1 /* SPDX-License-Identifier: GPL-2.0 */
5 #include <linux/linkage.h>
6 #include <linux/wait_bit.h>
7 #include <linux/kdev_t.h>
8 #include <linux/dcache.h>
9 #include <linux/path.h>
10 #include <linux/stat.h>
11 #include <linux/cache.h>
12 #include <linux/list.h>
13 #include <linux/list_lru.h>
14 #include <linux/llist.h>
15 #include <linux/radix-tree.h>
16 #include <linux/xarray.h>
17 #include <linux/rbtree.h>
18 #include <linux/init.h>
19 #include <linux/pid.h>
20 #include <linux/bug.h>
21 #include <linux/mutex.h>
22 #include <linux/rwsem.h>
23 #include <linux/mm_types.h>
24 #include <linux/capability.h>
25 #include <linux/semaphore.h>
26 #include <linux/fcntl.h>
27 #include <linux/rculist_bl.h>
28 #include <linux/atomic.h>
29 #include <linux/shrinker.h>
30 #include <linux/migrate_mode.h>
31 #include <linux/uidgid.h>
32 #include <linux/lockdep.h>
33 #include <linux/percpu-rwsem.h>
34 #include <linux/workqueue.h>
35 #include <linux/delayed_call.h>
36 #include <linux/uuid.h>
37 #include <linux/errseq.h>
38 #include <linux/ioprio.h>
39 #include <linux/fs_types.h>
40 #include <linux/build_bug.h>
41 #include <linux/stddef.h>
42 #include <linux/mount.h>
43 #include <linux/cred.h>
44 #include <linux/mnt_idmapping.h>
45 #include <linux/slab.h>
47 #include <asm/byteorder.h>
48 #include <uapi/linux/fs.h>
50 struct backing_dev_info;
54 struct export_operations;
55 struct fiemap_extent_info;
60 struct pipe_inode_info;
61 struct poll_table_struct;
63 struct vm_area_struct;
66 struct swap_info_struct;
68 struct workqueue_struct;
71 struct fscrypt_operations;
73 struct fsverity_operations;
75 struct fs_parameter_spec;
78 extern void __init inode_init(void);
79 extern void __init inode_init_early(void);
80 extern void __init files_init(void);
81 extern void __init files_maxfiles_init(void);
83 extern unsigned long get_max_files(void);
84 extern unsigned int sysctl_nr_open;
86 typedef __kernel_rwf_t rwf_t;
89 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
90 struct buffer_head *bh_result, int create);
91 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
92 ssize_t bytes, void *private);
94 #define MAY_EXEC 0x00000001
95 #define MAY_WRITE 0x00000002
96 #define MAY_READ 0x00000004
97 #define MAY_APPEND 0x00000008
98 #define MAY_ACCESS 0x00000010
99 #define MAY_OPEN 0x00000020
100 #define MAY_CHDIR 0x00000040
101 /* called from RCU mode, don't block */
102 #define MAY_NOT_BLOCK 0x00000080
105 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
106 * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
109 /* file is open for reading */
110 #define FMODE_READ ((__force fmode_t)0x1)
111 /* file is open for writing */
112 #define FMODE_WRITE ((__force fmode_t)0x2)
113 /* file is seekable */
114 #define FMODE_LSEEK ((__force fmode_t)0x4)
115 /* file can be accessed using pread */
116 #define FMODE_PREAD ((__force fmode_t)0x8)
117 /* file can be accessed using pwrite */
118 #define FMODE_PWRITE ((__force fmode_t)0x10)
119 /* File is opened for execution with sys_execve / sys_uselib */
120 #define FMODE_EXEC ((__force fmode_t)0x20)
121 /* File is opened with O_NDELAY (only set for block devices) */
122 #define FMODE_NDELAY ((__force fmode_t)0x40)
123 /* File is opened with O_EXCL (only set for block devices) */
124 #define FMODE_EXCL ((__force fmode_t)0x80)
125 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
126 (specialy hack for floppy.c) */
127 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
128 /* 32bit hashes as llseek() offset (for directories) */
129 #define FMODE_32BITHASH ((__force fmode_t)0x200)
130 /* 64bit hashes as llseek() offset (for directories) */
131 #define FMODE_64BITHASH ((__force fmode_t)0x400)
134 * Don't update ctime and mtime.
