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>
45 #include <asm/byteorder.h>
46 #include <uapi/linux/fs.h>
48 struct backing_dev_info;
51 struct export_operations;
52 struct fiemap_extent_info;
57 struct pipe_inode_info;
58 struct poll_table_struct;
60 struct vm_area_struct;
63 struct swap_info_struct;
65 struct workqueue_struct;
68 struct fscrypt_operations;
70 struct fsverity_operations;
72 struct fs_parameter_spec;
75 extern void __init inode_init(void);
76 extern void __init inode_init_early(void);
77 extern void __init files_init(void);
78 extern void __init files_maxfiles_init(void);
80 extern struct files_stat_struct files_stat;
81 extern unsigned long get_max_files(void);
82 extern unsigned int sysctl_nr_open;
83 extern struct inodes_stat_t inodes_stat;
84 extern int leases_enable, lease_break_time;
85 extern int sysctl_protected_symlinks;
86 extern int sysctl_protected_hardlinks;
87 extern int sysctl_protected_fifos;
88 extern int sysctl_protected_regular;
90 typedef __kernel_rwf_t rwf_t;
93 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
94 struct buffer_head *bh_result, int create);
95 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
96 ssize_t bytes, void *private);
98 #define MAY_EXEC 0x00000001
99 #define MAY_WRITE 0x00000002
100 #define MAY_READ 0x00000004
101 #define MAY_APPEND 0x00000008
102 #define MAY_ACCESS 0x00000010
103 #define MAY_OPEN 0x00000020
104 #define MAY_CHDIR 0x00000040
105 /* called from RCU mode, don't block */
106 #define MAY_NOT_BLOCK 0x00000080
109 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
110 * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
113 /* file is open for reading */
114 #define FMODE_READ ((__force fmode_t)0x1)
115 /* file is open for writing */
116 #define FMODE_WRITE ((__force fmode_t)0x2)
117 /* file is seekable */
118 #define FMODE_LSEEK ((__force fmode_t)0x4)
119 /* file can be accessed using pread */
120 #define FMODE_PREAD ((__force fmode_t)0x8)
121 /* file can be accessed using pwrite */
122 #define FMODE_PWRITE ((__force fmode_t)0x10)
123 /* File is opened for execution with sys_execve / sys_uselib */
124 #define FMODE_EXEC ((__force fmode_t)0x20)
125 /* File is opened with O_NDELAY (only set for block devices) */
126 #define FMODE_NDELAY ((__force fmode_t)0x40)
127 /* File is opened with O_EXCL (only set for block devices) */
128 #define FMODE_EXCL ((__force fmode_t)0x80)
129 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
130 (specialy hack for floppy.c) */
131 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
132 /* 32bit hashes as llseek() offset (for directories) */
133 #define FMODE_32BITHASH ((__force fmode_t)0x200)
134 /* 64bit hashes as llseek() offset (for directories) */
135 #define FMODE_64BITHASH ((__force fmode_t)0x400)
138 * Don't update ctime and mtime.
140 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
141 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
143 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
145 /* Expect random access pattern */
146 #define FMODE_RANDOM ((__force fmode_t)0x1000)
148 /* File is huge (eg. /dev/mem): treat loff_t as unsigned */
149 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
151 /* File is opened with O_PATH; almost nothing can be done with it */
152 #define FMODE_PATH ((__force fmode_t)0x4000)
154 /* File needs atomic accesses to f_pos */
155 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
156 /* Write access to underlying fs */
157 #define FMODE_WRITER ((__force fmode_t)0x10000)
158 /* Has read method(s) */
159 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
160 /* Has write method(s) */
161 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
163 #define FMODE_OPENED ((__force fmode_t)0x80000)
164 #define FMODE_CREATED ((__force fmode_t)0x100000)
166 /* File is stream-like */
167 #define FMODE_STREAM ((__force fmode_t)0x200000)
169 /* File was opened by fanotify and shouldn't generate fanotify events */
170 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
172 /* File is capable of returning -EAGAIN if I/O will block */
173 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
175 /* File represents mount that needs unmounting */
176 #define FMODE_NEED_UNMOUNT ((__force fmode_t)0x10000000)
178 /* File does not contribute to nr_files count */
179 #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000)
181 /* File supports async buffered reads */
182 #define FMODE_BUF_RASYNC ((__force fmode_t)0x40000000)
185 * Attribute flags. These should be or-ed together to figure out what
188 #define ATTR_MODE (1 << 0)
189 #define ATTR_UID (1 << 1)
190 #define ATTR_GID (1 << 2)
191 #define ATTR_SIZE (1 << 3)
192 #define ATTR_ATIME (1 << 4)
193 #define ATTR_MTIME (1 << 5)
194 #define ATTR_CTIME (1 << 6)
195 #define ATTR_ATIME_SET (1 << 7)
196 #define ATTR_MTIME_SET (1 << 8)
197 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
198 #define ATTR_KILL_SUID (1 << 11)
199 #define ATTR_KILL_SGID (1 << 12)
200 #define ATTR_FILE (1 << 13)
201 #define ATTR_KILL_PRIV (1 << 14)
202 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
203 #define ATTR_TIMES_SET (1 << 16)
204 #define ATTR_TOUCH (1 << 17)
207 * Whiteout is represented by a char device. The following constants define the
208 * mode and device number to use.
210 #define WHITEOUT_MODE 0
211 #define WHITEOUT_DEV 0
214 * This is the Inode Attributes structure, used for notify_change(). It
215 * uses the above definitions as flags, to know which values have changed.
216 * Also, in this manner, a Filesystem can look at only the values it cares
217 * about. Basically, these are the attributes that the VFS layer can
218 * request to change from the FS layer.
220 * Derek Atkins <warlord@MIT.EDU> 94-10-20
223 unsigned int ia_valid;
228 struct timespec64 ia_atime;
229 struct timespec64 ia_mtime;
230 struct timespec64 ia_ctime;
233 * Not an attribute, but an auxiliary info for filesystems wanting to
234 * implement an ftruncate() like method. NOTE: filesystem should
235 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
237 struct file *ia_file;
241 * Includes for diskquotas.
243 #include <linux/quota.h>
246 * Maximum number of layers of fs stack. Needs to be limited to
247 * prevent kernel stack overflow
249 #define FILESYSTEM_MAX_STACK_DEPTH 2
252 * enum positive_aop_returns - aop return codes with specific semantics
254 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
255 * completed, that the page is still locked, and
256 * should be considered active. The VM uses this hint
257 * to return the page to the active list -- it won't
258 * be a candidate for writeback again in the near
259 * future. Other callers must be careful to unlock
260 * the page if they get this return. Returned by
263 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
264 * unlocked it and the page might have been truncated.
265 * The caller should back up to acquiring a new page and
266 * trying again. The aop will be taking reasonable
267 * precautions not to livelock. If the caller held a page
268 * reference, it should drop it before retrying. Returned
271 * address_space_operation functions return these large constants to indicate
272 * special semantics to the caller. These are much larger than the bytes in a
273 * page to allow for functions that return the number of bytes operated on in a
277 enum positive_aop_returns {
278 AOP_WRITEPAGE_ACTIVATE = 0x80000,
279 AOP_TRUNCATED_PAGE = 0x80001,
282 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
283 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
284 * helper code (eg buffer layer)
285 * to clear GFP_FS from alloc */
288 * oh the beauties of C type declarations.
291 struct address_space;
292 struct writeback_control;
293 struct readahead_control;
296 * Write life time hint values.
297 * Stored in struct inode as u8.
300 WRITE_LIFE_NOT_SET = 0,
301 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
302 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
303 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
304 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
305 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
308 /* Match RWF_* bits to IOCB bits */
309 #define IOCB_HIPRI (__force int) RWF_HIPRI
310 #define IOCB_DSYNC (__force int) RWF_DSYNC
311 #define IOCB_SYNC (__force int) RWF_SYNC
312 #define IOCB_NOWAIT (__force int) RWF_NOWAIT
313 #define IOCB_APPEND (__force int) RWF_APPEND
315 /* non-RWF related bits - start at 16 */
316 #define IOCB_EVENTFD (1 << 16)
317 #define IOCB_DIRECT (1 << 17)
318 #define IOCB_WRITE (1 << 18)
319 /* iocb->ki_waitq is valid */
320 #define IOCB_WAITQ (1 << 19)
321 #define IOCB_NOIO (1 << 20)
322 /* can use bio alloc cache */
323 #define IOCB_ALLOC_CACHE (1 << 21)
326 struct file *ki_filp;
328 /* The 'ki_filp' pointer is shared in a union for aio */
329 randomized_struct_fields_start
332 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
336 u16 ki_ioprio; /* See linux/ioprio.h */
338 unsigned int ki_cookie; /* for ->iopoll */
339 struct wait_page_queue *ki_waitq; /* for async buffered IO */
342 randomized_struct_fields_end
345 static inline bool is_sync_kiocb(struct kiocb *kiocb)
347 return kiocb->ki_complete == NULL;
351 * "descriptor" for what we're up to with a read.
352 * This allows us to use the same read code yet
353 * have multiple different users of the data that
354 * we read from a file.
356 * The simplest case just copies the data to user
369 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
370 unsigned long, unsigned long);
372 struct address_space_operations {
373 int (*writepage)(struct page *page, struct writeback_control *wbc);
374 int (*readpage)(struct file *, struct page *);
376 /* Write back some dirty pages from this mapping. */
377 int (*writepages)(struct address_space *, struct writeback_control *);
379 /* Set a page dirty. Return true if this dirtied it */
380 int (*set_page_dirty)(struct page *page);
383 * Reads in the requested pages. Unlike ->readpage(), this is
384 * PURELY used for read-ahead!.
386 int (*readpages)(struct file *filp, struct address_space *mapping,
387 struct list_head *pages, unsigned nr_pages);
388 void (*readahead)(struct readahead_control *);
390 int (*write_begin)(struct file *, struct address_space *mapping,
391 loff_t pos, unsigned len, unsigned flags,
392 struct page **pagep, void **fsdata);
393 int (*write_end)(struct file *, struct address_space *mapping,
394 loff_t pos, unsigned len, unsigned copied,
395 struct page *page, void *fsdata);
397 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
398 sector_t (*bmap)(struct address_space *, sector_t);
399 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
400 int (*releasepage) (struct page *, gfp_t);
401 void (*freepage)(struct page *);
402 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
404 * migrate the contents of a page to the specified target. If
405 * migrate_mode is MIGRATE_ASYNC, it must not block.
407 int (*migratepage) (struct address_space *,
408 struct page *, struct page *, enum migrate_mode);
409 bool (*isolate_page)(struct page *, isolate_mode_t);
410 void (*putback_page)(struct page *);
411 int (*launder_page) (struct page *);
412 int (*is_partially_uptodate) (struct page *, unsigned long,
414 void (*is_dirty_writeback) (struct page *, bool *, bool *);
415 int (*error_remove_page)(struct address_space *, struct page *);
417 /* swapfile support */
418 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
420 void (*swap_deactivate)(struct file *file);
423 extern const struct address_space_operations empty_aops;
426 * pagecache_write_begin/pagecache_write_end must be used by general code
427 * to write into the pagecache.
429 int pagecache_write_begin(struct file *, struct address_space *mapping,
430 loff_t pos, unsigned len, unsigned flags,
431 struct page **pagep, void **fsdata);
433 int pagecache_write_end(struct file *, struct address_space *mapping,
434 loff_t pos, unsigned len, unsigned copied,
435 struct page *page, void *fsdata);
438 * struct address_space - Contents of a cacheable, mappable object.
439 * @host: Owner, either the inode or the block_device.
440 * @i_pages: Cached pages.
441 * @invalidate_lock: Guards coherency between page cache contents and
442 * file offset->disk block mappings in the filesystem during invalidates.
443 * It is also used to block modification of page cache contents through
445 * @gfp_mask: Memory allocation flags to use for allocating pages.
446 * @i_mmap_writable: Number of VM_SHARED mappings.
447 * @nr_thps: Number of THPs in the pagecache (non-shmem only).
448 * @i_mmap: Tree of private and shared mappings.
449 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
450 * @nrpages: Number of page entries, protected by the i_pages lock.
451 * @writeback_index: Writeback starts here.
453 * @flags: Error bits and flags (AS_*).
454 * @wb_err: The most recent error which has occurred.
455 * @private_lock: For use by the owner of the address_space.
456 * @private_list: For use by the owner of the address_space.
457 * @private_data: For use by the owner of the address_space.
459 struct address_space {
461 struct xarray i_pages;
462 struct rw_semaphore invalidate_lock;
464 atomic_t i_mmap_writable;
465 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
466 /* number of thp, only for non-shmem files */
469 struct rb_root_cached i_mmap;
470 struct rw_semaphore i_mmap_rwsem;
471 unsigned long nrpages;
472 pgoff_t writeback_index;
473 const struct address_space_operations *a_ops;
476 spinlock_t private_lock;
477 struct list_head private_list;
479 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
481 * On most architectures that alignment is already the case; but
482 * must be enforced here for CRIS, to let the least significant bit
483 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
486 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
487 #define PAGECACHE_TAG_DIRTY XA_MARK_0
488 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
489 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
492 * Returns true if any of the pages in the mapping are marked with the tag.
