1 /* SPDX-License-Identifier: GPL-2.0 */
5 #include <linux/linkage.h>
6 #include <linux/wait_bit.h>
7 #include <linux/kdev_t.h>
8 #include <linux/dcache.h>
9 #include <linux/path.h>
10 #include <linux/stat.h>
11 #include <linux/cache.h>
12 #include <linux/list.h>
13 #include <linux/list_lru.h>
14 #include <linux/llist.h>
15 #include <linux/radix-tree.h>
16 #include <linux/xarray.h>
17 #include <linux/rbtree.h>
18 #include <linux/init.h>
19 #include <linux/pid.h>
20 #include <linux/bug.h>
21 #include <linux/mutex.h>
22 #include <linux/rwsem.h>
23 #include <linux/mm_types.h>
24 #include <linux/capability.h>
25 #include <linux/semaphore.h>
26 #include <linux/fcntl.h>
27 #include <linux/rculist_bl.h>
28 #include <linux/atomic.h>
29 #include <linux/shrinker.h>
30 #include <linux/migrate_mode.h>
31 #include <linux/uidgid.h>
32 #include <linux/lockdep.h>
33 #include <linux/percpu-rwsem.h>
34 #include <linux/workqueue.h>
35 #include <linux/delayed_call.h>
36 #include <linux/uuid.h>
37 #include <linux/errseq.h>
38 #include <linux/ioprio.h>
39 #include <linux/fs_types.h>
40 #include <linux/build_bug.h>
41 #include <linux/stddef.h>
42 #include <linux/mount.h>
43 #include <linux/cred.h>
44 #include <linux/mnt_idmapping.h>
46 #include <asm/byteorder.h>
47 #include <uapi/linux/fs.h>
49 struct backing_dev_info;
53 struct export_operations;
54 struct fiemap_extent_info;
59 struct pipe_inode_info;
60 struct poll_table_struct;
62 struct vm_area_struct;
65 struct swap_info_struct;
67 struct workqueue_struct;
70 struct fscrypt_operations;
72 struct fsverity_operations;
74 struct fs_parameter_spec;
77 extern void __init inode_init(void);
78 extern void __init inode_init_early(void);
79 extern void __init files_init(void);
80 extern void __init files_maxfiles_init(void);
82 extern unsigned long get_max_files(void);
83 extern unsigned int sysctl_nr_open;
84 extern int leases_enable, lease_break_time;
86 typedef __kernel_rwf_t rwf_t;
89 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
90 struct buffer_head *bh_result, int create);
91 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
92 ssize_t bytes, void *private);
94 #define MAY_EXEC 0x00000001
95 #define MAY_WRITE 0x00000002
96 #define MAY_READ 0x00000004
97 #define MAY_APPEND 0x00000008
98 #define MAY_ACCESS 0x00000010
99 #define MAY_OPEN 0x00000020
100 #define MAY_CHDIR 0x00000040
101 /* called from RCU mode, don't block */
102 #define MAY_NOT_BLOCK 0x00000080
105 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
106 * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
109 /* file is open for reading */
110 #define FMODE_READ ((__force fmode_t)0x1)
111 /* file is open for writing */
112 #define FMODE_WRITE ((__force fmode_t)0x2)
113 /* file is seekable */
114 #define FMODE_LSEEK ((__force fmode_t)0x4)
115 /* file can be accessed using pread */
116 #define FMODE_PREAD ((__force fmode_t)0x8)
117 /* file can be accessed using pwrite */
118 #define FMODE_PWRITE ((__force fmode_t)0x10)
119 /* File is opened for execution with sys_execve / sys_uselib */
120 #define FMODE_EXEC ((__force fmode_t)0x20)
121 /* File is opened with O_NDELAY (only set for block devices) */
122 #define FMODE_NDELAY ((__force fmode_t)0x40)
123 /* File is opened with O_EXCL (only set for block devices) */
124 #define FMODE_EXCL ((__force fmode_t)0x80)
125 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
126 (specialy hack for floppy.c) */
127 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
128 /* 32bit hashes as llseek() offset (for directories) */
129 #define FMODE_32BITHASH ((__force fmode_t)0x200)
130 /* 64bit hashes as llseek() offset (for directories) */
131 #define FMODE_64BITHASH ((__force fmode_t)0x400)
134 * Don't update ctime and mtime.
136 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
137 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
139 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
141 /* Expect random access pattern */
142 #define FMODE_RANDOM ((__force fmode_t)0x1000)
144 /* File is huge (eg. /dev/mem): treat loff_t as unsigned */
145 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
147 /* File is opened with O_PATH; almost nothing can be done with it */
148 #define FMODE_PATH ((__force fmode_t)0x4000)
150 /* File needs atomic accesses to f_pos */
151 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
152 /* Write access to underlying fs */
153 #define FMODE_WRITER ((__force fmode_t)0x10000)
154 /* Has read method(s) */
155 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
156 /* Has write method(s) */
157 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
159 #define FMODE_OPENED ((__force fmode_t)0x80000)
160 #define FMODE_CREATED ((__force fmode_t)0x100000)
162 /* File is stream-like */
163 #define FMODE_STREAM ((__force fmode_t)0x200000)
165 /* File was opened by fanotify and shouldn't generate fanotify events */
166 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
168 /* File is capable of returning -EAGAIN if I/O will block */
169 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
171 /* File represents mount that needs unmounting */
172 #define FMODE_NEED_UNMOUNT ((__force fmode_t)0x10000000)
174 /* File does not contribute to nr_files count */
175 #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000)
177 /* File supports async buffered reads */
178 #define FMODE_BUF_RASYNC ((__force fmode_t)0x40000000)
181 * Attribute flags. These should be or-ed together to figure out what
184 #define ATTR_MODE (1 << 0)
185 #define ATTR_UID (1 << 1)
186 #define ATTR_GID (1 << 2)
187 #define ATTR_SIZE (1 << 3)
188 #define ATTR_ATIME (1 << 4)
189 #define ATTR_MTIME (1 << 5)
190 #define ATTR_CTIME (1 << 6)
191 #define ATTR_ATIME_SET (1 << 7)
192 #define ATTR_MTIME_SET (1 << 8)
193 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
194 #define ATTR_KILL_SUID (1 << 11)
195 #define ATTR_KILL_SGID (1 << 12)
196 #define ATTR_FILE (1 << 13)
197 #define ATTR_KILL_PRIV (1 << 14)
198 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
199 #define ATTR_TIMES_SET (1 << 16)
200 #define ATTR_TOUCH (1 << 17)
203 * Whiteout is represented by a char device. The following constants define the
204 * mode and device number to use.
206 #define WHITEOUT_MODE 0
207 #define WHITEOUT_DEV 0
210 * This is the Inode Attributes structure, used for notify_change(). It
211 * uses the above definitions as flags, to know which values have changed.
212 * Also, in this manner, a Filesystem can look at only the values it cares
213 * about. Basically, these are the attributes that the VFS layer can
214 * request to change from the FS layer.
216 * Derek Atkins <warlord@MIT.EDU> 94-10-20
219 unsigned int ia_valid;
224 struct timespec64 ia_atime;
225 struct timespec64 ia_mtime;
226 struct timespec64 ia_ctime;
229 * Not an attribute, but an auxiliary info for filesystems wanting to
230 * implement an ftruncate() like method. NOTE: filesystem should
231 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
233 struct file *ia_file;
237 * Includes for diskquotas.
239 #include <linux/quota.h>
242 * Maximum number of layers of fs stack. Needs to be limited to
243 * prevent kernel stack overflow
245 #define FILESYSTEM_MAX_STACK_DEPTH 2
248 * enum positive_aop_returns - aop return codes with specific semantics
250 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
251 * completed, that the page is still locked, and
252 * should be considered active. The VM uses this hint
253 * to return the page to the active list -- it won't
254 * be a candidate for writeback again in the near
255 * future. Other callers must be careful to unlock
256 * the page if they get this return. Returned by
259 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
260 * unlocked it and the page might have been truncated.
261 * The caller should back up to acquiring a new page and
262 * trying again. The aop will be taking reasonable
263 * precautions not to livelock. If the caller held a page
264 * reference, it should drop it before retrying. Returned
267 * address_space_operation functions return these large constants to indicate
268 * special semantics to the caller. These are much larger than the bytes in a
269 * page to allow for functions that return the number of bytes operated on in a
273 enum positive_aop_returns {
274 AOP_WRITEPAGE_ACTIVATE = 0x80000,
275 AOP_TRUNCATED_PAGE = 0x80001,
278 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
279 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
280 * helper code (eg buffer layer)
281 * to clear GFP_FS from alloc */
284 * oh the beauties of C type declarations.
287 struct address_space;
288 struct writeback_control;
289 struct readahead_control;
292 * Write life time hint values.
293 * Stored in struct inode as u8.
296 WRITE_LIFE_NOT_SET = 0,
297 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
298 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
299 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
300 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
301 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
304 /* Match RWF_* bits to IOCB bits */
305 #define IOCB_HIPRI (__force int) RWF_HIPRI
306 #define IOCB_DSYNC (__force int) RWF_DSYNC
307 #define IOCB_SYNC (__force int) RWF_SYNC
308 #define IOCB_NOWAIT (__force int) RWF_NOWAIT
309 #define IOCB_APPEND (__force int) RWF_APPEND
311 /* non-RWF related bits - start at 16 */
312 #define IOCB_EVENTFD (1 << 16)
313 #define IOCB_DIRECT (1 << 17)
314 #define IOCB_WRITE (1 << 18)
315 /* iocb->ki_waitq is valid */
316 #define IOCB_WAITQ (1 << 19)
317 #define IOCB_NOIO (1 << 20)
318 /* can use bio alloc cache */
319 #define IOCB_ALLOC_CACHE (1 << 21)
322 struct file *ki_filp;
324 /* The 'ki_filp' pointer is shared in a union for aio */
325 randomized_struct_fields_start
328 void (*ki_complete)(struct kiocb *iocb, long ret);
332 u16 ki_ioprio; /* See linux/ioprio.h */
333 struct wait_page_queue *ki_waitq; /* for async buffered IO */
334 randomized_struct_fields_end
337 static inline bool is_sync_kiocb(struct kiocb *kiocb)
339 return kiocb->ki_complete == NULL;
343 * "descriptor" for what we're up to with a read.
344 * This allows us to use the same read code yet
345 * have multiple different users of the data that
346 * we read from a file.
348 * The simplest case just copies the data to user
361 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
362 unsigned long, unsigned long);
364 struct address_space_operations {
365 int (*writepage)(struct page *page, struct writeback_control *wbc);
366 int (*readpage)(struct file *, struct page *);
368 /* Write back some dirty pages from this mapping. */
369 int (*writepages)(struct address_space *, struct writeback_control *);
371 /* Set a page dirty. Return true if this dirtied it */
372 int (*set_page_dirty)(struct page *page);
375 * Reads in the requested pages. Unlike ->readpage(), this is
376 * PURELY used for read-ahead!.
378 int (*readpages)(struct file *filp, struct address_space *mapping,
379 struct list_head *pages, unsigned nr_pages);
380 void (*readahead)(struct readahead_control *);
382 int (*write_begin)(struct file *, struct address_space *mapping,
383 loff_t pos, unsigned len, unsigned flags,
384 struct page **pagep, void **fsdata);
385 int (*write_end)(struct file *, struct address_space *mapping,
386 loff_t pos, unsigned len, unsigned copied,
387 struct page *page, void *fsdata);
389 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
390 sector_t (*bmap)(struct address_space *, sector_t);
391 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
392 int (*releasepage) (struct page *, gfp_t);
393 void (*freepage)(struct page *);
394 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
396 * migrate the contents of a page to the specified target. If
397 * migrate_mode is MIGRATE_ASYNC, it must not block.
399 int (*migratepage) (struct address_space *,
400 struct page *, struct page *, enum migrate_mode);
401 bool (*isolate_page)(struct page *, isolate_mode_t);
402 void (*putback_page)(struct page *);
403 int (*launder_page) (struct page *);
404 int (*is_partially_uptodate) (struct page *, unsigned long,
406 void (*is_dirty_writeback) (struct page *, bool *, bool *);
407 int (*error_remove_page)(struct address_space *, struct page *);
409 /* swapfile support */
410 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
412 void (*swap_deactivate)(struct file *file);
415 extern const struct address_space_operations empty_aops;
418 * pagecache_write_begin/pagecache_write_end must be used by general code
419 * to write into the pagecache.
421 int pagecache_write_begin(struct file *, struct address_space *mapping,
422 loff_t pos, unsigned len, unsigned flags,
423 struct page **pagep, void **fsdata);
425 int pagecache_write_end(struct file *, struct address_space *mapping,
426 loff_t pos, unsigned len, unsigned copied,
427 struct page *page, void *fsdata);
430 * struct address_space - Contents of a cacheable, mappable object.
431 * @host: Owner, either the inode or the block_device.
432 * @i_pages: Cached pages.
433 * @invalidate_lock: Guards coherency between page cache contents and
434 * file offset->disk block mappings in the filesystem during invalidates.