136 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
137 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
139 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
141 /* Expect random access pattern */
142 #define FMODE_RANDOM ((__force fmode_t)0x1000)
144 /* File is huge (eg. /dev/mem): treat loff_t as unsigned */
145 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
147 /* File is opened with O_PATH; almost nothing can be done with it */
148 #define FMODE_PATH ((__force fmode_t)0x4000)
150 /* File needs atomic accesses to f_pos */
151 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
152 /* Write access to underlying fs */
153 #define FMODE_WRITER ((__force fmode_t)0x10000)
154 /* Has read method(s) */
155 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
156 /* Has write method(s) */
157 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
159 #define FMODE_OPENED ((__force fmode_t)0x80000)
160 #define FMODE_CREATED ((__force fmode_t)0x100000)
162 /* File is stream-like */
163 #define FMODE_STREAM ((__force fmode_t)0x200000)
165 /* File supports DIRECT IO */
166 #define FMODE_CAN_ODIRECT ((__force fmode_t)0x400000)
168 /* File was opened by fanotify and shouldn't generate fanotify events */
169 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
171 /* File is capable of returning -EAGAIN if I/O will block */
172 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
174 /* File represents mount that needs unmounting */
175 #define FMODE_NEED_UNMOUNT ((__force fmode_t)0x10000000)
177 /* File does not contribute to nr_files count */
178 #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000)
180 /* File supports async buffered reads */
181 #define FMODE_BUF_RASYNC ((__force fmode_t)0x40000000)
184 * Attribute flags. These should be or-ed together to figure out what
187 #define ATTR_MODE (1 << 0)
188 #define ATTR_UID (1 << 1)
189 #define ATTR_GID (1 << 2)
190 #define ATTR_SIZE (1 << 3)
191 #define ATTR_ATIME (1 << 4)
192 #define ATTR_MTIME (1 << 5)
193 #define ATTR_CTIME (1 << 6)
194 #define ATTR_ATIME_SET (1 << 7)
195 #define ATTR_MTIME_SET (1 << 8)
196 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
197 #define ATTR_KILL_SUID (1 << 11)
198 #define ATTR_KILL_SGID (1 << 12)
199 #define ATTR_FILE (1 << 13)
200 #define ATTR_KILL_PRIV (1 << 14)
201 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
202 #define ATTR_TIMES_SET (1 << 16)
203 #define ATTR_TOUCH (1 << 17)
206 * Whiteout is represented by a char device. The following constants define the
207 * mode and device number to use.
209 #define WHITEOUT_MODE 0
210 #define WHITEOUT_DEV 0
213 * This is the Inode Attributes structure, used for notify_change(). It
214 * uses the above definitions as flags, to know which values have changed.
215 * Also, in this manner, a Filesystem can look at only the values it cares
216 * about. Basically, these are the attributes that the VFS layer can
217 * request to change from the FS layer.
219 * Derek Atkins <warlord@MIT.EDU> 94-10-20
222 unsigned int ia_valid;
227 struct timespec64 ia_atime;
228 struct timespec64 ia_mtime;
229 struct timespec64 ia_ctime;
232 * Not an attribute, but an auxiliary info for filesystems wanting to
233 * implement an ftruncate() like method. NOTE: filesystem should
234 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
236 struct file *ia_file;
240 * Includes for diskquotas.
242 #include <linux/quota.h>
245 * Maximum number of layers of fs stack. Needs to be limited to
246 * prevent kernel stack overflow
248 #define FILESYSTEM_MAX_STACK_DEPTH 2
251 * enum positive_aop_returns - aop return codes with specific semantics
253 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
254 * completed, that the page is still locked, and
255 * should be considered active. The VM uses this hint
256 * to return the page to the active list -- it won't
257 * be a candidate for writeback again in the near
258 * future. Other callers must be careful to unlock
259 * the page if they get this return. Returned by
262 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
263 * unlocked it and the page might have been truncated.
264 * The caller should back up to acquiring a new page and
265 * trying again. The aop will be taking reasonable
266 * precautions not to livelock. If the caller held a page
267 * reference, it should drop it before retrying. Returned
270 * address_space_operation functions return these large constants to indicate
271 * special semantics to the caller. These are much larger than the bytes in a
272 * page to allow for functions that return the number of bytes operated on in a
276 enum positive_aop_returns {
277 AOP_WRITEPAGE_ACTIVATE = 0x80000,
278 AOP_TRUNCATED_PAGE = 0x80001,
282 * oh the beauties of C type declarations.
285 struct address_space;
286 struct writeback_control;
287 struct readahead_control;
290 * Write life time hint values.
291 * Stored in struct inode as u8.