494 static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag)
496 return xa_marked(&mapping->i_pages, tag);
499 static inline void i_mmap_lock_write(struct address_space *mapping)
501 down_write(&mapping->i_mmap_rwsem);
504 static inline int i_mmap_trylock_write(struct address_space *mapping)
506 return down_write_trylock(&mapping->i_mmap_rwsem);
509 static inline void i_mmap_unlock_write(struct address_space *mapping)
511 up_write(&mapping->i_mmap_rwsem);
514 static inline void i_mmap_lock_read(struct address_space *mapping)
516 down_read(&mapping->i_mmap_rwsem);
519 static inline void i_mmap_unlock_read(struct address_space *mapping)
521 up_read(&mapping->i_mmap_rwsem);
524 static inline void i_mmap_assert_locked(struct address_space *mapping)
526 lockdep_assert_held(&mapping->i_mmap_rwsem);
529 static inline void i_mmap_assert_write_locked(struct address_space *mapping)
531 lockdep_assert_held_write(&mapping->i_mmap_rwsem);
535 * Might pages of this file be mapped into userspace?
537 static inline int mapping_mapped(struct address_space *mapping)
539 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
543 * Might pages of this file have been modified in userspace?
544 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap
545 * marks vma as VM_SHARED if it is shared, and the file was opened for
546 * writing i.e. vma may be mprotected writable even if now readonly.
548 * If i_mmap_writable is negative, no new writable mappings are allowed. You
549 * can only deny writable mappings, if none exists right now.
551 static inline int mapping_writably_mapped(struct address_space *mapping)
553 return atomic_read(&mapping->i_mmap_writable) > 0;
556 static inline int mapping_map_writable(struct address_space *mapping)
558 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
562 static inline void mapping_unmap_writable(struct address_space *mapping)
564 atomic_dec(&mapping->i_mmap_writable);
567 static inline int mapping_deny_writable(struct address_space *mapping)
569 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
573 static inline void mapping_allow_writable(struct address_space *mapping)
575 atomic_inc(&mapping->i_mmap_writable);
579 * Use sequence counter to get consistent i_size on 32-bit processors.
581 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
582 #include <linux/seqlock.h>
583 #define __NEED_I_SIZE_ORDERED
584 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
586 #define i_size_ordered_init(inode) do { } while (0)
590 #define ACL_NOT_CACHED ((void *)(-1))
591 #define ACL_DONT_CACHE ((void *)(-3))
593 static inline struct posix_acl *
594 uncached_acl_sentinel(struct task_struct *task)
596 return (void *)task + 1;
600 is_uncached_acl(struct posix_acl *acl)
602 return (long)acl & 1;
605 #define IOP_FASTPERM 0x0001
606 #define IOP_LOOKUP 0x0002
607 #define IOP_NOFOLLOW 0x0004
608 #define IOP_XATTR 0x0008
609 #define IOP_DEFAULT_READLINK 0x0010
611 struct fsnotify_mark_connector;
614 * Keep mostly read-only and often accessed (especially for
615 * the RCU path lookup and 'stat' data) fields at the beginning
616 * of the 'struct inode'
620 unsigned short i_opflags;
623 unsigned int i_flags;
625 #ifdef CONFIG_FS_POSIX_ACL
626 struct posix_acl *i_acl;
627 struct posix_acl *i_default_acl;
630 const struct inode_operations *i_op;
631 struct super_block *i_sb;
632 struct address_space *i_mapping;
634 #ifdef CONFIG_SECURITY
638 /* Stat data, not accessed from path walking */
641 * Filesystems may only read i_nlink directly. They shall use the
642 * following functions for modification:
644 * (set|clear|inc|drop)_nlink
645 * inode_(inc|dec)_link_count
648 const unsigned int i_nlink;
649 unsigned int __i_nlink;
653 struct timespec64 i_atime;
654 struct timespec64 i_mtime;
655 struct timespec64 i_ctime;
656 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
657 unsigned short i_bytes;
662 #ifdef __NEED_I_SIZE_ORDERED
663 seqcount_t i_size_seqcount;
667 unsigned long i_state;
668 struct rw_semaphore i_rwsem;
670 unsigned long dirtied_when; /* jiffies of first dirtying */
671 unsigned long dirtied_time_when;
673 struct hlist_node i_hash;
674 struct list_head i_io_list; /* backing dev IO list */
675 #ifdef CONFIG_CGROUP_WRITEBACK
676 struct bdi_writeback *i_wb; /* the associated cgroup wb */
678 /* foreign inode detection, see wbc_detach_inode() */
680 u16 i_wb_frn_avg_time;
681 u16 i_wb_frn_history;
683 struct list_head i_lru; /* inode LRU list */
684 struct list_head i_sb_list;
685 struct list_head i_wb_list; /* backing dev writeback list */
687 struct hlist_head i_dentry;
688 struct rcu_head i_rcu;
690 atomic64_t i_version;
691 atomic64_t i_sequence; /* see futex */
693 atomic_t i_dio_count;
694 atomic_t i_writecount;
695 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
696 atomic_t i_readcount; /* struct files open RO */
699 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
700 void (*free_inode)(struct inode *);
702 struct file_lock_context *i_flctx;
703 struct address_space i_data;
704 struct list_head i_devices;
706 struct pipe_inode_info *i_pipe;
714 #ifdef CONFIG_FSNOTIFY
715 __u32 i_fsnotify_mask; /* all events this inode cares about */
716 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
719 #ifdef CONFIG_FS_ENCRYPTION
720 struct fscrypt_info *i_crypt_info;
723 #ifdef CONFIG_FS_VERITY
724 struct fsverity_info *i_verity_info;
727 void *i_private; /* fs or device private pointer */
728 } __randomize_layout;
730 struct timespec64 timestamp_truncate(struct timespec64 t, struct inode *inode);
732 static inline unsigned int i_blocksize(const struct inode *node)
734 return (1 << node->i_blkbits);
737 static inline int inode_unhashed(struct inode *inode)
739 return hlist_unhashed(&inode->i_hash);
743 * __mark_inode_dirty expects inodes to be hashed. Since we don't
744 * want special inodes in the fileset inode space, we make them
745 * appear hashed, but do not put on any lists. hlist_del()
746 * will work fine and require no locking.
748 static inline void inode_fake_hash(struct inode *inode)
750 hlist_add_fake(&inode->i_hash);
754 * inode->i_mutex nesting subclasses for the lock validator:
756 * 0: the object of the current VFS operation
760 * 4: second non-directory
761 * 5: second parent (when locking independent directories in rename)
763 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
764 * non-directories at once.
766 * The locking order between these classes is
767 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
769 enum inode_i_mutex_lock_class
779 static inline void inode_lock(struct inode *inode)
781 down_write(&inode->i_rwsem);
784 static inline void inode_unlock(struct inode *inode)
786 up_write(&inode->i_rwsem);
789 static inline void inode_lock_shared(struct inode *inode)
791 down_read(&inode->i_rwsem);
794 static inline void inode_unlock_shared(struct inode *inode)
796 up_read(&inode->i_rwsem);
799 static inline int inode_trylock(struct inode *inode)
801 return down_write_trylock(&inode->i_rwsem);
804 static inline int inode_trylock_shared(struct inode *inode)
806 return down_read_trylock(&inode->i_rwsem);
809 static inline int inode_is_locked(struct inode *inode)
811 return rwsem_is_locked(&inode->i_rwsem);
814 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
816 down_write_nested(&inode->i_rwsem, subclass);
819 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
821 down_read_nested(&inode->i_rwsem, subclass);
824 static inline void filemap_invalidate_lock(struct address_space *mapping)
826 down_write(&mapping->invalidate_lock);
829 static inline void filemap_invalidate_unlock(struct address_space *mapping)
831 up_write(&mapping->invalidate_lock);
834 static inline void filemap_invalidate_lock_shared(struct address_space *mapping)
836 down_read(&mapping->invalidate_lock);
839 static inline int filemap_invalidate_trylock_shared(
840 struct address_space *mapping)
842 return down_read_trylock(&mapping->invalidate_lock);
845 static inline void filemap_invalidate_unlock_shared(
846 struct address_space *mapping)
848 up_read(&mapping->invalidate_lock);
851 void lock_two_nondirectories(struct inode *, struct inode*);
852 void unlock_two_nondirectories(struct inode *, struct inode*);
854 void filemap_invalidate_lock_two(struct address_space *mapping1,
855 struct address_space *mapping2);
856 void filemap_invalidate_unlock_two(struct address_space *mapping1,
857 struct address_space *mapping2);
861 * NOTE: in a 32bit arch with a preemptable kernel and
862 * an UP compile the i_size_read/write must be atomic
863 * with respect to the local cpu (unlike with preempt disabled),
864 * but they don't need to be atomic with respect to other cpus like in
865 * true SMP (so they need either to either locally disable irq around
866 * the read or for example on x86 they can be still implemented as a
867 * cmpxchg8b without the need of the lock prefix). For SMP compiles
868 * and 64bit archs it makes no difference if preempt is enabled or not.
870 static inline loff_t i_size_read(const struct inode *inode)
872 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
877 seq = read_seqcount_begin(&inode->i_size_seqcount);
878 i_size = inode->i_size;
879 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
881 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
885 i_size = inode->i_size;
889 return inode->i_size;
894 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
895 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
896 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
898 static inline void i_size_write(struct inode *inode, loff_t i_size)
900 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
902 write_seqcount_begin(&inode->i_size_seqcount);
903 inode->i_size = i_size;
904 write_seqcount_end(&inode->i_size_seqcount);
906 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
908 inode->i_size = i_size;
911 inode->i_size = i_size;
915 static inline unsigned iminor(const struct inode *inode)
917 return MINOR(inode->i_rdev);
920 static inline unsigned imajor(const struct inode *inode)
922 return MAJOR(inode->i_rdev);
926 rwlock_t lock; /* protects pid, uid, euid fields */
927 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
928 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
929 kuid_t uid, euid; /* uid/euid of process setting the owner */
930 int signum; /* posix.1b rt signal to be delivered on IO */
934 * struct file_ra_state - Track a file's readahead state.
935 * @start: Where the most recent readahead started.
936 * @size: Number of pages read in the most recent readahead.
937 * @async_size: Start next readahead when this many pages are left.
938 * @ra_pages: Maximum size of a readahead request.
939 * @mmap_miss: How many mmap accesses missed in the page cache.
940 * @prev_pos: The last byte in the most recent read request.
942 struct file_ra_state {
945 unsigned int async_size;
946 unsigned int ra_pages;
947 unsigned int mmap_miss;
952 * Check if @index falls in the readahead windows.
954 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
956 return (index >= ra->start &&
957 index < ra->start + ra->size);
962 struct llist_node fu_llist;
963 struct rcu_head fu_rcuhead;
966 struct inode *f_inode; /* cached value */
967 const struct file_operations *f_op;
970 * Protects f_ep, f_flags.
971 * Must not be taken from IRQ context.
974 enum rw_hint f_write_hint;
975 atomic_long_t f_count;
976 unsigned int f_flags;
978 struct mutex f_pos_lock;
980 struct fown_struct f_owner;
981 const struct cred *f_cred;
982 struct file_ra_state f_ra;
985 #ifdef CONFIG_SECURITY
988 /* needed for tty driver, and maybe others */
992 /* Used by fs/eventpoll.c to link all the hooks to this file */
993 struct hlist_head *f_ep;
994 #endif /* #ifdef CONFIG_EPOLL */
995 struct address_space *f_mapping;
997 errseq_t f_sb_err; /* for syncfs */
999 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
1001 struct file_handle {
1004 /* file identifier */
1005 unsigned char f_handle[];
1008 static inline struct file *get_file(struct file *f)
1010 atomic_long_inc(&f->f_count);
1013 #define get_file_rcu_many(x, cnt) \
1014 atomic_long_add_unless(&(x)->f_count, (cnt), 0)
1015 #define get_file_rcu(x) get_file_rcu_many((x), 1)
1016 #define file_count(x) atomic_long_read(&(x)->f_count)
1018 #define MAX_NON_LFS ((1UL<<31) - 1)
1020 /* Page cache limit. The filesystems should put that into their s_maxbytes
1021 limits, otherwise bad things can happen in VM. */
1022 #if BITS_PER_LONG==32
1023 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
1024 #elif BITS_PER_LONG==64
1025 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
1030 #define FL_DELEG 4 /* NFSv4 delegation */
1031 #define FL_ACCESS 8 /* not trying to lock, just looking */
1032 #define FL_EXISTS 16 /* when unlocking, test for existence */
1033 #define FL_LEASE 32 /* lease held on this file */
1034 #define FL_CLOSE 64 /* unlock on close */
1035 #define FL_SLEEP 128 /* A blocking lock */
1036 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
1037 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
1038 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
1039 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
1041 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
1044 * Special return value from posix_lock_file() and vfs_lock_file() for
1045 * asynchronous locking.
