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;
52 struct export_operations;
53 struct fiemap_extent_info;
58 struct pipe_inode_info;
59 struct poll_table_struct;
61 struct vm_area_struct;
64 struct swap_info_struct;
66 struct workqueue_struct;
69 struct fscrypt_operations;
71 struct fsverity_operations;
73 struct fs_parameter_spec;
76 extern void __init inode_init(void);
77 extern void __init inode_init_early(void);
78 extern void __init files_init(void);
79 extern void __init files_maxfiles_init(void);
81 extern struct files_stat_struct files_stat;
82 extern unsigned long get_max_files(void);
83 extern unsigned int sysctl_nr_open;
84 extern struct inodes_stat_t inodes_stat;
85 extern int leases_enable, lease_break_time;
86 extern int sysctl_protected_symlinks;
87 extern int sysctl_protected_hardlinks;
88 extern int sysctl_protected_fifos;
89 extern int sysctl_protected_regular;
91 typedef __kernel_rwf_t rwf_t;
94 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
95 struct buffer_head *bh_result, int create);
96 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
97 ssize_t bytes, void *private);
99 #define MAY_EXEC 0x00000001
100 #define MAY_WRITE 0x00000002
101 #define MAY_READ 0x00000004
102 #define MAY_APPEND 0x00000008
103 #define MAY_ACCESS 0x00000010
104 #define MAY_OPEN 0x00000020
105 #define MAY_CHDIR 0x00000040
106 /* called from RCU mode, don't block */
107 #define MAY_NOT_BLOCK 0x00000080
110 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
111 * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
114 /* file is open for reading */
115 #define FMODE_READ ((__force fmode_t)0x1)
116 /* file is open for writing */
117 #define FMODE_WRITE ((__force fmode_t)0x2)
118 /* file is seekable */
119 #define FMODE_LSEEK ((__force fmode_t)0x4)
120 /* file can be accessed using pread */
121 #define FMODE_PREAD ((__force fmode_t)0x8)
122 /* file can be accessed using pwrite */
123 #define FMODE_PWRITE ((__force fmode_t)0x10)
124 /* File is opened for execution with sys_execve / sys_uselib */
125 #define FMODE_EXEC ((__force fmode_t)0x20)
126 /* File is opened with O_NDELAY (only set for block devices) */
127 #define FMODE_NDELAY ((__force fmode_t)0x40)
128 /* File is opened with O_EXCL (only set for block devices) */
129 #define FMODE_EXCL ((__force fmode_t)0x80)
130 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
131 (specialy hack for floppy.c) */
132 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
133 /* 32bit hashes as llseek() offset (for directories) */
134 #define FMODE_32BITHASH ((__force fmode_t)0x200)
135 /* 64bit hashes as llseek() offset (for directories) */
136 #define FMODE_64BITHASH ((__force fmode_t)0x400)
139 * Don't update ctime and mtime.
141 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
142 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
144 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
146 /* Expect random access pattern */
147 #define FMODE_RANDOM ((__force fmode_t)0x1000)
149 /* File is huge (eg. /dev/mem): treat loff_t as unsigned */
150 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
152 /* File is opened with O_PATH; almost nothing can be done with it */
153 #define FMODE_PATH ((__force fmode_t)0x4000)
155 /* File needs atomic accesses to f_pos */
156 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
157 /* Write access to underlying fs */
158 #define FMODE_WRITER ((__force fmode_t)0x10000)
159 /* Has read method(s) */
160 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
161 /* Has write method(s) */
162 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
164 #define FMODE_OPENED ((__force fmode_t)0x80000)
165 #define FMODE_CREATED ((__force fmode_t)0x100000)
167 /* File is stream-like */
168 #define FMODE_STREAM ((__force fmode_t)0x200000)
170 /* File was opened by fanotify and shouldn't generate fanotify events */
171 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
173 /* File is capable of returning -EAGAIN if I/O will block */
174 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
176 /* File represents mount that needs unmounting */
177 #define FMODE_NEED_UNMOUNT ((__force fmode_t)0x10000000)
179 /* File does not contribute to nr_files count */
180 #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000)
182 /* File supports async buffered reads */
183 #define FMODE_BUF_RASYNC ((__force fmode_t)0x40000000)
186 * Attribute flags. These should be or-ed together to figure out what
189 #define ATTR_MODE (1 << 0)
190 #define ATTR_UID (1 << 1)
191 #define ATTR_GID (1 << 2)
192 #define ATTR_SIZE (1 << 3)
193 #define ATTR_ATIME (1 << 4)
194 #define ATTR_MTIME (1 << 5)
195 #define ATTR_CTIME (1 << 6)
196 #define ATTR_ATIME_SET (1 << 7)
197 #define ATTR_MTIME_SET (1 << 8)
198 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
199 #define ATTR_KILL_SUID (1 << 11)
200 #define ATTR_KILL_SGID (1 << 12)
201 #define ATTR_FILE (1 << 13)
202 #define ATTR_KILL_PRIV (1 << 14)
203 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
204 #define ATTR_TIMES_SET (1 << 16)
205 #define ATTR_TOUCH (1 << 17)
208 * Whiteout is represented by a char device. The following constants define the
209 * mode and device number to use.
211 #define WHITEOUT_MODE 0
212 #define WHITEOUT_DEV 0
215 * This is the Inode Attributes structure, used for notify_change(). It
216 * uses the above definitions as flags, to know which values have changed.
217 * Also, in this manner, a Filesystem can look at only the values it cares
218 * about. Basically, these are the attributes that the VFS layer can
219 * request to change from the FS layer.
221 * Derek Atkins <warlord@MIT.EDU> 94-10-20
224 unsigned int ia_valid;
229 struct timespec64 ia_atime;
230 struct timespec64 ia_mtime;
231 struct timespec64 ia_ctime;
234 * Not an attribute, but an auxiliary info for filesystems wanting to
235 * implement an ftruncate() like method. NOTE: filesystem should
236 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
238 struct file *ia_file;
242 * Includes for diskquotas.
244 #include <linux/quota.h>
247 * Maximum number of layers of fs stack. Needs to be limited to
248 * prevent kernel stack overflow
250 #define FILESYSTEM_MAX_STACK_DEPTH 2
253 * enum positive_aop_returns - aop return codes with specific semantics
255 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
256 * completed, that the page is still locked, and
257 * should be considered active. The VM uses this hint
258 * to return the page to the active list -- it won't
259 * be a candidate for writeback again in the near
260 * future. Other callers must be careful to unlock
261 * the page if they get this return. Returned by
264 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
265 * unlocked it and the page might have been truncated.
266 * The caller should back up to acquiring a new page and
267 * trying again. The aop will be taking reasonable
268 * precautions not to livelock. If the caller held a page
269 * reference, it should drop it before retrying. Returned
272 * address_space_operation functions return these large constants to indicate
273 * special semantics to the caller. These are much larger than the bytes in a
274 * page to allow for functions that return the number of bytes operated on in a
278 enum positive_aop_returns {
279 AOP_WRITEPAGE_ACTIVATE = 0x80000,
280 AOP_TRUNCATED_PAGE = 0x80001,
283 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
284 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
285 * helper code (eg buffer layer)
286 * to clear GFP_FS from alloc */
289 * oh the beauties of C type declarations.
292 struct address_space;
293 struct writeback_control;
294 struct readahead_control;
297 * Write life time hint values.
298 * Stored in struct inode as u8.
301 WRITE_LIFE_NOT_SET = 0,
302 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
303 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
304 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
305 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
306 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
309 /* Match RWF_* bits to IOCB bits */
310 #define IOCB_HIPRI (__force int) RWF_HIPRI
311 #define IOCB_DSYNC (__force int) RWF_DSYNC
312 #define IOCB_SYNC (__force int) RWF_SYNC
313 #define IOCB_NOWAIT (__force int) RWF_NOWAIT
314 #define IOCB_APPEND (__force int) RWF_APPEND
316 /* non-RWF related bits - start at 16 */
317 #define IOCB_EVENTFD (1 << 16)
318 #define IOCB_DIRECT (1 << 17)
319 #define IOCB_WRITE (1 << 18)
320 /* iocb->ki_waitq is valid */
321 #define IOCB_WAITQ (1 << 19)
322 #define IOCB_NOIO (1 << 20)
323 /* can use bio alloc cache */
324 #define IOCB_ALLOC_CACHE (1 << 21)
327 struct file *ki_filp;
329 /* The 'ki_filp' pointer is shared in a union for aio */
330 randomized_struct_fields_start
333 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
337 u16 ki_ioprio; /* See linux/ioprio.h */
339 unsigned int ki_cookie; /* for ->iopoll */
340 struct wait_page_queue *ki_waitq; /* for async buffered IO */
343 randomized_struct_fields_end
346 static inline bool is_sync_kiocb(struct kiocb *kiocb)
348 return kiocb->ki_complete == NULL;
352 * "descriptor" for what we're up to with a read.
353 * This allows us to use the same read code yet
354 * have multiple different users of the data that
355 * we read from a file.
357 * The simplest case just copies the data to user
370 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
371 unsigned long, unsigned long);
373 struct address_space_operations {
374 int (*writepage)(struct page *page, struct writeback_control *wbc);
375 int (*readpage)(struct file *, struct page *);
377 /* Write back some dirty pages from this mapping. */
378 int (*writepages)(struct address_space *, struct writeback_control *);
380 /* Set a page dirty. Return true if this dirtied it */
381 int (*set_page_dirty)(struct page *page);
384 * Reads in the requested pages. Unlike ->readpage(), this is
385 * PURELY used for read-ahead!.
387 int (*readpages)(struct file *filp, struct address_space *mapping,
388 struct list_head *pages, unsigned nr_pages);
389 void (*readahead)(struct readahead_control *);
391 int (*write_begin)(struct file *, struct address_space *mapping,
392 loff_t pos, unsigned len, unsigned flags,
393 struct page **pagep, void **fsdata);
394 int (*write_end)(struct file *, struct address_space *mapping,
395 loff_t pos, unsigned len, unsigned copied,
396 struct page *page, void *fsdata);
398 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
399 sector_t (*bmap)(struct address_space *, sector_t);
400 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
401 int (*releasepage) (struct page *, gfp_t);
402 void (*freepage)(struct page *);
403 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
405 * migrate the contents of a page to the specified target. If
406 * migrate_mode is MIGRATE_ASYNC, it must not block.
408 int (*migratepage) (struct address_space *,
409 struct page *, struct page *, enum migrate_mode);
410 bool (*isolate_page)(struct page *, isolate_mode_t);
411 void (*putback_page)(struct page *);
412 int (*launder_page) (struct page *);
413 int (*is_partially_uptodate) (struct page *, unsigned long,
415 void (*is_dirty_writeback) (struct page *, bool *, bool *);
416 int (*error_remove_page)(struct address_space *, struct page *);
418 /* swapfile support */
419 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
421 void (*swap_deactivate)(struct file *file);
424 extern const struct address_space_operations empty_aops;
427 * pagecache_write_begin/pagecache_write_end must be used by general code
428 * to write into the pagecache.
430 int pagecache_write_begin(struct file *, struct address_space *mapping,
431 loff_t pos, unsigned len, unsigned flags,
432 struct page **pagep, void **fsdata);
434 int pagecache_write_end(struct file *, struct address_space *mapping,
435 loff_t pos, unsigned len, unsigned copied,
436 struct page *page, void *fsdata);
439 * struct address_space - Contents of a cacheable, mappable object.
440 * @host: Owner, either the inode or the block_device.
441 * @i_pages: Cached pages.
442 * @invalidate_lock: Guards coherency between page cache contents and
443 * file offset->disk block mappings in the filesystem during invalidates.
444 * It is also used to block modification of page cache contents through
446 * @gfp_mask: Memory allocation flags to use for allocating pages.
447 * @i_mmap_writable: Number of VM_SHARED mappings.
448 * @nr_thps: Number of THPs in the pagecache (non-shmem only).
449 * @i_mmap: Tree of private and shared mappings.
450 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
451 * @nrpages: Number of page entries, protected by the i_pages lock.
452 * @writeback_index: Writeback starts here.
454 * @flags: Error bits and flags (AS_*).
455 * @wb_err: The most recent error which has occurred.
456 * @private_lock: For use by the owner of the address_space.
457 * @private_list: For use by the owner of the address_space.
458 * @private_data: For use by the owner of the address_space.
460 struct address_space {
462 struct xarray i_pages;
463 struct rw_semaphore invalidate_lock;
465 atomic_t i_mmap_writable;
466 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
467 /* number of thp, only for non-shmem files */
470 struct rb_root_cached i_mmap;
471 struct rw_semaphore i_mmap_rwsem;
472 unsigned long nrpages;
473 pgoff_t writeback_index;
474 const struct address_space_operations *a_ops;
477 spinlock_t private_lock;
478 struct list_head private_list;
480 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
482 * On most architectures that alignment is already the case; but
483 * must be enforced here for CRIS, to let the least significant bit
484 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
487 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
488 #define PAGECACHE_TAG_DIRTY XA_MARK_0
489 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
490 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
493 * Returns true if any of the pages in the mapping are marked with the tag.