435 * It is also used to block modification of page cache contents through
437 * @gfp_mask: Memory allocation flags to use for allocating pages.
438 * @i_mmap_writable: Number of VM_SHARED mappings.
439 * @nr_thps: Number of THPs in the pagecache (non-shmem only).
440 * @i_mmap: Tree of private and shared mappings.
441 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
442 * @nrpages: Number of page entries, protected by the i_pages lock.
443 * @writeback_index: Writeback starts here.
445 * @flags: Error bits and flags (AS_*).
446 * @wb_err: The most recent error which has occurred.
447 * @private_lock: For use by the owner of the address_space.
448 * @private_list: For use by the owner of the address_space.
449 * @private_data: For use by the owner of the address_space.
451 struct address_space {
453 struct xarray i_pages;
454 struct rw_semaphore invalidate_lock;
456 atomic_t i_mmap_writable;
457 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
458 /* number of thp, only for non-shmem files */
461 struct rb_root_cached i_mmap;
462 struct rw_semaphore i_mmap_rwsem;
463 unsigned long nrpages;
464 pgoff_t writeback_index;
465 const struct address_space_operations *a_ops;
468 spinlock_t private_lock;
469 struct list_head private_list;
471 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
473 * On most architectures that alignment is already the case; but
474 * must be enforced here for CRIS, to let the least significant bit
475 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
478 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
479 #define PAGECACHE_TAG_DIRTY XA_MARK_0
480 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
481 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
484 * Returns true if any of the pages in the mapping are marked with the tag.
486 static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag)
488 return xa_marked(&mapping->i_pages, tag);
491 static inline void i_mmap_lock_write(struct address_space *mapping)
493 down_write(&mapping->i_mmap_rwsem);
496 static inline int i_mmap_trylock_write(struct address_space *mapping)
498 return down_write_trylock(&mapping->i_mmap_rwsem);
501 static inline void i_mmap_unlock_write(struct address_space *mapping)
503 up_write(&mapping->i_mmap_rwsem);
506 static inline void i_mmap_lock_read(struct address_space *mapping)
508 down_read(&mapping->i_mmap_rwsem);
511 static inline void i_mmap_unlock_read(struct address_space *mapping)
513 up_read(&mapping->i_mmap_rwsem);
516 static inline void i_mmap_assert_locked(struct address_space *mapping)
518 lockdep_assert_held(&mapping->i_mmap_rwsem);
521 static inline void i_mmap_assert_write_locked(struct address_space *mapping)
523 lockdep_assert_held_write(&mapping->i_mmap_rwsem);
527 * Might pages of this file be mapped into userspace?
529 static inline int mapping_mapped(struct address_space *mapping)
531 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
535 * Might pages of this file have been modified in userspace?
536 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap
537 * marks vma as VM_SHARED if it is shared, and the file was opened for
538 * writing i.e. vma may be mprotected writable even if now readonly.
540 * If i_mmap_writable is negative, no new writable mappings are allowed. You
541 * can only deny writable mappings, if none exists right now.
543 static inline int mapping_writably_mapped(struct address_space *mapping)
545 return atomic_read(&mapping->i_mmap_writable) > 0;
548 static inline int mapping_map_writable(struct address_space *mapping)
550 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
554 static inline void mapping_unmap_writable(struct address_space *mapping)
556 atomic_dec(&mapping->i_mmap_writable);
559 static inline int mapping_deny_writable(struct address_space *mapping)
561 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
565 static inline void mapping_allow_writable(struct address_space *mapping)
567 atomic_inc(&mapping->i_mmap_writable);
571 * Use sequence counter to get consistent i_size on 32-bit processors.
573 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
574 #include <linux/seqlock.h>
575 #define __NEED_I_SIZE_ORDERED
576 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
578 #define i_size_ordered_init(inode) do { } while (0)
582 #define ACL_NOT_CACHED ((void *)(-1))
584 * ACL_DONT_CACHE is for stacked filesystems, that rely on underlying fs to
585 * cache the ACL. This also means that ->get_acl() can be called in RCU mode
586 * with the LOOKUP_RCU flag.
588 #define ACL_DONT_CACHE ((void *)(-3))
590 static inline struct posix_acl *
591 uncached_acl_sentinel(struct task_struct *task)
593 return (void *)task + 1;
597 is_uncached_acl(struct posix_acl *acl)
599 return (long)acl & 1;
602 #define IOP_FASTPERM 0x0001
603 #define IOP_LOOKUP 0x0002
604 #define IOP_NOFOLLOW 0x0004
605 #define IOP_XATTR 0x0008
606 #define IOP_DEFAULT_READLINK 0x0010
608 struct fsnotify_mark_connector;
611 * Keep mostly read-only and often accessed (especially for
612 * the RCU path lookup and 'stat' data) fields at the beginning
613 * of the 'struct inode'
617 unsigned short i_opflags;
620 unsigned int i_flags;
622 #ifdef CONFIG_FS_POSIX_ACL
623 struct posix_acl *i_acl;
624 struct posix_acl *i_default_acl;
627 const struct inode_operations *i_op;
628 struct super_block *i_sb;
629 struct address_space *i_mapping;
631 #ifdef CONFIG_SECURITY
635 /* Stat data, not accessed from path walking */
638 * Filesystems may only read i_nlink directly. They shall use the
639 * following functions for modification:
641 * (set|clear|inc|drop)_nlink
642 * inode_(inc|dec)_link_count
645 const unsigned int i_nlink;
646 unsigned int __i_nlink;
650 struct timespec64 i_atime;
651 struct timespec64 i_mtime;
652 struct timespec64 i_ctime;
653 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
654 unsigned short i_bytes;
659 #ifdef __NEED_I_SIZE_ORDERED
660 seqcount_t i_size_seqcount;
664 unsigned long i_state;
665 struct rw_semaphore i_rwsem;
667 unsigned long dirtied_when; /* jiffies of first dirtying */
668 unsigned long dirtied_time_when;
670 struct hlist_node i_hash;
671 struct list_head i_io_list; /* backing dev IO list */
672 #ifdef CONFIG_CGROUP_WRITEBACK
673 struct bdi_writeback *i_wb; /* the associated cgroup wb */
675 /* foreign inode detection, see wbc_detach_inode() */
677 u16 i_wb_frn_avg_time;
678 u16 i_wb_frn_history;
680 struct list_head i_lru; /* inode LRU list */
681 struct list_head i_sb_list;
682 struct list_head i_wb_list; /* backing dev writeback list */
684 struct hlist_head i_dentry;
685 struct rcu_head i_rcu;
687 atomic64_t i_version;
688 atomic64_t i_sequence; /* see futex */
690 atomic_t i_dio_count;
691 atomic_t i_writecount;
692 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
693 atomic_t i_readcount; /* struct files open RO */
696 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
697 void (*free_inode)(struct inode *);
699 struct file_lock_context *i_flctx;
700 struct address_space i_data;
701 struct list_head i_devices;
703 struct pipe_inode_info *i_pipe;
711 #ifdef CONFIG_FSNOTIFY
712 __u32 i_fsnotify_mask; /* all events this inode cares about */
713 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
716 #ifdef CONFIG_FS_ENCRYPTION
717 struct fscrypt_info *i_crypt_info;
720 #ifdef CONFIG_FS_VERITY
721 struct fsverity_info *i_verity_info;
724 void *i_private; /* fs or device private pointer */
725 } __randomize_layout;
727 struct timespec64 timestamp_truncate(struct timespec64 t, struct inode *inode);
729 static inline unsigned int i_blocksize(const struct inode *node)
731 return (1 << node->i_blkbits);
734 static inline int inode_unhashed(struct inode *inode)
736 return hlist_unhashed(&inode->i_hash);
740 * __mark_inode_dirty expects inodes to be hashed. Since we don't
741 * want special inodes in the fileset inode space, we make them
742 * appear hashed, but do not put on any lists. hlist_del()
743 * will work fine and require no locking.
745 static inline void inode_fake_hash(struct inode *inode)
747 hlist_add_fake(&inode->i_hash);
751 * inode->i_mutex nesting subclasses for the lock validator:
753 * 0: the object of the current VFS operation
757 * 4: second non-directory
758 * 5: second parent (when locking independent directories in rename)
760 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
761 * non-directories at once.
763 * The locking order between these classes is
764 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
766 enum inode_i_mutex_lock_class
776 static inline void inode_lock(struct inode *inode)
778 down_write(&inode->i_rwsem);
781 static inline void inode_unlock(struct inode *inode)
783 up_write(&inode->i_rwsem);
786 static inline void inode_lock_shared(struct inode *inode)
788 down_read(&inode->i_rwsem);
791 static inline void inode_unlock_shared(struct inode *inode)
793 up_read(&inode->i_rwsem);
796 static inline int inode_trylock(struct inode *inode)
798 return down_write_trylock(&inode->i_rwsem);
801 static inline int inode_trylock_shared(struct inode *inode)
803 return down_read_trylock(&inode->i_rwsem);
806 static inline int inode_is_locked(struct inode *inode)
808 return rwsem_is_locked(&inode->i_rwsem);
811 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
813 down_write_nested(&inode->i_rwsem, subclass);
816 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
818 down_read_nested(&inode->i_rwsem, subclass);
821 static inline void filemap_invalidate_lock(struct address_space *mapping)
823 down_write(&mapping->invalidate_lock);
826 static inline void filemap_invalidate_unlock(struct address_space *mapping)
828 up_write(&mapping->invalidate_lock);
831 static inline void filemap_invalidate_lock_shared(struct address_space *mapping)
833 down_read(&mapping->invalidate_lock);
836 static inline int filemap_invalidate_trylock_shared(
837 struct address_space *mapping)
839 return down_read_trylock(&mapping->invalidate_lock);
842 static inline void filemap_invalidate_unlock_shared(
843 struct address_space *mapping)
845 up_read(&mapping->invalidate_lock);
848 void lock_two_nondirectories(struct inode *, struct inode*);
849 void unlock_two_nondirectories(struct inode *, struct inode*);
851 void filemap_invalidate_lock_two(struct address_space *mapping1,
852 struct address_space *mapping2);
853 void filemap_invalidate_unlock_two(struct address_space *mapping1,
854 struct address_space *mapping2);
858 * NOTE: in a 32bit arch with a preemptable kernel and
859 * an UP compile the i_size_read/write must be atomic
860 * with respect to the local cpu (unlike with preempt disabled),
861 * but they don't need to be atomic with respect to other cpus like in
862 * true SMP (so they need either to either locally disable irq around
863 * the read or for example on x86 they can be still implemented as a
864 * cmpxchg8b without the need of the lock prefix). For SMP compiles
865 * and 64bit archs it makes no difference if preempt is enabled or not.
867 static inline loff_t i_size_read(const struct inode *inode)
869 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
874 seq = read_seqcount_begin(&inode->i_size_seqcount);
875 i_size = inode->i_size;
876 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
878 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
882 i_size = inode->i_size;
886 return inode->i_size;
891 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
892 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
893 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
895 static inline void i_size_write(struct inode *inode, loff_t i_size)
897 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
899 write_seqcount_begin(&inode->i_size_seqcount);
900 inode->i_size = i_size;
901 write_seqcount_end(&inode->i_size_seqcount);
903 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
905 inode->i_size = i_size;
908 inode->i_size = i_size;
912 static inline unsigned iminor(const struct inode *inode)
914 return MINOR(inode->i_rdev);
917 static inline unsigned imajor(const struct inode *inode)
919 return MAJOR(inode->i_rdev);
923 rwlock_t lock; /* protects pid, uid, euid fields */
924 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
925 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
926 kuid_t uid, euid; /* uid/euid of process setting the owner */
927 int signum; /* posix.1b rt signal to be delivered on IO */
931 * struct file_ra_state - Track a file's readahead state.
932 * @start: Where the most recent readahead started.
933 * @size: Number of pages read in the most recent readahead.
934 * @async_size: Start next readahead when this many pages are left.
935 * @ra_pages: Maximum size of a readahead request.
936 * @mmap_miss: How many mmap accesses missed in the page cache.
937 * @prev_pos: The last byte in the most recent read request.
939 struct file_ra_state {
942 unsigned int async_size;
943 unsigned int ra_pages;
944 unsigned int mmap_miss;
949 * Check if @index falls in the readahead windows.
951 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
953 return (index >= ra->start &&
954 index < ra->start + ra->size);
959 struct llist_node fu_llist;
960 struct rcu_head fu_rcuhead;
963 struct inode *f_inode; /* cached value */
964 const struct file_operations *f_op;
967 * Protects f_ep, f_flags.
968 * Must not be taken from IRQ context.
