1 /* SPDX-License-Identifier: GPL-2.0 */
5 #include <linux/linkage.h>
6 #include <linux/wait_bit.h>
7 #include <linux/kdev_t.h>
8 #include <linux/dcache.h>
9 #include <linux/path.h>
10 #include <linux/stat.h>
11 #include <linux/cache.h>
12 #include <linux/list.h>
13 #include <linux/list_lru.h>
14 #include <linux/llist.h>
15 #include <linux/radix-tree.h>
16 #include <linux/xarray.h>
17 #include <linux/rbtree.h>
18 #include <linux/init.h>
19 #include <linux/pid.h>
20 #include <linux/bug.h>
21 #include <linux/mutex.h>
22 #include <linux/rwsem.h>
23 #include <linux/mm_types.h>
24 #include <linux/capability.h>
25 #include <linux/semaphore.h>
26 #include <linux/fcntl.h>
27 #include <linux/rculist_bl.h>
28 #include <linux/atomic.h>
29 #include <linux/shrinker.h>
30 #include <linux/migrate_mode.h>
31 #include <linux/uidgid.h>
32 #include <linux/lockdep.h>
33 #include <linux/percpu-rwsem.h>
34 #include <linux/workqueue.h>
35 #include <linux/delayed_call.h>
36 #include <linux/uuid.h>
37 #include <linux/errseq.h>
38 #include <linux/ioprio.h>
39 #include <linux/fs_types.h>
40 #include <linux/build_bug.h>
41 #include <linux/stddef.h>
42 #include <linux/mount.h>
43 #include <linux/cred.h>
44 #include <linux/mnt_idmapping.h>
45 #include <linux/slab.h>
46 #include <linux/maple_tree.h>
47 #include <linux/rw_hint.h>
49 #include <asm/byteorder.h>
50 #include <uapi/linux/fs.h>
52 struct backing_dev_info;
56 struct export_operations;
57 struct fiemap_extent_info;
62 struct pipe_inode_info;
63 struct poll_table_struct;
65 struct vm_area_struct;
68 struct swap_info_struct;
70 struct workqueue_struct;
72 struct fscrypt_inode_info;
73 struct fscrypt_operations;
75 struct fsverity_operations;
77 struct fs_parameter_spec;
81 extern void __init inode_init(void);
82 extern void __init inode_init_early(void);
83 extern void __init files_init(void);
84 extern void __init files_maxfiles_init(void);
86 extern unsigned long get_max_files(void);
87 extern unsigned int sysctl_nr_open;
89 typedef __kernel_rwf_t rwf_t;
92 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
93 struct buffer_head *bh_result, int create);
94 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
95 ssize_t bytes, void *private);
97 #define MAY_EXEC 0x00000001
98 #define MAY_WRITE 0x00000002
99 #define MAY_READ 0x00000004
100 #define MAY_APPEND 0x00000008
101 #define MAY_ACCESS 0x00000010
102 #define MAY_OPEN 0x00000020
103 #define MAY_CHDIR 0x00000040
104 /* called from RCU mode, don't block */
105 #define MAY_NOT_BLOCK 0x00000080
108 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
109 * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
112 /* file is open for reading */
113 #define FMODE_READ ((__force fmode_t)(1 << 0))
114 /* file is open for writing */
115 #define FMODE_WRITE ((__force fmode_t)(1 << 1))
116 /* file is seekable */
117 #define FMODE_LSEEK ((__force fmode_t)(1 << 2))
118 /* file can be accessed using pread */
119 #define FMODE_PREAD ((__force fmode_t)(1 << 3))
120 /* file can be accessed using pwrite */
121 #define FMODE_PWRITE ((__force fmode_t)(1 << 4))
122 /* File is opened for execution with sys_execve / sys_uselib */
123 #define FMODE_EXEC ((__force fmode_t)(1 << 5))
124 /* File writes are restricted (block device specific) */
125 #define FMODE_WRITE_RESTRICTED ((__force fmode_t)(1 << 6))
127 /* FMODE_* bits 7 to 8 */
129 /* 32bit hashes as llseek() offset (for directories) */
130 #define FMODE_32BITHASH ((__force fmode_t)(1 << 9))
131 /* 64bit hashes as llseek() offset (for directories) */
132 #define FMODE_64BITHASH ((__force fmode_t)(1 << 10))
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)(1 << 11))
142 /* Expect random access pattern */
143 #define FMODE_RANDOM ((__force fmode_t)(1 << 12))
145 /* File is huge (eg. /dev/mem): treat loff_t as unsigned */
146 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)(1 << 13))
148 /* File is opened with O_PATH; almost nothing can be done with it */
149 #define FMODE_PATH ((__force fmode_t)(1 << 14))
151 /* File needs atomic accesses to f_pos */
152 #define FMODE_ATOMIC_POS ((__force fmode_t)(1 << 15))
153 /* Write access to underlying fs */
154 #define FMODE_WRITER ((__force fmode_t)(1 << 16))
155 /* Has read method(s) */
156 #define FMODE_CAN_READ ((__force fmode_t)(1 << 17))
157 /* Has write method(s) */
158 #define FMODE_CAN_WRITE ((__force fmode_t)(1 << 18))
160 #define FMODE_OPENED ((__force fmode_t)(1 << 19))
161 #define FMODE_CREATED ((__force fmode_t)(1 << 20))
163 /* File is stream-like */
164 #define FMODE_STREAM ((__force fmode_t)(1 << 21))
166 /* File supports DIRECT IO */
167 #define FMODE_CAN_ODIRECT ((__force fmode_t)(1 << 22))
169 #define FMODE_NOREUSE ((__force fmode_t)(1 << 23))
173 /* File is embedded in backing_file object */
174 #define FMODE_BACKING ((__force fmode_t)(1 << 25))
176 /* File was opened by fanotify and shouldn't generate fanotify events */
177 #define FMODE_NONOTIFY ((__force fmode_t)(1 << 26))
179 /* File is capable of returning -EAGAIN if I/O will block */
180 #define FMODE_NOWAIT ((__force fmode_t)(1 << 27))
182 /* File represents mount that needs unmounting */
183 #define FMODE_NEED_UNMOUNT ((__force fmode_t)(1 << 28))
185 /* File does not contribute to nr_files count */
186 #define FMODE_NOACCOUNT ((__force fmode_t)(1 << 29))
189 * Attribute flags. These should be or-ed together to figure out what
192 #define ATTR_MODE (1 << 0)
193 #define ATTR_UID (1 << 1)
194 #define ATTR_GID (1 << 2)
195 #define ATTR_SIZE (1 << 3)
196 #define ATTR_ATIME (1 << 4)
197 #define ATTR_MTIME (1 << 5)
198 #define ATTR_CTIME (1 << 6)
199 #define ATTR_ATIME_SET (1 << 7)
200 #define ATTR_MTIME_SET (1 << 8)
201 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
202 #define ATTR_KILL_SUID (1 << 11)
203 #define ATTR_KILL_SGID (1 << 12)
204 #define ATTR_FILE (1 << 13)
205 #define ATTR_KILL_PRIV (1 << 14)
206 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
207 #define ATTR_TIMES_SET (1 << 16)
208 #define ATTR_TOUCH (1 << 17)
211 * Whiteout is represented by a char device. The following constants define the
212 * mode and device number to use.
214 #define WHITEOUT_MODE 0
215 #define WHITEOUT_DEV 0
218 * This is the Inode Attributes structure, used for notify_change(). It
219 * uses the above definitions as flags, to know which values have changed.
220 * Also, in this manner, a Filesystem can look at only the values it cares
221 * about. Basically, these are the attributes that the VFS layer can
222 * request to change from the FS layer.
224 * Derek Atkins <warlord@MIT.EDU> 94-10-20
227 unsigned int ia_valid;
230 * The two anonymous unions wrap structures with the same member.
232 * Filesystems raising FS_ALLOW_IDMAP need to use ia_vfs{g,u}id which
233 * are a dedicated type requiring the filesystem to use the dedicated
234 * helpers. Other filesystem can continue to use ia_{g,u}id until they
237 * They always contain the same value. In other words FS_ALLOW_IDMAP
238 * pass down the same value on idmapped mounts as they would on regular
250 struct timespec64 ia_atime;
251 struct timespec64 ia_mtime;
252 struct timespec64 ia_ctime;
255 * Not an attribute, but an auxiliary info for filesystems wanting to
256 * implement an ftruncate() like method. NOTE: filesystem should
257 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
259 struct file *ia_file;
263 * Includes for diskquotas.
265 #include <linux/quota.h>
268 * Maximum number of layers of fs stack. Needs to be limited to
269 * prevent kernel stack overflow
271 #define FILESYSTEM_MAX_STACK_DEPTH 2
274 * enum positive_aop_returns - aop return codes with specific semantics
276 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
277 * completed, that the page is still locked, and
278 * should be considered active. The VM uses this hint
279 * to return the page to the active list -- it won't
280 * be a candidate for writeback again in the near
281 * future. Other callers must be careful to unlock
282 * the page if they get this return. Returned by
285 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
286 * unlocked it and the page might have been truncated.
287 * The caller should back up to acquiring a new page and
288 * trying again. The aop will be taking reasonable
289 * precautions not to livelock. If the caller held a page
290 * reference, it should drop it before retrying. Returned
293 * address_space_operation functions return these large constants to indicate
294 * special semantics to the caller. These are much larger than the bytes in a
295 * page to allow for functions that return the number of bytes operated on in a
299 enum positive_aop_returns {
300 AOP_WRITEPAGE_ACTIVATE = 0x80000,
301 AOP_TRUNCATED_PAGE = 0x80001,
305 * oh the beauties of C type declarations.
308 struct address_space;
309 struct writeback_control;
310 struct readahead_control;
312 /* Match RWF_* bits to IOCB bits */
313 #define IOCB_HIPRI (__force int) RWF_HIPRI
314 #define IOCB_DSYNC (__force int) RWF_DSYNC
315 #define IOCB_SYNC (__force int) RWF_SYNC
316 #define IOCB_NOWAIT (__force int) RWF_NOWAIT
317 #define IOCB_APPEND (__force int) RWF_APPEND
319 /* non-RWF related bits - start at 16 */
320 #define IOCB_EVENTFD (1 << 16)
321 #define IOCB_DIRECT (1 << 17)
322 #define IOCB_WRITE (1 << 18)
323 /* iocb->ki_waitq is valid */
324 #define IOCB_WAITQ (1 << 19)
325 #define IOCB_NOIO (1 << 20)
326 /* can use bio alloc cache */
327 #define IOCB_ALLOC_CACHE (1 << 21)
329 * IOCB_DIO_CALLER_COMP can be set by the iocb owner, to indicate that the
330 * iocb completion can be passed back to the owner for execution from a safe
331 * context rather than needing to be punted through a workqueue. If this
332 * flag is set, the bio completion handling may set iocb->dio_complete to a
333 * handler function and iocb->private to context information for that handler.
334 * The issuer should call the handler with that context information from task
335 * context to complete the processing of the iocb. Note that while this
336 * provides a task context for the dio_complete() callback, it should only be
337 * used on the completion side for non-IO generating completions. It's fine to
338 * call blocking functions from this callback, but they should not wait for
339 * unrelated IO (like cache flushing, new IO generation, etc).
341 #define IOCB_DIO_CALLER_COMP (1 << 22)
342 /* kiocb is a read or write operation submitted by fs/aio.c. */
343 #define IOCB_AIO_RW (1 << 23)
345 /* for use in trace events */
346 #define TRACE_IOCB_STRINGS \
347 { IOCB_HIPRI, "HIPRI" }, \
348 { IOCB_DSYNC, "DSYNC" }, \
349 { IOCB_SYNC, "SYNC" }, \
350 { IOCB_NOWAIT, "NOWAIT" }, \
351 { IOCB_APPEND, "APPEND" }, \
352 { IOCB_EVENTFD, "EVENTFD"}, \
353 { IOCB_DIRECT, "DIRECT" }, \
354 { IOCB_WRITE, "WRITE" }, \
355 { IOCB_WAITQ, "WAITQ" }, \
356 { IOCB_NOIO, "NOIO" }, \
357 { IOCB_ALLOC_CACHE, "ALLOC_CACHE" }, \
358 { IOCB_DIO_CALLER_COMP, "CALLER_COMP" }
361 struct file *ki_filp;
363 void (*ki_complete)(struct kiocb *iocb, long ret);
366 u16 ki_ioprio; /* See linux/ioprio.h */
369 * Only used for async buffered reads, where it denotes the
370 * page waitqueue associated with completing the read. Valid
371 * IFF IOCB_WAITQ is set.
373 struct wait_page_queue *ki_waitq;
375 * Can be used for O_DIRECT IO, where the completion handling
376 * is punted back to the issuer of the IO. May only be set
377 * if IOCB_DIO_CALLER_COMP is set by the issuer, and the issuer
378 * must then check for presence of this handler when ki_complete
379 * is invoked. The data passed in to this handler must be
380 * assigned to ->private when dio_complete is assigned.
382 ssize_t (*dio_complete)(void *data);
386 static inline bool is_sync_kiocb(struct kiocb *kiocb)
388 return kiocb->ki_complete == NULL;
391 struct address_space_operations {
392 int (*writepage)(struct page *page, struct writeback_control *wbc);
393 int (*read_folio)(struct file *, struct folio *);
395 /* Write back some dirty pages from this mapping. */
396 int (*writepages)(struct address_space *, struct writeback_control *);
398 /* Mark a folio dirty. Return true if this dirtied it */
399 bool (*dirty_folio)(struct address_space *, struct folio *);
401 void (*readahead)(struct readahead_control *);
403 int (*write_begin)(struct file *, struct address_space *mapping,
404 loff_t pos, unsigned len,
405 struct page **pagep, void **fsdata);
406 int (*write_end)(struct file *, struct address_space *mapping,
407 loff_t pos, unsigned len, unsigned copied,
408 struct page *page, void *fsdata);
410 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
411 sector_t (*bmap)(struct address_space *, sector_t);
412 void (*invalidate_folio) (struct folio *, size_t offset, size_t len);
413 bool (*release_folio)(struct folio *, gfp_t);
414 void (*free_folio)(struct folio *folio);
415 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
417 * migrate the contents of a folio to the specified target. If
418 * migrate_mode is MIGRATE_ASYNC, it must not block.
