1 /* SPDX-License-Identifier: GPL-2.0 */
5 #include <linux/linkage.h>
6 #include <linux/wait_bit.h>
7 #include <linux/kdev_t.h>
8 #include <linux/dcache.h>
9 #include <linux/path.h>
10 #include <linux/stat.h>
11 #include <linux/cache.h>
12 #include <linux/list.h>
13 #include <linux/list_lru.h>
14 #include <linux/llist.h>
15 #include <linux/radix-tree.h>
16 #include <linux/xarray.h>
17 #include <linux/rbtree.h>
18 #include <linux/init.h>
19 #include <linux/pid.h>
20 #include <linux/bug.h>
21 #include <linux/mutex.h>
22 #include <linux/rwsem.h>
23 #include <linux/mm_types.h>
24 #include <linux/capability.h>
25 #include <linux/semaphore.h>
26 #include <linux/fcntl.h>
27 #include <linux/fiemap.h>
28 #include <linux/rculist_bl.h>
29 #include <linux/atomic.h>
30 #include <linux/shrinker.h>
31 #include <linux/migrate_mode.h>
32 #include <linux/uidgid.h>
33 #include <linux/lockdep.h>
34 #include <linux/percpu-rwsem.h>
35 #include <linux/workqueue.h>
36 #include <linux/delayed_call.h>
37 #include <linux/uuid.h>
38 #include <linux/errseq.h>
39 #include <linux/ioprio.h>
40 #include <linux/fs_types.h>
41 #include <linux/build_bug.h>
42 #include <linux/stddef.h>
44 #include <asm/byteorder.h>
45 #include <uapi/linux/fs.h>
47 struct backing_dev_info;
50 struct export_operations;
55 struct pipe_inode_info;
56 struct poll_table_struct;
58 struct vm_area_struct;
61 struct swap_info_struct;
63 struct workqueue_struct;
66 struct fscrypt_operations;
68 struct fsverity_operations;
70 struct fs_parameter_description;
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;
297 * Write life time hint values.
298 * Stored in struct inode as u8.
301 WRITE_LIFE_NOT_SET = 0,
302 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
303 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
304 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
305 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
306 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
309 #define IOCB_EVENTFD (1 << 0)
310 #define IOCB_APPEND (1 << 1)
311 #define IOCB_DIRECT (1 << 2)
312 #define IOCB_HIPRI (1 << 3)
313 #define IOCB_DSYNC (1 << 4)
314 #define IOCB_SYNC (1 << 5)
315 #define IOCB_WRITE (1 << 6)
316 #define IOCB_NOWAIT (1 << 7)
319 struct file *ki_filp;
321 /* The 'ki_filp' pointer is shared in a union for aio */
322 randomized_struct_fields_start
325 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
329 u16 ki_ioprio; /* See linux/ioprio.h */
330 unsigned int ki_cookie; /* for ->iopoll */
332 randomized_struct_fields_end
335 static inline bool is_sync_kiocb(struct kiocb *kiocb)
337 return kiocb->ki_complete == NULL;
341 * "descriptor" for what we're up to with a read.
342 * This allows us to use the same read code yet
343 * have multiple different users of the data that
344 * we read from a file.
346 * The simplest case just copies the data to user
359 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
360 unsigned long, unsigned long);
362 struct address_space_operations {
363 int (*writepage)(struct page *page, struct writeback_control *wbc);
364 int (*readpage)(struct file *, struct page *);
366 /* Write back some dirty pages from this mapping. */
367 int (*writepages)(struct address_space *, struct writeback_control *);
369 /* Set a page dirty. Return true if this dirtied it */
370 int (*set_page_dirty)(struct page *page);
373 * Reads in the requested pages. Unlike ->readpage(), this is
374 * PURELY used for read-ahead!.
376 int (*readpages)(struct file *filp, struct address_space *mapping,
377 struct list_head *pages, unsigned nr_pages);
379 int (*write_begin)(struct file *, struct address_space *mapping,
380 loff_t pos, unsigned len, unsigned flags,
381 struct page **pagep, void **fsdata);
382 int (*write_end)(struct file *, struct address_space *mapping,
383 loff_t pos, unsigned len, unsigned copied,
384 struct page *page, void *fsdata);
386 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
387 sector_t (*bmap)(struct address_space *, sector_t);
388 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
389 int (*releasepage) (struct page *, gfp_t);
390 void (*freepage)(struct page *);
391 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
393 * migrate the contents of a page to the specified target. If
394 * migrate_mode is MIGRATE_ASYNC, it must not block.
396 int (*migratepage) (struct address_space *,
397 struct page *, struct page *, enum migrate_mode);
398 bool (*isolate_page)(struct page *, isolate_mode_t);
399 void (*putback_page)(struct page *);
400 int (*launder_page) (struct page *);
401 int (*is_partially_uptodate) (struct page *, unsigned long,
403 void (*is_dirty_writeback) (struct page *, bool *, bool *);
404 int (*error_remove_page)(struct address_space *, struct page *);
406 /* swapfile support */
407 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
409 void (*swap_deactivate)(struct file *file);
412 extern const struct address_space_operations empty_aops;
415 * pagecache_write_begin/pagecache_write_end must be used by general code
416 * to write into the pagecache.
418 int pagecache_write_begin(struct file *, struct address_space *mapping,
419 loff_t pos, unsigned len, unsigned flags,
420 struct page **pagep, void **fsdata);
422 int pagecache_write_end(struct file *, struct address_space *mapping,
423 loff_t pos, unsigned len, unsigned copied,
424 struct page *page, void *fsdata);
427 * struct address_space - Contents of a cacheable, mappable object.
428 * @host: Owner, either the inode or the block_device.
429 * @i_pages: Cached pages.
430 * @gfp_mask: Memory allocation flags to use for allocating pages.
431 * @i_mmap_writable: Number of VM_SHARED mappings.
432 * @nr_thps: Number of THPs in the pagecache (non-shmem only).
433 * @i_mmap: Tree of private and shared mappings.
434 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
435 * @nrpages: Number of page entries, protected by the i_pages lock.
436 * @nrexceptional: Shadow or DAX entries, protected by the i_pages lock.
437 * @writeback_index: Writeback starts here.
439 * @flags: Error bits and flags (AS_*).
440 * @wb_err: The most recent error which has occurred.
441 * @private_lock: For use by the owner of the address_space.
442 * @private_list: For use by the owner of the address_space.
443 * @private_data: For use by the owner of the address_space.
445 struct address_space {
447 struct xarray i_pages;
449 atomic_t i_mmap_writable;
450 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
451 /* number of thp, only for non-shmem files */
454 struct rb_root_cached i_mmap;
455 struct rw_semaphore i_mmap_rwsem;
456 unsigned long nrpages;
457 unsigned long nrexceptional;
458 pgoff_t writeback_index;
459 const struct address_space_operations *a_ops;
462 spinlock_t private_lock;
463 struct list_head private_list;
465 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
467 * On most architectures that alignment is already the case; but
468 * must be enforced here for CRIS, to let the least significant bit
469 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
471 struct request_queue;
473 struct block_device {
474 dev_t bd_dev; /* not a kdev_t - it's a search key */
476 struct inode * bd_inode; /* will die */
477 struct super_block * bd_super;
478 struct mutex bd_mutex; /* open/close mutex */
482 bool bd_write_holder;
484 struct list_head bd_holder_disks;
486 struct block_device * bd_contains;
487 unsigned bd_block_size;
489 struct hd_struct * bd_part;
490 /* number of times partitions within this device have been opened. */
491 unsigned bd_part_count;
493 struct gendisk * bd_disk;
494 struct request_queue * bd_queue;
495 struct backing_dev_info *bd_bdi;
496 struct list_head bd_list;
498 * Private data. You must have bd_claim'ed the block_device
499 * to use this. NOTE: bd_claim allows an owner to claim
500 * the same device multiple times, the owner must take special
501 * care to not mess up bd_private for that case.
503 unsigned long bd_private;
505 /* The counter of freeze processes */
506 int bd_fsfreeze_count;
507 /* Mutex for freeze */
508 struct mutex bd_fsfreeze_mutex;
509 } __randomize_layout;
511 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
512 #define PAGECACHE_TAG_DIRTY XA_MARK_0
513 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
514 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
517 * Returns true if any of the pages in the mapping are marked with the tag.
519 static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag)
521 return xa_marked(&mapping->i_pages, tag);
524 static inline void i_mmap_lock_write(struct address_space *mapping)
526 down_write(&mapping->i_mmap_rwsem);
529 static inline void i_mmap_unlock_write(struct address_space *mapping)
531 up_write(&mapping->i_mmap_rwsem);
534 static inline void i_mmap_lock_read(struct address_space *mapping)
536 down_read(&mapping->i_mmap_rwsem);
539 static inline void i_mmap_unlock_read(struct address_space *mapping)
541 up_read(&mapping->i_mmap_rwsem);
545 * Might pages of this file be mapped into userspace?
547 static inline int mapping_mapped(struct address_space *mapping)
549 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
553 * Might pages of this file have been modified in userspace?
554 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
555 * marks vma as VM_SHARED if it is shared, and the file was opened for
556 * writing i.e. vma may be mprotected writable even if now readonly.
558 * If i_mmap_writable is negative, no new writable mappings are allowed. You
559 * can only deny writable mappings, if none exists right now.
561 static inline int mapping_writably_mapped(struct address_space *mapping)
563 return atomic_read(&mapping->i_mmap_writable) > 0;
566 static inline int mapping_map_writable(struct address_space *mapping)
568 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
572 static inline void mapping_unmap_writable(struct address_space *mapping)
574 atomic_dec(&mapping->i_mmap_writable);
577 static inline int mapping_deny_writable(struct address_space *mapping)
579 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
583 static inline void mapping_allow_writable(struct address_space *mapping)
585 atomic_inc(&mapping->i_mmap_writable);
589 * Use sequence counter to get consistent i_size on 32-bit processors.