971 enum rw_hint f_write_hint;
972 atomic_long_t f_count;
973 unsigned int f_flags;
975 struct mutex f_pos_lock;
977 struct fown_struct f_owner;
978 const struct cred *f_cred;
979 struct file_ra_state f_ra;
982 #ifdef CONFIG_SECURITY
985 /* needed for tty driver, and maybe others */
989 /* Used by fs/eventpoll.c to link all the hooks to this file */
990 struct hlist_head *f_ep;
991 #endif /* #ifdef CONFIG_EPOLL */
992 struct address_space *f_mapping;
994 errseq_t f_sb_err; /* for syncfs */
996 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
1001 /* file identifier */
1002 unsigned char f_handle[];
1005 static inline struct file *get_file(struct file *f)
1007 atomic_long_inc(&f->f_count);
1010 #define get_file_rcu_many(x, cnt) \
1011 atomic_long_add_unless(&(x)->f_count, (cnt), 0)
1012 #define get_file_rcu(x) get_file_rcu_many((x), 1)
1013 #define file_count(x) atomic_long_read(&(x)->f_count)
1015 #define MAX_NON_LFS ((1UL<<31) - 1)
1017 /* Page cache limit. The filesystems should put that into their s_maxbytes
1018 limits, otherwise bad things can happen in VM. */
1019 #if BITS_PER_LONG==32
1020 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
1021 #elif BITS_PER_LONG==64
1022 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
1027 #define FL_DELEG 4 /* NFSv4 delegation */
1028 #define FL_ACCESS 8 /* not trying to lock, just looking */
1029 #define FL_EXISTS 16 /* when unlocking, test for existence */
1030 #define FL_LEASE 32 /* lease held on this file */
1031 #define FL_CLOSE 64 /* unlock on close */
1032 #define FL_SLEEP 128 /* A blocking lock */
1033 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
1034 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
1035 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
1036 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
1037 #define FL_RECLAIM 4096 /* reclaiming from a reboot server */
1039 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
1042 * Special return value from posix_lock_file() and vfs_lock_file() for
1043 * asynchronous locking.
1045 #define FILE_LOCK_DEFERRED 1
1047 /* legacy typedef, should eventually be removed */
1048 typedef void *fl_owner_t;
1052 struct file_lock_operations {
1053 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1054 void (*fl_release_private)(struct file_lock *);
1057 struct lock_manager_operations {
1058 fl_owner_t (*lm_get_owner)(fl_owner_t);
1059 void (*lm_put_owner)(fl_owner_t);
1060 void (*lm_notify)(struct file_lock *); /* unblock callback */
1061 int (*lm_grant)(struct file_lock *, int);
1062 bool (*lm_break)(struct file_lock *);
1063 int (*lm_change)(struct file_lock *, int, struct list_head *);
1064 void (*lm_setup)(struct file_lock *, void **);
1065 bool (*lm_breaker_owns_lease)(struct file_lock *);
1068 struct lock_manager {
1069 struct list_head list;
1071 * NFSv4 and up also want opens blocked during the grace period;
1078 void locks_start_grace(struct net *, struct lock_manager *);
1079 void locks_end_grace(struct lock_manager *);
1080 bool locks_in_grace(struct net *);
1081 bool opens_in_grace(struct net *);
1083 /* that will die - we need it for nfs_lock_info */
1084 #include <linux/nfs_fs_i.h>
1087 * struct file_lock represents a generic "file lock". It's used to represent
1088 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1089 * note that the same struct is used to represent both a request for a lock and
1090 * the lock itself, but the same object is never used for both.
1092 * FIXME: should we create a separate "struct lock_request" to help distinguish
1095 * The varous i_flctx lists are ordered by:
1098 * 2) lock range start
1101 * Obviously, the last two criteria only matter for POSIX locks.
1104 struct file_lock *fl_blocker; /* The lock, that is blocking us */
1105 struct list_head fl_list; /* link into file_lock_context */
1106 struct hlist_node fl_link; /* node in global lists */
1107 struct list_head fl_blocked_requests; /* list of requests with
1108 * ->fl_blocker pointing here
1110 struct list_head fl_blocked_member; /* node in
1111 * ->fl_blocker->fl_blocked_requests
1113 fl_owner_t fl_owner;
1114 unsigned int fl_flags;
1115 unsigned char fl_type;
1116 unsigned int fl_pid;
1117 int fl_link_cpu; /* what cpu's list is this on? */
1118 wait_queue_head_t fl_wait;
1119 struct file *fl_file;
1123 struct fasync_struct * fl_fasync; /* for lease break notifications */
1124 /* for lease breaks: */
1125 unsigned long fl_break_time;
1126 unsigned long fl_downgrade_time;
1128 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1129 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1131 struct nfs_lock_info nfs_fl;
1132 struct nfs4_lock_info nfs4_fl;
1134 struct list_head link; /* link in AFS vnode's pending_locks list */
1135 int state; /* state of grant or error if -ve */
1136 unsigned int debug_id;
1139 } __randomize_layout;
1141 struct file_lock_context {
1142 spinlock_t flc_lock;
1143 struct list_head flc_flock;
1144 struct list_head flc_posix;
1145 struct list_head flc_lease;
1148 /* The following constant reflects the upper bound of the file/locking space */
1150 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1151 #define OFFSET_MAX INT_LIMIT(loff_t)
1152 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1155 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1157 #define locks_inode(f) file_inode(f)
1159 #ifdef CONFIG_FILE_LOCKING
1160 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1161 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1164 #if BITS_PER_LONG == 32
1165 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1166 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1170 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1171 extern int fcntl_getlease(struct file *filp);
1174 void locks_free_lock_context(struct inode *inode);
1175 void locks_free_lock(struct file_lock *fl);
1176 extern void locks_init_lock(struct file_lock *);
1177 extern struct file_lock * locks_alloc_lock(void);
1178 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1179 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1180 extern void locks_remove_posix(struct file *, fl_owner_t);
1181 extern void locks_remove_file(struct file *);
1182 extern void locks_release_private(struct file_lock *);
1183 extern void posix_test_lock(struct file *, struct file_lock *);
1184 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1185 extern int locks_delete_block(struct file_lock *);
1186 extern int vfs_test_lock(struct file *, struct file_lock *);
1187 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1188 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1189 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1190 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1191 extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1192 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1193 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1194 extern int lease_modify(struct file_lock *, int, struct list_head *);
1196 struct notifier_block;
1197 extern int lease_register_notifier(struct notifier_block *);
1198 extern void lease_unregister_notifier(struct notifier_block *);
1200 struct files_struct;
1201 extern void show_fd_locks(struct seq_file *f,
1202 struct file *filp, struct files_struct *files);
1203 #else /* !CONFIG_FILE_LOCKING */
1204 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1205 struct flock __user *user)
1210 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1211 unsigned int cmd, struct flock __user *user)
1216 #if BITS_PER_LONG == 32
1217 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1218 struct flock64 *user)
1223 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1224 unsigned int cmd, struct flock64 *user)
1229 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1234 static inline int fcntl_getlease(struct file *filp)
1240 locks_free_lock_context(struct inode *inode)
1244 static inline void locks_init_lock(struct file_lock *fl)
1249 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1254 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1259 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1264 static inline void locks_remove_file(struct file *filp)
1269 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1274 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1275 struct file_lock *conflock)
1280 static inline int locks_delete_block(struct file_lock *waiter)
1285 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1290 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1291 struct file_lock *fl, struct file_lock *conf)
1296 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1301 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1306 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1311 static inline void lease_get_mtime(struct inode *inode,
1312 struct timespec64 *time)
1317 static inline int generic_setlease(struct file *filp, long arg,
1318 struct file_lock **flp, void **priv)
1323 static inline int vfs_setlease(struct file *filp, long arg,
1324 struct file_lock **lease, void **priv)
1329 static inline int lease_modify(struct file_lock *fl, int arg,
1330 struct list_head *dispose)
1335 struct files_struct;
1336 static inline void show_fd_locks(struct seq_file *f,
1337 struct file *filp, struct files_struct *files) {}
1338 #endif /* !CONFIG_FILE_LOCKING */
1340 static inline struct inode *file_inode(const struct file *f)
1345 static inline struct dentry *file_dentry(const struct file *file)
1347 return d_real(file->f_path.dentry, file_inode(file));
1350 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1352 return locks_lock_inode_wait(locks_inode(filp), fl);
1355 struct fasync_struct {
1359 struct fasync_struct *fa_next; /* singly linked list */
1360 struct file *fa_file;
1361 struct rcu_head fa_rcu;
1364 #define FASYNC_MAGIC 0x4601
1366 /* SMP safe fasync helpers: */
1367 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1368 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1369 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1370 extern struct fasync_struct *fasync_alloc(void);
1371 extern void fasync_free(struct fasync_struct *);
1373 /* can be called from interrupts */
1374 extern void kill_fasync(struct fasync_struct **, int, int);
1376 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1377 extern int f_setown(struct file *filp, unsigned long arg, int force);
1378 extern void f_delown(struct file *filp);
1379 extern pid_t f_getown(struct file *filp);
1380 extern int send_sigurg(struct fown_struct *fown);
1383 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1384 * represented in both.
1386 #define SB_RDONLY 1 /* Mount read-only */
1387 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1388 #define SB_NODEV 4 /* Disallow access to device special files */
1389 #define SB_NOEXEC 8 /* Disallow program execution */
1390 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1391 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1392 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1393 #define SB_NOATIME 1024 /* Do not update access times. */
1394 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1395 #define SB_SILENT 32768
1396 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1397 #define SB_INLINECRYPT (1<<17) /* Use blk-crypto for encrypted files */
1398 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1399 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1400 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1402 /* These sb flags are internal to the kernel */
1403 #define SB_SUBMOUNT (1<<26)
1404 #define SB_FORCE (1<<27)
1405 #define SB_NOSEC (1<<28)
1406 #define SB_BORN (1<<29)
1407 #define SB_ACTIVE (1<<30)
1408 #define SB_NOUSER (1<<31)
1410 /* These flags relate to encoding and casefolding */
1411 #define SB_ENC_STRICT_MODE_FL (1 << 0)
1413 #define sb_has_strict_encoding(sb) \
1414 (sb->s_encoding_flags & SB_ENC_STRICT_MODE_FL)
1420 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1421 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1422 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1423 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1424 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1427 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1428 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1429 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1430 #define SB_I_STABLE_WRITES 0x00000008 /* don't modify blks until WB is done */
1432 /* sb->s_iflags to limit user namespace mounts */
1433 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1434 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1435 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1437 #define SB_I_SKIP_SYNC 0x00000100 /* Skip superblock at global sync */
1438 #define SB_I_PERSB_BDI 0x00000200 /* has a per-sb bdi */
1440 /* Possible states of 'frozen' field */
1442 SB_UNFROZEN = 0, /* FS is unfrozen */
1443 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1444 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1445 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1446 * internal threads if needed) */
1447 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1450 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1453 int frozen; /* Is sb frozen? */
1454 wait_queue_head_t wait_unfrozen; /* wait for thaw */
1455 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1458 struct super_block {
1459 struct list_head s_list; /* Keep this first */
1460 dev_t s_dev; /* search index; _not_ kdev_t */
1461 unsigned char s_blocksize_bits;
1462 unsigned long s_blocksize;
1463 loff_t s_maxbytes; /* Max file size */
1464 struct file_system_type *s_type;
1465 const struct super_operations *s_op;
1466 const struct dquot_operations *dq_op;
1467 const struct quotactl_ops *s_qcop;
1468 const struct export_operations *s_export_op;
1469 unsigned long s_flags;
1470 unsigned long s_iflags; /* internal SB_I_* flags */
1471 unsigned long s_magic;
1472 struct dentry *s_root;
1473 struct rw_semaphore s_umount;
1476 #ifdef CONFIG_SECURITY
1479 const struct xattr_handler **s_xattr;
1480 #ifdef CONFIG_FS_ENCRYPTION
1481 const struct fscrypt_operations *s_cop;
1482 struct key *s_master_keys; /* master crypto keys in use */
1484 #ifdef CONFIG_FS_VERITY
1485 const struct fsverity_operations *s_vop;
1487 #ifdef CONFIG_UNICODE
1488 struct unicode_map *s_encoding;
1489 __u16 s_encoding_flags;
1491 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1492 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1493 struct block_device *s_bdev;
1494 struct backing_dev_info *s_bdi;
1495 struct mtd_info *s_mtd;
1496 struct hlist_node s_instances;
1497 unsigned int s_quota_types; /* Bitmask of supported quota types */
1498 struct quota_info s_dquot; /* Diskquota specific options */
1500 struct sb_writers s_writers;
1503 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1504 * s_fsnotify_marks together for cache efficiency. They are frequently
1505 * accessed and rarely modified.
1507 void *s_fs_info; /* Filesystem private info */
1509 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1511 /* Time limits for c/m/atime in seconds */
1512 time64_t s_time_min;
1513 time64_t s_time_max;
1514 #ifdef CONFIG_FSNOTIFY
1515 __u32 s_fsnotify_mask;
1516 struct fsnotify_mark_connector __rcu *s_fsnotify_marks;
1519 char s_id[32]; /* Informational name */
1520 uuid_t s_uuid; /* UUID */
1522 unsigned int s_max_links;
1526 * The next field is for VFS *only*. No filesystems have any business
1527 * even looking at it. You had been warned.