420 int (*migrate_folio)(struct address_space *, struct folio *dst,
421 struct folio *src, enum migrate_mode);
422 int (*launder_folio)(struct folio *);
423 bool (*is_partially_uptodate) (struct folio *, size_t from,
425 void (*is_dirty_writeback) (struct folio *, bool *dirty, bool *wb);
426 int (*error_remove_folio)(struct address_space *, struct folio *);
428 /* swapfile support */
429 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
431 void (*swap_deactivate)(struct file *file);
432 int (*swap_rw)(struct kiocb *iocb, struct iov_iter *iter);
435 extern const struct address_space_operations empty_aops;
438 * struct address_space - Contents of a cacheable, mappable object.
439 * @host: Owner, either the inode or the block_device.
440 * @i_pages: Cached pages.
441 * @invalidate_lock: Guards coherency between page cache contents and
442 * file offset->disk block mappings in the filesystem during invalidates.
443 * It is also used to block modification of page cache contents through
445 * @gfp_mask: Memory allocation flags to use for allocating pages.
446 * @i_mmap_writable: Number of VM_SHARED, VM_MAYWRITE mappings.
447 * @nr_thps: Number of THPs in the pagecache (non-shmem only).
448 * @i_mmap: Tree of private and shared mappings.
449 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
450 * @nrpages: Number of page entries, protected by the i_pages lock.
451 * @writeback_index: Writeback starts here.
453 * @flags: Error bits and flags (AS_*).
454 * @wb_err: The most recent error which has occurred.
455 * @i_private_lock: For use by the owner of the address_space.
456 * @i_private_list: For use by the owner of the address_space.
457 * @i_private_data: For use by the owner of the address_space.
459 struct address_space {
461 struct xarray i_pages;
462 struct rw_semaphore invalidate_lock;
464 atomic_t i_mmap_writable;
465 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
466 /* number of thp, only for non-shmem files */
469 struct rb_root_cached i_mmap;
470 unsigned long nrpages;
471 pgoff_t writeback_index;
472 const struct address_space_operations *a_ops;
475 spinlock_t i_private_lock;
476 struct list_head i_private_list;
477 struct rw_semaphore i_mmap_rwsem;
478 void * i_private_data;
479 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
481 * On most architectures that alignment is already the case; but
482 * must be enforced here for CRIS, to let the least significant bit
483 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
486 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
487 #define PAGECACHE_TAG_DIRTY XA_MARK_0
488 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
489 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
492 * Returns true if any of the pages in the mapping are marked with the tag.
494 static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag)
496 return xa_marked(&mapping->i_pages, tag);
499 static inline void i_mmap_lock_write(struct address_space *mapping)
501 down_write(&mapping->i_mmap_rwsem);
504 static inline int i_mmap_trylock_write(struct address_space *mapping)
506 return down_write_trylock(&mapping->i_mmap_rwsem);
509 static inline void i_mmap_unlock_write(struct address_space *mapping)
511 up_write(&mapping->i_mmap_rwsem);
514 static inline int i_mmap_trylock_read(struct address_space *mapping)
516 return down_read_trylock(&mapping->i_mmap_rwsem);
519 static inline void i_mmap_lock_read(struct address_space *mapping)
521 down_read(&mapping->i_mmap_rwsem);
524 static inline void i_mmap_unlock_read(struct address_space *mapping)
526 up_read(&mapping->i_mmap_rwsem);
529 static inline void i_mmap_assert_locked(struct address_space *mapping)
531 lockdep_assert_held(&mapping->i_mmap_rwsem);
534 static inline void i_mmap_assert_write_locked(struct address_space *mapping)
536 lockdep_assert_held_write(&mapping->i_mmap_rwsem);
540 * Might pages of this file be mapped into userspace?
542 static inline int mapping_mapped(struct address_space *mapping)
544 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
548 * Might pages of this file have been modified in userspace?
549 * Note that i_mmap_writable counts all VM_SHARED, VM_MAYWRITE vmas: do_mmap
550 * marks vma as VM_SHARED if it is shared, and the file was opened for
551 * writing i.e. vma may be mprotected writable even if now readonly.
553 * If i_mmap_writable is negative, no new writable mappings are allowed. You
554 * can only deny writable mappings, if none exists right now.
556 static inline int mapping_writably_mapped(struct address_space *mapping)
558 return atomic_read(&mapping->i_mmap_writable) > 0;
561 static inline int mapping_map_writable(struct address_space *mapping)
563 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
567 static inline void mapping_unmap_writable(struct address_space *mapping)
569 atomic_dec(&mapping->i_mmap_writable);
572 static inline int mapping_deny_writable(struct address_space *mapping)
574 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
578 static inline void mapping_allow_writable(struct address_space *mapping)
580 atomic_inc(&mapping->i_mmap_writable);
584 * Use sequence counter to get consistent i_size on 32-bit processors.
586 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
587 #include <linux/seqlock.h>
588 #define __NEED_I_SIZE_ORDERED
589 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
591 #define i_size_ordered_init(inode) do { } while (0)
595 #define ACL_NOT_CACHED ((void *)(-1))
597 * ACL_DONT_CACHE is for stacked filesystems, that rely on underlying fs to
598 * cache the ACL. This also means that ->get_inode_acl() can be called in RCU
599 * mode with the LOOKUP_RCU flag.
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; /* use inode_*_ctime accessors! */
666 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
667 unsigned short i_bytes;
669 enum rw_hint i_write_hint;
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;
724 #ifdef CONFIG_FSNOTIFY
725 __u32 i_fsnotify_mask; /* all events this inode cares about */
726 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
729 #ifdef CONFIG_FS_ENCRYPTION
730 struct fscrypt_inode_info *i_crypt_info;
733 #ifdef CONFIG_FS_VERITY
734 struct fsverity_info *i_verity_info;
737 void *i_private; /* fs or device private pointer */
738 } __randomize_layout;
740 struct timespec64 timestamp_truncate(struct timespec64 t, struct inode *inode);
742 static inline unsigned int i_blocksize(const struct inode *node)
744 return (1 << node->i_blkbits);
747 static inline int inode_unhashed(struct inode *inode)
749 return hlist_unhashed(&inode->i_hash);
753 * __mark_inode_dirty expects inodes to be hashed. Since we don't
754 * want special inodes in the fileset inode space, we make them
755 * appear hashed, but do not put on any lists. hlist_del()
756 * will work fine and require no locking.
758 static inline void inode_fake_hash(struct inode *inode)
760 hlist_add_fake(&inode->i_hash);
764 * inode->i_mutex nesting subclasses for the lock validator:
766 * 0: the object of the current VFS operation
770 * 4: second non-directory
771 * 5: second parent (when locking independent directories in rename)
773 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
774 * non-directories at once.
776 * The locking order between these classes is
777 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
779 enum inode_i_mutex_lock_class
789 static inline void inode_lock(struct inode *inode)
791 down_write(&inode->i_rwsem);
794 static inline void inode_unlock(struct inode *inode)
796 up_write(&inode->i_rwsem);
799 static inline void inode_lock_shared(struct inode *inode)
801 down_read(&inode->i_rwsem);
804 static inline void inode_unlock_shared(struct inode *inode)
806 up_read(&inode->i_rwsem);
809 static inline int inode_trylock(struct inode *inode)
811 return down_write_trylock(&inode->i_rwsem);
814 static inline int inode_trylock_shared(struct inode *inode)
816 return down_read_trylock(&inode->i_rwsem);
819 static inline int inode_is_locked(struct inode *inode)
821 return rwsem_is_locked(&inode->i_rwsem);
824 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
826 down_write_nested(&inode->i_rwsem, subclass);
829 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
831 down_read_nested(&inode->i_rwsem, subclass);
834 static inline void filemap_invalidate_lock(struct address_space *mapping)
836 down_write(&mapping->invalidate_lock);
839 static inline void filemap_invalidate_unlock(struct address_space *mapping)
841 up_write(&mapping->invalidate_lock);
844 static inline void filemap_invalidate_lock_shared(struct address_space *mapping)
846 down_read(&mapping->invalidate_lock);
849 static inline int filemap_invalidate_trylock_shared(
850 struct address_space *mapping)
852 return down_read_trylock(&mapping->invalidate_lock);
855 static inline void filemap_invalidate_unlock_shared(
856 struct address_space *mapping)
858 up_read(&mapping->invalidate_lock);
861 void lock_two_nondirectories(struct inode *, struct inode*);
862 void unlock_two_nondirectories(struct inode *, struct inode*);
864 void filemap_invalidate_lock_two(struct address_space *mapping1,
865 struct address_space *mapping2);
866 void filemap_invalidate_unlock_two(struct address_space *mapping1,
867 struct address_space *mapping2);
871 * NOTE: in a 32bit arch with a preemptable kernel and
872 * an UP compile the i_size_read/write must be atomic
873 * with respect to the local cpu (unlike with preempt disabled),
874 * but they don't need to be atomic with respect to other cpus like in
875 * true SMP (so they need either to either locally disable irq around
876 * the read or for example on x86 they can be still implemented as a
877 * cmpxchg8b without the need of the lock prefix). For SMP compiles
878 * and 64bit archs it makes no difference if preempt is enabled or not.
880 static inline loff_t i_size_read(const struct inode *inode)
882 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
887 seq = read_seqcount_begin(&inode->i_size_seqcount);
888 i_size = inode->i_size;
889 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
891 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
895 i_size = inode->i_size;
899 /* Pairs with smp_store_release() in i_size_write() */
900 return smp_load_acquire(&inode->i_size);
905 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
906 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
907 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
909 static inline void i_size_write(struct inode *inode, loff_t i_size)
911 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
913 write_seqcount_begin(&inode->i_size_seqcount);
914 inode->i_size = i_size;
915 write_seqcount_end(&inode->i_size_seqcount);
917 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
919 inode->i_size = i_size;
923 * Pairs with smp_load_acquire() in i_size_read() to ensure
924 * changes related to inode size (such as page contents) are
925 * visible before we see the changed inode size.
927 smp_store_release(&inode->i_size, i_size);
931 static inline unsigned iminor(const struct inode *inode)
933 return MINOR(inode->i_rdev);
936 static inline unsigned imajor(const struct inode *inode)
938 return MAJOR(inode->i_rdev);
942 rwlock_t lock; /* protects pid, uid, euid fields */
943 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
944 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
945 kuid_t uid, euid; /* uid/euid of process setting the owner */
946 int signum; /* posix.1b rt signal to be delivered on IO */
950 * struct file_ra_state - Track a file's readahead state.
951 * @start: Where the most recent readahead started.
952 * @size: Number of pages read in the most recent readahead.
953 * @async_size: Numer of pages that were/are not needed immediately
954 * and so were/are genuinely "ahead". Start next readahead when
955 * the first of these pages is accessed.
956 * @ra_pages: Maximum size of a readahead request, copied from the bdi.
957 * @mmap_miss: How many mmap accesses missed in the page cache.
958 * @prev_pos: The last byte in the most recent read request.
960 * When this structure is passed to ->readahead(), the "most recent"
961 * readahead means the current readahead.
963 struct file_ra_state {
966 unsigned int async_size;
967 unsigned int ra_pages;
968 unsigned int mmap_miss;
973 * Check if @index falls in the readahead windows.
975 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
977 return (index >= ra->start &&
978 index < ra->start + ra->size);
982 * f_{lock,count,pos_lock} members can be highly contended and share
983 * the same cacheline. f_{lock,mode} are very frequently used together
984 * and so share the same cacheline as well. The read-mostly
985 * f_{path,inode,op} are kept on a separate cacheline.
989 /* fput() uses task work when closing and freeing file (default). */
990 struct callback_head f_task_work;
991 /* fput() must use workqueue (most kernel threads). */
992 struct llist_node f_llist;
993 unsigned int f_iocb_flags;
997 * Protects f_ep, f_flags.
998 * Must not be taken from IRQ context.
1002 atomic_long_t f_count;
1003 struct mutex f_pos_lock;
1005 unsigned int f_flags;
1006 struct fown_struct f_owner;
1007 const struct cred *f_cred;
1008 struct file_ra_state f_ra;
1010 struct inode *f_inode; /* cached value */
1011 const struct file_operations *f_op;
1014 #ifdef CONFIG_SECURITY
1017 /* needed for tty driver, and maybe others */
1021 /* Used by fs/eventpoll.c to link all the hooks to this file */
1022 struct hlist_head *f_ep;
1023 #endif /* #ifdef CONFIG_EPOLL */
1024 struct address_space *f_mapping;
1026 errseq_t f_sb_err; /* for syncfs */
1027 } __randomize_layout
1028 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
1030 struct file_handle {
1033 /* file identifier */
1034 unsigned char f_handle[] __counted_by(handle_bytes);
1037 static inline struct file *get_file(struct file *f)
1039 long prior = atomic_long_fetch_inc_relaxed(&f->f_count);
1040 WARN_ONCE(!prior, "struct file::f_count incremented from zero; use-after-free condition present!\n");
1044 struct file *get_file_rcu(struct file __rcu **f);
1045 struct file *get_file_active(struct file **f);
1047 #define file_count(x) atomic_long_read(&(x)->f_count)
1049 #define MAX_NON_LFS ((1UL<<31) - 1)
1051 /* Page cache limit. The filesystems should put that into their s_maxbytes
1052 limits, otherwise bad things can happen in VM. */
1053 #if BITS_PER_LONG==32
1054 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
1055 #elif BITS_PER_LONG==64
1056 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
1059 /* legacy typedef, should eventually be removed */
1060 typedef void *fl_owner_t;
1065 /* The following constant reflects the upper bound of the file/locking space */
1067 #define OFFSET_MAX type_max(loff_t)
1068 #define OFFT_OFFSET_MAX type_max(off_t)
1071 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1073 static inline struct inode *file_inode(const struct file *f)
1079 * file_dentry() is a relic from the days that overlayfs was using files with a
1080 * "fake" path, meaning, f_path on overlayfs and f_inode on underlying fs.
