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>
47 #include <asm/byteorder.h>
48 #include <uapi/linux/fs.h>
50 struct backing_dev_info;
54 struct export_operations;
55 struct fiemap_extent_info;
60 struct pipe_inode_info;
61 struct poll_table_struct;
63 struct vm_area_struct;
66 struct swap_info_struct;
68 struct workqueue_struct;
71 struct fscrypt_operations;
73 struct fsverity_operations;
75 struct fs_parameter_spec;
78 extern void __init inode_init(void);
79 extern void __init inode_init_early(void);
80 extern void __init files_init(void);
81 extern void __init files_maxfiles_init(void);
83 extern unsigned long get_max_files(void);
84 extern unsigned int sysctl_nr_open;
86 typedef __kernel_rwf_t rwf_t;
89 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
90 struct buffer_head *bh_result, int create);
91 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
92 ssize_t bytes, void *private);
94 #define MAY_EXEC 0x00000001
95 #define MAY_WRITE 0x00000002
96 #define MAY_READ 0x00000004
97 #define MAY_APPEND 0x00000008
98 #define MAY_ACCESS 0x00000010
99 #define MAY_OPEN 0x00000020
100 #define MAY_CHDIR 0x00000040
101 /* called from RCU mode, don't block */
102 #define MAY_NOT_BLOCK 0x00000080
105 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
106 * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
109 /* file is open for reading */
110 #define FMODE_READ ((__force fmode_t)0x1)
111 /* file is open for writing */
112 #define FMODE_WRITE ((__force fmode_t)0x2)
113 /* file is seekable */
114 #define FMODE_LSEEK ((__force fmode_t)0x4)
115 /* file can be accessed using pread */
116 #define FMODE_PREAD ((__force fmode_t)0x8)
117 /* file can be accessed using pwrite */
118 #define FMODE_PWRITE ((__force fmode_t)0x10)
119 /* File is opened for execution with sys_execve / sys_uselib */
120 #define FMODE_EXEC ((__force fmode_t)0x20)
121 /* File is opened with O_NDELAY (only set for block devices) */
122 #define FMODE_NDELAY ((__force fmode_t)0x40)
123 /* File is opened with O_EXCL (only set for block devices) */
124 #define FMODE_EXCL ((__force fmode_t)0x80)
125 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
126 (specialy hack for floppy.c) */
127 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
128 /* 32bit hashes as llseek() offset (for directories) */
129 #define FMODE_32BITHASH ((__force fmode_t)0x200)
130 /* 64bit hashes as llseek() offset (for directories) */
131 #define FMODE_64BITHASH ((__force fmode_t)0x400)
134 * Don't update ctime and mtime.
136 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
137 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
139 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
141 /* Expect random access pattern */
142 #define FMODE_RANDOM ((__force fmode_t)0x1000)
144 /* File is huge (eg. /dev/mem): treat loff_t as unsigned */
145 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
147 /* File is opened with O_PATH; almost nothing can be done with it */
148 #define FMODE_PATH ((__force fmode_t)0x4000)
150 /* File needs atomic accesses to f_pos */
151 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
152 /* Write access to underlying fs */
153 #define FMODE_WRITER ((__force fmode_t)0x10000)
154 /* Has read method(s) */
155 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
156 /* Has write method(s) */
157 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
159 #define FMODE_OPENED ((__force fmode_t)0x80000)
160 #define FMODE_CREATED ((__force fmode_t)0x100000)
162 /* File is stream-like */
163 #define FMODE_STREAM ((__force fmode_t)0x200000)
165 /* File supports DIRECT IO */
166 #define FMODE_CAN_ODIRECT ((__force fmode_t)0x400000)
168 /* File was opened by fanotify and shouldn't generate fanotify events */
169 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
171 /* File is capable of returning -EAGAIN if I/O will block */
172 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
174 /* File represents mount that needs unmounting */
175 #define FMODE_NEED_UNMOUNT ((__force fmode_t)0x10000000)
177 /* File does not contribute to nr_files count */
178 #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000)
180 /* File supports async buffered reads */
181 #define FMODE_BUF_RASYNC ((__force fmode_t)0x40000000)
183 /* File supports async nowait buffered writes */
184 #define FMODE_BUF_WASYNC ((__force fmode_t)0x80000000)
187 * Attribute flags. These should be or-ed together to figure out what
190 #define ATTR_MODE (1 << 0)
191 #define ATTR_UID (1 << 1)
192 #define ATTR_GID (1 << 2)
193 #define ATTR_SIZE (1 << 3)
194 #define ATTR_ATIME (1 << 4)
195 #define ATTR_MTIME (1 << 5)
196 #define ATTR_CTIME (1 << 6)
197 #define ATTR_ATIME_SET (1 << 7)
198 #define ATTR_MTIME_SET (1 << 8)
199 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
200 #define ATTR_KILL_SUID (1 << 11)
201 #define ATTR_KILL_SGID (1 << 12)
202 #define ATTR_FILE (1 << 13)
203 #define ATTR_KILL_PRIV (1 << 14)
204 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
205 #define ATTR_TIMES_SET (1 << 16)
206 #define ATTR_TOUCH (1 << 17)
209 * Whiteout is represented by a char device. The following constants define the
210 * mode and device number to use.
212 #define WHITEOUT_MODE 0
213 #define WHITEOUT_DEV 0
216 * This is the Inode Attributes structure, used for notify_change(). It
217 * uses the above definitions as flags, to know which values have changed.
218 * Also, in this manner, a Filesystem can look at only the values it cares
219 * about. Basically, these are the attributes that the VFS layer can
220 * request to change from the FS layer.
222 * Derek Atkins <warlord@MIT.EDU> 94-10-20
225 unsigned int ia_valid;
228 * The two anonymous unions wrap structures with the same member.
230 * Filesystems raising FS_ALLOW_IDMAP need to use ia_vfs{g,u}id which
231 * are a dedicated type requiring the filesystem to use the dedicated
232 * helpers. Other filesystem can continue to use ia_{g,u}id until they
235 * They always contain the same value. In other words FS_ALLOW_IDMAP
236 * pass down the same value on idmapped mounts as they would on regular
248 struct timespec64 ia_atime;
249 struct timespec64 ia_mtime;
250 struct timespec64 ia_ctime;
253 * Not an attribute, but an auxiliary info for filesystems wanting to
254 * implement an ftruncate() like method. NOTE: filesystem should
255 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
257 struct file *ia_file;
261 * Includes for diskquotas.
263 #include <linux/quota.h>
266 * Maximum number of layers of fs stack. Needs to be limited to
267 * prevent kernel stack overflow
269 #define FILESYSTEM_MAX_STACK_DEPTH 2
272 * enum positive_aop_returns - aop return codes with specific semantics
274 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
275 * completed, that the page is still locked, and
276 * should be considered active. The VM uses this hint
277 * to return the page to the active list -- it won't
278 * be a candidate for writeback again in the near
279 * future. Other callers must be careful to unlock
280 * the page if they get this return. Returned by
283 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
284 * unlocked it and the page might have been truncated.
285 * The caller should back up to acquiring a new page and
286 * trying again. The aop will be taking reasonable
287 * precautions not to livelock. If the caller held a page
288 * reference, it should drop it before retrying. Returned
291 * address_space_operation functions return these large constants to indicate
292 * special semantics to the caller. These are much larger than the bytes in a
293 * page to allow for functions that return the number of bytes operated on in a
297 enum positive_aop_returns {
298 AOP_WRITEPAGE_ACTIVATE = 0x80000,
299 AOP_TRUNCATED_PAGE = 0x80001,
303 * oh the beauties of C type declarations.
306 struct address_space;
307 struct writeback_control;
308 struct readahead_control;
311 * Write life time hint values.
312 * Stored in struct inode as u8.
315 WRITE_LIFE_NOT_SET = 0,
316 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
317 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
318 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
319 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
320 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
323 /* Match RWF_* bits to IOCB bits */
324 #define IOCB_HIPRI (__force int) RWF_HIPRI
325 #define IOCB_DSYNC (__force int) RWF_DSYNC
326 #define IOCB_SYNC (__force int) RWF_SYNC
327 #define IOCB_NOWAIT (__force int) RWF_NOWAIT
328 #define IOCB_APPEND (__force int) RWF_APPEND
330 /* non-RWF related bits - start at 16 */
331 #define IOCB_EVENTFD (1 << 16)
332 #define IOCB_DIRECT (1 << 17)
333 #define IOCB_WRITE (1 << 18)
334 /* iocb->ki_waitq is valid */
335 #define IOCB_WAITQ (1 << 19)
336 #define IOCB_NOIO (1 << 20)
337 /* can use bio alloc cache */
338 #define IOCB_ALLOC_CACHE (1 << 21)
341 struct file *ki_filp;
343 /* The 'ki_filp' pointer is shared in a union for aio */
344 randomized_struct_fields_start
347 void (*ki_complete)(struct kiocb *iocb, long ret);
350 u16 ki_ioprio; /* See linux/ioprio.h */
351 struct wait_page_queue *ki_waitq; /* for async buffered IO */
352 randomized_struct_fields_end
355 static inline bool is_sync_kiocb(struct kiocb *kiocb)
357 return kiocb->ki_complete == NULL;
360 struct address_space_operations {
361 int (*writepage)(struct page *page, struct writeback_control *wbc);
362 int (*read_folio)(struct file *, struct folio *);
364 /* Write back some dirty pages from this mapping. */
365 int (*writepages)(struct address_space *, struct writeback_control *);
367 /* Mark a folio dirty. Return true if this dirtied it */
368 bool (*dirty_folio)(struct address_space *, struct folio *);
370 void (*readahead)(struct readahead_control *);
372 int (*write_begin)(struct file *, struct address_space *mapping,
373 loff_t pos, unsigned len,
374 struct page **pagep, void **fsdata);
375 int (*write_end)(struct file *, struct address_space *mapping,
376 loff_t pos, unsigned len, unsigned copied,
377 struct page *page, void *fsdata);
379 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
380 sector_t (*bmap)(struct address_space *, sector_t);
381 void (*invalidate_folio) (struct folio *, size_t offset, size_t len);
382 bool (*release_folio)(struct folio *, gfp_t);
383 void (*free_folio)(struct folio *folio);
384 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
386 * migrate the contents of a folio to the specified target. If
387 * migrate_mode is MIGRATE_ASYNC, it must not block.
389 int (*migrate_folio)(struct address_space *, struct folio *dst,
390 struct folio *src, enum migrate_mode);
391 int (*launder_folio)(struct folio *);
392 bool (*is_partially_uptodate) (struct folio *, size_t from,
394 void (*is_dirty_writeback) (struct folio *, bool *dirty, bool *wb);
395 int (*error_remove_page)(struct address_space *, struct page *);
397 /* swapfile support */
398 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
400 void (*swap_deactivate)(struct file *file);
401 int (*swap_rw)(struct kiocb *iocb, struct iov_iter *iter);
404 extern const struct address_space_operations empty_aops;
407 * struct address_space - Contents of a cacheable, mappable object.
408 * @host: Owner, either the inode or the block_device.
409 * @i_pages: Cached pages.
410 * @invalidate_lock: Guards coherency between page cache contents and
411 * file offset->disk block mappings in the filesystem during invalidates.
412 * It is also used to block modification of page cache contents through
414 * @gfp_mask: Memory allocation flags to use for allocating pages.
415 * @i_mmap_writable: Number of VM_SHARED mappings.
416 * @nr_thps: Number of THPs in the pagecache (non-shmem only).
417 * @i_mmap: Tree of private and shared mappings.
418 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
419 * @nrpages: Number of page entries, protected by the i_pages lock.
420 * @writeback_index: Writeback starts here.
422 * @flags: Error bits and flags (AS_*).
423 * @wb_err: The most recent error which has occurred.
424 * @private_lock: For use by the owner of the address_space.
425 * @private_list: For use by the owner of the address_space.
426 * @private_data: For use by the owner of the address_space.
428 struct address_space {
430 struct xarray i_pages;
431 struct rw_semaphore invalidate_lock;
433 atomic_t i_mmap_writable;
434 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
435 /* number of thp, only for non-shmem files */
438 struct rb_root_cached i_mmap;
439 struct rw_semaphore i_mmap_rwsem;
440 unsigned long nrpages;
441 pgoff_t writeback_index;
442 const struct address_space_operations *a_ops;
445 spinlock_t private_lock;
446 struct list_head private_list;
448 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
450 * On most architectures that alignment is already the case; but
451 * must be enforced here for CRIS, to let the least significant bit
452 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
455 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
456 #define PAGECACHE_TAG_DIRTY XA_MARK_0
457 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
458 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
461 * Returns true if any of the pages in the mapping are marked with the tag.