1529 struct mutex s_vfs_rename_mutex; /* Kludge */
1532 * Filesystem subtype. If non-empty the filesystem type field
1533 * in /proc/mounts will be "type.subtype"
1535 const char *s_subtype;
1537 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1540 * Saved pool identifier for cleancache (-1 means none)
1542 int cleancache_poolid;
1544 struct shrinker s_shrink; /* per-sb shrinker handle */
1546 /* Number of inodes with nlink == 0 but still referenced */
1547 atomic_long_t s_remove_count;
1550 * Number of inode/mount/sb objects that are being watched, note that
1551 * inodes objects are currently double-accounted.
1553 atomic_long_t s_fsnotify_connectors;
1555 /* Being remounted read-only */
1556 int s_readonly_remount;
1558 /* per-sb errseq_t for reporting writeback errors via syncfs */
1561 /* AIO completions deferred from interrupt context */
1562 struct workqueue_struct *s_dio_done_wq;
1563 struct hlist_head s_pins;
1566 * Owning user namespace and default context in which to
1567 * interpret filesystem uids, gids, quotas, device nodes,
1568 * xattrs and security labels.
1570 struct user_namespace *s_user_ns;
1573 * The list_lru structure is essentially just a pointer to a table
1574 * of per-node lru lists, each of which has its own spinlock.
1575 * There is no need to put them into separate cachelines.
1577 struct list_lru s_dentry_lru;
1578 struct list_lru s_inode_lru;
1579 struct rcu_head rcu;
1580 struct work_struct destroy_work;
1582 struct mutex s_sync_lock; /* sync serialisation lock */
1585 * Indicates how deep in a filesystem stack this SB is
1589 /* s_inode_list_lock protects s_inodes */
1590 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1591 struct list_head s_inodes; /* all inodes */
1593 spinlock_t s_inode_wblist_lock;
1594 struct list_head s_inodes_wb; /* writeback inodes */
1595 } __randomize_layout;
1597 static inline struct user_namespace *i_user_ns(const struct inode *inode)
1599 return inode->i_sb->s_user_ns;
1602 /* Helper functions so that in most cases filesystems will
1603 * not need to deal directly with kuid_t and kgid_t and can
1604 * instead deal with the raw numeric values that are stored
1605 * in the filesystem.
1607 static inline uid_t i_uid_read(const struct inode *inode)
1609 return from_kuid(i_user_ns(inode), inode->i_uid);
1612 static inline gid_t i_gid_read(const struct inode *inode)
1614 return from_kgid(i_user_ns(inode), inode->i_gid);
1617 static inline void i_uid_write(struct inode *inode, uid_t uid)
1619 inode->i_uid = make_kuid(i_user_ns(inode), uid);
1622 static inline void i_gid_write(struct inode *inode, gid_t gid)
1624 inode->i_gid = make_kgid(i_user_ns(inode), gid);
1628 * i_uid_into_mnt - map an inode's i_uid down into a mnt_userns
1629 * @mnt_userns: user namespace of the mount the inode was found from
1630 * @inode: inode to map
1632 * Return: the inode's i_uid mapped down according to @mnt_userns.
1633 * If the inode's i_uid has no mapping INVALID_UID is returned.
1635 static inline kuid_t i_uid_into_mnt(struct user_namespace *mnt_userns,
1636 const struct inode *inode)
1638 return mapped_kuid_fs(mnt_userns, i_user_ns(inode), inode->i_uid);
1642 * i_gid_into_mnt - map an inode's i_gid down into a mnt_userns
1643 * @mnt_userns: user namespace of the mount the inode was found from
1644 * @inode: inode to map
1646 * Return: the inode's i_gid mapped down according to @mnt_userns.
1647 * If the inode's i_gid has no mapping INVALID_GID is returned.
1649 static inline kgid_t i_gid_into_mnt(struct user_namespace *mnt_userns,
1650 const struct inode *inode)
1652 return mapped_kgid_fs(mnt_userns, i_user_ns(inode), inode->i_gid);
1656 * inode_fsuid_set - initialize inode's i_uid field with callers fsuid
1657 * @inode: inode to initialize
1658 * @mnt_userns: user namespace of the mount the inode was found from
1660 * Initialize the i_uid field of @inode. If the inode was found/created via
1661 * an idmapped mount map the caller's fsuid according to @mnt_users.
1663 static inline void inode_fsuid_set(struct inode *inode,
1664 struct user_namespace *mnt_userns)
1666 inode->i_uid = mapped_fsuid(mnt_userns, i_user_ns(inode));
1670 * inode_fsgid_set - initialize inode's i_gid field with callers fsgid
1671 * @inode: inode to initialize
1672 * @mnt_userns: user namespace of the mount the inode was found from
1674 * Initialize the i_gid field of @inode. If the inode was found/created via
1675 * an idmapped mount map the caller's fsgid according to @mnt_users.
1677 static inline void inode_fsgid_set(struct inode *inode,
1678 struct user_namespace *mnt_userns)
1680 inode->i_gid = mapped_fsgid(mnt_userns, i_user_ns(inode));
1684 * fsuidgid_has_mapping() - check whether caller's fsuid/fsgid is mapped
1685 * @sb: the superblock we want a mapping in
1686 * @mnt_userns: user namespace of the relevant mount
1688 * Check whether the caller's fsuid and fsgid have a valid mapping in the
1689 * s_user_ns of the superblock @sb. If the caller is on an idmapped mount map
1690 * the caller's fsuid and fsgid according to the @mnt_userns first.
1692 * Return: true if fsuid and fsgid is mapped, false if not.
1694 static inline bool fsuidgid_has_mapping(struct super_block *sb,
1695 struct user_namespace *mnt_userns)
1697 struct user_namespace *fs_userns = sb->s_user_ns;
1701 kuid = mapped_fsuid(mnt_userns, fs_userns);
1702 if (!uid_valid(kuid))
1704 kgid = mapped_fsgid(mnt_userns, fs_userns);
1705 if (!gid_valid(kgid))
1707 return kuid_has_mapping(fs_userns, kuid) &&
1708 kgid_has_mapping(fs_userns, kgid);
1711 extern struct timespec64 current_time(struct inode *inode);
1714 * Snapshotting support.
1718 * These are internal functions, please use sb_start_{write,pagefault,intwrite}
1721 static inline void __sb_end_write(struct super_block *sb, int level)
1723 percpu_up_read(sb->s_writers.rw_sem + level-1);
1726 static inline void __sb_start_write(struct super_block *sb, int level)
1728 percpu_down_read(sb->s_writers.rw_sem + level - 1);
1731 static inline bool __sb_start_write_trylock(struct super_block *sb, int level)
1733 return percpu_down_read_trylock(sb->s_writers.rw_sem + level - 1);
1736 #define __sb_writers_acquired(sb, lev) \
1737 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1738 #define __sb_writers_release(sb, lev) \
1739 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1742 * sb_end_write - drop write access to a superblock
1743 * @sb: the super we wrote to
1745 * Decrement number of writers to the filesystem. Wake up possible waiters
1746 * wanting to freeze the filesystem.
1748 static inline void sb_end_write(struct super_block *sb)
1750 __sb_end_write(sb, SB_FREEZE_WRITE);
1754 * sb_end_pagefault - drop write access to a superblock from a page fault
1755 * @sb: the super we wrote to
1757 * Decrement number of processes handling write page fault to the filesystem.
1758 * Wake up possible waiters wanting to freeze the filesystem.
1760 static inline void sb_end_pagefault(struct super_block *sb)
1762 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1766 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1767 * @sb: the super we wrote to
1769 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1770 * waiters wanting to freeze the filesystem.
1772 static inline void sb_end_intwrite(struct super_block *sb)
1774 __sb_end_write(sb, SB_FREEZE_FS);
1778 * sb_start_write - get write access to a superblock
1779 * @sb: the super we write to
1781 * When a process wants to write data or metadata to a file system (i.e. dirty
1782 * a page or an inode), it should embed the operation in a sb_start_write() -
1783 * sb_end_write() pair to get exclusion against file system freezing. This
1784 * function increments number of writers preventing freezing. If the file
1785 * system is already frozen, the function waits until the file system is
1788 * Since freeze protection behaves as a lock, users have to preserve
1789 * ordering of freeze protection and other filesystem locks. Generally,
1790 * freeze protection should be the outermost lock. In particular, we have:
1793 * -> i_mutex (write path, truncate, directory ops, ...)
1794 * -> s_umount (freeze_super, thaw_super)
1796 static inline void sb_start_write(struct super_block *sb)
1798 __sb_start_write(sb, SB_FREEZE_WRITE);
1801 static inline bool sb_start_write_trylock(struct super_block *sb)
1803 return __sb_start_write_trylock(sb, SB_FREEZE_WRITE);
1807 * sb_start_pagefault - get write access to a superblock from a page fault
1808 * @sb: the super we write to
1810 * When a process starts handling write page fault, it should embed the
1811 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1812 * exclusion against file system freezing. This is needed since the page fault
1813 * is going to dirty a page. This function increments number of running page
1814 * faults preventing freezing. If the file system is already frozen, the
1815 * function waits until the file system is thawed.
1817 * Since page fault freeze protection behaves as a lock, users have to preserve
1818 * ordering of freeze protection and other filesystem locks. It is advised to
1819 * put sb_start_pagefault() close to mmap_lock in lock ordering. Page fault
1820 * handling code implies lock dependency:
1823 * -> sb_start_pagefault
1825 static inline void sb_start_pagefault(struct super_block *sb)
1827 __sb_start_write(sb, SB_FREEZE_PAGEFAULT);
1831 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1832 * @sb: the super we write to
1834 * This is the third level of protection against filesystem freezing. It is
1835 * free for use by a filesystem. The only requirement is that it must rank
1836 * below sb_start_pagefault.
1838 * For example filesystem can call sb_start_intwrite() when starting a
1839 * transaction which somewhat eases handling of freezing for internal sources
1840 * of filesystem changes (internal fs threads, discarding preallocation on file
1843 static inline void sb_start_intwrite(struct super_block *sb)
1845 __sb_start_write(sb, SB_FREEZE_FS);
1848 static inline bool sb_start_intwrite_trylock(struct super_block *sb)
1850 return __sb_start_write_trylock(sb, SB_FREEZE_FS);
1853 bool inode_owner_or_capable(struct user_namespace *mnt_userns,
1854 const struct inode *inode);
1857 * VFS helper functions..
1859 int vfs_create(struct user_namespace *, struct inode *,
1860 struct dentry *, umode_t, bool);
1861 int vfs_mkdir(struct user_namespace *, struct inode *,
1862 struct dentry *, umode_t);
1863 int vfs_mknod(struct user_namespace *, struct inode *, struct dentry *,
1865 int vfs_symlink(struct user_namespace *, struct inode *,
1866 struct dentry *, const char *);
1867 int vfs_link(struct dentry *, struct user_namespace *, struct inode *,
1868 struct dentry *, struct inode **);
1869 int vfs_rmdir(struct user_namespace *, struct inode *, struct dentry *);
1870 int vfs_unlink(struct user_namespace *, struct inode *, struct dentry *,
1874 * struct renamedata - contains all information required for renaming
1875 * @old_mnt_userns: old user namespace of the mount the inode was found from
1876 * @old_dir: parent of source
1877 * @old_dentry: source
1878 * @new_mnt_userns: new user namespace of the mount the inode was found from
1879 * @new_dir: parent of destination
1880 * @new_dentry: destination
1881 * @delegated_inode: returns an inode needing a delegation break
1882 * @flags: rename flags
1885 struct user_namespace *old_mnt_userns;
1886 struct inode *old_dir;
1887 struct dentry *old_dentry;
1888 struct user_namespace *new_mnt_userns;
1889 struct inode *new_dir;
1890 struct dentry *new_dentry;
1891 struct inode **delegated_inode;
1893 } __randomize_layout;
1895 int vfs_rename(struct renamedata *);
1897 static inline int vfs_whiteout(struct user_namespace *mnt_userns,
1898 struct inode *dir, struct dentry *dentry)
1900 return vfs_mknod(mnt_userns, dir, dentry, S_IFCHR | WHITEOUT_MODE,
1904 struct dentry *vfs_tmpfile(struct user_namespace *mnt_userns,
1905 struct dentry *dentry, umode_t mode, int open_flag);
1907 int vfs_mkobj(struct dentry *, umode_t,
1908 int (*f)(struct dentry *, umode_t, void *),
1911 int vfs_fchown(struct file *file, uid_t user, gid_t group);
1912 int vfs_fchmod(struct file *file, umode_t mode);
1913 int vfs_utimes(const struct path *path, struct timespec64 *times);
1915 extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1917 #ifdef CONFIG_COMPAT
1918 extern long compat_ptr_ioctl(struct file *file, unsigned int cmd,
1921 #define compat_ptr_ioctl NULL
1925 * VFS file helper functions.
1927 void inode_init_owner(struct user_namespace *mnt_userns, struct inode *inode,
1928 const struct inode *dir, umode_t mode);
1929 extern bool may_open_dev(const struct path *path);
1932 * This is the "filldir" function type, used by readdir() to let
1933 * the kernel specify what kind of dirent layout it wants to have.
1934 * This allows the kernel to read directories into kernel space or
1935 * to have different dirent layouts depending on the binary type.