1081 * In those days, file_dentry() was needed to get the underlying fs dentry that
1083 * Files with "fake" path should not exist nowadays, so use an assertion to make
1084 * sure that file_dentry() was not papering over filesystem bugs.
1086 static inline struct dentry *file_dentry(const struct file *file)
1088 struct dentry *dentry = file->f_path.dentry;
1090 WARN_ON_ONCE(d_inode(dentry) != file_inode(file));
1094 struct fasync_struct {
1098 struct fasync_struct *fa_next; /* singly linked list */
1099 struct file *fa_file;
1100 struct rcu_head fa_rcu;
1103 #define FASYNC_MAGIC 0x4601
1105 /* SMP safe fasync helpers: */
1106 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1107 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1108 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1109 extern struct fasync_struct *fasync_alloc(void);
1110 extern void fasync_free(struct fasync_struct *);
1112 /* can be called from interrupts */
1113 extern void kill_fasync(struct fasync_struct **, int, int);
1115 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1116 extern int f_setown(struct file *filp, int who, int force);
1117 extern void f_delown(struct file *filp);
1118 extern pid_t f_getown(struct file *filp);
1119 extern int send_sigurg(struct fown_struct *fown);
1122 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1123 * represented in both.
1125 #define SB_RDONLY BIT(0) /* Mount read-only */
1126 #define SB_NOSUID BIT(1) /* Ignore suid and sgid bits */
1127 #define SB_NODEV BIT(2) /* Disallow access to device special files */
1128 #define SB_NOEXEC BIT(3) /* Disallow program execution */
1129 #define SB_SYNCHRONOUS BIT(4) /* Writes are synced at once */
1130 #define SB_MANDLOCK BIT(6) /* Allow mandatory locks on an FS */
1131 #define SB_DIRSYNC BIT(7) /* Directory modifications are synchronous */
1132 #define SB_NOATIME BIT(10) /* Do not update access times. */
1133 #define SB_NODIRATIME BIT(11) /* Do not update directory access times */
1134 #define SB_SILENT BIT(15)
1135 #define SB_POSIXACL BIT(16) /* Supports POSIX ACLs */
1136 #define SB_INLINECRYPT BIT(17) /* Use blk-crypto for encrypted files */
1137 #define SB_KERNMOUNT BIT(22) /* this is a kern_mount call */
1138 #define SB_I_VERSION BIT(23) /* Update inode I_version field */
1139 #define SB_LAZYTIME BIT(25) /* Update the on-disk [acm]times lazily */
1141 /* These sb flags are internal to the kernel */
1142 #define SB_DEAD BIT(21)
1143 #define SB_DYING BIT(24)
1144 #define SB_SUBMOUNT BIT(26)
1145 #define SB_FORCE BIT(27)
1146 #define SB_NOSEC BIT(28)
1147 #define SB_BORN BIT(29)
1148 #define SB_ACTIVE BIT(30)
1149 #define SB_NOUSER BIT(31)
1151 /* These flags relate to encoding and casefolding */
1152 #define SB_ENC_STRICT_MODE_FL (1 << 0)
1154 #define sb_has_strict_encoding(sb) \
1155 (sb->s_encoding_flags & SB_ENC_STRICT_MODE_FL)
1161 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1162 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1163 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1164 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1165 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1168 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1169 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1170 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1171 #define SB_I_STABLE_WRITES 0x00000008 /* don't modify blks until WB is done */
1173 /* sb->s_iflags to limit user namespace mounts */
1174 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1175 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1176 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1177 #define SB_I_EVM_HMAC_UNSUPPORTED 0x00000080
1179 #define SB_I_SKIP_SYNC 0x00000100 /* Skip superblock at global sync */
1180 #define SB_I_PERSB_BDI 0x00000200 /* has a per-sb bdi */
1181 #define SB_I_TS_EXPIRY_WARNED 0x00000400 /* warned about timestamp range expiry */
1182 #define SB_I_RETIRED 0x00000800 /* superblock shouldn't be reused */
1183 #define SB_I_NOUMASK 0x00001000 /* VFS does not apply umask */
1185 /* Possible states of 'frozen' field */
1187 SB_UNFROZEN = 0, /* FS is unfrozen */
1188 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1189 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1190 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1191 * internal threads if needed) */
1192 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1195 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1198 unsigned short frozen; /* Is sb frozen? */
1199 int freeze_kcount; /* How many kernel freeze requests? */
1200 int freeze_ucount; /* How many userspace freeze requests? */
1201 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1204 struct super_block {
1205 struct list_head s_list; /* Keep this first */
1206 dev_t s_dev; /* search index; _not_ kdev_t */
1207 unsigned char s_blocksize_bits;
1208 unsigned long s_blocksize;
1209 loff_t s_maxbytes; /* Max file size */
1210 struct file_system_type *s_type;
1211 const struct super_operations *s_op;
1212 const struct dquot_operations *dq_op;
1213 const struct quotactl_ops *s_qcop;
1214 const struct export_operations *s_export_op;
1215 unsigned long s_flags;
1216 unsigned long s_iflags; /* internal SB_I_* flags */
1217 unsigned long s_magic;
1218 struct dentry *s_root;
1219 struct rw_semaphore s_umount;
1222 #ifdef CONFIG_SECURITY
1225 const struct xattr_handler * const *s_xattr;
1226 #ifdef CONFIG_FS_ENCRYPTION
1227 const struct fscrypt_operations *s_cop;
1228 struct fscrypt_keyring *s_master_keys; /* master crypto keys in use */
1230 #ifdef CONFIG_FS_VERITY
1231 const struct fsverity_operations *s_vop;
1233 #if IS_ENABLED(CONFIG_UNICODE)
1234 struct unicode_map *s_encoding;
1235 __u16 s_encoding_flags;
1237 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1238 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1239 struct block_device *s_bdev; /* can go away once we use an accessor for @s_bdev_file */
1240 struct file *s_bdev_file;
1241 struct backing_dev_info *s_bdi;
1242 struct mtd_info *s_mtd;
1243 struct hlist_node s_instances;
1244 unsigned int s_quota_types; /* Bitmask of supported quota types */
1245 struct quota_info s_dquot; /* Diskquota specific options */
1247 struct sb_writers s_writers;
1250 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1251 * s_fsnotify_marks together for cache efficiency. They are frequently
1252 * accessed and rarely modified.
1254 void *s_fs_info; /* Filesystem private info */
1256 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1258 /* Time limits for c/m/atime in seconds */
1259 time64_t s_time_min;
1260 time64_t s_time_max;
1261 #ifdef CONFIG_FSNOTIFY
1262 __u32 s_fsnotify_mask;
1263 struct fsnotify_mark_connector __rcu *s_fsnotify_marks;
1267 * q: why are s_id and s_sysfs_name not the same? both are human
1268 * readable strings that identify the filesystem
1269 * a: s_id is allowed to change at runtime; it's used in log messages,
1270 * and we want to when a device starts out as single device (s_id is dev
1271 * name) but then a device is hot added and we have to switch to
1272 * identifying it by UUID
1273 * but s_sysfs_name is a handle for programmatic access, and can't
1276 char s_id[32]; /* Informational name */
1277 uuid_t s_uuid; /* UUID */
1278 u8 s_uuid_len; /* Default 16, possibly smaller for weird filesystems */
1280 /* if set, fs shows up under sysfs at /sys/fs/$FSTYP/s_sysfs_name */
1281 char s_sysfs_name[UUID_STRING_LEN + 1];
1283 unsigned int s_max_links;
1286 * The next field is for VFS *only*. No filesystems have any business
1287 * even looking at it. You had been warned.
1289 struct mutex s_vfs_rename_mutex; /* Kludge */
1292 * Filesystem subtype. If non-empty the filesystem type field
1293 * in /proc/mounts will be "type.subtype"
1295 const char *s_subtype;
1297 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1299 struct shrinker *s_shrink; /* per-sb shrinker handle */
1301 /* Number of inodes with nlink == 0 but still referenced */
1302 atomic_long_t s_remove_count;
1305 * Number of inode/mount/sb objects that are being watched, note that
1306 * inodes objects are currently double-accounted.
1308 atomic_long_t s_fsnotify_connectors;
1310 /* Read-only state of the superblock is being changed */
1311 int s_readonly_remount;
1313 /* per-sb errseq_t for reporting writeback errors via syncfs */
1316 /* AIO completions deferred from interrupt context */
1317 struct workqueue_struct *s_dio_done_wq;
1318 struct hlist_head s_pins;
1321 * Owning user namespace and default context in which to
1322 * interpret filesystem uids, gids, quotas, device nodes,
1323 * xattrs and security labels.
1325 struct user_namespace *s_user_ns;
1328 * The list_lru structure is essentially just a pointer to a table
1329 * of per-node lru lists, each of which has its own spinlock.
1330 * There is no need to put them into separate cachelines.
1332 struct list_lru s_dentry_lru;
1333 struct list_lru s_inode_lru;
1334 struct rcu_head rcu;
1335 struct work_struct destroy_work;
1337 struct mutex s_sync_lock; /* sync serialisation lock */
1340 * Indicates how deep in a filesystem stack this SB is
1344 /* s_inode_list_lock protects s_inodes */
1345 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1346 struct list_head s_inodes; /* all inodes */
1348 spinlock_t s_inode_wblist_lock;
1349 struct list_head s_inodes_wb; /* writeback inodes */
1350 } __randomize_layout;
1352 static inline struct user_namespace *i_user_ns(const struct inode *inode)
1354 return inode->i_sb->s_user_ns;
1357 /* Helper functions so that in most cases filesystems will
1358 * not need to deal directly with kuid_t and kgid_t and can
1359 * instead deal with the raw numeric values that are stored
1360 * in the filesystem.
1362 static inline uid_t i_uid_read(const struct inode *inode)
1364 return from_kuid(i_user_ns(inode), inode->i_uid);
1367 static inline gid_t i_gid_read(const struct inode *inode)
1369 return from_kgid(i_user_ns(inode), inode->i_gid);
1372 static inline void i_uid_write(struct inode *inode, uid_t uid)
1374 inode->i_uid = make_kuid(i_user_ns(inode), uid);
1377 static inline void i_gid_write(struct inode *inode, gid_t gid)
1379 inode->i_gid = make_kgid(i_user_ns(inode), gid);
1383 * i_uid_into_vfsuid - map an inode's i_uid down according to an idmapping
1384 * @idmap: idmap of the mount the inode was found from
1385 * @inode: inode to map
1387 * Return: whe inode's i_uid mapped down according to @idmap.
1388 * If the inode's i_uid has no mapping INVALID_VFSUID is returned.
1390 static inline vfsuid_t i_uid_into_vfsuid(struct mnt_idmap *idmap,
1391 const struct inode *inode)
1393 return make_vfsuid(idmap, i_user_ns(inode), inode->i_uid);
1397 * i_uid_needs_update - check whether inode's i_uid needs to be updated
1398 * @idmap: idmap of the mount the inode was found from
1399 * @attr: the new attributes of @inode
1400 * @inode: the inode to update
1402 * Check whether the $inode's i_uid field needs to be updated taking idmapped
1403 * mounts into account if the filesystem supports it.
1405 * Return: true if @inode's i_uid field needs to be updated, false if not.
1407 static inline bool i_uid_needs_update(struct mnt_idmap *idmap,
1408 const struct iattr *attr,
1409 const struct inode *inode)
1411 return ((attr->ia_valid & ATTR_UID) &&
1412 !vfsuid_eq(attr->ia_vfsuid,
1413 i_uid_into_vfsuid(idmap, inode)));
1417 * i_uid_update - update @inode's i_uid field
1418 * @idmap: idmap of the mount the inode was found from
1419 * @attr: the new attributes of @inode
1420 * @inode: the inode to update
1422 * Safely update @inode's i_uid field translating the vfsuid of any idmapped
1423 * mount into the filesystem kuid.
1425 static inline void i_uid_update(struct mnt_idmap *idmap,
1426 const struct iattr *attr,
1427 struct inode *inode)
1429 if (attr->ia_valid & ATTR_UID)
1430 inode->i_uid = from_vfsuid(idmap, i_user_ns(inode),
1435 * i_gid_into_vfsgid - map an inode's i_gid down according to an idmapping
1436 * @idmap: idmap of the mount the inode was found from
1437 * @inode: inode to map
1439 * Return: the inode's i_gid mapped down according to @idmap.
1440 * If the inode's i_gid has no mapping INVALID_VFSGID is returned.
1442 static inline vfsgid_t i_gid_into_vfsgid(struct mnt_idmap *idmap,
1443 const struct inode *inode)
1445 return make_vfsgid(idmap, i_user_ns(inode), inode->i_gid);
1449 * i_gid_needs_update - check whether inode's i_gid needs to be updated
1450 * @idmap: idmap of the mount the inode was found from
1451 * @attr: the new attributes of @inode
1452 * @inode: the inode to update
1454 * Check whether the $inode's i_gid field needs to be updated taking idmapped
1455 * mounts into account if the filesystem supports it.
1457 * Return: true if @inode's i_gid field needs to be updated, false if not.
1459 static inline bool i_gid_needs_update(struct mnt_idmap *idmap,
1460 const struct iattr *attr,
1461 const struct inode *inode)
1463 return ((attr->ia_valid & ATTR_GID) &&
1464 !vfsgid_eq(attr->ia_vfsgid,
1465 i_gid_into_vfsgid(idmap, inode)));
1469 * i_gid_update - update @inode's i_gid field
1470 * @idmap: idmap of the mount the inode was found from
1471 * @attr: the new attributes of @inode
1472 * @inode: the inode to update
1474 * Safely update @inode's i_gid field translating the vfsgid of any idmapped
1475 * mount into the filesystem kgid.
1477 static inline void i_gid_update(struct mnt_idmap *idmap,
1478 const struct iattr *attr,
1479 struct inode *inode)
1481 if (attr->ia_valid & ATTR_GID)
1482 inode->i_gid = from_vfsgid(idmap, i_user_ns(inode),
1487 * inode_fsuid_set - initialize inode's i_uid field with callers fsuid
1488 * @inode: inode to initialize
1489 * @idmap: idmap of the mount the inode was found from
1491 * Initialize the i_uid field of @inode. If the inode was found/created via
1492 * an idmapped mount map the caller's fsuid according to @idmap.