495 static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag)
497 return xa_marked(&mapping->i_pages, tag);
500 static inline void i_mmap_lock_write(struct address_space *mapping)
502 down_write(&mapping->i_mmap_rwsem);
505 static inline int i_mmap_trylock_write(struct address_space *mapping)
507 return down_write_trylock(&mapping->i_mmap_rwsem);
510 static inline void i_mmap_unlock_write(struct address_space *mapping)
512 up_write(&mapping->i_mmap_rwsem);
515 static inline void i_mmap_lock_read(struct address_space *mapping)
517 down_read(&mapping->i_mmap_rwsem);
520 static inline void i_mmap_unlock_read(struct address_space *mapping)
522 up_read(&mapping->i_mmap_rwsem);
525 static inline void i_mmap_assert_locked(struct address_space *mapping)
527 lockdep_assert_held(&mapping->i_mmap_rwsem);
530 static inline void i_mmap_assert_write_locked(struct address_space *mapping)
532 lockdep_assert_held_write(&mapping->i_mmap_rwsem);
536 * Might pages of this file be mapped into userspace?
538 static inline int mapping_mapped(struct address_space *mapping)
540 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
544 * Might pages of this file have been modified in userspace?
545 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap
546 * marks vma as VM_SHARED if it is shared, and the file was opened for
547 * writing i.e. vma may be mprotected writable even if now readonly.
549 * If i_mmap_writable is negative, no new writable mappings are allowed. You
550 * can only deny writable mappings, if none exists right now.
552 static inline int mapping_writably_mapped(struct address_space *mapping)
554 return atomic_read(&mapping->i_mmap_writable) > 0;
557 static inline int mapping_map_writable(struct address_space *mapping)
559 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
563 static inline void mapping_unmap_writable(struct address_space *mapping)
565 atomic_dec(&mapping->i_mmap_writable);
568 static inline int mapping_deny_writable(struct address_space *mapping)
570 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
574 static inline void mapping_allow_writable(struct address_space *mapping)
576 atomic_inc(&mapping->i_mmap_writable);
580 * Use sequence counter to get consistent i_size on 32-bit processors.
582 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
583 #include <linux/seqlock.h>
584 #define __NEED_I_SIZE_ORDERED
585 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
587 #define i_size_ordered_init(inode) do { } while (0)
591 #define ACL_NOT_CACHED ((void *)(-1))
593 * ACL_DONT_CACHE is for stacked filesystems, that rely on underlying fs to
594 * cache the ACL. This also means that ->get_acl() can be called in RCU mode
595 * with the LOOKUP_RCU flag.
597 #define ACL_DONT_CACHE ((void *)(-3))
599 static inline struct posix_acl *
600 uncached_acl_sentinel(struct task_struct *task)
602 return (void *)task + 1;
606 is_uncached_acl(struct posix_acl *acl)
608 return (long)acl & 1;
611 #define IOP_FASTPERM 0x0001
612 #define IOP_LOOKUP 0x0002
613 #define IOP_NOFOLLOW 0x0004
614 #define IOP_XATTR 0x0008
615 #define IOP_DEFAULT_READLINK 0x0010
617 struct fsnotify_mark_connector;
620 * Keep mostly read-only and often accessed (especially for
621 * the RCU path lookup and 'stat' data) fields at the beginning
622 * of the 'struct inode'
626 unsigned short i_opflags;
629 unsigned int i_flags;
631 #ifdef CONFIG_FS_POSIX_ACL
632 struct posix_acl *i_acl;
633 struct posix_acl *i_default_acl;
636 const struct inode_operations *i_op;
637 struct super_block *i_sb;
638 struct address_space *i_mapping;
640 #ifdef CONFIG_SECURITY
644 /* Stat data, not accessed from path walking */
647 * Filesystems may only read i_nlink directly. They shall use the
648 * following functions for modification:
650 * (set|clear|inc|drop)_nlink
651 * inode_(inc|dec)_link_count
654 const unsigned int i_nlink;
655 unsigned int __i_nlink;
659 struct timespec64 i_atime;
660 struct timespec64 i_mtime;
661 struct timespec64 i_ctime;
662 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
663 unsigned short i_bytes;
668 #ifdef __NEED_I_SIZE_ORDERED
669 seqcount_t i_size_seqcount;
673 unsigned long i_state;
674 struct rw_semaphore i_rwsem;
676 unsigned long dirtied_when; /* jiffies of first dirtying */
677 unsigned long dirtied_time_when;
679 struct hlist_node i_hash;
680 struct list_head i_io_list; /* backing dev IO list */
681 #ifdef CONFIG_CGROUP_WRITEBACK
682 struct bdi_writeback *i_wb; /* the associated cgroup wb */
684 /* foreign inode detection, see wbc_detach_inode() */
686 u16 i_wb_frn_avg_time;
687 u16 i_wb_frn_history;
689 struct list_head i_lru; /* inode LRU list */
690 struct list_head i_sb_list;
691 struct list_head i_wb_list; /* backing dev writeback list */
693 struct hlist_head i_dentry;
694 struct rcu_head i_rcu;
696 atomic64_t i_version;
697 atomic64_t i_sequence; /* see futex */
699 atomic_t i_dio_count;
700 atomic_t i_writecount;
701 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
702 atomic_t i_readcount; /* struct files open RO */
705 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
706 void (*free_inode)(struct inode *);
708 struct file_lock_context *i_flctx;
709 struct address_space i_data;
710 struct list_head i_devices;
712 struct pipe_inode_info *i_pipe;
720 #ifdef CONFIG_FSNOTIFY
721 __u32 i_fsnotify_mask; /* all events this inode cares about */
722 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
725 #ifdef CONFIG_FS_ENCRYPTION
726 struct fscrypt_info *i_crypt_info;
729 #ifdef CONFIG_FS_VERITY
730 struct fsverity_info *i_verity_info;
733 void *i_private; /* fs or device private pointer */
734 } __randomize_layout;
736 struct timespec64 timestamp_truncate(struct timespec64 t, struct inode *inode);
738 static inline unsigned int i_blocksize(const struct inode *node)
740 return (1 << node->i_blkbits);
743 static inline int inode_unhashed(struct inode *inode)
745 return hlist_unhashed(&inode->i_hash);
749 * __mark_inode_dirty expects inodes to be hashed. Since we don't
750 * want special inodes in the fileset inode space, we make them
751 * appear hashed, but do not put on any lists. hlist_del()
752 * will work fine and require no locking.
754 static inline void inode_fake_hash(struct inode *inode)
756 hlist_add_fake(&inode->i_hash);
760 * inode->i_mutex nesting subclasses for the lock validator:
762 * 0: the object of the current VFS operation
766 * 4: second non-directory
767 * 5: second parent (when locking independent directories in rename)
769 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
770 * non-directories at once.
772 * The locking order between these classes is
773 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
775 enum inode_i_mutex_lock_class
785 static inline void inode_lock(struct inode *inode)
787 down_write(&inode->i_rwsem);
790 static inline void inode_unlock(struct inode *inode)
792 up_write(&inode->i_rwsem);
795 static inline void inode_lock_shared(struct inode *inode)
797 down_read(&inode->i_rwsem);
800 static inline void inode_unlock_shared(struct inode *inode)
802 up_read(&inode->i_rwsem);
805 static inline int inode_trylock(struct inode *inode)
807 return down_write_trylock(&inode->i_rwsem);
810 static inline int inode_trylock_shared(struct inode *inode)
812 return down_read_trylock(&inode->i_rwsem);
815 static inline int inode_is_locked(struct inode *inode)
817 return rwsem_is_locked(&inode->i_rwsem);
820 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
822 down_write_nested(&inode->i_rwsem, subclass);
825 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
827 down_read_nested(&inode->i_rwsem, subclass);
830 static inline void filemap_invalidate_lock(struct address_space *mapping)
832 down_write(&mapping->invalidate_lock);
835 static inline void filemap_invalidate_unlock(struct address_space *mapping)
837 up_write(&mapping->invalidate_lock);
840 static inline void filemap_invalidate_lock_shared(struct address_space *mapping)
842 down_read(&mapping->invalidate_lock);
845 static inline int filemap_invalidate_trylock_shared(
846 struct address_space *mapping)
848 return down_read_trylock(&mapping->invalidate_lock);
851 static inline void filemap_invalidate_unlock_shared(
852 struct address_space *mapping)
854 up_read(&mapping->invalidate_lock);
857 void lock_two_nondirectories(struct inode *, struct inode*);
858 void unlock_two_nondirectories(struct inode *, struct inode*);
860 void filemap_invalidate_lock_two(struct address_space *mapping1,
861 struct address_space *mapping2);
862 void filemap_invalidate_unlock_two(struct address_space *mapping1,
863 struct address_space *mapping2);
867 * NOTE: in a 32bit arch with a preemptable kernel and
868 * an UP compile the i_size_read/write must be atomic
869 * with respect to the local cpu (unlike with preempt disabled),
870 * but they don't need to be atomic with respect to other cpus like in
871 * true SMP (so they need either to either locally disable irq around
872 * the read or for example on x86 they can be still implemented as a
873 * cmpxchg8b without the need of the lock prefix). For SMP compiles
874 * and 64bit archs it makes no difference if preempt is enabled or not.
876 static inline loff_t i_size_read(const struct inode *inode)
878 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
883 seq = read_seqcount_begin(&inode->i_size_seqcount);
884 i_size = inode->i_size;
885 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
887 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
891 i_size = inode->i_size;
895 return inode->i_size;
900 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
901 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
902 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
904 static inline void i_size_write(struct inode *inode, loff_t i_size)
906 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
908 write_seqcount_begin(&inode->i_size_seqcount);
909 inode->i_size = i_size;
910 write_seqcount_end(&inode->i_size_seqcount);
912 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
914 inode->i_size = i_size;
917 inode->i_size = i_size;
921 static inline unsigned iminor(const struct inode *inode)
923 return MINOR(inode->i_rdev);
926 static inline unsigned imajor(const struct inode *inode)
928 return MAJOR(inode->i_rdev);
932 rwlock_t lock; /* protects pid, uid, euid fields */
933 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
934 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
935 kuid_t uid, euid; /* uid/euid of process setting the owner */
936 int signum; /* posix.1b rt signal to be delivered on IO */
940 * struct file_ra_state - Track a file's readahead state.
941 * @start: Where the most recent readahead started.
942 * @size: Number of pages read in the most recent readahead.
943 * @async_size: Start next readahead when this many pages are left.
944 * @ra_pages: Maximum size of a readahead request.
945 * @mmap_miss: How many mmap accesses missed in the page cache.
946 * @prev_pos: The last byte in the most recent read request.
948 struct file_ra_state {
951 unsigned int async_size;
952 unsigned int ra_pages;
953 unsigned int mmap_miss;
958 * Check if @index falls in the readahead windows.
960 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
962 return (index >= ra->start &&
963 index < ra->start + ra->size);
968 struct llist_node fu_llist;
969 struct rcu_head fu_rcuhead;
972 struct inode *f_inode; /* cached value */
973 const struct file_operations *f_op;
976 * Protects f_ep, f_flags.
977 * Must not be taken from IRQ context.
980 enum rw_hint f_write_hint;
981 atomic_long_t f_count;
982 unsigned int f_flags;
984 struct mutex f_pos_lock;
986 struct fown_struct f_owner;
987 const struct cred *f_cred;
988 struct file_ra_state f_ra;
991 #ifdef CONFIG_SECURITY
994 /* needed for tty driver, and maybe others */
998 /* Used by fs/eventpoll.c to link all the hooks to this file */
999 struct hlist_head *f_ep;
1000 #endif /* #ifdef CONFIG_EPOLL */
1001 struct address_space *f_mapping;
1003 errseq_t f_sb_err; /* for syncfs */
1004 } __randomize_layout
1005 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
1007 struct file_handle {
1010 /* file identifier */
1011 unsigned char f_handle[];
1014 static inline struct file *get_file(struct file *f)
1016 atomic_long_inc(&f->f_count);
1019 #define get_file_rcu_many(x, cnt) \
1020 atomic_long_add_unless(&(x)->f_count, (cnt), 0)
1021 #define get_file_rcu(x) get_file_rcu_many((x), 1)
1022 #define file_count(x) atomic_long_read(&(x)->f_count)
1024 #define MAX_NON_LFS ((1UL<<31) - 1)
1026 /* Page cache limit. The filesystems should put that into their s_maxbytes
1027 limits, otherwise bad things can happen in VM. */
1028 #if BITS_PER_LONG==32
1029 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
1030 #elif BITS_PER_LONG==64
1031 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
1036 #define FL_DELEG 4 /* NFSv4 delegation */
1037 #define FL_ACCESS 8 /* not trying to lock, just looking */
1038 #define FL_EXISTS 16 /* when unlocking, test for existence */
1039 #define FL_LEASE 32 /* lease held on this file */
1040 #define FL_CLOSE 64 /* unlock on close */
1041 #define FL_SLEEP 128 /* A blocking lock */
1042 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
1043 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
1044 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
1045 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
1046 #define FL_RECLAIM 4096 /* reclaiming from a reboot server */
1048 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
1051 * Special return value from posix_lock_file() and vfs_lock_file() for
1052 * asynchronous locking.
