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 */
1040 #define FL_RECLAIM 4096 /* reclaiming from a reboot server */
1042 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
1045 * Special return value from posix_lock_file() and vfs_lock_file() for
1046 * asynchronous locking.
1048 #define FILE_LOCK_DEFERRED 1
1050 /* legacy typedef, should eventually be removed */
1051 typedef void *fl_owner_t;
1055 struct file_lock_operations {
1056 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1057 void (*fl_release_private)(struct file_lock *);
1060 struct lock_manager_operations {
1061 fl_owner_t (*lm_get_owner)(fl_owner_t);
1062 void (*lm_put_owner)(fl_owner_t);
1063 void (*lm_notify)(struct file_lock *); /* unblock callback */
1064 int (*lm_grant)(struct file_lock *, int);
1065 bool (*lm_break)(struct file_lock *);
1066 int (*lm_change)(struct file_lock *, int, struct list_head *);
1067 void (*lm_setup)(struct file_lock *, void **);
1068 bool (*lm_breaker_owns_lease)(struct file_lock *);
1071 struct lock_manager {
1072 struct list_head list;
1074 * NFSv4 and up also want opens blocked during the grace period;
1081 void locks_start_grace(struct net *, struct lock_manager *);
1082 void locks_end_grace(struct lock_manager *);
1083 bool locks_in_grace(struct net *);
1084 bool opens_in_grace(struct net *);
1086 /* that will die - we need it for nfs_lock_info */
1087 #include <linux/nfs_fs_i.h>
1090 * struct file_lock represents a generic "file lock". It's used to represent
1091 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1092 * note that the same struct is used to represent both a request for a lock and
1093 * the lock itself, but the same object is never used for both.
1095 * FIXME: should we create a separate "struct lock_request" to help distinguish
1098 * The varous i_flctx lists are ordered by:
1101 * 2) lock range start
1104 * Obviously, the last two criteria only matter for POSIX locks.
1107 struct file_lock *fl_blocker; /* The lock, that is blocking us */
1108 struct list_head fl_list; /* link into file_lock_context */
1109 struct hlist_node fl_link; /* node in global lists */
1110 struct list_head fl_blocked_requests; /* list of requests with
1111 * ->fl_blocker pointing here
1113 struct list_head fl_blocked_member; /* node in
1114 * ->fl_blocker->fl_blocked_requests
1116 fl_owner_t fl_owner;
1117 unsigned int fl_flags;
1118 unsigned char fl_type;
1119 unsigned int fl_pid;
1120 int fl_link_cpu; /* what cpu's list is this on? */
1121 wait_queue_head_t fl_wait;
1122 struct file *fl_file;
1126 struct fasync_struct * fl_fasync; /* for lease break notifications */
1127 /* for lease breaks: */
1128 unsigned long fl_break_time;
1129 unsigned long fl_downgrade_time;
1131 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1132 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1134 struct nfs_lock_info nfs_fl;
1135 struct nfs4_lock_info nfs4_fl;
1137 struct list_head link; /* link in AFS vnode's pending_locks list */
1138 int state; /* state of grant or error if -ve */
1139 unsigned int debug_id;
1142 } __randomize_layout;
1144 struct file_lock_context {
1145 spinlock_t flc_lock;
1146 struct list_head flc_flock;
1147 struct list_head flc_posix;
1148 struct list_head flc_lease;
1151 /* The following constant reflects the upper bound of the file/locking space */
1153 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1154 #define OFFSET_MAX INT_LIMIT(loff_t)
1155 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1158 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1160 #define locks_inode(f) file_inode(f)
1162 #ifdef CONFIG_FILE_LOCKING
1163 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1164 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1167 #if BITS_PER_LONG == 32
1168 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1169 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1173 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1174 extern int fcntl_getlease(struct file *filp);
1177 void locks_free_lock_context(struct inode *inode);
1178 void locks_free_lock(struct file_lock *fl);
1179 extern void locks_init_lock(struct file_lock *);
1180 extern struct file_lock * locks_alloc_lock(void);
1181 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1182 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1183 extern void locks_remove_posix(struct file *, fl_owner_t);
1184 extern void locks_remove_file(struct file *);
1185 extern void locks_release_private(struct file_lock *);
1186 extern void posix_test_lock(struct file *, struct file_lock *);
1187 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1188 extern int locks_delete_block(struct file_lock *);
1189 extern int vfs_test_lock(struct file *, struct file_lock *);
1190 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1191 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1192 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1193 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1194 extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1195 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1196 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1197 extern int lease_modify(struct file_lock *, int, struct list_head *);
1199 struct notifier_block;
1200 extern int lease_register_notifier(struct notifier_block *);
1201 extern void lease_unregister_notifier(struct notifier_block *);
1203 struct files_struct;
1204 extern void show_fd_locks(struct seq_file *f,
1205 struct file *filp, struct files_struct *files);
1206 #else /* !CONFIG_FILE_LOCKING */
1207 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1208 struct flock __user *user)
1213 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1214 unsigned int cmd, struct flock __user *user)
1219 #if BITS_PER_LONG == 32
1220 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1221 struct flock64 __user *user)
1226 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1227 unsigned int cmd, struct flock64 __user *user)
1232 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1237 static inline int fcntl_getlease(struct file *filp)
1243 locks_free_lock_context(struct inode *inode)
1247 static inline void locks_init_lock(struct file_lock *fl)
1252 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1257 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1262 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1267 static inline void locks_remove_file(struct file *filp)
1272 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1277 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1278 struct file_lock *conflock)
1283 static inline int locks_delete_block(struct file_lock *waiter)
1288 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1293 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1294 struct file_lock *fl, struct file_lock *conf)
1299 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1304 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1309 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1314 static inline void lease_get_mtime(struct inode *inode,
1315 struct timespec64 *time)
1320 static inline int generic_setlease(struct file *filp, long arg,
1321 struct file_lock **flp, void **priv)
1326 static inline int vfs_setlease(struct file *filp, long arg,
1327 struct file_lock **lease, void **priv)
1332 static inline int lease_modify(struct file_lock *fl, int arg,
1333 struct list_head *dispose)
1338 struct files_struct;
1339 static inline void show_fd_locks(struct seq_file *f,
1340 struct file *filp, struct files_struct *files) {}
1341 #endif /* !CONFIG_FILE_LOCKING */
1343 static inline struct inode *file_inode(const struct file *f)
1348 static inline struct dentry *file_dentry(const struct file *file)
1350 return d_real(file->f_path.dentry, file_inode(file));
1353 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1355 return locks_lock_inode_wait(locks_inode(filp), fl);
1358 struct fasync_struct {
1362 struct fasync_struct *fa_next; /* singly linked list */
1363 struct file *fa_file;
1364 struct rcu_head fa_rcu;
1367 #define FASYNC_MAGIC 0x4601
1369 /* SMP safe fasync helpers: */
1370 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1371 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1372 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1373 extern struct fasync_struct *fasync_alloc(void);
1374 extern void fasync_free(struct fasync_struct *);
1376 /* can be called from interrupts */
1377 extern void kill_fasync(struct fasync_struct **, int, int);
1379 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1380 extern int f_setown(struct file *filp, unsigned long arg, int force);
1381 extern void f_delown(struct file *filp);
1382 extern pid_t f_getown(struct file *filp);
1383 extern int send_sigurg(struct fown_struct *fown);
1386 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1387 * represented in both.
1389 #define SB_RDONLY 1 /* Mount read-only */
1390 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1391 #define SB_NODEV 4 /* Disallow access to device special files */
1392 #define SB_NOEXEC 8 /* Disallow program execution */
1393 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1394 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1395 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1396 #define SB_NOATIME 1024 /* Do not update access times. */
1397 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1398 #define SB_SILENT 32768
1399 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1400 #define SB_INLINECRYPT (1<<17) /* Use blk-crypto for encrypted files */
1401 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1402 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1403 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1405 /* These sb flags are internal to the kernel */
1406 #define SB_SUBMOUNT (1<<26)
1407 #define SB_FORCE (1<<27)
1408 #define SB_NOSEC (1<<28)
1409 #define SB_BORN (1<<29)
1410 #define SB_ACTIVE (1<<30)
1411 #define SB_NOUSER (1<<31)
1413 /* These flags relate to encoding and casefolding */
1414 #define SB_ENC_STRICT_MODE_FL (1 << 0)
1416 #define sb_has_strict_encoding(sb) \
1417 (sb->s_encoding_flags & SB_ENC_STRICT_MODE_FL)
1423 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1424 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1425 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1426 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1427 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1430 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1431 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1432 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1433 #define SB_I_STABLE_WRITES 0x00000008 /* don't modify blks until WB is done */
1435 /* sb->s_iflags to limit user namespace mounts */
1436 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1437 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1438 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1440 #define SB_I_SKIP_SYNC 0x00000100 /* Skip superblock at global sync */
1442 /* Possible states of 'frozen' field */
1444 SB_UNFROZEN = 0, /* FS is unfrozen */
1445 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1446 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1447 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1448 * internal threads if needed) */
1449 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1452 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1455 int frozen; /* Is sb frozen? */
1456 wait_queue_head_t wait_unfrozen; /* wait for thaw */
1457 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1460 struct super_block {
1461 struct list_head s_list; /* Keep this first */
1462 dev_t s_dev; /* search index; _not_ kdev_t */
1463 unsigned char s_blocksize_bits;
1464 unsigned long s_blocksize;
1465 loff_t s_maxbytes; /* Max file size */
1466 struct file_system_type *s_type;
1467 const struct super_operations *s_op;
1468 const struct dquot_operations *dq_op;
1469 const struct quotactl_ops *s_qcop;
1470 const struct export_operations *s_export_op;
1471 unsigned long s_flags;
1472 unsigned long s_iflags; /* internal SB_I_* flags */
1473 unsigned long s_magic;
1474 struct dentry *s_root;
1475 struct rw_semaphore s_umount;
1478 #ifdef CONFIG_SECURITY
1481 const struct xattr_handler **s_xattr;
1482 #ifdef CONFIG_FS_ENCRYPTION
1483 const struct fscrypt_operations *s_cop;
1484 struct key *s_master_keys; /* master crypto keys in use */
1486 #ifdef CONFIG_FS_VERITY
1487 const struct fsverity_operations *s_vop;
1489 #ifdef CONFIG_UNICODE
1490 struct unicode_map *s_encoding;
1491 __u16 s_encoding_flags;
1493 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1494 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1495 struct block_device *s_bdev;
1496 struct backing_dev_info *s_bdi;
1497 struct mtd_info *s_mtd;
1498 struct hlist_node s_instances;
1499 unsigned int s_quota_types; /* Bitmask of supported quota types */
1500 struct quota_info s_dquot; /* Diskquota specific options */
1502 struct sb_writers s_writers;
1505 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1506 * s_fsnotify_marks together for cache efficiency. They are frequently
1507 * accessed and rarely modified.