1938 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1941 struct dir_context {
1947 * These flags let !MMU mmap() govern direct device mapping vs immediate
1948 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1950 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1951 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1952 * NOMMU_MAP_READ: Can be mapped for reading
1953 * NOMMU_MAP_WRITE: Can be mapped for writing
1954 * NOMMU_MAP_EXEC: Can be mapped for execution
1956 #define NOMMU_MAP_COPY 0x00000001
1957 #define NOMMU_MAP_DIRECT 0x00000008
1958 #define NOMMU_MAP_READ VM_MAYREAD
1959 #define NOMMU_MAP_WRITE VM_MAYWRITE
1960 #define NOMMU_MAP_EXEC VM_MAYEXEC
1962 #define NOMMU_VMFLAGS \
1963 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1966 * These flags control the behavior of the remap_file_range function pointer.
1967 * If it is called with len == 0 that means "remap to end of source file".
1968 * See Documentation/filesystems/vfs.rst for more details about this call.
1970 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
1971 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
1973 #define REMAP_FILE_DEDUP (1 << 0)
1974 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
1977 * These flags signal that the caller is ok with altering various aspects of
1978 * the behavior of the remap operation. The changes must be made by the
1979 * implementation; the vfs remap helper functions can take advantage of them.
1980 * Flags in this category exist to preserve the quirky behavior of the hoisted
1981 * btrfs clone/dedupe ioctls.
1983 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
1987 struct file_operations {
1988 struct module *owner;
1989 loff_t (*llseek) (struct file *, loff_t, int);
1990 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1991 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1992 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1993 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1994 int (*iopoll)(struct kiocb *kiocb, struct io_comp_batch *,
1995 unsigned int flags);
1996 int (*iterate) (struct file *, struct dir_context *);
1997 int (*iterate_shared) (struct file *, struct dir_context *);
1998 __poll_t (*poll) (struct file *, struct poll_table_struct *);
1999 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
2000 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
2001 int (*mmap) (struct file *, struct vm_area_struct *);
2002 unsigned long mmap_supported_flags;
2003 int (*open) (struct inode *, struct file *);
2004 int (*flush) (struct file *, fl_owner_t id);
2005 int (*release) (struct inode *, struct file *);
2006 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
2007 int (*fasync) (int, struct file *, int);
2008 int (*lock) (struct file *, int, struct file_lock *);
2009 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
2010 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
2011 int (*check_flags)(int);
2012 int (*flock) (struct file *, int, struct file_lock *);
2013 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
2014 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
2015 int (*setlease)(struct file *, long, struct file_lock **, void **);
2016 long (*fallocate)(struct file *file, int mode, loff_t offset,
2018 void (*show_fdinfo)(struct seq_file *m, struct file *f);
2020 unsigned (*mmap_capabilities)(struct file *);
2022 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
2023 loff_t, size_t, unsigned int);
2024 loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
2025 struct file *file_out, loff_t pos_out,
2026 loff_t len, unsigned int remap_flags);
2027 int (*fadvise)(struct file *, loff_t, loff_t, int);
2028 } __randomize_layout;
2030 struct inode_operations {
2031 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
2032 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
2033 int (*permission) (struct user_namespace *, struct inode *, int);
2034 struct posix_acl * (*get_acl)(struct inode *, int, bool);
2036 int (*readlink) (struct dentry *, char __user *,int);
2038 int (*create) (struct user_namespace *, struct inode *,struct dentry *,
2040 int (*link) (struct dentry *,struct inode *,struct dentry *);
2041 int (*unlink) (struct inode *,struct dentry *);
2042 int (*symlink) (struct user_namespace *, struct inode *,struct dentry *,
2044 int (*mkdir) (struct user_namespace *, struct inode *,struct dentry *,
2046 int (*rmdir) (struct inode *,struct dentry *);
2047 int (*mknod) (struct user_namespace *, struct inode *,struct dentry *,
2049 int (*rename) (struct user_namespace *, struct inode *, struct dentry *,
2050 struct inode *, struct dentry *, unsigned int);
2051 int (*setattr) (struct user_namespace *, struct dentry *,
2053 int (*getattr) (struct user_namespace *, const struct path *,
2054 struct kstat *, u32, unsigned int);
2055 ssize_t (*listxattr) (struct dentry *, char *, size_t);
2056 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
2058 int (*update_time)(struct inode *, struct timespec64 *, int);
2059 int (*atomic_open)(struct inode *, struct dentry *,
2060 struct file *, unsigned open_flag,
2061 umode_t create_mode);
2062 int (*tmpfile) (struct user_namespace *, struct inode *,
2063 struct dentry *, umode_t);
2064 int (*set_acl)(struct user_namespace *, struct inode *,
2065 struct posix_acl *, int);
2066 int (*fileattr_set)(struct user_namespace *mnt_userns,
2067 struct dentry *dentry, struct fileattr *fa);
2068 int (*fileattr_get)(struct dentry *dentry, struct fileattr *fa);
2069 } ____cacheline_aligned;
2071 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
2072 struct iov_iter *iter)
2074 return file->f_op->read_iter(kio, iter);
2077 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
2078 struct iov_iter *iter)
2080 return file->f_op->write_iter(kio, iter);
2083 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
2085 return file->f_op->mmap(file, vma);
2088 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
2089 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
2090 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
2091 loff_t, size_t, unsigned int);
2092 extern ssize_t generic_copy_file_range(struct file *file_in, loff_t pos_in,
2093 struct file *file_out, loff_t pos_out,
2094 size_t len, unsigned int flags);
2095 extern int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
2096 struct file *file_out, loff_t pos_out,
2098 unsigned int remap_flags);
2099 extern loff_t do_clone_file_range(struct file *file_in, loff_t pos_in,
2100 struct file *file_out, loff_t pos_out,
2101 loff_t len, unsigned int remap_flags);
2102 extern loff_t vfs_clone_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 extern int vfs_dedupe_file_range(struct file *file,
2106 struct file_dedupe_range *same);
2107 extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
2108 struct file *dst_file, loff_t dst_pos,
2109 loff_t len, unsigned int remap_flags);
2112 struct super_operations {
2113 struct inode *(*alloc_inode)(struct super_block *sb);
2114 void (*destroy_inode)(struct inode *);
2115 void (*free_inode)(struct inode *);
2117 void (*dirty_inode) (struct inode *, int flags);
2118 int (*write_inode) (struct inode *, struct writeback_control *wbc);
2119 int (*drop_inode) (struct inode *);
2120 void (*evict_inode) (struct inode *);
2121 void (*put_super) (struct super_block *);
2122 int (*sync_fs)(struct super_block *sb, int wait);
2123 int (*freeze_super) (struct super_block *);
2124 int (*freeze_fs) (struct super_block *);
2125 int (*thaw_super) (struct super_block *);
2126 int (*unfreeze_fs) (struct super_block *);
2127 int (*statfs) (struct dentry *, struct kstatfs *);
2128 int (*remount_fs) (struct super_block *, int *, char *);
2129 void (*umount_begin) (struct super_block *);
2131 int (*show_options)(struct seq_file *, struct dentry *);
2132 int (*show_devname)(struct seq_file *, struct dentry *);
2133 int (*show_path)(struct seq_file *, struct dentry *);
2134 int (*show_stats)(struct seq_file *, struct dentry *);
2136 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
2137 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
2138 struct dquot **(*get_dquots)(struct inode *);
2140 long (*nr_cached_objects)(struct super_block *,
2141 struct shrink_control *);
2142 long (*free_cached_objects)(struct super_block *,
2143 struct shrink_control *);
2147 * Inode flags - they have no relation to superblock flags now
2149 #define S_SYNC (1 << 0) /* Writes are synced at once */
2150 #define S_NOATIME (1 << 1) /* Do not update access times */
2151 #define S_APPEND (1 << 2) /* Append-only file */
2152 #define S_IMMUTABLE (1 << 3) /* Immutable file */
2153 #define S_DEAD (1 << 4) /* removed, but still open directory */
2154 #define S_NOQUOTA (1 << 5) /* Inode is not counted to quota */
2155 #define S_DIRSYNC (1 << 6) /* Directory modifications are synchronous */
2156 #define S_NOCMTIME (1 << 7) /* Do not update file c/mtime */
2157 #define S_SWAPFILE (1 << 8) /* Do not truncate: swapon got its bmaps */
2158 #define S_PRIVATE (1 << 9) /* Inode is fs-internal */
2159 #define S_IMA (1 << 10) /* Inode has an associated IMA struct */
2160 #define S_AUTOMOUNT (1 << 11) /* Automount/referral quasi-directory */
2161 #define S_NOSEC (1 << 12) /* no suid or xattr security attributes */
2162 #ifdef CONFIG_FS_DAX
2163 #define S_DAX (1 << 13) /* Direct Access, avoiding the page cache */
2165 #define S_DAX 0 /* Make all the DAX code disappear */
2167 #define S_ENCRYPTED (1 << 14) /* Encrypted file (using fs/crypto/) */
2168 #define S_CASEFOLD (1 << 15) /* Casefolded file */
2169 #define S_VERITY (1 << 16) /* Verity file (using fs/verity/) */
2170 #define S_KERNEL_FILE (1 << 17) /* File is in use by the kernel (eg. fs/cachefiles) */
2173 * Note that nosuid etc flags are inode-specific: setting some file-system
2174 * flags just means all the inodes inherit those flags by default. It might be
2175 * possible to override it selectively if you really wanted to with some
2176 * ioctl() that is not currently implemented.
2178 * Exception: SB_RDONLY is always applied to the entire file system.
2180 * Unfortunately, it is possible to change a filesystems flags with it mounted
2181 * with files in use. This means that all of the inodes will not have their
2182 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
2183 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
2185 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
2187 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
2188 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
2189 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
2190 ((inode)->i_flags & S_SYNC))
2191 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
2192 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
2193 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
2194 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
2195 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
2197 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
2198 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
2199 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
2200 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
2202 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
2203 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
2204 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
2205 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
2206 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
2207 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2208 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2209 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2210 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2211 #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD)
2212 #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY)
2214 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2215 (inode)->i_rdev == WHITEOUT_DEV)
2217 static inline bool HAS_UNMAPPED_ID(struct user_namespace *mnt_userns,
2218 struct inode *inode)
2220 return !uid_valid(i_uid_into_mnt(mnt_userns, inode)) ||
2221 !gid_valid(i_gid_into_mnt(mnt_userns, inode));
2224 static inline enum rw_hint file_write_hint(struct file *file)
2226 if (file->f_write_hint != WRITE_LIFE_NOT_SET)
2227 return file->f_write_hint;
2229 return file_inode(file)->i_write_hint;
2232 static inline int iocb_flags(struct file *file);
2234 static inline u16 ki_hint_validate(enum rw_hint hint)
2236 typeof(((struct kiocb *)0)->ki_hint) max_hint = -1;
2238 if (hint <= max_hint)
2243 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
2245 *kiocb = (struct kiocb) {
2247 .ki_flags = iocb_flags(filp),
2248 .ki_hint = ki_hint_validate(file_write_hint(filp)),
2249 .ki_ioprio = get_current_ioprio(),
2253 static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src,
2256 *kiocb = (struct kiocb) {
2258 .ki_flags = kiocb_src->ki_flags,
2259 .ki_hint = kiocb_src->ki_hint,
2260 .ki_ioprio = kiocb_src->ki_ioprio,
2261 .ki_pos = kiocb_src->ki_pos,
2266 * Inode state bits. Protected by inode->i_lock
2268 * Four bits determine the dirty state of the inode: I_DIRTY_SYNC,
2269 * I_DIRTY_DATASYNC, I_DIRTY_PAGES, and I_DIRTY_TIME.
2271 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2272 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2273 * various stages of removing an inode.
2275 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2277 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2278 * fdatasync() (unless I_DIRTY_DATASYNC is also set).
2279 * Timestamp updates are the usual cause.
2280 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2281 * these changes separately from I_DIRTY_SYNC so that we
2282 * don't have to write inode on fdatasync() when only
2283 * e.g. the timestamps have changed.
2284 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2285 * I_DIRTY_TIME The inode itself only has dirty timestamps, and the
2286 * lazytime mount option is enabled. We keep track of this
2287 * separately from I_DIRTY_SYNC in order to implement
2288 * lazytime. This gets cleared if I_DIRTY_INODE
2289 * (I_DIRTY_SYNC and/or I_DIRTY_DATASYNC) gets set. I.e.
2290 * either I_DIRTY_TIME *or* I_DIRTY_INODE can be set in
2291 * i_state, but not both. I_DIRTY_PAGES may still be set.
2292 * I_NEW Serves as both a mutex and completion notification.
2293 * New inodes set I_NEW. If two processes both create
2294 * the same inode, one of them will release its inode and
2295 * wait for I_NEW to be released before returning.
2296 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2297 * also cause waiting on I_NEW, without I_NEW actually
2298 * being set. find_inode() uses this to prevent returning
2299 * nearly-dead inodes.
2300 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2301 * is zero. I_FREEING must be set when I_WILL_FREE is
2303 * I_FREEING Set when inode is about to be freed but still has dirty
2304 * pages or buffers attached or the inode itself is still
2306 * I_CLEAR Added by clear_inode(). In this state the inode is
2307 * clean and can be destroyed. Inode keeps I_FREEING.