591 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
592 #include <linux/seqlock.h>
593 #define __NEED_I_SIZE_ORDERED
594 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
596 #define i_size_ordered_init(inode) do { } while (0)
600 #define ACL_NOT_CACHED ((void *)(-1))
601 #define ACL_DONT_CACHE ((void *)(-3))
603 static inline struct posix_acl *
604 uncached_acl_sentinel(struct task_struct *task)
606 return (void *)task + 1;
610 is_uncached_acl(struct posix_acl *acl)
612 return (long)acl & 1;
615 #define IOP_FASTPERM 0x0001
616 #define IOP_LOOKUP 0x0002
617 #define IOP_NOFOLLOW 0x0004
618 #define IOP_XATTR 0x0008
619 #define IOP_DEFAULT_READLINK 0x0010
621 struct fsnotify_mark_connector;
624 * Keep mostly read-only and often accessed (especially for
625 * the RCU path lookup and 'stat' data) fields at the beginning
626 * of the 'struct inode'
630 unsigned short i_opflags;
633 unsigned int i_flags;
635 #ifdef CONFIG_FS_POSIX_ACL
636 struct posix_acl *i_acl;
637 struct posix_acl *i_default_acl;
640 const struct inode_operations *i_op;
641 struct super_block *i_sb;
642 struct address_space *i_mapping;
644 #ifdef CONFIG_SECURITY
648 /* Stat data, not accessed from path walking */
651 * Filesystems may only read i_nlink directly. They shall use the
652 * following functions for modification:
654 * (set|clear|inc|drop)_nlink
655 * inode_(inc|dec)_link_count
658 const unsigned int i_nlink;
659 unsigned int __i_nlink;
663 struct timespec64 i_atime;
664 struct timespec64 i_mtime;
665 struct timespec64 i_ctime;
666 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
667 unsigned short i_bytes;
672 #ifdef __NEED_I_SIZE_ORDERED
673 seqcount_t i_size_seqcount;
677 unsigned long i_state;
678 struct rw_semaphore i_rwsem;
680 unsigned long dirtied_when; /* jiffies of first dirtying */
681 unsigned long dirtied_time_when;
683 struct hlist_node i_hash;
684 struct list_head i_io_list; /* backing dev IO list */
685 #ifdef CONFIG_CGROUP_WRITEBACK
686 struct bdi_writeback *i_wb; /* the associated cgroup wb */
688 /* foreign inode detection, see wbc_detach_inode() */
690 u16 i_wb_frn_avg_time;
691 u16 i_wb_frn_history;
693 struct list_head i_lru; /* inode LRU list */
694 struct list_head i_sb_list;
695 struct list_head i_wb_list; /* backing dev writeback list */
697 struct hlist_head i_dentry;
698 struct rcu_head i_rcu;
700 atomic64_t i_version;
701 atomic64_t i_sequence; /* see futex */
703 atomic_t i_dio_count;
704 atomic_t i_writecount;
705 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
706 atomic_t i_readcount; /* struct files open RO */
709 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
710 void (*free_inode)(struct inode *);
712 struct file_lock_context *i_flctx;
713 struct address_space i_data;
714 struct list_head i_devices;
716 struct pipe_inode_info *i_pipe;
717 struct block_device *i_bdev;
725 #ifdef CONFIG_FSNOTIFY
726 __u32 i_fsnotify_mask; /* all events this inode cares about */
727 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
730 #ifdef CONFIG_FS_ENCRYPTION
731 struct fscrypt_info *i_crypt_info;
734 #ifdef CONFIG_FS_VERITY
735 struct fsverity_info *i_verity_info;
738 void *i_private; /* fs or device private pointer */
739 } __randomize_layout;
741 struct timespec64 timestamp_truncate(struct timespec64 t, struct inode *inode);
743 static inline unsigned int i_blocksize(const struct inode *node)
745 return (1 << node->i_blkbits);
748 static inline int inode_unhashed(struct inode *inode)
750 return hlist_unhashed(&inode->i_hash);
754 * __mark_inode_dirty expects inodes to be hashed. Since we don't
755 * want special inodes in the fileset inode space, we make them
756 * appear hashed, but do not put on any lists. hlist_del()
757 * will work fine and require no locking.
759 static inline void inode_fake_hash(struct inode *inode)
761 hlist_add_fake(&inode->i_hash);
765 * inode->i_mutex nesting subclasses for the lock validator:
767 * 0: the object of the current VFS operation
771 * 4: second non-directory
772 * 5: second parent (when locking independent directories in rename)
774 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
775 * non-directories at once.
777 * The locking order between these classes is
778 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
780 enum inode_i_mutex_lock_class
790 static inline void inode_lock(struct inode *inode)
792 down_write(&inode->i_rwsem);
795 static inline void inode_unlock(struct inode *inode)
797 up_write(&inode->i_rwsem);
800 static inline void inode_lock_shared(struct inode *inode)
802 down_read(&inode->i_rwsem);
805 static inline void inode_unlock_shared(struct inode *inode)
807 up_read(&inode->i_rwsem);
810 static inline int inode_trylock(struct inode *inode)
812 return down_write_trylock(&inode->i_rwsem);
815 static inline int inode_trylock_shared(struct inode *inode)
817 return down_read_trylock(&inode->i_rwsem);
820 static inline int inode_is_locked(struct inode *inode)
822 return rwsem_is_locked(&inode->i_rwsem);
825 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
827 down_write_nested(&inode->i_rwsem, subclass);
830 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
832 down_read_nested(&inode->i_rwsem, subclass);
835 void lock_two_nondirectories(struct inode *, struct inode*);
836 void unlock_two_nondirectories(struct inode *, struct inode*);
839 * NOTE: in a 32bit arch with a preemptable kernel and
840 * an UP compile the i_size_read/write must be atomic
841 * with respect to the local cpu (unlike with preempt disabled),
842 * but they don't need to be atomic with respect to other cpus like in
843 * true SMP (so they need either to either locally disable irq around
844 * the read or for example on x86 they can be still implemented as a
845 * cmpxchg8b without the need of the lock prefix). For SMP compiles
846 * and 64bit archs it makes no difference if preempt is enabled or not.
848 static inline loff_t i_size_read(const struct inode *inode)
850 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
855 seq = read_seqcount_begin(&inode->i_size_seqcount);
856 i_size = inode->i_size;
857 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
859 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
863 i_size = inode->i_size;
867 return inode->i_size;
872 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
873 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
874 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
876 static inline void i_size_write(struct inode *inode, loff_t i_size)
878 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
880 write_seqcount_begin(&inode->i_size_seqcount);
881 inode->i_size = i_size;
882 write_seqcount_end(&inode->i_size_seqcount);
884 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
886 inode->i_size = i_size;
889 inode->i_size = i_size;
893 static inline unsigned iminor(const struct inode *inode)
895 return MINOR(inode->i_rdev);
898 static inline unsigned imajor(const struct inode *inode)
900 return MAJOR(inode->i_rdev);
903 extern struct block_device *I_BDEV(struct inode *inode);
906 rwlock_t lock; /* protects pid, uid, euid fields */
907 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
908 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
909 kuid_t uid, euid; /* uid/euid of process setting the owner */
910 int signum; /* posix.1b rt signal to be delivered on IO */
914 * Track a single file's readahead state
916 struct file_ra_state {
917 pgoff_t start; /* where readahead started */
918 unsigned int size; /* # of readahead pages */
919 unsigned int async_size; /* do asynchronous readahead when
920 there are only # of pages ahead */
922 unsigned int ra_pages; /* Maximum readahead window */
923 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
924 loff_t prev_pos; /* Cache last read() position */
928 * Check if @index falls in the readahead windows.
930 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
932 return (index >= ra->start &&
933 index < ra->start + ra->size);
938 struct llist_node fu_llist;
939 struct rcu_head fu_rcuhead;
942 struct inode *f_inode; /* cached value */
943 const struct file_operations *f_op;
946 * Protects f_ep_links, f_flags.
947 * Must not be taken from IRQ context.
950 enum rw_hint f_write_hint;
951 atomic_long_t f_count;
952 unsigned int f_flags;
954 struct mutex f_pos_lock;
956 struct fown_struct f_owner;
957 const struct cred *f_cred;
958 struct file_ra_state f_ra;
961 #ifdef CONFIG_SECURITY
964 /* needed for tty driver, and maybe others */
968 /* Used by fs/eventpoll.c to link all the hooks to this file */
969 struct list_head f_ep_links;
970 struct list_head f_tfile_llink;
971 #endif /* #ifdef CONFIG_EPOLL */
972 struct address_space *f_mapping;
975 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
980 /* file identifier */
981 unsigned char f_handle[];
984 static inline struct file *get_file(struct file *f)
986 atomic_long_inc(&f->f_count);
989 #define get_file_rcu_many(x, cnt) \
990 atomic_long_add_unless(&(x)->f_count, (cnt), 0)
991 #define get_file_rcu(x) get_file_rcu_many((x), 1)
992 #define file_count(x) atomic_long_read(&(x)->f_count)
994 #define MAX_NON_LFS ((1UL<<31) - 1)
996 /* Page cache limit. The filesystems should put that into their s_maxbytes
997 limits, otherwise bad things can happen in VM. */
998 #if BITS_PER_LONG==32
999 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
1000 #elif BITS_PER_LONG==64
1001 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
1006 #define FL_DELEG 4 /* NFSv4 delegation */
1007 #define FL_ACCESS 8 /* not trying to lock, just looking */
1008 #define FL_EXISTS 16 /* when unlocking, test for existence */
1009 #define FL_LEASE 32 /* lease held on this file */
1010 #define FL_CLOSE 64 /* unlock on close */
1011 #define FL_SLEEP 128 /* A blocking lock */
1012 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
1013 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
1014 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
1015 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
1017 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
1020 * Special return value from posix_lock_file() and vfs_lock_file() for
1021 * asynchronous locking.
1023 #define FILE_LOCK_DEFERRED 1
1025 /* legacy typedef, should eventually be removed */
1026 typedef void *fl_owner_t;
1030 struct file_lock_operations {
1031 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1032 void (*fl_release_private)(struct file_lock *);
1035 struct lock_manager_operations {
1036 fl_owner_t (*lm_get_owner)(fl_owner_t);
1037 void (*lm_put_owner)(fl_owner_t);
1038 void (*lm_notify)(struct file_lock *); /* unblock callback */
1039 int (*lm_grant)(struct file_lock *, int);
1040 bool (*lm_break)(struct file_lock *);
1041 int (*lm_change)(struct file_lock *, int, struct list_head *);
1042 void (*lm_setup)(struct file_lock *, void **);
1045 struct lock_manager {
1046 struct list_head list;
1048 * NFSv4 and up also want opens blocked during the grace period;
1055 void locks_start_grace(struct net *, struct lock_manager *);
1056 void locks_end_grace(struct lock_manager *);
1057 bool locks_in_grace(struct net *);
1058 bool opens_in_grace(struct net *);
1060 /* that will die - we need it for nfs_lock_info */
1061 #include <linux/nfs_fs_i.h>
1064 * struct file_lock represents a generic "file lock". It's used to represent
1065 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1066 * note that the same struct is used to represent both a request for a lock and
1067 * the lock itself, but the same object is never used for both.
1069 * FIXME: should we create a separate "struct lock_request" to help distinguish
1072 * The varous i_flctx lists are ordered by:
1075 * 2) lock range start
1078 * Obviously, the last two criteria only matter for POSIX locks.