294 WRITE_LIFE_NOT_SET = 0,
295 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
296 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
297 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
298 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
299 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
302 /* Match RWF_* bits to IOCB bits */
303 #define IOCB_HIPRI (__force int) RWF_HIPRI
304 #define IOCB_DSYNC (__force int) RWF_DSYNC
305 #define IOCB_SYNC (__force int) RWF_SYNC
306 #define IOCB_NOWAIT (__force int) RWF_NOWAIT
307 #define IOCB_APPEND (__force int) RWF_APPEND
309 /* non-RWF related bits - start at 16 */
310 #define IOCB_EVENTFD (1 << 16)
311 #define IOCB_DIRECT (1 << 17)
312 #define IOCB_WRITE (1 << 18)
313 /* iocb->ki_waitq is valid */
314 #define IOCB_WAITQ (1 << 19)
315 #define IOCB_NOIO (1 << 20)
316 /* can use bio alloc cache */
317 #define IOCB_ALLOC_CACHE (1 << 21)
320 struct file *ki_filp;
322 /* The 'ki_filp' pointer is shared in a union for aio */
323 randomized_struct_fields_start
326 void (*ki_complete)(struct kiocb *iocb, long ret);
329 u16 ki_ioprio; /* See linux/ioprio.h */
330 struct wait_page_queue *ki_waitq; /* for async buffered IO */
331 randomized_struct_fields_end
334 static inline bool is_sync_kiocb(struct kiocb *kiocb)
336 return kiocb->ki_complete == NULL;
339 struct address_space_operations {
340 int (*writepage)(struct page *page, struct writeback_control *wbc);
341 int (*read_folio)(struct file *, struct folio *);
343 /* Write back some dirty pages from this mapping. */
344 int (*writepages)(struct address_space *, struct writeback_control *);
346 /* Mark a folio dirty. Return true if this dirtied it */
347 bool (*dirty_folio)(struct address_space *, struct folio *);
349 void (*readahead)(struct readahead_control *);
351 int (*write_begin)(struct file *, struct address_space *mapping,
352 loff_t pos, unsigned len,
353 struct page **pagep, void **fsdata);
354 int (*write_end)(struct file *, struct address_space *mapping,
355 loff_t pos, unsigned len, unsigned copied,
356 struct page *page, void *fsdata);
358 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
359 sector_t (*bmap)(struct address_space *, sector_t);
360 void (*invalidate_folio) (struct folio *, size_t offset, size_t len);
361 bool (*release_folio)(struct folio *, gfp_t);
362 void (*free_folio)(struct folio *folio);
363 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
365 * migrate the contents of a folio to the specified target. If
366 * migrate_mode is MIGRATE_ASYNC, it must not block.
368 int (*migrate_folio)(struct address_space *, struct folio *dst,
369 struct folio *src, enum migrate_mode);
370 int (*launder_folio)(struct folio *);
371 bool (*is_partially_uptodate) (struct folio *, size_t from,
373 void (*is_dirty_writeback) (struct folio *, bool *dirty, bool *wb);
374 int (*error_remove_page)(struct address_space *, struct page *);
376 /* swapfile support */
377 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
379 void (*swap_deactivate)(struct file *file);
380 int (*swap_rw)(struct kiocb *iocb, struct iov_iter *iter);
383 extern const struct address_space_operations empty_aops;
386 * struct address_space - Contents of a cacheable, mappable object.
387 * @host: Owner, either the inode or the block_device.
388 * @i_pages: Cached pages.
389 * @invalidate_lock: Guards coherency between page cache contents and
390 * file offset->disk block mappings in the filesystem during invalidates.
391 * It is also used to block modification of page cache contents through
393 * @gfp_mask: Memory allocation flags to use for allocating pages.
394 * @i_mmap_writable: Number of VM_SHARED mappings.
395 * @nr_thps: Number of THPs in the pagecache (non-shmem only).
396 * @i_mmap: Tree of private and shared mappings.
397 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
398 * @nrpages: Number of page entries, protected by the i_pages lock.
399 * @writeback_index: Writeback starts here.
401 * @flags: Error bits and flags (AS_*).
402 * @wb_err: The most recent error which has occurred.
403 * @private_lock: For use by the owner of the address_space.
404 * @private_list: For use by the owner of the address_space.
405 * @private_data: For use by the owner of the address_space.
407 struct address_space {
409 struct xarray i_pages;
410 struct rw_semaphore invalidate_lock;
412 atomic_t i_mmap_writable;
413 #ifdef CONFIG_READ_ONLY_THP_FOR_FS
414 /* number of thp, only for non-shmem files */
417 struct rb_root_cached i_mmap;
418 struct rw_semaphore i_mmap_rwsem;
419 unsigned long nrpages;
420 pgoff_t writeback_index;
421 const struct address_space_operations *a_ops;
424 spinlock_t private_lock;
425 struct list_head private_list;
427 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
429 * On most architectures that alignment is already the case; but
430 * must be enforced here for CRIS, to let the least significant bit
431 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
434 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
435 #define PAGECACHE_TAG_DIRTY XA_MARK_0
436 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
437 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
440 * Returns true if any of the pages in the mapping are marked with the tag.