1047 #define FILE_LOCK_DEFERRED 1
1049 /* legacy typedef, should eventually be removed */
1050 typedef void *fl_owner_t;
1054 struct file_lock_operations {
1055 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1056 void (*fl_release_private)(struct file_lock *);
1059 struct lock_manager_operations {
1060 fl_owner_t (*lm_get_owner)(fl_owner_t);
1061 void (*lm_put_owner)(fl_owner_t);
1062 void (*lm_notify)(struct file_lock *); /* unblock callback */
1063 int (*lm_grant)(struct file_lock *, int);
1064 bool (*lm_break)(struct file_lock *);
1065 int (*lm_change)(struct file_lock *, int, struct list_head *);
1066 void (*lm_setup)(struct file_lock *, void **);
1067 bool (*lm_breaker_owns_lease)(struct file_lock *);
1070 struct lock_manager {
1071 struct list_head list;
1073 * NFSv4 and up also want opens blocked during the grace period;
1080 void locks_start_grace(struct net *, struct lock_manager *);
1081 void locks_end_grace(struct lock_manager *);
1082 bool locks_in_grace(struct net *);
1083 bool opens_in_grace(struct net *);
1085 /* that will die - we need it for nfs_lock_info */
1086 #include <linux/nfs_fs_i.h>
1089 * struct file_lock represents a generic "file lock". It's used to represent
1090 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1091 * note that the same struct is used to represent both a request for a lock and
1092 * the lock itself, but the same object is never used for both.
1094 * FIXME: should we create a separate "struct lock_request" to help distinguish
1097 * The varous i_flctx lists are ordered by:
1100 * 2) lock range start
1103 * Obviously, the last two criteria only matter for POSIX locks.
1106 struct file_lock *fl_blocker; /* The lock, that is blocking us */
1107 struct list_head fl_list; /* link into file_lock_context */
1108 struct hlist_node fl_link; /* node in global lists */
1109 struct list_head fl_blocked_requests; /* list of requests with
1110 * ->fl_blocker pointing here
1112 struct list_head fl_blocked_member; /* node in
1113 * ->fl_blocker->fl_blocked_requests
1115 fl_owner_t fl_owner;
1116 unsigned int fl_flags;
1117 unsigned char fl_type;
1118 unsigned int fl_pid;
1119 int fl_link_cpu; /* what cpu's list is this on? */
1120 wait_queue_head_t fl_wait;
1121 struct file *fl_file;
1125 struct fasync_struct * fl_fasync; /* for lease break notifications */
1126 /* for lease breaks: */
1127 unsigned long fl_break_time;
1128 unsigned long fl_downgrade_time;
1130 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1131 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1133 struct nfs_lock_info nfs_fl;
1134 struct nfs4_lock_info nfs4_fl;
1136 struct list_head link; /* link in AFS vnode's pending_locks list */
1137 int state; /* state of grant or error if -ve */
1138 unsigned int debug_id;
1141 } __randomize_layout;
1143 struct file_lock_context {
1144 spinlock_t flc_lock;
1145 struct list_head flc_flock;
1146 struct list_head flc_posix;
1147 struct list_head flc_lease;
1150 /* The following constant reflects the upper bound of the file/locking space */
1152 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1153 #define OFFSET_MAX INT_LIMIT(loff_t)
1154 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1157 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1159 #define locks_inode(f) file_inode(f)
1161 #ifdef CONFIG_FILE_LOCKING
1162 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1163 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1166 #if BITS_PER_LONG == 32
1167 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1168 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1172 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1173 extern int fcntl_getlease(struct file *filp);
1176 void locks_free_lock_context(struct inode *inode);
1177 void locks_free_lock(struct file_lock *fl);
1178 extern void locks_init_lock(struct file_lock *);
1179 extern struct file_lock * locks_alloc_lock(void);
1180 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1181 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1182 extern void locks_remove_posix(struct file *, fl_owner_t);
1183 extern void locks_remove_file(struct file *);
1184 extern void locks_release_private(struct file_lock *);
1185 extern void posix_test_lock(struct file *, struct file_lock *);
1186 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1187 extern int locks_delete_block(struct file_lock *);
1188 extern int vfs_test_lock(struct file *, struct file_lock *);
1189 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1190 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1191 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1192 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1193 extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1194 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1195 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1196 extern int lease_modify(struct file_lock *, int, struct list_head *);
1198 struct notifier_block;
1199 extern int lease_register_notifier(struct notifier_block *);
1200 extern void lease_unregister_notifier(struct notifier_block *);
1202 struct files_struct;
1203 extern void show_fd_locks(struct seq_file *f,
1204 struct file *filp, struct files_struct *files);
1205 #else /* !CONFIG_FILE_LOCKING */
1206 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1207 struct flock __user *user)
1212 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1213 unsigned int cmd, struct flock __user *user)
1218 #if BITS_PER_LONG == 32
1219 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1220 struct flock64 __user *user)
1225 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1226 unsigned int cmd, struct flock64 __user *user)
1231 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1236 static inline int fcntl_getlease(struct file *filp)
1242 locks_free_lock_context(struct inode *inode)
1246 static inline void locks_init_lock(struct file_lock *fl)
1251 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1256 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1261 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1266 static inline void locks_remove_file(struct file *filp)
1271 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1276 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1277 struct file_lock *conflock)
1282 static inline int locks_delete_block(struct file_lock *waiter)
1287 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1292 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1293 struct file_lock *fl, struct file_lock *conf)
1298 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1303 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1308 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1313 static inline void lease_get_mtime(struct inode *inode,
1314 struct timespec64 *time)
1319 static inline int generic_setlease(struct file *filp, long arg,
1320 struct file_lock **flp, void **priv)
1325 static inline int vfs_setlease(struct file *filp, long arg,
1326 struct file_lock **lease, void **priv)
1331 static inline int lease_modify(struct file_lock *fl, int arg,
1332 struct list_head *dispose)
1337 struct files_struct;
1338 static inline void show_fd_locks(struct seq_file *f,
1339 struct file *filp, struct files_struct *files) {}
1340 #endif /* !CONFIG_FILE_LOCKING */
1342 static inline struct inode *file_inode(const struct file *f)
1347 static inline struct dentry *file_dentry(const struct file *file)
1349 return d_real(file->f_path.dentry, file_inode(file));
1352 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1354 return locks_lock_inode_wait(locks_inode(filp), fl);
1357 struct fasync_struct {
1361 struct fasync_struct *fa_next; /* singly linked list */
1362 struct file *fa_file;
1363 struct rcu_head fa_rcu;
1366 #define FASYNC_MAGIC 0x4601
1368 /* SMP safe fasync helpers: */
1369 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1370 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1371 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1372 extern struct fasync_struct *fasync_alloc(void);
1373 extern void fasync_free(struct fasync_struct *);
1375 /* can be called from interrupts */
1376 extern void kill_fasync(struct fasync_struct **, int, int);
1378 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1379 extern int f_setown(struct file *filp, unsigned long arg, int force);
1380 extern void f_delown(struct file *filp);
1381 extern pid_t f_getown(struct file *filp);
1382 extern int send_sigurg(struct fown_struct *fown);
1385 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1386 * represented in both.
1388 #define SB_RDONLY 1 /* Mount read-only */
1389 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1390 #define SB_NODEV 4 /* Disallow access to device special files */
1391 #define SB_NOEXEC 8 /* Disallow program execution */
1392 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1393 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1394 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1395 #define SB_NOATIME 1024 /* Do not update access times. */
1396 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1397 #define SB_SILENT 32768
1398 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1399 #define SB_INLINECRYPT (1<<17) /* Use blk-crypto for encrypted files */
1400 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1401 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1402 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1404 /* These sb flags are internal to the kernel */
1405 #define SB_SUBMOUNT (1<<26)
1406 #define SB_FORCE (1<<27)
1407 #define SB_NOSEC (1<<28)
1408 #define SB_BORN (1<<29)
1409 #define SB_ACTIVE (1<<30)
1410 #define SB_NOUSER (1<<31)
1412 /* These flags relate to encoding and casefolding */
1413 #define SB_ENC_STRICT_MODE_FL (1 << 0)
1415 #define sb_has_strict_encoding(sb) \
1416 (sb->s_encoding_flags & SB_ENC_STRICT_MODE_FL)
1422 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1423 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1424 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1425 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1426 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1429 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1430 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1431 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1432 #define SB_I_STABLE_WRITES 0x00000008 /* don't modify blks until WB is done */
1434 /* sb->s_iflags to limit user namespace mounts */
1435 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1436 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1437 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1439 #define SB_I_SKIP_SYNC 0x00000100 /* Skip superblock at global sync */
1441 /* Possible states of 'frozen' field */
1443 SB_UNFROZEN = 0, /* FS is unfrozen */
1444 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1445 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1446 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1447 * internal threads if needed) */
1448 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1451 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1454 int frozen; /* Is sb frozen? */
1455 wait_queue_head_t wait_unfrozen; /* wait for thaw */
1456 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1459 struct super_block {
1460 struct list_head s_list; /* Keep this first */
1461 dev_t s_dev; /* search index; _not_ kdev_t */
1462 unsigned char s_blocksize_bits;
1463 unsigned long s_blocksize;
1464 loff_t s_maxbytes; /* Max file size */
1465 struct file_system_type *s_type;
1466 const struct super_operations *s_op;
1467 const struct dquot_operations *dq_op;
1468 const struct quotactl_ops *s_qcop;
1469 const struct export_operations *s_export_op;
1470 unsigned long s_flags;
1471 unsigned long s_iflags; /* internal SB_I_* flags */
1472 unsigned long s_magic;
1473 struct dentry *s_root;
1474 struct rw_semaphore s_umount;
1477 #ifdef CONFIG_SECURITY
1480 const struct xattr_handler **s_xattr;
1481 #ifdef CONFIG_FS_ENCRYPTION
1482 const struct fscrypt_operations *s_cop;
1483 struct key *s_master_keys; /* master crypto keys in use */
1485 #ifdef CONFIG_FS_VERITY
1486 const struct fsverity_operations *s_vop;
1488 #ifdef CONFIG_UNICODE
1489 struct unicode_map *s_encoding;
1490 __u16 s_encoding_flags;
1492 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1493 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1494 struct block_device *s_bdev;
1495 struct backing_dev_info *s_bdi;
1496 struct mtd_info *s_mtd;
1497 struct hlist_node s_instances;
1498 unsigned int s_quota_types; /* Bitmask of supported quota types */
1499 struct quota_info s_dquot; /* Diskquota specific options */
1501 struct sb_writers s_writers;
1504 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1505 * s_fsnotify_marks together for cache efficiency. They are frequently
1506 * accessed and rarely modified.
1508 void *s_fs_info; /* Filesystem private info */
1510 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1512 /* Time limits for c/m/atime in seconds */
1513 time64_t s_time_min;
1514 time64_t s_time_max;
1515 #ifdef CONFIG_FSNOTIFY
1516 __u32 s_fsnotify_mask;
1517 struct fsnotify_mark_connector __rcu *s_fsnotify_marks;
1520 char s_id[32]; /* Informational name */
1521 uuid_t s_uuid; /* UUID */
1523 unsigned int s_max_links;
1527 * The next field is for VFS *only*. No filesystems have any business
1528 * even looking at it. You had been warned.
1530 struct mutex s_vfs_rename_mutex; /* Kludge */
1533 * Filesystem subtype. If non-empty the filesystem type field
1534 * in /proc/mounts will be "type.subtype"
1536 const char *s_subtype;
1538 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1541 * Saved pool identifier for cleancache (-1 means none)
1543 int cleancache_poolid;
1545 struct shrinker s_shrink; /* per-sb shrinker handle */
1547 /* Number of inodes with nlink == 0 but still referenced */
1548 atomic_long_t s_remove_count;
1551 * Number of inode/mount/sb objects that are being watched, note that
1552 * inodes objects are currently double-accounted.
1554 atomic_long_t s_fsnotify_connectors;
1556 /* Being remounted read-only */
1557 int s_readonly_remount;
1559 /* per-sb errseq_t for reporting writeback errors via syncfs */
1562 /* AIO completions deferred from interrupt context */
1563 struct workqueue_struct *s_dio_done_wq;
1564 struct hlist_head s_pins;
1567 * Owning user namespace and default context in which to
1568 * interpret filesystem uids, gids, quotas, device nodes,
1569 * xattrs and security labels.
1571 struct user_namespace *s_user_ns;
1574 * The list_lru structure is essentially just a pointer to a table
1575 * of per-node lru lists, each of which has its own spinlock.
1576 * There is no need to put them into separate cachelines.
1578 struct list_lru s_dentry_lru;
1579 struct list_lru s_inode_lru;
1580 struct rcu_head rcu;
1581 struct work_struct destroy_work;
1583 struct mutex s_sync_lock; /* sync serialisation lock */
1586 * Indicates how deep in a filesystem stack this SB is
1590 /* s_inode_list_lock protects s_inodes */
1591 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1592 struct list_head s_inodes; /* all inodes */
1594 spinlock_t s_inode_wblist_lock;
1595 struct list_head s_inodes_wb; /* writeback inodes */
1596 } __randomize_layout;
1598 /* Helper functions so that in most cases filesystems will
1599 * not need to deal directly with kuid_t and kgid_t and can
1600 * instead deal with the raw numeric values that are stored
1601 * in the filesystem.