1054 #define FILE_LOCK_DEFERRED 1
1056 /* legacy typedef, should eventually be removed */
1057 typedef void *fl_owner_t;
1061 struct file_lock_operations {
1062 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1063 void (*fl_release_private)(struct file_lock *);
1066 struct lock_manager_operations {
1067 fl_owner_t (*lm_get_owner)(fl_owner_t);
1068 void (*lm_put_owner)(fl_owner_t);
1069 void (*lm_notify)(struct file_lock *); /* unblock callback */
1070 int (*lm_grant)(struct file_lock *, int);
1071 bool (*lm_break)(struct file_lock *);
1072 int (*lm_change)(struct file_lock *, int, struct list_head *);
1073 void (*lm_setup)(struct file_lock *, void **);
1074 bool (*lm_breaker_owns_lease)(struct file_lock *);
1077 struct lock_manager {
1078 struct list_head list;
1080 * NFSv4 and up also want opens blocked during the grace period;
1087 void locks_start_grace(struct net *, struct lock_manager *);
1088 void locks_end_grace(struct lock_manager *);
1089 bool locks_in_grace(struct net *);
1090 bool opens_in_grace(struct net *);
1092 /* that will die - we need it for nfs_lock_info */
1093 #include <linux/nfs_fs_i.h>
1096 * struct file_lock represents a generic "file lock". It's used to represent
1097 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1098 * note that the same struct is used to represent both a request for a lock and
1099 * the lock itself, but the same object is never used for both.
1101 * FIXME: should we create a separate "struct lock_request" to help distinguish
1104 * The varous i_flctx lists are ordered by:
1107 * 2) lock range start
1110 * Obviously, the last two criteria only matter for POSIX locks.
1113 struct file_lock *fl_blocker; /* The lock, that is blocking us */
1114 struct list_head fl_list; /* link into file_lock_context */
1115 struct hlist_node fl_link; /* node in global lists */
1116 struct list_head fl_blocked_requests; /* list of requests with
1117 * ->fl_blocker pointing here
1119 struct list_head fl_blocked_member; /* node in
1120 * ->fl_blocker->fl_blocked_requests
1122 fl_owner_t fl_owner;
1123 unsigned int fl_flags;
1124 unsigned char fl_type;
1125 unsigned int fl_pid;
1126 int fl_link_cpu; /* what cpu's list is this on? */
1127 wait_queue_head_t fl_wait;
1128 struct file *fl_file;
1132 struct fasync_struct * fl_fasync; /* for lease break notifications */
1133 /* for lease breaks: */
1134 unsigned long fl_break_time;
1135 unsigned long fl_downgrade_time;
1137 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1138 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1140 struct nfs_lock_info nfs_fl;
1141 struct nfs4_lock_info nfs4_fl;
1143 struct list_head link; /* link in AFS vnode's pending_locks list */
1144 int state; /* state of grant or error if -ve */
1145 unsigned int debug_id;
1148 } __randomize_layout;
1150 struct file_lock_context {
1151 spinlock_t flc_lock;
1152 struct list_head flc_flock;
1153 struct list_head flc_posix;
1154 struct list_head flc_lease;
1157 /* The following constant reflects the upper bound of the file/locking space */
1159 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1160 #define OFFSET_MAX INT_LIMIT(loff_t)
1161 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1164 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1166 #define locks_inode(f) file_inode(f)
1168 #ifdef CONFIG_FILE_LOCKING
1169 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1170 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1173 #if BITS_PER_LONG == 32
1174 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1175 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1179 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1180 extern int fcntl_getlease(struct file *filp);
1183 void locks_free_lock_context(struct inode *inode);
1184 void locks_free_lock(struct file_lock *fl);
1185 extern void locks_init_lock(struct file_lock *);
1186 extern struct file_lock * locks_alloc_lock(void);
1187 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1188 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1189 extern void locks_remove_posix(struct file *, fl_owner_t);
1190 extern void locks_remove_file(struct file *);
1191 extern void locks_release_private(struct file_lock *);
1192 extern void posix_test_lock(struct file *, struct file_lock *);
1193 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1194 extern int locks_delete_block(struct file_lock *);
1195 extern int vfs_test_lock(struct file *, struct file_lock *);
1196 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1197 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1198 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1199 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1200 extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1201 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1202 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1203 extern int lease_modify(struct file_lock *, int, struct list_head *);
1205 struct notifier_block;
1206 extern int lease_register_notifier(struct notifier_block *);
1207 extern void lease_unregister_notifier(struct notifier_block *);
1209 struct files_struct;
1210 extern void show_fd_locks(struct seq_file *f,
1211 struct file *filp, struct files_struct *files);
1212 #else /* !CONFIG_FILE_LOCKING */
1213 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1214 struct flock __user *user)
1219 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1220 unsigned int cmd, struct flock __user *user)
1225 #if BITS_PER_LONG == 32
1226 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1227 struct flock64 __user *user)
1232 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1233 unsigned int cmd, struct flock64 __user *user)
1238 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1243 static inline int fcntl_getlease(struct file *filp)
1249 locks_free_lock_context(struct inode *inode)
1253 static inline void locks_init_lock(struct file_lock *fl)
1258 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1263 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1268 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1273 static inline void locks_remove_file(struct file *filp)
1278 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1283 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1284 struct file_lock *conflock)
1289 static inline int locks_delete_block(struct file_lock *waiter)
1294 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1299 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1300 struct file_lock *fl, struct file_lock *conf)
1305 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1310 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1315 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1320 static inline void lease_get_mtime(struct inode *inode,
1321 struct timespec64 *time)
1326 static inline int generic_setlease(struct file *filp, long arg,
1327 struct file_lock **flp, void **priv)
1332 static inline int vfs_setlease(struct file *filp, long arg,
1333 struct file_lock **lease, void **priv)
1338 static inline int lease_modify(struct file_lock *fl, int arg,
1339 struct list_head *dispose)
1344 struct files_struct;
1345 static inline void show_fd_locks(struct seq_file *f,
1346 struct file *filp, struct files_struct *files) {}
1347 #endif /* !CONFIG_FILE_LOCKING */
1349 static inline struct inode *file_inode(const struct file *f)
1354 static inline struct dentry *file_dentry(const struct file *file)
1356 return d_real(file->f_path.dentry, file_inode(file));
1359 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1361 return locks_lock_inode_wait(locks_inode(filp), fl);
1364 struct fasync_struct {
1368 struct fasync_struct *fa_next; /* singly linked list */
1369 struct file *fa_file;
1370 struct rcu_head fa_rcu;
1373 #define FASYNC_MAGIC 0x4601
1375 /* SMP safe fasync helpers: */
1376 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1377 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1378 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1379 extern struct fasync_struct *fasync_alloc(void);
1380 extern void fasync_free(struct fasync_struct *);
1382 /* can be called from interrupts */
1383 extern void kill_fasync(struct fasync_struct **, int, int);
1385 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1386 extern int f_setown(struct file *filp, unsigned long arg, int force);
1387 extern void f_delown(struct file *filp);
1388 extern pid_t f_getown(struct file *filp);
1389 extern int send_sigurg(struct fown_struct *fown);
1392 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1393 * represented in both.
1395 #define SB_RDONLY 1 /* Mount read-only */
1396 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1397 #define SB_NODEV 4 /* Disallow access to device special files */
1398 #define SB_NOEXEC 8 /* Disallow program execution */
1399 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1400 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1401 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1402 #define SB_NOATIME 1024 /* Do not update access times. */
1403 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1404 #define SB_SILENT 32768
1405 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1406 #define SB_INLINECRYPT (1<<17) /* Use blk-crypto for encrypted files */
1407 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1408 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1409 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1411 /* These sb flags are internal to the kernel */
1412 #define SB_SUBMOUNT (1<<26)
1413 #define SB_FORCE (1<<27)
1414 #define SB_NOSEC (1<<28)
1415 #define SB_BORN (1<<29)
1416 #define SB_ACTIVE (1<<30)
1417 #define SB_NOUSER (1<<31)
1419 /* These flags relate to encoding and casefolding */
1420 #define SB_ENC_STRICT_MODE_FL (1 << 0)
1422 #define sb_has_strict_encoding(sb) \
1423 (sb->s_encoding_flags & SB_ENC_STRICT_MODE_FL)
1429 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1430 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1431 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1432 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1433 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1436 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1437 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1438 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1439 #define SB_I_STABLE_WRITES 0x00000008 /* don't modify blks until WB is done */
1441 /* sb->s_iflags to limit user namespace mounts */
1442 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1443 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1444 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1446 #define SB_I_SKIP_SYNC 0x00000100 /* Skip superblock at global sync */
1448 /* Possible states of 'frozen' field */
1450 SB_UNFROZEN = 0, /* FS is unfrozen */
1451 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1452 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1453 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1454 * internal threads if needed) */
1455 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1458 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1461 int frozen; /* Is sb frozen? */
1462 wait_queue_head_t wait_unfrozen; /* wait for thaw */
1463 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1466 struct super_block {
1467 struct list_head s_list; /* Keep this first */
1468 dev_t s_dev; /* search index; _not_ kdev_t */
1469 unsigned char s_blocksize_bits;
1470 unsigned long s_blocksize;
1471 loff_t s_maxbytes; /* Max file size */
1472 struct file_system_type *s_type;
1473 const struct super_operations *s_op;
1474 const struct dquot_operations *dq_op;
1475 const struct quotactl_ops *s_qcop;
1476 const struct export_operations *s_export_op;
1477 unsigned long s_flags;
1478 unsigned long s_iflags; /* internal SB_I_* flags */
1479 unsigned long s_magic;
1480 struct dentry *s_root;
1481 struct rw_semaphore s_umount;
1484 #ifdef CONFIG_SECURITY
1487 const struct xattr_handler **s_xattr;
1488 #ifdef CONFIG_FS_ENCRYPTION
1489 const struct fscrypt_operations *s_cop;
1490 struct key *s_master_keys; /* master crypto keys in use */
1492 #ifdef CONFIG_FS_VERITY
1493 const struct fsverity_operations *s_vop;
1495 #ifdef CONFIG_UNICODE
1496 struct unicode_map *s_encoding;
1497 __u16 s_encoding_flags;
1499 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1500 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1501 struct block_device *s_bdev;
1502 struct backing_dev_info *s_bdi;
1503 struct mtd_info *s_mtd;
1504 struct hlist_node s_instances;
1505 unsigned int s_quota_types; /* Bitmask of supported quota types */
1506 struct quota_info s_dquot; /* Diskquota specific options */
1508 struct sb_writers s_writers;
1511 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1512 * s_fsnotify_marks together for cache efficiency. They are frequently
1513 * accessed and rarely modified.
1515 void *s_fs_info; /* Filesystem private info */
1517 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1519 /* Time limits for c/m/atime in seconds */
1520 time64_t s_time_min;
1521 time64_t s_time_max;
1522 #ifdef CONFIG_FSNOTIFY
1523 __u32 s_fsnotify_mask;
1524 struct fsnotify_mark_connector __rcu *s_fsnotify_marks;
1527 char s_id[32]; /* Informational name */
1528 uuid_t s_uuid; /* UUID */
1530 unsigned int s_max_links;
1534 * The next field is for VFS *only*. No filesystems have any business
1535 * even looking at it. You had been warned.
1537 struct mutex s_vfs_rename_mutex; /* Kludge */
1540 * Filesystem subtype. If non-empty the filesystem type field
1541 * in /proc/mounts will be "type.subtype"
1543 const char *s_subtype;
1545 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1548 * Saved pool identifier for cleancache (-1 means none)
1550 int cleancache_poolid;
1552 struct shrinker s_shrink; /* per-sb shrinker handle */
1554 /* Number of inodes with nlink == 0 but still referenced */
1555 atomic_long_t s_remove_count;
1558 * Number of inode/mount/sb objects that are being watched, note that
1559 * inodes objects are currently double-accounted.
1561 atomic_long_t s_fsnotify_connectors;
1563 /* Being remounted read-only */
1564 int s_readonly_remount;
1566 /* per-sb errseq_t for reporting writeback errors via syncfs */
1569 /* AIO completions deferred from interrupt context */
1570 struct workqueue_struct *s_dio_done_wq;
1571 struct hlist_head s_pins;
1574 * Owning user namespace and default context in which to
1575 * interpret filesystem uids, gids, quotas, device nodes,
1576 * xattrs and security labels.
1578 struct user_namespace *s_user_ns;
1581 * The list_lru structure is essentially just a pointer to a table
1582 * of per-node lru lists, each of which has its own spinlock.