463 static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag)
465 return xa_marked(&mapping->i_pages, tag);
468 static inline void i_mmap_lock_write(struct address_space *mapping)
470 down_write(&mapping->i_mmap_rwsem);
473 static inline int i_mmap_trylock_write(struct address_space *mapping)
475 return down_write_trylock(&mapping->i_mmap_rwsem);
478 static inline void i_mmap_unlock_write(struct address_space *mapping)
480 up_write(&mapping->i_mmap_rwsem);
483 static inline int i_mmap_trylock_read(struct address_space *mapping)
485 return down_read_trylock(&mapping->i_mmap_rwsem);
488 static inline void i_mmap_lock_read(struct address_space *mapping)
490 down_read(&mapping->i_mmap_rwsem);
493 static inline void i_mmap_unlock_read(struct address_space *mapping)
495 up_read(&mapping->i_mmap_rwsem);
498 static inline void i_mmap_assert_locked(struct address_space *mapping)
500 lockdep_assert_held(&mapping->i_mmap_rwsem);
503 static inline void i_mmap_assert_write_locked(struct address_space *mapping)
505 lockdep_assert_held_write(&mapping->i_mmap_rwsem);
509 * Might pages of this file be mapped into userspace?
511 static inline int mapping_mapped(struct address_space *mapping)
513 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
517 * Might pages of this file have been modified in userspace?
518 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap
519 * marks vma as VM_SHARED if it is shared, and the file was opened for
520 * writing i.e. vma may be mprotected writable even if now readonly.
522 * If i_mmap_writable is negative, no new writable mappings are allowed. You
523 * can only deny writable mappings, if none exists right now.
525 static inline int mapping_writably_mapped(struct address_space *mapping)
527 return atomic_read(&mapping->i_mmap_writable) > 0;
530 static inline int mapping_map_writable(struct address_space *mapping)
532 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
536 static inline void mapping_unmap_writable(struct address_space *mapping)
538 atomic_dec(&mapping->i_mmap_writable);
541 static inline int mapping_deny_writable(struct address_space *mapping)
543 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
547 static inline void mapping_allow_writable(struct address_space *mapping)
549 atomic_inc(&mapping->i_mmap_writable);
553 * Use sequence counter to get consistent i_size on 32-bit processors.
555 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
556 #include <linux/seqlock.h>
557 #define __NEED_I_SIZE_ORDERED
558 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
560 #define i_size_ordered_init(inode) do { } while (0)
564 #define ACL_NOT_CACHED ((void *)(-1))
566 * ACL_DONT_CACHE is for stacked filesystems, that rely on underlying fs to
567 * cache the ACL. This also means that ->get_acl() can be called in RCU mode
568 * with the LOOKUP_RCU flag.
570 #define ACL_DONT_CACHE ((void *)(-3))
572 static inline struct posix_acl *
573 uncached_acl_sentinel(struct task_struct *task)
575 return (void *)task + 1;
579 is_uncached_acl(struct posix_acl *acl)
581 return (long)acl & 1;
584 #define IOP_FASTPERM 0x0001
585 #define IOP_LOOKUP 0x0002
586 #define IOP_NOFOLLOW 0x0004
587 #define IOP_XATTR 0x0008
588 #define IOP_DEFAULT_READLINK 0x0010
590 struct fsnotify_mark_connector;
593 * Keep mostly read-only and often accessed (especially for
594 * the RCU path lookup and 'stat' data) fields at the beginning
595 * of the 'struct inode'
599 unsigned short i_opflags;
602 unsigned int i_flags;
604 #ifdef CONFIG_FS_POSIX_ACL
605 struct posix_acl *i_acl;
606 struct posix_acl *i_default_acl;
609 const struct inode_operations *i_op;
610 struct super_block *i_sb;
611 struct address_space *i_mapping;
613 #ifdef CONFIG_SECURITY
617 /* Stat data, not accessed from path walking */
620 * Filesystems may only read i_nlink directly. They shall use the
621 * following functions for modification:
623 * (set|clear|inc|drop)_nlink
624 * inode_(inc|dec)_link_count
627 const unsigned int i_nlink;
628 unsigned int __i_nlink;
632 struct timespec64 i_atime;
633 struct timespec64 i_mtime;
634 struct timespec64 i_ctime;
635 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
636 unsigned short i_bytes;
641 #ifdef __NEED_I_SIZE_ORDERED
642 seqcount_t i_size_seqcount;
646 unsigned long i_state;
647 struct rw_semaphore i_rwsem;
649 unsigned long dirtied_when; /* jiffies of first dirtying */
650 unsigned long dirtied_time_when;
652 struct hlist_node i_hash;
653 struct list_head i_io_list; /* backing dev IO list */
654 #ifdef CONFIG_CGROUP_WRITEBACK
655 struct bdi_writeback *i_wb; /* the associated cgroup wb */
657 /* foreign inode detection, see wbc_detach_inode() */
659 u16 i_wb_frn_avg_time;
660 u16 i_wb_frn_history;
662 struct list_head i_lru; /* inode LRU list */
663 struct list_head i_sb_list;
664 struct list_head i_wb_list; /* backing dev writeback list */
666 struct hlist_head i_dentry;
667 struct rcu_head i_rcu;
669 atomic64_t i_version;
670 atomic64_t i_sequence; /* see futex */
672 atomic_t i_dio_count;
673 atomic_t i_writecount;
674 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
675 atomic_t i_readcount; /* struct files open RO */
678 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
679 void (*free_inode)(struct inode *);
681 struct file_lock_context *i_flctx;
682 struct address_space i_data;
683 struct list_head i_devices;
685 struct pipe_inode_info *i_pipe;
693 #ifdef CONFIG_FSNOTIFY
694 __u32 i_fsnotify_mask; /* all events this inode cares about */
695 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
698 #ifdef CONFIG_FS_ENCRYPTION
699 struct fscrypt_info *i_crypt_info;
702 #ifdef CONFIG_FS_VERITY
703 struct fsverity_info *i_verity_info;
706 void *i_private; /* fs or device private pointer */
707 } __randomize_layout;
709 struct timespec64 timestamp_truncate(struct timespec64 t, struct inode *inode);
711 static inline unsigned int i_blocksize(const struct inode *node)
713 return (1 << node->i_blkbits);
716 static inline int inode_unhashed(struct inode *inode)
718 return hlist_unhashed(&inode->i_hash);
722 * __mark_inode_dirty expects inodes to be hashed. Since we don't
723 * want special inodes in the fileset inode space, we make them
724 * appear hashed, but do not put on any lists. hlist_del()
725 * will work fine and require no locking.
727 static inline void inode_fake_hash(struct inode *inode)
729 hlist_add_fake(&inode->i_hash);
733 * inode->i_mutex nesting subclasses for the lock validator:
735 * 0: the object of the current VFS operation
739 * 4: second non-directory
740 * 5: second parent (when locking independent directories in rename)
742 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
743 * non-directories at once.
745 * The locking order between these classes is
746 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
748 enum inode_i_mutex_lock_class
758 static inline void inode_lock(struct inode *inode)
760 down_write(&inode->i_rwsem);
763 static inline void inode_unlock(struct inode *inode)
765 up_write(&inode->i_rwsem);
768 static inline void inode_lock_shared(struct inode *inode)
770 down_read(&inode->i_rwsem);
773 static inline void inode_unlock_shared(struct inode *inode)
775 up_read(&inode->i_rwsem);
778 static inline int inode_trylock(struct inode *inode)
780 return down_write_trylock(&inode->i_rwsem);
783 static inline int inode_trylock_shared(struct inode *inode)
785 return down_read_trylock(&inode->i_rwsem);
788 static inline int inode_is_locked(struct inode *inode)
790 return rwsem_is_locked(&inode->i_rwsem);
793 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
795 down_write_nested(&inode->i_rwsem, subclass);
798 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
800 down_read_nested(&inode->i_rwsem, subclass);
803 static inline void filemap_invalidate_lock(struct address_space *mapping)
805 down_write(&mapping->invalidate_lock);
808 static inline void filemap_invalidate_unlock(struct address_space *mapping)
810 up_write(&mapping->invalidate_lock);
813 static inline void filemap_invalidate_lock_shared(struct address_space *mapping)
815 down_read(&mapping->invalidate_lock);
818 static inline int filemap_invalidate_trylock_shared(
819 struct address_space *mapping)
821 return down_read_trylock(&mapping->invalidate_lock);
824 static inline void filemap_invalidate_unlock_shared(
825 struct address_space *mapping)
827 up_read(&mapping->invalidate_lock);
830 void lock_two_nondirectories(struct inode *, struct inode*);
831 void unlock_two_nondirectories(struct inode *, struct inode*);
833 void filemap_invalidate_lock_two(struct address_space *mapping1,
834 struct address_space *mapping2);
835 void filemap_invalidate_unlock_two(struct address_space *mapping1,
836 struct address_space *mapping2);
840 * NOTE: in a 32bit arch with a preemptable kernel and
841 * an UP compile the i_size_read/write must be atomic
842 * with respect to the local cpu (unlike with preempt disabled),
843 * but they don't need to be atomic with respect to other cpus like in
844 * true SMP (so they need either to either locally disable irq around
845 * the read or for example on x86 they can be still implemented as a
846 * cmpxchg8b without the need of the lock prefix). For SMP compiles
847 * and 64bit archs it makes no difference if preempt is enabled or not.
849 static inline loff_t i_size_read(const struct inode *inode)
851 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
856 seq = read_seqcount_begin(&inode->i_size_seqcount);
857 i_size = inode->i_size;
858 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
860 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
864 i_size = inode->i_size;
868 return inode->i_size;
873 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
874 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
875 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
877 static inline void i_size_write(struct inode *inode, loff_t i_size)
879 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
881 write_seqcount_begin(&inode->i_size_seqcount);
882 inode->i_size = i_size;
883 write_seqcount_end(&inode->i_size_seqcount);
885 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
887 inode->i_size = i_size;
890 inode->i_size = i_size;
894 static inline unsigned iminor(const struct inode *inode)
896 return MINOR(inode->i_rdev);
899 static inline unsigned imajor(const struct inode *inode)
901 return MAJOR(inode->i_rdev);
905 rwlock_t lock; /* protects pid, uid, euid fields */
906 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
907 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
908 kuid_t uid, euid; /* uid/euid of process setting the owner */
909 int signum; /* posix.1b rt signal to be delivered on IO */
913 * struct file_ra_state - Track a file's readahead state.
914 * @start: Where the most recent readahead started.
915 * @size: Number of pages read in the most recent readahead.
916 * @async_size: Numer of pages that were/are not needed immediately
917 * and so were/are genuinely "ahead". Start next readahead when
918 * the first of these pages is accessed.
919 * @ra_pages: Maximum size of a readahead request, copied from the bdi.
920 * @mmap_miss: How many mmap accesses missed in the page cache.
921 * @prev_pos: The last byte in the most recent read request.
923 * When this structure is passed to ->readahead(), the "most recent"
924 * readahead means the current readahead.
926 struct file_ra_state {
929 unsigned int async_size;
930 unsigned int ra_pages;
931 unsigned int mmap_miss;
936 * Check if @index falls in the readahead windows.
938 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
940 return (index >= ra->start &&
941 index < ra->start + ra->size);
946 struct llist_node f_llist;
947 struct rcu_head f_rcuhead;
948 unsigned int f_iocb_flags;
951 struct inode *f_inode; /* cached value */
952 const struct file_operations *f_op;
955 * Protects f_ep, f_flags.
956 * Must not be taken from IRQ context.
959 atomic_long_t f_count;
960 unsigned int f_flags;
962 struct mutex f_pos_lock;
964 struct fown_struct f_owner;
965 const struct cred *f_cred;
966 struct file_ra_state f_ra;
969 #ifdef CONFIG_SECURITY
972 /* needed for tty driver, and maybe others */
976 /* Used by fs/eventpoll.c to link all the hooks to this file */
977 struct hlist_head *f_ep;
978 #endif /* #ifdef CONFIG_EPOLL */
979 struct address_space *f_mapping;
981 errseq_t f_sb_err; /* for syncfs */
983 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
988 /* file identifier */
989 unsigned char f_handle[];
992 static inline struct file *get_file(struct file *f)
994 atomic_long_inc(&f->f_count);
997 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
998 #define file_count(x) atomic_long_read(&(x)->f_count)
1000 #define MAX_NON_LFS ((1UL<<31) - 1)
1002 /* Page cache limit. The filesystems should put that into their s_maxbytes
1003 limits, otherwise bad things can happen in VM. */
1004 #if BITS_PER_LONG==32
1005 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
1006 #elif BITS_PER_LONG==64
1007 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
1012 #define FL_DELEG 4 /* NFSv4 delegation */
1013 #define FL_ACCESS 8 /* not trying to lock, just looking */
1014 #define FL_EXISTS 16 /* when unlocking, test for existence */
1015 #define FL_LEASE 32 /* lease held on this file */
1016 #define FL_CLOSE 64 /* unlock on close */
1017 #define FL_SLEEP 128 /* A blocking lock */
1018 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
1019 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
1020 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
1021 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
1022 #define FL_RECLAIM 4096 /* reclaiming from a reboot server */
1024 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
1027 * Special return value from posix_lock_file() and vfs_lock_file() for
1028 * asynchronous locking.