1494 static inline void inode_fsuid_set(struct inode *inode,
1495 struct mnt_idmap *idmap)
1497 inode->i_uid = mapped_fsuid(idmap, i_user_ns(inode));
1501 * inode_fsgid_set - initialize inode's i_gid field with callers fsgid
1502 * @inode: inode to initialize
1503 * @idmap: idmap of the mount the inode was found from
1505 * Initialize the i_gid field of @inode. If the inode was found/created via
1506 * an idmapped mount map the caller's fsgid according to @idmap.
1508 static inline void inode_fsgid_set(struct inode *inode,
1509 struct mnt_idmap *idmap)
1511 inode->i_gid = mapped_fsgid(idmap, i_user_ns(inode));
1515 * fsuidgid_has_mapping() - check whether caller's fsuid/fsgid is mapped
1516 * @sb: the superblock we want a mapping in
1517 * @idmap: idmap of the relevant mount
1519 * Check whether the caller's fsuid and fsgid have a valid mapping in the
1520 * s_user_ns of the superblock @sb. If the caller is on an idmapped mount map
1521 * the caller's fsuid and fsgid according to the @idmap first.
1523 * Return: true if fsuid and fsgid is mapped, false if not.
1525 static inline bool fsuidgid_has_mapping(struct super_block *sb,
1526 struct mnt_idmap *idmap)
1528 struct user_namespace *fs_userns = sb->s_user_ns;
1532 kuid = mapped_fsuid(idmap, fs_userns);
1533 if (!uid_valid(kuid))
1535 kgid = mapped_fsgid(idmap, fs_userns);
1536 if (!gid_valid(kgid))
1538 return kuid_has_mapping(fs_userns, kuid) &&
1539 kgid_has_mapping(fs_userns, kgid);
1542 struct timespec64 current_time(struct inode *inode);
1543 struct timespec64 inode_set_ctime_current(struct inode *inode);
1545 static inline time64_t inode_get_atime_sec(const struct inode *inode)
1547 return inode->__i_atime.tv_sec;
1550 static inline long inode_get_atime_nsec(const struct inode *inode)
1552 return inode->__i_atime.tv_nsec;
1555 static inline struct timespec64 inode_get_atime(const struct inode *inode)
1557 return inode->__i_atime;
1560 static inline struct timespec64 inode_set_atime_to_ts(struct inode *inode,
1561 struct timespec64 ts)
1563 inode->__i_atime = ts;
1567 static inline struct timespec64 inode_set_atime(struct inode *inode,
1568 time64_t sec, long nsec)
1570 struct timespec64 ts = { .tv_sec = sec,
1572 return inode_set_atime_to_ts(inode, ts);
1575 static inline time64_t inode_get_mtime_sec(const struct inode *inode)
1577 return inode->__i_mtime.tv_sec;
1580 static inline long inode_get_mtime_nsec(const struct inode *inode)
1582 return inode->__i_mtime.tv_nsec;
1585 static inline struct timespec64 inode_get_mtime(const struct inode *inode)
1587 return inode->__i_mtime;
1590 static inline struct timespec64 inode_set_mtime_to_ts(struct inode *inode,
1591 struct timespec64 ts)
1593 inode->__i_mtime = ts;
1597 static inline struct timespec64 inode_set_mtime(struct inode *inode,
1598 time64_t sec, long nsec)
1600 struct timespec64 ts = { .tv_sec = sec,
1602 return inode_set_mtime_to_ts(inode, ts);
1605 static inline time64_t inode_get_ctime_sec(const struct inode *inode)
1607 return inode->__i_ctime.tv_sec;
1610 static inline long inode_get_ctime_nsec(const struct inode *inode)
1612 return inode->__i_ctime.tv_nsec;
1615 static inline struct timespec64 inode_get_ctime(const struct inode *inode)
1617 return inode->__i_ctime;
1620 static inline struct timespec64 inode_set_ctime_to_ts(struct inode *inode,
1621 struct timespec64 ts)
1623 inode->__i_ctime = ts;
1628 * inode_set_ctime - set the ctime in the inode
1629 * @inode: inode in which to set the ctime
1630 * @sec: tv_sec value to set
1631 * @nsec: tv_nsec value to set
1633 * Set the ctime in @inode to { @sec, @nsec }
1635 static inline struct timespec64 inode_set_ctime(struct inode *inode,
1636 time64_t sec, long nsec)
1638 struct timespec64 ts = { .tv_sec = sec,
1641 return inode_set_ctime_to_ts(inode, ts);
1644 struct timespec64 simple_inode_init_ts(struct inode *inode);
1647 * Snapshotting support.
1651 * These are internal functions, please use sb_start_{write,pagefault,intwrite}
1654 static inline void __sb_end_write(struct super_block *sb, int level)
1656 percpu_up_read(sb->s_writers.rw_sem + level-1);
1659 static inline void __sb_start_write(struct super_block *sb, int level)
1661 percpu_down_read(sb->s_writers.rw_sem + level - 1);
1664 static inline bool __sb_start_write_trylock(struct super_block *sb, int level)
1666 return percpu_down_read_trylock(sb->s_writers.rw_sem + level - 1);
1669 #define __sb_writers_acquired(sb, lev) \
1670 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1671 #define __sb_writers_release(sb, lev) \
1672 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1675 * __sb_write_started - check if sb freeze level is held
1676 * @sb: the super we write to
1677 * @level: the freeze level
1679 * * > 0 - sb freeze level is held
1680 * * 0 - sb freeze level is not held
1681 * * < 0 - !CONFIG_LOCKDEP/LOCK_STATE_UNKNOWN
1683 static inline int __sb_write_started(const struct super_block *sb, int level)
1685 return lockdep_is_held_type(sb->s_writers.rw_sem + level - 1, 1);
1689 * sb_write_started - check if SB_FREEZE_WRITE is held
1690 * @sb: the super we write to
1692 * May be false positive with !CONFIG_LOCKDEP/LOCK_STATE_UNKNOWN.
1694 static inline bool sb_write_started(const struct super_block *sb)
1696 return __sb_write_started(sb, SB_FREEZE_WRITE);
1700 * sb_write_not_started - check if SB_FREEZE_WRITE is not held
1701 * @sb: the super we write to
1703 * May be false positive with !CONFIG_LOCKDEP/LOCK_STATE_UNKNOWN.
1705 static inline bool sb_write_not_started(const struct super_block *sb)
1707 return __sb_write_started(sb, SB_FREEZE_WRITE) <= 0;
1711 * file_write_started - check if SB_FREEZE_WRITE is held
1712 * @file: the file we write to
1714 * May be false positive with !CONFIG_LOCKDEP/LOCK_STATE_UNKNOWN.
1715 * May be false positive with !S_ISREG, because file_start_write() has
1716 * no effect on !S_ISREG.
1718 static inline bool file_write_started(const struct file *file)
1720 if (!S_ISREG(file_inode(file)->i_mode))
1722 return sb_write_started(file_inode(file)->i_sb);
1726 * file_write_not_started - check if SB_FREEZE_WRITE is not held
1727 * @file: the file we write to
1729 * May be false positive with !CONFIG_LOCKDEP/LOCK_STATE_UNKNOWN.
1730 * May be false positive with !S_ISREG, because file_start_write() has
1731 * no effect on !S_ISREG.
1733 static inline bool file_write_not_started(const struct file *file)
1735 if (!S_ISREG(file_inode(file)->i_mode))
1737 return sb_write_not_started(file_inode(file)->i_sb);
1741 * sb_end_write - drop write access to a superblock
1742 * @sb: the super we wrote to
1744 * Decrement number of writers to the filesystem. Wake up possible waiters
1745 * wanting to freeze the filesystem.
1747 static inline void sb_end_write(struct super_block *sb)
1749 __sb_end_write(sb, SB_FREEZE_WRITE);
1753 * sb_end_pagefault - drop write access to a superblock from a page fault
1754 * @sb: the super we wrote to
1756 * Decrement number of processes handling write page fault to the filesystem.
1757 * Wake up possible waiters wanting to freeze the filesystem.
1759 static inline void sb_end_pagefault(struct super_block *sb)
1761 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1765 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1766 * @sb: the super we wrote to
1768 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1769 * waiters wanting to freeze the filesystem.
1771 static inline void sb_end_intwrite(struct super_block *sb)
1773 __sb_end_write(sb, SB_FREEZE_FS);
1777 * sb_start_write - get write access to a superblock
1778 * @sb: the super we write to
1780 * When a process wants to write data or metadata to a file system (i.e. dirty
1781 * a page or an inode), it should embed the operation in a sb_start_write() -
1782 * sb_end_write() pair to get exclusion against file system freezing. This
1783 * function increments number of writers preventing freezing. If the file
1784 * system is already frozen, the function waits until the file system is
1787 * Since freeze protection behaves as a lock, users have to preserve
1788 * ordering of freeze protection and other filesystem locks. Generally,
1789 * freeze protection should be the outermost lock. In particular, we have:
1792 * -> i_mutex (write path, truncate, directory ops, ...)
1793 * -> s_umount (freeze_super, thaw_super)
1795 static inline void sb_start_write(struct super_block *sb)
1797 __sb_start_write(sb, SB_FREEZE_WRITE);
1800 static inline bool sb_start_write_trylock(struct super_block *sb)
1802 return __sb_start_write_trylock(sb, SB_FREEZE_WRITE);
1806 * sb_start_pagefault - get write access to a superblock from a page fault
1807 * @sb: the super we write to
1809 * When a process starts handling write page fault, it should embed the
1810 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1811 * exclusion against file system freezing. This is needed since the page fault
1812 * is going to dirty a page. This function increments number of running page
1813 * faults preventing freezing. If the file system is already frozen, the
1814 * function waits until the file system is thawed.
1816 * Since page fault freeze protection behaves as a lock, users have to preserve
1817 * ordering of freeze protection and other filesystem locks. It is advised to
1818 * put sb_start_pagefault() close to mmap_lock in lock ordering. Page fault
1819 * handling code implies lock dependency:
1822 * -> sb_start_pagefault
1824 static inline void sb_start_pagefault(struct super_block *sb)
1826 __sb_start_write(sb, SB_FREEZE_PAGEFAULT);
1830 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1831 * @sb: the super we write to
1833 * This is the third level of protection against filesystem freezing. It is
1834 * free for use by a filesystem. The only requirement is that it must rank
1835 * below sb_start_pagefault.
1837 * For example filesystem can call sb_start_intwrite() when starting a
1838 * transaction which somewhat eases handling of freezing for internal sources
1839 * of filesystem changes (internal fs threads, discarding preallocation on file
1842 static inline void sb_start_intwrite(struct super_block *sb)
1844 __sb_start_write(sb, SB_FREEZE_FS);
1847 static inline bool sb_start_intwrite_trylock(struct super_block *sb)
1849 return __sb_start_write_trylock(sb, SB_FREEZE_FS);
1852 bool inode_owner_or_capable(struct mnt_idmap *idmap,
1853 const struct inode *inode);
1856 * VFS helper functions..
1858 int vfs_create(struct mnt_idmap *, struct inode *,
1859 struct dentry *, umode_t, bool);
1860 int vfs_mkdir(struct mnt_idmap *, struct inode *,
1861 struct dentry *, umode_t);
1862 int vfs_mknod(struct mnt_idmap *, struct inode *, struct dentry *,
1864 int vfs_symlink(struct mnt_idmap *, struct inode *,
1865 struct dentry *, const char *);
1866 int vfs_link(struct dentry *, struct mnt_idmap *, struct inode *,
1867 struct dentry *, struct inode **);
1868 int vfs_rmdir(struct mnt_idmap *, struct inode *, struct dentry *);
1869 int vfs_unlink(struct mnt_idmap *, struct inode *, struct dentry *,
1873 * struct renamedata - contains all information required for renaming
1874 * @old_mnt_idmap: idmap of the old mount the inode was found from
1875 * @old_dir: parent of source
1876 * @old_dentry: source
1877 * @new_mnt_idmap: idmap of the new mount the inode was found from
1878 * @new_dir: parent of destination
1879 * @new_dentry: destination
1880 * @delegated_inode: returns an inode needing a delegation break
1881 * @flags: rename flags
1884 struct mnt_idmap *old_mnt_idmap;
1885 struct inode *old_dir;
1886 struct dentry *old_dentry;
1887 struct mnt_idmap *new_mnt_idmap;
1888 struct inode *new_dir;
1889 struct dentry *new_dentry;
1890 struct inode **delegated_inode;
1892 } __randomize_layout;
1894 int vfs_rename(struct renamedata *);
1896 static inline int vfs_whiteout(struct mnt_idmap *idmap,
1897 struct inode *dir, struct dentry *dentry)
1899 return vfs_mknod(idmap, dir, dentry, S_IFCHR | WHITEOUT_MODE,
1903 struct file *kernel_tmpfile_open(struct mnt_idmap *idmap,
1904 const struct path *parentpath,
1905 umode_t mode, int open_flag,
1906 const struct cred *cred);
1907 struct file *kernel_file_open(const struct path *path, int flags,
1908 struct inode *inode, const struct cred *cred);
1910 int vfs_mkobj(struct dentry *, umode_t,
1911 int (*f)(struct dentry *, umode_t, void *),
1914 int vfs_fchown(struct file *file, uid_t user, gid_t group);
1915 int vfs_fchmod(struct file *file, umode_t mode);
1916 int vfs_utimes(const struct path *path, struct timespec64 *times);
1918 extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1920 #ifdef CONFIG_COMPAT
1921 extern long compat_ptr_ioctl(struct file *file, unsigned int cmd,
1924 #define compat_ptr_ioctl NULL
1928 * VFS file helper functions.
1930 void inode_init_owner(struct mnt_idmap *idmap, struct inode *inode,
1931 const struct inode *dir, umode_t mode);
1932 extern bool may_open_dev(const struct path *path);
1933 umode_t mode_strip_sgid(struct mnt_idmap *idmap,
1934 const struct inode *dir, umode_t mode);
1937 * This is the "filldir" function type, used by readdir() to let
1938 * the kernel specify what kind of dirent layout it wants to have.