1583 * There is no need to put them into separate cachelines.
1585 struct list_lru s_dentry_lru;
1586 struct list_lru s_inode_lru;
1587 struct rcu_head rcu;
1588 struct work_struct destroy_work;
1590 struct mutex s_sync_lock; /* sync serialisation lock */
1593 * Indicates how deep in a filesystem stack this SB is
1597 /* s_inode_list_lock protects s_inodes */
1598 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1599 struct list_head s_inodes; /* all inodes */
1601 spinlock_t s_inode_wblist_lock;
1602 struct list_head s_inodes_wb; /* writeback inodes */
1603 } __randomize_layout;
1605 static inline struct user_namespace *i_user_ns(const struct inode *inode)
1607 return inode->i_sb->s_user_ns;
1610 /* Helper functions so that in most cases filesystems will
1611 * not need to deal directly with kuid_t and kgid_t and can
1612 * instead deal with the raw numeric values that are stored
1613 * in the filesystem.
1615 static inline uid_t i_uid_read(const struct inode *inode)
1617 return from_kuid(i_user_ns(inode), inode->i_uid);
1620 static inline gid_t i_gid_read(const struct inode *inode)
1622 return from_kgid(i_user_ns(inode), inode->i_gid);
1625 static inline void i_uid_write(struct inode *inode, uid_t uid)
1627 inode->i_uid = make_kuid(i_user_ns(inode), uid);
1630 static inline void i_gid_write(struct inode *inode, gid_t gid)
1632 inode->i_gid = make_kgid(i_user_ns(inode), gid);
1636 * i_uid_into_mnt - map an inode's i_uid down into a mnt_userns
1637 * @mnt_userns: user namespace of the mount the inode was found from
1638 * @inode: inode to map
1640 * Return: the inode's i_uid mapped down according to @mnt_userns.
1641 * If the inode's i_uid has no mapping INVALID_UID is returned.
1643 static inline kuid_t i_uid_into_mnt(struct user_namespace *mnt_userns,
1644 const struct inode *inode)
1646 return mapped_kuid_fs(mnt_userns, i_user_ns(inode), inode->i_uid);
1650 * i_gid_into_mnt - map an inode's i_gid down into a mnt_userns
1651 * @mnt_userns: user namespace of the mount the inode was found from
1652 * @inode: inode to map
1654 * Return: the inode's i_gid mapped down according to @mnt_userns.
1655 * If the inode's i_gid has no mapping INVALID_GID is returned.
1657 static inline kgid_t i_gid_into_mnt(struct user_namespace *mnt_userns,
1658 const struct inode *inode)
1660 return mapped_kgid_fs(mnt_userns, i_user_ns(inode), inode->i_gid);
1664 * inode_fsuid_set - initialize inode's i_uid field with callers fsuid
1665 * @inode: inode to initialize
1666 * @mnt_userns: user namespace of the mount the inode was found from
1668 * Initialize the i_uid field of @inode. If the inode was found/created via
1669 * an idmapped mount map the caller's fsuid according to @mnt_users.
1671 static inline void inode_fsuid_set(struct inode *inode,
1672 struct user_namespace *mnt_userns)
1674 inode->i_uid = mapped_fsuid(mnt_userns, i_user_ns(inode));
1678 * inode_fsgid_set - initialize inode's i_gid field with callers fsgid
1679 * @inode: inode to initialize
1680 * @mnt_userns: user namespace of the mount the inode was found from
1682 * Initialize the i_gid field of @inode. If the inode was found/created via
1683 * an idmapped mount map the caller's fsgid according to @mnt_users.
1685 static inline void inode_fsgid_set(struct inode *inode,
1686 struct user_namespace *mnt_userns)
1688 inode->i_gid = mapped_fsgid(mnt_userns, i_user_ns(inode));
1692 * fsuidgid_has_mapping() - check whether caller's fsuid/fsgid is mapped
1693 * @sb: the superblock we want a mapping in
1694 * @mnt_userns: user namespace of the relevant mount
1696 * Check whether the caller's fsuid and fsgid have a valid mapping in the
1697 * s_user_ns of the superblock @sb. If the caller is on an idmapped mount map
1698 * the caller's fsuid and fsgid according to the @mnt_userns first.
1700 * Return: true if fsuid and fsgid is mapped, false if not.
1702 static inline bool fsuidgid_has_mapping(struct super_block *sb,
1703 struct user_namespace *mnt_userns)
1705 struct user_namespace *fs_userns = sb->s_user_ns;
1709 kuid = mapped_fsuid(mnt_userns, fs_userns);
1710 if (!uid_valid(kuid))
1712 kgid = mapped_fsgid(mnt_userns, fs_userns);
1713 if (!gid_valid(kgid))
1715 return kuid_has_mapping(fs_userns, kuid) &&
1716 kgid_has_mapping(fs_userns, kgid);
1719 extern struct timespec64 current_time(struct inode *inode);
1722 * Snapshotting support.
1726 * These are internal functions, please use sb_start_{write,pagefault,intwrite}
1729 static inline void __sb_end_write(struct super_block *sb, int level)
1731 percpu_up_read(sb->s_writers.rw_sem + level-1);
1734 static inline void __sb_start_write(struct super_block *sb, int level)
1736 percpu_down_read(sb->s_writers.rw_sem + level - 1);
1739 static inline bool __sb_start_write_trylock(struct super_block *sb, int level)
1741 return percpu_down_read_trylock(sb->s_writers.rw_sem + level - 1);
1744 #define __sb_writers_acquired(sb, lev) \
1745 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1746 #define __sb_writers_release(sb, lev) \
1747 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1750 * sb_end_write - drop write access to a superblock
1751 * @sb: the super we wrote to
1753 * Decrement number of writers to the filesystem. Wake up possible waiters
1754 * wanting to freeze the filesystem.
1756 static inline void sb_end_write(struct super_block *sb)
1758 __sb_end_write(sb, SB_FREEZE_WRITE);
1762 * sb_end_pagefault - drop write access to a superblock from a page fault
1763 * @sb: the super we wrote to
1765 * Decrement number of processes handling write page fault to the filesystem.
1766 * Wake up possible waiters wanting to freeze the filesystem.
1768 static inline void sb_end_pagefault(struct super_block *sb)
1770 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1774 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1775 * @sb: the super we wrote to
1777 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1778 * waiters wanting to freeze the filesystem.
1780 static inline void sb_end_intwrite(struct super_block *sb)
1782 __sb_end_write(sb, SB_FREEZE_FS);
1786 * sb_start_write - get write access to a superblock
1787 * @sb: the super we write to
1789 * When a process wants to write data or metadata to a file system (i.e. dirty
1790 * a page or an inode), it should embed the operation in a sb_start_write() -
1791 * sb_end_write() pair to get exclusion against file system freezing. This
1792 * function increments number of writers preventing freezing. If the file
1793 * system is already frozen, the function waits until the file system is
1796 * Since freeze protection behaves as a lock, users have to preserve
1797 * ordering of freeze protection and other filesystem locks. Generally,
1798 * freeze protection should be the outermost lock. In particular, we have:
1801 * -> i_mutex (write path, truncate, directory ops, ...)
1802 * -> s_umount (freeze_super, thaw_super)
1804 static inline void sb_start_write(struct super_block *sb)
1806 __sb_start_write(sb, SB_FREEZE_WRITE);
1809 static inline bool sb_start_write_trylock(struct super_block *sb)
1811 return __sb_start_write_trylock(sb, SB_FREEZE_WRITE);
1815 * sb_start_pagefault - get write access to a superblock from a page fault
1816 * @sb: the super we write to
1818 * When a process starts handling write page fault, it should embed the
1819 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1820 * exclusion against file system freezing. This is needed since the page fault
1821 * is going to dirty a page. This function increments number of running page
1822 * faults preventing freezing. If the file system is already frozen, the
1823 * function waits until the file system is thawed.
1825 * Since page fault freeze protection behaves as a lock, users have to preserve
1826 * ordering of freeze protection and other filesystem locks. It is advised to
1827 * put sb_start_pagefault() close to mmap_lock in lock ordering. Page fault
1828 * handling code implies lock dependency:
1831 * -> sb_start_pagefault
1833 static inline void sb_start_pagefault(struct super_block *sb)
1835 __sb_start_write(sb, SB_FREEZE_PAGEFAULT);
1839 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1840 * @sb: the super we write to
1842 * This is the third level of protection against filesystem freezing. It is
1843 * free for use by a filesystem. The only requirement is that it must rank
1844 * below sb_start_pagefault.
1846 * For example filesystem can call sb_start_intwrite() when starting a
1847 * transaction which somewhat eases handling of freezing for internal sources
1848 * of filesystem changes (internal fs threads, discarding preallocation on file
1851 static inline void sb_start_intwrite(struct super_block *sb)
1853 __sb_start_write(sb, SB_FREEZE_FS);
1856 static inline bool sb_start_intwrite_trylock(struct super_block *sb)
1858 return __sb_start_write_trylock(sb, SB_FREEZE_FS);
1861 bool inode_owner_or_capable(struct user_namespace *mnt_userns,
1862 const struct inode *inode);
1865 * VFS helper functions..
1867 int vfs_create(struct user_namespace *, struct inode *,
1868 struct dentry *, umode_t, bool);
1869 int vfs_mkdir(struct user_namespace *, struct inode *,
1870 struct dentry *, umode_t);
1871 int vfs_mknod(struct user_namespace *, struct inode *, struct dentry *,
1873 int vfs_symlink(struct user_namespace *, struct inode *,
1874 struct dentry *, const char *);
1875 int vfs_link(struct dentry *, struct user_namespace *, struct inode *,
1876 struct dentry *, struct inode **);
1877 int vfs_rmdir(struct user_namespace *, struct inode *, struct dentry *);
1878 int vfs_unlink(struct user_namespace *, struct inode *, struct dentry *,
1882 * struct renamedata - contains all information required for renaming
1883 * @old_mnt_userns: old user namespace of the mount the inode was found from
1884 * @old_dir: parent of source
1885 * @old_dentry: source
1886 * @new_mnt_userns: new user namespace of the mount the inode was found from
1887 * @new_dir: parent of destination
1888 * @new_dentry: destination
1889 * @delegated_inode: returns an inode needing a delegation break
1890 * @flags: rename flags
1893 struct user_namespace *old_mnt_userns;
1894 struct inode *old_dir;
1895 struct dentry *old_dentry;
1896 struct user_namespace *new_mnt_userns;
1897 struct inode *new_dir;
1898 struct dentry *new_dentry;
1899 struct inode **delegated_inode;
1901 } __randomize_layout;
1903 int vfs_rename(struct renamedata *);
1905 static inline int vfs_whiteout(struct user_namespace *mnt_userns,
1906 struct inode *dir, struct dentry *dentry)
1908 return vfs_mknod(mnt_userns, dir, dentry, S_IFCHR | WHITEOUT_MODE,
1912 struct dentry *vfs_tmpfile(struct user_namespace *mnt_userns,
1913 struct dentry *dentry, umode_t mode, int open_flag);
1915 int vfs_mkobj(struct dentry *, umode_t,
1916 int (*f)(struct dentry *, umode_t, void *),
1919 int vfs_fchown(struct file *file, uid_t user, gid_t group);
1920 int vfs_fchmod(struct file *file, umode_t mode);
1921 int vfs_utimes(const struct path *path, struct timespec64 *times);
1923 extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1925 #ifdef CONFIG_COMPAT
1926 extern long compat_ptr_ioctl(struct file *file, unsigned int cmd,
1929 #define compat_ptr_ioctl NULL
1933 * VFS file helper functions.
1935 void inode_init_owner(struct user_namespace *mnt_userns, struct inode *inode,
1936 const struct inode *dir, umode_t mode);
1937 extern bool may_open_dev(const struct path *path);
1940 * This is the "filldir" function type, used by readdir() to let
1941 * the kernel specify what kind of dirent layout it wants to have.
1942 * This allows the kernel to read directories into kernel space or
1943 * to have different dirent layouts depending on the binary type.
1946 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1949 struct dir_context {
1955 * These flags let !MMU mmap() govern direct device mapping vs immediate
1956 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1958 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1959 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1960 * NOMMU_MAP_READ: Can be mapped for reading
1961 * NOMMU_MAP_WRITE: Can be mapped for writing
1962 * NOMMU_MAP_EXEC: Can be mapped for execution
1964 #define NOMMU_MAP_COPY 0x00000001
1965 #define NOMMU_MAP_DIRECT 0x00000008
1966 #define NOMMU_MAP_READ VM_MAYREAD
1967 #define NOMMU_MAP_WRITE VM_MAYWRITE
1968 #define NOMMU_MAP_EXEC VM_MAYEXEC
1970 #define NOMMU_VMFLAGS \
1971 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1974 * These flags control the behavior of the remap_file_range function pointer.
1975 * If it is called with len == 0 that means "remap to end of source file".