442 static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag)
444 return xa_marked(&mapping->i_pages, tag);
447 static inline void i_mmap_lock_write(struct address_space *mapping)
449 down_write(&mapping->i_mmap_rwsem);
452 static inline int i_mmap_trylock_write(struct address_space *mapping)
454 return down_write_trylock(&mapping->i_mmap_rwsem);
457 static inline void i_mmap_unlock_write(struct address_space *mapping)
459 up_write(&mapping->i_mmap_rwsem);
462 static inline int i_mmap_trylock_read(struct address_space *mapping)
464 return down_read_trylock(&mapping->i_mmap_rwsem);
467 static inline void i_mmap_lock_read(struct address_space *mapping)
469 down_read(&mapping->i_mmap_rwsem);
472 static inline void i_mmap_unlock_read(struct address_space *mapping)
474 up_read(&mapping->i_mmap_rwsem);
477 static inline void i_mmap_assert_locked(struct address_space *mapping)
479 lockdep_assert_held(&mapping->i_mmap_rwsem);
482 static inline void i_mmap_assert_write_locked(struct address_space *mapping)
484 lockdep_assert_held_write(&mapping->i_mmap_rwsem);
488 * Might pages of this file be mapped into userspace?
490 static inline int mapping_mapped(struct address_space *mapping)
492 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
496 * Might pages of this file have been modified in userspace?
497 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap
498 * marks vma as VM_SHARED if it is shared, and the file was opened for
499 * writing i.e. vma may be mprotected writable even if now readonly.
501 * If i_mmap_writable is negative, no new writable mappings are allowed. You
502 * can only deny writable mappings, if none exists right now.
504 static inline int mapping_writably_mapped(struct address_space *mapping)
506 return atomic_read(&mapping->i_mmap_writable) > 0;
509 static inline int mapping_map_writable(struct address_space *mapping)
511 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
515 static inline void mapping_unmap_writable(struct address_space *mapping)
517 atomic_dec(&mapping->i_mmap_writable);
520 static inline int mapping_deny_writable(struct address_space *mapping)
522 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
526 static inline void mapping_allow_writable(struct address_space *mapping)
528 atomic_inc(&mapping->i_mmap_writable);
532 * Use sequence counter to get consistent i_size on 32-bit processors.
534 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
535 #include <linux/seqlock.h>
536 #define __NEED_I_SIZE_ORDERED
537 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
539 #define i_size_ordered_init(inode) do { } while (0)
543 #define ACL_NOT_CACHED ((void *)(-1))
545 * ACL_DONT_CACHE is for stacked filesystems, that rely on underlying fs to
546 * cache the ACL. This also means that ->get_acl() can be called in RCU mode
547 * with the LOOKUP_RCU flag.
549 #define ACL_DONT_CACHE ((void *)(-3))
551 static inline struct posix_acl *
552 uncached_acl_sentinel(struct task_struct *task)
554 return (void *)task + 1;
558 is_uncached_acl(struct posix_acl *acl)
560 return (long)acl & 1;
563 #define IOP_FASTPERM 0x0001
564 #define IOP_LOOKUP 0x0002
565 #define IOP_NOFOLLOW 0x0004
566 #define IOP_XATTR 0x0008
567 #define IOP_DEFAULT_READLINK 0x0010
569 struct fsnotify_mark_connector;
572 * Keep mostly read-only and often accessed (especially for
573 * the RCU path lookup and 'stat' data) fields at the beginning
574 * of the 'struct inode'
578 unsigned short i_opflags;
581 unsigned int i_flags;
583 #ifdef CONFIG_FS_POSIX_ACL
584 struct posix_acl *i_acl;
585 struct posix_acl *i_default_acl;
588 const struct inode_operations *i_op;
589 struct super_block *i_sb;
590 struct address_space *i_mapping;
592 #ifdef CONFIG_SECURITY
596 /* Stat data, not accessed from path walking */
599 * Filesystems may only read i_nlink directly. They shall use the
600 * following functions for modification:
602 * (set|clear|inc|drop)_nlink
603 * inode_(inc|dec)_link_count
606 const unsigned int i_nlink;
607 unsigned int __i_nlink;
611 struct timespec64 i_atime;
612 struct timespec64 i_mtime;
613 struct timespec64 i_ctime;
614 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
615 unsigned short i_bytes;
620 #ifdef __NEED_I_SIZE_ORDERED
621 seqcount_t i_size_seqcount;
625 unsigned long i_state;
626 struct rw_semaphore i_rwsem;
628 unsigned long dirtied_when; /* jiffies of first dirtying */
629 unsigned long dirtied_time_when;
631 struct hlist_node i_hash;
632 struct list_head i_io_list; /* backing dev IO list */
633 #ifdef CONFIG_CGROUP_WRITEBACK
634 struct bdi_writeback *i_wb; /* the associated cgroup wb */
636 /* foreign inode detection, see wbc_detach_inode() */