1509 void *s_fs_info; /* Filesystem private info */
1511 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1513 /* Time limits for c/m/atime in seconds */
1514 time64_t s_time_min;
1515 time64_t s_time_max;
1516 #ifdef CONFIG_FSNOTIFY
1517 __u32 s_fsnotify_mask;
1518 struct fsnotify_mark_connector __rcu *s_fsnotify_marks;
1521 char s_id[32]; /* Informational name */
1522 uuid_t s_uuid; /* UUID */
1524 unsigned int s_max_links;
1528 * The next field is for VFS *only*. No filesystems have any business
1529 * even looking at it. You had been warned.
1531 struct mutex s_vfs_rename_mutex; /* Kludge */
1534 * Filesystem subtype. If non-empty the filesystem type field
1535 * in /proc/mounts will be "type.subtype"
1537 const char *s_subtype;
1539 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1542 * Saved pool identifier for cleancache (-1 means none)
1544 int cleancache_poolid;
1546 struct shrinker s_shrink; /* per-sb shrinker handle */
1548 /* Number of inodes with nlink == 0 but still referenced */
1549 atomic_long_t s_remove_count;
1552 * Number of inode/mount/sb objects that are being watched, note that
1553 * inodes objects are currently double-accounted.
1555 atomic_long_t s_fsnotify_connectors;
1557 /* Being remounted read-only */
1558 int s_readonly_remount;
1560 /* per-sb errseq_t for reporting writeback errors via syncfs */
1563 /* AIO completions deferred from interrupt context */
1564 struct workqueue_struct *s_dio_done_wq;
1565 struct hlist_head s_pins;
1568 * Owning user namespace and default context in which to
1569 * interpret filesystem uids, gids, quotas, device nodes,
1570 * xattrs and security labels.
1572 struct user_namespace *s_user_ns;
1575 * The list_lru structure is essentially just a pointer to a table
1576 * of per-node lru lists, each of which has its own spinlock.
1577 * There is no need to put them into separate cachelines.
1579 struct list_lru s_dentry_lru;
1580 struct list_lru s_inode_lru;
1581 struct rcu_head rcu;
1582 struct work_struct destroy_work;
1584 struct mutex s_sync_lock; /* sync serialisation lock */
1587 * Indicates how deep in a filesystem stack this SB is
1591 /* s_inode_list_lock protects s_inodes */
1592 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1593 struct list_head s_inodes; /* all inodes */
1595 spinlock_t s_inode_wblist_lock;
1596 struct list_head s_inodes_wb; /* writeback inodes */
1597 } __randomize_layout;
1599 /* Helper functions so that in most cases filesystems will
1600 * not need to deal directly with kuid_t and kgid_t and can
1601 * instead deal with the raw numeric values that are stored
1602 * in the filesystem.
1604 static inline uid_t i_uid_read(const struct inode *inode)
1606 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1609 static inline gid_t i_gid_read(const struct inode *inode)
1611 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1614 static inline void i_uid_write(struct inode *inode, uid_t uid)
1616 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1619 static inline void i_gid_write(struct inode *inode, gid_t gid)
1621 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1625 * kuid_into_mnt - map a kuid down into a mnt_userns
1626 * @mnt_userns: user namespace of the relevant mount
1627 * @kuid: kuid to be mapped
1629 * Return: @kuid mapped according to @mnt_userns.
1630 * If @kuid has no mapping INVALID_UID is returned.
1632 static inline kuid_t kuid_into_mnt(struct user_namespace *mnt_userns,
1635 return make_kuid(mnt_userns, __kuid_val(kuid));
1639 * kgid_into_mnt - map a kgid down into a mnt_userns
1640 * @mnt_userns: user namespace of the relevant mount
1641 * @kgid: kgid to be mapped
1643 * Return: @kgid mapped according to @mnt_userns.
1644 * If @kgid has no mapping INVALID_GID is returned.
1646 static inline kgid_t kgid_into_mnt(struct user_namespace *mnt_userns,
1649 return make_kgid(mnt_userns, __kgid_val(kgid));
1653 * i_uid_into_mnt - map an inode's i_uid down into a mnt_userns
1654 * @mnt_userns: user namespace of the mount the inode was found from
1655 * @inode: inode to map
1657 * Return: the inode's i_uid mapped down according to @mnt_userns.
1658 * If the inode's i_uid has no mapping INVALID_UID is returned.
1660 static inline kuid_t i_uid_into_mnt(struct user_namespace *mnt_userns,
1661 const struct inode *inode)
1663 return kuid_into_mnt(mnt_userns, inode->i_uid);
1667 * i_gid_into_mnt - map an inode's i_gid down into a mnt_userns
1668 * @mnt_userns: user namespace of the mount the inode was found from
1669 * @inode: inode to map
1671 * Return: the inode's i_gid mapped down according to @mnt_userns.
1672 * If the inode's i_gid has no mapping INVALID_GID is returned.
1674 static inline kgid_t i_gid_into_mnt(struct user_namespace *mnt_userns,
1675 const struct inode *inode)
1677 return kgid_into_mnt(mnt_userns, inode->i_gid);
1681 * kuid_from_mnt - map a kuid up into a mnt_userns
1682 * @mnt_userns: user namespace of the relevant mount
1683 * @kuid: kuid to be mapped
1685 * Return: @kuid mapped up according to @mnt_userns.
1686 * If @kuid has no mapping INVALID_UID is returned.
1688 static inline kuid_t kuid_from_mnt(struct user_namespace *mnt_userns,
1691 return KUIDT_INIT(from_kuid(mnt_userns, kuid));
1695 * kgid_from_mnt - map a kgid up into a mnt_userns
1696 * @mnt_userns: user namespace of the relevant mount
1697 * @kgid: kgid to be mapped
1699 * Return: @kgid mapped up according to @mnt_userns.
1700 * If @kgid has no mapping INVALID_GID is returned.
1702 static inline kgid_t kgid_from_mnt(struct user_namespace *mnt_userns,
1705 return KGIDT_INIT(from_kgid(mnt_userns, kgid));
1709 * mapped_fsuid - return caller's fsuid mapped up into a mnt_userns
1710 * @mnt_userns: user namespace of the relevant mount
1712 * Use this helper to initialize a new vfs or filesystem object based on
1713 * the caller's fsuid. A common example is initializing the i_uid field of
1714 * a newly allocated inode triggered by a creation event such as mkdir or
1715 * O_CREAT. Other examples include the allocation of quotas for a specific
1718 * Return: the caller's current fsuid mapped up according to @mnt_userns.
1720 static inline kuid_t mapped_fsuid(struct user_namespace *mnt_userns)
1722 return kuid_from_mnt(mnt_userns, current_fsuid());
1726 * mapped_fsgid - return caller's fsgid mapped up into a mnt_userns
1727 * @mnt_userns: user namespace of the relevant mount
1729 * Use this helper to initialize a new vfs or filesystem object based on
1730 * the caller's fsgid. A common example is initializing the i_gid field of
1731 * a newly allocated inode triggered by a creation event such as mkdir or
1732 * O_CREAT. Other examples include the allocation of quotas for a specific
1735 * Return: the caller's current fsgid mapped up according to @mnt_userns.
1737 static inline kgid_t mapped_fsgid(struct user_namespace *mnt_userns)
1739 return kgid_from_mnt(mnt_userns, current_fsgid());
1743 * inode_fsuid_set - initialize inode's i_uid field with callers fsuid
1744 * @inode: inode to initialize
1745 * @mnt_userns: user namespace of the mount the inode was found from
1747 * Initialize the i_uid field of @inode. If the inode was found/created via
1748 * an idmapped mount map the caller's fsuid according to @mnt_users.
1750 static inline void inode_fsuid_set(struct inode *inode,
1751 struct user_namespace *mnt_userns)
1753 inode->i_uid = mapped_fsuid(mnt_userns);
1757 * inode_fsgid_set - initialize inode's i_gid field with callers fsgid
1758 * @inode: inode to initialize
1759 * @mnt_userns: user namespace of the mount the inode was found from
1761 * Initialize the i_gid field of @inode. If the inode was found/created via
1762 * an idmapped mount map the caller's fsgid according to @mnt_users.
1764 static inline void inode_fsgid_set(struct inode *inode,
1765 struct user_namespace *mnt_userns)
1767 inode->i_gid = mapped_fsgid(mnt_userns);
1771 * fsuidgid_has_mapping() - check whether caller's fsuid/fsgid is mapped
1772 * @sb: the superblock we want a mapping in
1773 * @mnt_userns: user namespace of the relevant mount
1775 * Check whether the caller's fsuid and fsgid have a valid mapping in the
1776 * s_user_ns of the superblock @sb. If the caller is on an idmapped mount map
1777 * the caller's fsuid and fsgid according to the @mnt_userns first.
1779 * Return: true if fsuid and fsgid is mapped, false if not.
