)
2956 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2958 static inline int deny_write_access(struct file *file)
2960 struct inode *inode = file_inode(file);
2961 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2963 static inline void put_write_access(struct inode * inode)
2965 atomic_dec(&inode->i_writecount);
2967 static inline void allow_write_access(struct file *file)
2970 atomic_inc(&file_inode(file)->i_writecount);
2972 static inline bool inode_is_open_for_write(const struct inode *inode)
2974 return atomic_read(&inode->i_writecount) > 0;
2977 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
2978 static inline void i_readcount_dec(struct inode *inode)
2980 BUG_ON(!atomic_read(&inode->i_readcount));
2981 atomic_dec(&inode->i_readcount);
2983 static inline void i_readcount_inc(struct inode *inode)
2985 atomic_inc(&inode->i_readcount);
2988 static inline void i_readcount_dec(struct inode *inode)
2992 static inline void i_readcount_inc(struct inode *inode)
2997 extern int do_pipe_flags(int *, int);
2999 #define __kernel_read_file_id(id) \
3000 id(UNKNOWN, unknown) \
3001 id(FIRMWARE, firmware) \
3002 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
3003 id(FIRMWARE_EFI_EMBEDDED, firmware) \
3004 id(MODULE, kernel-module) \
3005 id(KEXEC_IMAGE, kexec-image) \
3006 id(KEXEC_INITRAMFS, kexec-initramfs) \
3007 id(POLICY, security-policy) \
3008 id(X509_CERTIFICATE, x509-certificate) \
3011 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
3012 #define __fid_stringify(dummy, str) #str,
3014 enum kernel_read_file_id {
3015 __kernel_read_file_id(__fid_enumify)
3018 static const char * const kernel_read_file_str[] = {
3019 __kernel_read_file_id(__fid_stringify)
3022 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
3024 if ((unsigned)id >= READING_MAX_ID)
3025 return kernel_read_file_str[READING_UNKNOWN];
3027 return kernel_read_file_str[id];
3030 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
3031 enum kernel_read_file_id);
3032 extern int kernel_read_file_from_path(const char *, void **, loff_t *, loff_t,
3033 enum kernel_read_file_id);
3034 extern int kernel_read_file_from_path_initns(const char *, void **, loff_t *, loff_t,
3035 enum kernel_read_file_id);
3036 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
3037 enum kernel_read_file_id);
3038 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
3039 ssize_t __kernel_read(struct file *file, void *buf, size_t count, loff_t *pos);
3040 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
3041 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
3042 extern struct file * open_exec(const char *);
3044 /* fs/dcache.c -- generic fs support functions */
3045 extern bool is_subdir(struct dentry *, struct dentry *);
3046 extern bool path_is_under(const struct path *, const struct path *);
3048 extern char *file_path(struct file *, char *, int);
3050 #include <linux/err.h>
3052 /* needed for stackable file system support */
3053 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
3055 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
3057 extern int inode_init_always(struct super_block *, struct inode *);
3058 extern void inode_init_once(struct inode *);
3059 extern void address_space_init_once(struct address_space *mapping);
3060 extern struct inode * igrab(struct inode *);
3061 extern ino_t iunique(struct super_block *, ino_t);
3062 extern int inode_needs_sync(struct inode *inode);
3063 extern int generic_delete_inode(struct inode *inode);
3064 static inline int generic_drop_inode(struct inode *inode)
3066 return !inode->i_nlink || inode_unhashed(inode) ||
3067 (inode->i_state & I_DONTCACHE);
3069 extern void d_mark_dontcache(struct inode *inode);
3071 extern struct inode *ilookup5_nowait(struct super_block *sb,
3072 unsigned long hashval, int (*test)(struct inode *, void *),
3074 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
3075 int (*test)(struct inode *, void *), void *data);
3076 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
3078 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
3079 int (*test)(struct inode *, void *),
3080 int (*set)(struct inode *, void *),
3082 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
3083 extern struct inode * iget_locked(struct super_block *, unsigned long);
3084 extern struct inode *find_inode_nowait(struct super_block *,
3086 int (*match)(struct inode *,
3087 unsigned long, void *),
3089 extern struct inode *find_inode_rcu(struct super_block *, unsigned long,
3090 int (*)(struct inode *, void *), void *);
3091 extern struct inode *find_inode_by_ino_rcu(struct super_block *, unsigned long);
3092 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
3093 extern int insert_inode_locked(struct inode *);
3094 #ifdef CONFIG_DEBUG_LOCK_ALLOC
3095 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
3097 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
3099 extern void unlock_new_inode(struct inode *);
3100 extern void discard_new_inode(struct inode *);
3101 extern unsigned int get_next_ino(void);