638 u16 i_wb_frn_avg_time;
639 u16 i_wb_frn_history;
641 struct list_head i_lru; /* inode LRU list */
642 struct list_head i_sb_list;
643 struct list_head i_wb_list; /* backing dev writeback list */
645 struct hlist_head i_dentry;
646 struct rcu_head i_rcu;
648 atomic64_t i_version;
649 atomic64_t i_sequence; /* see futex */
651 atomic_t i_dio_count;
652 atomic_t i_writecount;
653 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
654 atomic_t i_readcount; /* struct files open RO */
657 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
658 void (*free_inode)(struct inode *);
660 struct file_lock_context *i_flctx;
661 struct address_space i_data;
662 struct list_head i_devices;
664 struct pipe_inode_info *i_pipe;
672 #ifdef CONFIG_FSNOTIFY
673 __u32 i_fsnotify_mask; /* all events this inode cares about */
674 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
677 #ifdef CONFIG_FS_ENCRYPTION
678 struct fscrypt_info *i_crypt_info;
681 #ifdef CONFIG_FS_VERITY
682 struct fsverity_info *i_verity_info;
685 void *i_private; /* fs or device private pointer */
686 } __randomize_layout;
688 struct timespec64 timestamp_truncate(struct timespec64 t, struct inode *inode);
690 static inline unsigned int i_blocksize(const struct inode *node)
692 return (1 << node->i_blkbits);
695 static inline int inode_unhashed(struct inode *inode)
697 return hlist_unhashed(&inode->i_hash);
701 * __mark_inode_dirty expects inodes to be hashed. Since we don't
702 * want special inodes in the fileset inode space, we make them
703 * appear hashed, but do not put on any lists. hlist_del()
704 * will work fine and require no locking.
706 static inline void inode_fake_hash(struct inode *inode)
708 hlist_add_fake(&inode->i_hash);
712 * inode->i_mutex nesting subclasses for the lock validator:
714 * 0: the object of the current VFS operation
718 * 4: second non-directory
719 * 5: second parent (when locking independent directories in rename)
721 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
722 * non-directories at once.
724 * The locking order between these classes is
725 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
727 enum inode_i_mutex_lock_class
737 static inline void inode_lock(struct inode *inode)
739 down_write(&inode->i_rwsem);
742 static inline void inode_unlock(struct inode *inode)
744 up_write(&inode->i_rwsem);
747 static inline void inode_lock_shared(struct inode *inode)
749 down_read(&inode->i_rwsem);
752 static inline void inode_unlock_shared(struct inode *inode)
754 up_read(&inode->i_rwsem);
757 static inline int inode_trylock(struct inode *inode)
759 return down_write_trylock(&inode->i_rwsem);
762 static inline int inode_trylock_shared(struct inode *inode)
764 return down_read_trylock(&inode->i_rwsem);
767 static inline int inode_is_locked(struct inode *inode)
769 return rwsem_is_locked(&inode->i_rwsem);
772 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
774 down_write_nested(&inode->i_rwsem, subclass);
777 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
779 down_read_nested(&inode->i_rwsem, subclass);
782 static inline void filemap_invalidate_lock(struct address_space *mapping)
784 down_write(&mapping->invalidate_lock);
787 static inline void filemap_invalidate_unlock(struct address_space *mapping)
789 up_write(&mapping->invalidate_lock);
792 static inline void filemap_invalidate_lock_shared(struct address_space *mapping)
794 down_read(&mapping->invalidate_lock);
797 static inline int filemap_invalidate_trylock_shared(
798 struct address_space *mapping)
800 return down_read_trylock(&mapping->invalidate_lock);
803 static inline void filemap_invalidate_unlock_shared(
804 struct address_space *mapping)
806 up_read(&mapping->invalidate_lock);
809 void lock_two_nondirectories(struct inode *, struct inode*);
810 void unlock_two_nondirectories(struct inode *, struct inode*);
812 void filemap_invalidate_lock_two(struct address_space *mapping1,
813 struct address_space *mapping2);
814 void filemap_invalidate_unlock_two(struct address_space *mapping1,
815 struct address_space *mapping2);
819 * NOTE: in a 32bit arch with a preemptable kernel and
820 * an UP compile the i_size_read/write must be atomic
821 * with respect to the local cpu (unlike with preempt disabled),
822 * but they don't need to be atomic with respect to other cpus like in
823 * true SMP (so they need either to either locally disable irq around
824 * the read or for example on x86 they can be still implemented as a
825 * cmpxchg8b without the need of the lock prefix). For SMP compiles
826 * and 64bit archs it makes no difference if preempt is enabled or not.