1781 static inline bool fsuidgid_has_mapping(struct super_block *sb,
1782 struct user_namespace *mnt_userns)
1784 struct user_namespace *s_user_ns = sb->s_user_ns;
1786 return kuid_has_mapping(s_user_ns, mapped_fsuid(mnt_userns)) &&
1787 kgid_has_mapping(s_user_ns, mapped_fsgid(mnt_userns));
1790 extern struct timespec64 current_time(struct inode *inode);
1793 * Snapshotting support.
1797 * These are internal functions, please use sb_start_{write,pagefault,intwrite}
1800 static inline void __sb_end_write(struct super_block *sb, int level)
1802 percpu_up_read(sb->s_writers.rw_sem + level-1);
1805 static inline void __sb_start_write(struct super_block *sb, int level)
1807 percpu_down_read(sb->s_writers.rw_sem + level - 1);
1810 static inline bool __sb_start_write_trylock(struct super_block *sb, int level)
1812 return percpu_down_read_trylock(sb->s_writers.rw_sem + level - 1);
1815 #define __sb_writers_acquired(sb, lev) \
1816 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1817 #define __sb_writers_release(sb, lev) \
1818 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1821 * sb_end_write - drop write access to a superblock
1822 * @sb: the super we wrote to
1824 * Decrement number of writers to the filesystem. Wake up possible waiters
1825 * wanting to freeze the filesystem.
1827 static inline void sb_end_write(struct super_block *sb)
1829 __sb_end_write(sb, SB_FREEZE_WRITE);
1833 * sb_end_pagefault - drop write access to a superblock from a page fault
1834 * @sb: the super we wrote to
1836 * Decrement number of processes handling write page fault to the filesystem.
1837 * Wake up possible waiters wanting to freeze the filesystem.
1839 static inline void sb_end_pagefault(struct super_block *sb)
1841 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1845 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1846 * @sb: the super we wrote to
1848 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1849 * waiters wanting to freeze the filesystem.
1851 static inline void sb_end_intwrite(struct super_block *sb)
1853 __sb_end_write(sb, SB_FREEZE_FS);
1857 * sb_start_write - get write access to a superblock
1858 * @sb: the super we write to
1860 * When a process wants to write data or metadata to a file system (i.e. dirty
1861 * a page or an inode), it should embed the operation in a sb_start_write() -
1862 * sb_end_write() pair to get exclusion against file system freezing. This
1863 * function increments number of writers preventing freezing. If the file
1864 * system is already frozen, the function waits until the file system is
1867 * Since freeze protection behaves as a lock, users have to preserve
1868 * ordering of freeze protection and other filesystem locks. Generally,
1869 * freeze protection should be the outermost lock. In particular, we have:
1872 * -> i_mutex (write path, truncate, directory ops, ...)
1873 * -> s_umount (freeze_super, thaw_super)
1875 static inline void sb_start_write(struct super_block *sb)
1877 __sb_start_write(sb, SB_FREEZE_WRITE);
1880 static inline bool sb_start_write_trylock(struct super_block *sb)
1882 return __sb_start_write_trylock(sb, SB_FREEZE_WRITE);
1886 * sb_start_pagefault - get write access to a superblock from a page fault
1887 * @sb: the super we write to
1889 * When a process starts handling write page fault, it should embed the
1890 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1891 * exclusion against file system freezing. This is needed since the page fault
1892 * is going to dirty a page. This function increments number of running page
1893 * faults preventing freezing. If the file system is already frozen, the
1894 * function waits until the file system is thawed.
1896 * Since page fault freeze protection behaves as a lock, users have to preserve
1897 * ordering of freeze protection and other filesystem locks. It is advised to
1898 * put sb_start_pagefault() close to mmap_lock in lock ordering. Page fault
1899 * handling code implies lock dependency:
1902 * -> sb_start_pagefault
1904 static inline void sb_start_pagefault(struct super_block *sb)
1906 __sb_start_write(sb, SB_FREEZE_PAGEFAULT);
1910 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1911 * @sb: the super we write to
1913 * This is the third level of protection against filesystem freezing. It is
1914 * free for use by a filesystem. The only requirement is that it must rank
1915 * below sb_start_pagefault.
1917 * For example filesystem can call sb_start_intwrite() when starting a
1918 * transaction which somewhat eases handling of freezing for internal sources
1919 * of filesystem changes (internal fs threads, discarding preallocation on file
1922 static inline void sb_start_intwrite(struct super_block *sb)
1924 __sb_start_write(sb, SB_FREEZE_FS);
1927 static inline bool sb_start_intwrite_trylock(struct super_block *sb)
1929 return __sb_start_write_trylock(sb, SB_FREEZE_FS);
1932 bool inode_owner_or_capable(struct user_namespace *mnt_userns,
1933 const struct inode *inode);
1936 * VFS helper functions..
1938 int vfs_create(struct user_namespace *, struct inode *,
1939 struct dentry *, umode_t, bool);
1940 int vfs_mkdir(struct user_namespace *, struct inode *,
1941 struct dentry *, umode_t);
1942 int vfs_mknod(struct user_namespace *, struct inode *, struct dentry *,
1944 int vfs_symlink(struct user_namespace *, struct inode *,
1945 struct dentry *, const char *);
1946 int vfs_link(struct dentry *, struct user_namespace *, struct inode *,
1947 struct dentry *, struct inode **);
1948 int vfs_rmdir(struct user_namespace *, struct inode *, struct dentry *);
1949 int vfs_unlink(struct user_namespace *, struct inode *, struct dentry *,
1953 * struct renamedata - contains all information required for renaming
1954 * @old_mnt_userns: old user namespace of the mount the inode was found from
1955 * @old_dir: parent of source
1956 * @old_dentry: source
1957 * @new_mnt_userns: new user namespace of the mount the inode was found from
1958 * @new_dir: parent of destination
1959 * @new_dentry: destination
1960 * @delegated_inode: returns an inode needing a delegation break
1961 * @flags: rename flags
1964 struct user_namespace *old_mnt_userns;
1965 struct inode *old_dir;
1966 struct dentry *old_dentry;
1967 struct user_namespace *new_mnt_userns;
1968 struct inode *new_dir;
1969 struct dentry *new_dentry;
1970 struct inode **delegated_inode;
1972 } __randomize_layout;
1974 int vfs_rename(struct renamedata *);
1976 static inline int vfs_whiteout(struct user_namespace *mnt_userns,
1977 struct inode *dir, struct dentry *dentry)
1979 return vfs_mknod(mnt_userns, dir, dentry, S_IFCHR | WHITEOUT_MODE,
1983 struct dentry *vfs_tmpfile(struct user_namespace *mnt_userns,
1984 struct dentry *dentry, umode_t mode, int open_flag);
1986 int vfs_mkobj(struct dentry *, umode_t,
1987 int (*f)(struct dentry *, umode_t, void *),
1990 int vfs_fchown(struct file *file, uid_t user, gid_t group);
1991 int vfs_fchmod(struct file *file, umode_t mode);
1992 int vfs_utimes(const struct path *path, struct timespec64 *times);
1994 extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1996 #ifdef CONFIG_COMPAT
1997 extern long compat_ptr_ioctl(struct file *file, unsigned int cmd,
2000 #define compat_ptr_ioctl NULL
2004 * VFS file helper functions.
2006 void inode_init_owner(struct user_namespace *mnt_userns, struct inode *inode,
2007 const struct inode *dir, umode_t mode);
2008 extern bool may_open_dev(const struct path *path);
2011 * This is the "filldir" function type, used by readdir() to let
2012 * the kernel specify what kind of dirent layout it wants to have.
2013 * This allows the kernel to read directories into kernel space or
2014 * to have different dirent layouts depending on the binary type.
2017 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
2020 struct dir_context {
2026 * These flags let !MMU mmap() govern direct device mapping vs immediate
2027 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
2029 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
2030 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
2031 * NOMMU_MAP_READ: Can be mapped for reading
2032 * NOMMU_MAP_WRITE: Can be mapped for writing
2033 * NOMMU_MAP_EXEC: Can be mapped for execution
2035 #define NOMMU_MAP_COPY 0x00000001
2036 #define NOMMU_MAP_DIRECT 0x00000008
2037 #define NOMMU_MAP_READ VM_MAYREAD
2038 #define NOMMU_MAP_WRITE VM_MAYWRITE
2039 #define NOMMU_MAP_EXEC VM_MAYEXEC
2041 #define NOMMU_VMFLAGS \
2042 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
2045 * These flags control the behavior of the remap_file_range function pointer.
2046 * If it is called with len == 0 that means "remap to end of source file".
2047 * See Documentation/filesystems/vfs.rst for more details about this call.
2049 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
2050 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
2052 #define REMAP_FILE_DEDUP (1 << 0)
2053 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
2056 * These flags signal that the caller is ok with altering various aspects of
2057 * the behavior of the remap operation. The changes must be made by the
2058 * implementation; the vfs remap helper functions can take advantage of them.
2059 * Flags in this category exist to preserve the quirky behavior of the hoisted
2060 * btrfs clone/dedupe ioctls.