2309 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2310 * prohibited for many purposes. iget() must wait for
2311 * the inode to be completely released, then create it
2312 * anew. Other functions will just ignore such inodes,
2313 * if appropriate. I_NEW is used for waiting.
2315 * I_SYNC Writeback of inode is running. The bit is set during
2316 * data writeback, and cleared with a wakeup on the bit
2317 * address once it is done. The bit is also used to pin
2318 * the inode in memory for flusher thread.
2320 * I_REFERENCED Marks the inode as recently references on the LRU list.
2322 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2324 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2325 * synchronize competing switching instances and to tell
2326 * wb stat updates to grab the i_pages lock. See
2327 * inode_switch_wbs_work_fn() for details.
2329 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2330 * and work dirs among overlayfs mounts.
2332 * I_CREATING New object's inode in the middle of setting up.
2334 * I_DONTCACHE Evict inode as soon as it is not used anymore.
2336 * I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists.
2337 * Used to detect that mark_inode_dirty() should not move
2338 * inode between dirty lists.
2340 * I_PINNING_FSCACHE_WB Inode is pinning an fscache object for writeback.
2342 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2344 #define I_DIRTY_SYNC (1 << 0)
2345 #define I_DIRTY_DATASYNC (1 << 1)
2346 #define I_DIRTY_PAGES (1 << 2)
2348 #define I_NEW (1 << __I_NEW)
2349 #define I_WILL_FREE (1 << 4)
2350 #define I_FREEING (1 << 5)
2351 #define I_CLEAR (1 << 6)
2353 #define I_SYNC (1 << __I_SYNC)
2354 #define I_REFERENCED (1 << 8)
2355 #define __I_DIO_WAKEUP 9
2356 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2357 #define I_LINKABLE (1 << 10)
2358 #define I_DIRTY_TIME (1 << 11)
2359 #define I_WB_SWITCH (1 << 13)
2360 #define I_OVL_INUSE (1 << 14)
2361 #define I_CREATING (1 << 15)
2362 #define I_DONTCACHE (1 << 16)
2363 #define I_SYNC_QUEUED (1 << 17)
2364 #define I_PINNING_FSCACHE_WB (1 << 18)
2366 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2367 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2368 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2370 extern void __mark_inode_dirty(struct inode *, int);
2371 static inline void mark_inode_dirty(struct inode *inode)
2373 __mark_inode_dirty(inode, I_DIRTY);
2376 static inline void mark_inode_dirty_sync(struct inode *inode)
2378 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2382 * Returns true if the given inode itself only has dirty timestamps (its pages
2383 * may still be dirty) and isn't currently being allocated or freed.
2384 * Filesystems should call this if when writing an inode when lazytime is
2385 * enabled, they want to opportunistically write the timestamps of other inodes
2386 * located very nearby on-disk, e.g. in the same inode block. This returns true
2387 * if the given inode is in need of such an opportunistic update. Requires
2388 * i_lock, or at least later re-checking under i_lock.
2390 static inline bool inode_is_dirtytime_only(struct inode *inode)
2392 return (inode->i_state & (I_DIRTY_TIME | I_NEW |
2393 I_FREEING | I_WILL_FREE)) == I_DIRTY_TIME;
2396 extern void inc_nlink(struct inode *inode);
2397 extern void drop_nlink(struct inode *inode);
2398 extern void clear_nlink(struct inode *inode);
2399 extern void set_nlink(struct inode *inode, unsigned int nlink);
2401 static inline void inode_inc_link_count(struct inode *inode)
2404 mark_inode_dirty(inode);
2407 static inline void inode_dec_link_count(struct inode *inode)
2410 mark_inode_dirty(inode);
2413 enum file_time_flags {
2420 extern bool atime_needs_update(const struct path *, struct inode *);
2421 extern void touch_atime(const struct path *);
2422 int inode_update_time(struct inode *inode, struct timespec64 *time, int flags);
2424 static inline void file_accessed(struct file *file)
2426 if (!(file->f_flags & O_NOATIME))
2427 touch_atime(&file->f_path);
2430 extern int file_modified(struct file *file);
2432 int sync_inode_metadata(struct inode *inode, int wait);
2434 struct file_system_type {
2437 #define FS_REQUIRES_DEV 1
2438 #define FS_BINARY_MOUNTDATA 2
2439 #define FS_HAS_SUBTYPE 4
2440 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2441 #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */
2442 #define FS_ALLOW_IDMAP 32 /* FS has been updated to handle vfs idmappings. */
2443 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2444 int (*init_fs_context)(struct fs_context *);
2445 const struct fs_parameter_spec *parameters;
2446 struct dentry *(*mount) (struct file_system_type *, int,
2447 const char *, void *);
2448 void (*kill_sb) (struct super_block *);
2449 struct module *owner;
2450 struct file_system_type * next;
2451 struct hlist_head fs_supers;
2453 struct lock_class_key s_lock_key;
2454 struct lock_class_key s_umount_key;
2455 struct lock_class_key s_vfs_rename_key;
2456 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2458 struct lock_class_key i_lock_key;
2459 struct lock_class_key i_mutex_key;
2460 struct lock_class_key invalidate_lock_key;
2461 struct lock_class_key i_mutex_dir_key;
2464 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2466 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2467 int flags, const char *dev_name, void *data,
2468 int (*fill_super)(struct super_block *, void *, int));
2469 extern struct dentry *mount_single(struct file_system_type *fs_type,
2470 int flags, void *data,
2471 int (*fill_super)(struct super_block *, void *, int));
2472 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2473 int flags, void *data,
2474 int (*fill_super)(struct super_block *, void *, int));
2475 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2476 void generic_shutdown_super(struct super_block *sb);
2477 void kill_block_super(struct super_block *sb);
2478 void kill_anon_super(struct super_block *sb);
2479 void kill_litter_super(struct super_block *sb);
2480 void deactivate_super(struct super_block *sb);
2481 void deactivate_locked_super(struct super_block *sb);
2482 int set_anon_super(struct super_block *s, void *data);
2483 int set_anon_super_fc(struct super_block *s, struct fs_context *fc);
2484 int get_anon_bdev(dev_t *);
2485 void free_anon_bdev(dev_t);
2486 struct super_block *sget_fc(struct fs_context *fc,
2487 int (*test)(struct super_block *, struct fs_context *),
2488 int (*set)(struct super_block *, struct fs_context *));
2489 struct super_block *sget(struct file_system_type *type,
2490 int (*test)(struct super_block *,void *),
2491 int (*set)(struct super_block *,void *),
2492 int flags, void *data);
2494 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2495 #define fops_get(fops) \
2496 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2497 #define fops_put(fops) \
2498 do { if (fops) module_put((fops)->owner); } while(0)
2500 * This one is to be used *ONLY* from ->open() instances.
2501 * fops must be non-NULL, pinned down *and* module dependencies
2502 * should be sufficient to pin the caller down as well.
2504 #define replace_fops(f, fops) \
2506 struct file *__file = (f); \
2507 fops_put(__file->f_op); \
2508 BUG_ON(!(__file->f_op = (fops))); \
2511 extern int register_filesystem(struct file_system_type *);
2512 extern int unregister_filesystem(struct file_system_type *);
2513 extern struct vfsmount *kern_mount(struct file_system_type *);
2514 extern void kern_unmount(struct vfsmount *mnt);
2515 extern int may_umount_tree(struct vfsmount *);
2516 extern int may_umount(struct vfsmount *);
2517 extern long do_mount(const char *, const char __user *,
2518 const char *, unsigned long, void *);
2519 extern struct vfsmount *collect_mounts(const struct path *);
2520 extern void drop_collected_mounts(struct vfsmount *);
2521 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2523 extern int vfs_statfs(const struct path *, struct kstatfs *);
2524 extern int user_statfs(const char __user *, struct kstatfs *);
2525 extern int fd_statfs(int, struct kstatfs *);
2526 extern int freeze_super(struct super_block *super);
2527 extern int thaw_super(struct super_block *super);
2528 extern bool our_mnt(struct vfsmount *mnt);
2529 extern __printf(2, 3)
2530 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2531 extern int super_setup_bdi(struct super_block *sb);
2533 extern int current_umask(void);
2535 extern void ihold(struct inode * inode);
2536 extern void iput(struct inode *);
2537 extern int generic_update_time(struct inode *, struct timespec64 *, int);
2540 extern struct kobject *fs_kobj;
2542 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2544 #ifdef CONFIG_FILE_LOCKING
2545 static inline int break_lease(struct inode *inode, unsigned int mode)
2548 * Since this check is lockless, we must ensure that any refcounts
2549 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2550 * could end up racing with tasks trying to set a new lease on this
2554 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2555 return __break_lease(inode, mode, FL_LEASE);
2559 static inline int break_deleg(struct inode *inode, unsigned int mode)
2562 * Since this check is lockless, we must ensure that any refcounts
2563 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2564 * could end up racing with tasks trying to set a new lease on this
2568 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2569 return __break_lease(inode, mode, FL_DELEG);
2573 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2577 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2578 if (ret == -EWOULDBLOCK && delegated_inode) {
2579 *delegated_inode = inode;
2585 static inline int break_deleg_wait(struct inode **delegated_inode)
2589 ret = break_deleg(*delegated_inode, O_WRONLY);
2590 iput(*delegated_inode);
2591 *delegated_inode = NULL;
2595 static inline int break_layout(struct inode *inode, bool wait)
2598 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2599 return __break_lease(inode,
2600 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2605 #else /* !CONFIG_FILE_LOCKING */
2606 static inline int break_lease(struct inode *inode, unsigned int mode)
2611 static inline int break_deleg(struct inode *inode, unsigned int mode)
2616 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2621 static inline int break_deleg_wait(struct inode **delegated_inode)
2627 static inline int break_layout(struct inode *inode, bool wait)
2632 #endif /* CONFIG_FILE_LOCKING */
2637 const char *name; /* pointer to actual string */
2638 const __user char *uptr; /* original userland pointer */
2640 struct audit_names *aname;
2643 static_assert(offsetof(struct filename, iname) % sizeof(long) == 0);
2645 static inline struct user_namespace *file_mnt_user_ns(struct file *file)
2647 return mnt_user_ns(file->f_path.mnt);
2651 * is_idmapped_mnt - check whether a mount is mapped
2652 * @mnt: the mount to check
2654 * If @mnt has an idmapping attached different from the
2655 * filesystem's idmapping then @mnt is mapped.
2657 * Return: true if mount is mapped, false if not.