1081 struct file_lock *fl_blocker; /* The lock, that is blocking us */
1082 struct list_head fl_list; /* link into file_lock_context */
1083 struct hlist_node fl_link; /* node in global lists */
1084 struct list_head fl_blocked_requests; /* list of requests with
1085 * ->fl_blocker pointing here
1087 struct list_head fl_blocked_member; /* node in
1088 * ->fl_blocker->fl_blocked_requests
1090 fl_owner_t fl_owner;
1091 unsigned int fl_flags;
1092 unsigned char fl_type;
1093 unsigned int fl_pid;
1094 int fl_link_cpu; /* what cpu's list is this on? */
1095 wait_queue_head_t fl_wait;
1096 struct file *fl_file;
1100 struct fasync_struct * fl_fasync; /* for lease break notifications */
1101 /* for lease breaks: */
1102 unsigned long fl_break_time;
1103 unsigned long fl_downgrade_time;
1105 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1106 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1108 struct nfs_lock_info nfs_fl;
1109 struct nfs4_lock_info nfs4_fl;
1111 struct list_head link; /* link in AFS vnode's pending_locks list */
1112 int state; /* state of grant or error if -ve */
1113 unsigned int debug_id;
1116 } __randomize_layout;
1118 struct file_lock_context {
1119 spinlock_t flc_lock;
1120 struct list_head flc_flock;
1121 struct list_head flc_posix;
1122 struct list_head flc_lease;
1125 /* The following constant reflects the upper bound of the file/locking space */
1127 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1128 #define OFFSET_MAX INT_LIMIT(loff_t)
1129 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1132 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1134 #define locks_inode(f) file_inode(f)
1136 #ifdef CONFIG_FILE_LOCKING
1137 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1138 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1141 #if BITS_PER_LONG == 32
1142 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1143 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1147 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1148 extern int fcntl_getlease(struct file *filp);
1151 void locks_free_lock_context(struct inode *inode);
1152 void locks_free_lock(struct file_lock *fl);
1153 extern void locks_init_lock(struct file_lock *);
1154 extern struct file_lock * locks_alloc_lock(void);
1155 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1156 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1157 extern void locks_remove_posix(struct file *, fl_owner_t);
1158 extern void locks_remove_file(struct file *);
1159 extern void locks_release_private(struct file_lock *);
1160 extern void posix_test_lock(struct file *, struct file_lock *);
1161 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1162 extern int locks_delete_block(struct file_lock *);
1163 extern int vfs_test_lock(struct file *, struct file_lock *);
1164 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1165 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1166 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1167 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1168 extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1169 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1170 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1171 extern int lease_modify(struct file_lock *, int, struct list_head *);
1173 struct notifier_block;
1174 extern int lease_register_notifier(struct notifier_block *);
1175 extern void lease_unregister_notifier(struct notifier_block *);
1177 struct files_struct;
1178 extern void show_fd_locks(struct seq_file *f,
1179 struct file *filp, struct files_struct *files);
1180 #else /* !CONFIG_FILE_LOCKING */
1181 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1182 struct flock __user *user)
1187 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1188 unsigned int cmd, struct flock __user *user)
1193 #if BITS_PER_LONG == 32
1194 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1195 struct flock64 __user *user)
1200 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1201 unsigned int cmd, struct flock64 __user *user)
1206 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1211 static inline int fcntl_getlease(struct file *filp)
1217 locks_free_lock_context(struct inode *inode)
1221 static inline void locks_init_lock(struct file_lock *fl)
1226 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1231 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1236 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1241 static inline void locks_remove_file(struct file *filp)
1246 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1251 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1252 struct file_lock *conflock)
1257 static inline int locks_delete_block(struct file_lock *waiter)
1262 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1267 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1268 struct file_lock *fl, struct file_lock *conf)
1273 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1278 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1283 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1288 static inline void lease_get_mtime(struct inode *inode,
1289 struct timespec64 *time)
1294 static inline int generic_setlease(struct file *filp, long arg,
1295 struct file_lock **flp, void **priv)
1300 static inline int vfs_setlease(struct file *filp, long arg,
1301 struct file_lock **lease, void **priv)
1306 static inline int lease_modify(struct file_lock *fl, int arg,
1307 struct list_head *dispose)
1312 struct files_struct;
1313 static inline void show_fd_locks(struct seq_file *f,
1314 struct file *filp, struct files_struct *files) {}
1315 #endif /* !CONFIG_FILE_LOCKING */
1317 static inline struct inode *file_inode(const struct file *f)
1322 static inline struct dentry *file_dentry(const struct file *file)
1324 return d_real(file->f_path.dentry, file_inode(file));
1327 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1329 return locks_lock_inode_wait(locks_inode(filp), fl);
1332 struct fasync_struct {
1336 struct fasync_struct *fa_next; /* singly linked list */
1337 struct file *fa_file;
1338 struct rcu_head fa_rcu;
1341 #define FASYNC_MAGIC 0x4601
1343 /* SMP safe fasync helpers: */
1344 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1345 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1346 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1347 extern struct fasync_struct *fasync_alloc(void);
1348 extern void fasync_free(struct fasync_struct *);
1350 /* can be called from interrupts */
1351 extern void kill_fasync(struct fasync_struct **, int, int);
1353 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1354 extern int f_setown(struct file *filp, unsigned long arg, int force);
1355 extern void f_delown(struct file *filp);
1356 extern pid_t f_getown(struct file *filp);
1357 extern int send_sigurg(struct fown_struct *fown);
1360 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1361 * represented in both.
1363 #define SB_RDONLY 1 /* Mount read-only */
1364 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1365 #define SB_NODEV 4 /* Disallow access to device special files */
1366 #define SB_NOEXEC 8 /* Disallow program execution */
1367 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1368 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1369 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1370 #define SB_NOATIME 1024 /* Do not update access times. */
1371 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1372 #define SB_SILENT 32768
1373 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1374 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1375 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1376 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1378 /* These sb flags are internal to the kernel */
1379 #define SB_SUBMOUNT (1<<26)
1380 #define SB_FORCE (1<<27)
1381 #define SB_NOSEC (1<<28)
1382 #define SB_BORN (1<<29)
1383 #define SB_ACTIVE (1<<30)
1384 #define SB_NOUSER (1<<31)
1390 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1391 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1392 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1393 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1394 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1397 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1398 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1399 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1400 #define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */
1402 /* sb->s_iflags to limit user namespace mounts */
1403 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1404 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1405 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1407 /* Possible states of 'frozen' field */
1409 SB_UNFROZEN = 0, /* FS is unfrozen */
1410 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1411 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1412 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1413 * internal threads if needed) */
1414 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1417 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1420 int frozen; /* Is sb frozen? */
1421 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1422 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1425 struct super_block {
1426 struct list_head s_list; /* Keep this first */
1427 dev_t s_dev; /* search index; _not_ kdev_t */
1428 unsigned char s_blocksize_bits;
1429 unsigned long s_blocksize;
1430 loff_t s_maxbytes; /* Max file size */
1431 struct file_system_type *s_type;
1432 const struct super_operations *s_op;
1433 const struct dquot_operations *dq_op;
1434 const struct quotactl_ops *s_qcop;
1435 const struct export_operations *s_export_op;
1436 unsigned long s_flags;
1437 unsigned long s_iflags; /* internal SB_I_* flags */
1438 unsigned long s_magic;
1439 struct dentry *s_root;
1440 struct rw_semaphore s_umount;
1443 #ifdef CONFIG_SECURITY
1446 const struct xattr_handler **s_xattr;
1447 #ifdef CONFIG_FS_ENCRYPTION
1448 const struct fscrypt_operations *s_cop;
1449 struct key *s_master_keys; /* master crypto keys in use */
1451 #ifdef CONFIG_FS_VERITY
1452 const struct fsverity_operations *s_vop;
1454 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1455 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1456 struct block_device *s_bdev;
1457 struct backing_dev_info *s_bdi;
1458 struct mtd_info *s_mtd;
1459 struct hlist_node s_instances;
1460 unsigned int s_quota_types; /* Bitmask of supported quota types */
1461 struct quota_info s_dquot; /* Diskquota specific options */
1463 struct sb_writers s_writers;
1466 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1467 * s_fsnotify_marks together for cache efficiency. They are frequently
1468 * accessed and rarely modified.
1470 void *s_fs_info; /* Filesystem private info */
1472 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1474 /* Time limits for c/m/atime in seconds */
1475 time64_t s_time_min;
1476 time64_t s_time_max;
1477 #ifdef CONFIG_FSNOTIFY
1478 __u32 s_fsnotify_mask;
1479 struct fsnotify_mark_connector __rcu *s_fsnotify_marks;
1482 char s_id[32]; /* Informational name */
1483 uuid_t s_uuid; /* UUID */
1485 unsigned int s_max_links;
1489 * The next field is for VFS *only*. No filesystems have any business
1490 * even looking at it. You had been warned.
1492 struct mutex s_vfs_rename_mutex; /* Kludge */
1495 * Filesystem subtype. If non-empty the filesystem type field
1496 * in /proc/mounts will be "type.subtype"
1498 const char *s_subtype;
1500 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1503 * Saved pool identifier for cleancache (-1 means none)
1505 int cleancache_poolid;
1507 struct shrinker s_shrink; /* per-sb shrinker handle */
1509 /* Number of inodes with nlink == 0 but still referenced */
1510 atomic_long_t s_remove_count;
1512 /* Pending fsnotify inode refs */
1513 atomic_long_t s_fsnotify_inode_refs;
1515 /* Being remounted read-only */
1516 int s_readonly_remount;
1518 /* AIO completions deferred from interrupt context */
1519 struct workqueue_struct *s_dio_done_wq;
1520 struct hlist_head s_pins;
1523 * Owning user namespace and default context in which to
1524 * interpret filesystem uids, gids, quotas, device nodes,
1525 * xattrs and security labels.
1527 struct user_namespace *s_user_ns;
1530 * The list_lru structure is essentially just a pointer to a table
1531 * of per-node lru lists, each of which has its own spinlock.
1532 * There is no need to put them into separate cachelines.
1534 struct list_lru s_dentry_lru;
1535 struct list_lru s_inode_lru;
1536 struct rcu_head rcu;
1537 struct work_struct destroy_work;
1539 struct mutex s_sync_lock; /* sync serialisation lock */
1542 * Indicates how deep in a filesystem stack this SB is
1546 /* s_inode_list_lock protects s_inodes */
1547 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1548 struct list_head s_inodes; /* all inodes */
1550 spinlock_t s_inode_wblist_lock;
1551 struct list_head s_inodes_wb; /* writeback inodes */
1552 } __randomize_layout;
1554 /* Helper functions so that in most cases filesystems will
1555 * not need to deal directly with kuid_t and kgid_t and can
1556 * instead deal with the raw numeric values that are stored
1557 * in the filesystem.
1559 static inline uid_t i_uid_read(const struct inode *inode)
1561 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1564 static inline gid_t i_gid_read(const struct inode *inode)
1566 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1569 static inline void i_uid_write(struct inode *inode, uid_t uid)
1571 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1574 static inline void i_gid_write(struct inode *inode, gid_t gid)
1576 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1579 extern struct timespec64 timespec64_trunc(struct timespec64 t, unsigned gran);
1580 extern struct timespec64 current_time(struct inode *inode);
1583 * Snapshotting support.
1586 void __sb_end_write(struct super_block *sb, int level);
1587 int __sb_start_write(struct super_block *sb, int level, bool wait);
1589 #define __sb_writers_acquired(sb, lev) \
1590 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1591 #define __sb_writers_release(sb, lev) \
1592 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1595 * sb_end_write - drop write access to a superblock
1596 * @sb: the super we wrote to
1598 * Decrement number of writers to the filesystem. Wake up possible waiters
1599 * wanting to freeze the filesystem.
1601 static inline void sb_end_write(struct super_block *sb)
1603 __sb_end_write(sb, SB_FREEZE_WRITE);
1607 * sb_end_pagefault - drop write access to a superblock from a page fault
1608 * @sb: the super we wrote to
1610 * Decrement number of processes handling write page fault to the filesystem.
1611 * Wake up possible waiters wanting to freeze the filesystem.
1613 static inline void sb_end_pagefault(struct super_block *sb)
1615 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1619 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1620 * @sb: the super we wrote to
1622 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1623 * waiters wanting to freeze the filesystem.