2062 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
2066 struct file_operations {
2067 struct module *owner;
2068 loff_t (*llseek) (struct file *, loff_t, int);
2069 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
2070 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
2071 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
2072 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
2073 int (*iopoll)(struct kiocb *kiocb, bool spin);
2074 int (*iterate) (struct file *, struct dir_context *);
2075 int (*iterate_shared) (struct file *, struct dir_context *);
2076 __poll_t (*poll) (struct file *, struct poll_table_struct *);
2077 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
2078 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
2079 int (*mmap) (struct file *, struct vm_area_struct *);
2080 unsigned long mmap_supported_flags;
2081 int (*open) (struct inode *, struct file *);
2082 int (*flush) (struct file *, fl_owner_t id);
2083 int (*release) (struct inode *, struct file *);
2084 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
2085 int (*fasync) (int, struct file *, int);
2086 int (*lock) (struct file *, int, struct file_lock *);
2087 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
2088 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
2089 int (*check_flags)(int);
2090 int (*flock) (struct file *, int, struct file_lock *);
2091 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
2092 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
2093 int (*setlease)(struct file *, long, struct file_lock **, void **);
2094 long (*fallocate)(struct file *file, int mode, loff_t offset,
2096 void (*show_fdinfo)(struct seq_file *m, struct file *f);
2098 unsigned (*mmap_capabilities)(struct file *);
2100 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
2101 loff_t, size_t, unsigned int);
2102 loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
2103 struct file *file_out, loff_t pos_out,
2104 loff_t len, unsigned int remap_flags);
2105 int (*fadvise)(struct file *, loff_t, loff_t, int);
2106 } __randomize_layout;
2108 struct inode_operations {
2109 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
2110 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
2111 int (*permission) (struct user_namespace *, struct inode *, int);
2112 struct posix_acl * (*get_acl)(struct inode *, int);
2114 int (*readlink) (struct dentry *, char __user *,int);
2116 int (*create) (struct user_namespace *, struct inode *,struct dentry *,
2118 int (*link) (struct dentry *,struct inode *,struct dentry *);
2119 int (*unlink) (struct inode *,struct dentry *);
2120 int (*symlink) (struct user_namespace *, struct inode *,struct dentry *,
2122 int (*mkdir) (struct user_namespace *, struct inode *,struct dentry *,
2124 int (*rmdir) (struct inode *,struct dentry *);
2125 int (*mknod) (struct user_namespace *, struct inode *,struct dentry *,
2127 int (*rename) (struct user_namespace *, struct inode *, struct dentry *,
2128 struct inode *, struct dentry *, unsigned int);
2129 int (*setattr) (struct user_namespace *, struct dentry *,
2131 int (*getattr) (struct user_namespace *, const struct path *,
2132 struct kstat *, u32, unsigned int);
2133 ssize_t (*listxattr) (struct dentry *, char *, size_t);
2134 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
2136 int (*update_time)(struct inode *, struct timespec64 *, int);
2137 int (*atomic_open)(struct inode *, struct dentry *,
2138 struct file *, unsigned open_flag,
2139 umode_t create_mode);
2140 int (*tmpfile) (struct user_namespace *, struct inode *,
2141 struct dentry *, umode_t);
2142 int (*set_acl)(struct user_namespace *, struct inode *,
2143 struct posix_acl *, int);
2144 int (*fileattr_set)(struct user_namespace *mnt_userns,
2145 struct dentry *dentry, struct fileattr *fa);
2146 int (*fileattr_get)(struct dentry *dentry, struct fileattr *fa);
2147 } ____cacheline_aligned;
2149 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
2150 struct iov_iter *iter)
2152 return file->f_op->read_iter(kio, iter);
2155 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
2156 struct iov_iter *iter)
2158 return file->f_op->write_iter(kio, iter);
2161 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
2163 return file->f_op->mmap(file, vma);
2166 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
2167 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
2168 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
2169 loff_t, size_t, unsigned int);
2170 extern ssize_t generic_copy_file_range(struct file *file_in, loff_t pos_in,
2171 struct file *file_out, loff_t pos_out,
2172 size_t len, unsigned int flags);
2173 extern int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
2174 struct file *file_out, loff_t pos_out,
2176 unsigned int remap_flags);
2177 extern loff_t do_clone_file_range(struct file *file_in, loff_t pos_in,
2178 struct file *file_out, loff_t pos_out,
2179 loff_t len, unsigned int remap_flags);
2180 extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in,
2181 struct file *file_out, loff_t pos_out,
2182 loff_t len, unsigned int remap_flags);
2183 extern int vfs_dedupe_file_range(struct file *file,
2184 struct file_dedupe_range *same);
2185 extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
2186 struct file *dst_file, loff_t dst_pos,
2187 loff_t len, unsigned int remap_flags);
2190 struct super_operations {
2191 struct inode *(*alloc_inode)(struct super_block *sb);
2192 void (*destroy_inode)(struct inode *);
2193 void (*free_inode)(struct inode *);
2195 void (*dirty_inode) (struct inode *, int flags);
2196 int (*write_inode) (struct inode *, struct writeback_control *wbc);
2197 int (*drop_inode) (struct inode *);
2198 void (*evict_inode) (struct inode *);
2199 void (*put_super) (struct super_block *);
2200 int (*sync_fs)(struct super_block *sb, int wait);
2201 int (*freeze_super) (struct super_block *);
2202 int (*freeze_fs) (struct super_block *);
2203 int (*thaw_super) (struct super_block *);
2204 int (*unfreeze_fs) (struct super_block *);
2205 int (*statfs) (struct dentry *, struct kstatfs *);
2206 int (*remount_fs) (struct super_block *, int *, char *);
2207 void (*umount_begin) (struct super_block *);
2209 int (*show_options)(struct seq_file *, struct dentry *);
2210 int (*show_devname)(struct seq_file *, struct dentry *);
2211 int (*show_path)(struct seq_file *, struct dentry *);
2212 int (*show_stats)(struct seq_file *, struct dentry *);
2214 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
2215 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
2216 struct dquot **(*get_dquots)(struct inode *);
2218 long (*nr_cached_objects)(struct super_block *,
2219 struct shrink_control *);
2220 long (*free_cached_objects)(struct super_block *,
2221 struct shrink_control *);
2225 * Inode flags - they have no relation to superblock flags now
2227 #define S_SYNC (1 << 0) /* Writes are synced at once */
2228 #define S_NOATIME (1 << 1) /* Do not update access times */
2229 #define S_APPEND (1 << 2) /* Append-only file */
2230 #define S_IMMUTABLE (1 << 3) /* Immutable file */
2231 #define S_DEAD (1 << 4) /* removed, but still open directory */
2232 #define S_NOQUOTA (1 << 5) /* Inode is not counted to quota */
2233 #define S_DIRSYNC (1 << 6) /* Directory modifications are synchronous */
2234 #define S_NOCMTIME (1 << 7) /* Do not update file c/mtime */
2235 #define S_SWAPFILE (1 << 8) /* Do not truncate: swapon got its bmaps */
2236 #define S_PRIVATE (1 << 9) /* Inode is fs-internal */
2237 #define S_IMA (1 << 10) /* Inode has an associated IMA struct */
2238 #define S_AUTOMOUNT (1 << 11) /* Automount/referral quasi-directory */
2239 #define S_NOSEC (1 << 12) /* no suid or xattr security attributes */
2240 #ifdef CONFIG_FS_DAX
2241 #define S_DAX (1 << 13) /* Direct Access, avoiding the page cache */
2243 #define S_DAX 0 /* Make all the DAX code disappear */
2245 #define S_ENCRYPTED (1 << 14) /* Encrypted file (using fs/crypto/) */
2246 #define S_CASEFOLD (1 << 15) /* Casefolded file */
2247 #define S_VERITY (1 << 16) /* Verity file (using fs/verity/) */
2250 * Note that nosuid etc flags are inode-specific: setting some file-system
2251 * flags just means all the inodes inherit those flags by default. It might be
2252 * possible to override it selectively if you really wanted to with some
2253 * ioctl() that is not currently implemented.
2255 * Exception: SB_RDONLY is always applied to the entire file system.
2257 * Unfortunately, it is possible to change a filesystems flags with it mounted
2258 * with files in use. This means that all of the inodes will not have their
2259 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
2260 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
2262 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
2264 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
2265 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
2266 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
2267 ((inode)->i_flags & S_SYNC))
2268 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
2269 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
2270 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
2271 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
2272 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
2274 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
2275 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
2276 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
2277 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
2279 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
2280 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
2281 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
2282 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
2283 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
2284 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2285 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2286 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2287 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2288 #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD)
2289 #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY)
2291 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2292 (inode)->i_rdev == WHITEOUT_DEV)
2294 static inline bool HAS_UNMAPPED_ID(struct user_namespace *mnt_userns,
2295 struct inode *inode)
2297 return !uid_valid(i_uid_into_mnt(mnt_userns, inode)) ||
2298 !gid_valid(i_gid_into_mnt(mnt_userns, inode));
2301 static inline enum rw_hint file_write_hint(struct file *file)
2303 if (file->f_write_hint != WRITE_LIFE_NOT_SET)
2304 return file->f_write_hint;
2306 return file_inode(file)->i_write_hint;
2309 static inline int iocb_flags(struct file *file);
2311 static inline u16 ki_hint_validate(enum rw_hint hint)
2313 typeof(((struct kiocb *)0)->ki_hint) max_hint = -1;
2315 if (hint <= max_hint)
2320 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
2322 *kiocb = (struct kiocb) {
2324 .ki_flags = iocb_flags(filp),
2325 .ki_hint = ki_hint_validate(file_write_hint(filp)),
2326 .ki_ioprio = get_current_ioprio(),
2330 static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src,
2333 *kiocb = (struct kiocb) {
2335 .ki_flags = kiocb_src->ki_flags,
2336 .ki_hint = kiocb_src->ki_hint,
2337 .ki_ioprio = kiocb_src->ki_ioprio,
2338 .ki_pos = kiocb_src->ki_pos,
2343 * Inode state bits. Protected by inode->i_lock
2345 * Four bits determine the dirty state of the inode: I_DIRTY_SYNC,
2346 * I_DIRTY_DATASYNC, I_DIRTY_PAGES, and I_DIRTY_TIME.
2348 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2349 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2350 * various stages of removing an inode.
2352 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2354 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2355 * fdatasync() (unless I_DIRTY_DATASYNC is also set).
2356 * Timestamp updates are the usual cause.
2357 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2358 * these changes separately from I_DIRTY_SYNC so that we
2359 * don't have to write inode on fdatasync() when only
2360 * e.g. the timestamps have changed.
2361 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2362 * I_DIRTY_TIME The inode itself only has dirty timestamps, and the
2363 * lazytime mount option is enabled. We keep track of this
2364 * separately from I_DIRTY_SYNC in order to implement
2365 * lazytime. This gets cleared if I_DIRTY_INODE
2366 * (I_DIRTY_SYNC and/or I_DIRTY_DATASYNC) gets set. I.e.