2659 static inline bool is_idmapped_mnt(const struct vfsmount *mnt)
2661 return mnt_user_ns(mnt) != mnt->mnt_sb->s_user_ns;
2664 extern long vfs_truncate(const struct path *, loff_t);
2665 int do_truncate(struct user_namespace *, struct dentry *, loff_t start,
2666 unsigned int time_attrs, struct file *filp);
2667 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2669 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2671 extern struct file *file_open_name(struct filename *, int, umode_t);
2672 extern struct file *filp_open(const char *, int, umode_t);
2673 extern struct file *file_open_root(const struct path *,
2674 const char *, int, umode_t);
2675 static inline struct file *file_open_root_mnt(struct vfsmount *mnt,
2676 const char *name, int flags, umode_t mode)
2678 return file_open_root(&(struct path){.mnt = mnt, .dentry = mnt->mnt_root},
2681 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2682 extern struct file * open_with_fake_path(const struct path *, int,
2683 struct inode*, const struct cred *);
2684 static inline struct file *file_clone_open(struct file *file)
2686 return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2688 extern int filp_close(struct file *, fl_owner_t id);
2690 extern struct filename *getname_flags(const char __user *, int, int *);
2691 extern struct filename *getname_uflags(const char __user *, int);
2692 extern struct filename *getname(const char __user *);
2693 extern struct filename *getname_kernel(const char *);
2694 extern void putname(struct filename *name);
2696 extern int finish_open(struct file *file, struct dentry *dentry,
2697 int (*open)(struct inode *, struct file *));
2698 extern int finish_no_open(struct file *file, struct dentry *dentry);
2701 extern void __init vfs_caches_init_early(void);
2702 extern void __init vfs_caches_init(void);
2704 extern struct kmem_cache *names_cachep;
2706 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2707 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2709 extern struct super_block *blockdev_superblock;
2710 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2712 return IS_ENABLED(CONFIG_BLOCK) && sb == blockdev_superblock;
2715 void emergency_thaw_all(void);
2716 extern int sync_filesystem(struct super_block *);
2717 extern const struct file_operations def_blk_fops;
2718 extern const struct file_operations def_chr_fops;
2721 #define CHRDEV_MAJOR_MAX 512
2722 /* Marks the bottom of the first segment of free char majors */
2723 #define CHRDEV_MAJOR_DYN_END 234
2724 /* Marks the top and bottom of the second segment of free char majors */
2725 #define CHRDEV_MAJOR_DYN_EXT_START 511
2726 #define CHRDEV_MAJOR_DYN_EXT_END 384
2728 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2729 extern int register_chrdev_region(dev_t, unsigned, const char *);
2730 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2731 unsigned int count, const char *name,
2732 const struct file_operations *fops);
2733 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2734 unsigned int count, const char *name);
2735 extern void unregister_chrdev_region(dev_t, unsigned);
2736 extern void chrdev_show(struct seq_file *,off_t);
2738 static inline int register_chrdev(unsigned int major, const char *name,
2739 const struct file_operations *fops)
2741 return __register_chrdev(major, 0, 256, name, fops);
2744 static inline void unregister_chrdev(unsigned int major, const char *name)
2746 __unregister_chrdev(major, 0, 256, name);
2749 extern void init_special_inode(struct inode *, umode_t, dev_t);
2751 /* Invalid inode operations -- fs/bad_inode.c */
2752 extern void make_bad_inode(struct inode *);
2753 extern bool is_bad_inode(struct inode *);
2755 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2756 pgoff_t start, pgoff_t end);
2758 void invalidate_mapping_pagevec(struct address_space *mapping,
2759 pgoff_t start, pgoff_t end,
2760 unsigned long *nr_pagevec);
2762 static inline void invalidate_remote_inode(struct inode *inode)
2764 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2765 S_ISLNK(inode->i_mode))
2766 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2768 extern int invalidate_inode_pages2(struct address_space *mapping);
2769 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2770 pgoff_t start, pgoff_t end);
2771 extern int write_inode_now(struct inode *, int);
2772 extern int filemap_fdatawrite(struct address_space *);
2773 extern int filemap_flush(struct address_space *);
2774 extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2775 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2777 extern int filemap_fdatawait_range_keep_errors(struct address_space *mapping,
2778 loff_t start_byte, loff_t end_byte);
2780 static inline int filemap_fdatawait(struct address_space *mapping)
2782 return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2785 extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2787 extern int filemap_write_and_wait_range(struct address_space *mapping,
2788 loff_t lstart, loff_t lend);
2789 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2790 loff_t start, loff_t end, int sync_mode);
2791 extern int filemap_fdatawrite_range(struct address_space *mapping,
2792 loff_t start, loff_t end);
2793 extern int filemap_check_errors(struct address_space *mapping);
2794 extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2795 int filemap_fdatawrite_wbc(struct address_space *mapping,
2796 struct writeback_control *wbc);
2798 static inline int filemap_write_and_wait(struct address_space *mapping)
2800 return filemap_write_and_wait_range(mapping, 0, LLONG_MAX);
2803 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2805 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2806 extern int __must_check file_write_and_wait_range(struct file *file,
2807 loff_t start, loff_t end);
2809 static inline int file_write_and_wait(struct file *file)
2811 return file_write_and_wait_range(file, 0, LLONG_MAX);
2815 * filemap_set_wb_err - set a writeback error on an address_space
2816 * @mapping: mapping in which to set writeback error
2817 * @err: error to be set in mapping
2819 * When writeback fails in some way, we must record that error so that
2820 * userspace can be informed when fsync and the like are called. We endeavor
2821 * to report errors on any file that was open at the time of the error. Some
2822 * internal callers also need to know when writeback errors have occurred.
2824 * When a writeback error occurs, most filesystems will want to call
2825 * filemap_set_wb_err to record the error in the mapping so that it will be
2826 * automatically reported whenever fsync is called on the file.
2828 static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2830 /* Fastpath for common case of no error */
2832 __filemap_set_wb_err(mapping, err);
2836 * filemap_check_wb_err - has an error occurred since the mark was sampled?
2837 * @mapping: mapping to check for writeback errors
2838 * @since: previously-sampled errseq_t
2840 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2841 * the given value was sampled.
2843 * If it has then report the latest error set, otherwise return 0.
2845 static inline int filemap_check_wb_err(struct address_space *mapping,
2848 return errseq_check(&mapping->wb_err, since);
2852 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2853 * @mapping: mapping to be sampled
2855 * Writeback errors are always reported relative to a particular sample point
2856 * in the past. This function provides those sample points.
2858 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2860 return errseq_sample(&mapping->wb_err);
2864 * file_sample_sb_err - sample the current errseq_t to test for later errors
2865 * @file: file pointer to be sampled
2867 * Grab the most current superblock-level errseq_t value for the given
2870 static inline errseq_t file_sample_sb_err(struct file *file)
2872 return errseq_sample(&file->f_path.dentry->d_sb->s_wb_err);
2875 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2877 extern int vfs_fsync(struct file *file, int datasync);
2879 extern int sync_file_range(struct file *file, loff_t offset, loff_t nbytes,
2880 unsigned int flags);
2883 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2884 * to already be updated for the write, and will return either the amount
2885 * of bytes passed in, or an error if syncing the file failed.
2887 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2889 if (iocb->ki_flags & IOCB_DSYNC) {
2890 int ret = vfs_fsync_range(iocb->ki_filp,
2891 iocb->ki_pos - count, iocb->ki_pos - 1,
2892 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2900 extern void emergency_sync(void);
2901 extern void emergency_remount(void);
2904 extern int bmap(struct inode *inode, sector_t *block);
2906 static inline int bmap(struct inode *inode, sector_t *block)
2912 int notify_change(struct user_namespace *, struct dentry *,
2913 struct iattr *, struct inode **);
2914 int inode_permission(struct user_namespace *, struct inode *, int);
2915 int generic_permission(struct user_namespace *, struct inode *, int);
2916 static inline int file_permission(struct file *file, int mask)
2918 return inode_permission(file_mnt_user_ns(file),
2919 file_inode(file), mask);
2921 static inline int path_permission(const struct path *path, int mask)
2923 return inode_permission(mnt_user_ns(path->mnt),
2924 d_inode(path->dentry), mask);
2926 int __check_sticky(struct user_namespace *mnt_userns, struct inode *dir,
2927 struct inode *inode);
2929 static inline bool execute_ok(struct inode *inode)
2931 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2934 static inline bool inode_wrong_type(const struct inode *inode, umode_t mode)
2936 return (inode->i_mode ^ mode) & S_IFMT;
2939 static inline void file_start_write(struct file *file)
2941 if (!S_ISREG(file_inode(file)->i_mode))
2943 sb_start_write(file_inode(file)->i_sb);
2946 static inline bool file_start_write_trylock(struct file *file)
2948 if (!S_ISREG(file_inode(file)->i_mode))
2950 return sb_start_write_trylock(file_inode(file)->i_sb);
2953 static inline void file_end_write(struct file *file)
2955 if (!S_ISREG(file_inode(file)->i_mode))
2957 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2961 * This is used for regular files where some users -- especially the
2962 * currently executed binary in a process, previously handled via
2963 * VM_DENYWRITE -- cannot handle concurrent write (and maybe mmap
2964 * read-write shared) accesses.
2966 * get_write_access() gets write permission for a file.
2967 * put_write_access() releases this write permission.
2968 * deny_write_access() denies write access to a file.
2969 * allow_write_access() re-enables write access to a file.
2971 * The i_writecount field of an inode can have the following values:
2972 * 0: no write access, no denied write access
2973 * < 0: (-i_writecount) users that denied write access to the file.
2974 * > 0: (i_writecount) users that have write access to the file.
2976 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2977 * except for the cases where we don't hold i_writecount yet. Then we need to
2978 * use {get,deny}_write_access() - these functions check the sign and refuse
2979 * to do the change if sign is wrong.
2981 static inline int get_write_access(struct inode *inode)
2983 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2985 static inline int deny_write_access(struct file *file)
2987 struct inode *inode = file_inode(file);
2988 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2990 static inline void put_write_access(struct inode * inode)
2992 atomic_dec(&inode->i_writecount);
2994 static inline void allow_write_access(struct file *file)
2997 atomic_inc(&file_inode(file)->i_writecount);
2999 static inline bool inode_is_open_for_write(const struct inode *inode)
3001 return atomic_read(&inode->i_writecount) > 0;
3004 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
3005 static inline void i_readcount_dec(struct inode *inode)
3007 BUG_ON(!atomic_read(&inode->i_readcount));
3008 atomic_dec(&inode->i_readcount);
3010 static inline void i_readcount_inc(struct inode *inode)
3012 atomic_inc(&inode->i_readcount);
3015 static inline void i_readcount_dec(struct inode *inode)
3019 static inline void i_readcount_inc(struct inode *inode)
3024 extern int do_pipe_flags(int *, int);
3026 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
3027 ssize_t __kernel_read(struct file *file, void *buf, size_t count, loff_t *pos);
3028 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
3029 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
3030 extern struct file * open_exec(const char *);
3032 /* fs/dcache.c -- generic fs support functions */
3033 extern bool is_subdir(struct dentry *, struct dentry *);
3034 extern bool path_is_under(const struct path *, const struct path *);
3036 extern char *file_path(struct file *, char *, int);
3038 #include <linux/err.h>
3040 /* needed for stackable file system support */
3041 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
3043 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
3045 extern int inode_init_always(struct super_block *, struct inode *);
3046 extern void inode_init_once(struct inode *);
3047 extern void address_space_init_once(struct address_space *mapping);
3048 extern struct inode * igrab(struct inode *);
3049 extern ino_t iunique(struct super_block *, ino_t);
3050 extern int inode_needs_sync(struct inode *inode);
3051 extern int generic_delete_inode(struct inode *inode);
3052 static inline int generic_drop_inode(struct inode *inode)
3054 return !inode->i_nlink || inode_unhashed(inode);
3056 extern void d_mark_dontcache(struct inode *inode);
3058 extern struct inode *ilookup5_nowait(struct super_block *sb,
3059 unsigned long hashval, int (*test)(struct inode *, void *),
3061 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
3062 int (*test)(struct inode *, void *), void *data);
3063 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
3065 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
3066 int (*test)(struct inode *, void *),
3067 int (*set)(struct inode *, void *),
3069 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
3070 extern struct inode * iget_locked(struct super_block *, unsigned long);
3071 extern struct inode *find_inode_nowait(struct super_block *,
3073 int (*match)(struct inode *,
3074 unsigned long, void *),
3076 extern struct inode *find_inode_rcu(struct super_block *, unsigned long,
3077 int (*)(struct inode *, void *), void *);
3078 extern struct inode *find_inode_by_ino_rcu(struct super_block *, unsigned long);
3079 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
3080 extern int insert_inode_locked(struct inode *);
3081 #ifdef CONFIG_DEBUG_LOCK_ALLOC
3082 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
3084 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
3086 extern void unlock_new_inode(struct inode *);
3087 extern void discard_new_inode(struct inode *);
3088 extern unsigned int get_next_ino(void);
3089 extern void evict_inodes(struct super_block *sb);
3090 void dump_mapping(const struct address_space *);
3093 * Userspace may rely on the the inode number being non-zero. For example, glibc
3094 * simply ignores files with zero i_ino in unlink() and other places.
3096 * As an additional complication, if userspace was compiled with
3097 * _FILE_OFFSET_BITS=32 on a 64-bit kernel we'll only end up reading out the
3098 * lower 32 bits, so we need to check that those aren't zero explicitly. With
3099 * _FILE_OFFSET_BITS=64, this may cause some harmless false-negatives, but
3100 * better safe than sorry.
3102 static inline bool is_zero_ino(ino_t ino)
3104 return (u32)ino == 0;
3107 extern void __iget(struct inode * inode);
3108 extern void iget_failed(struct inode *);
3109 extern void clear_inode(struct inode *);
3110 extern void __destroy_inode(struct inode *);
3111 extern struct inode *new_inode_pseudo(struct super_block *sb);
3112 extern struct inode *new_inode(struct super_block *sb);
3113 extern void free_inode_nonrcu(struct inode *inode);
3114 extern int should_remove_suid(struct dentry *);
3115 extern int file_remove_privs(struct file *);
3117 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
3118 static inline void insert_inode_hash(struct inode *inode)
3120 __insert_inode_hash(inode, inode->i_ino);
3123 extern void __remove_inode_hash(struct inode *);
3124 static inline void remove_inode_hash(struct inode *inode)
3126 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
3127 __remove_inode_hash(inode);
3130 extern void inode_sb_list_add(struct inode *inode);
3131 extern void inode_add_lru(struct inode *inode);
3133 extern int sb_set_blocksize(struct super_block *, int);
3134 extern int sb_min_blocksize(struct super_block *, int);
3136 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
3137 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
3138 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
3139 extern int generic_write_check_limits(struct file *file, loff_t pos,
3141 extern int generic_file_rw_checks(struct file *file_in, struct file *file_out);
3142 ssize_t filemap_read(struct kiocb *iocb, struct iov_iter *to,
3143 ssize_t already_read);
3144 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
3145 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
3146 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
3147 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
3148 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
3150 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
3152 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
3154 ssize_t vfs_iocb_iter_read(struct file *file, struct kiocb *iocb,
3155 struct iov_iter *iter);
3156 ssize_t vfs_iocb_iter_write(struct file *file, struct kiocb *iocb,
3157 struct iov_iter *iter);
3160 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
3161 struct pipe_inode_info *, size_t, unsigned int);
3162 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
3163 struct file *, loff_t *, size_t, unsigned int);
3164 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
3165 struct file *out, loff_t *, size_t len, unsigned int flags);
3166 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
3167 loff_t *opos, size_t len, unsigned int flags);
3171 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3172 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3173 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
3174 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3175 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3176 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3177 int whence, loff_t maxsize, loff_t eof);
3178 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3179 int whence, loff_t size);
3180 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3181 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3182 extern int generic_file_open(struct inode * inode, struct file * filp);
3183 extern int nonseekable_open(struct inode * inode, struct file * filp);
3184 extern int stream_open(struct inode * inode, struct file * filp);
3187 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3188 loff_t file_offset);
3191 /* need locking between buffered and direct access */
3194 /* filesystem does not support filling holes */
3195 DIO_SKIP_HOLES = 0x02,
3198 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3199 struct block_device *bdev, struct iov_iter *iter,
3200 get_block_t get_block,
3201 dio_iodone_t end_io, dio_submit_t submit_io,
3204 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3205 struct inode *inode,
3206 struct iov_iter *iter,
3207 get_block_t get_block)
3209 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3210 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3214 void inode_dio_wait(struct inode *inode);
3217 * inode_dio_begin - signal start of a direct I/O requests
3218 * @inode: inode the direct I/O happens on
3220 * This is called once we've finished processing a direct I/O request,
3221 * and is used to wake up callers waiting for direct I/O to be quiesced.