1625 static inline void sb_end_intwrite(struct super_block *sb)
1627 __sb_end_write(sb, SB_FREEZE_FS);
1631 * sb_start_write - get write access to a superblock
1632 * @sb: the super we write to
1634 * When a process wants to write data or metadata to a file system (i.e. dirty
1635 * a page or an inode), it should embed the operation in a sb_start_write() -
1636 * sb_end_write() pair to get exclusion against file system freezing. This
1637 * function increments number of writers preventing freezing. If the file
1638 * system is already frozen, the function waits until the file system is
1641 * Since freeze protection behaves as a lock, users have to preserve
1642 * ordering of freeze protection and other filesystem locks. Generally,
1643 * freeze protection should be the outermost lock. In particular, we have:
1646 * -> i_mutex (write path, truncate, directory ops, ...)
1647 * -> s_umount (freeze_super, thaw_super)
1649 static inline void sb_start_write(struct super_block *sb)
1651 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1654 static inline int sb_start_write_trylock(struct super_block *sb)
1656 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1660 * sb_start_pagefault - get write access to a superblock from a page fault
1661 * @sb: the super we write to
1663 * When a process starts handling write page fault, it should embed the
1664 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1665 * exclusion against file system freezing. This is needed since the page fault
1666 * is going to dirty a page. This function increments number of running page
1667 * faults preventing freezing. If the file system is already frozen, the
1668 * function waits until the file system is thawed.
1670 * Since page fault freeze protection behaves as a lock, users have to preserve
1671 * ordering of freeze protection and other filesystem locks. It is advised to
1672 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1673 * handling code implies lock dependency:
1676 * -> sb_start_pagefault
1678 static inline void sb_start_pagefault(struct super_block *sb)
1680 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1684 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1685 * @sb: the super we write to
1687 * This is the third level of protection against filesystem freezing. It is
1688 * free for use by a filesystem. The only requirement is that it must rank
1689 * below sb_start_pagefault.
1691 * For example filesystem can call sb_start_intwrite() when starting a
1692 * transaction which somewhat eases handling of freezing for internal sources
1693 * of filesystem changes (internal fs threads, discarding preallocation on file
1696 static inline void sb_start_intwrite(struct super_block *sb)
1698 __sb_start_write(sb, SB_FREEZE_FS, true);
1701 static inline int sb_start_intwrite_trylock(struct super_block *sb)
1703 return __sb_start_write(sb, SB_FREEZE_FS, false);
1707 extern bool inode_owner_or_capable(const struct inode *inode);
1710 * VFS helper functions..
1712 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1713 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1714 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1715 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1716 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1717 extern int vfs_rmdir(struct inode *, struct dentry *);
1718 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1719 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1720 extern int vfs_whiteout(struct inode *, struct dentry *);
1722 extern struct dentry *vfs_tmpfile(struct dentry *dentry, umode_t mode,
1725 int vfs_mkobj(struct dentry *, umode_t,
1726 int (*f)(struct dentry *, umode_t, void *),
1729 extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1731 #ifdef CONFIG_COMPAT
1732 extern long compat_ptr_ioctl(struct file *file, unsigned int cmd,
1735 #define compat_ptr_ioctl NULL
1739 * VFS file helper functions.
1741 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1743 extern bool may_open_dev(const struct path *path);
1745 * VFS FS_IOC_FIEMAP helper definitions.
1747 struct fiemap_extent_info {
1748 unsigned int fi_flags; /* Flags as passed from user */
1749 unsigned int fi_extents_mapped; /* Number of mapped extents */
1750 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1751 struct fiemap_extent __user *fi_extents_start; /* Start of
1752 fiemap_extent array */
1754 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1755 u64 phys, u64 len, u32 flags);
1756 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1759 * This is the "filldir" function type, used by readdir() to let
1760 * the kernel specify what kind of dirent layout it wants to have.
1761 * This allows the kernel to read directories into kernel space or
1762 * to have different dirent layouts depending on the binary type.
1765 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1768 struct dir_context {
1773 struct block_device_operations;
1775 /* These macros are for out of kernel modules to test that
1776 * the kernel supports the unlocked_ioctl and compat_ioctl
1777 * fields in struct file_operations. */
1778 #define HAVE_COMPAT_IOCTL 1
1779 #define HAVE_UNLOCKED_IOCTL 1
1782 * These flags let !MMU mmap() govern direct device mapping vs immediate
1783 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1785 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1786 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1787 * NOMMU_MAP_READ: Can be mapped for reading
1788 * NOMMU_MAP_WRITE: Can be mapped for writing
1789 * NOMMU_MAP_EXEC: Can be mapped for execution
1791 #define NOMMU_MAP_COPY 0x00000001
1792 #define NOMMU_MAP_DIRECT 0x00000008
1793 #define NOMMU_MAP_READ VM_MAYREAD
1794 #define NOMMU_MAP_WRITE VM_MAYWRITE
1795 #define NOMMU_MAP_EXEC VM_MAYEXEC
1797 #define NOMMU_VMFLAGS \
1798 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1801 * These flags control the behavior of the remap_file_range function pointer.
1802 * If it is called with len == 0 that means "remap to end of source file".
1803 * See Documentation/filesystems/vfs.rst for more details about this call.
1805 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
1806 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
1808 #define REMAP_FILE_DEDUP (1 << 0)
1809 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
1812 * These flags signal that the caller is ok with altering various aspects of
1813 * the behavior of the remap operation. The changes must be made by the
1814 * implementation; the vfs remap helper functions can take advantage of them.
1815 * Flags in this category exist to preserve the quirky behavior of the hoisted
1816 * btrfs clone/dedupe ioctls.
1818 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
1822 struct file_operations {
1823 struct module *owner;
1824 loff_t (*llseek) (struct file *, loff_t, int);
1825 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1826 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1827 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1828 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1829 int (*iopoll)(struct kiocb *kiocb, bool spin);
1830 int (*iterate) (struct file *, struct dir_context *);
1831 int (*iterate_shared) (struct file *, struct dir_context *);
1832 __poll_t (*poll) (struct file *, struct poll_table_struct *);
1833 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1834 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1835 int (*mmap) (struct file *, struct vm_area_struct *);
1836 unsigned long mmap_supported_flags;
1837 int (*open) (struct inode *, struct file *);
1838 int (*flush) (struct file *, fl_owner_t id);
1839 int (*release) (struct inode *, struct file *);
1840 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1841 int (*fasync) (int, struct file *, int);
1842 int (*lock) (struct file *, int, struct file_lock *);
1843 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1844 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1845 int (*check_flags)(int);
1846 int (*flock) (struct file *, int, struct file_lock *);
1847 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1848 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1849 int (*setlease)(struct file *, long, struct file_lock **, void **);
1850 long (*fallocate)(struct file *file, int mode, loff_t offset,
1852 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1854 unsigned (*mmap_capabilities)(struct file *);
1856 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1857 loff_t, size_t, unsigned int);
1858 loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
1859 struct file *file_out, loff_t pos_out,
1860 loff_t len, unsigned int remap_flags);
1861 int (*fadvise)(struct file *, loff_t, loff_t, int);
1863 } __randomize_layout;
1865 struct inode_operations {
1866 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1867 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1868 int (*permission) (struct inode *, int);
1869 struct posix_acl * (*get_acl)(struct inode *, int);
1871 int (*readlink) (struct dentry *, char __user *,int);
1873 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1874 int (*link) (struct dentry *,struct inode *,struct dentry *);
1875 int (*unlink) (struct inode *,struct dentry *);
1876 int (*symlink) (struct inode *,struct dentry *,const char *);
1877 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1878 int (*rmdir) (struct inode *,struct dentry *);
1879 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1880 int (*rename) (struct inode *, struct dentry *,
1881 struct inode *, struct dentry *, unsigned int);
1882 int (*setattr) (struct dentry *, struct iattr *);
1883 int (*getattr) (const struct path *, struct kstat *, u32, unsigned int);
1884 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1885 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1887 int (*update_time)(struct inode *, struct timespec64 *, int);
1888 int (*atomic_open)(struct inode *, struct dentry *,
1889 struct file *, unsigned open_flag,
1890 umode_t create_mode);
1891 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1892 int (*set_acl)(struct inode *, struct posix_acl *, int);
1893 } ____cacheline_aligned;
1895 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
1896 struct iov_iter *iter)
1898 return file->f_op->read_iter(kio, iter);
1901 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
1902 struct iov_iter *iter)
1904 return file->f_op->write_iter(kio, iter);
1907 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
1909 return file->f_op->mmap(file, vma);
1912 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1913 unsigned long nr_segs, unsigned long fast_segs,
1914 struct iovec *fast_pointer,
1915 struct iovec **ret_pointer);
1917 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1918 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1919 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1920 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1921 unsigned long, loff_t *, rwf_t);
1922 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1923 loff_t, size_t, unsigned int);
1924 extern ssize_t generic_copy_file_range(struct file *file_in, loff_t pos_in,
1925 struct file *file_out, loff_t pos_out,
1926 size_t len, unsigned int flags);
1927 extern int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
1928 struct file *file_out, loff_t pos_out,
1930 unsigned int remap_flags);
1931 extern loff_t do_clone_file_range(struct file *file_in, loff_t pos_in,
1932 struct file *file_out, loff_t pos_out,
1933 loff_t len, unsigned int remap_flags);
1934 extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1935 struct file *file_out, loff_t pos_out,
1936 loff_t len, unsigned int remap_flags);
1937 extern int vfs_dedupe_file_range(struct file *file,
1938 struct file_dedupe_range *same);
1939 extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
1940 struct file *dst_file, loff_t dst_pos,
1941 loff_t len, unsigned int remap_flags);
1944 struct super_operations {
1945 struct inode *(*alloc_inode)(struct super_block *sb);
1946 void (*destroy_inode)(struct inode *);
1947 void (*free_inode)(struct inode *);
1949 void (*dirty_inode) (struct inode *, int flags);
1950 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1951 int (*drop_inode) (struct inode *);
1952 void (*evict_inode) (struct inode *);
1953 void (*put_super) (struct super_block *);
1954 int (*sync_fs)(struct super_block *sb, int wait);
1955 int (*freeze_super) (struct super_block *);
1956 int (*freeze_fs) (struct super_block *);
1957 int (*thaw_super) (struct super_block *);
1958 int (*unfreeze_fs) (struct super_block *);
1959 int (*statfs) (struct dentry *, struct kstatfs *);
1960 int (*remount_fs) (struct super_block *, int *, char *);
1961 void (*umount_begin) (struct super_block *);
1963 int (*show_options)(struct seq_file *, struct dentry *);
1964 int (*show_devname)(struct seq_file *, struct dentry *);
1965 int (*show_path)(struct seq_file *, struct dentry *);
1966 int (*show_stats)(struct seq_file *, struct dentry *);
1968 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1969 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1970 struct dquot **(*get_dquots)(struct inode *);
1972 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1973 long (*nr_cached_objects)(struct super_block *,
1974 struct shrink_control *);
1975 long (*free_cached_objects)(struct super_block *,
1976 struct shrink_control *);
1980 * Inode flags - they have no relation to superblock flags now
1982 #define S_SYNC 1 /* Writes are synced at once */
1983 #define S_NOATIME 2 /* Do not update access times */
1984 #define S_APPEND 4 /* Append-only file */
1985 #define S_IMMUTABLE 8 /* Immutable file */
1986 #define S_DEAD 16 /* removed, but still open directory */
1987 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1988 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1989 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1990 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1991 #define S_PRIVATE 512 /* Inode is fs-internal */
1992 #define S_IMA 1024 /* Inode has an associated IMA struct */
1993 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1994 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1995 #ifdef CONFIG_FS_DAX
1996 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1998 #define S_DAX 0 /* Make all the DAX code disappear */
2000 #define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */
2001 #define S_CASEFOLD 32768 /* Casefolded file */
2002 #define S_VERITY 65536 /* Verity file (using fs/verity/) */
2005 * Note that nosuid etc flags are inode-specific: setting some file-system
2006 * flags just means all the inodes inherit those flags by default. It might be
2007 * possible to override it selectively if you really wanted to with some
2008 * ioctl() that is not currently implemented.