1976 * See Documentation/filesystems/vfs.rst for more details about this call.
1978 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
1979 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
1981 #define REMAP_FILE_DEDUP (1 << 0)
1982 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
1985 * These flags signal that the caller is ok with altering various aspects of
1986 * the behavior of the remap operation. The changes must be made by the
1987 * implementation; the vfs remap helper functions can take advantage of them.
1988 * Flags in this category exist to preserve the quirky behavior of the hoisted
1989 * btrfs clone/dedupe ioctls.
1991 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
1995 struct file_operations {
1996 struct module *owner;
1997 loff_t (*llseek) (struct file *, loff_t, int);
1998 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1999 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
2000 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
2001 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
2002 int (*iopoll)(struct kiocb *kiocb, bool spin);
2003 int (*iterate) (struct file *, struct dir_context *);
2004 int (*iterate_shared) (struct file *, struct dir_context *);
2005 __poll_t (*poll) (struct file *, struct poll_table_struct *);
2006 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
2007 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
2008 int (*mmap) (struct file *, struct vm_area_struct *);
2009 unsigned long mmap_supported_flags;
2010 int (*open) (struct inode *, struct file *);
2011 int (*flush) (struct file *, fl_owner_t id);
2012 int (*release) (struct inode *, struct file *);
2013 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
2014 int (*fasync) (int, struct file *, int);
2015 int (*lock) (struct file *, int, struct file_lock *);
2016 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
2017 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
2018 int (*check_flags)(int);
2019 int (*flock) (struct file *, int, struct file_lock *);
2020 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
2021 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
2022 int (*setlease)(struct file *, long, struct file_lock **, void **);
2023 long (*fallocate)(struct file *file, int mode, loff_t offset,
2025 void (*show_fdinfo)(struct seq_file *m, struct file *f);
2027 unsigned (*mmap_capabilities)(struct file *);
2029 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
2030 loff_t, size_t, unsigned int);
2031 loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
2032 struct file *file_out, loff_t pos_out,
2033 loff_t len, unsigned int remap_flags);
2034 int (*fadvise)(struct file *, loff_t, loff_t, int);
2035 } __randomize_layout;
2037 struct inode_operations {
2038 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
2039 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
2040 int (*permission) (struct user_namespace *, struct inode *, int);
2041 struct posix_acl * (*get_acl)(struct inode *, int, bool);
2043 int (*readlink) (struct dentry *, char __user *,int);
2045 int (*create) (struct user_namespace *, struct inode *,struct dentry *,
2047 int (*link) (struct dentry *,struct inode *,struct dentry *);
2048 int (*unlink) (struct inode *,struct dentry *);
2049 int (*symlink) (struct user_namespace *, struct inode *,struct dentry *,
2051 int (*mkdir) (struct user_namespace *, struct inode *,struct dentry *,
2053 int (*rmdir) (struct inode *,struct dentry *);
2054 int (*mknod) (struct user_namespace *, struct inode *,struct dentry *,
2056 int (*rename) (struct user_namespace *, struct inode *, struct dentry *,
2057 struct inode *, struct dentry *, unsigned int);
2058 int (*setattr) (struct user_namespace *, struct dentry *,
2060 int (*getattr) (struct user_namespace *, const struct path *,
2061 struct kstat *, u32, unsigned int);
2062 ssize_t (*listxattr) (struct dentry *, char *, size_t);
2063 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
2065 int (*update_time)(struct inode *, struct timespec64 *, int);
2066 int (*atomic_open)(struct inode *, struct dentry *,
2067 struct file *, unsigned open_flag,
2068 umode_t create_mode);
2069 int (*tmpfile) (struct user_namespace *, struct inode *,
2070 struct dentry *, umode_t);
2071 int (*set_acl)(struct user_namespace *, struct inode *,
2072 struct posix_acl *, int);
2073 int (*fileattr_set)(struct user_namespace *mnt_userns,
2074 struct dentry *dentry, struct fileattr *fa);
2075 int (*fileattr_get)(struct dentry *dentry, struct fileattr *fa);
2076 } ____cacheline_aligned;
2078 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
2079 struct iov_iter *iter)
2081 return file->f_op->read_iter(kio, iter);
2084 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
2085 struct iov_iter *iter)
2087 return file->f_op->write_iter(kio, iter);
2090 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
2092 return file->f_op->mmap(file, vma);
2095 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
2096 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
2097 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
2098 loff_t, size_t, unsigned int);
2099 extern ssize_t generic_copy_file_range(struct file *file_in, loff_t pos_in,
2100 struct file *file_out, loff_t pos_out,
2101 size_t len, unsigned int flags);
2102 extern int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
2103 struct file *file_out, loff_t pos_out,
2105 unsigned int remap_flags);
2106 extern loff_t do_clone_file_range(struct file *file_in, loff_t pos_in,
2107 struct file *file_out, loff_t pos_out,
2108 loff_t len, unsigned int remap_flags);
2109 extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in,
2110 struct file *file_out, loff_t pos_out,
2111 loff_t len, unsigned int remap_flags);
2112 extern int vfs_dedupe_file_range(struct file *file,
2113 struct file_dedupe_range *same);
2114 extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
2115 struct file *dst_file, loff_t dst_pos,
2116 loff_t len, unsigned int remap_flags);
2119 struct super_operations {
2120 struct inode *(*alloc_inode)(struct super_block *sb);
2121 void (*destroy_inode)(struct inode *);
2122 void (*free_inode)(struct inode *);
2124 void (*dirty_inode) (struct inode *, int flags);
2125 int (*write_inode) (struct inode *, struct writeback_control *wbc);
2126 int (*drop_inode) (struct inode *);
2127 void (*evict_inode) (struct inode *);
2128 void (*put_super) (struct super_block *);
2129 int (*sync_fs)(struct super_block *sb, int wait);
2130 int (*freeze_super) (struct super_block *);
2131 int (*freeze_fs) (struct super_block *);
2132 int (*thaw_super) (struct super_block *);
2133 int (*unfreeze_fs) (struct super_block *);
2134 int (*statfs) (struct dentry *, struct kstatfs *);
2135 int (*remount_fs) (struct super_block *, int *, char *);
2136 void (*umount_begin) (struct super_block *);
2138 int (*show_options)(struct seq_file *, struct dentry *);
2139 int (*show_devname)(struct seq_file *, struct dentry *);
2140 int (*show_path)(struct seq_file *, struct dentry *);
2141 int (*show_stats)(struct seq_file *, struct dentry *);
2143 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
2144 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
2145 struct dquot **(*get_dquots)(struct inode *);
2147 long (*nr_cached_objects)(struct super_block *,
2148 struct shrink_control *);
2149 long (*free_cached_objects)(struct super_block *,
2150 struct shrink_control *);
2154 * Inode flags - they have no relation to superblock flags now
2156 #define S_SYNC (1 << 0) /* Writes are synced at once */
2157 #define S_NOATIME (1 << 1) /* Do not update access times */
2158 #define S_APPEND (1 << 2) /* Append-only file */
2159 #define S_IMMUTABLE (1 << 3) /* Immutable file */
2160 #define S_DEAD (1 << 4) /* removed, but still open directory */
2161 #define S_NOQUOTA (1 << 5) /* Inode is not counted to quota */
2162 #define S_DIRSYNC (1 << 6) /* Directory modifications are synchronous */
2163 #define S_NOCMTIME (1 << 7) /* Do not update file c/mtime */
2164 #define S_SWAPFILE (1 << 8) /* Do not truncate: swapon got its bmaps */
2165 #define S_PRIVATE (1 << 9) /* Inode is fs-internal */
2166 #define S_IMA (1 << 10) /* Inode has an associated IMA struct */
2167 #define S_AUTOMOUNT (1 << 11) /* Automount/referral quasi-directory */
2168 #define S_NOSEC (1 << 12) /* no suid or xattr security attributes */
2169 #ifdef CONFIG_FS_DAX
2170 #define S_DAX (1 << 13) /* Direct Access, avoiding the page cache */
2172 #define S_DAX 0 /* Make all the DAX code disappear */
2174 #define S_ENCRYPTED (1 << 14) /* Encrypted file (using fs/crypto/) */
2175 #define S_CASEFOLD (1 << 15) /* Casefolded file */
2176 #define S_VERITY (1 << 16) /* Verity file (using fs/verity/) */
2179 * Note that nosuid etc flags are inode-specific: setting some file-system
2180 * flags just means all the inodes inherit those flags by default. It might be
2181 * possible to override it selectively if you really wanted to with some
2182 * ioctl() that is not currently implemented.
2184 * Exception: SB_RDONLY is always applied to the entire file system.
2186 * Unfortunately, it is possible to change a filesystems flags with it mounted
2187 * with files in use. This means that all of the inodes will not have their
2188 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
2189 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
2191 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
2193 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
2194 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
2195 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
2196 ((inode)->i_flags & S_SYNC))
2197 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
2198 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
2199 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
2200 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
2201 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
2203 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
2204 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
2205 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
2206 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
2208 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
2209 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
2210 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
2211 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
2212 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
2213 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2214 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2215 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2216 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2217 #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD)
2218 #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY)
2220 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2221 (inode)->i_rdev == WHITEOUT_DEV)
2223 static inline bool HAS_UNMAPPED_ID(struct user_namespace *mnt_userns,
2224 struct inode *inode)
2226 return !uid_valid(i_uid_into_mnt(mnt_userns, inode)) ||
2227 !gid_valid(i_gid_into_mnt(mnt_userns, inode));
2230 static inline enum rw_hint file_write_hint(struct file *file)
2232 if (file->f_write_hint != WRITE_LIFE_NOT_SET)
2233 return file->f_write_hint;
2235 return file_inode(file)->i_write_hint;
2238 static inline int iocb_flags(struct file *file);
2240 static inline u16 ki_hint_validate(enum rw_hint hint)
2242 typeof(((struct kiocb *)0)->ki_hint) max_hint = -1;
2244 if (hint <= max_hint)
2249 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
2251 *kiocb = (struct kiocb) {
2253 .ki_flags = iocb_flags(filp),
2254 .ki_hint = ki_hint_validate(file_write_hint(filp)),
2255 .ki_ioprio = get_current_ioprio(),
2259 static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src,
2262 *kiocb = (struct kiocb) {
2264 .ki_flags = kiocb_src->ki_flags,
2265 .ki_hint = kiocb_src->ki_hint,
2266 .ki_ioprio = kiocb_src->ki_ioprio,
2267 .ki_pos = kiocb_src->ki_pos,
2272 * Inode state bits. Protected by inode->i_lock
2274 * Four bits determine the dirty state of the inode: I_DIRTY_SYNC,
2275 * I_DIRTY_DATASYNC, I_DIRTY_PAGES, and I_DIRTY_TIME.
2277 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2278 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2279 * various stages of removing an inode.
2281 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2283 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2284 * fdatasync() (unless I_DIRTY_DATASYNC is also set).
2285 * Timestamp updates are the usual cause.
2286 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2287 * these changes separately from I_DIRTY_SYNC so that we
2288 * don't have to write inode on fdatasync() when only
2289 * e.g. the timestamps have changed.
2290 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2291 * I_DIRTY_TIME The inode itself only has dirty timestamps, and the
2292 * lazytime mount option is enabled. We keep track of this
2293 * separately from I_DIRTY_SYNC in order to implement
2294 * lazytime. This gets cleared if I_DIRTY_INODE
2295 * (I_DIRTY_SYNC and/or I_DIRTY_DATASYNC) gets set. I.e.
2296 * either I_DIRTY_TIME *or* I_DIRTY_INODE can be set in
2297 * i_state, but not both. I_DIRTY_PAGES may still be set.
2298 * I_NEW Serves as both a mutex and completion notification.
2299 * New inodes set I_NEW. If two processes both create
2300 * the same inode, one of them will release its inode and
2301 * wait for I_NEW to be released before returning.
2302 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2303 * also cause waiting on I_NEW, without I_NEW actually
2304 * being set. find_inode() uses this to prevent returning
2305 * nearly-dead inodes.
2306 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2307 * is zero. I_FREEING must be set when I_WILL_FREE is
2309 * I_FREEING Set when inode is about to be freed but still has dirty
2310 * pages or buffers attached or the inode itself is still
2312 * I_CLEAR Added by clear_inode(). In this state the inode is
2313 * clean and can be destroyed. Inode keeps I_FREEING.
2315 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2316 * prohibited for many purposes. iget() must wait for
2317 * the inode to be completely released, then create it
2318 * anew. Other functions will just ignore such inodes,
2319 * if appropriate. I_NEW is used for waiting.
2321 * I_SYNC Writeback of inode is running. The bit is set during
2322 * data writeback, and cleared with a wakeup on the bit
2323 * address once it is done. The bit is also used to pin
2324 * the inode in memory for flusher thread.
2326 * I_REFERENCED Marks the inode as recently references on the LRU list.