3223 static inline void inode_dio_begin(struct inode *inode)
3225 atomic_inc(&inode->i_dio_count);
3229 * inode_dio_end - signal finish of a direct I/O requests
3230 * @inode: inode the direct I/O happens on
3232 * This is called once we've finished processing a direct I/O request,
3233 * and is used to wake up callers waiting for direct I/O to be quiesced.
3235 static inline void inode_dio_end(struct inode *inode)
3237 if (atomic_dec_and_test(&inode->i_dio_count))
3238 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3242 * Warn about a page cache invalidation failure diring a direct I/O write.
3244 void dio_warn_stale_pagecache(struct file *filp);
3246 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3249 extern const struct file_operations generic_ro_fops;
3251 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3253 extern int readlink_copy(char __user *, int, const char *);
3254 extern int page_readlink(struct dentry *, char __user *, int);
3255 extern const char *page_get_link(struct dentry *, struct inode *,
3256 struct delayed_call *);
3257 extern void page_put_link(void *);
3258 extern int __page_symlink(struct inode *inode, const char *symname, int len,
3260 extern int page_symlink(struct inode *inode, const char *symname, int len);
3261 extern const struct inode_operations page_symlink_inode_operations;
3262 extern void kfree_link(void *);
3263 void generic_fillattr(struct user_namespace *, struct inode *, struct kstat *);
3264 void generic_fill_statx_attr(struct inode *inode, struct kstat *stat);
3265 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3266 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3267 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3268 void inode_add_bytes(struct inode *inode, loff_t bytes);
3269 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3270 void inode_sub_bytes(struct inode *inode, loff_t bytes);
3271 static inline loff_t __inode_get_bytes(struct inode *inode)
3273 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3275 loff_t inode_get_bytes(struct inode *inode);
3276 void inode_set_bytes(struct inode *inode, loff_t bytes);
3277 const char *simple_get_link(struct dentry *, struct inode *,
3278 struct delayed_call *);
3279 extern const struct inode_operations simple_symlink_inode_operations;
3281 extern int iterate_dir(struct file *, struct dir_context *);
3283 int vfs_fstatat(int dfd, const char __user *filename, struct kstat *stat,
3285 int vfs_fstat(int fd, struct kstat *stat);
3287 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3289 return vfs_fstatat(AT_FDCWD, filename, stat, 0);
3291 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3293 return vfs_fstatat(AT_FDCWD, name, stat, AT_SYMLINK_NOFOLLOW);
3296 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3297 extern int vfs_readlink(struct dentry *, char __user *, int);
3299 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3300 extern void put_filesystem(struct file_system_type *fs);
3301 extern struct file_system_type *get_fs_type(const char *name);
3302 extern struct super_block *get_super(struct block_device *);
3303 extern struct super_block *get_active_super(struct block_device *bdev);
3304 extern void drop_super(struct super_block *sb);
3305 extern void drop_super_exclusive(struct super_block *sb);
3306 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3307 extern void iterate_supers_type(struct file_system_type *,
3308 void (*)(struct super_block *, void *), void *);
3310 extern int dcache_dir_open(struct inode *, struct file *);
3311 extern int dcache_dir_close(struct inode *, struct file *);
3312 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3313 extern int dcache_readdir(struct file *, struct dir_context *);
3314 extern int simple_setattr(struct user_namespace *, struct dentry *,
3316 extern int simple_getattr(struct user_namespace *, const struct path *,
3317 struct kstat *, u32, unsigned int);
3318 extern int simple_statfs(struct dentry *, struct kstatfs *);
3319 extern int simple_open(struct inode *inode, struct file *file);
3320 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3321 extern int simple_unlink(struct inode *, struct dentry *);
3322 extern int simple_rmdir(struct inode *, struct dentry *);
3323 extern int simple_rename_exchange(struct inode *old_dir, struct dentry *old_dentry,
3324 struct inode *new_dir, struct dentry *new_dentry);
3325 extern int simple_rename(struct user_namespace *, struct inode *,
3326 struct dentry *, struct inode *, struct dentry *,
3328 extern void simple_recursive_removal(struct dentry *,
3329 void (*callback)(struct dentry *));
3330 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3331 extern void noop_invalidatepage(struct page *page, unsigned int offset,
3332 unsigned int length);
3333 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3334 extern int simple_empty(struct dentry *);
3335 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3336 loff_t pos, unsigned len, unsigned flags,
3337 struct page **pagep, void **fsdata);
3338 extern const struct address_space_operations ram_aops;
3339 extern int always_delete_dentry(const struct dentry *);
3340 extern struct inode *alloc_anon_inode(struct super_block *);
3341 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3342 extern const struct dentry_operations simple_dentry_operations;
3344 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3345 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3346 extern const struct file_operations simple_dir_operations;
3347 extern const struct inode_operations simple_dir_inode_operations;
3348 extern void make_empty_dir_inode(struct inode *inode);
3349 extern bool is_empty_dir_inode(struct inode *inode);
3350 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3351 struct dentry *d_alloc_name(struct dentry *, const char *);
3352 extern int simple_fill_super(struct super_block *, unsigned long,
3353 const struct tree_descr *);
3354 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3355 extern void simple_release_fs(struct vfsmount **mount, int *count);
3357 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3358 loff_t *ppos, const void *from, size_t available);
3359 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3360 const void __user *from, size_t count);
3362 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3363 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3365 extern int generic_check_addressable(unsigned, u64);
3367 extern void generic_set_encrypted_ci_d_ops(struct dentry *dentry);
3369 #ifdef CONFIG_MIGRATION
3370 extern int buffer_migrate_page(struct address_space *,
3371 struct page *, struct page *,
3373 extern int buffer_migrate_page_norefs(struct address_space *,
3374 struct page *, struct page *,
3377 #define buffer_migrate_page NULL
3378 #define buffer_migrate_page_norefs NULL
3381 int may_setattr(struct user_namespace *mnt_userns, struct inode *inode,
3382 unsigned int ia_valid);
3383 int setattr_prepare(struct user_namespace *, struct dentry *, struct iattr *);
3384 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3385 void setattr_copy(struct user_namespace *, struct inode *inode,
3386 const struct iattr *attr);
3388 extern int file_update_time(struct file *file);
3390 static inline bool vma_is_dax(const struct vm_area_struct *vma)
3392 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3395 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3397 struct inode *inode;
3399 if (!IS_ENABLED(CONFIG_FS_DAX) || !vma->vm_file)
3401 if (!vma_is_dax(vma))
3403 inode = file_inode(vma->vm_file);
3404 if (S_ISCHR(inode->i_mode))
3405 return false; /* device-dax */
3409 static inline int iocb_flags(struct file *file)
3412 if (file->f_flags & O_APPEND)
3414 if (file->f_flags & O_DIRECT)
3416 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3418 if (file->f_flags & __O_SYNC)
3423 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3425 int kiocb_flags = 0;
3427 /* make sure there's no overlap between RWF and private IOCB flags */
3428 BUILD_BUG_ON((__force int) RWF_SUPPORTED & IOCB_EVENTFD);
3432 if (unlikely(flags & ~RWF_SUPPORTED))
3435 if (flags & RWF_NOWAIT) {
3436 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3438 kiocb_flags |= IOCB_NOIO;
3440 kiocb_flags |= (__force int) (flags & RWF_SUPPORTED);
3441 if (flags & RWF_SYNC)
3442 kiocb_flags |= IOCB_DSYNC;
3444 ki->ki_flags |= kiocb_flags;
3448 static inline ino_t parent_ino(struct dentry *dentry)
3453 * Don't strictly need d_lock here? If the parent ino could change
3454 * then surely we'd have a deeper race in the caller?
3456 spin_lock(&dentry->d_lock);
3457 res = dentry->d_parent->d_inode->i_ino;
3458 spin_unlock(&dentry->d_lock);
3462 /* Transaction based IO helpers */
3465 * An argresp is stored in an allocated page and holds the
3466 * size of the argument or response, along with its content
3468 struct simple_transaction_argresp {
3473 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3475 char *simple_transaction_get(struct file *file, const char __user *buf,
3477 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3478 size_t size, loff_t *pos);
3479 int simple_transaction_release(struct inode *inode, struct file *file);
3481 void simple_transaction_set(struct file *file, size_t n);
3484 * simple attribute files
3486 * These attributes behave similar to those in sysfs:
3488 * Writing to an attribute immediately sets a value, an open file can be
3489 * written to multiple times.
3491 * Reading from an attribute creates a buffer from the value that might get
3492 * read with multiple read calls. When the attribute has been read
3493 * completely, no further read calls are possible until the file is opened
3496 * All attributes contain a text representation of a numeric value
3497 * that are accessed with the get() and set() functions.
3499 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3500 static int __fops ## _open(struct inode *inode, struct file *file) \
3502 __simple_attr_check_format(__fmt, 0ull); \
3503 return simple_attr_open(inode, file, __get, __set, __fmt); \
3505 static const struct file_operations __fops = { \
3506 .owner = THIS_MODULE, \
3507 .open = __fops ## _open, \
3508 .release = simple_attr_release, \
3509 .read = simple_attr_read, \
3510 .write = simple_attr_write, \
3511 .llseek = generic_file_llseek, \
3514 static inline __printf(1, 2)
3515 void __simple_attr_check_format(const char *fmt, ...)
3517 /* don't do anything, just let the compiler check the arguments; */
3520 int simple_attr_open(struct inode *inode, struct file *file,
3521 int (*get)(void *, u64 *), int (*set)(void *, u64),
3523 int simple_attr_release(struct inode *inode, struct file *file);
3524 ssize_t simple_attr_read(struct file *file, char __user *buf,
3525 size_t len, loff_t *ppos);
3526 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3527 size_t len, loff_t *ppos);
3530 int __init list_bdev_fs_names(char *buf, size_t size);
3532 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3533 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3535 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3536 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3537 (flag & __FMODE_NONOTIFY)))
3539 static inline bool is_sxid(umode_t mode)
3541 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3544 static inline int check_sticky(struct user_namespace *mnt_userns,
3545 struct inode *dir, struct inode *inode)
3547 if (!(dir->i_mode & S_ISVTX))
3550 return __check_sticky(mnt_userns, dir, inode);
3553 static inline void inode_has_no_xattr(struct inode *inode)
3555 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3556 inode->i_flags |= S_NOSEC;
3559 static inline bool is_root_inode(struct inode *inode)
3561 return inode == inode->i_sb->s_root->d_inode;
3564 static inline bool dir_emit(struct dir_context *ctx,
3565 const char *name, int namelen,
3566 u64 ino, unsigned type)
3568 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3570 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3572 return ctx->actor(ctx, ".", 1, ctx->pos,
3573 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3575 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3577 return ctx->actor(ctx, "..", 2, ctx->pos,
3578 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3580 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3582 if (ctx->pos == 0) {
3583 if (!dir_emit_dot(file, ctx))
3587 if (ctx->pos == 1) {
3588 if (!dir_emit_dotdot(file, ctx))
3594 static inline bool dir_relax(struct inode *inode)
3596 inode_unlock(inode);
3598 return !IS_DEADDIR(inode);
3601 static inline bool dir_relax_shared(struct inode *inode)
3603 inode_unlock_shared(inode);
3604 inode_lock_shared(inode);
3605 return !IS_DEADDIR(inode);
3608 extern bool path_noexec(const struct path *path);
3609 extern void inode_nohighmem(struct inode *inode);
3612 extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len,
3614 extern int generic_fadvise(struct file *file, loff_t offset, loff_t len,
3618 * Flush file data before changing attributes. Caller must hold any locks
3619 * required to prevent further writes to this file until we're done setting
3622 static inline int inode_drain_writes(struct inode *inode)
3624 inode_dio_wait(inode);
3625 return filemap_write_and_wait(inode->i_mapping);
3628 #endif /* _LINUX_FS_H */