2010 * Exception: SB_RDONLY is always applied to the entire file system.
2012 * Unfortunately, it is possible to change a filesystems flags with it mounted
2013 * with files in use. This means that all of the inodes will not have their
2014 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
2015 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
2017 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
2019 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
2020 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
2021 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
2022 ((inode)->i_flags & S_SYNC))
2023 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
2024 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
2025 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
2026 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
2027 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
2029 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
2030 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
2031 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
2032 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
2034 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
2035 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
2036 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
2037 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
2038 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
2039 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2040 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2041 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2042 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2043 #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD)
2044 #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY)
2046 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2047 (inode)->i_rdev == WHITEOUT_DEV)
2049 static inline bool HAS_UNMAPPED_ID(struct inode *inode)
2051 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
2054 static inline enum rw_hint file_write_hint(struct file *file)
2056 if (file->f_write_hint != WRITE_LIFE_NOT_SET)
2057 return file->f_write_hint;
2059 return file_inode(file)->i_write_hint;
2062 static inline int iocb_flags(struct file *file);
2064 static inline u16 ki_hint_validate(enum rw_hint hint)
2066 typeof(((struct kiocb *)0)->ki_hint) max_hint = -1;
2068 if (hint <= max_hint)
2073 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
2075 *kiocb = (struct kiocb) {
2077 .ki_flags = iocb_flags(filp),
2078 .ki_hint = ki_hint_validate(file_write_hint(filp)),
2079 .ki_ioprio = get_current_ioprio(),
2084 * Inode state bits. Protected by inode->i_lock
2086 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
2087 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
2089 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2090 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2091 * various stages of removing an inode.
2093 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2095 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2096 * fdatasync(). i_atime is the usual cause.
2097 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2098 * these changes separately from I_DIRTY_SYNC so that we
2099 * don't have to write inode on fdatasync() when only
2100 * mtime has changed in it.
2101 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2102 * I_NEW Serves as both a mutex and completion notification.
2103 * New inodes set I_NEW. If two processes both create
2104 * the same inode, one of them will release its inode and
2105 * wait for I_NEW to be released before returning.
2106 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2107 * also cause waiting on I_NEW, without I_NEW actually
2108 * being set. find_inode() uses this to prevent returning
2109 * nearly-dead inodes.
2110 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2111 * is zero. I_FREEING must be set when I_WILL_FREE is
2113 * I_FREEING Set when inode is about to be freed but still has dirty
2114 * pages or buffers attached or the inode itself is still
2116 * I_CLEAR Added by clear_inode(). In this state the inode is
2117 * clean and can be destroyed. Inode keeps I_FREEING.
2119 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2120 * prohibited for many purposes. iget() must wait for
2121 * the inode to be completely released, then create it
2122 * anew. Other functions will just ignore such inodes,
2123 * if appropriate. I_NEW is used for waiting.
2125 * I_SYNC Writeback of inode is running. The bit is set during
2126 * data writeback, and cleared with a wakeup on the bit
2127 * address once it is done. The bit is also used to pin
2128 * the inode in memory for flusher thread.
2130 * I_REFERENCED Marks the inode as recently references on the LRU list.
2132 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2134 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2135 * synchronize competing switching instances and to tell
2136 * wb stat updates to grab the i_pages lock. See
2137 * inode_switch_wbs_work_fn() for details.
2139 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2140 * and work dirs among overlayfs mounts.
2142 * I_CREATING New object's inode in the middle of setting up.
2144 * I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists.
2145 * Used to detect that mark_inode_dirty() should not move
2146 * inode between dirty lists.
2148 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2150 #define I_DIRTY_SYNC (1 << 0)
2151 #define I_DIRTY_DATASYNC (1 << 1)
2152 #define I_DIRTY_PAGES (1 << 2)
2154 #define I_NEW (1 << __I_NEW)
2155 #define I_WILL_FREE (1 << 4)
2156 #define I_FREEING (1 << 5)
2157 #define I_CLEAR (1 << 6)
2159 #define I_SYNC (1 << __I_SYNC)
2160 #define I_REFERENCED (1 << 8)
2161 #define __I_DIO_WAKEUP 9
2162 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2163 #define I_LINKABLE (1 << 10)
2164 #define I_DIRTY_TIME (1 << 11)
2165 #define I_WB_SWITCH (1 << 13)
2166 #define I_OVL_INUSE (1 << 14)
2167 #define I_CREATING (1 << 15)
2168 #define I_SYNC_QUEUED (1 << 17)
2170 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2171 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2172 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2174 extern void __mark_inode_dirty(struct inode *, int);
2175 static inline void mark_inode_dirty(struct inode *inode)
2177 __mark_inode_dirty(inode, I_DIRTY);
2180 static inline void mark_inode_dirty_sync(struct inode *inode)
2182 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2185 extern void inc_nlink(struct inode *inode);
2186 extern void drop_nlink(struct inode *inode);
2187 extern void clear_nlink(struct inode *inode);
2188 extern void set_nlink(struct inode *inode, unsigned int nlink);
2190 static inline void inode_inc_link_count(struct inode *inode)
2193 mark_inode_dirty(inode);
2196 static inline void inode_dec_link_count(struct inode *inode)
2199 mark_inode_dirty(inode);
2202 enum file_time_flags {
2209 extern bool atime_needs_update(const struct path *, struct inode *);
2210 extern void touch_atime(const struct path *);
2211 static inline void file_accessed(struct file *file)
2213 if (!(file->f_flags & O_NOATIME))
2214 touch_atime(&file->f_path);
2217 extern int file_modified(struct file *file);
2219 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2220 int sync_inode_metadata(struct inode *inode, int wait);
2222 struct file_system_type {
2225 #define FS_REQUIRES_DEV 1
2226 #define FS_BINARY_MOUNTDATA 2
2227 #define FS_HAS_SUBTYPE 4
2228 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2229 #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */
2230 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2231 int (*init_fs_context)(struct fs_context *);
2232 const struct fs_parameter_description *parameters;
2233 struct dentry *(*mount) (struct file_system_type *, int,
2234 const char *, void *);
2235 void (*kill_sb) (struct super_block *);
2236 struct module *owner;
2237 struct file_system_type * next;
2238 struct hlist_head fs_supers;
2240 struct lock_class_key s_lock_key;
2241 struct lock_class_key s_umount_key;
2242 struct lock_class_key s_vfs_rename_key;
2243 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2245 struct lock_class_key i_lock_key;
2246 struct lock_class_key i_mutex_key;
2247 struct lock_class_key i_mutex_dir_key;
2250 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2253 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2254 int flags, const char *dev_name, void *data,
2255 int (*fill_super)(struct super_block *, void *, int));
2257 static inline struct dentry *mount_bdev(struct file_system_type *fs_type,
2258 int flags, const char *dev_name, void *data,
2259 int (*fill_super)(struct super_block *, void *, int))
2261 return ERR_PTR(-ENODEV);
2264 extern struct dentry *mount_single(struct file_system_type *fs_type,
2265 int flags, void *data,
2266 int (*fill_super)(struct super_block *, void *, int));
2267 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2268 int flags, void *data,
2269 int (*fill_super)(struct super_block *, void *, int));
2270 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2271 void generic_shutdown_super(struct super_block *sb);
2273 void kill_block_super(struct super_block *sb);
2275 static inline void kill_block_super(struct super_block *sb)
2280 void kill_anon_super(struct super_block *sb);
2281 void kill_litter_super(struct super_block *sb);
2282 void deactivate_super(struct super_block *sb);
2283 void deactivate_locked_super(struct super_block *sb);
2284 int set_anon_super(struct super_block *s, void *data);
2285 int set_anon_super_fc(struct super_block *s, struct fs_context *fc);
2286 int get_anon_bdev(dev_t *);
2287 void free_anon_bdev(dev_t);
2288 struct super_block *sget_fc(struct fs_context *fc,
2289 int (*test)(struct super_block *, struct fs_context *),
2290 int (*set)(struct super_block *, struct fs_context *));
2291 struct super_block *sget(struct file_system_type *type,
2292 int (*test)(struct super_block *,void *),
2293 int (*set)(struct super_block *,void *),
2294 int flags, void *data);
2296 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2297 #define fops_get(fops) \
2298 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2299 #define fops_put(fops) \
2300 do { if (fops) module_put((fops)->owner); } while(0)
2302 * This one is to be used *ONLY* from ->open() instances.
2303 * fops must be non-NULL, pinned down *and* module dependencies
2304 * should be sufficient to pin the caller down as well.
2306 #define replace_fops(f, fops) \
2308 struct file *__file = (f); \
2309 fops_put(__file->f_op); \
2310 BUG_ON(!(__file->f_op = (fops))); \
2313 extern int register_filesystem(struct file_system_type *);
2314 extern int unregister_filesystem(struct file_system_type *);
2315 extern struct vfsmount *kern_mount(struct file_system_type *);
2316 extern void kern_unmount(struct vfsmount *mnt);
2317 extern int may_umount_tree(struct vfsmount *);
2318 extern int may_umount(struct vfsmount *);
2319 extern long do_mount(const char *, const char __user *,
2320 const char *, unsigned long, void *);
2321 extern struct vfsmount *collect_mounts(const struct path *);
2322 extern void drop_collected_mounts(struct vfsmount *);
2323 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2325 extern int vfs_statfs(const struct path *, struct kstatfs *);
2326 extern int user_statfs(const char __user *, struct kstatfs *);
2327 extern int fd_statfs(int, struct kstatfs *);
2328 extern int freeze_super(struct super_block *super);
2329 extern int thaw_super(struct super_block *super);
2330 extern bool our_mnt(struct vfsmount *mnt);
2331 extern __printf(2, 3)
2332 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2333 extern int super_setup_bdi(struct super_block *sb);
2335 extern int current_umask(void);
2337 extern void ihold(struct inode * inode);
2338 extern void iput(struct inode *);
2339 extern int generic_update_time(struct inode *, struct timespec64 *, int);
2342 extern struct kobject *fs_kobj;
2344 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2346 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2347 extern int locks_mandatory_locked(struct file *);
2348 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2351 * Candidates for mandatory locking have the setgid bit set
2352 * but no group execute bit - an otherwise meaningless combination.