1603 static inline uid_t i_uid_read(const struct inode *inode)
1605 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1608 static inline gid_t i_gid_read(const struct inode *inode)
1610 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1613 static inline void i_uid_write(struct inode *inode, uid_t uid)
1615 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1618 static inline void i_gid_write(struct inode *inode, gid_t gid)
1620 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1624 * kuid_into_mnt - map a kuid down into a mnt_userns
1625 * @mnt_userns: user namespace of the relevant mount
1626 * @kuid: kuid to be mapped
1628 * Return: @kuid mapped according to @mnt_userns.
1629 * If @kuid has no mapping INVALID_UID is returned.
1631 static inline kuid_t kuid_into_mnt(struct user_namespace *mnt_userns,
1634 return make_kuid(mnt_userns, __kuid_val(kuid));
1638 * kgid_into_mnt - map a kgid down into a mnt_userns
1639 * @mnt_userns: user namespace of the relevant mount
1640 * @kgid: kgid to be mapped
1642 * Return: @kgid mapped according to @mnt_userns.
1643 * If @kgid has no mapping INVALID_GID is returned.
1645 static inline kgid_t kgid_into_mnt(struct user_namespace *mnt_userns,
1648 return make_kgid(mnt_userns, __kgid_val(kgid));
1652 * i_uid_into_mnt - map an inode's i_uid down into a mnt_userns
1653 * @mnt_userns: user namespace of the mount the inode was found from
1654 * @inode: inode to map
1656 * Return: the inode's i_uid mapped down according to @mnt_userns.
1657 * If the inode's i_uid has no mapping INVALID_UID is returned.
1659 static inline kuid_t i_uid_into_mnt(struct user_namespace *mnt_userns,
1660 const struct inode *inode)
1662 return kuid_into_mnt(mnt_userns, inode->i_uid);
1666 * i_gid_into_mnt - map an inode's i_gid down into a mnt_userns
1667 * @mnt_userns: user namespace of the mount the inode was found from
1668 * @inode: inode to map
1670 * Return: the inode's i_gid mapped down according to @mnt_userns.
1671 * If the inode's i_gid has no mapping INVALID_GID is returned.
1673 static inline kgid_t i_gid_into_mnt(struct user_namespace *mnt_userns,
1674 const struct inode *inode)
1676 return kgid_into_mnt(mnt_userns, inode->i_gid);
1680 * kuid_from_mnt - map a kuid up into a mnt_userns
1681 * @mnt_userns: user namespace of the relevant mount
1682 * @kuid: kuid to be mapped
1684 * Return: @kuid mapped up according to @mnt_userns.
1685 * If @kuid has no mapping INVALID_UID is returned.
1687 static inline kuid_t kuid_from_mnt(struct user_namespace *mnt_userns,
1690 return KUIDT_INIT(from_kuid(mnt_userns, kuid));
1694 * kgid_from_mnt - map a kgid up into a mnt_userns
1695 * @mnt_userns: user namespace of the relevant mount
1696 * @kgid: kgid to be mapped
1698 * Return: @kgid mapped up according to @mnt_userns.
1699 * If @kgid has no mapping INVALID_GID is returned.
1701 static inline kgid_t kgid_from_mnt(struct user_namespace *mnt_userns,
1704 return KGIDT_INIT(from_kgid(mnt_userns, kgid));
1708 * mapped_fsuid - return caller's fsuid mapped up into a mnt_userns
1709 * @mnt_userns: user namespace of the relevant mount
1711 * Use this helper to initialize a new vfs or filesystem object based on
1712 * the caller's fsuid. A common example is initializing the i_uid field of
1713 * a newly allocated inode triggered by a creation event such as mkdir or
1714 * O_CREAT. Other examples include the allocation of quotas for a specific
1717 * Return: the caller's current fsuid mapped up according to @mnt_userns.
1719 static inline kuid_t mapped_fsuid(struct user_namespace *mnt_userns)
1721 return kuid_from_mnt(mnt_userns, current_fsuid());
1725 * mapped_fsgid - return caller's fsgid mapped up into a mnt_userns
1726 * @mnt_userns: user namespace of the relevant mount
1728 * Use this helper to initialize a new vfs or filesystem object based on
1729 * the caller's fsgid. A common example is initializing the i_gid field of
1730 * a newly allocated inode triggered by a creation event such as mkdir or
1731 * O_CREAT. Other examples include the allocation of quotas for a specific
1734 * Return: the caller's current fsgid mapped up according to @mnt_userns.
1736 static inline kgid_t mapped_fsgid(struct user_namespace *mnt_userns)
1738 return kgid_from_mnt(mnt_userns, current_fsgid());
1742 * inode_fsuid_set - initialize inode's i_uid field with callers fsuid
1743 * @inode: inode to initialize
1744 * @mnt_userns: user namespace of the mount the inode was found from
1746 * Initialize the i_uid field of @inode. If the inode was found/created via
1747 * an idmapped mount map the caller's fsuid according to @mnt_users.
1749 static inline void inode_fsuid_set(struct inode *inode,
1750 struct user_namespace *mnt_userns)
1752 inode->i_uid = mapped_fsuid(mnt_userns);
1756 * inode_fsgid_set - initialize inode's i_gid field with callers fsgid
1757 * @inode: inode to initialize
1758 * @mnt_userns: user namespace of the mount the inode was found from
1760 * Initialize the i_gid field of @inode. If the inode was found/created via
1761 * an idmapped mount map the caller's fsgid according to @mnt_users.
1763 static inline void inode_fsgid_set(struct inode *inode,
1764 struct user_namespace *mnt_userns)
1766 inode->i_gid = mapped_fsgid(mnt_userns);
1770 * fsuidgid_has_mapping() - check whether caller's fsuid/fsgid is mapped
1771 * @sb: the superblock we want a mapping in
1772 * @mnt_userns: user namespace of the relevant mount
1774 * Check whether the caller's fsuid and fsgid have a valid mapping in the
1775 * s_user_ns of the superblock @sb. If the caller is on an idmapped mount map
1776 * the caller's fsuid and fsgid according to the @mnt_userns first.
1778 * Return: true if fsuid and fsgid is mapped, false if not.
1780 static inline bool fsuidgid_has_mapping(struct super_block *sb,
1781 struct user_namespace *mnt_userns)
1783 struct user_namespace *s_user_ns = sb->s_user_ns;
1785 return kuid_has_mapping(s_user_ns, mapped_fsuid(mnt_userns)) &&
1786 kgid_has_mapping(s_user_ns, mapped_fsgid(mnt_userns));
1789 extern struct timespec64 current_time(struct inode *inode);
1792 * Snapshotting support.
1796 * These are internal functions, please use sb_start_{write,pagefault,intwrite}
1799 static inline void __sb_end_write(struct super_block *sb, int level)
1801 percpu_up_read(sb->s_writers.rw_sem + level-1);
1804 static inline void __sb_start_write(struct super_block *sb, int level)
1806 percpu_down_read(sb->s_writers.rw_sem + level - 1);
1809 static inline bool __sb_start_write_trylock(struct super_block *sb, int level)
1811 return percpu_down_read_trylock(sb->s_writers.rw_sem + level - 1);
1814 #define __sb_writers_acquired(sb, lev) \
1815 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1816 #define __sb_writers_release(sb, lev) \
1817 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1820 * sb_end_write - drop write access to a superblock
1821 * @sb: the super we wrote to
1823 * Decrement number of writers to the filesystem. Wake up possible waiters
1824 * wanting to freeze the filesystem.
1826 static inline void sb_end_write(struct super_block *sb)
1828 __sb_end_write(sb, SB_FREEZE_WRITE);
1832 * sb_end_pagefault - drop write access to a superblock from a page fault
1833 * @sb: the super we wrote to
1835 * Decrement number of processes handling write page fault to the filesystem.
1836 * Wake up possible waiters wanting to freeze the filesystem.
1838 static inline void sb_end_pagefault(struct super_block *sb)
1840 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1844 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1845 * @sb: the super we wrote to
1847 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1848 * waiters wanting to freeze the filesystem.
1850 static inline void sb_end_intwrite(struct super_block *sb)
1852 __sb_end_write(sb, SB_FREEZE_FS);
1856 * sb_start_write - get write access to a superblock
1857 * @sb: the super we write to
1859 * When a process wants to write data or metadata to a file system (i.e. dirty
1860 * a page or an inode), it should embed the operation in a sb_start_write() -
1861 * sb_end_write() pair to get exclusion against file system freezing. This
1862 * function increments number of writers preventing freezing. If the file
1863 * system is already frozen, the function waits until the file system is
1866 * Since freeze protection behaves as a lock, users have to preserve
1867 * ordering of freeze protection and other filesystem locks. Generally,
1868 * freeze protection should be the outermost lock. In particular, we have:
1871 * -> i_mutex (write path, truncate, directory ops, ...)
1872 * -> s_umount (freeze_super, thaw_super)
1874 static inline void sb_start_write(struct super_block *sb)
1876 __sb_start_write(sb, SB_FREEZE_WRITE);
1879 static inline bool sb_start_write_trylock(struct super_block *sb)
1881 return __sb_start_write_trylock(sb, SB_FREEZE_WRITE);
1885 * sb_start_pagefault - get write access to a superblock from a page fault
1886 * @sb: the super we write to
1888 * When a process starts handling write page fault, it should embed the
1889 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1890 * exclusion against file system freezing. This is needed since the page fault
1891 * is going to dirty a page. This function increments number of running page
1892 * faults preventing freezing. If the file system is already frozen, the
1893 * function waits until the file system is thawed.
1895 * Since page fault freeze protection behaves as a lock, users have to preserve
1896 * ordering of freeze protection and other filesystem locks. It is advised to
1897 * put sb_start_pagefault() close to mmap_lock in lock ordering. Page fault
1898 * handling code implies lock dependency:
1901 * -> sb_start_pagefault
1903 static inline void sb_start_pagefault(struct super_block *sb)
1905 __sb_start_write(sb, SB_FREEZE_PAGEFAULT);
1909 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1910 * @sb: the super we write to
1912 * This is the third level of protection against filesystem freezing. It is
1913 * free for use by a filesystem. The only requirement is that it must rank
1914 * below sb_start_pagefault.
1916 * For example filesystem can call sb_start_intwrite() when starting a
1917 * transaction which somewhat eases handling of freezing for internal sources
1918 * of filesystem changes (internal fs threads, discarding preallocation on file
1921 static inline void sb_start_intwrite(struct super_block *sb)
1923 __sb_start_write(sb, SB_FREEZE_FS);
1926 static inline bool sb_start_intwrite_trylock(struct super_block *sb)
1928 return __sb_start_write_trylock(sb, SB_FREEZE_FS);
1931 bool inode_owner_or_capable(struct user_namespace *mnt_userns,
1932 const struct inode *inode);
1935 * VFS helper functions..
1937 int vfs_create(struct user_namespace *, struct inode *,
1938 struct dentry *, umode_t, bool);
1939 int vfs_mkdir(struct user_namespace *, struct inode *,
1940 struct dentry *, umode_t);
1941 int vfs_mknod(struct user_namespace *, struct inode *, struct dentry *,
1943 int vfs_symlink(struct user_namespace *, struct inode *,
1944 struct dentry *, const char *);
1945 int vfs_link(struct dentry *, struct user_namespace *, struct inode *,
1946 struct dentry *, struct inode **);
1947 int vfs_rmdir(struct user_namespace *, struct inode *, struct dentry *);
1948 int vfs_unlink(struct user_namespace *, struct inode *, struct dentry *,
1952 * struct renamedata - contains all information required for renaming
1953 * @old_mnt_userns: old user namespace of the mount the inode was found from
1954 * @old_dir: parent of source
1955 * @old_dentry: source
1956 * @new_mnt_userns: new user namespace of the mount the inode was found from
1957 * @new_dir: parent of destination
1958 * @new_dentry: destination
1959 * @delegated_inode: returns an inode needing a delegation break
1960 * @flags: rename flags
1963 struct user_namespace *old_mnt_userns;
1964 struct inode *old_dir;
1965 struct dentry *old_dentry;
1966 struct user_namespace *new_mnt_userns;
1967 struct inode *new_dir;
1968 struct dentry *new_dentry;
1969 struct inode **delegated_inode;
1971 } __randomize_layout;
1973 int vfs_rename(struct renamedata *);
1975 static inline int vfs_whiteout(struct user_namespace *mnt_userns,
1976 struct inode *dir, struct dentry *dentry)
1978 return vfs_mknod(mnt_userns, dir, dentry, S_IFCHR | WHITEOUT_MODE,
1982 struct dentry *vfs_tmpfile(struct user_namespace *mnt_userns,
1983 struct dentry *dentry, umode_t mode, int open_flag);
1985 int vfs_mkobj(struct dentry *, umode_t,
1986 int (*f)(struct dentry *, umode_t, void *),
1989 int vfs_fchown(struct file *file, uid_t user, gid_t group);
1990 int vfs_fchmod(struct file *file, umode_t mode);
1991 int vfs_utimes(const struct path *path, struct timespec64 *times);
1993 extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1995 #ifdef CONFIG_COMPAT
1996 extern long compat_ptr_ioctl(struct file *file, unsigned int cmd,
1999 #define compat_ptr_ioctl NULL
2003 * VFS file helper functions.