2328 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2330 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2331 * synchronize competing switching instances and to tell
2332 * wb stat updates to grab the i_pages lock. See
2333 * inode_switch_wbs_work_fn() for details.
2335 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2336 * and work dirs among overlayfs mounts.
2338 * I_CREATING New object's inode in the middle of setting up.
2340 * I_DONTCACHE Evict inode as soon as it is not used anymore.
2342 * I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists.
2343 * Used to detect that mark_inode_dirty() should not move
2344 * inode between dirty lists.
2346 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2348 #define I_DIRTY_SYNC (1 << 0)
2349 #define I_DIRTY_DATASYNC (1 << 1)
2350 #define I_DIRTY_PAGES (1 << 2)
2352 #define I_NEW (1 << __I_NEW)
2353 #define I_WILL_FREE (1 << 4)
2354 #define I_FREEING (1 << 5)
2355 #define I_CLEAR (1 << 6)
2357 #define I_SYNC (1 << __I_SYNC)
2358 #define I_REFERENCED (1 << 8)
2359 #define __I_DIO_WAKEUP 9
2360 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2361 #define I_LINKABLE (1 << 10)
2362 #define I_DIRTY_TIME (1 << 11)
2363 #define I_WB_SWITCH (1 << 13)
2364 #define I_OVL_INUSE (1 << 14)
2365 #define I_CREATING (1 << 15)
2366 #define I_DONTCACHE (1 << 16)
2367 #define I_SYNC_QUEUED (1 << 17)
2369 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2370 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2371 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2373 extern void __mark_inode_dirty(struct inode *, int);
2374 static inline void mark_inode_dirty(struct inode *inode)
2376 __mark_inode_dirty(inode, I_DIRTY);
2379 static inline void mark_inode_dirty_sync(struct inode *inode)
2381 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2385 * Returns true if the given inode itself only has dirty timestamps (its pages
2386 * may still be dirty) and isn't currently being allocated or freed.
2387 * Filesystems should call this if when writing an inode when lazytime is
2388 * enabled, they want to opportunistically write the timestamps of other inodes
2389 * located very nearby on-disk, e.g. in the same inode block. This returns true
2390 * if the given inode is in need of such an opportunistic update. Requires
2391 * i_lock, or at least later re-checking under i_lock.
2393 static inline bool inode_is_dirtytime_only(struct inode *inode)
2395 return (inode->i_state & (I_DIRTY_TIME | I_NEW |
2396 I_FREEING | I_WILL_FREE)) == I_DIRTY_TIME;
2399 extern void inc_nlink(struct inode *inode);
2400 extern void drop_nlink(struct inode *inode);
2401 extern void clear_nlink(struct inode *inode);
2402 extern void set_nlink(struct inode *inode, unsigned int nlink);
2404 static inline void inode_inc_link_count(struct inode *inode)
2407 mark_inode_dirty(inode);
2410 static inline void inode_dec_link_count(struct inode *inode)
2413 mark_inode_dirty(inode);
2416 enum file_time_flags {
2423 extern bool atime_needs_update(const struct path *, struct inode *);
2424 extern void touch_atime(const struct path *);
2425 int inode_update_time(struct inode *inode, struct timespec64 *time, int flags);
2427 static inline void file_accessed(struct file *file)
2429 if (!(file->f_flags & O_NOATIME))
2430 touch_atime(&file->f_path);
2433 extern int file_modified(struct file *file);
2435 int sync_inode_metadata(struct inode *inode, int wait);
2437 struct file_system_type {
2440 #define FS_REQUIRES_DEV 1
2441 #define FS_BINARY_MOUNTDATA 2
2442 #define FS_HAS_SUBTYPE 4
2443 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2444 #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */
2445 #define FS_ALLOW_IDMAP 32 /* FS has been updated to handle vfs idmappings. */
2446 #define FS_THP_SUPPORT 8192 /* Remove once all fs converted */
2447 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2448 int (*init_fs_context)(struct fs_context *);
2449 const struct fs_parameter_spec *parameters;
2450 struct dentry *(*mount) (struct file_system_type *, int,
2451 const char *, void *);
2452 void (*kill_sb) (struct super_block *);
2453 struct module *owner;
2454 struct file_system_type * next;
2455 struct hlist_head fs_supers;
2457 struct lock_class_key s_lock_key;
2458 struct lock_class_key s_umount_key;
2459 struct lock_class_key s_vfs_rename_key;
2460 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2462 struct lock_class_key i_lock_key;
2463 struct lock_class_key i_mutex_key;
2464 struct lock_class_key invalidate_lock_key;
2465 struct lock_class_key i_mutex_dir_key;
2468 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2470 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2471 int flags, const char *dev_name, void *data,
2472 int (*fill_super)(struct super_block *, void *, int));
2473 extern struct dentry *mount_single(struct file_system_type *fs_type,
2474 int flags, void *data,
2475 int (*fill_super)(struct super_block *, void *, int));
2476 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2477 int flags, void *data,
2478 int (*fill_super)(struct super_block *, void *, int));
2479 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2480 void generic_shutdown_super(struct super_block *sb);
2481 void kill_block_super(struct super_block *sb);
2482 void kill_anon_super(struct super_block *sb);
2483 void kill_litter_super(struct super_block *sb);
2484 void deactivate_super(struct super_block *sb);
2485 void deactivate_locked_super(struct super_block *sb);
2486 int set_anon_super(struct super_block *s, void *data);
2487 int set_anon_super_fc(struct super_block *s, struct fs_context *fc);
2488 int get_anon_bdev(dev_t *);
2489 void free_anon_bdev(dev_t);
2490 struct super_block *sget_fc(struct fs_context *fc,
2491 int (*test)(struct super_block *, struct fs_context *),
2492 int (*set)(struct super_block *, struct fs_context *));
2493 struct super_block *sget(struct file_system_type *type,
2494 int (*test)(struct super_block *,void *),
2495 int (*set)(struct super_block *,void *),
2496 int flags, void *data);
2498 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2499 #define fops_get(fops) \
2500 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2501 #define fops_put(fops) \
2502 do { if (fops) module_put((fops)->owner); } while(0)
2504 * This one is to be used *ONLY* from ->open() instances.
2505 * fops must be non-NULL, pinned down *and* module dependencies
2506 * should be sufficient to pin the caller down as well.
2508 #define replace_fops(f, fops) \
2510 struct file *__file = (f); \
2511 fops_put(__file->f_op); \
2512 BUG_ON(!(__file->f_op = (fops))); \
2515 extern int register_filesystem(struct file_system_type *);
2516 extern int unregister_filesystem(struct file_system_type *);
2517 extern struct vfsmount *kern_mount(struct file_system_type *);
2518 extern void kern_unmount(struct vfsmount *mnt);
2519 extern int may_umount_tree(struct vfsmount *);
2520 extern int may_umount(struct vfsmount *);
2521 extern long do_mount(const char *, const char __user *,
2522 const char *, unsigned long, void *);
2523 extern struct vfsmount *collect_mounts(const struct path *);
2524 extern void drop_collected_mounts(struct vfsmount *);
2525 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2527 extern int vfs_statfs(const struct path *, struct kstatfs *);
2528 extern int user_statfs(const char __user *, struct kstatfs *);
2529 extern int fd_statfs(int, struct kstatfs *);
2530 extern int freeze_super(struct super_block *super);
2531 extern int thaw_super(struct super_block *super);
2532 extern bool our_mnt(struct vfsmount *mnt);
2533 extern __printf(2, 3)
2534 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2535 extern int super_setup_bdi(struct super_block *sb);
2537 extern int current_umask(void);
2539 extern void ihold(struct inode * inode);
2540 extern void iput(struct inode *);
2541 extern int generic_update_time(struct inode *, struct timespec64 *, int);
2544 extern struct kobject *fs_kobj;
2546 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2548 #ifdef CONFIG_FILE_LOCKING
2549 static inline int break_lease(struct inode *inode, unsigned int mode)
2552 * Since this check is lockless, we must ensure that any refcounts
2553 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2554 * could end up racing with tasks trying to set a new lease on this
2558 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2559 return __break_lease(inode, mode, FL_LEASE);
2563 static inline int break_deleg(struct inode *inode, unsigned int mode)
2566 * Since this check is lockless, we must ensure that any refcounts
2567 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2568 * could end up racing with tasks trying to set a new lease on this
2572 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2573 return __break_lease(inode, mode, FL_DELEG);
2577 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2581 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2582 if (ret == -EWOULDBLOCK && delegated_inode) {
2583 *delegated_inode = inode;
2589 static inline int break_deleg_wait(struct inode **delegated_inode)
2593 ret = break_deleg(*delegated_inode, O_WRONLY);
2594 iput(*delegated_inode);
2595 *delegated_inode = NULL;
2599 static inline int break_layout(struct inode *inode, bool wait)
2602 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2603 return __break_lease(inode,
2604 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2609 #else /* !CONFIG_FILE_LOCKING */
2610 static inline int break_lease(struct inode *inode, unsigned int mode)
2615 static inline int break_deleg(struct inode *inode, unsigned int mode)
2620 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2625 static inline int break_deleg_wait(struct inode **delegated_inode)
2631 static inline int break_layout(struct inode *inode, bool wait)
2636 #endif /* CONFIG_FILE_LOCKING */
2641 const char *name; /* pointer to actual string */
2642 const __user char *uptr; /* original userland pointer */
2644 struct audit_names *aname;
2647 static_assert(offsetof(struct filename, iname) % sizeof(long) == 0);
2649 static inline struct user_namespace *file_mnt_user_ns(struct file *file)
2651 return mnt_user_ns(file->f_path.mnt);
2655 * is_idmapped_mnt - check whether a mount is mapped
2656 * @mnt: the mount to check
2658 * If @mnt has an idmapping attached different from the
2659 * filesystem's idmapping then @mnt is mapped.
2661 * Return: true if mount is mapped, false if not.
2663 static inline bool is_idmapped_mnt(const struct vfsmount *mnt)
2665 return mnt_user_ns(mnt) != mnt->mnt_sb->s_user_ns;
2668 extern long vfs_truncate(const struct path *, loff_t);
2669 int do_truncate(struct user_namespace *, struct dentry *, loff_t start,
2670 unsigned int time_attrs, struct file *filp);
2671 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2673 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2675 extern struct file *file_open_name(struct filename *, int, umode_t);
2676 extern struct file *filp_open(const char *, int, umode_t);
2677 extern struct file *file_open_root(const struct path *,
2678 const char *, int, umode_t);
2679 static inline struct file *file_open_root_mnt(struct vfsmount *mnt,
2680 const char *name, int flags, umode_t mode)
2682 return file_open_root(&(struct path){.mnt = mnt, .dentry = mnt->mnt_root},
2685 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2686 extern struct file * open_with_fake_path(const struct path *, int,
2687 struct inode*, const struct cred *);
2688 static inline struct file *file_clone_open(struct file *file)
2690 return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2692 extern int filp_close(struct file *, fl_owner_t id);
2694 extern struct filename *getname_flags(const char __user *, int, int *);
2695 extern struct filename *getname_uflags(const char __user *, int);
2696 extern struct filename *getname(const char __user *);
2697 extern struct filename *getname_kernel(const char *);
2698 extern void putname(struct filename *name);
2700 extern int finish_open(struct file *file, struct dentry *dentry,
2701 int (*open)(struct inode *, struct file *));
2702 extern int finish_no_open(struct file *file, struct dentry *dentry);
2705 extern void __init vfs_caches_init_early(void);
2706 extern void __init vfs_caches_init(void);
2708 extern struct kmem_cache *names_cachep;
2710 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2711 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2713 extern struct super_block *blockdev_superblock;
2714 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2716 return IS_ENABLED(CONFIG_BLOCK) && sb == blockdev_superblock;
2719 void emergency_thaw_all(void);
2720 extern int sync_filesystem(struct super_block *);
2721 extern const struct file_operations def_blk_fops;
2722 extern const struct file_operations def_chr_fops;
2725 #define CHRDEV_MAJOR_MAX 512
2726 /* Marks the bottom of the first segment of free char majors */
2727 #define CHRDEV_MAJOR_DYN_END 234
2728 /* Marks the top and bottom of the second segment of free char majors */
2729 #define CHRDEV_MAJOR_DYN_EXT_START 511
2730 #define CHRDEV_MAJOR_DYN_EXT_END 384
2732 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2733 extern int register_chrdev_region(dev_t, unsigned, const char *);
2734 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2735 unsigned int count, const char *name,
2736 const struct file_operations *fops);
2737 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2738 unsigned int count, const char *name);
2739 extern void unregister_chrdev_region(dev_t, unsigned);
2740 extern void chrdev_show(struct seq_file *,off_t);
2742 static inline int register_chrdev(unsigned int major, const char *name,
2743 const struct file_operations *fops)
2745 return __register_chrdev(major, 0, 256, name, fops);
2748 static inline void unregister_chrdev(unsigned int major, const char *name)
2750 __unregister_chrdev(major, 0, 256, name);
2753 extern void init_special_inode(struct inode *, umode_t, dev_t);
2755 /* Invalid inode operations -- fs/bad_inode.c */
2756 extern void make_bad_inode(struct inode *);
2757 extern bool is_bad_inode(struct inode *);
2759 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2760 pgoff_t start, pgoff_t end);
2762 void invalidate_mapping_pagevec(struct address_space *mapping,
2763 pgoff_t start, pgoff_t end,
2764 unsigned long *nr_pagevec);
2766 static inline void invalidate_remote_inode(struct inode *inode)
2768 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2769 S_ISLNK(inode->i_mode))
2770 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2772 extern int invalidate_inode_pages2(struct address_space *mapping);
2773 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2774 pgoff_t start, pgoff_t end);
2775 extern int write_inode_now(struct inode *, int);
2776 extern int filemap_fdatawrite(struct address_space *);
2777 extern int filemap_flush(struct address_space *);
2778 extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2779 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2781 extern int filemap_fdatawait_range_keep_errors(struct address_space *mapping,
2782 loff_t start_byte, loff_t end_byte);
2784 static inline int filemap_fdatawait(struct address_space *mapping)
2786 return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2789 extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2791 extern bool filemap_range_needs_writeback(struct address_space *,
2792 loff_t lstart, loff_t lend);
2793 extern int filemap_write_and_wait_range(struct address_space *mapping,
2794 loff_t lstart, loff_t lend);
2795 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2796 loff_t start, loff_t end, int sync_mode);
2797 extern int filemap_fdatawrite_range(struct address_space *mapping,
2798 loff_t start, loff_t end);
2799 extern int filemap_check_errors(struct address_space *mapping);
2800 extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2801 int filemap_fdatawrite_wbc(struct address_space *mapping,
2802 struct writeback_control *wbc);
2804 static inline int filemap_write_and_wait(struct address_space *mapping)
2806 return filemap_write_and_wait_range(mapping, 0, LLONG_MAX);
2809 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2811 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2812 extern int __must_check file_write_and_wait_range(struct file *file,
2813 loff_t start, loff_t end);
2815 static inline int file_write_and_wait(struct file *file)
2817 return file_write_and_wait_range(file, 0, LLONG_MAX);
2821 * filemap_set_wb_err - set a writeback error on an address_space
2822 * @mapping: mapping in which to set writeback error
2823 * @err: error to be set in mapping
2825 * When writeback fails in some way, we must record that error so that
2826 * userspace can be informed when fsync and the like are called. We endeavor
2827 * to report errors on any file that was open at the time of the error. Some
2828 * internal callers also need to know when writeback errors have occurred.