3102 extern void evict_inodes(struct super_block *sb);
3104 extern void __iget(struct inode * inode);
3105 extern void iget_failed(struct inode *);
3106 extern void clear_inode(struct inode *);
3107 extern void __destroy_inode(struct inode *);
3108 extern struct inode *new_inode_pseudo(struct super_block *sb);
3109 extern struct inode *new_inode(struct super_block *sb);
3110 extern void free_inode_nonrcu(struct inode *inode);
3111 extern int should_remove_suid(struct dentry *);
3112 extern int file_remove_privs(struct file *);
3114 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
3115 static inline void insert_inode_hash(struct inode *inode)
3117 __insert_inode_hash(inode, inode->i_ino);
3120 extern void __remove_inode_hash(struct inode *);
3121 static inline void remove_inode_hash(struct inode *inode)
3123 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
3124 __remove_inode_hash(inode);
3127 extern void inode_sb_list_add(struct inode *inode);
3130 extern int bdev_read_only(struct block_device *);
3132 extern int set_blocksize(struct block_device *, int);
3133 extern int sb_set_blocksize(struct super_block *, int);
3134 extern int sb_min_blocksize(struct super_block *, int);
3136 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
3137 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
3138 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
3139 extern int generic_remap_checks(struct file *file_in, loff_t pos_in,
3140 struct file *file_out, loff_t pos_out,
3141 loff_t *count, unsigned int remap_flags);
3142 extern int generic_file_rw_checks(struct file *file_in, struct file *file_out);
3143 extern int generic_copy_file_checks(struct file *file_in, loff_t pos_in,
3144 struct file *file_out, loff_t pos_out,
3145 size_t *count, unsigned int flags);
3146 extern ssize_t generic_file_buffered_read(struct kiocb *iocb,
3147 struct iov_iter *to, ssize_t already_read);
3148 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
3149 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
3150 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
3151 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
3152 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
3154 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
3156 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
3158 ssize_t vfs_iocb_iter_read(struct file *file, struct kiocb *iocb,
3159 struct iov_iter *iter);
3160 ssize_t vfs_iocb_iter_write(struct file *file, struct kiocb *iocb,
3161 struct iov_iter *iter);
3163 /* fs/block_dev.c */
3164 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
3165 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
3166 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
3168 extern void block_sync_page(struct page *page);
3171 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
3172 struct pipe_inode_info *, size_t, unsigned int);
3173 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
3174 struct file *, loff_t *, size_t, unsigned int);
3175 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
3176 struct file *out, loff_t *, size_t len, unsigned int flags);
3177 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
3178 loff_t *opos, size_t len, unsigned int flags);
3182 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3183 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3184 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
3185 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3186 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3187 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3188 int whence, loff_t maxsize, loff_t eof);
3189 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3190 int whence, loff_t size);
3191 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3192 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3193 extern int generic_file_open(struct inode * inode, struct file * filp);
3194 extern int nonseekable_open(struct inode * inode, struct file * filp);
3195 extern int stream_open(struct inode * inode, struct file * filp);
3198 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3199 loff_t file_offset);
3202 /* need locking between buffered and direct access */
3205 /* filesystem does not support filling holes */
3206 DIO_SKIP_HOLES = 0x02,
3209 void dio_end_io(struct bio *bio);
3211 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3212 struct block_device *bdev, struct iov_iter *iter,
3213 get_block_t get_block,
3214 dio_iodone_t end_io, dio_submit_t submit_io,
3217 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3218 struct inode *inode,
3219 struct iov_iter *iter,
3220 get_block_t get_block)
3222 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3223 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3227 void inode_dio_wait(struct inode *inode);
3230 * inode_dio_begin - signal start of a direct I/O requests
3231 * @inode: inode the direct I/O happens on
3233 * This is called once we've finished processing a direct I/O request,
3234 * and is used to wake up callers waiting for direct I/O to be quiesced.