2005 void inode_init_owner(struct user_namespace *mnt_userns, struct inode *inode,
2006 const struct inode *dir, umode_t mode);
2007 extern bool may_open_dev(const struct path *path);
2010 * This is the "filldir" function type, used by readdir() to let
2011 * the kernel specify what kind of dirent layout it wants to have.
2012 * This allows the kernel to read directories into kernel space or
2013 * to have different dirent layouts depending on the binary type.
2016 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
2019 struct dir_context {
2025 * These flags let !MMU mmap() govern direct device mapping vs immediate
2026 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
2028 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
2029 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
2030 * NOMMU_MAP_READ: Can be mapped for reading
2031 * NOMMU_MAP_WRITE: Can be mapped for writing
2032 * NOMMU_MAP_EXEC: Can be mapped for execution
2034 #define NOMMU_MAP_COPY 0x00000001
2035 #define NOMMU_MAP_DIRECT 0x00000008
2036 #define NOMMU_MAP_READ VM_MAYREAD
2037 #define NOMMU_MAP_WRITE VM_MAYWRITE
2038 #define NOMMU_MAP_EXEC VM_MAYEXEC
2040 #define NOMMU_VMFLAGS \
2041 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
2044 * These flags control the behavior of the remap_file_range function pointer.
2045 * If it is called with len == 0 that means "remap to end of source file".
2046 * See Documentation/filesystems/vfs.rst for more details about this call.
2048 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
2049 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
2051 #define REMAP_FILE_DEDUP (1 << 0)
2052 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
2055 * These flags signal that the caller is ok with altering various aspects of
2056 * the behavior of the remap operation. The changes must be made by the
2057 * implementation; the vfs remap helper functions can take advantage of them.
2058 * Flags in this category exist to preserve the quirky behavior of the hoisted
2059 * btrfs clone/dedupe ioctls.
2061 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
2065 struct file_operations {
2066 struct module *owner;
2067 loff_t (*llseek) (struct file *, loff_t, int);
2068 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
2069 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
2070 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
2071 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
2072 int (*iopoll)(struct kiocb *kiocb, bool spin);
2073 int (*iterate) (struct file *, struct dir_context *);
2074 int (*iterate_shared) (struct file *, struct dir_context *);
2075 __poll_t (*poll) (struct file *, struct poll_table_struct *);
2076 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
2077 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
2078 int (*mmap) (struct file *, struct vm_area_struct *);
2079 unsigned long mmap_supported_flags;
2080 int (*open) (struct inode *, struct file *);
2081 int (*flush) (struct file *, fl_owner_t id);
2082 int (*release) (struct inode *, struct file *);
2083 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
2084 int (*fasync) (int, struct file *, int);
2085 int (*lock) (struct file *, int, struct file_lock *);
2086 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
2087 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
2088 int (*check_flags)(int);
2089 int (*flock) (struct file *, int, struct file_lock *);
2090 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
2091 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
2092 int (*setlease)(struct file *, long, struct file_lock **, void **);
2093 long (*fallocate)(struct file *file, int mode, loff_t offset,
2095 void (*show_fdinfo)(struct seq_file *m, struct file *f);
2097 unsigned (*mmap_capabilities)(struct file *);
2099 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
2100 loff_t, size_t, unsigned int);
2101 loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
2102 struct file *file_out, loff_t pos_out,
2103 loff_t len, unsigned int remap_flags);
2104 int (*fadvise)(struct file *, loff_t, loff_t, int);
2105 } __randomize_layout;
2107 struct inode_operations {
2108 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
2109 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
2110 int (*permission) (struct user_namespace *, struct inode *, int);
2111 struct posix_acl * (*get_acl)(struct inode *, int);
2113 int (*readlink) (struct dentry *, char __user *,int);
2115 int (*create) (struct user_namespace *, struct inode *,struct dentry *,
2117 int (*link) (struct dentry *,struct inode *,struct dentry *);
2118 int (*unlink) (struct inode *,struct dentry *);
2119 int (*symlink) (struct user_namespace *, struct inode *,struct dentry *,
2121 int (*mkdir) (struct user_namespace *, struct inode *,struct dentry *,
2123 int (*rmdir) (struct inode *,struct dentry *);
2124 int (*mknod) (struct user_namespace *, struct inode *,struct dentry *,
2126 int (*rename) (struct user_namespace *, struct inode *, struct dentry *,
2127 struct inode *, struct dentry *, unsigned int);
2128 int (*setattr) (struct user_namespace *, struct dentry *,
2130 int (*getattr) (struct user_namespace *, const struct path *,
2131 struct kstat *, u32, unsigned int);
2132 ssize_t (*listxattr) (struct dentry *, char *, size_t);
2133 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
2135 int (*update_time)(struct inode *, struct timespec64 *, int);
2136 int (*atomic_open)(struct inode *, struct dentry *,
2137 struct file *, unsigned open_flag,
2138 umode_t create_mode);
2139 int (*tmpfile) (struct user_namespace *, struct inode *,
2140 struct dentry *, umode_t);
2141 int (*set_acl)(struct user_namespace *, struct inode *,
2142 struct posix_acl *, int);
2143 int (*fileattr_set)(struct user_namespace *mnt_userns,
2144 struct dentry *dentry, struct fileattr *fa);
2145 int (*fileattr_get)(struct dentry *dentry, struct fileattr *fa);
2146 } ____cacheline_aligned;
2148 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
2149 struct iov_iter *iter)
2151 return file->f_op->read_iter(kio, iter);
2154 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
2155 struct iov_iter *iter)
2157 return file->f_op->write_iter(kio, iter);
2160 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
2162 return file->f_op->mmap(file, vma);
2165 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
2166 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
2167 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
2168 loff_t, size_t, unsigned int);
2169 extern ssize_t generic_copy_file_range(struct file *file_in, loff_t pos_in,
2170 struct file *file_out, loff_t pos_out,
2171 size_t len, unsigned int flags);
2172 extern int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
2173 struct file *file_out, loff_t pos_out,
2175 unsigned int remap_flags);
2176 extern loff_t do_clone_file_range(struct file *file_in, loff_t pos_in,
2177 struct file *file_out, loff_t pos_out,
2178 loff_t len, unsigned int remap_flags);
2179 extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in,
2180 struct file *file_out, loff_t pos_out,
2181 loff_t len, unsigned int remap_flags);
2182 extern int vfs_dedupe_file_range(struct file *file,
2183 struct file_dedupe_range *same);
2184 extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
2185 struct file *dst_file, loff_t dst_pos,
2186 loff_t len, unsigned int remap_flags);
2189 struct super_operations {
2190 struct inode *(*alloc_inode)(struct super_block *sb);
2191 void (*destroy_inode)(struct inode *);
2192 void (*free_inode)(struct inode *);
2194 void (*dirty_inode) (struct inode *, int flags);
2195 int (*write_inode) (struct inode *, struct writeback_control *wbc);
2196 int (*drop_inode) (struct inode *);
2197 void (*evict_inode) (struct inode *);
2198 void (*put_super) (struct super_block *);
2199 int (*sync_fs)(struct super_block *sb, int wait);
2200 int (*freeze_super) (struct super_block *);
2201 int (*freeze_fs) (struct super_block *);
2202 int (*thaw_super) (struct super_block *);
2203 int (*unfreeze_fs) (struct super_block *);
2204 int (*statfs) (struct dentry *, struct kstatfs *);
2205 int (*remount_fs) (struct super_block *, int *, char *);
2206 void (*umount_begin) (struct super_block *);
2208 int (*show_options)(struct seq_file *, struct dentry *);
2209 int (*show_devname)(struct seq_file *, struct dentry *);
2210 int (*show_path)(struct seq_file *, struct dentry *);
2211 int (*show_stats)(struct seq_file *, struct dentry *);
2213 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
2214 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
2215 struct dquot **(*get_dquots)(struct inode *);
2217 long (*nr_cached_objects)(struct super_block *,
2218 struct shrink_control *);
2219 long (*free_cached_objects)(struct super_block *,
2220 struct shrink_control *);
2224 * Inode flags - they have no relation to superblock flags now
2226 #define S_SYNC (1 << 0) /* Writes are synced at once */
2227 #define S_NOATIME (1 << 1) /* Do not update access times */
2228 #define S_APPEND (1 << 2) /* Append-only file */
2229 #define S_IMMUTABLE (1 << 3) /* Immutable file */
2230 #define S_DEAD (1 << 4) /* removed, but still open directory */
2231 #define S_NOQUOTA (1 << 5) /* Inode is not counted to quota */
2232 #define S_DIRSYNC (1 << 6) /* Directory modifications are synchronous */
2233 #define S_NOCMTIME (1 << 7) /* Do not update file c/mtime */
2234 #define S_SWAPFILE (1 << 8) /* Do not truncate: swapon got its bmaps */
2235 #define S_PRIVATE (1 << 9) /* Inode is fs-internal */
2236 #define S_IMA (1 << 10) /* Inode has an associated IMA struct */
2237 #define S_AUTOMOUNT (1 << 11) /* Automount/referral quasi-directory */
2238 #define S_NOSEC (1 << 12) /* no suid or xattr security attributes */
2239 #ifdef CONFIG_FS_DAX
2240 #define S_DAX (1 << 13) /* Direct Access, avoiding the page cache */
2242 #define S_DAX 0 /* Make all the DAX code disappear */
2244 #define S_ENCRYPTED (1 << 14) /* Encrypted file (using fs/crypto/) */
2245 #define S_CASEFOLD (1 << 15) /* Casefolded file */
2246 #define S_VERITY (1 << 16) /* Verity file (using fs/verity/) */
2249 * Note that nosuid etc flags are inode-specific: setting some file-system
2250 * flags just means all the inodes inherit those flags by default. It might be
2251 * possible to override it selectively if you really wanted to with some
2252 * ioctl() that is not currently implemented.
2254 * Exception: SB_RDONLY is always applied to the entire file system.
2256 * Unfortunately, it is possible to change a filesystems flags with it mounted
2257 * with files in use. This means that all of the inodes will not have their
2258 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
2259 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
2261 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
2263 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
2264 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
2265 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
2266 ((inode)->i_flags & S_SYNC))
2267 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
2268 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
2269 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
2270 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
2271 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
2273 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
2274 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
2275 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
2276 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
2278 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
2279 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
2280 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
2281 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
2282 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
2283 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2284 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2285 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2286 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2287 #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD)
2288 #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY)
2290 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2291 (inode)->i_rdev == WHITEOUT_DEV)
2293 static inline bool HAS_UNMAPPED_ID(struct user_namespace *mnt_userns,
2294 struct inode *inode)
2296 return !uid_valid(i_uid_into_mnt(mnt_userns, inode)) ||
2297 !gid_valid(i_gid_into_mnt(mnt_userns, inode));
2300 static inline enum rw_hint file_write_hint(struct file *file)
2302 if (file->f_write_hint != WRITE_LIFE_NOT_SET)
2303 return file->f_write_hint;
2305 return file_inode(file)->i_write_hint;
2308 static inline int iocb_flags(struct file *file);
2310 static inline u16 ki_hint_validate(enum rw_hint hint)
2312 typeof(((struct kiocb *)0)->ki_hint) max_hint = -1;
2314 if (hint <= max_hint)
2319 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
2321 *kiocb = (struct kiocb) {
2323 .ki_flags = iocb_flags(filp),
2324 .ki_hint = ki_hint_validate(file_write_hint(filp)),
2325 .ki_ioprio = get_current_ioprio(),
2329 static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src,
2332 *kiocb = (struct kiocb) {
2334 .ki_flags = kiocb_src->ki_flags,
2335 .ki_hint = kiocb_src->ki_hint,
2336 .ki_ioprio = kiocb_src->ki_ioprio,
2337 .ki_pos = kiocb_src->ki_pos,
2342 * Inode state bits. Protected by inode->i_lock
2344 * Four bits determine the dirty state of the inode: I_DIRTY_SYNC,
2345 * I_DIRTY_DATASYNC, I_DIRTY_PAGES, and I_DIRTY_TIME.
2347 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2348 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2349 * various stages of removing an inode.
2351 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2353 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2354 * fdatasync() (unless I_DIRTY_DATASYNC is also set).
2355 * Timestamp updates are the usual cause.
2356 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2357 * these changes separately from I_DIRTY_SYNC so that we
2358 * don't have to write inode on fdatasync() when only
2359 * e.g. the timestamps have changed.
2360 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2361 * I_DIRTY_TIME The inode itself only has dirty timestamps, and the
2362 * lazytime mount option is enabled. We keep track of this
2363 * separately from I_DIRTY_SYNC in order to implement
2364 * lazytime. This gets cleared if I_DIRTY_INODE
2365 * (I_DIRTY_SYNC and/or I_DIRTY_DATASYNC) gets set. I.e.