2830 * When a writeback error occurs, most filesystems will want to call
2831 * filemap_set_wb_err to record the error in the mapping so that it will be
2832 * automatically reported whenever fsync is called on the file.
2834 static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2836 /* Fastpath for common case of no error */
2838 __filemap_set_wb_err(mapping, err);
2842 * filemap_check_wb_err - has an error occurred since the mark was sampled?
2843 * @mapping: mapping to check for writeback errors
2844 * @since: previously-sampled errseq_t
2846 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2847 * the given value was sampled.
2849 * If it has then report the latest error set, otherwise return 0.
2851 static inline int filemap_check_wb_err(struct address_space *mapping,
2854 return errseq_check(&mapping->wb_err, since);
2858 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2859 * @mapping: mapping to be sampled
2861 * Writeback errors are always reported relative to a particular sample point
2862 * in the past. This function provides those sample points.
2864 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2866 return errseq_sample(&mapping->wb_err);
2870 * file_sample_sb_err - sample the current errseq_t to test for later errors
2871 * @file: file pointer to be sampled
2873 * Grab the most current superblock-level errseq_t value for the given
2876 static inline errseq_t file_sample_sb_err(struct file *file)
2878 return errseq_sample(&file->f_path.dentry->d_sb->s_wb_err);
2881 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2883 extern int vfs_fsync(struct file *file, int datasync);
2885 extern int sync_file_range(struct file *file, loff_t offset, loff_t nbytes,
2886 unsigned int flags);
2889 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2890 * to already be updated for the write, and will return either the amount
2891 * of bytes passed in, or an error if syncing the file failed.
2893 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2895 if (iocb->ki_flags & IOCB_DSYNC) {
2896 int ret = vfs_fsync_range(iocb->ki_filp,
2897 iocb->ki_pos - count, iocb->ki_pos - 1,
2898 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2906 extern void emergency_sync(void);
2907 extern void emergency_remount(void);
2910 extern int bmap(struct inode *inode, sector_t *block);
2912 static inline int bmap(struct inode *inode, sector_t *block)
2918 int notify_change(struct user_namespace *, struct dentry *,
2919 struct iattr *, struct inode **);
2920 int inode_permission(struct user_namespace *, struct inode *, int);
2921 int generic_permission(struct user_namespace *, struct inode *, int);
2922 static inline int file_permission(struct file *file, int mask)
2924 return inode_permission(file_mnt_user_ns(file),
2925 file_inode(file), mask);
2927 static inline int path_permission(const struct path *path, int mask)
2929 return inode_permission(mnt_user_ns(path->mnt),
2930 d_inode(path->dentry), mask);
2932 int __check_sticky(struct user_namespace *mnt_userns, struct inode *dir,
2933 struct inode *inode);
2935 static inline bool execute_ok(struct inode *inode)
2937 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2940 static inline bool inode_wrong_type(const struct inode *inode, umode_t mode)
2942 return (inode->i_mode ^ mode) & S_IFMT;
2945 static inline void file_start_write(struct file *file)
2947 if (!S_ISREG(file_inode(file)->i_mode))
2949 sb_start_write(file_inode(file)->i_sb);
2952 static inline bool file_start_write_trylock(struct file *file)
2954 if (!S_ISREG(file_inode(file)->i_mode))
2956 return sb_start_write_trylock(file_inode(file)->i_sb);
2959 static inline void file_end_write(struct file *file)
2961 if (!S_ISREG(file_inode(file)->i_mode))
2963 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2967 * This is used for regular files where some users -- especially the
2968 * currently executed binary in a process, previously handled via
2969 * VM_DENYWRITE -- cannot handle concurrent write (and maybe mmap
2970 * read-write shared) accesses.
2972 * get_write_access() gets write permission for a file.
2973 * put_write_access() releases this write permission.
2974 * deny_write_access() denies write access to a file.
2975 * allow_write_access() re-enables write access to a file.
2977 * The i_writecount field of an inode can have the following values:
2978 * 0: no write access, no denied write access
2979 * < 0: (-i_writecount) users that denied write access to the file.
2980 * > 0: (i_writecount) users that have write access to the file.
2982 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2983 * except for the cases where we don't hold i_writecount yet. Then we need to
2984 * use {get,deny}_write_access() - these functions check the sign and refuse
2985 * to do the change if sign is wrong.
2987 static inline int get_write_access(struct inode *inode)
2989 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2991 static inline int deny_write_access(struct file *file)
2993 struct inode *inode = file_inode(file);
2994 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2996 static inline void put_write_access(struct inode * inode)
2998 atomic_dec(&inode->i_writecount);
3000 static inline void allow_write_access(struct file *file)
3003 atomic_inc(&file_inode(file)->i_writecount);
3005 static inline bool inode_is_open_for_write(const struct inode *inode)
3007 return atomic_read(&inode->i_writecount) > 0;
3010 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
3011 static inline void i_readcount_dec(struct inode *inode)
3013 BUG_ON(!atomic_read(&inode->i_readcount));
3014 atomic_dec(&inode->i_readcount);
3016 static inline void i_readcount_inc(struct inode *inode)
3018 atomic_inc(&inode->i_readcount);
3021 static inline void i_readcount_dec(struct inode *inode)
3025 static inline void i_readcount_inc(struct inode *inode)
3030 extern int do_pipe_flags(int *, int);
3032 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
3033 ssize_t __kernel_read(struct file *file, void *buf, size_t count, loff_t *pos);
3034 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
3035 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
3036 extern struct file * open_exec(const char *);
3038 /* fs/dcache.c -- generic fs support functions */
3039 extern bool is_subdir(struct dentry *, struct dentry *);
3040 extern bool path_is_under(const struct path *, const struct path *);
3042 extern char *file_path(struct file *, char *, int);
3044 #include <linux/err.h>
3046 /* needed for stackable file system support */
3047 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
3049 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
3051 extern int inode_init_always(struct super_block *, struct inode *);
3052 extern void inode_init_once(struct inode *);
3053 extern void address_space_init_once(struct address_space *mapping);
3054 extern struct inode * igrab(struct inode *);
3055 extern ino_t iunique(struct super_block *, ino_t);
3056 extern int inode_needs_sync(struct inode *inode);
3057 extern int generic_delete_inode(struct inode *inode);
3058 static inline int generic_drop_inode(struct inode *inode)
3060 return !inode->i_nlink || inode_unhashed(inode);
3062 extern void d_mark_dontcache(struct inode *inode);
3064 extern struct inode *ilookup5_nowait(struct super_block *sb,
3065 unsigned long hashval, int (*test)(struct inode *, void *),
3067 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
3068 int (*test)(struct inode *, void *), void *data);
3069 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
3071 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
3072 int (*test)(struct inode *, void *),
3073 int (*set)(struct inode *, void *),
3075 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
3076 extern struct inode * iget_locked(struct super_block *, unsigned long);
3077 extern struct inode *find_inode_nowait(struct super_block *,
3079 int (*match)(struct inode *,
3080 unsigned long, void *),
3082 extern struct inode *find_inode_rcu(struct super_block *, unsigned long,
3083 int (*)(struct inode *, void *), void *);
3084 extern struct inode *find_inode_by_ino_rcu(struct super_block *, unsigned long);
3085 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
3086 extern int insert_inode_locked(struct inode *);
3087 #ifdef CONFIG_DEBUG_LOCK_ALLOC
3088 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
3090 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
3092 extern void unlock_new_inode(struct inode *);
3093 extern void discard_new_inode(struct inode *);
3094 extern unsigned int get_next_ino(void);
3095 extern void evict_inodes(struct super_block *sb);
3098 * Userspace may rely on the the inode number being non-zero. For example, glibc
3099 * simply ignores files with zero i_ino in unlink() and other places.
3101 * As an additional complication, if userspace was compiled with
3102 * _FILE_OFFSET_BITS=32 on a 64-bit kernel we'll only end up reading out the
3103 * lower 32 bits, so we need to check that those aren't zero explicitly. With
3104 * _FILE_OFFSET_BITS=64, this may cause some harmless false-negatives, but
3105 * better safe than sorry.
3107 static inline bool is_zero_ino(ino_t ino)
3109 return (u32)ino == 0;
3112 extern void __iget(struct inode * inode);
3113 extern void iget_failed(struct inode *);
3114 extern void clear_inode(struct inode *);
3115 extern void __destroy_inode(struct inode *);
3116 extern struct inode *new_inode_pseudo(struct super_block *sb);
3117 extern struct inode *new_inode(struct super_block *sb);
3118 extern void free_inode_nonrcu(struct inode *inode);
3119 extern int should_remove_suid(struct dentry *);
3120 extern int file_remove_privs(struct file *);
3122 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
3123 static inline void insert_inode_hash(struct inode *inode)
3125 __insert_inode_hash(inode, inode->i_ino);
3128 extern void __remove_inode_hash(struct inode *);
3129 static inline void remove_inode_hash(struct inode *inode)
3131 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
3132 __remove_inode_hash(inode);
3135 extern void inode_sb_list_add(struct inode *inode);
3137 extern int sb_set_blocksize(struct super_block *, int);
3138 extern int sb_min_blocksize(struct super_block *, int);
3140 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
3141 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
3142 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
3143 extern int generic_write_check_limits(struct file *file, loff_t pos,
3145 extern int generic_file_rw_checks(struct file *file_in, struct file *file_out);
3146 ssize_t filemap_read(struct kiocb *iocb, struct iov_iter *to,
3147 ssize_t already_read);
3148 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
3149 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
3150 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
3151 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
3152 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
3154 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
3156 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
3158 ssize_t vfs_iocb_iter_read(struct file *file, struct kiocb *iocb,
3159 struct iov_iter *iter);
3160 ssize_t vfs_iocb_iter_write(struct file *file, struct kiocb *iocb,
3161 struct iov_iter *iter);
3164 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
3165 struct pipe_inode_info *, size_t, unsigned int);
3166 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
3167 struct file *, loff_t *, size_t, unsigned int);
3168 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
3169 struct file *out, loff_t *, size_t len, unsigned int flags);
3170 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
3171 loff_t *opos, size_t len, unsigned int flags);
3175 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3176 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3177 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
3178 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3179 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3180 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3181 int whence, loff_t maxsize, loff_t eof);
3182 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3183 int whence, loff_t size);
3184 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3185 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3186 extern int generic_file_open(struct inode * inode, struct file * filp);
3187 extern int nonseekable_open(struct inode * inode, struct file * filp);
3188 extern int stream_open(struct inode * inode, struct file * filp);
3191 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3192 loff_t file_offset);
3195 /* need locking between buffered and direct access */
3198 /* filesystem does not support filling holes */
3199 DIO_SKIP_HOLES = 0x02,
3202 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3203 struct block_device *bdev, struct iov_iter *iter,
3204 get_block_t get_block,
3205 dio_iodone_t end_io, dio_submit_t submit_io,
3208 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3209 struct inode *inode,
3210 struct iov_iter *iter,
3211 get_block_t get_block)
3213 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3214 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3218 void inode_dio_wait(struct inode *inode);
3221 * inode_dio_begin - signal start of a direct I/O requests
3222 * @inode: inode the direct I/O happens on
3224 * This is called once we've finished processing a direct I/O request,
3225 * and is used to wake up callers waiting for direct I/O to be quiesced.