3236 static inline void inode_dio_begin(struct inode *inode)
3238 atomic_inc(&inode->i_dio_count);
3242 * inode_dio_end - signal finish of a direct I/O requests
3243 * @inode: inode the direct I/O happens on
3245 * This is called once we've finished processing a direct I/O request,
3246 * and is used to wake up callers waiting for direct I/O to be quiesced.
3248 static inline void inode_dio_end(struct inode *inode)
3250 if (atomic_dec_and_test(&inode->i_dio_count))
3251 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3255 * Warn about a page cache invalidation failure diring a direct I/O write.
3257 void dio_warn_stale_pagecache(struct file *filp);
3259 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3262 extern const struct file_operations generic_ro_fops;
3264 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3266 extern int readlink_copy(char __user *, int, const char *);
3267 extern int page_readlink(struct dentry *, char __user *, int);
3268 extern const char *page_get_link(struct dentry *, struct inode *,
3269 struct delayed_call *);
3270 extern void page_put_link(void *);
3271 extern int __page_symlink(struct inode *inode, const char *symname, int len,
3273 extern int page_symlink(struct inode *inode, const char *symname, int len);
3274 extern const struct inode_operations page_symlink_inode_operations;
3275 extern void kfree_link(void *);
3276 extern void generic_fillattr(struct inode *, struct kstat *);
3277 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3278 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3279 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3280 void inode_add_bytes(struct inode *inode, loff_t bytes);
3281 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3282 void inode_sub_bytes(struct inode *inode, loff_t bytes);
3283 static inline loff_t __inode_get_bytes(struct inode *inode)
3285 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3287 loff_t inode_get_bytes(struct inode *inode);
3288 void inode_set_bytes(struct inode *inode, loff_t bytes);
3289 const char *simple_get_link(struct dentry *, struct inode *,
3290 struct delayed_call *);
3291 extern const struct inode_operations simple_symlink_inode_operations;
3293 extern int iterate_dir(struct file *, struct dir_context *);
3295 extern int vfs_statx(int, const char __user *, int, struct kstat *, u32);
3296 extern int vfs_statx_fd(unsigned int, struct kstat *, u32, unsigned int);
3298 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3300 return vfs_statx(AT_FDCWD, filename, AT_NO_AUTOMOUNT,
3301 stat, STATX_BASIC_STATS);
3303 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3305 return vfs_statx(AT_FDCWD, name, AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT,
3306 stat, STATX_BASIC_STATS);
3308 static inline int vfs_fstatat(int dfd, const char __user *filename,
3309 struct kstat *stat, int flags)
3311 return vfs_statx(dfd, filename, flags | AT_NO_AUTOMOUNT,
3312 stat, STATX_BASIC_STATS);
3314 static inline int vfs_fstat(int fd, struct kstat *stat)
3316 return vfs_statx_fd(fd, stat, STATX_BASIC_STATS, 0);
3320 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3321 extern int vfs_readlink(struct dentry *, char __user *, int);
3323 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3324 extern void put_filesystem(struct file_system_type *fs);
3325 extern struct file_system_type *get_fs_type(const char *name);
3326 extern struct super_block *get_super(struct block_device *);
3327 extern struct super_block *get_super_thawed(struct block_device *);
3328 extern struct super_block *get_super_exclusive_thawed(struct block_device *bdev);
3329 extern struct super_block *get_active_super(struct block_device *bdev);
3330 extern void drop_super(struct super_block *sb);
3331 extern void drop_super_exclusive(struct super_block *sb);
3332 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3333 extern void iterate_supers_type(struct file_system_type *,
3334 void (*)(struct super_block *, void *), void *);
3336 extern int dcache_dir_open(struct inode *, struct file *);
3337 extern int dcache_dir_close(struct inode *, struct file *);
3338 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3339 extern int dcache_readdir(struct file *, struct dir_context *);
3340 extern int simple_setattr(struct dentry *, struct iattr *);
3341 extern int simple_getattr(const struct path *, struct kstat *, u32, unsigned int);
3342 extern int simple_statfs(struct dentry *, struct kstatfs *);
3343 extern int simple_open(struct inode *inode, struct file *file);
3344 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3345 extern int simple_unlink(struct inode *, struct dentry *);
3346 extern int simple_rmdir(struct inode *, struct dentry *);
3347 extern int simple_rename(struct inode *, struct dentry *,
3348 struct inode *, struct dentry *, unsigned int);
3349 extern void simple_recursive_removal(struct dentry *,
3350 void (*callback)(struct dentry *));
3351 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3352 extern int noop_set_page_dirty(struct page *page);