828 static inline loff_t i_size_read(const struct inode *inode)
830 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
835 seq = read_seqcount_begin(&inode->i_size_seqcount);
836 i_size = inode->i_size;
837 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
839 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
843 i_size = inode->i_size;
847 return inode->i_size;
852 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
853 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
854 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
856 static inline void i_size_write(struct inode *inode, loff_t i_size)
858 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
860 write_seqcount_begin(&inode->i_size_seqcount);
861 inode->i_size = i_size;
862 write_seqcount_end(&inode->i_size_seqcount);
864 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
866 inode->i_size = i_size;
869 inode->i_size = i_size;
873 static inline unsigned iminor(const struct inode *inode)
875 return MINOR(inode->i_rdev);
878 static inline unsigned imajor(const struct inode *inode)
880 return MAJOR(inode->i_rdev);
884 rwlock_t lock; /* protects pid, uid, euid fields */
885 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
886 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
887 kuid_t uid, euid; /* uid/euid of process setting the owner */
888 int signum; /* posix.1b rt signal to be delivered on IO */
892 * struct file_ra_state - Track a file's readahead state.
893 * @start: Where the most recent readahead started.
894 * @size: Number of pages read in the most recent readahead.
895 * @async_size: Numer of pages that were/are not needed immediately
896 * and so were/are genuinely "ahead". Start next readahead when
897 * the first of these pages is accessed.
898 * @ra_pages: Maximum size of a readahead request, copied from the bdi.
899 * @mmap_miss: How many mmap accesses missed in the page cache.
900 * @prev_pos: The last byte in the most recent read request.
902 * When this structure is passed to ->readahead(), the "most recent"
903 * readahead means the current readahead.
905 struct file_ra_state {
908 unsigned int async_size;
909 unsigned int ra_pages;
910 unsigned int mmap_miss;
915 * Check if @index falls in the readahead windows.
917 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
919 return (index >= ra->start &&
920 index < ra->start + ra->size);
925 struct llist_node fu_llist;
926 struct rcu_head fu_rcuhead;
929 struct inode *f_inode; /* cached value */
930 const struct file_operations *f_op;
933 * Protects f_ep, f_flags.
934 * Must not be taken from IRQ context.
937 atomic_long_t f_count;
938 unsigned int f_flags;
940 struct mutex f_pos_lock;
942 struct fown_struct f_owner;
943 const struct cred *f_cred;
944 struct file_ra_state f_ra;
947 #ifdef CONFIG_SECURITY
950 /* needed for tty driver, and maybe others */
954 /* Used by fs/eventpoll.c to link all the hooks to this file */
955 struct hlist_head *f_ep;
956 #endif /* #ifdef CONFIG_EPOLL */
957 struct address_space *f_mapping;
959 errseq_t f_sb_err; /* for syncfs */
961 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
966 /* file identifier */
967 unsigned char f_handle[];
970 static inline struct file *get_file(struct file *f)
972 atomic_long_inc(&f->f_count);
975 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
976 #define file_count(x) atomic_long_read(&(x)->f_count)
978 #define MAX_NON_LFS ((1UL<<31) - 1)
980 /* Page cache limit. The filesystems should put that into their s_maxbytes
981 limits, otherwise bad things can happen in VM. */
982 #if BITS_PER_LONG==32
983 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
984 #elif BITS_PER_LONG==64
985 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
990 #define FL_DELEG 4 /* NFSv4 delegation */
991 #define FL_ACCESS 8 /* not trying to lock, just looking */
992 #define FL_EXISTS 16 /* when unlocking, test for existence */
993 #define FL_LEASE 32 /* lease held on this file */
994 #define FL_CLOSE 64 /* unlock on close */
995 #define FL_SLEEP 128 /* A blocking lock */
996 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
997 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
998 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
999 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
1000 #define FL_RECLAIM 4096 /* reclaiming from a reboot server */
1002 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
1005 * Special return value from posix_lock_file() and vfs_lock_file() for
1006 * asynchronous locking.