3227 static inline void inode_dio_begin(struct inode *inode)
3229 atomic_inc(&inode->i_dio_count);
3233 * inode_dio_end - signal finish of a direct I/O requests
3234 * @inode: inode the direct I/O happens on
3236 * This is called once we've finished processing a direct I/O request,
3237 * and is used to wake up callers waiting for direct I/O to be quiesced.
3239 static inline void inode_dio_end(struct inode *inode)
3241 if (atomic_dec_and_test(&inode->i_dio_count))
3242 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3246 * Warn about a page cache invalidation failure diring a direct I/O write.
3248 void dio_warn_stale_pagecache(struct file *filp);
3250 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3253 extern const struct file_operations generic_ro_fops;
3255 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3257 extern int readlink_copy(char __user *, int, const char *);
3258 extern int page_readlink(struct dentry *, char __user *, int);
3259 extern const char *page_get_link(struct dentry *, struct inode *,
3260 struct delayed_call *);
3261 extern void page_put_link(void *);
3262 extern int __page_symlink(struct inode *inode, const char *symname, int len,
3264 extern int page_symlink(struct inode *inode, const char *symname, int len);
3265 extern const struct inode_operations page_symlink_inode_operations;
3266 extern void kfree_link(void *);
3267 void generic_fillattr(struct user_namespace *, struct inode *, struct kstat *);
3268 void generic_fill_statx_attr(struct inode *inode, struct kstat *stat);
3269 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3270 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3271 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3272 void inode_add_bytes(struct inode *inode, loff_t bytes);
3273 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3274 void inode_sub_bytes(struct inode *inode, loff_t bytes);
3275 static inline loff_t __inode_get_bytes(struct inode *inode)
3277 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3279 loff_t inode_get_bytes(struct inode *inode);
3280 void inode_set_bytes(struct inode *inode, loff_t bytes);
3281 const char *simple_get_link(struct dentry *, struct inode *,
3282 struct delayed_call *);
3283 extern const struct inode_operations simple_symlink_inode_operations;
3285 extern int iterate_dir(struct file *, struct dir_context *);
3287 int vfs_fstatat(int dfd, const char __user *filename, struct kstat *stat,
3289 int vfs_fstat(int fd, struct kstat *stat);
3291 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3293 return vfs_fstatat(AT_FDCWD, filename, stat, 0);
3295 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3297 return vfs_fstatat(AT_FDCWD, name, stat, AT_SYMLINK_NOFOLLOW);
3300 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3301 extern int vfs_readlink(struct dentry *, char __user *, int);
3303 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3304 extern void put_filesystem(struct file_system_type *fs);
3305 extern struct file_system_type *get_fs_type(const char *name);
3306 extern struct super_block *get_super(struct block_device *);
3307 extern struct super_block *get_active_super(struct block_device *bdev);
3308 extern void drop_super(struct super_block *sb);
3309 extern void drop_super_exclusive(struct super_block *sb);
3310 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3311 extern void iterate_supers_type(struct file_system_type *,
3312 void (*)(struct super_block *, void *), void *);
3314 extern int dcache_dir_open(struct inode *, struct file *);
3315 extern int dcache_dir_close(struct inode *, struct file *);
3316 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3317 extern int dcache_readdir(struct file *, struct dir_context *);
3318 extern int simple_setattr(struct user_namespace *, struct dentry *,
3320 extern int simple_getattr(struct user_namespace *, const struct path *,
3321 struct kstat *, u32, unsigned int);
3322 extern int simple_statfs(struct dentry *, struct kstatfs *);
3323 extern int simple_open(struct inode *inode, struct file *file);
3324 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3325 extern int simple_unlink(struct inode *, struct dentry *);
3326 extern int simple_rmdir(struct inode *, struct dentry *);
3327 extern int simple_rename(struct user_namespace *, struct inode *,
3328 struct dentry *, struct inode *, struct dentry *,
3330 extern void simple_recursive_removal(struct dentry *,
3331 void (*callback)(struct dentry *));
3332 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3333 extern void noop_invalidatepage(struct page *page, unsigned int offset,
3334 unsigned int length);
3335 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3336 extern int simple_empty(struct dentry *);
3337 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3338 loff_t pos, unsigned len, unsigned flags,
3339 struct page **pagep, void **fsdata);
3340 extern const struct address_space_operations ram_aops;
3341 extern int always_delete_dentry(const struct dentry *);
3342 extern struct inode *alloc_anon_inode(struct super_block *);
3343 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3344 extern const struct dentry_operations simple_dentry_operations;
3346 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3347 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3348 extern const struct file_operations simple_dir_operations;
3349 extern const struct inode_operations simple_dir_inode_operations;
3350 extern void make_empty_dir_inode(struct inode *inode);
3351 extern bool is_empty_dir_inode(struct inode *inode);
3352 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3353 struct dentry *d_alloc_name(struct dentry *, const char *);
3354 extern int simple_fill_super(struct super_block *, unsigned long,
3355 const struct tree_descr *);
3356 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3357 extern void simple_release_fs(struct vfsmount **mount, int *count);
3359 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3360 loff_t *ppos, const void *from, size_t available);
3361 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3362 const void __user *from, size_t count);
3364 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3365 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3367 extern int generic_check_addressable(unsigned, u64);
3369 extern void generic_set_encrypted_ci_d_ops(struct dentry *dentry);
3371 #ifdef CONFIG_MIGRATION
3372 extern int buffer_migrate_page(struct address_space *,
3373 struct page *, struct page *,
3375 extern int buffer_migrate_page_norefs(struct address_space *,
3376 struct page *, struct page *,
3379 #define buffer_migrate_page NULL
3380 #define buffer_migrate_page_norefs NULL
3383 int may_setattr(struct user_namespace *mnt_userns, struct inode *inode,
3384 unsigned int ia_valid);
3385 int setattr_prepare(struct user_namespace *, struct dentry *, struct iattr *);
3386 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3387 void setattr_copy(struct user_namespace *, struct inode *inode,
3388 const struct iattr *attr);
3390 extern int file_update_time(struct file *file);
3392 static inline bool vma_is_dax(const struct vm_area_struct *vma)
3394 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3397 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3399 struct inode *inode;
3401 if (!IS_ENABLED(CONFIG_FS_DAX) || !vma->vm_file)
3403 if (!vma_is_dax(vma))
3405 inode = file_inode(vma->vm_file);
3406 if (S_ISCHR(inode->i_mode))
3407 return false; /* device-dax */
3411 static inline int iocb_flags(struct file *file)
3414 if (file->f_flags & O_APPEND)
3416 if (file->f_flags & O_DIRECT)
3418 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3420 if (file->f_flags & __O_SYNC)
3425 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3427 int kiocb_flags = 0;
3429 /* make sure there's no overlap between RWF and private IOCB flags */
3430 BUILD_BUG_ON((__force int) RWF_SUPPORTED & IOCB_EVENTFD);
3434 if (unlikely(flags & ~RWF_SUPPORTED))
3437 if (flags & RWF_NOWAIT) {
3438 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3440 kiocb_flags |= IOCB_NOIO;
3442 kiocb_flags |= (__force int) (flags & RWF_SUPPORTED);
3443 if (flags & RWF_SYNC)
3444 kiocb_flags |= IOCB_DSYNC;
3446 ki->ki_flags |= kiocb_flags;
3450 static inline ino_t parent_ino(struct dentry *dentry)
3455 * Don't strictly need d_lock here? If the parent ino could change
3456 * then surely we'd have a deeper race in the caller?
3458 spin_lock(&dentry->d_lock);
3459 res = dentry->d_parent->d_inode->i_ino;
3460 spin_unlock(&dentry->d_lock);
3464 /* Transaction based IO helpers */
3467 * An argresp is stored in an allocated page and holds the
3468 * size of the argument or response, along with its content
3470 struct simple_transaction_argresp {
3475 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3477 char *simple_transaction_get(struct file *file, const char __user *buf,
3479 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3480 size_t size, loff_t *pos);
3481 int simple_transaction_release(struct inode *inode, struct file *file);
3483 void simple_transaction_set(struct file *file, size_t n);
3486 * simple attribute files
3488 * These attributes behave similar to those in sysfs:
3490 * Writing to an attribute immediately sets a value, an open file can be
3491 * written to multiple times.
3493 * Reading from an attribute creates a buffer from the value that might get
3494 * read with multiple read calls. When the attribute has been read
3495 * completely, no further read calls are possible until the file is opened
3498 * All attributes contain a text representation of a numeric value
3499 * that are accessed with the get() and set() functions.
3501 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3502 static int __fops ## _open(struct inode *inode, struct file *file) \
3504 __simple_attr_check_format(__fmt, 0ull); \
3505 return simple_attr_open(inode, file, __get, __set, __fmt); \
3507 static const struct file_operations __fops = { \
3508 .owner = THIS_MODULE, \
3509 .open = __fops ## _open, \
3510 .release = simple_attr_release, \
3511 .read = simple_attr_read, \
3512 .write = simple_attr_write, \
3513 .llseek = generic_file_llseek, \
3516 static inline __printf(1, 2)
3517 void __simple_attr_check_format(const char *fmt, ...)
3519 /* don't do anything, just let the compiler check the arguments; */
3522 int simple_attr_open(struct inode *inode, struct file *file,
3523 int (*get)(void *, u64 *), int (*set)(void *, u64),
3525 int simple_attr_release(struct inode *inode, struct file *file);
3526 ssize_t simple_attr_read(struct file *file, char __user *buf,
3527 size_t len, loff_t *ppos);
3528 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3529 size_t len, loff_t *ppos);
3532 int proc_nr_files(struct ctl_table *table, int write,
3533 void *buffer, size_t *lenp, loff_t *ppos);
3534 int proc_nr_dentry(struct ctl_table *table, int write,
3535 void *buffer, size_t *lenp, loff_t *ppos);
3536 int proc_nr_inodes(struct ctl_table *table, int write,
3537 void *buffer, size_t *lenp, loff_t *ppos);
3538 int __init list_bdev_fs_names(char *buf, size_t size);
3540 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3541 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3543 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3544 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3545 (flag & __FMODE_NONOTIFY)))
3547 static inline bool is_sxid(umode_t mode)
3549 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3552 static inline int check_sticky(struct user_namespace *mnt_userns,
3553 struct inode *dir, struct inode *inode)
3555 if (!(dir->i_mode & S_ISVTX))
3558 return __check_sticky(mnt_userns, dir, inode);
3561 static inline void inode_has_no_xattr(struct inode *inode)
3563 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3564 inode->i_flags |= S_NOSEC;
3567 static inline bool is_root_inode(struct inode *inode)
3569 return inode == inode->i_sb->s_root->d_inode;
3572 static inline bool dir_emit(struct dir_context *ctx,
3573 const char *name, int namelen,
3574 u64 ino, unsigned type)
3576 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3578 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3580 return ctx->actor(ctx, ".", 1, ctx->pos,
3581 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3583 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3585 return ctx->actor(ctx, "..", 2, ctx->pos,
3586 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3588 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3590 if (ctx->pos == 0) {
3591 if (!dir_emit_dot(file, ctx))
3595 if (ctx->pos == 1) {
3596 if (!dir_emit_dotdot(file, ctx))
3602 static inline bool dir_relax(struct inode *inode)
3604 inode_unlock(inode);
3606 return !IS_DEADDIR(inode);
3609 static inline bool dir_relax_shared(struct inode *inode)
3611 inode_unlock_shared(inode);
3612 inode_lock_shared(inode);
3613 return !IS_DEADDIR(inode);
3616 extern bool path_noexec(const struct path *path);
3617 extern void inode_nohighmem(struct inode *inode);
3620 extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len,
3622 extern int generic_fadvise(struct file *file, loff_t offset, loff_t len,
3626 * Flush file data before changing attributes. Caller must hold any locks
3627 * required to prevent further writes to this file until we're done setting
3630 static inline int inode_drain_writes(struct inode *inode)
3632 inode_dio_wait(inode);
3633 return filemap_write_and_wait(inode->i_mapping);
3636 #endif /* _LINUX_FS_H */