1939 * This allows the kernel to read directories into kernel space or
1940 * to have different dirent layouts depending on the binary type.
1941 * Return 'true' to keep going and 'false' if there are no more entries.
1944 typedef bool (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1947 struct dir_context {
1953 * These flags let !MMU mmap() govern direct device mapping vs immediate
1954 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1956 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1957 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1958 * NOMMU_MAP_READ: Can be mapped for reading
1959 * NOMMU_MAP_WRITE: Can be mapped for writing
1960 * NOMMU_MAP_EXEC: Can be mapped for execution
1962 #define NOMMU_MAP_COPY 0x00000001
1963 #define NOMMU_MAP_DIRECT 0x00000008
1964 #define NOMMU_MAP_READ VM_MAYREAD
1965 #define NOMMU_MAP_WRITE VM_MAYWRITE
1966 #define NOMMU_MAP_EXEC VM_MAYEXEC
1968 #define NOMMU_VMFLAGS \
1969 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1972 * These flags control the behavior of the remap_file_range function pointer.
1973 * If it is called with len == 0 that means "remap to end of source file".
1974 * See Documentation/filesystems/vfs.rst for more details about this call.
1976 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
1977 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
1979 #define REMAP_FILE_DEDUP (1 << 0)
1980 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
1983 * These flags signal that the caller is ok with altering various aspects of
1984 * the behavior of the remap operation. The changes must be made by the
1985 * implementation; the vfs remap helper functions can take advantage of them.
1986 * Flags in this category exist to preserve the quirky behavior of the hoisted
1987 * btrfs clone/dedupe ioctls.
1989 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
1992 * These flags control the behavior of vfs_copy_file_range().
1993 * They are not available to the user via syscall.
1995 * COPY_FILE_SPLICE: call splice direct instead of fs clone/copy ops
1997 #define COPY_FILE_SPLICE (1 << 0)
2000 struct io_uring_cmd;
2003 typedef unsigned int __bitwise fop_flags_t;
2005 struct file_operations {
2006 struct module *owner;
2007 fop_flags_t fop_flags;
2008 loff_t (*llseek) (struct file *, loff_t, int);
2009 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
2010 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
2011 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
2012 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
2013 int (*iopoll)(struct kiocb *kiocb, struct io_comp_batch *,
2014 unsigned int flags);
2015 int (*iterate_shared) (struct file *, struct dir_context *);
2016 __poll_t (*poll) (struct file *, struct poll_table_struct *);
2017 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
2018 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
2019 int (*mmap) (struct file *, struct vm_area_struct *);
2020 int (*open) (struct inode *, struct file *);
2021 int (*flush) (struct file *, fl_owner_t id);
2022 int (*release) (struct inode *, struct file *);
2023 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
2024 int (*fasync) (int, struct file *, int);
2025 int (*lock) (struct file *, int, struct file_lock *);
2026 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
2027 int (*check_flags)(int);
2028 int (*flock) (struct file *, int, struct file_lock *);
2029 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
2030 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
2031 void (*splice_eof)(struct file *file);
2032 int (*setlease)(struct file *, int, struct file_lease **, void **);
2033 long (*fallocate)(struct file *file, int mode, loff_t offset,
2035 void (*show_fdinfo)(struct seq_file *m, struct file *f);
2037 unsigned (*mmap_capabilities)(struct file *);
2039 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
2040 loff_t, size_t, unsigned int);
2041 loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
2042 struct file *file_out, loff_t pos_out,
2043 loff_t len, unsigned int remap_flags);
2044 int (*fadvise)(struct file *, loff_t, loff_t, int);
2045 int (*uring_cmd)(struct io_uring_cmd *ioucmd, unsigned int issue_flags);
2046 int (*uring_cmd_iopoll)(struct io_uring_cmd *, struct io_comp_batch *,
2047 unsigned int poll_flags);
2048 } __randomize_layout;
2050 /* Supports async buffered reads */
2051 #define FOP_BUFFER_RASYNC ((__force fop_flags_t)(1 << 0))
2052 /* Supports async buffered writes */
2053 #define FOP_BUFFER_WASYNC ((__force fop_flags_t)(1 << 1))
2054 /* Supports synchronous page faults for mappings */
2055 #define FOP_MMAP_SYNC ((__force fop_flags_t)(1 << 2))
2056 /* Supports non-exclusive O_DIRECT writes from multiple threads */
2057 #define FOP_DIO_PARALLEL_WRITE ((__force fop_flags_t)(1 << 3))
2058 /* Contains huge pages */
2059 #define FOP_HUGE_PAGES ((__force fop_flags_t)(1 << 4))
2061 /* Wrap a directory iterator that needs exclusive inode access */
2062 int wrap_directory_iterator(struct file *, struct dir_context *,
2063 int (*) (struct file *, struct dir_context *));
2064 #define WRAP_DIR_ITER(x) \
2065 static int shared_##x(struct file *file , struct dir_context *ctx) \
2066 { return wrap_directory_iterator(file, ctx, x); }
2068 struct inode_operations {
2069 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
2070 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
2071 int (*permission) (struct mnt_idmap *, struct inode *, int);
2072 struct posix_acl * (*get_inode_acl)(struct inode *, int, bool);
2074 int (*readlink) (struct dentry *, char __user *,int);
2076 int (*create) (struct mnt_idmap *, struct inode *,struct dentry *,
2078 int (*link) (struct dentry *,struct inode *,struct dentry *);
2079 int (*unlink) (struct inode *,struct dentry *);
2080 int (*symlink) (struct mnt_idmap *, struct inode *,struct dentry *,
2082 int (*mkdir) (struct mnt_idmap *, struct inode *,struct dentry *,
2084 int (*rmdir) (struct inode *,struct dentry *);
2085 int (*mknod) (struct mnt_idmap *, struct inode *,struct dentry *,
2087 int (*rename) (struct mnt_idmap *, struct inode *, struct dentry *,
2088 struct inode *, struct dentry *, unsigned int);
2089 int (*setattr) (struct mnt_idmap *, struct dentry *, struct iattr *);
2090 int (*getattr) (struct mnt_idmap *, const struct path *,
2091 struct kstat *, u32, unsigned int);
2092 ssize_t (*listxattr) (struct dentry *, char *, size_t);
2093 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
2095 int (*update_time)(struct inode *, int);
2096 int (*atomic_open)(struct inode *, struct dentry *,
2097 struct file *, unsigned open_flag,
2098 umode_t create_mode);
2099 int (*tmpfile) (struct mnt_idmap *, struct inode *,
2100 struct file *, umode_t);
2101 struct posix_acl *(*get_acl)(struct mnt_idmap *, struct dentry *,
2103 int (*set_acl)(struct mnt_idmap *, struct dentry *,
2104 struct posix_acl *, int);
2105 int (*fileattr_set)(struct mnt_idmap *idmap,
2106 struct dentry *dentry, struct fileattr *fa);
2107 int (*fileattr_get)(struct dentry *dentry, struct fileattr *fa);
2108 struct offset_ctx *(*get_offset_ctx)(struct inode *inode);
2109 } ____cacheline_aligned;
2111 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
2112 struct iov_iter *iter)
2114 return file->f_op->read_iter(kio, iter);
2117 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
2118 struct iov_iter *iter)
2120 return file->f_op->write_iter(kio, iter);
2123 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
2125 return file->f_op->mmap(file, vma);
2128 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
2129 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
2130 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
2131 loff_t, size_t, unsigned int);
2132 int __generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
2133 struct file *file_out, loff_t pos_out,
2134 loff_t *len, unsigned int remap_flags,
2135 const struct iomap_ops *dax_read_ops);
2136 int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
2137 struct file *file_out, loff_t pos_out,
2138 loff_t *count, unsigned int remap_flags);
2139 extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in,
2140 struct file *file_out, loff_t pos_out,
2141 loff_t len, unsigned int remap_flags);
2142 extern int vfs_dedupe_file_range(struct file *file,
2143 struct file_dedupe_range *same);
2144 extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
2145 struct file *dst_file, loff_t dst_pos,
2146 loff_t len, unsigned int remap_flags);
2149 * enum freeze_holder - holder of the freeze
2150 * @FREEZE_HOLDER_KERNEL: kernel wants to freeze or thaw filesystem
2151 * @FREEZE_HOLDER_USERSPACE: userspace wants to freeze or thaw filesystem
2152 * @FREEZE_MAY_NEST: whether nesting freeze and thaw requests is allowed
2154 * Indicate who the owner of the freeze or thaw request is and whether
2155 * the freeze needs to be exclusive or can nest.
2156 * Without @FREEZE_MAY_NEST, multiple freeze and thaw requests from the
2157 * same holder aren't allowed. It is however allowed to hold a single
2158 * @FREEZE_HOLDER_USERSPACE and a single @FREEZE_HOLDER_KERNEL freeze at
2159 * the same time. This is relied upon by some filesystems during online
2160 * repair or similar.
2162 enum freeze_holder {
2163 FREEZE_HOLDER_KERNEL = (1U << 0),
2164 FREEZE_HOLDER_USERSPACE = (1U << 1),
2165 FREEZE_MAY_NEST = (1U << 2),
2168 struct super_operations {
2169 struct inode *(*alloc_inode)(struct super_block *sb);
2170 void (*destroy_inode)(struct inode *);
2171 void (*free_inode)(struct inode *);
2173 void (*dirty_inode) (struct inode *, int flags);
2174 int (*write_inode) (struct inode *, struct writeback_control *wbc);
2175 int (*drop_inode) (struct inode *);
2176 void (*evict_inode) (struct inode *);
2177 void (*put_super) (struct super_block *);
2178 int (*sync_fs)(struct super_block *sb, int wait);
2179 int (*freeze_super) (struct super_block *, enum freeze_holder who);
2180 int (*freeze_fs) (struct super_block *);
2181 int (*thaw_super) (struct super_block *, enum freeze_holder who);
2182 int (*unfreeze_fs) (struct super_block *);
2183 int (*statfs) (struct dentry *, struct kstatfs *);
2184 int (*remount_fs) (struct super_block *, int *, char *);
2185 void (*umount_begin) (struct super_block *);
2187 int (*show_options)(struct seq_file *, struct dentry *);
2188 int (*show_devname)(struct seq_file *, struct dentry *);
2189 int (*show_path)(struct seq_file *, struct dentry *);
2190 int (*show_stats)(struct seq_file *, struct dentry *);
2192 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
2193 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
2194 struct dquot __rcu **(*get_dquots)(struct inode *);
2196 long (*nr_cached_objects)(struct super_block *,
2197 struct shrink_control *);
2198 long (*free_cached_objects)(struct super_block *,
2199 struct shrink_control *);
2200 void (*shutdown)(struct super_block *sb);
2204 * Inode flags - they have no relation to superblock flags now
2206 #define S_SYNC (1 << 0) /* Writes are synced at once */
2207 #define S_NOATIME (1 << 1) /* Do not update access times */
2208 #define S_APPEND (1 << 2) /* Append-only file */
2209 #define S_IMMUTABLE (1 << 3) /* Immutable file */
2210 #define S_DEAD (1 << 4) /* removed, but still open directory */
2211 #define S_NOQUOTA (1 << 5) /* Inode is not counted to quota */
2212 #define S_DIRSYNC (1 << 6) /* Directory modifications are synchronous */
2213 #define S_NOCMTIME (1 << 7) /* Do not update file c/mtime */
2214 #define S_SWAPFILE (1 << 8) /* Do not truncate: swapon got its bmaps */
2215 #define S_PRIVATE (1 << 9) /* Inode is fs-internal */
2216 #define S_IMA (1 << 10) /* Inode has an associated IMA struct */
2217 #define S_AUTOMOUNT (1 << 11) /* Automount/referral quasi-directory */
2218 #define S_NOSEC (1 << 12) /* no suid or xattr security attributes */
2219 #ifdef CONFIG_FS_DAX
2220 #define S_DAX (1 << 13) /* Direct Access, avoiding the page cache */
2222 #define S_DAX 0 /* Make all the DAX code disappear */
2224 #define S_ENCRYPTED (1 << 14) /* Encrypted file (using fs/crypto/) */
2225 #define S_CASEFOLD (1 << 15) /* Casefolded file */
2226 #define S_VERITY (1 << 16) /* Verity file (using fs/verity/) */
2227 #define S_KERNEL_FILE (1 << 17) /* File is in use by the kernel (eg. fs/cachefiles) */
2230 * Note that nosuid etc flags are inode-specific: setting some file-system
2231 * flags just means all the inodes inherit those flags by default. It might be
2232 * possible to override it selectively if you really wanted to with some
2233 * ioctl() that is not currently implemented.
2235 * Exception: SB_RDONLY is always applied to the entire file system.
2237 * Unfortunately, it is possible to change a filesystems flags with it mounted
2238 * with files in use. This means that all of the inodes will not have their
2239 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
2240 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
2242 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
2244 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
2245 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
2246 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
2247 ((inode)->i_flags & S_SYNC))
2248 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
2249 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
2250 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
2251 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
2252 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
2254 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
2255 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
2256 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
2258 #ifdef CONFIG_FS_POSIX_ACL
2259 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
2261 #define IS_POSIXACL(inode) 0
2264 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
2265 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
2268 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
2270 #define IS_SWAPFILE(inode) ((void)(inode), 0U)
2273 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
2274 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
2275 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2276 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2277 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2278 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2279 #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD)
2280 #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY)
2282 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2283 (inode)->i_rdev == WHITEOUT_DEV)
2285 static inline bool HAS_UNMAPPED_ID(struct mnt_idmap *idmap,
2286 struct inode *inode)
2288 return !vfsuid_valid(i_uid_into_vfsuid(idmap, inode)) ||
2289 !vfsgid_valid(i_gid_into_vfsgid(idmap, inode));
2292 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
2294 *kiocb = (struct kiocb) {
2296 .ki_flags = filp->f_iocb_flags,
2297 .ki_ioprio = get_current_ioprio(),
2301 static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src,
2304 *kiocb = (struct kiocb) {
2306 .ki_flags = kiocb_src->ki_flags,
2307 .ki_ioprio = kiocb_src->ki_ioprio,
2308 .ki_pos = kiocb_src->ki_pos,
2313 * Inode state bits. Protected by inode->i_lock
2315 * Four bits determine the dirty state of the inode: I_DIRTY_SYNC,
2316 * I_DIRTY_DATASYNC, I_DIRTY_PAGES, and I_DIRTY_TIME.