3353 extern void noop_invalidatepage(struct page *page, unsigned int offset,
3354 unsigned int length);
3355 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3356 extern int simple_empty(struct dentry *);
3357 extern int simple_readpage(struct file *file, struct page *page);
3358 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3359 loff_t pos, unsigned len, unsigned flags,
3360 struct page **pagep, void **fsdata);
3361 extern int simple_write_end(struct file *file, struct address_space *mapping,
3362 loff_t pos, unsigned len, unsigned copied,
3363 struct page *page, void *fsdata);
3364 extern int always_delete_dentry(const struct dentry *);
3365 extern struct inode *alloc_anon_inode(struct super_block *);
3366 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3367 extern const struct dentry_operations simple_dentry_operations;
3369 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3370 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3371 extern const struct file_operations simple_dir_operations;
3372 extern const struct inode_operations simple_dir_inode_operations;
3373 extern void make_empty_dir_inode(struct inode *inode);
3374 extern bool is_empty_dir_inode(struct inode *inode);
3375 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3376 struct dentry *d_alloc_name(struct dentry *, const char *);
3377 extern int simple_fill_super(struct super_block *, unsigned long,
3378 const struct tree_descr *);
3379 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3380 extern void simple_release_fs(struct vfsmount **mount, int *count);
3382 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3383 loff_t *ppos, const void *from, size_t available);
3384 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3385 const void __user *from, size_t count);
3387 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3388 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3390 extern int generic_check_addressable(unsigned, u64);
3392 #ifdef CONFIG_MIGRATION
3393 extern int buffer_migrate_page(struct address_space *,
3394 struct page *, struct page *,
3396 extern int buffer_migrate_page_norefs(struct address_space *,
3397 struct page *, struct page *,
3400 #define buffer_migrate_page NULL
3401 #define buffer_migrate_page_norefs NULL
3404 extern int setattr_prepare(struct dentry *, struct iattr *);
3405 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3406 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
3408 extern int file_update_time(struct file *file);
3410 static inline bool vma_is_dax(const struct vm_area_struct *vma)
3412 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3415 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3417 struct inode *inode;
3421 if (!vma_is_dax(vma))
3423 inode = file_inode(vma->vm_file);
3424 if (S_ISCHR(inode->i_mode))
3425 return false; /* device-dax */
3429 static inline int iocb_flags(struct file *file)
3432 if (file->f_flags & O_APPEND)
3434 if (file->f_flags & O_DIRECT)
3436 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3438 if (file->f_flags & __O_SYNC)
3443 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3445 if (unlikely(flags & ~RWF_SUPPORTED))
3448 if (flags & RWF_NOWAIT) {
3449 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3451 ki->ki_flags |= IOCB_NOWAIT;
3453 if (flags & RWF_HIPRI)
3454 ki->ki_flags |= IOCB_HIPRI;
3455 if (flags & RWF_DSYNC)
3456 ki->ki_flags |= IOCB_DSYNC;
3457 if (flags & RWF_SYNC)
3458 ki->ki_flags |= (IOCB_DSYNC | IOCB_SYNC);
3459 if (flags & RWF_APPEND)
3460 ki->ki_flags |= IOCB_APPEND;
3464 static inline ino_t parent_ino(struct dentry *dentry)
3469 * Don't strictly need d_lock here? If the parent ino could change
3470 * then surely we'd have a deeper race in the caller?
3472 spin_lock(&dentry->d_lock);
3473 res = dentry->d_parent->d_inode->i_ino;
3474 spin_unlock(&dentry->d_lock);
3478 /* Transaction based IO helpers */
3481 * An argresp is stored in an allocated page and holds the
3482 * size of the argument or response, along with its content
3484 struct simple_transaction_argresp {
3489 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3491 char *simple_transaction_get(struct file *file, const char __user *buf,
3493 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3494 size_t size, loff_t *pos);
3495 int simple_transaction_release(struct inode *inode, struct file *file);
3497 void simple_transaction_set(struct file *file, size_t n);
3500 * simple attribute files
3502 * These attributes behave similar to those in sysfs:
3504 * Writing to an attribute immediately sets a value, an open file can be
3505 * written to multiple times.
3507 * Reading from an attribute creates a buffer from the value that might get
3508 * read with multiple read calls. When the attribute has been read
3509 * completely, no further read calls are possible until the file is opened
3512 * All attributes contain a text representation of a numeric value
3513 * that are accessed with the get() and set() functions.