2367 * either I_DIRTY_TIME *or* I_DIRTY_INODE can be set in
2368 * i_state, but not both. I_DIRTY_PAGES may still be set.
2369 * I_NEW Serves as both a mutex and completion notification.
2370 * New inodes set I_NEW. If two processes both create
2371 * the same inode, one of them will release its inode and
2372 * wait for I_NEW to be released before returning.
2373 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2374 * also cause waiting on I_NEW, without I_NEW actually
2375 * being set. find_inode() uses this to prevent returning
2376 * nearly-dead inodes.
2377 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2378 * is zero. I_FREEING must be set when I_WILL_FREE is
2380 * I_FREEING Set when inode is about to be freed but still has dirty
2381 * pages or buffers attached or the inode itself is still
2383 * I_CLEAR Added by clear_inode(). In this state the inode is
2384 * clean and can be destroyed. Inode keeps I_FREEING.
2386 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2387 * prohibited for many purposes. iget() must wait for
2388 * the inode to be completely released, then create it
2389 * anew. Other functions will just ignore such inodes,
2390 * if appropriate. I_NEW is used for waiting.
2392 * I_SYNC Writeback of inode is running. The bit is set during
2393 * data writeback, and cleared with a wakeup on the bit
2394 * address once it is done. The bit is also used to pin
2395 * the inode in memory for flusher thread.
2397 * I_REFERENCED Marks the inode as recently references on the LRU list.
2399 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2401 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2402 * synchronize competing switching instances and to tell
2403 * wb stat updates to grab the i_pages lock. See
2404 * inode_switch_wbs_work_fn() for details.
2406 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2407 * and work dirs among overlayfs mounts.
2409 * I_CREATING New object's inode in the middle of setting up.
2411 * I_DONTCACHE Evict inode as soon as it is not used anymore.
2413 * I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists.
2414 * Used to detect that mark_inode_dirty() should not move
2415 * inode between dirty lists.
2417 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2419 #define I_DIRTY_SYNC (1 << 0)
2420 #define I_DIRTY_DATASYNC (1 << 1)
2421 #define I_DIRTY_PAGES (1 << 2)
2423 #define I_NEW (1 << __I_NEW)
2424 #define I_WILL_FREE (1 << 4)
2425 #define I_FREEING (1 << 5)
2426 #define I_CLEAR (1 << 6)
2428 #define I_SYNC (1 << __I_SYNC)
2429 #define I_REFERENCED (1 << 8)
2430 #define __I_DIO_WAKEUP 9
2431 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2432 #define I_LINKABLE (1 << 10)
2433 #define I_DIRTY_TIME (1 << 11)
2434 #define I_WB_SWITCH (1 << 13)
2435 #define I_OVL_INUSE (1 << 14)
2436 #define I_CREATING (1 << 15)
2437 #define I_DONTCACHE (1 << 16)
2438 #define I_SYNC_QUEUED (1 << 17)
2440 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2441 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2442 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2444 extern void __mark_inode_dirty(struct inode *, int);
2445 static inline void mark_inode_dirty(struct inode *inode)
2447 __mark_inode_dirty(inode, I_DIRTY);
2450 static inline void mark_inode_dirty_sync(struct inode *inode)
2452 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2456 * Returns true if the given inode itself only has dirty timestamps (its pages
2457 * may still be dirty) and isn't currently being allocated or freed.
2458 * Filesystems should call this if when writing an inode when lazytime is
2459 * enabled, they want to opportunistically write the timestamps of other inodes
2460 * located very nearby on-disk, e.g. in the same inode block. This returns true
2461 * if the given inode is in need of such an opportunistic update. Requires
2462 * i_lock, or at least later re-checking under i_lock.
2464 static inline bool inode_is_dirtytime_only(struct inode *inode)
2466 return (inode->i_state & (I_DIRTY_TIME | I_NEW |
2467 I_FREEING | I_WILL_FREE)) == I_DIRTY_TIME;
2470 extern void inc_nlink(struct inode *inode);
2471 extern void drop_nlink(struct inode *inode);
2472 extern void clear_nlink(struct inode *inode);
2473 extern void set_nlink(struct inode *inode, unsigned int nlink);
2475 static inline void inode_inc_link_count(struct inode *inode)
2478 mark_inode_dirty(inode);
2481 static inline void inode_dec_link_count(struct inode *inode)
2484 mark_inode_dirty(inode);
2487 enum file_time_flags {
2494 extern bool atime_needs_update(const struct path *, struct inode *);
2495 extern void touch_atime(const struct path *);
2496 static inline void file_accessed(struct file *file)
2498 if (!(file->f_flags & O_NOATIME))
2499 touch_atime(&file->f_path);
2502 extern int file_modified(struct file *file);
2504 int sync_inode_metadata(struct inode *inode, int wait);
2506 struct file_system_type {
2509 #define FS_REQUIRES_DEV 1
2510 #define FS_BINARY_MOUNTDATA 2
2511 #define FS_HAS_SUBTYPE 4
2512 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2513 #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */
2514 #define FS_ALLOW_IDMAP 32 /* FS has been updated to handle vfs idmappings. */
2515 #define FS_THP_SUPPORT 8192 /* Remove once all fs converted */
2516 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2517 int (*init_fs_context)(struct fs_context *);
2518 const struct fs_parameter_spec *parameters;
2519 struct dentry *(*mount) (struct file_system_type *, int,
2520 const char *, void *);
2521 void (*kill_sb) (struct super_block *);
2522 struct module *owner;
2523 struct file_system_type * next;
2524 struct hlist_head fs_supers;
2526 struct lock_class_key s_lock_key;
2527 struct lock_class_key s_umount_key;
2528 struct lock_class_key s_vfs_rename_key;
2529 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2531 struct lock_class_key i_lock_key;
2532 struct lock_class_key i_mutex_key;
2533 struct lock_class_key invalidate_lock_key;
2534 struct lock_class_key i_mutex_dir_key;
2537 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2539 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2540 int flags, const char *dev_name, void *data,
2541 int (*fill_super)(struct super_block *, void *, int));
2542 extern struct dentry *mount_single(struct file_system_type *fs_type,
2543 int flags, void *data,
2544 int (*fill_super)(struct super_block *, void *, int));
2545 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2546 int flags, void *data,
2547 int (*fill_super)(struct super_block *, void *, int));
2548 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2549 void generic_shutdown_super(struct super_block *sb);
2550 void kill_block_super(struct super_block *sb);
2551 void kill_anon_super(struct super_block *sb);
2552 void kill_litter_super(struct super_block *sb);
2553 void deactivate_super(struct super_block *sb);
2554 void deactivate_locked_super(struct super_block *sb);
2555 int set_anon_super(struct super_block *s, void *data);
2556 int set_anon_super_fc(struct super_block *s, struct fs_context *fc);
2557 int get_anon_bdev(dev_t *);
2558 void free_anon_bdev(dev_t);
2559 struct super_block *sget_fc(struct fs_context *fc,
2560 int (*test)(struct super_block *, struct fs_context *),
2561 int (*set)(struct super_block *, struct fs_context *));
2562 struct super_block *sget(struct file_system_type *type,
2563 int (*test)(struct super_block *,void *),
2564 int (*set)(struct super_block *,void *),
2565 int flags, void *data);
2567 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2568 #define fops_get(fops) \
2569 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2570 #define fops_put(fops) \
2571 do { if (fops) module_put((fops)->owner); } while(0)
2573 * This one is to be used *ONLY* from ->open() instances.
2574 * fops must be non-NULL, pinned down *and* module dependencies
2575 * should be sufficient to pin the caller down as well.