2318 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2319 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2320 * various stages of removing an inode.
2322 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2324 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2325 * fdatasync() (unless I_DIRTY_DATASYNC is also set).
2326 * Timestamp updates are the usual cause.
2327 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2328 * these changes separately from I_DIRTY_SYNC so that we
2329 * don't have to write inode on fdatasync() when only
2330 * e.g. the timestamps have changed.
2331 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2332 * I_DIRTY_TIME The inode itself has dirty timestamps, and the
2333 * lazytime mount option is enabled. We keep track of this
2334 * separately from I_DIRTY_SYNC in order to implement
2335 * lazytime. This gets cleared if I_DIRTY_INODE
2336 * (I_DIRTY_SYNC and/or I_DIRTY_DATASYNC) gets set. But
2337 * I_DIRTY_TIME can still be set if I_DIRTY_SYNC is already
2338 * in place because writeback might already be in progress
2339 * and we don't want to lose the time update
2340 * I_NEW Serves as both a mutex and completion notification.
2341 * New inodes set I_NEW. If two processes both create
2342 * the same inode, one of them will release its inode and
2343 * wait for I_NEW to be released before returning.
2344 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2345 * also cause waiting on I_NEW, without I_NEW actually
2346 * being set. find_inode() uses this to prevent returning
2347 * nearly-dead inodes.
2348 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2349 * is zero. I_FREEING must be set when I_WILL_FREE is
2351 * I_FREEING Set when inode is about to be freed but still has dirty
2352 * pages or buffers attached or the inode itself is still
2354 * I_CLEAR Added by clear_inode(). In this state the inode is
2355 * clean and can be destroyed. Inode keeps I_FREEING.
2357 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2358 * prohibited for many purposes. iget() must wait for
2359 * the inode to be completely released, then create it
2360 * anew. Other functions will just ignore such inodes,
2361 * if appropriate. I_NEW is used for waiting.
2363 * I_SYNC Writeback of inode is running. The bit is set during
2364 * data writeback, and cleared with a wakeup on the bit
2365 * address once it is done. The bit is also used to pin
2366 * the inode in memory for flusher thread.
2368 * I_REFERENCED Marks the inode as recently references on the LRU list.
2370 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2372 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2373 * synchronize competing switching instances and to tell
2374 * wb stat updates to grab the i_pages lock. See
2375 * inode_switch_wbs_work_fn() for details.
2377 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2378 * and work dirs among overlayfs mounts.
2380 * I_CREATING New object's inode in the middle of setting up.
2382 * I_DONTCACHE Evict inode as soon as it is not used anymore.
2384 * I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists.
2385 * Used to detect that mark_inode_dirty() should not move
2386 * inode between dirty lists.
2388 * I_PINNING_FSCACHE_WB Inode is pinning an fscache object for writeback.
2390 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2392 #define I_DIRTY_SYNC (1 << 0)
2393 #define I_DIRTY_DATASYNC (1 << 1)
2394 #define I_DIRTY_PAGES (1 << 2)
2396 #define I_NEW (1 << __I_NEW)
2397 #define I_WILL_FREE (1 << 4)
2398 #define I_FREEING (1 << 5)
2399 #define I_CLEAR (1 << 6)
2401 #define I_SYNC (1 << __I_SYNC)
2402 #define I_REFERENCED (1 << 8)
2403 #define __I_DIO_WAKEUP 9
2404 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2405 #define I_LINKABLE (1 << 10)
2406 #define I_DIRTY_TIME (1 << 11)
2407 #define I_WB_SWITCH (1 << 13)
2408 #define I_OVL_INUSE (1 << 14)
2409 #define I_CREATING (1 << 15)
2410 #define I_DONTCACHE (1 << 16)
2411 #define I_SYNC_QUEUED (1 << 17)
2412 #define I_PINNING_NETFS_WB (1 << 18)
2414 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2415 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2416 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2418 extern void __mark_inode_dirty(struct inode *, int);
2419 static inline void mark_inode_dirty(struct inode *inode)
2421 __mark_inode_dirty(inode, I_DIRTY);
2424 static inline void mark_inode_dirty_sync(struct inode *inode)
2426 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2430 * Returns true if the given inode itself only has dirty timestamps (its pages
2431 * may still be dirty) and isn't currently being allocated or freed.
2432 * Filesystems should call this if when writing an inode when lazytime is
2433 * enabled, they want to opportunistically write the timestamps of other inodes
2434 * located very nearby on-disk, e.g. in the same inode block. This returns true
2435 * if the given inode is in need of such an opportunistic update. Requires
2436 * i_lock, or at least later re-checking under i_lock.
2438 static inline bool inode_is_dirtytime_only(struct inode *inode)
2440 return (inode->i_state & (I_DIRTY_TIME | I_NEW |
2441 I_FREEING | I_WILL_FREE)) == I_DIRTY_TIME;
2444 extern void inc_nlink(struct inode *inode);
2445 extern void drop_nlink(struct inode *inode);
2446 extern void clear_nlink(struct inode *inode);
2447 extern void set_nlink(struct inode *inode, unsigned int nlink);
2449 static inline void inode_inc_link_count(struct inode *inode)
2452 mark_inode_dirty(inode);
2455 static inline void inode_dec_link_count(struct inode *inode)
2458 mark_inode_dirty(inode);
2461 enum file_time_flags {
2468 extern bool atime_needs_update(const struct path *, struct inode *);
2469 extern void touch_atime(const struct path *);
2470 int inode_update_time(struct inode *inode, int flags);
2472 static inline void file_accessed(struct file *file)
2474 if (!(file->f_flags & O_NOATIME))
2475 touch_atime(&file->f_path);
2478 extern int file_modified(struct file *file);
2479 int kiocb_modified(struct kiocb *iocb);
2481 int sync_inode_metadata(struct inode *inode, int wait);
2483 struct file_system_type {
2486 #define FS_REQUIRES_DEV 1
2487 #define FS_BINARY_MOUNTDATA 2
2488 #define FS_HAS_SUBTYPE 4
2489 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2490 #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */
2491 #define FS_ALLOW_IDMAP 32 /* FS has been updated to handle vfs idmappings. */
2492 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2493 int (*init_fs_context)(struct fs_context *);
2494 const struct fs_parameter_spec *parameters;
2495 struct dentry *(*mount) (struct file_system_type *, int,
2496 const char *, void *);
2497 void (*kill_sb) (struct super_block *);
2498 struct module *owner;
2499 struct file_system_type * next;
2500 struct hlist_head fs_supers;
2502 struct lock_class_key s_lock_key;
2503 struct lock_class_key s_umount_key;
2504 struct lock_class_key s_vfs_rename_key;
2505 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2507 struct lock_class_key i_lock_key;
2508 struct lock_class_key i_mutex_key;
2509 struct lock_class_key invalidate_lock_key;
2510 struct lock_class_key i_mutex_dir_key;
2513 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2515 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2516 int flags, const char *dev_name, void *data,
2517 int (*fill_super)(struct super_block *, void *, int));
2518 extern struct dentry *mount_single(struct file_system_type *fs_type,
2519 int flags, void *data,
2520 int (*fill_super)(struct super_block *, void *, int));
2521 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2522 int flags, void *data,
2523 int (*fill_super)(struct super_block *, void *, int));
2524 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2525 void retire_super(struct super_block *sb);
2526 void generic_shutdown_super(struct super_block *sb);
2527 void kill_block_super(struct super_block *sb);
2528 void kill_anon_super(struct super_block *sb);
2529 void kill_litter_super(struct super_block *sb);
2530 void deactivate_super(struct super_block *sb);
2531 void deactivate_locked_super(struct super_block *sb);
2532 int set_anon_super(struct super_block *s, void *data);
2533 int set_anon_super_fc(struct super_block *s, struct fs_context *fc);
2534 int get_anon_bdev(dev_t *);
2535 void free_anon_bdev(dev_t);
2536 struct super_block *sget_fc(struct fs_context *fc,
2537 int (*test)(struct super_block *, struct fs_context *),
2538 int (*set)(struct super_block *, struct fs_context *));
2539 struct super_block *sget(struct file_system_type *type,
2540 int (*test)(struct super_block *,void *),
2541 int (*set)(struct super_block *,void *),
2542 int flags, void *data);
2543 struct super_block *sget_dev(struct fs_context *fc, dev_t dev);
2545 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2546 #define fops_get(fops) \
2547 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2548 #define fops_put(fops) \
2549 do { if (fops) module_put((fops)->owner); } while(0)
2551 * This one is to be used *ONLY* from ->open() instances.
2552 * fops must be non-NULL, pinned down *and* module dependencies
2553 * should be sufficient to pin the caller down as well.
2555 #define replace_fops(f, fops) \
2557 struct file *__file = (f); \
2558 fops_put(__file->f_op); \
2559 BUG_ON(!(__file->f_op = (fops))); \
2562 extern int register_filesystem(struct file_system_type *);
2563 extern int unregister_filesystem(struct file_system_type *);
2564 extern int vfs_statfs(const struct path *, struct kstatfs *);
2565 extern int user_statfs(const char __user *, struct kstatfs *);
2566 extern int fd_statfs(int, struct kstatfs *);
2567 int freeze_super(struct super_block *super, enum freeze_holder who);
2568 int thaw_super(struct super_block *super, enum freeze_holder who);
2569 extern __printf(2, 3)
2570 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2571 extern int super_setup_bdi(struct super_block *sb);
2573 static inline void super_set_uuid(struct super_block *sb, const u8 *uuid, unsigned len)
2575 if (WARN_ON(len > sizeof(sb->s_uuid)))
2576 len = sizeof(sb->s_uuid);
2577 sb->s_uuid_len = len;
2578 memcpy(&sb->s_uuid, uuid, len);
2581 /* set sb sysfs name based on sb->s_bdev */
2582 static inline void super_set_sysfs_name_bdev(struct super_block *sb)
2584 snprintf(sb->s_sysfs_name, sizeof(sb->s_sysfs_name), "%pg", sb->s_bdev);
2587 /* set sb sysfs name based on sb->s_uuid */
2588 static inline void super_set_sysfs_name_uuid(struct super_block *sb)
2590 WARN_ON(sb->s_uuid_len != sizeof(sb->s_uuid));
2591 snprintf(sb->s_sysfs_name, sizeof(sb->s_sysfs_name), "%pU", sb->s_uuid.b);
2594 /* set sb sysfs name based on sb->s_id */
2595 static inline void super_set_sysfs_name_id(struct super_block *sb)
2597 strscpy(sb->s_sysfs_name, sb->s_id, sizeof(sb->s_sysfs_name));
2600 /* try to use something standard before you use this */
2602 static inline void super_set_sysfs_name_generic(struct super_block *sb, const char *fmt, ...)
2606 va_start(args, fmt);
2607 vsnprintf(sb->s_sysfs_name, sizeof(sb->s_sysfs_name), fmt, args);
2611 extern int current_umask(void);
2613 extern void ihold(struct inode * inode);
2614 extern void iput(struct inode *);
2615 int inode_update_timestamps(struct inode *inode, int flags);
2616 int generic_update_time(struct inode *, int);
2619 extern struct kobject *fs_kobj;
2621 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2626 const char *name; /* pointer to actual string */
2627 const __user char *uptr; /* original userland pointer */
2629 struct audit_names *aname;
2632 static_assert(offsetof(struct filename, iname) % sizeof(long) == 0);
2634 static inline struct mnt_idmap *file_mnt_idmap(const struct file *file)
2636 return mnt_idmap(file->f_path.mnt);
2640 * is_idmapped_mnt - check whether a mount is mapped
2641 * @mnt: the mount to check
2643 * If @mnt has an non @nop_mnt_idmap attached to it then @mnt is mapped.
2645 * Return: true if mount is mapped, false if not.
2647 static inline bool is_idmapped_mnt(const struct vfsmount *mnt)
2649 return mnt_idmap(mnt) != &nop_mnt_idmap;
2652 extern long vfs_truncate(const struct path *, loff_t);
2653 int do_truncate(struct mnt_idmap *, struct dentry *, loff_t start,
2654 unsigned int time_attrs, struct file *filp);
2655 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2657 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2659 extern struct file *file_open_name(struct filename *, int, umode_t);
2660 extern struct file *filp_open(const char *, int, umode_t);
2661 extern struct file *file_open_root(const struct path *,
2662 const char *, int, umode_t);
2663 static inline struct file *file_open_root_mnt(struct vfsmount *mnt,
2664 const char *name, int flags, umode_t mode)
2666 return file_open_root(&(struct path){.mnt = mnt, .dentry = mnt->mnt_root},
2669 struct file *dentry_open(const struct path *path, int flags,
2670 const struct cred *creds);
2671 struct file *dentry_create(const struct path *path, int flags, umode_t mode,
2672 const struct cred *cred);
2673 struct path *backing_file_user_path(struct file *f);
2676 * When mmapping a file on a stackable filesystem (e.g., overlayfs), the file
2677 * stored in ->vm_file is a backing file whose f_inode is on the underlying
2678 * filesystem. When the mapped file path and inode number are displayed to
2679 * user (e.g. via /proc/<pid>/maps), these helpers should be used to get the
2680 * path and inode number to display to the user, which is the path of the fd
2681 * that user has requested to map and the inode number that would be returned
2682 * by fstat() on that same fd.