1030 #define FILE_LOCK_DEFERRED 1
1032 /* legacy typedef, should eventually be removed */
1033 typedef void *fl_owner_t;
1037 struct file_lock_operations {
1038 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1039 void (*fl_release_private)(struct file_lock *);
1042 struct lock_manager_operations {
1044 fl_owner_t (*lm_get_owner)(fl_owner_t);
1045 void (*lm_put_owner)(fl_owner_t);
1046 void (*lm_notify)(struct file_lock *); /* unblock callback */
1047 int (*lm_grant)(struct file_lock *, int);
1048 bool (*lm_break)(struct file_lock *);
1049 int (*lm_change)(struct file_lock *, int, struct list_head *);
1050 void (*lm_setup)(struct file_lock *, void **);
1051 bool (*lm_breaker_owns_lease)(struct file_lock *);
1052 bool (*lm_lock_expirable)(struct file_lock *cfl);
1053 void (*lm_expire_lock)(void);
1056 struct lock_manager {
1057 struct list_head list;
1059 * NFSv4 and up also want opens blocked during the grace period;
1066 void locks_start_grace(struct net *, struct lock_manager *);
1067 void locks_end_grace(struct lock_manager *);
1068 bool locks_in_grace(struct net *);
1069 bool opens_in_grace(struct net *);
1071 /* that will die - we need it for nfs_lock_info */
1072 #include <linux/nfs_fs_i.h>
1075 * struct file_lock represents a generic "file lock". It's used to represent
1076 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1077 * note that the same struct is used to represent both a request for a lock and
1078 * the lock itself, but the same object is never used for both.
1080 * FIXME: should we create a separate "struct lock_request" to help distinguish
1083 * The varous i_flctx lists are ordered by:
1086 * 2) lock range start
1089 * Obviously, the last two criteria only matter for POSIX locks.
1092 struct file_lock *fl_blocker; /* The lock, that is blocking us */
1093 struct list_head fl_list; /* link into file_lock_context */
1094 struct hlist_node fl_link; /* node in global lists */
1095 struct list_head fl_blocked_requests; /* list of requests with
1096 * ->fl_blocker pointing here
1098 struct list_head fl_blocked_member; /* node in
1099 * ->fl_blocker->fl_blocked_requests
1101 fl_owner_t fl_owner;
1102 unsigned int fl_flags;
1103 unsigned char fl_type;
1104 unsigned int fl_pid;
1105 int fl_link_cpu; /* what cpu's list is this on? */
1106 wait_queue_head_t fl_wait;
1107 struct file *fl_file;
1111 struct fasync_struct * fl_fasync; /* for lease break notifications */
1112 /* for lease breaks: */
1113 unsigned long fl_break_time;
1114 unsigned long fl_downgrade_time;
1116 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1117 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1119 struct nfs_lock_info nfs_fl;
1120 struct nfs4_lock_info nfs4_fl;
1122 struct list_head link; /* link in AFS vnode's pending_locks list */
1123 int state; /* state of grant or error if -ve */
1124 unsigned int debug_id;
1127 } __randomize_layout;
1129 struct file_lock_context {
1130 spinlock_t flc_lock;
1131 struct list_head flc_flock;
1132 struct list_head flc_posix;
1133 struct list_head flc_lease;
1136 /* The following constant reflects the upper bound of the file/locking space */
1138 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1139 #define OFFSET_MAX INT_LIMIT(loff_t)
1140 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1143 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1145 #define locks_inode(f) file_inode(f)
1147 #ifdef CONFIG_FILE_LOCKING
1148 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1149 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1152 #if BITS_PER_LONG == 32
1153 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1154 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1158 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1159 extern int fcntl_getlease(struct file *filp);
1162 void locks_free_lock_context(struct inode *inode);
1163 void locks_free_lock(struct file_lock *fl);
1164 extern void locks_init_lock(struct file_lock *);
1165 extern struct file_lock * locks_alloc_lock(void);
1166 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1167 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1168 extern void locks_remove_posix(struct file *, fl_owner_t);
1169 extern void locks_remove_file(struct file *);
1170 extern void locks_release_private(struct file_lock *);
1171 extern void posix_test_lock(struct file *, struct file_lock *);
1172 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1173 extern int locks_delete_block(struct file_lock *);
1174 extern int vfs_test_lock(struct file *, struct file_lock *);
1175 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1176 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1177 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1178 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1179 extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1180 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1181 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1182 extern int lease_modify(struct file_lock *, int, struct list_head *);
1184 struct notifier_block;
1185 extern int lease_register_notifier(struct notifier_block *);
1186 extern void lease_unregister_notifier(struct notifier_block *);
1188 struct files_struct;
1189 extern void show_fd_locks(struct seq_file *f,
1190 struct file *filp, struct files_struct *files);
1191 extern bool locks_owner_has_blockers(struct file_lock_context *flctx,
1193 #else /* !CONFIG_FILE_LOCKING */
1194 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1195 struct flock __user *user)
1200 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1201 unsigned int cmd, struct flock __user *user)
1206 #if BITS_PER_LONG == 32
1207 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1208 struct flock64 *user)
1213 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1214 unsigned int cmd, struct flock64 *user)
1219 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1224 static inline int fcntl_getlease(struct file *filp)
1230 locks_free_lock_context(struct inode *inode)
1234 static inline void locks_init_lock(struct file_lock *fl)
1239 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1244 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1249 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1254 static inline void locks_remove_file(struct file *filp)
1259 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1264 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1265 struct file_lock *conflock)
1270 static inline int locks_delete_block(struct file_lock *waiter)
1275 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1280 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1281 struct file_lock *fl, struct file_lock *conf)
1286 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1291 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1296 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1301 static inline void lease_get_mtime(struct inode *inode,
1302 struct timespec64 *time)
1307 static inline int generic_setlease(struct file *filp, long arg,
1308 struct file_lock **flp, void **priv)
1313 static inline int vfs_setlease(struct file *filp, long arg,
1314 struct file_lock **lease, void **priv)
1319 static inline int lease_modify(struct file_lock *fl, int arg,
1320 struct list_head *dispose)
1325 struct files_struct;
1326 static inline void show_fd_locks(struct seq_file *f,
1327 struct file *filp, struct files_struct *files) {}
1328 static inline bool locks_owner_has_blockers(struct file_lock_context *flctx,
1333 #endif /* !CONFIG_FILE_LOCKING */
1335 static inline struct inode *file_inode(const struct file *f)
1340 static inline struct dentry *file_dentry(const struct file *file)
1342 return d_real(file->f_path.dentry, file_inode(file));
1345 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1347 return locks_lock_inode_wait(locks_inode(filp), fl);
1350 struct fasync_struct {
1354 struct fasync_struct *fa_next; /* singly linked list */
1355 struct file *fa_file;
1356 struct rcu_head fa_rcu;
1359 #define FASYNC_MAGIC 0x4601
1361 /* SMP safe fasync helpers: */
1362 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1363 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1364 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1365 extern struct fasync_struct *fasync_alloc(void);
1366 extern void fasync_free(struct fasync_struct *);
1368 /* can be called from interrupts */
1369 extern void kill_fasync(struct fasync_struct **, int, int);
1371 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1372 extern int f_setown(struct file *filp, unsigned long arg, int force);
1373 extern void f_delown(struct file *filp);
1374 extern pid_t f_getown(struct file *filp);
1375 extern int send_sigurg(struct fown_struct *fown);
1378 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1379 * represented in both.
1381 #define SB_RDONLY 1 /* Mount read-only */
1382 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1383 #define SB_NODEV 4 /* Disallow access to device special files */
1384 #define SB_NOEXEC 8 /* Disallow program execution */
1385 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1386 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1387 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1388 #define SB_NOATIME 1024 /* Do not update access times. */
1389 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1390 #define SB_SILENT 32768
1391 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1392 #define SB_INLINECRYPT (1<<17) /* Use blk-crypto for encrypted files */
1393 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1394 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1395 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1397 /* These sb flags are internal to the kernel */
1398 #define SB_SUBMOUNT (1<<26)
1399 #define SB_FORCE (1<<27)
1400 #define SB_NOSEC (1<<28)
1401 #define SB_BORN (1<<29)
1402 #define SB_ACTIVE (1<<30)
1403 #define SB_NOUSER (1<<31)
1405 /* These flags relate to encoding and casefolding */
1406 #define SB_ENC_STRICT_MODE_FL (1 << 0)
1408 #define sb_has_strict_encoding(sb) \
1409 (sb->s_encoding_flags & SB_ENC_STRICT_MODE_FL)
1415 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1416 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1417 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1418 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1419 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1422 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1423 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1424 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1425 #define SB_I_STABLE_WRITES 0x00000008 /* don't modify blks until WB is done */
1427 /* sb->s_iflags to limit user namespace mounts */
1428 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1429 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1430 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1432 #define SB_I_SKIP_SYNC 0x00000100 /* Skip superblock at global sync */
1433 #define SB_I_PERSB_BDI 0x00000200 /* has a per-sb bdi */
1434 #define SB_I_TS_EXPIRY_WARNED 0x00000400 /* warned about timestamp range expiry */
1436 /* Possible states of 'frozen' field */
1438 SB_UNFROZEN = 0, /* FS is unfrozen */
1439 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1440 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1441 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1442 * internal threads if needed) */
1443 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1446 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1449 int frozen; /* Is sb frozen? */
1450 wait_queue_head_t wait_unfrozen; /* wait for thaw */
1451 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1454 struct super_block {
1455 struct list_head s_list; /* Keep this first */
1456 dev_t s_dev; /* search index; _not_ kdev_t */
1457 unsigned char s_blocksize_bits;
1458 unsigned long s_blocksize;
1459 loff_t s_maxbytes; /* Max file size */
1460 struct file_system_type *s_type;
1461 const struct super_operations *s_op;
1462 const struct dquot_operations *dq_op;
1463 const struct quotactl_ops *s_qcop;
1464 const struct export_operations *s_export_op;
1465 unsigned long s_flags;
1466 unsigned long s_iflags; /* internal SB_I_* flags */
1467 unsigned long s_magic;
1468 struct dentry *s_root;
1469 struct rw_semaphore s_umount;
1472 #ifdef CONFIG_SECURITY
1475 const struct xattr_handler **s_xattr;
1476 #ifdef CONFIG_FS_ENCRYPTION
1477 const struct fscrypt_operations *s_cop;
1478 struct key *s_master_keys; /* master crypto keys in use */
1480 #ifdef CONFIG_FS_VERITY
1481 const struct fsverity_operations *s_vop;
1483 #if IS_ENABLED(CONFIG_UNICODE)
1484 struct unicode_map *s_encoding;
1485 __u16 s_encoding_flags;
1487 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1488 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1489 struct block_device *s_bdev;
1490 struct backing_dev_info *s_bdi;
1491 struct mtd_info *s_mtd;
1492 struct hlist_node s_instances;
1493 unsigned int s_quota_types; /* Bitmask of supported quota types */
1494 struct quota_info s_dquot; /* Diskquota specific options */
1496 struct sb_writers s_writers;
1499 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1500 * s_fsnotify_marks together for cache efficiency. They are frequently
1501 * accessed and rarely modified.
1503 void *s_fs_info; /* Filesystem private info */
1505 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1507 /* Time limits for c/m/atime in seconds */
1508 time64_t s_time_min;
1509 time64_t s_time_max;
1510 #ifdef CONFIG_FSNOTIFY
1511 __u32 s_fsnotify_mask;
1512 struct fsnotify_mark_connector __rcu *s_fsnotify_marks;
1515 char s_id[32]; /* Informational name */
1516 uuid_t s_uuid; /* UUID */
1518 unsigned int s_max_links;
1522 * The next field is for VFS *only*. No filesystems have any business
1523 * even looking at it. You had been warned.