2577 #define replace_fops(f, fops) \
2579 struct file *__file = (f); \
2580 fops_put(__file->f_op); \
2581 BUG_ON(!(__file->f_op = (fops))); \
2584 extern int register_filesystem(struct file_system_type *);
2585 extern int unregister_filesystem(struct file_system_type *);
2586 extern struct vfsmount *kern_mount(struct file_system_type *);
2587 extern void kern_unmount(struct vfsmount *mnt);
2588 extern int may_umount_tree(struct vfsmount *);
2589 extern int may_umount(struct vfsmount *);
2590 extern long do_mount(const char *, const char __user *,
2591 const char *, unsigned long, void *);
2592 extern struct vfsmount *collect_mounts(const struct path *);
2593 extern void drop_collected_mounts(struct vfsmount *);
2594 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2596 extern int vfs_statfs(const struct path *, struct kstatfs *);
2597 extern int user_statfs(const char __user *, struct kstatfs *);
2598 extern int fd_statfs(int, struct kstatfs *);
2599 extern int freeze_super(struct super_block *super);
2600 extern int thaw_super(struct super_block *super);
2601 extern bool our_mnt(struct vfsmount *mnt);
2602 extern __printf(2, 3)
2603 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2604 extern int super_setup_bdi(struct super_block *sb);
2606 extern int current_umask(void);
2608 extern void ihold(struct inode * inode);
2609 extern void iput(struct inode *);
2610 extern int generic_update_time(struct inode *, struct timespec64 *, int);
2613 extern struct kobject *fs_kobj;
2615 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2617 #ifdef CONFIG_FILE_LOCKING
2618 static inline int break_lease(struct inode *inode, unsigned int mode)
2621 * Since this check is lockless, we must ensure that any refcounts
2622 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2623 * could end up racing with tasks trying to set a new lease on this
2627 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2628 return __break_lease(inode, mode, FL_LEASE);
2632 static inline int break_deleg(struct inode *inode, unsigned int mode)
2635 * Since this check is lockless, we must ensure that any refcounts
2636 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2637 * could end up racing with tasks trying to set a new lease on this
2641 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2642 return __break_lease(inode, mode, FL_DELEG);
2646 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2650 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2651 if (ret == -EWOULDBLOCK && delegated_inode) {
2652 *delegated_inode = inode;
2658 static inline int break_deleg_wait(struct inode **delegated_inode)
2662 ret = break_deleg(*delegated_inode, O_WRONLY);
2663 iput(*delegated_inode);
2664 *delegated_inode = NULL;
2668 static inline int break_layout(struct inode *inode, bool wait)
2671 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2672 return __break_lease(inode,
2673 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2678 #else /* !CONFIG_FILE_LOCKING */
2679 static inline int break_lease(struct inode *inode, unsigned int mode)
2684 static inline int break_deleg(struct inode *inode, unsigned int mode)
2689 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2694 static inline int break_deleg_wait(struct inode **delegated_inode)
2700 static inline int break_layout(struct inode *inode, bool wait)
2705 #endif /* CONFIG_FILE_LOCKING */
2710 const char *name; /* pointer to actual string */
2711 const __user char *uptr; /* original userland pointer */
2713 struct audit_names *aname;
2716 static_assert(offsetof(struct filename, iname) % sizeof(long) == 0);
2718 static inline struct user_namespace *file_mnt_user_ns(struct file *file)
2720 return mnt_user_ns(file->f_path.mnt);
2722 extern long vfs_truncate(const struct path *, loff_t);
2723 int do_truncate(struct user_namespace *, struct dentry *, loff_t start,
2724 unsigned int time_attrs, struct file *filp);
2725 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2727 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2729 extern struct file *file_open_name(struct filename *, int, umode_t);
2730 extern struct file *filp_open(const char *, int, umode_t);
2731 extern struct file *file_open_root(const struct path *,
2732 const char *, int, umode_t);
2733 static inline struct file *file_open_root_mnt(struct vfsmount *mnt,
2734 const char *name, int flags, umode_t mode)
2736 return file_open_root(&(struct path){.mnt = mnt, .dentry = mnt->mnt_root},
2739 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2740 extern struct file * open_with_fake_path(const struct path *, int,
2741 struct inode*, const struct cred *);
2742 static inline struct file *file_clone_open(struct file *file)
2744 return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2746 extern int filp_close(struct file *, fl_owner_t id);
2748 extern struct filename *getname_flags(const char __user *, int, int *);
2749 extern struct filename *getname_uflags(const char __user *, int);
2750 extern struct filename *getname(const char __user *);
2751 extern struct filename *getname_kernel(const char *);
2752 extern void putname(struct filename *name);
2754 extern int finish_open(struct file *file, struct dentry *dentry,
2755 int (*open)(struct inode *, struct file *));
2756 extern int finish_no_open(struct file *file, struct dentry *dentry);
2759 extern void __init vfs_caches_init_early(void);
2760 extern void __init vfs_caches_init(void);
2762 extern struct kmem_cache *names_cachep;
2764 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2765 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2767 extern struct super_block *blockdev_superblock;
2768 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2770 return IS_ENABLED(CONFIG_BLOCK) && sb == blockdev_superblock;
2773 void emergency_thaw_all(void);
2774 extern int sync_filesystem(struct super_block *);
2775 extern const struct file_operations def_blk_fops;
2776 extern const struct file_operations def_chr_fops;
2779 #define CHRDEV_MAJOR_MAX 512
2780 /* Marks the bottom of the first segment of free char majors */
2781 #define CHRDEV_MAJOR_DYN_END 234
2782 /* Marks the top and bottom of the second segment of free char majors */
2783 #define CHRDEV_MAJOR_DYN_EXT_START 511
2784 #define CHRDEV_MAJOR_DYN_EXT_END 384
2786 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2787 extern int register_chrdev_region(dev_t, unsigned, const char *);
2788 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2789 unsigned int count, const char *name,
2790 const struct file_operations *fops);
2791 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2792 unsigned int count, const char *name);
2793 extern void unregister_chrdev_region(dev_t, unsigned);
2794 extern void chrdev_show(struct seq_file *,off_t);
2796 static inline int register_chrdev(unsigned int major, const char *name,
2797 const struct file_operations *fops)
2799 return __register_chrdev(major, 0, 256, name, fops);
2802 static inline void unregister_chrdev(unsigned int major, const char *name)
2804 __unregister_chrdev(major, 0, 256, name);
2807 extern void init_special_inode(struct inode *, umode_t, dev_t);
2809 /* Invalid inode operations -- fs/bad_inode.c */
2810 extern void make_bad_inode(struct inode *);
2811 extern bool is_bad_inode(struct inode *);
2813 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2814 pgoff_t start, pgoff_t end);
2816 void invalidate_mapping_pagevec(struct address_space *mapping,
2817 pgoff_t start, pgoff_t end,
2818 unsigned long *nr_pagevec);
2820 static inline void invalidate_remote_inode(struct inode *inode)
2822 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2823 S_ISLNK(inode->i_mode))
2824 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2826 extern int invalidate_inode_pages2(struct address_space *mapping);
2827 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2828 pgoff_t start, pgoff_t end);
2829 extern int write_inode_now(struct inode *, int);
2830 extern int filemap_fdatawrite(struct address_space *);
2831 extern int filemap_flush(struct address_space *);
2832 extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2833 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2835 extern int filemap_fdatawait_range_keep_errors(struct address_space *mapping,
2836 loff_t start_byte, loff_t end_byte);
2838 static inline int filemap_fdatawait(struct address_space *mapping)
2840 return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2843 extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2845 extern bool filemap_range_needs_writeback(struct address_space *,
2846 loff_t lstart, loff_t lend);
2847 extern int filemap_write_and_wait_range(struct address_space *mapping,
2848 loff_t lstart, loff_t lend);
2849 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2850 loff_t start, loff_t end, int sync_mode);
2851 extern int filemap_fdatawrite_range(struct address_space *mapping,
2852 loff_t start, loff_t end);
2853 extern int filemap_check_errors(struct address_space *mapping);
2854 extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2855 int filemap_fdatawrite_wbc(struct address_space *mapping,
2856 struct writeback_control *wbc);
2858 static inline int filemap_write_and_wait(struct address_space *mapping)
2860 return filemap_write_and_wait_range(mapping, 0, LLONG_MAX);
2863 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2865 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2866 extern int __must_check file_write_and_wait_range(struct file *file,
2867 loff_t start, loff_t end);
2869 static inline int file_write_and_wait(struct file *file)
2871 return file_write_and_wait_range(file, 0, LLONG_MAX);
2875 * filemap_set_wb_err - set a writeback error on an address_space
2876 * @mapping: mapping in which to set writeback error
2877 * @err: error to be set in mapping
2879 * When writeback fails in some way, we must record that error so that
2880 * userspace can be informed when fsync and the like are called. We endeavor
2881 * to report errors on any file that was open at the time of the error. Some
2882 * internal callers also need to know when writeback errors have occurred.
2884 * When a writeback error occurs, most filesystems will want to call
2885 * filemap_set_wb_err to record the error in the mapping so that it will be
2886 * automatically reported whenever fsync is called on the file.
2888 static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2890 /* Fastpath for common case of no error */
2892 __filemap_set_wb_err(mapping, err);
2896 * filemap_check_wb_err - has an error occurred since the mark was sampled?
2897 * @mapping: mapping to check for writeback errors
2898 * @since: previously-sampled errseq_t
2900 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2901 * the given value was sampled.
2903 * If it has then report the latest error set, otherwise return 0.
2905 static inline int filemap_check_wb_err(struct address_space *mapping,
2908 return errseq_check(&mapping->wb_err, since);
2912 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2913 * @mapping: mapping to be sampled
2915 * Writeback errors are always reported relative to a particular sample point
2916 * in the past. This function provides those sample points.
2918 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2920 return errseq_sample(&mapping->wb_err);
2924 * file_sample_sb_err - sample the current errseq_t to test for later errors
2925 * @file: file pointer to be sampled
2927 * Grab the most current superblock-level errseq_t value for the given
2930 static inline errseq_t file_sample_sb_err(struct file *file)
2932 return errseq_sample(&file->f_path.dentry->d_sb->s_wb_err);
2935 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2937 extern int vfs_fsync(struct file *file, int datasync);
2939 extern int sync_file_range(struct file *file, loff_t offset, loff_t nbytes,
2940 unsigned int flags);
2943 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2944 * to already be updated for the write, and will return either the amount
2945 * of bytes passed in, or an error if syncing the file failed.
2947 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2949 if (iocb->ki_flags & IOCB_DSYNC) {
2950 int ret = vfs_fsync_range(iocb->ki_filp,
2951 iocb->ki_pos - count, iocb->ki_pos - 1,
2952 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2960 extern void emergency_sync(void);
2961 extern void emergency_remount(void);
2964 extern int bmap(struct inode *inode, sector_t *block);
2966 static inline int bmap(struct inode *inode, sector_t *block)
2972 int notify_change(struct user_namespace *, struct dentry *,
2973 struct iattr *, struct inode **);
2974 int inode_permission(struct user_namespace *, struct inode *, int);
2975 int generic_permission(struct user_namespace *, struct inode *, int);
2976 static inline int file_permission(struct file *file, int mask)
2978 return inode_permission(file_mnt_user_ns(file),
2979 file_inode(file), mask);
2981 static inline int path_permission(const struct path *path, int mask)
2983 return inode_permission(mnt_user_ns(path->mnt),
2984 d_inode(path->dentry), mask);
2986 int __check_sticky(struct user_namespace *mnt_userns, struct inode *dir,
2987 struct inode *inode);
2989 static inline bool execute_ok(struct inode *inode)
2991 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2994 static inline bool inode_wrong_type(const struct inode *inode, umode_t mode)
2996 return (inode->i_mode ^ mode) & S_IFMT;
2999 static inline void file_start_write(struct file *file)
3001 if (!S_ISREG(file_inode(file)->i_mode))
3003 sb_start_write(file_inode(file)->i_sb);
3006 static inline bool file_start_write_trylock(struct file *file)
3008 if (!S_ISREG(file_inode(file)->i_mode))
3010 return sb_start_write_trylock(file_inode(file)->i_sb);
3013 static inline void file_end_write(struct file *file)
3015 if (!S_ISREG(file_inode(file)->i_mode))
3017 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
3021 * get_write_access() gets write permission for a file.