2684 /* Get the path to display in /proc/<pid>/maps */
2685 static inline const struct path *file_user_path(struct file *f)
2687 if (unlikely(f->f_mode & FMODE_BACKING))
2688 return backing_file_user_path(f);
2691 /* Get the inode whose inode number to display in /proc/<pid>/maps */
2692 static inline const struct inode *file_user_inode(struct file *f)
2694 if (unlikely(f->f_mode & FMODE_BACKING))
2695 return d_inode(backing_file_user_path(f)->dentry);
2696 return file_inode(f);
2699 static inline struct file *file_clone_open(struct file *file)
2701 return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2703 extern int filp_close(struct file *, fl_owner_t id);
2705 extern struct filename *getname_flags(const char __user *, int, int *);
2706 extern struct filename *getname_uflags(const char __user *, int);
2707 extern struct filename *getname(const char __user *);
2708 extern struct filename *getname_kernel(const char *);
2709 extern void putname(struct filename *name);
2711 extern int finish_open(struct file *file, struct dentry *dentry,
2712 int (*open)(struct inode *, struct file *));
2713 extern int finish_no_open(struct file *file, struct dentry *dentry);
2715 /* Helper for the simple case when original dentry is used */
2716 static inline int finish_open_simple(struct file *file, int error)
2721 return finish_open(file, file->f_path.dentry, NULL);
2725 extern void __init vfs_caches_init_early(void);
2726 extern void __init vfs_caches_init(void);
2728 extern struct kmem_cache *names_cachep;
2730 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2731 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2733 extern struct super_block *blockdev_superblock;
2734 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2736 return IS_ENABLED(CONFIG_BLOCK) && sb == blockdev_superblock;
2739 void emergency_thaw_all(void);
2740 extern int sync_filesystem(struct super_block *);
2741 extern const struct file_operations def_blk_fops;
2742 extern const struct file_operations def_chr_fops;
2745 #define CHRDEV_MAJOR_MAX 512
2746 /* Marks the bottom of the first segment of free char majors */
2747 #define CHRDEV_MAJOR_DYN_END 234
2748 /* Marks the top and bottom of the second segment of free char majors */
2749 #define CHRDEV_MAJOR_DYN_EXT_START 511
2750 #define CHRDEV_MAJOR_DYN_EXT_END 384
2752 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2753 extern int register_chrdev_region(dev_t, unsigned, const char *);
2754 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2755 unsigned int count, const char *name,
2756 const struct file_operations *fops);
2757 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2758 unsigned int count, const char *name);
2759 extern void unregister_chrdev_region(dev_t, unsigned);
2760 extern void chrdev_show(struct seq_file *,off_t);
2762 static inline int register_chrdev(unsigned int major, const char *name,
2763 const struct file_operations *fops)
2765 return __register_chrdev(major, 0, 256, name, fops);
2768 static inline void unregister_chrdev(unsigned int major, const char *name)
2770 __unregister_chrdev(major, 0, 256, name);
2773 extern void init_special_inode(struct inode *, umode_t, dev_t);
2775 /* Invalid inode operations -- fs/bad_inode.c */
2776 extern void make_bad_inode(struct inode *);
2777 extern bool is_bad_inode(struct inode *);
2779 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2781 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2782 extern int __must_check file_write_and_wait_range(struct file *file,
2783 loff_t start, loff_t end);
2785 static inline int file_write_and_wait(struct file *file)
2787 return file_write_and_wait_range(file, 0, LLONG_MAX);
2790 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2792 extern int vfs_fsync(struct file *file, int datasync);
2794 extern int sync_file_range(struct file *file, loff_t offset, loff_t nbytes,
2795 unsigned int flags);
2797 static inline bool iocb_is_dsync(const struct kiocb *iocb)
2799 return (iocb->ki_flags & IOCB_DSYNC) ||
2800 IS_SYNC(iocb->ki_filp->f_mapping->host);
2804 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2805 * to already be updated for the write, and will return either the amount
2806 * of bytes passed in, or an error if syncing the file failed.
2808 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2810 if (iocb_is_dsync(iocb)) {
2811 int ret = vfs_fsync_range(iocb->ki_filp,
2812 iocb->ki_pos - count, iocb->ki_pos - 1,
2813 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2821 extern void emergency_sync(void);
2822 extern void emergency_remount(void);
2825 extern int bmap(struct inode *inode, sector_t *block);
2827 static inline int bmap(struct inode *inode, sector_t *block)
2833 int notify_change(struct mnt_idmap *, struct dentry *,
2834 struct iattr *, struct inode **);
2835 int inode_permission(struct mnt_idmap *, struct inode *, int);
2836 int generic_permission(struct mnt_idmap *, struct inode *, int);
2837 static inline int file_permission(struct file *file, int mask)
2839 return inode_permission(file_mnt_idmap(file),
2840 file_inode(file), mask);
2842 static inline int path_permission(const struct path *path, int mask)
2844 return inode_permission(mnt_idmap(path->mnt),
2845 d_inode(path->dentry), mask);
2847 int __check_sticky(struct mnt_idmap *idmap, struct inode *dir,
2848 struct inode *inode);
2850 static inline bool execute_ok(struct inode *inode)
2852 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2855 static inline bool inode_wrong_type(const struct inode *inode, umode_t mode)
2857 return (inode->i_mode ^ mode) & S_IFMT;
2861 * file_start_write - get write access to a superblock for regular file io
2862 * @file: the file we want to write to
2864 * This is a variant of sb_start_write() which is a noop on non-regualr file.
2865 * Should be matched with a call to file_end_write().
2867 static inline void file_start_write(struct file *file)
2869 if (!S_ISREG(file_inode(file)->i_mode))
2871 sb_start_write(file_inode(file)->i_sb);
2874 static inline bool file_start_write_trylock(struct file *file)
2876 if (!S_ISREG(file_inode(file)->i_mode))
2878 return sb_start_write_trylock(file_inode(file)->i_sb);
2882 * file_end_write - drop write access to a superblock of a regular file
2883 * @file: the file we wrote to
2885 * Should be matched with a call to file_start_write().
2887 static inline void file_end_write(struct file *file)
2889 if (!S_ISREG(file_inode(file)->i_mode))
2891 sb_end_write(file_inode(file)->i_sb);
2895 * kiocb_start_write - get write access to a superblock for async file io
2896 * @iocb: the io context we want to submit the write with
2898 * This is a variant of sb_start_write() for async io submission.
2899 * Should be matched with a call to kiocb_end_write().
2901 static inline void kiocb_start_write(struct kiocb *iocb)
2903 struct inode *inode = file_inode(iocb->ki_filp);
2905 sb_start_write(inode->i_sb);
2907 * Fool lockdep by telling it the lock got released so that it
2908 * doesn't complain about the held lock when we return to userspace.
2910 __sb_writers_release(inode->i_sb, SB_FREEZE_WRITE);
2914 * kiocb_end_write - drop write access to a superblock after async file io
2915 * @iocb: the io context we sumbitted the write with
2917 * Should be matched with a call to kiocb_start_write().
2919 static inline void kiocb_end_write(struct kiocb *iocb)
2921 struct inode *inode = file_inode(iocb->ki_filp);
2924 * Tell lockdep we inherited freeze protection from submission thread.
2926 __sb_writers_acquired(inode->i_sb, SB_FREEZE_WRITE);
2927 sb_end_write(inode->i_sb);
2931 * This is used for regular files where some users -- especially the
2932 * currently executed binary in a process, previously handled via
2933 * VM_DENYWRITE -- cannot handle concurrent write (and maybe mmap
2934 * read-write shared) accesses.
2936 * get_write_access() gets write permission for a file.
2937 * put_write_access() releases this write permission.
2938 * deny_write_access() denies write access to a file.
2939 * allow_write_access() re-enables write access to a file.
2941 * The i_writecount field of an inode can have the following values:
2942 * 0: no write access, no denied write access
2943 * < 0: (-i_writecount) users that denied write access to the file.
2944 * > 0: (i_writecount) users that have write access to the file.
2946 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2947 * except for the cases where we don't hold i_writecount yet. Then we need to
2948 * use {get,deny}_write_access() - these functions check the sign and refuse
2949 * to do the change if sign is wrong.
2951 static inline int get_write_access(struct inode *inode)
2953 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2955 static inline int deny_write_access(struct file *file)
2957 struct inode *inode = file_inode(file);
2958 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2960 static inline void put_write_access(struct inode * inode)
2962 atomic_dec(&inode->i_writecount);
2964 static inline void allow_write_access(struct file *file)
2967 atomic_inc(&file_inode(file)->i_writecount);
2969 static inline bool inode_is_open_for_write(const struct inode *inode)
2971 return atomic_read(&inode->i_writecount) > 0;
2974 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
2975 static inline void i_readcount_dec(struct inode *inode)
2977 BUG_ON(atomic_dec_return(&inode->i_readcount) < 0);
2979 static inline void i_readcount_inc(struct inode *inode)
2981 atomic_inc(&inode->i_readcount);
2984 static inline void i_readcount_dec(struct inode *inode)
2988 static inline void i_readcount_inc(struct inode *inode)
2993 extern int do_pipe_flags(int *, int);
2995 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
2996 ssize_t __kernel_read(struct file *file, void *buf, size_t count, loff_t *pos);
2997 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
2998 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
2999 extern struct file * open_exec(const char *);
3001 /* fs/dcache.c -- generic fs support functions */
3002 extern bool is_subdir(struct dentry *, struct dentry *);
3003 extern bool path_is_under(const struct path *, const struct path *);
3005 extern char *file_path(struct file *, char *, int);
3008 * is_dot_dotdot - returns true only if @name is "." or ".."
3009 * @name: file name to check
3010 * @len: length of file name, in bytes
3012 static inline bool is_dot_dotdot(const char *name, size_t len)
3014 return len && unlikely(name[0] == '.') &&
3015 (len == 1 || (len == 2 && name[1] == '.'));
3018 #include <linux/err.h>
3020 /* needed for stackable file system support */
3021 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
3023 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
3025 extern int inode_init_always(struct super_block *, struct inode *);
3026 extern void inode_init_once(struct inode *);
3027 extern void address_space_init_once(struct address_space *mapping);
3028 extern struct inode * igrab(struct inode *);
3029 extern ino_t iunique(struct super_block *, ino_t);
3030 extern int inode_needs_sync(struct inode *inode);
3031 extern int generic_delete_inode(struct inode *inode);
3032 static inline int generic_drop_inode(struct inode *inode)
3034 return !inode->i_nlink || inode_unhashed(inode);
3036 extern void d_mark_dontcache(struct inode *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 struct inode *find_inode_rcu(struct super_block *, unsigned long,
3057 int (*)(struct inode *, void *), void *);
3058 extern struct inode *find_inode_by_ino_rcu(struct super_block *, unsigned long);
3059 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
3060 extern int insert_inode_locked(struct inode *);
3061 #ifdef CONFIG_DEBUG_LOCK_ALLOC
3062 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
3064 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
3066 extern void unlock_new_inode(struct inode *);
3067 extern void discard_new_inode(struct inode *);
3068 extern unsigned int get_next_ino(void);
3069 extern void evict_inodes(struct super_block *sb);
3070 void dump_mapping(const struct address_space *);
3073 * Userspace may rely on the inode number being non-zero. For example, glibc
3074 * simply ignores files with zero i_ino in unlink() and other places.
3076 * As an additional complication, if userspace was compiled with
3077 * _FILE_OFFSET_BITS=32 on a 64-bit kernel we'll only end up reading out the
3078 * lower 32 bits, so we need to check that those aren't zero explicitly. With
3079 * _FILE_OFFSET_BITS=64, this may cause some harmless false-negatives, but
3080 * better safe than sorry.
3082 static inline bool is_zero_ino(ino_t ino)
3084 return (u32)ino == 0;
3087 extern void __iget(struct inode * inode);
3088 extern void iget_failed(struct inode *);
3089 extern void clear_inode(struct inode *);
3090 extern void __destroy_inode(struct inode *);
3091 extern struct inode *new_inode_pseudo(struct super_block *sb);
3092 extern struct inode *new_inode(struct super_block *sb);
3093 extern void free_inode_nonrcu(struct inode *inode);
3094 extern int setattr_should_drop_suidgid(struct mnt_idmap *, struct inode *);
3095 extern int file_remove_privs_flags(struct file *file, unsigned int flags);
3096 extern int file_remove_privs(struct file *);
3097 int setattr_should_drop_sgid(struct mnt_idmap *idmap,
3098 const struct inode *inode);
3101 * This must be used for allocating filesystems specific inodes to set
3102 * up the inode reclaim context correctly.
3104 #define alloc_inode_sb(_sb, _cache, _gfp) kmem_cache_alloc_lru(_cache, &_sb->s_inode_lru, _gfp)
3106 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
3107 static inline void insert_inode_hash(struct inode *inode)
3109 __insert_inode_hash(inode, inode->i_ino);
3112 extern void __remove_inode_hash(struct inode *);
3113 static inline void remove_inode_hash(struct inode *inode)
3115 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
3116 __remove_inode_hash(inode);
3119 extern void inode_sb_list_add(struct inode *inode);
3120 extern void inode_add_lru(struct inode *inode);
3122 extern int sb_set_blocksize(struct super_block *, int);
3123 extern int sb_min_blocksize(struct super_block *, int);
3125 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
3126 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
3127 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
3128 int generic_write_checks_count(struct kiocb *iocb, loff_t *count);
3129 extern int generic_write_check_limits(struct file *file, loff_t pos,
3131 extern int generic_file_rw_checks(struct file *file_in, struct file *file_out);
3132 ssize_t filemap_read(struct kiocb *iocb, struct iov_iter *to,
3133 ssize_t already_read);
3134 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
3135 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
3136 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
3137 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
3138 ssize_t generic_perform_write(struct kiocb *, struct iov_iter *);
3139 ssize_t direct_write_fallback(struct kiocb *iocb, struct iov_iter *iter,
3140 ssize_t direct_written, ssize_t buffered_written);
3142 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
3144 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
3146 ssize_t vfs_iocb_iter_read(struct file *file, struct kiocb *iocb,
3147 struct iov_iter *iter);
3148 ssize_t vfs_iocb_iter_write(struct file *file, struct kiocb *iocb,
3149 struct iov_iter *iter);
3152 ssize_t filemap_splice_read(struct file *in, loff_t *ppos,
3153 struct pipe_inode_info *pipe,
3154 size_t len, unsigned int flags);
3155 ssize_t copy_splice_read(struct file *in, loff_t *ppos,
3156 struct pipe_inode_info *pipe,
3157 size_t len, unsigned int flags);
3158 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
3159 struct file *, loff_t *, size_t, unsigned int);
3163 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3164 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3165 #define no_llseek NULL
3166 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3167 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3168 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3169 int whence, loff_t maxsize, loff_t eof);
3170 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3171 int whence, loff_t size);
3172 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3173 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3174 int rw_verify_area(int, struct file *, const loff_t *, size_t);
3175 extern int generic_file_open(struct inode * inode, struct file * filp);
3176 extern int nonseekable_open(struct inode * inode, struct file * filp);
3177 extern int stream_open(struct inode * inode, struct file * filp);
3180 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3181 loff_t file_offset);
3184 /* need locking between buffered and direct access */
3187 /* filesystem does not support filling holes */
3188 DIO_SKIP_HOLES = 0x02,
3191 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3192 struct block_device *bdev, struct iov_iter *iter,
3193 get_block_t get_block,
3194 dio_iodone_t end_io,
3197 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3198 struct inode *inode,
3199 struct iov_iter *iter,
3200 get_block_t get_block)
3202 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3203 get_block, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3207 void inode_dio_wait(struct inode *inode);
3210 * inode_dio_begin - signal start of a direct I/O requests
3211 * @inode: inode the direct I/O happens on
3213 * This is called once we've finished processing a direct I/O request,
3214 * and is used to wake up callers waiting for direct I/O to be quiesced.