1525 struct mutex s_vfs_rename_mutex; /* Kludge */
1528 * Filesystem subtype. If non-empty the filesystem type field
1529 * in /proc/mounts will be "type.subtype"
1531 const char *s_subtype;
1533 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1535 struct shrinker s_shrink; /* per-sb shrinker handle */
1537 /* Number of inodes with nlink == 0 but still referenced */
1538 atomic_long_t s_remove_count;
1541 * Number of inode/mount/sb objects that are being watched, note that
1542 * inodes objects are currently double-accounted.
1544 atomic_long_t s_fsnotify_connectors;
1546 /* Being remounted read-only */
1547 int s_readonly_remount;
1549 /* per-sb errseq_t for reporting writeback errors via syncfs */
1552 /* AIO completions deferred from interrupt context */
1553 struct workqueue_struct *s_dio_done_wq;
1554 struct hlist_head s_pins;
1557 * Owning user namespace and default context in which to
1558 * interpret filesystem uids, gids, quotas, device nodes,
1559 * xattrs and security labels.
1561 struct user_namespace *s_user_ns;
1564 * The list_lru structure is essentially just a pointer to a table
1565 * of per-node lru lists, each of which has its own spinlock.
1566 * There is no need to put them into separate cachelines.
1568 struct list_lru s_dentry_lru;
1569 struct list_lru s_inode_lru;
1570 struct rcu_head rcu;
1571 struct work_struct destroy_work;
1573 struct mutex s_sync_lock; /* sync serialisation lock */
1576 * Indicates how deep in a filesystem stack this SB is
1580 /* s_inode_list_lock protects s_inodes */
1581 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1582 struct list_head s_inodes; /* all inodes */
1584 spinlock_t s_inode_wblist_lock;
1585 struct list_head s_inodes_wb; /* writeback inodes */
1586 } __randomize_layout;
1588 static inline struct user_namespace *i_user_ns(const struct inode *inode)
1590 return inode->i_sb->s_user_ns;
1593 /* Helper functions so that in most cases filesystems will
1594 * not need to deal directly with kuid_t and kgid_t and can
1595 * instead deal with the raw numeric values that are stored
1596 * in the filesystem.
1598 static inline uid_t i_uid_read(const struct inode *inode)
1600 return from_kuid(i_user_ns(inode), inode->i_uid);
1603 static inline gid_t i_gid_read(const struct inode *inode)
1605 return from_kgid(i_user_ns(inode), inode->i_gid);
1608 static inline void i_uid_write(struct inode *inode, uid_t uid)
1610 inode->i_uid = make_kuid(i_user_ns(inode), uid);
1613 static inline void i_gid_write(struct inode *inode, gid_t gid)
1615 inode->i_gid = make_kgid(i_user_ns(inode), gid);
1619 * i_uid_into_mnt - map an inode's i_uid down into a mnt_userns
1620 * @mnt_userns: user namespace of the mount the inode was found from
1621 * @inode: inode to map
1623 * Note, this will eventually be removed completely in favor of the type-safe
1624 * i_uid_into_vfsuid().
1626 * Return: the inode's i_uid mapped down according to @mnt_userns.
1627 * If the inode's i_uid has no mapping INVALID_UID is returned.
1629 static inline kuid_t i_uid_into_mnt(struct user_namespace *mnt_userns,
1630 const struct inode *inode)
1632 return AS_KUIDT(make_vfsuid(mnt_userns, i_user_ns(inode), inode->i_uid));
1636 * i_uid_into_vfsuid - map an inode's i_uid down into a mnt_userns
1637 * @mnt_userns: user namespace of the mount the inode was found from
1638 * @inode: inode to map
1640 * Return: whe inode's i_uid mapped down according to @mnt_userns.
1641 * If the inode's i_uid has no mapping INVALID_VFSUID is returned.
1643 static inline vfsuid_t i_uid_into_vfsuid(struct user_namespace *mnt_userns,
1644 const struct inode *inode)
1646 return make_vfsuid(mnt_userns, i_user_ns(inode), inode->i_uid);
1650 * i_uid_needs_update - check whether inode's i_uid needs to be updated
1651 * @mnt_userns: user namespace of the mount the inode was found from
1652 * @attr: the new attributes of @inode
1653 * @inode: the inode to update
1655 * Check whether the $inode's i_uid field needs to be updated taking idmapped
1656 * mounts into account if the filesystem supports it.
1658 * Return: true if @inode's i_uid field needs to be updated, false if not.
1660 static inline bool i_uid_needs_update(struct user_namespace *mnt_userns,
1661 const struct iattr *attr,
1662 const struct inode *inode)
1664 return ((attr->ia_valid & ATTR_UID) &&
1665 !vfsuid_eq(attr->ia_vfsuid,
1666 i_uid_into_vfsuid(mnt_userns, inode)));
1670 * i_uid_update - update @inode's i_uid field
1671 * @mnt_userns: user namespace of the mount the inode was found from
1672 * @attr: the new attributes of @inode
1673 * @inode: the inode to update
1675 * Safely update @inode's i_uid field translating the vfsuid of any idmapped
1676 * mount into the filesystem kuid.
1678 static inline void i_uid_update(struct user_namespace *mnt_userns,
1679 const struct iattr *attr,
1680 struct inode *inode)
1682 if (attr->ia_valid & ATTR_UID)
1683 inode->i_uid = from_vfsuid(mnt_userns, i_user_ns(inode),
1688 * i_gid_into_mnt - map an inode's i_gid down into a mnt_userns
1689 * @mnt_userns: user namespace of the mount the inode was found from
1690 * @inode: inode to map
1692 * Note, this will eventually be removed completely in favor of the type-safe
1693 * i_gid_into_vfsgid().
1695 * Return: the inode's i_gid mapped down according to @mnt_userns.
1696 * If the inode's i_gid has no mapping INVALID_GID is returned.
1698 static inline kgid_t i_gid_into_mnt(struct user_namespace *mnt_userns,
1699 const struct inode *inode)
1701 return AS_KGIDT(make_vfsgid(mnt_userns, i_user_ns(inode), inode->i_gid));
1705 * i_gid_into_vfsgid - map an inode's i_gid down into a mnt_userns
1706 * @mnt_userns: user namespace of the mount the inode was found from
1707 * @inode: inode to map
1709 * Return: the inode's i_gid mapped down according to @mnt_userns.
1710 * If the inode's i_gid has no mapping INVALID_VFSGID is returned.
1712 static inline vfsgid_t i_gid_into_vfsgid(struct user_namespace *mnt_userns,
1713 const struct inode *inode)
1715 return make_vfsgid(mnt_userns, i_user_ns(inode), inode->i_gid);
1719 * i_gid_needs_update - check whether inode's i_gid needs to be updated
1720 * @mnt_userns: user namespace of the mount the inode was found from
1721 * @attr: the new attributes of @inode
1722 * @inode: the inode to update
1724 * Check whether the $inode's i_gid field needs to be updated taking idmapped
1725 * mounts into account if the filesystem supports it.
1727 * Return: true if @inode's i_gid field needs to be updated, false if not.
1729 static inline bool i_gid_needs_update(struct user_namespace *mnt_userns,
1730 const struct iattr *attr,
1731 const struct inode *inode)
1733 return ((attr->ia_valid & ATTR_GID) &&
1734 !vfsgid_eq(attr->ia_vfsgid,
1735 i_gid_into_vfsgid(mnt_userns, inode)));
1739 * i_gid_update - update @inode's i_gid field
1740 * @mnt_userns: user namespace of the mount the inode was found from
1741 * @attr: the new attributes of @inode
1742 * @inode: the inode to update
1744 * Safely update @inode's i_gid field translating the vfsgid of any idmapped
1745 * mount into the filesystem kgid.
1747 static inline void i_gid_update(struct user_namespace *mnt_userns,
1748 const struct iattr *attr,
1749 struct inode *inode)
1751 if (attr->ia_valid & ATTR_GID)
1752 inode->i_gid = from_vfsgid(mnt_userns, i_user_ns(inode),
1757 * inode_fsuid_set - initialize inode's i_uid field with callers fsuid
1758 * @inode: inode to initialize
1759 * @mnt_userns: user namespace of the mount the inode was found from
1761 * Initialize the i_uid field of @inode. If the inode was found/created via
1762 * an idmapped mount map the caller's fsuid according to @mnt_users.
1764 static inline void inode_fsuid_set(struct inode *inode,
1765 struct user_namespace *mnt_userns)
1767 inode->i_uid = mapped_fsuid(mnt_userns, i_user_ns(inode));
1771 * inode_fsgid_set - initialize inode's i_gid field with callers fsgid
1772 * @inode: inode to initialize
1773 * @mnt_userns: user namespace of the mount the inode was found from
1775 * Initialize the i_gid field of @inode. If the inode was found/created via
1776 * an idmapped mount map the caller's fsgid according to @mnt_users.
1778 static inline void inode_fsgid_set(struct inode *inode,
1779 struct user_namespace *mnt_userns)
1781 inode->i_gid = mapped_fsgid(mnt_userns, i_user_ns(inode));
1785 * fsuidgid_has_mapping() - check whether caller's fsuid/fsgid is mapped
1786 * @sb: the superblock we want a mapping in
1787 * @mnt_userns: user namespace of the relevant mount
1789 * Check whether the caller's fsuid and fsgid have a valid mapping in the
1790 * s_user_ns of the superblock @sb. If the caller is on an idmapped mount map
1791 * the caller's fsuid and fsgid according to the @mnt_userns first.
1793 * Return: true if fsuid and fsgid is mapped, false if not.
1795 static inline bool fsuidgid_has_mapping(struct super_block *sb,
1796 struct user_namespace *mnt_userns)
1798 struct user_namespace *fs_userns = sb->s_user_ns;
1802 kuid = mapped_fsuid(mnt_userns, fs_userns);
1803 if (!uid_valid(kuid))
1805 kgid = mapped_fsgid(mnt_userns, fs_userns);
1806 if (!gid_valid(kgid))
1808 return kuid_has_mapping(fs_userns, kuid) &&
1809 kgid_has_mapping(fs_userns, kgid);
1812 extern struct timespec64 current_time(struct inode *inode);
1815 * Snapshotting support.
1819 * These are internal functions, please use sb_start_{write,pagefault,intwrite}
1822 static inline void __sb_end_write(struct super_block *sb, int level)
1824 percpu_up_read(sb->s_writers.rw_sem + level-1);
1827 static inline void __sb_start_write(struct super_block *sb, int level)
1829 percpu_down_read(sb->s_writers.rw_sem + level - 1);
1832 static inline bool __sb_start_write_trylock(struct super_block *sb, int level)
1834 return percpu_down_read_trylock(sb->s_writers.rw_sem + level - 1);
1837 #define __sb_writers_acquired(sb, lev) \
1838 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1839 #define __sb_writers_release(sb, lev) \
1840 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1842 static inline bool sb_write_started(const struct super_block *sb)
1844 return lockdep_is_held_type(sb->s_writers.rw_sem + SB_FREEZE_WRITE - 1, 1);
1848 * sb_end_write - drop write access to a superblock
1849 * @sb: the super we wrote to
1851 * Decrement number of writers to the filesystem. Wake up possible waiters
1852 * wanting to freeze the filesystem.
1854 static inline void sb_end_write(struct super_block *sb)
1856 __sb_end_write(sb, SB_FREEZE_WRITE);
1860 * sb_end_pagefault - drop write access to a superblock from a page fault
1861 * @sb: the super we wrote to
1863 * Decrement number of processes handling write page fault to the filesystem.
1864 * Wake up possible waiters wanting to freeze the filesystem.
1866 static inline void sb_end_pagefault(struct super_block *sb)
1868 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1872 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1873 * @sb: the super we wrote to
1875 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1876 * waiters wanting to freeze the filesystem.
1878 static inline void sb_end_intwrite(struct super_block *sb)
1880 __sb_end_write(sb, SB_FREEZE_FS);
1884 * sb_start_write - get write access to a superblock
1885 * @sb: the super we write to
1887 * When a process wants to write data or metadata to a file system (i.e. dirty
1888 * a page or an inode), it should embed the operation in a sb_start_write() -
1889 * sb_end_write() pair to get exclusion against file system freezing. This
1890 * function increments number of writers preventing freezing. If the file
1891 * system is already frozen, the function waits until the file system is
1894 * Since freeze protection behaves as a lock, users have to preserve
1895 * ordering of freeze protection and other filesystem locks. Generally,
1896 * freeze protection should be the outermost lock. In particular, we have:
1899 * -> i_mutex (write path, truncate, directory ops, ...)