2366 * either I_DIRTY_TIME *or* I_DIRTY_INODE can be set in
2367 * i_state, but not both. I_DIRTY_PAGES may still be set.
2368 * I_NEW Serves as both a mutex and completion notification.
2369 * New inodes set I_NEW. If two processes both create
2370 * the same inode, one of them will release its inode and
2371 * wait for I_NEW to be released before returning.
2372 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2373 * also cause waiting on I_NEW, without I_NEW actually
2374 * being set. find_inode() uses this to prevent returning
2375 * nearly-dead inodes.
2376 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2377 * is zero. I_FREEING must be set when I_WILL_FREE is
2379 * I_FREEING Set when inode is about to be freed but still has dirty
2380 * pages or buffers attached or the inode itself is still
2382 * I_CLEAR Added by clear_inode(). In this state the inode is
2383 * clean and can be destroyed. Inode keeps I_FREEING.
2385 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2386 * prohibited for many purposes. iget() must wait for
2387 * the inode to be completely released, then create it
2388 * anew. Other functions will just ignore such inodes,
2389 * if appropriate. I_NEW is used for waiting.
2391 * I_SYNC Writeback of inode is running. The bit is set during
2392 * data writeback, and cleared with a wakeup on the bit
2393 * address once it is done. The bit is also used to pin
2394 * the inode in memory for flusher thread.
2396 * I_REFERENCED Marks the inode as recently references on the LRU list.
2398 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2400 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2401 * synchronize competing switching instances and to tell
2402 * wb stat updates to grab the i_pages lock. See
2403 * inode_switch_wbs_work_fn() for details.
2405 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2406 * and work dirs among overlayfs mounts.
2408 * I_CREATING New object's inode in the middle of setting up.
2410 * I_DONTCACHE Evict inode as soon as it is not used anymore.
2412 * I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists.
2413 * Used to detect that mark_inode_dirty() should not move
2414 * inode between dirty lists.
2416 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2418 #define I_DIRTY_SYNC (1 << 0)
2419 #define I_DIRTY_DATASYNC (1 << 1)
2420 #define I_DIRTY_PAGES (1 << 2)
2422 #define I_NEW (1 << __I_NEW)
2423 #define I_WILL_FREE (1 << 4)
2424 #define I_FREEING (1 << 5)
2425 #define I_CLEAR (1 << 6)
2427 #define I_SYNC (1 << __I_SYNC)
2428 #define I_REFERENCED (1 << 8)
2429 #define __I_DIO_WAKEUP 9
2430 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2431 #define I_LINKABLE (1 << 10)
2432 #define I_DIRTY_TIME (1 << 11)
2433 #define I_WB_SWITCH (1 << 13)
2434 #define I_OVL_INUSE (1 << 14)
2435 #define I_CREATING (1 << 15)
2436 #define I_DONTCACHE (1 << 16)
2437 #define I_SYNC_QUEUED (1 << 17)
2439 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2440 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2441 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2443 extern void __mark_inode_dirty(struct inode *, int);
2444 static inline void mark_inode_dirty(struct inode *inode)
2446 __mark_inode_dirty(inode, I_DIRTY);
2449 static inline void mark_inode_dirty_sync(struct inode *inode)
2451 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2455 * Returns true if the given inode itself only has dirty timestamps (its pages
2456 * may still be dirty) and isn't currently being allocated or freed.
2457 * Filesystems should call this if when writing an inode when lazytime is
2458 * enabled, they want to opportunistically write the timestamps of other inodes
2459 * located very nearby on-disk, e.g. in the same inode block. This returns true
2460 * if the given inode is in need of such an opportunistic update. Requires
2461 * i_lock, or at least later re-checking under i_lock.
2463 static inline bool inode_is_dirtytime_only(struct inode *inode)
2465 return (inode->i_state & (I_DIRTY_TIME | I_NEW |
2466 I_FREEING | I_WILL_FREE)) == I_DIRTY_TIME;
2469 extern void inc_nlink(struct inode *inode);
2470 extern void drop_nlink(struct inode *inode);
2471 extern void clear_nlink(struct inode *inode);
2472 extern void set_nlink(struct inode *inode, unsigned int nlink);
2474 static inline void inode_inc_link_count(struct inode *inode)
2477 mark_inode_dirty(inode);
2480 static inline void inode_dec_link_count(struct inode *inode)
2483 mark_inode_dirty(inode);
2486 enum file_time_flags {
2493 extern bool atime_needs_update(const struct path *, struct inode *);
2494 extern void touch_atime(const struct path *);
2495 static inline void file_accessed(struct file *file)
2497 if (!(file->f_flags & O_NOATIME))
2498 touch_atime(&file->f_path);
2501 extern int file_modified(struct file *file);
2503 int sync_inode_metadata(struct inode *inode, int wait);
2505 struct file_system_type {
2508 #define FS_REQUIRES_DEV 1
2509 #define FS_BINARY_MOUNTDATA 2
2510 #define FS_HAS_SUBTYPE 4
2511 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2512 #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */
2513 #define FS_ALLOW_IDMAP 32 /* FS has been updated to handle vfs idmappings. */
2514 #define FS_THP_SUPPORT 8192 /* Remove once all fs converted */
2515 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2516 int (*init_fs_context)(struct fs_context *);
2517 const struct fs_parameter_spec *parameters;
2518 struct dentry *(*mount) (struct file_system_type *, int,
2519 const char *, void *);
2520 void (*kill_sb) (struct super_block *);
2521 struct module *owner;
2522 struct file_system_type * next;
2523 struct hlist_head fs_supers;
2525 struct lock_class_key s_lock_key;
2526 struct lock_class_key s_umount_key;
2527 struct lock_class_key s_vfs_rename_key;
2528 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2530 struct lock_class_key i_lock_key;
2531 struct lock_class_key i_mutex_key;
2532 struct lock_class_key invalidate_lock_key;
2533 struct lock_class_key i_mutex_dir_key;
2536 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2538 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2539 int flags, const char *dev_name, void *data,
2540 int (*fill_super)(struct super_block *, void *, int));
2541 extern struct dentry *mount_single(struct file_system_type *fs_type,
2542 int flags, void *data,
2543 int (*fill_super)(struct super_block *, void *, int));
2544 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2545 int flags, void *data,
2546 int (*fill_super)(struct super_block *, void *, int));
2547 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2548 void generic_shutdown_super(struct super_block *sb);
2549 void kill_block_super(struct super_block *sb);
2550 void kill_anon_super(struct super_block *sb);
2551 void kill_litter_super(struct super_block *sb);
2552 void deactivate_super(struct super_block *sb);
2553 void deactivate_locked_super(struct super_block *sb);
2554 int set_anon_super(struct super_block *s, void *data);
2555 int set_anon_super_fc(struct super_block *s, struct fs_context *fc);
2556 int get_anon_bdev(dev_t *);
2557 void free_anon_bdev(dev_t);
2558 struct super_block *sget_fc(struct fs_context *fc,
2559 int (*test)(struct super_block *, struct fs_context *),
2560 int (*set)(struct super_block *, struct fs_context *));
2561 struct super_block *sget(struct file_system_type *type,
2562 int (*test)(struct super_block *,void *),
2563 int (*set)(struct super_block *,void *),
2564 int flags, void *data);
2566 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2567 #define fops_get(fops) \
2568 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2569 #define fops_put(fops) \
2570 do { if (fops) module_put((fops)->owner); } while(0)
2572 * This one is to be used *ONLY* from ->open() instances.
2573 * fops must be non-NULL, pinned down *and* module dependencies
2574 * should be sufficient to pin the caller down as well.
2576 #define replace_fops(f, fops) \
2578 struct file *__file = (f); \
2579 fops_put(__file->f_op); \
2580 BUG_ON(!(__file->f_op = (fops))); \
2583 extern int register_filesystem(struct file_system_type *);
2584 extern int unregister_filesystem(struct file_system_type *);
2585 extern struct vfsmount *kern_mount(struct file_system_type *);
2586 extern void kern_unmount(struct vfsmount *mnt);
2587 extern int may_umount_tree(struct vfsmount *);
2588 extern int may_umount(struct vfsmount *);
2589 extern long do_mount(const char *, const char __user *,
2590 const char *, unsigned long, void *);
2591 extern struct vfsmount *collect_mounts(const struct path *);
2592 extern void drop_collected_mounts(struct vfsmount *);
2593 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2595 extern int vfs_statfs(const struct path *, struct kstatfs *);
2596 extern int user_statfs(const char __user *, struct kstatfs *);
2597 extern int fd_statfs(int, struct kstatfs *);
2598 extern int freeze_super(struct super_block *super);
2599 extern int thaw_super(struct super_block *super);
2600 extern bool our_mnt(struct vfsmount *mnt);
2601 extern __printf(2, 3)
2602 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2603 extern int super_setup_bdi(struct super_block *sb);
2605 extern int current_umask(void);
2607 extern void ihold(struct inode * inode);
2608 extern void iput(struct inode *);
2609 extern int generic_update_time(struct inode *, struct timespec64 *, int);
2612 extern struct kobject *fs_kobj;
2614 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2616 #ifdef CONFIG_FILE_LOCKING
2617 static inline int break_lease(struct inode *inode, unsigned int mode)
2620 * Since this check is lockless, we must ensure that any refcounts
2621 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2622 * could end up racing with tasks trying to set a new lease on this
2626 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2627 return __break_lease(inode, mode, FL_LEASE);
2631 static inline int break_deleg(struct inode *inode, unsigned int mode)
2634 * Since this check is lockless, we must ensure that any refcounts
2635 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2636 * could end up racing with tasks trying to set a new lease on this
2640 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2641 return __break_lease(inode, mode, FL_DELEG);
2645 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2649 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2650 if (ret == -EWOULDBLOCK && delegated_inode) {
2651 *delegated_inode = inode;
2657 static inline int break_deleg_wait(struct inode **delegated_inode)
2661 ret = break_deleg(*delegated_inode, O_WRONLY);
2662 iput(*delegated_inode);
2663 *delegated_inode = NULL;
2667 static inline int break_layout(struct inode *inode, bool wait)
2670 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2671 return __break_lease(inode,
2672 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2677 #else /* !CONFIG_FILE_LOCKING */
2678 static inline int break_lease(struct inode *inode, unsigned int mode)
2683 static inline int break_deleg(struct inode *inode, unsigned int mode)
2688 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2693 static inline int break_deleg_wait(struct inode **delegated_inode)
2699 static inline int break_layout(struct inode *inode, bool wait)
2704 #endif /* CONFIG_FILE_LOCKING */
2709 const char *name; /* pointer to actual string */
2710 const __user char *uptr; /* original userland pointer */
2712 struct audit_names *aname;
2715 static_assert(offsetof(struct filename, iname) % sizeof(long) == 0);
2717 static inline struct user_namespace *file_mnt_user_ns(struct file *file)
2719 return mnt_user_ns(file->f_path.mnt);
2721 extern long vfs_truncate(const struct path *, loff_t);
2722 int do_truncate(struct user_namespace *, struct dentry *, loff_t start,
2723 unsigned int time_attrs, struct file *filp);
2724 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2726 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2728 extern struct file *file_open_name(struct filename *, int, umode_t);
2729 extern struct file *filp_open(const char *, int, umode_t);
2730 extern struct file *file_open_root(const struct path *,
2731 const char *, int, umode_t);
2732 static inline struct file *file_open_root_mnt(struct vfsmount *mnt,
2733 const char *name, int flags, umode_t mode)
2735 return file_open_root(&(struct path){.mnt = mnt, .dentry = mnt->mnt_root},
2738 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2739 extern struct file * open_with_fake_path(const struct path *, int,
2740 struct inode*, const struct cred *);
2741 static inline struct file *file_clone_open(struct file *file)
2743 return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2745 extern int filp_close(struct file *, fl_owner_t id);
2747 extern struct filename *getname_flags(const char __user *, int, int *);
2748 extern struct filename *getname_uflags(const char __user *, int);
2749 extern struct filename *getname(const char __user *);
2750 extern struct filename *getname_kernel(const char *);
2751 extern void putname(struct filename *name);
2753 extern int finish_open(struct file *file, struct dentry *dentry,
2754 int (*open)(struct inode *, struct file *));
2755 extern int finish_no_open(struct file *file, struct dentry *dentry);
2758 extern void __init vfs_caches_init_early(void);
2759 extern void __init vfs_caches_init(void);
2761 extern struct kmem_cache *names_cachep;
2763 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2764 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2766 extern struct super_block *blockdev_superblock;
2767 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2769 return IS_ENABLED(CONFIG_BLOCK) && sb == blockdev_superblock;
2772 void emergency_thaw_all(void);
2773 extern int sync_filesystem(struct super_block *);
2774 extern const struct file_operations def_blk_fops;
2775 extern const struct file_operations def_chr_fops;
2778 #define CHRDEV_MAJOR_MAX 512
2779 /* Marks the bottom of the first segment of free char majors */
2780 #define CHRDEV_MAJOR_DYN_END 234
2781 /* Marks the top and bottom of the second segment of free char majors */
2782 #define CHRDEV_MAJOR_DYN_EXT_START 511
2783 #define CHRDEV_MAJOR_DYN_EXT_END 384
2785 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2786 extern int register_chrdev_region(dev_t, unsigned, const char *);
2787 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2788 unsigned int count, const char *name,
2789 const struct file_operations *fops);
2790 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2791 unsigned int count, const char *name);
2792 extern void unregister_chrdev_region(dev_t, unsigned);
2793 extern void chrdev_show(struct seq_file *,off_t);
2795 static inline int register_chrdev(unsigned int major, const char *name,
2796 const struct file_operations *fops)
2798 return __register_chrdev(major, 0, 256, name, fops);
2801 static inline void unregister_chrdev(unsigned int major, const char *name)
2803 __unregister_chrdev(major, 0, 256, name);
2806 extern void init_special_inode(struct inode *, umode_t, dev_t);
2808 /* Invalid inode operations -- fs/bad_inode.c */
2809 extern void make_bad_inode(struct inode *);
2810 extern bool is_bad_inode(struct inode *);
2812 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2813 pgoff_t start, pgoff_t end);
2815 void invalidate_mapping_pagevec(struct address_space *mapping,
2816 pgoff_t start, pgoff_t end,
2817 unsigned long *nr_pagevec);
2819 static inline void invalidate_remote_inode(struct inode *inode)
2821 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2822 S_ISLNK(inode->i_mode))
2823 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2825 extern int invalidate_inode_pages2(struct address_space *mapping);
2826 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2827 pgoff_t start, pgoff_t end);
2828 extern int write_inode_now(struct inode *, int);
2829 extern int filemap_fdatawrite(struct address_space *);
2830 extern int filemap_flush(struct address_space *);
2831 extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2832 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2834 extern int filemap_fdatawait_range_keep_errors(struct address_space *mapping,
2835 loff_t start_byte, loff_t end_byte);
2837 static inline int filemap_fdatawait(struct address_space *mapping)
2839 return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2842 extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2844 extern bool filemap_range_needs_writeback(struct address_space *,
2845 loff_t lstart, loff_t lend);
2846 extern int filemap_write_and_wait_range(struct address_space *mapping,
2847 loff_t lstart, loff_t lend);
2848 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2849 loff_t start, loff_t end, int sync_mode);
2850 extern int filemap_fdatawrite_range(struct address_space *mapping,
2851 loff_t start, loff_t end);
2852 extern int filemap_check_errors(struct address_space *mapping);
2853 extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2854 int filemap_fdatawrite_wbc(struct address_space *mapping,
2855 struct writeback_control *wbc);
2857 static inline int filemap_write_and_wait(struct address_space *mapping)
2859 return filemap_write_and_wait_range(mapping, 0, LLONG_MAX);
2862 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2864 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2865 extern int __must_check file_write_and_wait_range(struct file *file,
2866 loff_t start, loff_t end);
2868 static inline int file_write_and_wait(struct file *file)
2870 return file_write_and_wait_range(file, 0, LLONG_MAX);
2874 * filemap_set_wb_err - set a writeback error on an address_space
2875 * @mapping: mapping in which to set writeback error
2876 * @err: error to be set in mapping
2878 * When writeback fails in some way, we must record that error so that
2879 * userspace can be informed when fsync and the like are called. We endeavor
2880 * to report errors on any file that was open at the time of the error. Some
2881 * internal callers also need to know when writeback errors have occurred.