3022 * put_write_access() releases this write permission.
3023 * This is used for regular files.
3024 * We cannot support write (and maybe mmap read-write shared) accesses and
3025 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
3026 * can have the following values:
3027 * 0: no writers, no VM_DENYWRITE mappings
3028 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
3029 * > 0: (i_writecount) users are writing to the file.
3031 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
3032 * except for the cases where we don't hold i_writecount yet. Then we need to
3033 * use {get,deny}_write_access() - these functions check the sign and refuse
3034 * to do the change if sign is wrong.
3036 static inline int get_write_access(struct inode *inode)
3038 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
3040 static inline int deny_write_access(struct file *file)
3042 struct inode *inode = file_inode(file);
3043 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
3045 static inline void put_write_access(struct inode * inode)
3047 atomic_dec(&inode->i_writecount);
3049 static inline void allow_write_access(struct file *file)
3052 atomic_inc(&file_inode(file)->i_writecount);
3054 static inline bool inode_is_open_for_write(const struct inode *inode)
3056 return atomic_read(&inode->i_writecount) > 0;
3059 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
3060 static inline void i_readcount_dec(struct inode *inode)
3062 BUG_ON(!atomic_read(&inode->i_readcount));
3063 atomic_dec(&inode->i_readcount);
3065 static inline void i_readcount_inc(struct inode *inode)
3067 atomic_inc(&inode->i_readcount);
3070 static inline void i_readcount_dec(struct inode *inode)
3074 static inline void i_readcount_inc(struct inode *inode)
3079 extern int do_pipe_flags(int *, int);
3081 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
3082 ssize_t __kernel_read(struct file *file, void *buf, size_t count, loff_t *pos);
3083 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
3084 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
3085 extern struct file * open_exec(const char *);
3087 /* fs/dcache.c -- generic fs support functions */
3088 extern bool is_subdir(struct dentry *, struct dentry *);
3089 extern bool path_is_under(const struct path *, const struct path *);
3091 extern char *file_path(struct file *, char *, int);
3093 #include <linux/err.h>
3095 /* needed for stackable file system support */
3096 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
3098 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
3100 extern int inode_init_always(struct super_block *, struct inode *);
3101 extern void inode_init_once(struct inode *);
3102 extern void address_space_init_once(struct address_space *mapping);
3103 extern struct inode * igrab(struct inode *);
3104 extern ino_t iunique(struct super_block *, ino_t);
3105 extern int inode_needs_sync(struct inode *inode);
3106 extern int generic_delete_inode(struct inode *inode);
3107 static inline int generic_drop_inode(struct inode *inode)
3109 return !inode->i_nlink || inode_unhashed(inode);
3111 extern void d_mark_dontcache(struct inode *inode);
3113 extern struct inode *ilookup5_nowait(struct super_block *sb,
3114 unsigned long hashval, int (*test)(struct inode *, void *),
3116 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
3117 int (*test)(struct inode *, void *), void *data);
3118 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
3120 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
3121 int (*test)(struct inode *, void *),
3122 int (*set)(struct inode *, void *),
3124 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
3125 extern struct inode * iget_locked(struct super_block *, unsigned long);
3126 extern struct inode *find_inode_nowait(struct super_block *,
3128 int (*match)(struct inode *,
3129 unsigned long, void *),
3131 extern struct inode *find_inode_rcu(struct super_block *, unsigned long,
3132 int (*)(struct inode *, void *), void *);
3133 extern struct inode *find_inode_by_ino_rcu(struct super_block *, unsigned long);
3134 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
3135 extern int insert_inode_locked(struct inode *);
3136 #ifdef CONFIG_DEBUG_LOCK_ALLOC
3137 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
3139 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
3141 extern void unlock_new_inode(struct inode *);
3142 extern void discard_new_inode(struct inode *);
3143 extern unsigned int get_next_ino(void);
3144 extern void evict_inodes(struct super_block *sb);
3147 * Userspace may rely on the the inode number being non-zero. For example, glibc
3148 * simply ignores files with zero i_ino in unlink() and other places.
3150 * As an additional complication, if userspace was compiled with
3151 * _FILE_OFFSET_BITS=32 on a 64-bit kernel we'll only end up reading out the
3152 * lower 32 bits, so we need to check that those aren't zero explicitly. With
3153 * _FILE_OFFSET_BITS=64, this may cause some harmless false-negatives, but
3154 * better safe than sorry.
3156 static inline bool is_zero_ino(ino_t ino)
3158 return (u32)ino == 0;
3161 extern void __iget(struct inode * inode);
3162 extern void iget_failed(struct inode *);
3163 extern void clear_inode(struct inode *);
3164 extern void __destroy_inode(struct inode *);
3165 extern struct inode *new_inode_pseudo(struct super_block *sb);
3166 extern struct inode *new_inode(struct super_block *sb);
3167 extern void free_inode_nonrcu(struct inode *inode);
3168 extern int should_remove_suid(struct dentry *);
3169 extern int file_remove_privs(struct file *);
3171 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
3172 static inline void insert_inode_hash(struct inode *inode)
3174 __insert_inode_hash(inode, inode->i_ino);
3177 extern void __remove_inode_hash(struct inode *);
3178 static inline void remove_inode_hash(struct inode *inode)
3180 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
3181 __remove_inode_hash(inode);
3184 extern void inode_sb_list_add(struct inode *inode);
3186 extern int sb_set_blocksize(struct super_block *, int);
3187 extern int sb_min_blocksize(struct super_block *, int);
3189 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
3190 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
3191 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
3192 extern int generic_write_check_limits(struct file *file, loff_t pos,
3194 extern int generic_file_rw_checks(struct file *file_in, struct file *file_out);
3195 ssize_t filemap_read(struct kiocb *iocb, struct iov_iter *to,
3196 ssize_t already_read);
3197 extern ssize_t generic_file_read_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_write_iter(struct kiocb *, struct iov_iter *);
3200 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
3201 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
3203 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
3205 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
3207 ssize_t vfs_iocb_iter_read(struct file *file, struct kiocb *iocb,
3208 struct iov_iter *iter);
3209 ssize_t vfs_iocb_iter_write(struct file *file, struct kiocb *iocb,
3210 struct iov_iter *iter);
3213 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
3214 struct pipe_inode_info *, size_t, unsigned int);
3215 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
3216 struct file *, loff_t *, size_t, unsigned int);
3217 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
3218 struct file *out, loff_t *, size_t len, unsigned int flags);
3219 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
3220 loff_t *opos, size_t len, unsigned int flags);
3224 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3225 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3226 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
3227 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3228 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3229 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3230 int whence, loff_t maxsize, loff_t eof);
3231 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3232 int whence, loff_t size);
3233 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3234 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3235 extern int generic_file_open(struct inode * inode, struct file * filp);
3236 extern int nonseekable_open(struct inode * inode, struct file * filp);
3237 extern int stream_open(struct inode * inode, struct file * filp);
3240 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3241 loff_t file_offset);
3244 /* need locking between buffered and direct access */
3247 /* filesystem does not support filling holes */
3248 DIO_SKIP_HOLES = 0x02,
3251 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3252 struct block_device *bdev, struct iov_iter *iter,
3253 get_block_t get_block,
3254 dio_iodone_t end_io, dio_submit_t submit_io,
3257 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3258 struct inode *inode,
3259 struct iov_iter *iter,
3260 get_block_t get_block)
3262 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3263 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3267 void inode_dio_wait(struct inode *inode);
3270 * inode_dio_begin - signal start of a direct I/O requests
3271 * @inode: inode the direct I/O happens on
3273 * This is called once we've finished processing a direct I/O request,
3274 * and is used to wake up callers waiting for direct I/O to be quiesced.
3276 static inline void inode_dio_begin(struct inode *inode)
3278 atomic_inc(&inode->i_dio_count);
3282 * inode_dio_end - signal finish of a direct I/O requests
3283 * @inode: inode the direct I/O happens on
3285 * This is called once we've finished processing a direct I/O request,
3286 * and is used to wake up callers waiting for direct I/O to be quiesced.
3288 static inline void inode_dio_end(struct inode *inode)
3290 if (atomic_dec_and_test(&inode->i_dio_count))
3291 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3295 * Warn about a page cache invalidation failure diring a direct I/O write.