1900 * -> s_umount (freeze_super, thaw_super)
1902 static inline void sb_start_write(struct super_block *sb)
1904 __sb_start_write(sb, SB_FREEZE_WRITE);
1907 static inline bool sb_start_write_trylock(struct super_block *sb)
1909 return __sb_start_write_trylock(sb, SB_FREEZE_WRITE);
1913 * sb_start_pagefault - get write access to a superblock from a page fault
1914 * @sb: the super we write to
1916 * When a process starts handling write page fault, it should embed the
1917 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1918 * exclusion against file system freezing. This is needed since the page fault
1919 * is going to dirty a page. This function increments number of running page
1920 * faults preventing freezing. If the file system is already frozen, the
1921 * function waits until the file system is thawed.
1923 * Since page fault freeze protection behaves as a lock, users have to preserve
1924 * ordering of freeze protection and other filesystem locks. It is advised to
1925 * put sb_start_pagefault() close to mmap_lock in lock ordering. Page fault
1926 * handling code implies lock dependency:
1929 * -> sb_start_pagefault
1931 static inline void sb_start_pagefault(struct super_block *sb)
1933 __sb_start_write(sb, SB_FREEZE_PAGEFAULT);
1937 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1938 * @sb: the super we write to
1940 * This is the third level of protection against filesystem freezing. It is
1941 * free for use by a filesystem. The only requirement is that it must rank
1942 * below sb_start_pagefault.
1944 * For example filesystem can call sb_start_intwrite() when starting a
1945 * transaction which somewhat eases handling of freezing for internal sources
1946 * of filesystem changes (internal fs threads, discarding preallocation on file
1949 static inline void sb_start_intwrite(struct super_block *sb)
1951 __sb_start_write(sb, SB_FREEZE_FS);
1954 static inline bool sb_start_intwrite_trylock(struct super_block *sb)
1956 return __sb_start_write_trylock(sb, SB_FREEZE_FS);
1959 bool inode_owner_or_capable(struct user_namespace *mnt_userns,
1960 const struct inode *inode);
1963 * VFS helper functions..
1965 int vfs_create(struct user_namespace *, struct inode *,
1966 struct dentry *, umode_t, bool);
1967 int vfs_mkdir(struct user_namespace *, struct inode *,
1968 struct dentry *, umode_t);
1969 int vfs_mknod(struct user_namespace *, struct inode *, struct dentry *,
1971 int vfs_symlink(struct user_namespace *, struct inode *,
1972 struct dentry *, const char *);
1973 int vfs_link(struct dentry *, struct user_namespace *, struct inode *,
1974 struct dentry *, struct inode **);
1975 int vfs_rmdir(struct user_namespace *, struct inode *, struct dentry *);
1976 int vfs_unlink(struct user_namespace *, struct inode *, struct dentry *,
1980 * struct renamedata - contains all information required for renaming
1981 * @old_mnt_userns: old user namespace of the mount the inode was found from
1982 * @old_dir: parent of source
1983 * @old_dentry: source
1984 * @new_mnt_userns: new user namespace of the mount the inode was found from
1985 * @new_dir: parent of destination
1986 * @new_dentry: destination
1987 * @delegated_inode: returns an inode needing a delegation break
1988 * @flags: rename flags
1991 struct user_namespace *old_mnt_userns;
1992 struct inode *old_dir;
1993 struct dentry *old_dentry;
1994 struct user_namespace *new_mnt_userns;
1995 struct inode *new_dir;
1996 struct dentry *new_dentry;
1997 struct inode **delegated_inode;
1999 } __randomize_layout;
2001 int vfs_rename(struct renamedata *);
2003 static inline int vfs_whiteout(struct user_namespace *mnt_userns,
2004 struct inode *dir, struct dentry *dentry)
2006 return vfs_mknod(mnt_userns, dir, dentry, S_IFCHR | WHITEOUT_MODE,
2010 struct dentry *vfs_tmpfile(struct user_namespace *mnt_userns,
2011 struct dentry *dentry, umode_t mode, int open_flag);
2013 int vfs_mkobj(struct dentry *, umode_t,
2014 int (*f)(struct dentry *, umode_t, void *),
2017 int vfs_fchown(struct file *file, uid_t user, gid_t group);
2018 int vfs_fchmod(struct file *file, umode_t mode);
2019 int vfs_utimes(const struct path *path, struct timespec64 *times);
2021 extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
2023 #ifdef CONFIG_COMPAT
2024 extern long compat_ptr_ioctl(struct file *file, unsigned int cmd,
2027 #define compat_ptr_ioctl NULL
2031 * VFS file helper functions.
2033 void inode_init_owner(struct user_namespace *mnt_userns, struct inode *inode,
2034 const struct inode *dir, umode_t mode);
2035 extern bool may_open_dev(const struct path *path);
2038 * This is the "filldir" function type, used by readdir() to let
2039 * the kernel specify what kind of dirent layout it wants to have.
2040 * This allows the kernel to read directories into kernel space or
2041 * to have different dirent layouts depending on the binary type.
2044 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
2047 struct dir_context {
2053 * These flags let !MMU mmap() govern direct device mapping vs immediate
2054 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
2056 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
2057 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
2058 * NOMMU_MAP_READ: Can be mapped for reading
2059 * NOMMU_MAP_WRITE: Can be mapped for writing
2060 * NOMMU_MAP_EXEC: Can be mapped for execution
2062 #define NOMMU_MAP_COPY 0x00000001
2063 #define NOMMU_MAP_DIRECT 0x00000008
2064 #define NOMMU_MAP_READ VM_MAYREAD
2065 #define NOMMU_MAP_WRITE VM_MAYWRITE
2066 #define NOMMU_MAP_EXEC VM_MAYEXEC
2068 #define NOMMU_VMFLAGS \
2069 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
2072 * These flags control the behavior of the remap_file_range function pointer.
2073 * If it is called with len == 0 that means "remap to end of source file".
2074 * See Documentation/filesystems/vfs.rst for more details about this call.
2076 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
2077 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
2079 #define REMAP_FILE_DEDUP (1 << 0)
2080 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
2083 * These flags signal that the caller is ok with altering various aspects of
2084 * the behavior of the remap operation. The changes must be made by the
2085 * implementation; the vfs remap helper functions can take advantage of them.
2086 * Flags in this category exist to preserve the quirky behavior of the hoisted
2087 * btrfs clone/dedupe ioctls.
2089 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
2092 struct io_uring_cmd;
2094 struct file_operations {
2095 struct module *owner;
2096 loff_t (*llseek) (struct file *, loff_t, int);
2097 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
2098 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
2099 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
2100 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
2101 int (*iopoll)(struct kiocb *kiocb, struct io_comp_batch *,
2102 unsigned int flags);
2103 int (*iterate) (struct file *, struct dir_context *);
2104 int (*iterate_shared) (struct file *, struct dir_context *);
2105 __poll_t (*poll) (struct file *, struct poll_table_struct *);
2106 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
2107 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
2108 int (*mmap) (struct file *, struct vm_area_struct *);
2109 unsigned long mmap_supported_flags;
2110 int (*open) (struct inode *, struct file *);
2111 int (*flush) (struct file *, fl_owner_t id);
2112 int (*release) (struct inode *, struct file *);
2113 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
2114 int (*fasync) (int, struct file *, int);
2115 int (*lock) (struct file *, int, struct file_lock *);
2116 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
2117 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
2118 int (*check_flags)(int);
2119 int (*flock) (struct file *, int, struct file_lock *);
2120 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
2121 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
2122 int (*setlease)(struct file *, long, struct file_lock **, void **);
2123 long (*fallocate)(struct file *file, int mode, loff_t offset,
2125 void (*show_fdinfo)(struct seq_file *m, struct file *f);
2127 unsigned (*mmap_capabilities)(struct file *);
2129 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
2130 loff_t, size_t, unsigned int);
2131 loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
2132 struct file *file_out, loff_t pos_out,
2133 loff_t len, unsigned int remap_flags);
2134 int (*fadvise)(struct file *, loff_t, loff_t, int);
2135 int (*uring_cmd)(struct io_uring_cmd *ioucmd, unsigned int issue_flags);
2136 } __randomize_layout;
2138 struct inode_operations {
2139 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
2140 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
2141 int (*permission) (struct user_namespace *, struct inode *, int);
2142 struct posix_acl * (*get_acl)(struct inode *, int, bool);
2144 int (*readlink) (struct dentry *, char __user *,int);
2146 int (*create) (struct user_namespace *, struct inode *,struct dentry *,
2148 int (*link) (struct dentry *,struct inode *,struct dentry *);
2149 int (*unlink) (struct inode *,struct dentry *);
2150 int (*symlink) (struct user_namespace *, struct inode *,struct dentry *,
2152 int (*mkdir) (struct user_namespace *, struct inode *,struct dentry *,
2154 int (*rmdir) (struct inode *,struct dentry *);
2155 int (*mknod) (struct user_namespace *, struct inode *,struct dentry *,
2157 int (*rename) (struct user_namespace *, struct inode *, struct dentry *,
2158 struct inode *, struct dentry *, unsigned int);
2159 int (*setattr) (struct user_namespace *, struct dentry *,
2161 int (*getattr) (struct user_namespace *, const struct path *,
2162 struct kstat *, u32, unsigned int);
2163 ssize_t (*listxattr) (struct dentry *, char *, size_t);
2164 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
2166 int (*update_time)(struct inode *, struct timespec64 *, int);
2167 int (*atomic_open)(struct inode *, struct dentry *,
2168 struct file *, unsigned open_flag,
2169 umode_t create_mode);
2170 int (*tmpfile) (struct user_namespace *, struct inode *,
2171 struct dentry *, umode_t);
2172 int (*set_acl)(struct user_namespace *, struct inode *,
2173 struct posix_acl *, int);
2174 int (*fileattr_set)(struct user_namespace *mnt_userns,
2175 struct dentry *dentry, struct fileattr *fa);
2176 int (*fileattr_get)(struct dentry *dentry, struct fileattr *fa);
2177 } ____cacheline_aligned;
2179 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
2180 struct iov_iter *iter)
2182 return file->f_op->read_iter(kio, iter);
2185 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
2186 struct iov_iter *iter)
2188 return file->f_op->write_iter(kio, iter);
2191 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
2193 return file->f_op->mmap(file, vma);
2196 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
2197 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
2198 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
2199 loff_t, size_t, unsigned int);
2200 extern ssize_t generic_copy_file_range(struct file *file_in, loff_t pos_in,
2201 struct file *file_out, loff_t pos_out,
2202 size_t len, unsigned int flags);
2203 extern int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
2204 struct file *file_out, loff_t pos_out,
2206 unsigned int remap_flags);
2207 extern loff_t do_clone_file_range(struct file *file_in, loff_t pos_in,
2208 struct file *file_out, loff_t pos_out,
2209 loff_t len, unsigned int remap_flags);
2210 extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in,
2211 struct file *file_out, loff_t pos_out,
2212 loff_t len, unsigned int remap_flags);
2213 extern int vfs_dedupe_file_range(struct file *file,
2214 struct file_dedupe_range *same);
2215 extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
2216 struct file *dst_file, loff_t dst_pos,
2217 loff_t len, unsigned int remap_flags);
2220 struct super_operations {
2221 struct inode *(*alloc_inode)(struct super_block *sb);
2222 void (*destroy_inode)(struct inode *);
2223 void (*free_inode)(struct inode *);
2225 void (*dirty_inode) (struct inode *, int flags);
2226 int (*write_inode) (struct inode *, struct writeback_control *wbc);
2227 int (*drop_inode) (struct inode *);
2228 void (*evict_inode) (struct inode *);
2229 void (*put_super) (struct super_block *);
2230 int (*sync_fs)(struct super_block *sb, int wait);
2231 int (*freeze_super) (struct super_block *);
2232 int (*freeze_fs) (struct super_block *);
2233 int (*thaw_super) (struct super_block *);
2234 int (*unfreeze_fs) (struct super_block *);
2235 int (*statfs) (struct dentry *, struct kstatfs *);
2236 int (*remount_fs) (struct super_block *, int *, char *);
2237 void (*umount_begin) (struct super_block *);
2239 int (*show_options)(struct seq_file *, struct dentry *);
2240 int (*show_devname)(struct seq_file *, struct dentry *);
2241 int (*show_path)(struct seq_file *, struct dentry *);
2242 int (*show_stats)(struct seq_file *, struct dentry *);
2244 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
2245 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
2246 struct dquot **(*get_dquots)(struct inode *);
2248 long (*nr_cached_objects)(struct super_block *,
2249 struct shrink_control *);
2250 long (*free_cached_objects)(struct super_block *,
2251 struct shrink_control *);
2255 * Inode flags - they have no relation to superblock flags now
2257 #define S_SYNC (1 << 0) /* Writes are synced at once */
2258 #define S_NOATIME (1 << 1) /* Do not update access times */
2259 #define S_APPEND (1 << 2) /* Append-only file */
2260 #define S_IMMUTABLE (1 << 3) /* Immutable file */
2261 #define S_DEAD (1 << 4) /* removed, but still open directory */
2262 #define S_NOQUOTA (1 << 5) /* Inode is not counted to quota */
2263 #define S_DIRSYNC (1 << 6) /* Directory modifications are synchronous */
2264 #define S_NOCMTIME (1 << 7) /* Do not update file c/mtime */
2265 #define S_SWAPFILE (1 << 8) /* Do not truncate: swapon got its bmaps */
2266 #define S_PRIVATE (1 << 9) /* Inode is fs-internal */
2267 #define S_IMA (1 << 10) /* Inode has an associated IMA struct */
2268 #define S_AUTOMOUNT (1 << 11) /* Automount/referral quasi-directory */
2269 #define S_NOSEC (1 << 12) /* no suid or xattr security attributes */
2270 #ifdef CONFIG_FS_DAX
2271 #define S_DAX (1 << 13) /* Direct Access, avoiding the page cache */
2273 #define S_DAX 0 /* Make all the DAX code disappear */
2275 #define S_ENCRYPTED (1 << 14) /* Encrypted file (using fs/crypto/) */
2276 #define S_CASEFOLD (1 << 15) /* Casefolded file */
2277 #define S_VERITY (1 << 16) /* Verity file (using fs/verity/) */
2278 #define S_KERNEL_FILE (1 << 17) /* File is in use by the kernel (eg. fs/cachefiles) */
2281 * Note that nosuid etc flags are inode-specific: setting some file-system
2282 * flags just means all the inodes inherit those flags by default. It might be
2283 * possible to override it selectively if you really wanted to with some
2284 * ioctl() that is not currently implemented.