2883 * When a writeback error occurs, most filesystems will want to call
2884 * filemap_set_wb_err to record the error in the mapping so that it will be
2885 * automatically reported whenever fsync is called on the file.
2887 static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2889 /* Fastpath for common case of no error */
2891 __filemap_set_wb_err(mapping, err);
2895 * filemap_check_wb_err - has an error occurred since the mark was sampled?
2896 * @mapping: mapping to check for writeback errors
2897 * @since: previously-sampled errseq_t
2899 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2900 * the given value was sampled.
2902 * If it has then report the latest error set, otherwise return 0.
2904 static inline int filemap_check_wb_err(struct address_space *mapping,
2907 return errseq_check(&mapping->wb_err, since);
2911 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2912 * @mapping: mapping to be sampled
2914 * Writeback errors are always reported relative to a particular sample point
2915 * in the past. This function provides those sample points.
2917 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2919 return errseq_sample(&mapping->wb_err);
2923 * file_sample_sb_err - sample the current errseq_t to test for later errors
2924 * @file: file pointer to be sampled
2926 * Grab the most current superblock-level errseq_t value for the given
2929 static inline errseq_t file_sample_sb_err(struct file *file)
2931 return errseq_sample(&file->f_path.dentry->d_sb->s_wb_err);
2934 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2936 extern int vfs_fsync(struct file *file, int datasync);
2938 extern int sync_file_range(struct file *file, loff_t offset, loff_t nbytes,
2939 unsigned int flags);
2942 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2943 * to already be updated for the write, and will return either the amount
2944 * of bytes passed in, or an error if syncing the file failed.
2946 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2948 if (iocb->ki_flags & IOCB_DSYNC) {
2949 int ret = vfs_fsync_range(iocb->ki_filp,
2950 iocb->ki_pos - count, iocb->ki_pos - 1,
2951 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2959 extern void emergency_sync(void);
2960 extern void emergency_remount(void);
2963 extern int bmap(struct inode *inode, sector_t *block);
2965 static inline int bmap(struct inode *inode, sector_t *block)
2971 int notify_change(struct user_namespace *, struct dentry *,
2972 struct iattr *, struct inode **);
2973 int inode_permission(struct user_namespace *, struct inode *, int);
2974 int generic_permission(struct user_namespace *, struct inode *, int);
2975 static inline int file_permission(struct file *file, int mask)
2977 return inode_permission(file_mnt_user_ns(file),
2978 file_inode(file), mask);
2980 static inline int path_permission(const struct path *path, int mask)
2982 return inode_permission(mnt_user_ns(path->mnt),
2983 d_inode(path->dentry), mask);
2985 int __check_sticky(struct user_namespace *mnt_userns, struct inode *dir,
2986 struct inode *inode);
2988 static inline bool execute_ok(struct inode *inode)
2990 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2993 static inline bool inode_wrong_type(const struct inode *inode, umode_t mode)
2995 return (inode->i_mode ^ mode) & S_IFMT;
2998 static inline void file_start_write(struct file *file)
3000 if (!S_ISREG(file_inode(file)->i_mode))
3002 sb_start_write(file_inode(file)->i_sb);
3005 static inline bool file_start_write_trylock(struct file *file)
3007 if (!S_ISREG(file_inode(file)->i_mode))
3009 return sb_start_write_trylock(file_inode(file)->i_sb);
3012 static inline void file_end_write(struct file *file)
3014 if (!S_ISREG(file_inode(file)->i_mode))
3016 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
3020 * get_write_access() gets write permission for a file.
3021 * put_write_access() releases this write permission.
3022 * This is used for regular files.
3023 * We cannot support write (and maybe mmap read-write shared) accesses and
3024 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
3025 * can have the following values:
3026 * 0: no writers, no VM_DENYWRITE mappings
3027 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
3028 * > 0: (i_writecount) users are writing to the file.
3030 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
3031 * except for the cases where we don't hold i_writecount yet. Then we need to
3032 * use {get,deny}_write_access() - these functions check the sign and refuse
3033 * to do the change if sign is wrong.
3035 static inline int get_write_access(struct inode *inode)
3037 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
3039 static inline int deny_write_access(struct file *file)
3041 struct inode *inode = file_inode(file);
3042 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
3044 static inline void put_write_access(struct inode * inode)
3046 atomic_dec(&inode->i_writecount);
3048 static inline void allow_write_access(struct file *file)
3051 atomic_inc(&file_inode(file)->i_writecount);
3053 static inline bool inode_is_open_for_write(const struct inode *inode)
3055 return atomic_read(&inode->i_writecount) > 0;
3058 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
3059 static inline void i_readcount_dec(struct inode *inode)
3061 BUG_ON(!atomic_read(&inode->i_readcount));
3062 atomic_dec(&inode->i_readcount);
3064 static inline void i_readcount_inc(struct inode *inode)
3066 atomic_inc(&inode->i_readcount);
3069 static inline void i_readcount_dec(struct inode *inode)
3073 static inline void i_readcount_inc(struct inode *inode)
3078 extern int do_pipe_flags(int *, int);
3080 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
3081 ssize_t __kernel_read(struct file *file, void *buf, size_t count, loff_t *pos);
3082 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
3083 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
3084 extern struct file * open_exec(const char *);
3086 /* fs/dcache.c -- generic fs support functions */
3087 extern bool is_subdir(struct dentry *, struct dentry *);
3088 extern bool path_is_under(const struct path *, const struct path *);
3090 extern char *file_path(struct file *, char *, int);
3092 #include <linux/err.h>
3094 /* needed for stackable file system support */
3095 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
3097 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
3099 extern int inode_init_always(struct super_block *, struct inode *);
3100 extern void inode_init_once(struct inode *);
3101 extern void address_space_init_once(struct address_space *mapping);
3102 extern struct inode * igrab(struct inode *);
3103 extern ino_t iunique(struct super_block *, ino_t);
3104 extern int inode_needs_sync(struct inode *inode);
3105 extern int generic_delete_inode(struct inode *inode);
3106 static inline int generic_drop_inode(struct inode *inode)
3108 return !inode->i_nlink || inode_unhashed(inode);
3110 extern void d_mark_dontcache(struct inode *inode);
3112 extern struct inode *ilookup5_nowait(struct super_block *sb,
3113 unsigned long hashval, int (*test)(struct inode *, void *),
3115 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
3116 int (*test)(struct inode *, void *), void *data);
3117 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
3119 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
3120 int (*test)(struct inode *, void *),
3121 int (*set)(struct inode *, void *),
3123 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
3124 extern struct inode * iget_locked(struct super_block *, unsigned long);
3125 extern struct inode *find_inode_nowait(struct super_block *,
3127 int (*match)(struct inode *,
3128 unsigned long, void *),
3130 extern struct inode *find_inode_rcu(struct super_block *, unsigned long,
3131 int (*)(struct inode *, void *), void *);
3132 extern struct inode *find_inode_by_ino_rcu(struct super_block *, unsigned long);
3133 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
3134 extern int insert_inode_locked(struct inode *);
3135 #ifdef CONFIG_DEBUG_LOCK_ALLOC
3136 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
3138 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
3140 extern void unlock_new_inode(struct inode *);
3141 extern void discard_new_inode(struct inode *);
3142 extern unsigned int get_next_ino(void);
3143 extern void evict_inodes(struct super_block *sb);
3146 * Userspace may rely on the the inode number being non-zero. For example, glibc
3147 * simply ignores files with zero i_ino in unlink() and other places.
3149 * As an additional complication, if userspace was compiled with
3150 * _FILE_OFFSET_BITS=32 on a 64-bit kernel we'll only end up reading out the
3151 * lower 32 bits, so we need to check that those aren't zero explicitly. With
3152 * _FILE_OFFSET_BITS=64, this may cause some harmless false-negatives, but
3153 * better safe than sorry.
3155 static inline bool is_zero_ino(ino_t ino)
3157 return (u32)ino == 0;
3160 extern void __iget(struct inode * inode);
3161 extern void iget_failed(struct inode *);
3162 extern void clear_inode(struct inode *);
3163 extern void __destroy_inode(struct inode *);
3164 extern struct inode *new_inode_pseudo(struct super_block *sb);
3165 extern struct inode *new_inode(struct super_block *sb);
3166 extern void free_inode_nonrcu(struct inode *inode);
3167 extern int should_remove_suid(struct dentry *);
3168 extern int file_remove_privs(struct file *);
3170 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
3171 static inline void insert_inode_hash(struct inode *inode)
3173 __insert_inode_hash(inode, inode->i_ino);
3176 extern void __remove_inode_hash(struct inode *);
3177 static inline void remove_inode_hash(struct inode *inode)
3179 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
3180 __remove_inode_hash(inode);
3183 extern void inode_sb_list_add(struct inode *inode);
3185 extern int sb_set_blocksize(struct super_block *, int);
3186 extern int sb_min_blocksize(struct super_block *, int);
3188 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
3189 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
3190 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
3191 extern int generic_write_check_limits(struct file *file, loff_t pos,
3193 extern int generic_file_rw_checks(struct file *file_in, struct file *file_out);
3194 ssize_t filemap_read(struct kiocb *iocb, struct iov_iter *to,
3195 ssize_t already_read);
3196 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
3197 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
3198 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
3199 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
3200 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
3202 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
3204 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
3206 ssize_t vfs_iocb_iter_read(struct file *file, struct kiocb *iocb,
3207 struct iov_iter *iter);
3208 ssize_t vfs_iocb_iter_write(struct file *file, struct kiocb *iocb,
3209 struct iov_iter *iter);
3212 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
3213 struct pipe_inode_info *, size_t, unsigned int);
3214 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
3215 struct file *, loff_t *, size_t, unsigned int);
3216 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
3217 struct file *out, loff_t *, size_t len, unsigned int flags);
3218 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
3219 loff_t *opos, size_t len, unsigned int flags);
3223 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3224 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3225 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
3226 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3227 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3228 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3229 int whence, loff_t maxsize, loff_t eof);
3230 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3231 int whence, loff_t size);
3232 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3233 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3234 extern int generic_file_open(struct inode * inode, struct file * filp);
3235 extern int nonseekable_open(struct inode * inode, struct file * filp);
3236 extern int stream_open(struct inode * inode, struct file * filp);
3239 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3240 loff_t file_offset);
3243 /* need locking between buffered and direct access */
3246 /* filesystem does not support filling holes */
3247 DIO_SKIP_HOLES = 0x02,
3250 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3251 struct block_device *bdev, struct iov_iter *iter,
3252 get_block_t get_block,
3253 dio_iodone_t end_io, dio_submit_t submit_io,
3256 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3257 struct inode *inode,
3258 struct iov_iter *iter,
3259 get_block_t get_block)
3261 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3262 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3266 void inode_dio_wait(struct inode *inode);
3269 * inode_dio_begin - signal start of a direct I/O requests
3270 * @inode: inode the direct I/O happens on
3272 * This is called once we've finished processing a direct I/O request,
3273 * and is used to wake up callers waiting for direct I/O to be quiesced.