1008 #define FILE_LOCK_DEFERRED 1
1010 /* legacy typedef, should eventually be removed */
1011 typedef void *fl_owner_t;
1015 struct file_lock_operations {
1016 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1017 void (*fl_release_private)(struct file_lock *);
1020 struct lock_manager_operations {
1022 fl_owner_t (*lm_get_owner)(fl_owner_t);
1023 void (*lm_put_owner)(fl_owner_t);
1024 void (*lm_notify)(struct file_lock *); /* unblock callback */
1025 int (*lm_grant)(struct file_lock *, int);
1026 bool (*lm_break)(struct file_lock *);
1027 int (*lm_change)(struct file_lock *, int, struct list_head *);
1028 void (*lm_setup)(struct file_lock *, void **);
1029 bool (*lm_breaker_owns_lease)(struct file_lock *);
1030 bool (*lm_lock_expirable)(struct file_lock *cfl);
1031 void (*lm_expire_lock)(void);
1034 struct lock_manager {
1035 struct list_head list;
1037 * NFSv4 and up also want opens blocked during the grace period;
1044 void locks_start_grace(struct net *, struct lock_manager *);
1045 void locks_end_grace(struct lock_manager *);
1046 bool locks_in_grace(struct net *);
1047 bool opens_in_grace(struct net *);
1049 /* that will die - we need it for nfs_lock_info */
1050 #include <linux/nfs_fs_i.h>
1053 * struct file_lock represents a generic "file lock". It's used to represent
1054 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1055 * note that the same struct is used to represent both a request for a lock and
1056 * the lock itself, but the same object is never used for both.
1058 * FIXME: should we create a separate "struct lock_request" to help distinguish
1061 * The varous i_flctx lists are ordered by:
1064 * 2) lock range start
1067 * Obviously, the last two criteria only matter for POSIX locks.
1070 struct file_lock *fl_blocker; /* The lock, that is blocking us */
1071 struct list_head fl_list; /* link into file_lock_context */
1072 struct hlist_node fl_link; /* node in global lists */
1073 struct list_head fl_blocked_requests; /* list of requests with
1074 * ->fl_blocker pointing here
1076 struct list_head fl_blocked_member; /* node in
1077 * ->fl_blocker->fl_blocked_requests
1079 fl_owner_t fl_owner;
1080 unsigned int fl_flags;
1081 unsigned char fl_type;
1082 unsigned int fl_pid;
1083 int fl_link_cpu; /* what cpu's list is this on? */
1084 wait_queue_head_t fl_wait;
1085 struct file *fl_file;
1089 struct fasync_struct * fl_fasync; /* for lease break notifications */
1090 /* for lease breaks: */
1091 unsigned long fl_break_time;
1092 unsigned long fl_downgrade_time;
1094 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1095 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1097 struct nfs_lock_info nfs_fl;
1098 struct nfs4_lock_info nfs4_fl;
1100 struct list_head link; /* link in AFS vnode's pending_locks list */
1101 int state; /* state of grant or error if -ve */
1102 unsigned int debug_id;
1105 } __randomize_layout;
1107 struct file_lock_context {
1108 spinlock_t flc_lock;
1109 struct list_head flc_flock;
1110 struct list_head flc_posix;
1111 struct list_head flc_lease;
1114 /* The following constant reflects the upper bound of the file/locking space */
1116 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1117 #define OFFSET_MAX INT_LIMIT(loff_t)
1118 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1121 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1123 #define locks_inode(f) file_inode(f)
1125 #ifdef CONFIG_FILE_LOCKING
1126 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1127 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1130 #if BITS_PER_LONG == 32
1131 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1132 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1136 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1137 extern int fcntl_getlease(struct file *filp);
1140 void locks_free_lock_context(struct inode *inode);
1141 void locks_free_lock(struct file_lock *fl);
1142 extern void locks_init_lock(struct file_lock *);
1143 extern struct file_lock * locks_alloc_lock(void);
1144 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1145 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1146 extern void locks_remove_posix(struct file *, fl_owner_t);
1147 extern void locks_remove_file(struct file *);
1148 extern void locks_release_private(struct file_lock *);
1149 extern void posix_test_lock(struct file *, struct file_lock *);
1150 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1151 extern int locks_delete_block(struct file_lock *);
1152 extern int vfs_test_lock(struct file *, struct file_lock *);
1153 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1154 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1155 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1156 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1157 extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1158 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1159 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1160 extern int lease_modify(struct file_lock *, int, struct list_head *);
1162 struct notifier_block;
1163 extern int lease_register_notifier(struct notifier_block *);
1164 extern void lease_unregister_notifier(struct notifier_block *);
1166 struct files_struct;
1167 extern void show_fd_locks(struct seq_file *f,
1168 struct file *filp, struct files_struct *files);
1169 extern bool locks_owner_has_blockers(struct file_lock_context *flctx,
1171 #else /* !CONFIG_FILE_LOCKING */
1172 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1173 struct flock __user *user)
1178 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1179 unsigned int cmd, struct flock __user *user)
1184 #if BITS_PER_LONG == 32
1185 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1186 struct flock64 *user)
1191 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1192 unsigned int cmd, struct flock64 *user)