2286 * Exception: SB_RDONLY is always applied to the entire file system.
2288 * Unfortunately, it is possible to change a filesystems flags with it mounted
2289 * with files in use. This means that all of the inodes will not have their
2290 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
2291 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
2293 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
2295 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
2296 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
2297 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
2298 ((inode)->i_flags & S_SYNC))
2299 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
2300 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
2301 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
2302 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
2303 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
2305 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
2306 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
2307 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
2308 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
2310 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
2311 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
2312 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
2313 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
2314 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
2315 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2316 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2317 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2318 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2319 #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD)
2320 #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY)
2322 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2323 (inode)->i_rdev == WHITEOUT_DEV)
2325 static inline bool HAS_UNMAPPED_ID(struct user_namespace *mnt_userns,
2326 struct inode *inode)
2328 return !vfsuid_valid(i_uid_into_vfsuid(mnt_userns, inode)) ||
2329 !vfsgid_valid(i_gid_into_vfsgid(mnt_userns, inode));
2332 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
2334 *kiocb = (struct kiocb) {
2336 .ki_flags = filp->f_iocb_flags,
2337 .ki_ioprio = get_current_ioprio(),
2341 static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src,
2344 *kiocb = (struct kiocb) {
2346 .ki_flags = kiocb_src->ki_flags,
2347 .ki_ioprio = kiocb_src->ki_ioprio,
2348 .ki_pos = kiocb_src->ki_pos,
2353 * Inode state bits. Protected by inode->i_lock
2355 * Four bits determine the dirty state of the inode: I_DIRTY_SYNC,
2356 * I_DIRTY_DATASYNC, I_DIRTY_PAGES, and I_DIRTY_TIME.
2358 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2359 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2360 * various stages of removing an inode.
2362 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2364 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2365 * fdatasync() (unless I_DIRTY_DATASYNC is also set).
2366 * Timestamp updates are the usual cause.
2367 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2368 * these changes separately from I_DIRTY_SYNC so that we
2369 * don't have to write inode on fdatasync() when only
2370 * e.g. the timestamps have changed.
2371 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2372 * I_DIRTY_TIME The inode itself only has dirty timestamps, and the
2373 * lazytime mount option is enabled. We keep track of this
2374 * separately from I_DIRTY_SYNC in order to implement
2375 * lazytime. This gets cleared if I_DIRTY_INODE
2376 * (I_DIRTY_SYNC and/or I_DIRTY_DATASYNC) gets set. I.e.
2377 * either I_DIRTY_TIME *or* I_DIRTY_INODE can be set in
2378 * i_state, but not both. I_DIRTY_PAGES may still be set.
2379 * I_NEW Serves as both a mutex and completion notification.
2380 * New inodes set I_NEW. If two processes both create
2381 * the same inode, one of them will release its inode and
2382 * wait for I_NEW to be released before returning.
2383 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2384 * also cause waiting on I_NEW, without I_NEW actually
2385 * being set. find_inode() uses this to prevent returning
2386 * nearly-dead inodes.
2387 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2388 * is zero. I_FREEING must be set when I_WILL_FREE is
2390 * I_FREEING Set when inode is about to be freed but still has dirty
2391 * pages or buffers attached or the inode itself is still
2393 * I_CLEAR Added by clear_inode(). In this state the inode is
2394 * clean and can be destroyed. Inode keeps I_FREEING.
2396 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2397 * prohibited for many purposes. iget() must wait for
2398 * the inode to be completely released, then create it
2399 * anew. Other functions will just ignore such inodes,
2400 * if appropriate. I_NEW is used for waiting.
2402 * I_SYNC Writeback of inode is running. The bit is set during
2403 * data writeback, and cleared with a wakeup on the bit
2404 * address once it is done. The bit is also used to pin
2405 * the inode in memory for flusher thread.
2407 * I_REFERENCED Marks the inode as recently references on the LRU list.
2409 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2411 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2412 * synchronize competing switching instances and to tell
2413 * wb stat updates to grab the i_pages lock. See
2414 * inode_switch_wbs_work_fn() for details.
2416 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2417 * and work dirs among overlayfs mounts.
2419 * I_CREATING New object's inode in the middle of setting up.
2421 * I_DONTCACHE Evict inode as soon as it is not used anymore.
2423 * I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists.
2424 * Used to detect that mark_inode_dirty() should not move
2425 * inode between dirty lists.
2427 * I_PINNING_FSCACHE_WB Inode is pinning an fscache object for writeback.
2429 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2431 #define I_DIRTY_SYNC (1 << 0)
2432 #define I_DIRTY_DATASYNC (1 << 1)
2433 #define I_DIRTY_PAGES (1 << 2)
2435 #define I_NEW (1 << __I_NEW)
2436 #define I_WILL_FREE (1 << 4)
2437 #define I_FREEING (1 << 5)
2438 #define I_CLEAR (1 << 6)
2440 #define I_SYNC (1 << __I_SYNC)
2441 #define I_REFERENCED (1 << 8)
2442 #define __I_DIO_WAKEUP 9
2443 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2444 #define I_LINKABLE (1 << 10)
2445 #define I_DIRTY_TIME (1 << 11)
2446 #define I_WB_SWITCH (1 << 13)
2447 #define I_OVL_INUSE (1 << 14)
2448 #define I_CREATING (1 << 15)
2449 #define I_DONTCACHE (1 << 16)
2450 #define I_SYNC_QUEUED (1 << 17)
2451 #define I_PINNING_FSCACHE_WB (1 << 18)
2453 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2454 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2455 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2457 extern void __mark_inode_dirty(struct inode *, int);
2458 static inline void mark_inode_dirty(struct inode *inode)
2460 __mark_inode_dirty(inode, I_DIRTY);
2463 static inline void mark_inode_dirty_sync(struct inode *inode)
2465 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2469 * Returns true if the given inode itself only has dirty timestamps (its pages
2470 * may still be dirty) and isn't currently being allocated or freed.
2471 * Filesystems should call this if when writing an inode when lazytime is
2472 * enabled, they want to opportunistically write the timestamps of other inodes
2473 * located very nearby on-disk, e.g. in the same inode block. This returns true
2474 * if the given inode is in need of such an opportunistic update. Requires
2475 * i_lock, or at least later re-checking under i_lock.
2477 static inline bool inode_is_dirtytime_only(struct inode *inode)
2479 return (inode->i_state & (I_DIRTY_TIME | I_NEW |
2480 I_FREEING | I_WILL_FREE)) == I_DIRTY_TIME;
2483 extern void inc_nlink(struct inode *inode);
2484 extern void drop_nlink(struct inode *inode);
2485 extern void clear_nlink(struct inode *inode);
2486 extern void set_nlink(struct inode *inode, unsigned int nlink);
2488 static inline void inode_inc_link_count(struct inode *inode)
2491 mark_inode_dirty(inode);
2494 static inline void inode_dec_link_count(struct inode *inode)
2497 mark_inode_dirty(inode);
2500 enum file_time_flags {
2507 extern bool atime_needs_update(const struct path *, struct inode *);
2508 extern void touch_atime(const struct path *);
2509 int inode_update_time(struct inode *inode, struct timespec64 *time, int flags);
2511 static inline void file_accessed(struct file *file)
2513 if (!(file->f_flags & O_NOATIME))
2514 touch_atime(&file->f_path);
2517 extern int file_modified(struct file *file);
2518 int kiocb_modified(struct kiocb *iocb);
2520 int sync_inode_metadata(struct inode *inode, int wait);
2522 struct file_system_type {
2525 #define FS_REQUIRES_DEV 1
2526 #define FS_BINARY_MOUNTDATA 2
2527 #define FS_HAS_SUBTYPE 4
2528 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2529 #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */
2530 #define FS_ALLOW_IDMAP 32 /* FS has been updated to handle vfs idmappings. */
2531 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2532 int (*init_fs_context)(struct fs_context *);
2533 const struct fs_parameter_spec *parameters;
2534 struct dentry *(*mount) (struct file_system_type *, int,
2535 const char *, void *);
2536 void (*kill_sb) (struct super_block *);
2537 struct module *owner;
2538 struct file_system_type * next;
2539 struct hlist_head fs_supers;
2541 struct lock_class_key s_lock_key;
2542 struct lock_class_key s_umount_key;
2543 struct lock_class_key s_vfs_rename_key;
2544 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2546 struct lock_class_key i_lock_key;
2547 struct lock_class_key i_mutex_key;
2548 struct lock_class_key invalidate_lock_key;
2549 struct lock_class_key i_mutex_dir_key;
2552 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2554 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2555 int flags, const char *dev_name, void *data,
2556 int (*fill_super)(struct super_block *, void *, int));
2557 extern struct dentry *mount_single(struct file_system_type *fs_type,
2558 int flags, void *data,
2559 int (*fill_super)(struct super_block *, void *, int));
2560 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2561 int flags, void *data,
2562 int (*fill_super)(struct super_block *, void *, int));
2563 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2564 void generic_shutdown_super(struct super_block *sb);
2565 void kill_block_super(struct super_block *sb);
2566 void kill_anon_super(struct super_block *sb);
2567 void kill_litter_super(struct super_block *sb);
2568 void deactivate_super(struct super_block *sb);
2569 void deactivate_locked_super(struct super_block *sb);
2570 int set_anon_super(struct super_block *s, void *data);
2571 int set_anon_super_fc(struct super_block *s, struct fs_context *fc);
2572 int get_anon_bdev(dev_t *);
2573 void free_anon_bdev(dev_t);
2574 struct super_block *sget_fc(struct fs_context *fc,
2575 int (*test)(struct super_block *, struct fs_context *),
2576 int (*set)(struct super_block *, struct fs_context *));
2577 struct super_block *sget(struct file_system_type *type,
2578 int (*test)(struct super_block *,void *),
2579 int (*set)(struct super_block *,void *),
2580 int flags, void *data);
2582 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2583 #define fops_get(fops) \
2584 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2585 #define fops_put(fops) \
2586 do { if (fops) module_put((fops)->owner); } while(0)
2588 * This one is to be used *ONLY* from ->open() instances.
2589 * fops must be non-NULL, pinned down *and* module dependencies
2590 * should be sufficient to pin the caller down as well.