2355 static inline int __mandatory_lock(struct inode *ino)
2357 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2361 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2362 * otherwise these will be advisory locks
2365 static inline int mandatory_lock(struct inode *ino)
2367 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2370 static inline int locks_verify_locked(struct file *file)
2372 if (mandatory_lock(locks_inode(file)))
2373 return locks_mandatory_locked(file);
2377 static inline int locks_verify_truncate(struct inode *inode,
2381 if (!inode->i_flctx || !mandatory_lock(inode))
2384 if (size < inode->i_size) {
2385 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2388 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2393 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2395 static inline int locks_mandatory_locked(struct file *file)
2400 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2401 loff_t start, loff_t end, unsigned char type)
2406 static inline int __mandatory_lock(struct inode *inode)
2411 static inline int mandatory_lock(struct inode *inode)
2416 static inline int locks_verify_locked(struct file *file)
2421 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2427 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2430 #ifdef CONFIG_FILE_LOCKING
2431 static inline int break_lease(struct inode *inode, unsigned int mode)
2434 * Since this check is lockless, we must ensure that any refcounts
2435 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2436 * could end up racing with tasks trying to set a new lease on this
2440 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2441 return __break_lease(inode, mode, FL_LEASE);
2445 static inline int break_deleg(struct inode *inode, unsigned int mode)
2448 * Since this check is lockless, we must ensure that any refcounts
2449 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2450 * could end up racing with tasks trying to set a new lease on this
2454 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2455 return __break_lease(inode, mode, FL_DELEG);
2459 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2463 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2464 if (ret == -EWOULDBLOCK && delegated_inode) {
2465 *delegated_inode = inode;
2471 static inline int break_deleg_wait(struct inode **delegated_inode)
2475 ret = break_deleg(*delegated_inode, O_WRONLY);
2476 iput(*delegated_inode);
2477 *delegated_inode = NULL;
2481 static inline int break_layout(struct inode *inode, bool wait)
2484 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2485 return __break_lease(inode,
2486 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2491 #else /* !CONFIG_FILE_LOCKING */
2492 static inline int break_lease(struct inode *inode, unsigned int mode)
2497 static inline int break_deleg(struct inode *inode, unsigned int mode)
2502 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2507 static inline int break_deleg_wait(struct inode **delegated_inode)
2513 static inline int break_layout(struct inode *inode, bool wait)
2518 #endif /* CONFIG_FILE_LOCKING */
2523 const char *name; /* pointer to actual string */
2524 const __user char *uptr; /* original userland pointer */
2526 struct audit_names *aname;
2529 static_assert(offsetof(struct filename, iname) % sizeof(long) == 0);
2531 extern long vfs_truncate(const struct path *, loff_t);
2532 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2534 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2536 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2538 extern struct file *file_open_name(struct filename *, int, umode_t);
2539 extern struct file *filp_open(const char *, int, umode_t);
2540 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2541 const char *, int, umode_t);
2542 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2543 extern struct file * open_with_fake_path(const struct path *, int,
2544 struct inode*, const struct cred *);
2545 static inline struct file *file_clone_open(struct file *file)
2547 return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2549 extern int filp_close(struct file *, fl_owner_t id);
2551 extern struct filename *getname_flags(const char __user *, int, int *);
2552 extern struct filename *getname(const char __user *);
2553 extern struct filename *getname_kernel(const char *);
2554 extern void putname(struct filename *name);
2556 extern int finish_open(struct file *file, struct dentry *dentry,
2557 int (*open)(struct inode *, struct file *));
2558 extern int finish_no_open(struct file *file, struct dentry *dentry);
2562 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2565 extern void __init vfs_caches_init_early(void);
2566 extern void __init vfs_caches_init(void);
2568 extern struct kmem_cache *names_cachep;
2570 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2571 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2574 extern int register_blkdev(unsigned int, const char *);
2575 extern void unregister_blkdev(unsigned int, const char *);
2576 extern void bdev_unhash_inode(dev_t dev);
2577 extern struct block_device *bdget(dev_t);
2578 extern struct block_device *bdgrab(struct block_device *bdev);
2579 extern void bd_set_size(struct block_device *, loff_t size);
2580 extern void bd_forget(struct inode *inode);
2581 extern void bdput(struct block_device *);
2582 extern void invalidate_bdev(struct block_device *);
2583 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2584 extern int sync_blockdev(struct block_device *bdev);
2585 extern void kill_bdev(struct block_device *);
2586 extern struct super_block *freeze_bdev(struct block_device *);
2587 extern void emergency_thaw_all(void);
2588 extern void emergency_thaw_bdev(struct super_block *sb);
2589 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2590 extern int fsync_bdev(struct block_device *);
2592 extern struct super_block *blockdev_superblock;
2594 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2596 return sb == blockdev_superblock;
2599 static inline void bd_forget(struct inode *inode) {}
2600 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2601 static inline void kill_bdev(struct block_device *bdev) {}
2602 static inline void invalidate_bdev(struct block_device *bdev) {}
2604 static inline struct super_block *freeze_bdev(struct block_device *sb)
2609 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2614 static inline int emergency_thaw_bdev(struct super_block *sb)
2619 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2623 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2628 extern int sync_filesystem(struct super_block *);
2629 extern const struct file_operations def_blk_fops;
2630 extern const struct file_operations def_chr_fops;
2632 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2633 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2634 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2635 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2636 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2638 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2640 extern struct block_device *bd_start_claiming(struct block_device *bdev,
2642 extern void bd_finish_claiming(struct block_device *bdev,
2643 struct block_device *whole, void *holder);
2644 extern void bd_abort_claiming(struct block_device *bdev,
2645 struct block_device *whole, void *holder);
2646 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2647 extern int __blkdev_reread_part(struct block_device *bdev);
2648 extern int blkdev_reread_part(struct block_device *bdev);
2651 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2652 extern void bd_unlink_disk_holder(struct block_device *bdev,
2653 struct gendisk *disk);
2655 static inline int bd_link_disk_holder(struct block_device *bdev,
2656 struct gendisk *disk)
2660 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2661 struct gendisk *disk)
2668 #define CHRDEV_MAJOR_MAX 512
2669 /* Marks the bottom of the first segment of free char majors */
2670 #define CHRDEV_MAJOR_DYN_END 234
2671 /* Marks the top and bottom of the second segment of free char majors */
2672 #define CHRDEV_MAJOR_DYN_EXT_START 511
2673 #define CHRDEV_MAJOR_DYN_EXT_END 384
2675 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2676 extern int register_chrdev_region(dev_t, unsigned, const char *);
2677 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2678 unsigned int count, const char *name,
2679 const struct file_operations *fops);
2680 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2681 unsigned int count, const char *name);
2682 extern void unregister_chrdev_region(dev_t, unsigned);
2683 extern void chrdev_show(struct seq_file *,off_t);
2685 static inline int register_chrdev(unsigned int major, const char *name,
2686 const struct file_operations *fops)
2688 return __register_chrdev(major, 0, 256, name, fops);
2691 static inline void unregister_chrdev(unsigned int major, const char *name)
2693 __unregister_chrdev(major, 0, 256, name);
2696 /* fs/block_dev.c */
2697 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2698 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2701 #define BLKDEV_MAJOR_MAX 512
2702 extern const char *__bdevname(dev_t, char *buffer);
2703 extern const char *bdevname(struct block_device *bdev, char *buffer);
2704 extern struct block_device *lookup_bdev(const char *);
2705 extern void blkdev_show(struct seq_file *,off_t);
2708 #define BLKDEV_MAJOR_MAX 0
2711 extern void init_special_inode(struct inode *, umode_t, dev_t);
2713 /* Invalid inode operations -- fs/bad_inode.c */
2714 extern void make_bad_inode(struct inode *);
2715 extern bool is_bad_inode(struct inode *);
2718 extern void check_disk_size_change(struct gendisk *disk,
2719 struct block_device *bdev, bool verbose);
2720 extern int revalidate_disk(struct gendisk *);
2721 extern int check_disk_change(struct block_device *);
2722 extern int __invalidate_device(struct block_device *, bool);
2723 extern int invalidate_partition(struct gendisk *, int);
2725 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2726 pgoff_t start, pgoff_t end);
2728 static inline void invalidate_remote_inode(struct inode *inode)
2730 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2731 S_ISLNK(inode->i_mode))
2732 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2734 extern int invalidate_inode_pages2(struct address_space *mapping);
2735 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2736 pgoff_t start, pgoff_t end);
2737 extern int write_inode_now(struct inode *, int);
2738 extern int filemap_fdatawrite(struct address_space *);
2739 extern int filemap_flush(struct address_space *);
2740 extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2741 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2743 extern int filemap_fdatawait_range_keep_errors(struct address_space *mapping,
2744 loff_t start_byte, loff_t end_byte);
2746 static inline int filemap_fdatawait(struct address_space *mapping)
2748 return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2751 extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2753 extern int filemap_write_and_wait(struct address_space *mapping);
2754 extern int filemap_write_and_wait_range(struct address_space *mapping,
2755 loff_t lstart, loff_t lend);
2756 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2757 loff_t start, loff_t end, int sync_mode);
2758 extern int filemap_fdatawrite_range(struct address_space *mapping,
2759 loff_t start, loff_t end);
2760 extern int filemap_check_errors(struct address_space *mapping);
2761 extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2763 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2765 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2766 extern int __must_check file_write_and_wait_range(struct file *file,
2767 loff_t start, loff_t end);
2769 static inline int file_write_and_wait(struct file *file)
2771 return file_write_and_wait_range(file, 0, LLONG_MAX);
2775 * filemap_set_wb_err - set a writeback error on an address_space
2776 * @mapping: mapping in which to set writeback error
2777 * @err: error to be set in mapping
2779 * When writeback fails in some way, we must record that error so that
2780 * userspace can be informed when fsync and the like are called. We endeavor
2781 * to report errors on any file that was open at the time of the error. Some
2782 * internal callers also need to know when writeback errors have occurred.
2784 * When a writeback error occurs, most filesystems will want to call
2785 * filemap_set_wb_err to record the error in the mapping so that it will be
2786 * automatically reported whenever fsync is called on the file.
2788 static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2790 /* Fastpath for common case of no error */
2792 __filemap_set_wb_err(mapping, err);
2796 * filemap_check_wb_error - has an error occurred since the mark was sampled?
2797 * @mapping: mapping to check for writeback errors
2798 * @since: previously-sampled errseq_t
2800 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2801 * the given value was sampled.
2803 * If it has then report the latest error set, otherwise return 0.
2805 static inline int filemap_check_wb_err(struct address_space *mapping,
2808 return errseq_check(&mapping->wb_err, since);
2812 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2813 * @mapping: mapping to be sampled
2815 * Writeback errors are always reported relative to a particular sample point
2816 * in the past. This function provides those sample points.
2818 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2820 return errseq_sample(&mapping->wb_err);
2823 static inline int filemap_nr_thps(struct address_space *mapping)
2825 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
2826 return atomic_read(&mapping->nr_thps);
2832 static inline void filemap_nr_thps_inc(struct address_space *mapping)
2834 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
2835 atomic_inc(&mapping->nr_thps);
2841 static inline void filemap_nr_thps_dec(struct address_space *mapping)
2843 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
2844 atomic_dec(&mapping->nr_thps);
2850 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2852 extern int vfs_fsync(struct file *file, int datasync);
2854 extern int sync_file_range(struct file *file, loff_t offset, loff_t nbytes,
2855 unsigned int flags);
2858 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2859 * to already be updated for the write, and will return either the amount
2860 * of bytes passed in, or an error if syncing the file failed.