3216 static inline void inode_dio_begin(struct inode *inode)
3218 atomic_inc(&inode->i_dio_count);
3222 * inode_dio_end - signal finish of a direct I/O requests
3223 * @inode: inode the direct I/O happens on
3225 * This is called once we've finished processing a direct I/O request,
3226 * and is used to wake up callers waiting for direct I/O to be quiesced.
3228 static inline void inode_dio_end(struct inode *inode)
3230 if (atomic_dec_and_test(&inode->i_dio_count))
3231 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3234 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3237 extern const struct file_operations generic_ro_fops;
3239 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3241 extern int readlink_copy(char __user *, int, const char *);
3242 extern int page_readlink(struct dentry *, char __user *, int);
3243 extern const char *page_get_link(struct dentry *, struct inode *,
3244 struct delayed_call *);
3245 extern void page_put_link(void *);
3246 extern int page_symlink(struct inode *inode, const char *symname, int len);
3247 extern const struct inode_operations page_symlink_inode_operations;
3248 extern void kfree_link(void *);
3249 void generic_fillattr(struct mnt_idmap *, u32, struct inode *, struct kstat *);
3250 void generic_fill_statx_attr(struct inode *inode, struct kstat *stat);
3251 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3252 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3253 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3254 void inode_add_bytes(struct inode *inode, loff_t bytes);
3255 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3256 void inode_sub_bytes(struct inode *inode, loff_t bytes);
3257 static inline loff_t __inode_get_bytes(struct inode *inode)
3259 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3261 loff_t inode_get_bytes(struct inode *inode);
3262 void inode_set_bytes(struct inode *inode, loff_t bytes);
3263 const char *simple_get_link(struct dentry *, struct inode *,
3264 struct delayed_call *);
3265 extern const struct inode_operations simple_symlink_inode_operations;
3267 extern int iterate_dir(struct file *, struct dir_context *);
3269 int vfs_fstatat(int dfd, const char __user *filename, struct kstat *stat,
3271 int vfs_fstat(int fd, struct kstat *stat);
3273 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3275 return vfs_fstatat(AT_FDCWD, filename, stat, 0);
3277 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3279 return vfs_fstatat(AT_FDCWD, name, stat, AT_SYMLINK_NOFOLLOW);
3282 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3283 extern int vfs_readlink(struct dentry *, char __user *, int);
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 void drop_super(struct super_block *sb);
3289 extern void drop_super_exclusive(struct super_block *sb);
3290 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3291 extern void iterate_supers_type(struct file_system_type *,
3292 void (*)(struct super_block *, void *), void *);
3294 extern int dcache_dir_open(struct inode *, struct file *);
3295 extern int dcache_dir_close(struct inode *, struct file *);
3296 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3297 extern int dcache_readdir(struct file *, struct dir_context *);
3298 extern int simple_setattr(struct mnt_idmap *, struct dentry *,
3300 extern int simple_getattr(struct mnt_idmap *, const struct path *,
3301 struct kstat *, u32, unsigned int);
3302 extern int simple_statfs(struct dentry *, struct kstatfs *);
3303 extern int simple_open(struct inode *inode, struct file *file);
3304 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3305 extern int simple_unlink(struct inode *, struct dentry *);
3306 extern int simple_rmdir(struct inode *, struct dentry *);
3307 void simple_rename_timestamp(struct inode *old_dir, struct dentry *old_dentry,
3308 struct inode *new_dir, struct dentry *new_dentry);
3309 extern int simple_rename_exchange(struct inode *old_dir, struct dentry *old_dentry,
3310 struct inode *new_dir, struct dentry *new_dentry);
3311 extern int simple_rename(struct mnt_idmap *, struct inode *,
3312 struct dentry *, struct inode *, struct dentry *,
3314 extern void simple_recursive_removal(struct dentry *,
3315 void (*callback)(struct dentry *));
3316 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3317 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3318 extern int simple_empty(struct dentry *);
3319 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3320 loff_t pos, unsigned len,
3321 struct page **pagep, void **fsdata);
3322 extern const struct address_space_operations ram_aops;
3323 extern int always_delete_dentry(const struct dentry *);
3324 extern struct inode *alloc_anon_inode(struct super_block *);
3325 extern int simple_nosetlease(struct file *, int, struct file_lease **, void **);
3326 extern const struct dentry_operations simple_dentry_operations;
3328 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3329 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3330 extern const struct file_operations simple_dir_operations;
3331 extern const struct inode_operations simple_dir_inode_operations;
3332 extern void make_empty_dir_inode(struct inode *inode);
3333 extern bool is_empty_dir_inode(struct inode *inode);
3334 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3335 struct dentry *d_alloc_name(struct dentry *, const char *);
3336 extern int simple_fill_super(struct super_block *, unsigned long,
3337 const struct tree_descr *);
3338 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3339 extern void simple_release_fs(struct vfsmount **mount, int *count);
3341 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3342 loff_t *ppos, const void *from, size_t available);
3343 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3344 const void __user *from, size_t count);
3347 struct maple_tree mt;
3348 unsigned long next_offset;
3351 void simple_offset_init(struct offset_ctx *octx);
3352 int simple_offset_add(struct offset_ctx *octx, struct dentry *dentry);
3353 void simple_offset_remove(struct offset_ctx *octx, struct dentry *dentry);
3354 int simple_offset_empty(struct dentry *dentry);
3355 int simple_offset_rename(struct inode *old_dir, struct dentry *old_dentry,
3356 struct inode *new_dir, struct dentry *new_dentry);
3357 int simple_offset_rename_exchange(struct inode *old_dir,
3358 struct dentry *old_dentry,
3359 struct inode *new_dir,
3360 struct dentry *new_dentry);
3361 void simple_offset_destroy(struct offset_ctx *octx);
3363 extern const struct file_operations simple_offset_dir_operations;
3365 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3366 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3368 extern int generic_check_addressable(unsigned, u64);
3370 extern void generic_set_sb_d_ops(struct super_block *sb);
3372 static inline bool sb_has_encoding(const struct super_block *sb)
3374 #if IS_ENABLED(CONFIG_UNICODE)
3375 return !!sb->s_encoding;
3381 int may_setattr(struct mnt_idmap *idmap, struct inode *inode,
3382 unsigned int ia_valid);
3383 int setattr_prepare(struct mnt_idmap *, struct dentry *, struct iattr *);
3384 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3385 void setattr_copy(struct mnt_idmap *, struct inode *inode,
3386 const struct iattr *attr);
3388 extern int file_update_time(struct file *file);
3390 static inline bool vma_is_dax(const struct vm_area_struct *vma)
3392 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3395 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3397 struct inode *inode;
3399 if (!IS_ENABLED(CONFIG_FS_DAX) || !vma->vm_file)
3401 if (!vma_is_dax(vma))
3403 inode = file_inode(vma->vm_file);
3404 if (S_ISCHR(inode->i_mode))
3405 return false; /* device-dax */
3409 static inline int iocb_flags(struct file *file)
3412 if (file->f_flags & O_APPEND)
3414 if (file->f_flags & O_DIRECT)
3416 if (file->f_flags & O_DSYNC)
3418 if (file->f_flags & __O_SYNC)
3423 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3425 int kiocb_flags = 0;
3427 /* make sure there's no overlap between RWF and private IOCB flags */
3428 BUILD_BUG_ON((__force int) RWF_SUPPORTED & IOCB_EVENTFD);
3432 if (unlikely(flags & ~RWF_SUPPORTED))
3434 if (unlikely((flags & RWF_APPEND) && (flags & RWF_NOAPPEND)))
3437 if (flags & RWF_NOWAIT) {
3438 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3440 kiocb_flags |= IOCB_NOIO;
3442 kiocb_flags |= (__force int) (flags & RWF_SUPPORTED);
3443 if (flags & RWF_SYNC)
3444 kiocb_flags |= IOCB_DSYNC;
3446 if ((flags & RWF_NOAPPEND) && (ki->ki_flags & IOCB_APPEND)) {
3447 if (IS_APPEND(file_inode(ki->ki_filp)))
3449 ki->ki_flags &= ~IOCB_APPEND;
3452 ki->ki_flags |= kiocb_flags;
3456 static inline ino_t parent_ino(struct dentry *dentry)
3461 * Don't strictly need d_lock here? If the parent ino could change
3462 * then surely we'd have a deeper race in the caller?
3464 spin_lock(&dentry->d_lock);
3465 res = dentry->d_parent->d_inode->i_ino;
3466 spin_unlock(&dentry->d_lock);
3470 /* Transaction based IO helpers */
3473 * An argresp is stored in an allocated page and holds the
3474 * size of the argument or response, along with its content
3476 struct simple_transaction_argresp {
3481 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3483 char *simple_transaction_get(struct file *file, const char __user *buf,
3485 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3486 size_t size, loff_t *pos);
3487 int simple_transaction_release(struct inode *inode, struct file *file);
3489 void simple_transaction_set(struct file *file, size_t n);
3492 * simple attribute files
3494 * These attributes behave similar to those in sysfs:
3496 * Writing to an attribute immediately sets a value, an open file can be
3497 * written to multiple times.
3499 * Reading from an attribute creates a buffer from the value that might get
3500 * read with multiple read calls. When the attribute has been read
3501 * completely, no further read calls are possible until the file is opened
3504 * All attributes contain a text representation of a numeric value
3505 * that are accessed with the get() and set() functions.
3507 #define DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, __is_signed) \
3508 static int __fops ## _open(struct inode *inode, struct file *file) \
3510 __simple_attr_check_format(__fmt, 0ull); \
3511 return simple_attr_open(inode, file, __get, __set, __fmt); \
3513 static const struct file_operations __fops = { \
3514 .owner = THIS_MODULE, \
3515 .open = __fops ## _open, \
3516 .release = simple_attr_release, \
3517 .read = simple_attr_read, \
3518 .write = (__is_signed) ? simple_attr_write_signed : simple_attr_write, \
3519 .llseek = generic_file_llseek, \
3522 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3523 DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, false)
3525 #define DEFINE_SIMPLE_ATTRIBUTE_SIGNED(__fops, __get, __set, __fmt) \
3526 DEFINE_SIMPLE_ATTRIBUTE_XSIGNED(__fops, __get, __set, __fmt, true)
3528 static inline __printf(1, 2)
3529 void __simple_attr_check_format(const char *fmt, ...)
3531 /* don't do anything, just let the compiler check the arguments; */
3534 int simple_attr_open(struct inode *inode, struct file *file,
3535 int (*get)(void *, u64 *), int (*set)(void *, u64),
3537 int simple_attr_release(struct inode *inode, struct file *file);
3538 ssize_t simple_attr_read(struct file *file, char __user *buf,
3539 size_t len, loff_t *ppos);
3540 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3541 size_t len, loff_t *ppos);
3542 ssize_t simple_attr_write_signed(struct file *file, const char __user *buf,
3543 size_t len, loff_t *ppos);
3546 int __init list_bdev_fs_names(char *buf, size_t size);
3548 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3549 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3551 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3552 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3553 (flag & __FMODE_NONOTIFY)))
3555 static inline bool is_sxid(umode_t mode)
3557 return mode & (S_ISUID | S_ISGID);
3560 static inline int check_sticky(struct mnt_idmap *idmap,
3561 struct inode *dir, struct inode *inode)
3563 if (!(dir->i_mode & S_ISVTX))
3566 return __check_sticky(idmap, dir, inode);
3569 static inline void inode_has_no_xattr(struct inode *inode)
3571 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3572 inode->i_flags |= S_NOSEC;
3575 static inline bool is_root_inode(struct inode *inode)
3577 return inode == inode->i_sb->s_root->d_inode;
3580 static inline bool dir_emit(struct dir_context *ctx,
3581 const char *name, int namelen,
3582 u64 ino, unsigned type)
3584 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type);
3586 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3588 return ctx->actor(ctx, ".", 1, ctx->pos,
3589 file->f_path.dentry->d_inode->i_ino, DT_DIR);
3591 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3593 return ctx->actor(ctx, "..", 2, ctx->pos,
3594 parent_ino(file->f_path.dentry), DT_DIR);
3596 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3598 if (ctx->pos == 0) {
3599 if (!dir_emit_dot(file, ctx))
3603 if (ctx->pos == 1) {
3604 if (!dir_emit_dotdot(file, ctx))
3610 static inline bool dir_relax(struct inode *inode)
3612 inode_unlock(inode);
3614 return !IS_DEADDIR(inode);
3617 static inline bool dir_relax_shared(struct inode *inode)
3619 inode_unlock_shared(inode);
3620 inode_lock_shared(inode);
3621 return !IS_DEADDIR(inode);
3624 extern bool path_noexec(const struct path *path);
3625 extern void inode_nohighmem(struct inode *inode);
3628 extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len,
3630 extern int generic_fadvise(struct file *file, loff_t offset, loff_t len,
3633 #endif /* _LINUX_FS_H */