2592 #define replace_fops(f, fops) \
2594 struct file *__file = (f); \
2595 fops_put(__file->f_op); \
2596 BUG_ON(!(__file->f_op = (fops))); \
2599 extern int register_filesystem(struct file_system_type *);
2600 extern int unregister_filesystem(struct file_system_type *);
2601 extern int vfs_statfs(const struct path *, struct kstatfs *);
2602 extern int user_statfs(const char __user *, struct kstatfs *);
2603 extern int fd_statfs(int, struct kstatfs *);
2604 extern int freeze_super(struct super_block *super);
2605 extern int thaw_super(struct super_block *super);
2606 extern __printf(2, 3)
2607 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2608 extern int super_setup_bdi(struct super_block *sb);
2610 extern int current_umask(void);
2612 extern void ihold(struct inode * inode);
2613 extern void iput(struct inode *);
2614 extern int generic_update_time(struct inode *, struct timespec64 *, int);
2617 extern struct kobject *fs_kobj;
2619 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2621 #ifdef CONFIG_FILE_LOCKING
2622 static inline int break_lease(struct inode *inode, unsigned int mode)
2625 * Since this check is lockless, we must ensure that any refcounts
2626 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2627 * could end up racing with tasks trying to set a new lease on this
2631 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2632 return __break_lease(inode, mode, FL_LEASE);
2636 static inline int break_deleg(struct inode *inode, unsigned int mode)
2639 * Since this check is lockless, we must ensure that any refcounts
2640 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2641 * could end up racing with tasks trying to set a new lease on this
2645 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2646 return __break_lease(inode, mode, FL_DELEG);
2650 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2654 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2655 if (ret == -EWOULDBLOCK && delegated_inode) {
2656 *delegated_inode = inode;
2662 static inline int break_deleg_wait(struct inode **delegated_inode)
2666 ret = break_deleg(*delegated_inode, O_WRONLY);
2667 iput(*delegated_inode);
2668 *delegated_inode = NULL;
2672 static inline int break_layout(struct inode *inode, bool wait)
2675 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2676 return __break_lease(inode,
2677 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2682 #else /* !CONFIG_FILE_LOCKING */
2683 static inline int break_lease(struct inode *inode, unsigned int mode)
2688 static inline int break_deleg(struct inode *inode, unsigned int mode)
2693 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2698 static inline int break_deleg_wait(struct inode **delegated_inode)
2704 static inline int break_layout(struct inode *inode, bool wait)
2709 #endif /* CONFIG_FILE_LOCKING */
2714 const char *name; /* pointer to actual string */
2715 const __user char *uptr; /* original userland pointer */
2717 struct audit_names *aname;
2720 static_assert(offsetof(struct filename, iname) % sizeof(long) == 0);
2722 static inline struct user_namespace *file_mnt_user_ns(struct file *file)
2724 return mnt_user_ns(file->f_path.mnt);
2728 * is_idmapped_mnt - check whether a mount is mapped
2729 * @mnt: the mount to check
2731 * If @mnt has an idmapping attached different from the
2732 * filesystem's idmapping then @mnt is mapped.
2734 * Return: true if mount is mapped, false if not.
2736 static inline bool is_idmapped_mnt(const struct vfsmount *mnt)
2738 return mnt_user_ns(mnt) != mnt->mnt_sb->s_user_ns;
2741 extern long vfs_truncate(const struct path *, loff_t);
2742 int do_truncate(struct user_namespace *, struct dentry *, loff_t start,
2743 unsigned int time_attrs, struct file *filp);
2744 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2746 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2748 extern struct file *file_open_name(struct filename *, int, umode_t);
2749 extern struct file *filp_open(const char *, int, umode_t);
2750 extern struct file *file_open_root(const struct path *,
2751 const char *, int, umode_t);
2752 static inline struct file *file_open_root_mnt(struct vfsmount *mnt,
2753 const char *name, int flags, umode_t mode)
2755 return file_open_root(&(struct path){.mnt = mnt, .dentry = mnt->mnt_root},
2758 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2759 extern struct file *dentry_create(const struct path *path, int flags,
2760 umode_t mode, const struct cred *cred);
2761 extern struct file * open_with_fake_path(const struct path *, int,
2762 struct inode*, const struct cred *);
2763 static inline struct file *file_clone_open(struct file *file)
2765 return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2767 extern int filp_close(struct file *, fl_owner_t id);
2769 extern struct filename *getname_flags(const char __user *, int, int *);
2770 extern struct filename *getname_uflags(const char __user *, int);
2771 extern struct filename *getname(const char __user *);
2772 extern struct filename *getname_kernel(const char *);
2773 extern void putname(struct filename *name);
2775 extern int finish_open(struct file *file, struct dentry *dentry,
2776 int (*open)(struct inode *, struct file *));
2777 extern int finish_no_open(struct file *file, struct dentry *dentry);
2780 extern void __init vfs_caches_init_early(void);
2781 extern void __init vfs_caches_init(void);
2783 extern struct kmem_cache *names_cachep;
2785 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2786 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2788 extern struct super_block *blockdev_superblock;
2789 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2791 return IS_ENABLED(CONFIG_BLOCK) && sb == blockdev_superblock;
2794 void emergency_thaw_all(void);
2795 extern int sync_filesystem(struct super_block *);
2796 extern const struct file_operations def_blk_fops;
2797 extern const struct file_operations def_chr_fops;
2800 #define CHRDEV_MAJOR_MAX 512
2801 /* Marks the bottom of the first segment of free char majors */
2802 #define CHRDEV_MAJOR_DYN_END 234
2803 /* Marks the top and bottom of the second segment of free char majors */
2804 #define CHRDEV_MAJOR_DYN_EXT_START 511
2805 #define CHRDEV_MAJOR_DYN_EXT_END 384
2807 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2808 extern int register_chrdev_region(dev_t, unsigned, const char *);
2809 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2810 unsigned int count, const char *name,
2811 const struct file_operations *fops);
2812 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2813 unsigned int count, const char *name);
2814 extern void unregister_chrdev_region(dev_t, unsigned);
2815 extern void chrdev_show(struct seq_file *,off_t);
2817 static inline int register_chrdev(unsigned int major, const char *name,
2818 const struct file_operations *fops)
2820 return __register_chrdev(major, 0, 256, name, fops);
2823 static inline void unregister_chrdev(unsigned int major, const char *name)
2825 __unregister_chrdev(major, 0, 256, name);
2828 extern void init_special_inode(struct inode *, umode_t, dev_t);
2830 /* Invalid inode operations -- fs/bad_inode.c */
2831 extern void make_bad_inode(struct inode *);
2832 extern bool is_bad_inode(struct inode *);
2834 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2836 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2837 extern int __must_check file_write_and_wait_range(struct file *file,
2838 loff_t start, loff_t end);
2840 static inline int file_write_and_wait(struct file *file)
2842 return file_write_and_wait_range(file, 0, LLONG_MAX);
2845 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2847 extern int vfs_fsync(struct file *file, int datasync);
2849 extern int sync_file_range(struct file *file, loff_t offset, loff_t nbytes,
2850 unsigned int flags);
2852 static inline bool iocb_is_dsync(const struct kiocb *iocb)
2854 return (iocb->ki_flags & IOCB_DSYNC) ||
2855 IS_SYNC(iocb->ki_filp->f_mapping->host);
2859 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2860 * to already be updated for the write, and will return either the amount
2861 * of bytes passed in, or an error if syncing the file failed.
2863 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2865 if (iocb_is_dsync(iocb)) {
2866 int ret = vfs_fsync_range(iocb->ki_filp,
2867 iocb->ki_pos - count, iocb->ki_pos - 1,
2868 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2876 extern void emergency_sync(void);
2877 extern void emergency_remount(void);
2880 extern int bmap(struct inode *inode, sector_t *block);
2882 static inline int bmap(struct inode *inode, sector_t *block)
2888 int notify_change(struct user_namespace *, struct dentry *,
2889 struct iattr *, struct inode **);
2890 int inode_permission(struct user_namespace *, struct inode *, int);
2891 int generic_permission(struct user_namespace *, struct inode *, int);
2892 static inline int file_permission(struct file *file, int mask)
2894 return inode_permission(file_mnt_user_ns(file),
2895 file_inode(file), mask);
2897 static inline int path_permission(const struct path *path, int mask)
2899 return inode_permission(mnt_user_ns(path->mnt),
2900 d_inode(path->dentry), mask);
2902 int __check_sticky(struct user_namespace *mnt_userns, struct inode *dir,
2903 struct inode *inode);
2905 static inline bool execute_ok(struct inode *inode)
2907 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2910 static inline bool inode_wrong_type(const struct inode *inode, umode_t mode)
2912 return (inode->i_mode ^ mode) & S_IFMT;
2915 static inline void file_start_write(struct file *file)
2917 if (!S_ISREG(file_inode(file)->i_mode))
2919 sb_start_write(file_inode(file)->i_sb);
2922 static inline bool file_start_write_trylock(struct file *file)
2924 if (!S_ISREG(file_inode(file)->i_mode))
2926 return sb_start_write_trylock(file_inode(file)->i_sb);
2929 static inline void file_end_write(struct file *file)
2931 if (!S_ISREG(file_inode(file)->i_mode))
2933 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2937 * This is used for regular files where some users -- especially the
2938 * currently executed binary in a process, previously handled via
2939 * VM_DENYWRITE -- cannot handle concurrent write (and maybe mmap
2940 * read-write shared) accesses.
2942 * get_write_access() gets write permission for a file.
2943 * put_write_access() releases this write permission.
2944 * deny_write_access() denies write access to a file.
2945 * allow_write_access() re-enables write access to a file.
2947 * The i_writecount field of an inode can have the following values:
2948 * 0: no write access, no denied write access
2949 * < 0: (-i_writecount) users that denied write access to the file.
2950 * > 0: (i_writecount) users that have write access to the file.
2952 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2953 * except for the cases where we don't hold i_writecount yet. Then we need to
2954 * use {get,deny}_write_access() - these functions check the sign and refuse
2955 * to do the change if sign is wrong.
2957 static inline int get_write_access(struct inode *inode)
2959 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2961 static inline int deny_write_access(struct file *file)
2963 struct inode *inode = file_inode(file);
2964 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2966 static inline void put_write_access(struct inode * inode)
2968 atomic_dec(&inode->i_writecount);
2970 static inline void allow_write_access(struct file *file)
2973 atomic_inc(&file_inode(file)->i_writecount);
2975 static inline bool inode_is_open_for_write(const struct inode *inode)
2977 return atomic_read(&inode->i_writecount) > 0;
2980 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
2981 static inline void i_readcount_dec(struct inode *inode)
2983 BUG_ON(!atomic_read(&inode->i_readcount));
2984 atomic_dec(&inode->i_readcount);
2986 static inline void i_readcount_inc(struct inode *inode)
2988 atomic_inc(&inode->i_readcount);
2991 static inline void i_readcount_dec(struct inode *inode)
2995 static inline void i_readcount_inc(struct inode *inode)
3000 extern int do_pipe_flags(int *, int);
3002 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
3003 ssize_t __kernel_read(struct file *file, void *buf, size_t count, loff_t *pos);
3004 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
3005 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
3006 extern struct file * open_exec(const char *);
3008 /* fs/dcache.c -- generic fs support functions */
3009 extern bool is_subdir(struct dentry *, struct dentry *);
3010 extern bool path_is_under(const struct path *, const struct path *);
3012 extern char *file_path(struct file *, char *, int);
3014 #include <linux/err.h>
3016 /* needed for stackable file system support */
3017 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
3019 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
3021 extern int inode_init_always(struct super_block *, struct inode *);
3022 extern void inode_init_once(struct inode *);
3023 extern void address_space_init_once(struct address_space *mapping);
3024 extern struct inode * igrab(struct inode *);
3025 extern ino_t iunique(struct super_block *, ino_t);
3026 extern int inode_needs_sync(struct inode *inode);
3027 extern int generic_delete_inode(struct inode *inode);
3028 static inline int generic_drop_inode(struct inode *inode)
3030 return !inode->i_nlink || inode_unhashed(inode);
3032 extern void d_mark_dontcache(struct inode *inode);
3034 extern struct inode *ilookup5_nowait(struct super_block *sb,
3035 unsigned long hashval, int (*test)(struct inode *, void *),
3037 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
3038 int (*test)(struct inode *, void *), void *data);
3039 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
3041 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
3042 int (*test)(struct inode *, void *),
3043 int (*set)(struct inode *, void *),
3045 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
3046 extern struct inode * iget_locked(struct super_block *, unsigned long);
3047 extern struct inode *find_inode_nowait(struct super_block *,
3049 int (*match)(struct inode *,
3050 unsigned long, void *),
3052 extern struct inode *find_inode_rcu(struct super_block *, unsigned long,
3053 int (*)(struct inode *, void *), void *);
3054 extern struct inode *find_inode_by_ino_rcu(struct super_block *, unsigned long);
3055 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
3056 extern int insert_inode_locked(struct inode *);
3057 #ifdef CONFIG_DEBUG_LOCK_ALLOC
3058 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
3060 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
3062 extern void unlock_new_inode(struct inode *);
3063 extern void discard_new_inode(struct inode *);
3064 extern unsigned int get_next_ino(void);
3065 extern void evict_inodes(struct super_block *sb);
3066 void dump_mapping(const struct address_space *);
3069 * Userspace may rely on the the inode number being non-zero. For example, glibc
3070 * simply ignores files with zero i_ino in unlink() and other places.