2862 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2864 if (iocb->ki_flags & IOCB_DSYNC) {
2865 int ret = vfs_fsync_range(iocb->ki_filp,
2866 iocb->ki_pos - count, iocb->ki_pos - 1,
2867 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2875 extern void emergency_sync(void);
2876 extern void emergency_remount(void);
2878 extern sector_t bmap(struct inode *, sector_t);
2880 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2881 extern int inode_permission(struct inode *, int);
2882 extern int generic_permission(struct inode *, int);
2883 extern int __check_sticky(struct inode *dir, struct inode *inode);
2885 static inline bool execute_ok(struct inode *inode)
2887 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2890 static inline void file_start_write(struct file *file)
2892 if (!S_ISREG(file_inode(file)->i_mode))
2894 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2897 static inline bool file_start_write_trylock(struct file *file)
2899 if (!S_ISREG(file_inode(file)->i_mode))
2901 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2904 static inline void file_end_write(struct file *file)
2906 if (!S_ISREG(file_inode(file)->i_mode))
2908 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2912 * get_write_access() gets write permission for a file.
2913 * put_write_access() releases this write permission.
2914 * This is used for regular files.
2915 * We cannot support write (and maybe mmap read-write shared) accesses and
2916 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2917 * can have the following values:
2918 * 0: no writers, no VM_DENYWRITE mappings
2919 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2920 * > 0: (i_writecount) users are writing to the file.
2922 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2923 * except for the cases where we don't hold i_writecount yet. Then we need to
2924 * use {get,deny}_write_access() - these functions check the sign and refuse
2925 * to do the change if sign is wrong.
2927 static inline int get_write_access(struct inode *inode)
2929 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2931 static inline int deny_write_access(struct file *file)
2933 struct inode *inode = file_inode(file);
2934 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2936 static inline void put_write_access(struct inode * inode)
2938 atomic_dec(&inode->i_writecount);
2940 static inline void allow_write_access(struct file *file)
2943 atomic_inc(&file_inode(file)->i_writecount);
2945 static inline bool inode_is_open_for_write(const struct inode *inode)
2947 return atomic_read(&inode->i_writecount) > 0;
2950 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
2951 static inline void i_readcount_dec(struct inode *inode)
2953 BUG_ON(!atomic_read(&inode->i_readcount));
2954 atomic_dec(&inode->i_readcount);
2956 static inline void i_readcount_inc(struct inode *inode)
2958 atomic_inc(&inode->i_readcount);
2961 static inline void i_readcount_dec(struct inode *inode)
2965 static inline void i_readcount_inc(struct inode *inode)
2970 extern int do_pipe_flags(int *, int);
2972 #define __kernel_read_file_id(id) \
2973 id(UNKNOWN, unknown) \
2974 id(FIRMWARE, firmware) \
2975 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2976 id(MODULE, kernel-module) \
2977 id(KEXEC_IMAGE, kexec-image) \
2978 id(KEXEC_INITRAMFS, kexec-initramfs) \
2979 id(POLICY, security-policy) \
2980 id(X509_CERTIFICATE, x509-certificate) \
2983 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2984 #define __fid_stringify(dummy, str) #str,
2986 enum kernel_read_file_id {
2987 __kernel_read_file_id(__fid_enumify)
2990 static const char * const kernel_read_file_str[] = {
2991 __kernel_read_file_id(__fid_stringify)
2994 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
2996 if ((unsigned)id >= READING_MAX_ID)
2997 return kernel_read_file_str[READING_UNKNOWN];
2999 return kernel_read_file_str[id];
3002 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
3003 enum kernel_read_file_id);
3004 extern int kernel_read_file_from_path(const char *, void **, loff_t *, loff_t,
3005 enum kernel_read_file_id);
3006 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
3007 enum kernel_read_file_id);
3008 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
3009 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
3010 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
3011 extern struct file * open_exec(const char *);
3013 /* fs/dcache.c -- generic fs support functions */
3014 extern bool is_subdir(struct dentry *, struct dentry *);
3015 extern bool path_is_under(const struct path *, const struct path *);
3017 extern char *file_path(struct file *, char *, int);
3019 #include <linux/err.h>
3021 /* needed for stackable file system support */
3022 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
3024 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
3026 extern int inode_init_always(struct super_block *, struct inode *);
3027 extern void inode_init_once(struct inode *);
3028 extern void address_space_init_once(struct address_space *mapping);
3029 extern struct inode * igrab(struct inode *);
3030 extern ino_t iunique(struct super_block *, ino_t);
3031 extern int inode_needs_sync(struct inode *inode);
3032 extern int generic_delete_inode(struct inode *inode);
3033 static inline int generic_drop_inode(struct inode *inode)
3035 return !inode->i_nlink || inode_unhashed(inode);
3038 extern struct inode *ilookup5_nowait(struct super_block *sb,
3039 unsigned long hashval, int (*test)(struct inode *, void *),
3041 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
3042 int (*test)(struct inode *, void *), void *data);
3043 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
3045 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
3046 int (*test)(struct inode *, void *),
3047 int (*set)(struct inode *, void *),
3049 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
3050 extern struct inode * iget_locked(struct super_block *, unsigned long);
3051 extern struct inode *find_inode_nowait(struct super_block *,
3053 int (*match)(struct inode *,
3054 unsigned long, void *),
3056 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
3057 extern int insert_inode_locked(struct inode *);
3058 #ifdef CONFIG_DEBUG_LOCK_ALLOC
3059 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
3061 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
3063 extern void unlock_new_inode(struct inode *);
3064 extern void discard_new_inode(struct inode *);
3065 extern unsigned int get_next_ino(void);
3066 extern void evict_inodes(struct super_block *sb);
3068 extern void __iget(struct inode * inode);
3069 extern void iget_failed(struct inode *);
3070 extern void clear_inode(struct inode *);
3071 extern void __destroy_inode(struct inode *);
3072 extern struct inode *new_inode_pseudo(struct super_block *sb);
3073 extern struct inode *new_inode(struct super_block *sb);
3074 extern void free_inode_nonrcu(struct inode *inode);
3075 extern int should_remove_suid(struct dentry *);
3076 extern int file_remove_privs(struct file *);
3078 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
3079 static inline void insert_inode_hash(struct inode *inode)
3081 __insert_inode_hash(inode, inode->i_ino);
3084 extern void __remove_inode_hash(struct inode *);
3085 static inline void remove_inode_hash(struct inode *inode)
3087 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
3088 __remove_inode_hash(inode);
3091 extern void inode_sb_list_add(struct inode *inode);
3094 extern int bdev_read_only(struct block_device *);
3096 extern int set_blocksize(struct block_device *, int);
3097 extern int sb_set_blocksize(struct super_block *, int);
3098 extern int sb_min_blocksize(struct super_block *, int);
3100 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
3101 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
3102 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
3103 extern int generic_remap_checks(struct file *file_in, loff_t pos_in,
3104 struct file *file_out, loff_t pos_out,
3105 loff_t *count, unsigned int remap_flags);
3106 extern int generic_file_rw_checks(struct file *file_in, struct file *file_out);
3107 extern int generic_copy_file_checks(struct file *file_in, loff_t pos_in,
3108 struct file *file_out, loff_t pos_out,
3109 size_t *count, unsigned int flags);
3110 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
3111 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
3112 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
3113 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
3114 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
3116 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
3118 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
3121 /* fs/block_dev.c */
3122 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
3123 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
3124 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
3126 extern void block_sync_page(struct page *page);
3129 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
3130 struct pipe_inode_info *, size_t, unsigned int);
3131 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
3132 struct file *, loff_t *, size_t, unsigned int);
3133 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
3134 struct file *out, loff_t *, size_t len, unsigned int flags);
3135 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
3136 loff_t *opos, size_t len, unsigned int flags);
3140 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3141 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3142 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
3143 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3144 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3145 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3146 int whence, loff_t maxsize, loff_t eof);
3147 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3148 int whence, loff_t size);
3149 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3150 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3151 extern int generic_file_open(struct inode * inode, struct file * filp);
3152 extern int nonseekable_open(struct inode * inode, struct file * filp);
3153 extern int stream_open(struct inode * inode, struct file * filp);
3156 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3157 loff_t file_offset);
3160 /* need locking between buffered and direct access */
3163 /* filesystem does not support filling holes */
3164 DIO_SKIP_HOLES = 0x02,
3167 void dio_end_io(struct bio *bio);
3168 void dio_warn_stale_pagecache(struct file *filp);
3170 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3171 struct block_device *bdev, struct iov_iter *iter,
3172 get_block_t get_block,
3173 dio_iodone_t end_io, dio_submit_t submit_io,
3176 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3177 struct inode *inode,
3178 struct iov_iter *iter,
3179 get_block_t get_block)
3181 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3182 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3186 void inode_dio_wait(struct inode *inode);
3189 * inode_dio_begin - signal start of a direct I/O requests
3190 * @inode: inode the direct I/O happens on
3192 * This is called once we've finished processing a direct I/O request,
3193 * and is used to wake up callers waiting for direct I/O to be quiesced.
3195 static inline void inode_dio_begin(struct inode *inode)
3197 atomic_inc(&inode->i_dio_count);
3201 * inode_dio_end - signal finish of a direct I/O requests
3202 * @inode: inode the direct I/O happens on
3204 * This is called once we've finished processing a direct I/O request,
3205 * and is used to wake up callers waiting for direct I/O to be quiesced.