3275 static inline void inode_dio_begin(struct inode *inode)
3277 atomic_inc(&inode->i_dio_count);
3281 * inode_dio_end - signal finish of a direct I/O requests
3282 * @inode: inode the direct I/O happens on
3284 * This is called once we've finished processing a direct I/O request,
3285 * and is used to wake up callers waiting for direct I/O to be quiesced.
3287 static inline void inode_dio_end(struct inode *inode)
3289 if (atomic_dec_and_test(&inode->i_dio_count))
3290 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3294 * Warn about a page cache invalidation failure diring a direct I/O write.
3296 void dio_warn_stale_pagecache(struct file *filp);
3298 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3301 extern const struct file_operations generic_ro_fops;
3303 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3305 extern int readlink_copy(char __user *, int, const char *);
3306 extern int page_readlink(struct dentry *, char __user *, int);
3307 extern const char *page_get_link(struct dentry *, struct inode *,
3308 struct delayed_call *);
3309 extern void page_put_link(void *);
3310 extern int __page_symlink(struct inode *inode, const char *symname, int len,
3312 extern int page_symlink(struct inode *inode, const char *symname, int len);
3313 extern const struct inode_operations page_symlink_inode_operations;
3314 extern void kfree_link(void *);
3315 void generic_fillattr(struct user_namespace *, struct inode *, struct kstat *);
3316 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3317 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3318 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3319 void inode_add_bytes(struct inode *inode, loff_t bytes);
3320 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3321 void inode_sub_bytes(struct inode *inode, loff_t bytes);
3322 static inline loff_t __inode_get_bytes(struct inode *inode)
3324 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3326 loff_t inode_get_bytes(struct inode *inode);
3327 void inode_set_bytes(struct inode *inode, loff_t bytes);
3328 const char *simple_get_link(struct dentry *, struct inode *,
3329 struct delayed_call *);
3330 extern const struct inode_operations simple_symlink_inode_operations;
3332 extern int iterate_dir(struct file *, struct dir_context *);
3334 int vfs_fstatat(int dfd, const char __user *filename, struct kstat *stat,
3336 int vfs_fstat(int fd, struct kstat *stat);
3338 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3340 return vfs_fstatat(AT_FDCWD, filename, stat, 0);
3342 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3344 return vfs_fstatat(AT_FDCWD, name, stat, AT_SYMLINK_NOFOLLOW);
3347 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3348 extern int vfs_readlink(struct dentry *, char __user *, int);
3350 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3351 extern void put_filesystem(struct file_system_type *fs);
3352 extern struct file_system_type *get_fs_type(const char *name);
3353 extern struct super_block *get_super(struct block_device *);
3354 extern struct super_block *get_active_super(struct block_device *bdev);
3355 extern void drop_super(struct super_block *sb);
3356 extern void drop_super_exclusive(struct super_block *sb);
3357 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3358 extern void iterate_supers_type(struct file_system_type *,
3359 void (*)(struct super_block *, void *), void *);
3361 extern int dcache_dir_open(struct inode *, struct file *);
3362 extern int dcache_dir_close(struct inode *, struct file *);
3363 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3364 extern int dcache_readdir(struct file *, struct dir_context *);
3365 extern int simple_setattr(struct user_namespace *, struct dentry *,
3367 extern int simple_getattr(struct user_namespace *, const struct path *,
3368 struct kstat *, u32, unsigned int);
3369 extern int simple_statfs(struct dentry *, struct kstatfs *);
3370 extern int simple_open(struct inode *inode, struct file *file);
3371 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3372 extern int simple_unlink(struct inode *, struct dentry *);
3373 extern int simple_rmdir(struct inode *, struct dentry *);
3374 extern int simple_rename(struct user_namespace *, struct inode *,
3375 struct dentry *, struct inode *, struct dentry *,
3377 extern void simple_recursive_removal(struct dentry *,
3378 void (*callback)(struct dentry *));
3379 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3380 extern void noop_invalidatepage(struct page *page, unsigned int offset,
3381 unsigned int length);
3382 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3383 extern int simple_empty(struct dentry *);
3384 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3385 loff_t pos, unsigned len, unsigned flags,
3386 struct page **pagep, void **fsdata);
3387 extern const struct address_space_operations ram_aops;
3388 extern int always_delete_dentry(const struct dentry *);
3389 extern struct inode *alloc_anon_inode(struct super_block *);
3390 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3391 extern const struct dentry_operations simple_dentry_operations;
3393 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3394 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3395 extern const struct file_operations simple_dir_operations;
3396 extern const struct inode_operations simple_dir_inode_operations;
3397 extern void make_empty_dir_inode(struct inode *inode);
3398 extern bool is_empty_dir_inode(struct inode *inode);
3399 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3400 struct dentry *d_alloc_name(struct dentry *, const char *);
3401 extern int simple_fill_super(struct super_block *, unsigned long,
3402 const struct tree_descr *);
3403 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3404 extern void simple_release_fs(struct vfsmount **mount, int *count);
3406 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3407 loff_t *ppos, const void *from, size_t available);
3408 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3409 const void __user *from, size_t count);
3411 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3412 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3414 extern int generic_check_addressable(unsigned, u64);
3416 extern void generic_set_encrypted_ci_d_ops(struct dentry *dentry);
3418 #ifdef CONFIG_MIGRATION
3419 extern int buffer_migrate_page(struct address_space *,
3420 struct page *, struct page *,
3422 extern int buffer_migrate_page_norefs(struct address_space *,
3423 struct page *, struct page *,
3426 #define buffer_migrate_page NULL
3427 #define buffer_migrate_page_norefs NULL
3430 int setattr_prepare(struct user_namespace *, struct dentry *, struct iattr *);
3431 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3432 void setattr_copy(struct user_namespace *, struct inode *inode,
3433 const struct iattr *attr);
3435 extern int file_update_time(struct file *file);
3437 static inline bool vma_is_dax(const struct vm_area_struct *vma)
3439 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3442 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3444 struct inode *inode;
3446 if (!IS_ENABLED(CONFIG_FS_DAX) || !vma->vm_file)
3448 if (!vma_is_dax(vma))
3450 inode = file_inode(vma->vm_file);
3451 if (S_ISCHR(inode->i_mode))
3452 return false; /* device-dax */
3456 static inline int iocb_flags(struct file *file)
3459 if (file->f_flags & O_APPEND)
3461 if (file->f_flags & O_DIRECT)
3463 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3465 if (file->f_flags & __O_SYNC)
3470 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3472 int kiocb_flags = 0;
3474 /* make sure there's no overlap between RWF and private IOCB flags */
3475 BUILD_BUG_ON((__force int) RWF_SUPPORTED & IOCB_EVENTFD);
3479 if (unlikely(flags & ~RWF_SUPPORTED))
3482 if (flags & RWF_NOWAIT) {
3483 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3485 kiocb_flags |= IOCB_NOIO;
3487 kiocb_flags |= (__force int) (flags & RWF_SUPPORTED);
3488 if (flags & RWF_SYNC)
3489 kiocb_flags |= IOCB_DSYNC;
3491 ki->ki_flags |= kiocb_flags;
3495 static inline ino_t parent_ino(struct dentry *dentry)
3500 * Don't strictly need d_lock here? If the parent ino could change
3501 * then surely we'd have a deeper race in the caller?
3503 spin_lock(&dentry->d_lock);
3504 res = dentry->d_parent->d_inode->i_ino;
3505 spin_unlock(&dentry->d_lock);
3509 /* Transaction based IO helpers */
3512 * An argresp is stored in an allocated page and holds the
3513 * size of the argument or response, along with its content
3515 struct simple_transaction_argresp {
3520 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3522 char *simple_transaction_get(struct file *file, const char __user *buf,
3524 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3525 size_t size, loff_t *pos);
3526 int simple_transaction_release(struct inode *inode, struct file *file);
3528 void simple_transaction_set(struct file *file, size_t n);
3531 * simple attribute files
3533 * These attributes behave similar to those in sysfs:
3535 * Writing to an attribute immediately sets a value, an open file can be
3536 * written to multiple times.
3538 * Reading from an attribute creates a buffer from the value that might get
3539 * read with multiple read calls. When the attribute has been read
3540 * completely, no further read calls are possible until the file is opened
3543 * All attributes contain a text representation of a numeric value
3544 * that are accessed with the get() and set() functions.
3546 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3547 static int __fops ## _open(struct inode *inode, struct file *file) \
3549 __simple_attr_check_format(__fmt, 0ull); \
3550 return simple_attr_open(inode, file, __get, __set, __fmt); \
3552 static const struct file_operations __fops = { \
3553 .owner = THIS_MODULE, \
3554 .open = __fops ## _open, \
3555 .release = simple_attr_release, \
3556 .read = simple_attr_read, \
3557 .write = simple_attr_write, \
3558 .llseek = generic_file_llseek, \
3561 static inline __printf(1, 2)
3562 void __simple_attr_check_format(const char *fmt, ...)
3564 /* don't do anything, just let the compiler check the arguments; */
3567 int simple_attr_open(struct inode *inode, struct file *file,
3568 int (*get)(void *, u64 *), int (*set)(void *, u64),
3570 int simple_attr_release(struct inode *inode, struct file *file);
3571 ssize_t simple_attr_read(struct file *file, char __user *buf,
3572 size_t len, loff_t *ppos);
3573 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3574 size_t len, loff_t *ppos);
3577 int proc_nr_files(struct ctl_table *table, int write,
3578 void *buffer, size_t *lenp, loff_t *ppos);
3579 int proc_nr_dentry(struct ctl_table *table, int write,
3580 void *buffer, size_t *lenp, loff_t *ppos);
3581 int proc_nr_inodes(struct ctl_table *table, int write,
3582 void *buffer, size_t *lenp, loff_t *ppos);
3583 int __init get_filesystem_list(char *buf);
3585 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3586 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3588 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3589 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3590 (flag & __FMODE_NONOTIFY)))
3592 static inline bool is_sxid(umode_t mode)
3594 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3597 static inline int check_sticky(struct user_namespace *mnt_userns,
3598 struct inode *dir, struct inode *inode)
3600 if (!(dir->i_mode & S_ISVTX))
3603 return __check_sticky(mnt_userns, dir, inode);
3606 static inline void inode_has_no_xattr(struct inode *inode)
3608 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3609 inode->i_flags |= S_NOSEC;
3612 static inline bool is_root_inode(struct inode *inode)
3614 return inode == inode->i_sb->s_root->d_inode;
3617 static inline bool dir_emit(struct dir_context *ctx,
3618 const char *name, int namelen,
3619 u64 ino, unsigned type)
3621 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3623 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3625 return ctx->actor(ctx, ".", 1, ctx->pos,
3626 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3628 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3630 return ctx->actor(ctx, "..", 2, ctx->pos,
3631 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3633 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3635 if (ctx->pos == 0) {
3636 if (!dir_emit_dot(file, ctx))
3640 if (ctx->pos == 1) {
3641 if (!dir_emit_dotdot(file, ctx))
3647 static inline bool dir_relax(struct inode *inode)
3649 inode_unlock(inode);
3651 return !IS_DEADDIR(inode);
3654 static inline bool dir_relax_shared(struct inode *inode)
3656 inode_unlock_shared(inode);
3657 inode_lock_shared(inode);
3658 return !IS_DEADDIR(inode);
3661 extern bool path_noexec(const struct path *path);
3662 extern void inode_nohighmem(struct inode *inode);
3665 extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len,
3667 extern int generic_fadvise(struct file *file, loff_t offset, loff_t len,
3671 * Flush file data before changing attributes. Caller must hold any locks
3672 * required to prevent further writes to this file until we're done setting
3675 static inline int inode_drain_writes(struct inode *inode)
3677 inode_dio_wait(inode);
3678 return filemap_write_and_wait(inode->i_mapping);
3681 #endif /* _LINUX_FS_H */