3515 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3516 static int __fops ## _open(struct inode *inode, struct file *file) \
3518 __simple_attr_check_format(__fmt, 0ull); \
3519 return simple_attr_open(inode, file, __get, __set, __fmt); \
3521 static const struct file_operations __fops = { \
3522 .owner = THIS_MODULE, \
3523 .open = __fops ## _open, \
3524 .release = simple_attr_release, \
3525 .read = simple_attr_read, \
3526 .write = simple_attr_write, \
3527 .llseek = generic_file_llseek, \
3530 static inline __printf(1, 2)
3531 void __simple_attr_check_format(const char *fmt, ...)
3533 /* don't do anything, just let the compiler check the arguments; */
3536 int simple_attr_open(struct inode *inode, struct file *file,
3537 int (*get)(void *, u64 *), int (*set)(void *, u64),
3539 int simple_attr_release(struct inode *inode, struct file *file);
3540 ssize_t simple_attr_read(struct file *file, char __user *buf,
3541 size_t len, loff_t *ppos);
3542 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3543 size_t len, loff_t *ppos);
3546 int proc_nr_files(struct ctl_table *table, int write,
3547 void *buffer, size_t *lenp, loff_t *ppos);
3548 int proc_nr_dentry(struct ctl_table *table, int write,
3549 void *buffer, size_t *lenp, loff_t *ppos);
3550 int proc_nr_inodes(struct ctl_table *table, int write,
3551 void *buffer, size_t *lenp, loff_t *ppos);
3552 int __init get_filesystem_list(char *buf);
3554 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3555 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3557 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3558 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3559 (flag & __FMODE_NONOTIFY)))
3561 static inline bool is_sxid(umode_t mode)
3563 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3566 static inline int check_sticky(struct inode *dir, struct inode *inode)
3568 if (!(dir->i_mode & S_ISVTX))
3571 return __check_sticky(dir, inode);
3574 static inline void inode_has_no_xattr(struct inode *inode)
3576 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3577 inode->i_flags |= S_NOSEC;
3580 static inline bool is_root_inode(struct inode *inode)
3582 return inode == inode->i_sb->s_root->d_inode;
3585 static inline bool dir_emit(struct dir_context *ctx,
3586 const char *name, int namelen,
3587 u64 ino, unsigned type)
3589 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3591 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3593 return ctx->actor(ctx, ".", 1, ctx->pos,
3594 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3596 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3598 return ctx->actor(ctx, "..", 2, ctx->pos,
3599 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3601 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3603 if (ctx->pos == 0) {
3604 if (!dir_emit_dot(file, ctx))
3608 if (ctx->pos == 1) {
3609 if (!dir_emit_dotdot(file, ctx))
3615 static inline bool dir_relax(struct inode *inode)
3617 inode_unlock(inode);
3619 return !IS_DEADDIR(inode);
3622 static inline bool dir_relax_shared(struct inode *inode)
3624 inode_unlock_shared(inode);
3625 inode_lock_shared(inode);
3626 return !IS_DEADDIR(inode);
3629 extern bool path_noexec(const struct path *path);
3630 extern void inode_nohighmem(struct inode *inode);
3633 extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len,
3635 extern int generic_fadvise(struct file *file, loff_t offset, loff_t len,
3638 #if defined(CONFIG_IO_URING)
3639 extern struct sock *io_uring_get_socket(struct file *file);
3641 static inline struct sock *io_uring_get_socket(struct file *file)
3647 int vfs_ioc_setflags_prepare(struct inode *inode, unsigned int oldflags,
3648 unsigned int flags);
3650 int vfs_ioc_fssetxattr_check(struct inode *inode, const struct fsxattr *old_fa,
3651 struct fsxattr *fa);
3653 static inline void simple_fill_fsxattr(struct fsxattr *fa, __u32 xflags)
3655 memset(fa, 0, sizeof(*fa));
3656 fa->fsx_xflags = xflags;
3660 * Flush file data before changing attributes. Caller must hold any locks
3661 * required to prevent further writes to this file until we're done setting
3664 static inline int inode_drain_writes(struct inode *inode)
3666 inode_dio_wait(inode);
3667 return filemap_write_and_wait(inode->i_mapping);
3670 #endif /* _LINUX_FS_H */