3297 void dio_warn_stale_pagecache(struct file *filp);
3299 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3302 extern const struct file_operations generic_ro_fops;
3304 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3306 extern int readlink_copy(char __user *, int, const char *);
3307 extern int page_readlink(struct dentry *, char __user *, int);
3308 extern const char *page_get_link(struct dentry *, struct inode *,
3309 struct delayed_call *);
3310 extern void page_put_link(void *);
3311 extern int __page_symlink(struct inode *inode, const char *symname, int len,
3313 extern int page_symlink(struct inode *inode, const char *symname, int len);
3314 extern const struct inode_operations page_symlink_inode_operations;
3315 extern void kfree_link(void *);
3316 void generic_fillattr(struct user_namespace *, struct inode *, struct kstat *);
3317 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3318 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3319 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3320 void inode_add_bytes(struct inode *inode, loff_t bytes);
3321 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3322 void inode_sub_bytes(struct inode *inode, loff_t bytes);
3323 static inline loff_t __inode_get_bytes(struct inode *inode)
3325 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3327 loff_t inode_get_bytes(struct inode *inode);
3328 void inode_set_bytes(struct inode *inode, loff_t bytes);
3329 const char *simple_get_link(struct dentry *, struct inode *,
3330 struct delayed_call *);
3331 extern const struct inode_operations simple_symlink_inode_operations;
3333 extern int iterate_dir(struct file *, struct dir_context *);
3335 int vfs_fstatat(int dfd, const char __user *filename, struct kstat *stat,
3337 int vfs_fstat(int fd, struct kstat *stat);
3339 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3341 return vfs_fstatat(AT_FDCWD, filename, stat, 0);
3343 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3345 return vfs_fstatat(AT_FDCWD, name, stat, AT_SYMLINK_NOFOLLOW);
3348 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3349 extern int vfs_readlink(struct dentry *, char __user *, int);
3351 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3352 extern void put_filesystem(struct file_system_type *fs);
3353 extern struct file_system_type *get_fs_type(const char *name);
3354 extern struct super_block *get_super(struct block_device *);
3355 extern struct super_block *get_active_super(struct block_device *bdev);
3356 extern void drop_super(struct super_block *sb);
3357 extern void drop_super_exclusive(struct super_block *sb);
3358 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3359 extern void iterate_supers_type(struct file_system_type *,
3360 void (*)(struct super_block *, void *), void *);
3362 extern int dcache_dir_open(struct inode *, struct file *);
3363 extern int dcache_dir_close(struct inode *, struct file *);
3364 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3365 extern int dcache_readdir(struct file *, struct dir_context *);
3366 extern int simple_setattr(struct user_namespace *, struct dentry *,
3368 extern int simple_getattr(struct user_namespace *, const struct path *,
3369 struct kstat *, u32, unsigned int);
3370 extern int simple_statfs(struct dentry *, struct kstatfs *);
3371 extern int simple_open(struct inode *inode, struct file *file);
3372 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3373 extern int simple_unlink(struct inode *, struct dentry *);
3374 extern int simple_rmdir(struct inode *, struct dentry *);
3375 extern int simple_rename(struct user_namespace *, struct inode *,
3376 struct dentry *, struct inode *, struct dentry *,
3378 extern void simple_recursive_removal(struct dentry *,
3379 void (*callback)(struct dentry *));
3380 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3381 extern void noop_invalidatepage(struct page *page, unsigned int offset,
3382 unsigned int length);
3383 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3384 extern int simple_empty(struct dentry *);
3385 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3386 loff_t pos, unsigned len, unsigned flags,
3387 struct page **pagep, void **fsdata);
3388 extern const struct address_space_operations ram_aops;
3389 extern int always_delete_dentry(const struct dentry *);
3390 extern struct inode *alloc_anon_inode(struct super_block *);
3391 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3392 extern const struct dentry_operations simple_dentry_operations;
3394 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3395 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3396 extern const struct file_operations simple_dir_operations;
3397 extern const struct inode_operations simple_dir_inode_operations;
3398 extern void make_empty_dir_inode(struct inode *inode);
3399 extern bool is_empty_dir_inode(struct inode *inode);
3400 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3401 struct dentry *d_alloc_name(struct dentry *, const char *);
3402 extern int simple_fill_super(struct super_block *, unsigned long,
3403 const struct tree_descr *);
3404 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3405 extern void simple_release_fs(struct vfsmount **mount, int *count);
3407 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3408 loff_t *ppos, const void *from, size_t available);
3409 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3410 const void __user *from, size_t count);
3412 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3413 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3415 extern int generic_check_addressable(unsigned, u64);
3417 extern void generic_set_encrypted_ci_d_ops(struct dentry *dentry);
3419 #ifdef CONFIG_MIGRATION
3420 extern int buffer_migrate_page(struct address_space *,
3421 struct page *, struct page *,
3423 extern int buffer_migrate_page_norefs(struct address_space *,
3424 struct page *, struct page *,
3427 #define buffer_migrate_page NULL
3428 #define buffer_migrate_page_norefs NULL
3431 int setattr_prepare(struct user_namespace *, struct dentry *, struct iattr *);
3432 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3433 void setattr_copy(struct user_namespace *, struct inode *inode,
3434 const struct iattr *attr);
3436 extern int file_update_time(struct file *file);
3438 static inline bool vma_is_dax(const struct vm_area_struct *vma)
3440 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3443 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3445 struct inode *inode;
3447 if (!IS_ENABLED(CONFIG_FS_DAX) || !vma->vm_file)
3449 if (!vma_is_dax(vma))
3451 inode = file_inode(vma->vm_file);
3452 if (S_ISCHR(inode->i_mode))
3453 return false; /* device-dax */
3457 static inline int iocb_flags(struct file *file)
3460 if (file->f_flags & O_APPEND)
3462 if (file->f_flags & O_DIRECT)
3464 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3466 if (file->f_flags & __O_SYNC)
3471 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3473 int kiocb_flags = 0;
3475 /* make sure there's no overlap between RWF and private IOCB flags */
3476 BUILD_BUG_ON((__force int) RWF_SUPPORTED & IOCB_EVENTFD);
3480 if (unlikely(flags & ~RWF_SUPPORTED))
3483 if (flags & RWF_NOWAIT) {
3484 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3486 kiocb_flags |= IOCB_NOIO;
3488 kiocb_flags |= (__force int) (flags & RWF_SUPPORTED);
3489 if (flags & RWF_SYNC)
3490 kiocb_flags |= IOCB_DSYNC;
3492 ki->ki_flags |= kiocb_flags;
3496 static inline ino_t parent_ino(struct dentry *dentry)
3501 * Don't strictly need d_lock here? If the parent ino could change
3502 * then surely we'd have a deeper race in the caller?
3504 spin_lock(&dentry->d_lock);
3505 res = dentry->d_parent->d_inode->i_ino;
3506 spin_unlock(&dentry->d_lock);
3510 /* Transaction based IO helpers */
3513 * An argresp is stored in an allocated page and holds the
3514 * size of the argument or response, along with its content
3516 struct simple_transaction_argresp {
3521 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3523 char *simple_transaction_get(struct file *file, const char __user *buf,
3525 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3526 size_t size, loff_t *pos);
3527 int simple_transaction_release(struct inode *inode, struct file *file);
3529 void simple_transaction_set(struct file *file, size_t n);
3532 * simple attribute files
3534 * These attributes behave similar to those in sysfs:
3536 * Writing to an attribute immediately sets a value, an open file can be
3537 * written to multiple times.
3539 * Reading from an attribute creates a buffer from the value that might get
3540 * read with multiple read calls. When the attribute has been read
3541 * completely, no further read calls are possible until the file is opened
3544 * All attributes contain a text representation of a numeric value
3545 * that are accessed with the get() and set() functions.
3547 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3548 static int __fops ## _open(struct inode *inode, struct file *file) \
3550 __simple_attr_check_format(__fmt, 0ull); \
3551 return simple_attr_open(inode, file, __get, __set, __fmt); \
3553 static const struct file_operations __fops = { \
3554 .owner = THIS_MODULE, \
3555 .open = __fops ## _open, \
3556 .release = simple_attr_release, \
3557 .read = simple_attr_read, \
3558 .write = simple_attr_write, \
3559 .llseek = generic_file_llseek, \
3562 static inline __printf(1, 2)
3563 void __simple_attr_check_format(const char *fmt, ...)
3565 /* don't do anything, just let the compiler check the arguments; */
3568 int simple_attr_open(struct inode *inode, struct file *file,
3569 int (*get)(void *, u64 *), int (*set)(void *, u64),
3571 int simple_attr_release(struct inode *inode, struct file *file);
3572 ssize_t simple_attr_read(struct file *file, char __user *buf,
3573 size_t len, loff_t *ppos);
3574 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3575 size_t len, loff_t *ppos);
3578 int proc_nr_files(struct ctl_table *table, int write,
3579 void *buffer, size_t *lenp, loff_t *ppos);
3580 int proc_nr_dentry(struct ctl_table *table, int write,
3581 void *buffer, size_t *lenp, loff_t *ppos);
3582 int proc_nr_inodes(struct ctl_table *table, int write,
3583 void *buffer, size_t *lenp, loff_t *ppos);
3584 int __init get_filesystem_list(char *buf);
3586 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3587 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3589 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3590 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3591 (flag & __FMODE_NONOTIFY)))
3593 static inline bool is_sxid(umode_t mode)
3595 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3598 static inline int check_sticky(struct user_namespace *mnt_userns,
3599 struct inode *dir, struct inode *inode)
3601 if (!(dir->i_mode & S_ISVTX))
3604 return __check_sticky(mnt_userns, dir, inode);
3607 static inline void inode_has_no_xattr(struct inode *inode)
3609 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3610 inode->i_flags |= S_NOSEC;
3613 static inline bool is_root_inode(struct inode *inode)
3615 return inode == inode->i_sb->s_root->d_inode;
3618 static inline bool dir_emit(struct dir_context *ctx,
3619 const char *name, int namelen,
3620 u64 ino, unsigned type)
3622 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3624 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3626 return ctx->actor(ctx, ".", 1, ctx->pos,
3627 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3629 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3631 return ctx->actor(ctx, "..", 2, ctx->pos,
3632 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3634 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3636 if (ctx->pos == 0) {
3637 if (!dir_emit_dot(file, ctx))
3641 if (ctx->pos == 1) {
3642 if (!dir_emit_dotdot(file, ctx))
3648 static inline bool dir_relax(struct inode *inode)
3650 inode_unlock(inode);
3652 return !IS_DEADDIR(inode);
3655 static inline bool dir_relax_shared(struct inode *inode)
3657 inode_unlock_shared(inode);
3658 inode_lock_shared(inode);
3659 return !IS_DEADDIR(inode);
3662 extern bool path_noexec(const struct path *path);
3663 extern void inode_nohighmem(struct inode *inode);
3666 extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len,
3668 extern int generic_fadvise(struct file *file, loff_t offset, loff_t len,
3672 * Flush file data before changing attributes. Caller must hold any locks
3673 * required to prevent further writes to this file until we're done setting
3676 static inline int inode_drain_writes(struct inode *inode)
3678 inode_dio_wait(inode);
3679 return filemap_write_and_wait(inode->i_mapping);
3682 #endif /* _LINUX_FS_H */