3072 * As an additional complication, if userspace was compiled with
3073 * _FILE_OFFSET_BITS=32 on a 64-bit kernel we'll only end up reading out the
3074 * lower 32 bits, so we need to check that those aren't zero explicitly. With
3075 * _FILE_OFFSET_BITS=64, this may cause some harmless false-negatives, but
3076 * better safe than sorry.
3078 static inline bool is_zero_ino(ino_t ino)
3080 return (u32)ino == 0;
3083 extern void __iget(struct inode * inode);
3084 extern void iget_failed(struct inode *);
3085 extern void clear_inode(struct inode *);
3086 extern void __destroy_inode(struct inode *);
3087 extern struct inode *new_inode_pseudo(struct super_block *sb);
3088 extern struct inode *new_inode(struct super_block *sb);
3089 extern void free_inode_nonrcu(struct inode *inode);
3090 extern int should_remove_suid(struct dentry *);
3091 extern int file_remove_privs(struct file *);
3094 * This must be used for allocating filesystems specific inodes to set
3095 * up the inode reclaim context correctly.
3097 static inline void *
3098 alloc_inode_sb(struct super_block *sb, struct kmem_cache *cache, gfp_t gfp)
3100 return kmem_cache_alloc_lru(cache, &sb->s_inode_lru, gfp);
3103 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
3104 static inline void insert_inode_hash(struct inode *inode)
3106 __insert_inode_hash(inode, inode->i_ino);
3109 extern void __remove_inode_hash(struct inode *);
3110 static inline void remove_inode_hash(struct inode *inode)
3112 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
3113 __remove_inode_hash(inode);
3116 extern void inode_sb_list_add(struct inode *inode);
3117 extern void inode_add_lru(struct inode *inode);
3119 extern int sb_set_blocksize(struct super_block *, int);
3120 extern int sb_min_blocksize(struct super_block *, int);
3122 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
3123 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
3124 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
3125 int generic_write_checks_count(struct kiocb *iocb, loff_t *count);
3126 extern int generic_write_check_limits(struct file *file, loff_t pos,
3128 extern int generic_file_rw_checks(struct file *file_in, struct file *file_out);
3129 ssize_t filemap_read(struct kiocb *iocb, struct iov_iter *to,
3130 ssize_t already_read);
3131 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
3132 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
3133 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
3134 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
3135 ssize_t generic_perform_write(struct kiocb *, struct iov_iter *);
3137 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
3139 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
3141 ssize_t vfs_iocb_iter_read(struct file *file, struct kiocb *iocb,
3142 struct iov_iter *iter);
3143 ssize_t vfs_iocb_iter_write(struct file *file, struct kiocb *iocb,
3144 struct iov_iter *iter);
3147 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
3148 struct pipe_inode_info *, size_t, unsigned int);
3149 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
3150 struct file *, loff_t *, size_t, unsigned int);
3151 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
3152 struct file *out, loff_t *, size_t len, unsigned int flags);
3153 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
3154 loff_t *opos, size_t len, unsigned int flags);
3158 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3159 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3160 #define no_llseek NULL
3161 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3162 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3163 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3164 int whence, loff_t maxsize, loff_t eof);
3165 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3166 int whence, loff_t size);
3167 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3168 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3169 int rw_verify_area(int, struct file *, const loff_t *, size_t);
3170 extern int generic_file_open(struct inode * inode, struct file * filp);
3171 extern int nonseekable_open(struct inode * inode, struct file * filp);
3172 extern int stream_open(struct inode * inode, struct file * filp);
3175 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3176 loff_t file_offset);
3179 /* need locking between buffered and direct access */
3182 /* filesystem does not support filling holes */
3183 DIO_SKIP_HOLES = 0x02,
3186 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3187 struct block_device *bdev, struct iov_iter *iter,
3188 get_block_t get_block,
3189 dio_iodone_t end_io, dio_submit_t submit_io,
3192 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3193 struct inode *inode,
3194 struct iov_iter *iter,
3195 get_block_t get_block)
3197 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3198 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3202 void inode_dio_wait(struct inode *inode);
3205 * inode_dio_begin - signal start of a direct I/O requests
3206 * @inode: inode the direct I/O happens on
3208 * This is called once we've finished processing a direct I/O request,
3209 * and is used to wake up callers waiting for direct I/O to be quiesced.
3211 static inline void inode_dio_begin(struct inode *inode)
3213 atomic_inc(&inode->i_dio_count);
3217 * inode_dio_end - signal finish of a direct I/O requests
3218 * @inode: inode the direct I/O happens on
3220 * This is called once we've finished processing a direct I/O request,
3221 * and is used to wake up callers waiting for direct I/O to be quiesced.
3223 static inline void inode_dio_end(struct inode *inode)
3225 if (atomic_dec_and_test(&inode->i_dio_count))
3226 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3230 * Warn about a page cache invalidation failure diring a direct I/O write.
3232 void dio_warn_stale_pagecache(struct file *filp);
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 user_namespace *, 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 struct super_block *get_super(struct block_device *);
3289 extern struct super_block *get_active_super(struct block_device *bdev);
3290 extern void drop_super(struct super_block *sb);
3291 extern void drop_super_exclusive(struct super_block *sb);
3292 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3293 extern void iterate_supers_type(struct file_system_type *,
3294 void (*)(struct super_block *, void *), void *);
3296 extern int dcache_dir_open(struct inode *, struct file *);
3297 extern int dcache_dir_close(struct inode *, struct file *);
3298 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3299 extern int dcache_readdir(struct file *, struct dir_context *);
3300 extern int simple_setattr(struct user_namespace *, struct dentry *,
3302 extern int simple_getattr(struct user_namespace *, const struct path *,
3303 struct kstat *, u32, unsigned int);
3304 extern int simple_statfs(struct dentry *, struct kstatfs *);
3305 extern int simple_open(struct inode *inode, struct file *file);
3306 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3307 extern int simple_unlink(struct inode *, struct dentry *);
3308 extern int simple_rmdir(struct inode *, struct dentry *);
3309 extern int simple_rename_exchange(struct inode *old_dir, struct dentry *old_dentry,
3310 struct inode *new_dir, struct dentry *new_dentry);
3311 extern int simple_rename(struct user_namespace *, 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 *, long, struct file_lock **, 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);
3346 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3347 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3349 extern int generic_check_addressable(unsigned, u64);
3351 extern void generic_set_encrypted_ci_d_ops(struct dentry *dentry);
3353 int may_setattr(struct user_namespace *mnt_userns, struct inode *inode,
3354 unsigned int ia_valid);
3355 int setattr_prepare(struct user_namespace *, struct dentry *, struct iattr *);
3356 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3357 void setattr_copy(struct user_namespace *, struct inode *inode,
3358 const struct iattr *attr);
3360 extern int file_update_time(struct file *file);
3362 static inline bool vma_is_dax(const struct vm_area_struct *vma)
3364 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3367 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3369 struct inode *inode;
3371 if (!IS_ENABLED(CONFIG_FS_DAX) || !vma->vm_file)
3373 if (!vma_is_dax(vma))
3375 inode = file_inode(vma->vm_file);
3376 if (S_ISCHR(inode->i_mode))
3377 return false; /* device-dax */
3381 static inline int iocb_flags(struct file *file)
3384 if (file->f_flags & O_APPEND)
3386 if (file->f_flags & O_DIRECT)
3388 if (file->f_flags & O_DSYNC)
3390 if (file->f_flags & __O_SYNC)
3395 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3397 int kiocb_flags = 0;
3399 /* make sure there's no overlap between RWF and private IOCB flags */
3400 BUILD_BUG_ON((__force int) RWF_SUPPORTED & IOCB_EVENTFD);
3404 if (unlikely(flags & ~RWF_SUPPORTED))
3407 if (flags & RWF_NOWAIT) {
3408 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3410 kiocb_flags |= IOCB_NOIO;
3412 kiocb_flags |= (__force int) (flags & RWF_SUPPORTED);
3413 if (flags & RWF_SYNC)
3414 kiocb_flags |= IOCB_DSYNC;
3416 ki->ki_flags |= kiocb_flags;
3420 static inline ino_t parent_ino(struct dentry *dentry)
3425 * Don't strictly need d_lock here? If the parent ino could change
3426 * then surely we'd have a deeper race in the caller?
3428 spin_lock(&dentry->d_lock);
3429 res = dentry->d_parent->d_inode->i_ino;
3430 spin_unlock(&dentry->d_lock);
3434 /* Transaction based IO helpers */
3437 * An argresp is stored in an allocated page and holds the
3438 * size of the argument or response, along with its content
3440 struct simple_transaction_argresp {
3445 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3447 char *simple_transaction_get(struct file *file, const char __user *buf,
3449 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3450 size_t size, loff_t *pos);
3451 int simple_transaction_release(struct inode *inode, struct file *file);
3453 void simple_transaction_set(struct file *file, size_t n);
3456 * simple attribute files
3458 * These attributes behave similar to those in sysfs:
3460 * Writing to an attribute immediately sets a value, an open file can be
3461 * written to multiple times.
3463 * Reading from an attribute creates a buffer from the value that might get
3464 * read with multiple read calls. When the attribute has been read
3465 * completely, no further read calls are possible until the file is opened
3468 * All attributes contain a text representation of a numeric value
3469 * that are accessed with the get() and set() functions.
3471 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3472 static int __fops ## _open(struct inode *inode, struct file *file) \
3474 __simple_attr_check_format(__fmt, 0ull); \
3475 return simple_attr_open(inode, file, __get, __set, __fmt); \
3477 static const struct file_operations __fops = { \
3478 .owner = THIS_MODULE, \
3479 .open = __fops ## _open, \
3480 .release = simple_attr_release, \
3481 .read = simple_attr_read, \
3482 .write = simple_attr_write, \
3483 .llseek = generic_file_llseek, \
3486 static inline __printf(1, 2)
3487 void __simple_attr_check_format(const char *fmt, ...)
3489 /* don't do anything, just let the compiler check the arguments; */
3492 int simple_attr_open(struct inode *inode, struct file *file,
3493 int (*get)(void *, u64 *), int (*set)(void *, u64),
3495 int simple_attr_release(struct inode *inode, struct file *file);
3496 ssize_t simple_attr_read(struct file *file, char __user *buf,
3497 size_t len, loff_t *ppos);
3498 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3499 size_t len, loff_t *ppos);
3502 int __init list_bdev_fs_names(char *buf, size_t size);
3504 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3505 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3507 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3508 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3509 (flag & __FMODE_NONOTIFY)))
3511 static inline bool is_sxid(umode_t mode)
3513 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3516 static inline int check_sticky(struct user_namespace *mnt_userns,
3517 struct inode *dir, struct inode *inode)
3519 if (!(dir->i_mode & S_ISVTX))
3522 return __check_sticky(mnt_userns, dir, inode);
3525 static inline void inode_has_no_xattr(struct inode *inode)
3527 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3528 inode->i_flags |= S_NOSEC;
3531 static inline bool is_root_inode(struct inode *inode)
3533 return inode == inode->i_sb->s_root->d_inode;
3536 static inline bool dir_emit(struct dir_context *ctx,
3537 const char *name, int namelen,
3538 u64 ino, unsigned type)
3540 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3542 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3544 return ctx->actor(ctx, ".", 1, ctx->pos,
3545 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3547 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3549 return ctx->actor(ctx, "..", 2, ctx->pos,
3550 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3552 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3554 if (ctx->pos == 0) {
3555 if (!dir_emit_dot(file, ctx))
3559 if (ctx->pos == 1) {
3560 if (!dir_emit_dotdot(file, ctx))
3566 static inline bool dir_relax(struct inode *inode)
3568 inode_unlock(inode);
3570 return !IS_DEADDIR(inode);
3573 static inline bool dir_relax_shared(struct inode *inode)
3575 inode_unlock_shared(inode);
3576 inode_lock_shared(inode);
3577 return !IS_DEADDIR(inode);
3580 extern bool path_noexec(const struct path *path);
3581 extern void inode_nohighmem(struct inode *inode);
3584 extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len,
3586 extern int generic_fadvise(struct file *file, loff_t offset, loff_t len,
3589 #endif /* _LINUX_FS_H */