3207 static inline void inode_dio_end(struct inode *inode)
3209 if (atomic_dec_and_test(&inode->i_dio_count))
3210 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3213 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3216 extern const struct file_operations generic_ro_fops;
3218 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3220 extern int readlink_copy(char __user *, int, const char *);
3221 extern int page_readlink(struct dentry *, char __user *, int);
3222 extern const char *page_get_link(struct dentry *, struct inode *,
3223 struct delayed_call *);
3224 extern void page_put_link(void *);
3225 extern int __page_symlink(struct inode *inode, const char *symname, int len,
3227 extern int page_symlink(struct inode *inode, const char *symname, int len);
3228 extern const struct inode_operations page_symlink_inode_operations;
3229 extern void kfree_link(void *);
3230 extern void generic_fillattr(struct inode *, struct kstat *);
3231 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3232 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3233 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3234 void inode_add_bytes(struct inode *inode, loff_t bytes);
3235 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3236 void inode_sub_bytes(struct inode *inode, loff_t bytes);
3237 static inline loff_t __inode_get_bytes(struct inode *inode)
3239 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3241 loff_t inode_get_bytes(struct inode *inode);
3242 void inode_set_bytes(struct inode *inode, loff_t bytes);
3243 const char *simple_get_link(struct dentry *, struct inode *,
3244 struct delayed_call *);
3245 extern const struct inode_operations simple_symlink_inode_operations;
3247 extern int iterate_dir(struct file *, struct dir_context *);
3249 extern int vfs_statx(int, const char __user *, int, struct kstat *, u32);
3250 extern int vfs_statx_fd(unsigned int, struct kstat *, u32, unsigned int);
3252 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3254 return vfs_statx(AT_FDCWD, filename, AT_NO_AUTOMOUNT,
3255 stat, STATX_BASIC_STATS);
3257 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3259 return vfs_statx(AT_FDCWD, name, AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT,
3260 stat, STATX_BASIC_STATS);
3262 static inline int vfs_fstatat(int dfd, const char __user *filename,
3263 struct kstat *stat, int flags)
3265 return vfs_statx(dfd, filename, flags | AT_NO_AUTOMOUNT,
3266 stat, STATX_BASIC_STATS);
3268 static inline int vfs_fstat(int fd, struct kstat *stat)
3270 return vfs_statx_fd(fd, stat, STATX_BASIC_STATS, 0);
3274 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3275 extern int vfs_readlink(struct dentry *, char __user *, int);
3277 extern int __generic_block_fiemap(struct inode *inode,
3278 struct fiemap_extent_info *fieinfo,
3279 loff_t start, loff_t len,
3280 get_block_t *get_block);
3281 extern int generic_block_fiemap(struct inode *inode,
3282 struct fiemap_extent_info *fieinfo, u64 start,
3283 u64 len, get_block_t *get_block);
3285 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3286 extern void put_filesystem(struct file_system_type *fs);
3287 extern struct file_system_type *get_fs_type(const char *name);
3288 extern struct super_block *get_super(struct block_device *);
3289 extern struct super_block *get_super_thawed(struct block_device *);
3290 extern struct super_block *get_super_exclusive_thawed(struct block_device *bdev);
3291 extern struct super_block *get_active_super(struct block_device *bdev);
3292 extern void drop_super(struct super_block *sb);
3293 extern void drop_super_exclusive(struct super_block *sb);
3294 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3295 extern void iterate_supers_type(struct file_system_type *,
3296 void (*)(struct super_block *, void *), void *);
3298 extern int dcache_dir_open(struct inode *, struct file *);
3299 extern int dcache_dir_close(struct inode *, struct file *);
3300 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3301 extern int dcache_readdir(struct file *, struct dir_context *);
3302 extern int simple_setattr(struct dentry *, struct iattr *);
3303 extern int simple_getattr(const struct path *, struct kstat *, u32, unsigned int);
3304 extern int simple_statfs(struct dentry *, struct kstatfs *);
3305 extern int simple_open(struct inode *inode, struct file *file);
3306 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3307 extern int simple_unlink(struct inode *, struct dentry *);
3308 extern int simple_rmdir(struct inode *, struct dentry *);
3309 extern int simple_rename(struct inode *, struct dentry *,
3310 struct inode *, struct dentry *, unsigned int);
3311 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3312 extern int noop_set_page_dirty(struct page *page);
3313 extern void noop_invalidatepage(struct page *page, unsigned int offset,
3314 unsigned int length);
3315 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3316 extern int simple_empty(struct dentry *);
3317 extern int simple_readpage(struct file *file, struct page *page);
3318 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3319 loff_t pos, unsigned len, unsigned flags,
3320 struct page **pagep, void **fsdata);
3321 extern int simple_write_end(struct file *file, struct address_space *mapping,
3322 loff_t pos, unsigned len, unsigned copied,
3323 struct page *page, void *fsdata);
3324 extern int always_delete_dentry(const struct dentry *);
3325 extern struct inode *alloc_anon_inode(struct super_block *);
3326 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3327 extern const struct dentry_operations simple_dentry_operations;
3329 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3330 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3331 extern const struct file_operations simple_dir_operations;
3332 extern const struct inode_operations simple_dir_inode_operations;
3333 extern void make_empty_dir_inode(struct inode *inode);
3334 extern bool is_empty_dir_inode(struct inode *inode);
3335 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3336 struct dentry *d_alloc_name(struct dentry *, const char *);
3337 extern int simple_fill_super(struct super_block *, unsigned long,
3338 const struct tree_descr *);
3339 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3340 extern void simple_release_fs(struct vfsmount **mount, int *count);
3342 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3343 loff_t *ppos, const void *from, size_t available);
3344 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3345 const void __user *from, size_t count);
3347 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3348 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3350 extern int generic_check_addressable(unsigned, u64);
3352 #ifdef CONFIG_MIGRATION
3353 extern int buffer_migrate_page(struct address_space *,
3354 struct page *, struct page *,
3356 extern int buffer_migrate_page_norefs(struct address_space *,
3357 struct page *, struct page *,
3360 #define buffer_migrate_page NULL
3361 #define buffer_migrate_page_norefs NULL
3364 extern int setattr_prepare(struct dentry *, struct iattr *);
3365 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3366 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
3368 extern int file_update_time(struct file *file);
3370 static inline bool io_is_direct(struct file *filp)
3372 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
3375 static inline bool vma_is_dax(struct vm_area_struct *vma)
3377 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3380 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3382 struct inode *inode;
3386 if (!vma_is_dax(vma))
3388 inode = file_inode(vma->vm_file);
3389 if (S_ISCHR(inode->i_mode))
3390 return false; /* device-dax */
3394 static inline int iocb_flags(struct file *file)
3397 if (file->f_flags & O_APPEND)
3399 if (io_is_direct(file))
3401 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3403 if (file->f_flags & __O_SYNC)
3408 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3410 if (unlikely(flags & ~RWF_SUPPORTED))
3413 if (flags & RWF_NOWAIT) {
3414 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3416 ki->ki_flags |= IOCB_NOWAIT;
3418 if (flags & RWF_HIPRI)
3419 ki->ki_flags |= IOCB_HIPRI;
3420 if (flags & RWF_DSYNC)
3421 ki->ki_flags |= IOCB_DSYNC;
3422 if (flags & RWF_SYNC)
3423 ki->ki_flags |= (IOCB_DSYNC | IOCB_SYNC);
3424 if (flags & RWF_APPEND)
3425 ki->ki_flags |= IOCB_APPEND;
3429 static inline ino_t parent_ino(struct dentry *dentry)
3434 * Don't strictly need d_lock here? If the parent ino could change
3435 * then surely we'd have a deeper race in the caller?
3437 spin_lock(&dentry->d_lock);
3438 res = dentry->d_parent->d_inode->i_ino;
3439 spin_unlock(&dentry->d_lock);
3443 /* Transaction based IO helpers */
3446 * An argresp is stored in an allocated page and holds the
3447 * size of the argument or response, along with its content
3449 struct simple_transaction_argresp {
3454 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3456 char *simple_transaction_get(struct file *file, const char __user *buf,
3458 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3459 size_t size, loff_t *pos);
3460 int simple_transaction_release(struct inode *inode, struct file *file);
3462 void simple_transaction_set(struct file *file, size_t n);
3465 * simple attribute files
3467 * These attributes behave similar to those in sysfs:
3469 * Writing to an attribute immediately sets a value, an open file can be
3470 * written to multiple times.
3472 * Reading from an attribute creates a buffer from the value that might get
3473 * read with multiple read calls. When the attribute has been read
3474 * completely, no further read calls are possible until the file is opened
3477 * All attributes contain a text representation of a numeric value
3478 * that are accessed with the get() and set() functions.
3480 #define DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, __is_signed) \
3481 static int __fops ## _open(struct inode *inode, struct file *file) \
3483 __simple_attr_check_format(__fmt, 0ull); \
3484 return simple_attr_open(inode, file, __get, __set, __fmt); \
3486 static const struct file_operations __fops = { \
3487 .owner = THIS_MODULE, \
3488 .open = __fops ## _open, \
3489 .release = simple_attr_release, \
3490 .read = simple_attr_read, \
3491 .write = (__is_signed) ? simple_attr_write_signed : simple_attr_write, \
3492 .llseek = generic_file_llseek, \
3495 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3496 DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, false)
3498 #define DEFINE_SIMPLE_ATTRIBUTE_SIGNED(__fops, __get, __set, __fmt) \
3499 DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, true)
3501 static inline __printf(1, 2)
3502 void __simple_attr_check_format(const char *fmt, ...)
3504 /* don't do anything, just let the compiler check the arguments; */
3507 int simple_attr_open(struct inode *inode, struct file *file,
3508 int (*get)(void *, u64 *), int (*set)(void *, u64),
3510 int simple_attr_release(struct inode *inode, struct file *file);
3511 ssize_t simple_attr_read(struct file *file, char __user *buf,
3512 size_t len, loff_t *ppos);
3513 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3514 size_t len, loff_t *ppos);
3515 ssize_t simple_attr_write_signed(struct file *file, const char __user *buf,
3516 size_t len, loff_t *ppos);
3519 int proc_nr_files(struct ctl_table *table, int write,
3520 void __user *buffer, size_t *lenp, loff_t *ppos);
3521 int proc_nr_dentry(struct ctl_table *table, int write,
3522 void __user *buffer, size_t *lenp, loff_t *ppos);
3523 int proc_nr_inodes(struct ctl_table *table, int write,
3524 void __user *buffer, size_t *lenp, loff_t *ppos);
3525 int __init get_filesystem_list(char *buf);
3527 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3528 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3530 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3531 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3532 (flag & __FMODE_NONOTIFY)))
3534 static inline bool is_sxid(umode_t mode)
3536 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3539 static inline int check_sticky(struct inode *dir, struct inode *inode)
3541 if (!(dir->i_mode & S_ISVTX))
3544 return __check_sticky(dir, inode);
3547 static inline void inode_has_no_xattr(struct inode *inode)
3549 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3550 inode->i_flags |= S_NOSEC;
3553 static inline bool is_root_inode(struct inode *inode)
3555 return inode == inode->i_sb->s_root->d_inode;
3558 static inline bool dir_emit(struct dir_context *ctx,
3559 const char *name, int namelen,
3560 u64 ino, unsigned type)
3562 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3564 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3566 return ctx->actor(ctx, ".", 1, ctx->pos,
3567 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3569 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3571 return ctx->actor(ctx, "..", 2, ctx->pos,
3572 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3574 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3576 if (ctx->pos == 0) {
3577 if (!dir_emit_dot(file, ctx))
3581 if (ctx->pos == 1) {
3582 if (!dir_emit_dotdot(file, ctx))
3588 static inline bool dir_relax(struct inode *inode)
3590 inode_unlock(inode);
3592 return !IS_DEADDIR(inode);
3595 static inline bool dir_relax_shared(struct inode *inode)
3597 inode_unlock_shared(inode);
3598 inode_lock_shared(inode);
3599 return !IS_DEADDIR(inode);
3602 extern bool path_noexec(const struct path *path);
3603 extern void inode_nohighmem(struct inode *inode);
3606 extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len,
3608 extern int generic_fadvise(struct file *file, loff_t offset, loff_t len,
3611 #if defined(CONFIG_IO_URING)
3612 extern struct sock *io_uring_get_socket(struct file *file);
3614 static inline struct sock *io_uring_get_socket(struct file *file)
3620 int vfs_ioc_setflags_prepare(struct inode *inode, unsigned int oldflags,
3621 unsigned int flags);
3623 int vfs_ioc_fssetxattr_check(struct inode *inode, const struct fsxattr *old_fa,
3624 struct fsxattr *fa);
3626 static inline void simple_fill_fsxattr(struct fsxattr *fa, __u32 xflags)
3628 memset(fa, 0, sizeof(*fa));
3629 fa->fsx_xflags = xflags;
3633 * Flush file data before changing attributes. Caller must hold any locks
3634 * required to prevent further writes to this file until we're done setting
3637 static inline int inode_drain_writes(struct inode *inode)
3639 inode_dio_wait(inode);
3640 return filemap_write_and_wait(inode->i_mapping);
3643 #endif /* _LINUX_FS_H */