1197 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1202 static inline int fcntl_getlease(struct file *filp)
1208 locks_free_lock_context(struct inode *inode)
1212 static inline void locks_init_lock(struct file_lock *fl)
1217 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1222 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1227 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1232 static inline void locks_remove_file(struct file *filp)
1237 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1242 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1243 struct file_lock *conflock)
1248 static inline int locks_delete_block(struct file_lock *waiter)
1253 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1258 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1259 struct file_lock *fl, struct file_lock *conf)
1264 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1269 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1274 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1279 static inline void lease_get_mtime(struct inode *inode,
1280 struct timespec64 *time)
1285 static inline int generic_setlease(struct file *filp, long arg,
1286 struct file_lock **flp, void **priv)
1291 static inline int vfs_setlease(struct file *filp, long arg,
1292 struct file_lock **lease, void **priv)
1297 static inline int lease_modify(struct file_lock *fl, int arg,
1298 struct list_head *dispose)
1303 struct files_struct;
1304 static inline void show_fd_locks(struct seq_file *f,
1305 struct file *filp, struct files_struct *files) {}
1306 static inline bool locks_owner_has_blockers(struct file_lock_context *flctx,
1311 #endif /* !CONFIG_FILE_LOCKING */
1313 static inline struct inode *file_inode(const struct file *f)
1318 static inline struct dentry *file_dentry(const struct file *file)
1320 return d_real(file->f_path.dentry, file_inode(file));
1323 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1325 return locks_lock_inode_wait(locks_inode(filp), fl);
1328 struct fasync_struct {
1332 struct fasync_struct *fa_next; /* singly linked list */
1333 struct file *fa_file;
1334 struct rcu_head fa_rcu;
1337 #define FASYNC_MAGIC 0x4601
1339 /* SMP safe fasync helpers: */
1340 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1341 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1342 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1343 extern struct fasync_struct *fasync_alloc(void);
1344 extern void fasync_free(struct fasync_struct *);
1346 /* can be called from interrupts */
1347 extern void kill_fasync(struct fasync_struct **, int, int);
1349 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1350 extern int f_setown(struct file *filp, unsigned long arg, int force);
1351 extern void f_delown(struct file *filp);
1352 extern pid_t f_getown(struct file *filp);
1353 extern int send_sigurg(struct fown_struct *fown);
1356 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1357 * represented in both.
1359 #define SB_RDONLY 1 /* Mount read-only */
1360 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1361 #define SB_NODEV 4 /* Disallow access to device special files */
1362 #define SB_NOEXEC 8 /* Disallow program execution */
1363 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1364 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1365 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1366 #define SB_NOATIME 1024 /* Do not update access times. */
1367 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1368 #define SB_SILENT 32768
1369 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1370 #define SB_INLINECRYPT (1<<17) /* Use blk-crypto for encrypted files */
1371 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1372 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1373 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1375 /* These sb flags are internal to the kernel */
1376 #define SB_SUBMOUNT (1<<26)
1377 #define SB_FORCE (1<<27)
1378 #define SB_NOSEC (1<<28)
1379 #define SB_BORN (1<<29)
1380 #define SB_ACTIVE (1<<30)
1381 #define SB_NOUSER (1<<31)
1383 /* These flags relate to encoding and casefolding */
1384 #define SB_ENC_STRICT_MODE_FL (1 << 0)
1386 #define sb_has_strict_encoding(sb) \
1387 (sb->s_encoding_flags & SB_ENC_STRICT_MODE_FL)
1393 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1394 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1395 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1396 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1397 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1400 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1401 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1402 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1403 #define SB_I_STABLE_WRITES 0x00000008 /* don't modify blks until WB is done */
1405 /* sb->s_iflags to limit user namespace mounts */
1406 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1407 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1408 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1410 #define SB_I_SKIP_SYNC 0x00000100 /* Skip superblock at global sync */
1411 #define SB_I_PERSB_BDI 0x00000200 /* has a per-sb bdi */
1412 #define SB_I_TS_EXPIRY_WARNED 0x00000400 /* warned about timestamp range expiry */
1414 /* Possible states of 'frozen' field */
1416 SB_UNFROZEN = 0, /* FS is unfrozen */
1417 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1418 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1419 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1420 * internal threads if needed) */
1421 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1424 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1427 int frozen; /* Is sb frozen? */
1428 wait_queue_head_t wait_unfrozen; /* wait for thaw */
1429 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1432 struct super_block {
1433 struct list_head s_list; /* Keep this first */
1434 dev_t s_dev; /* search index; _not_ kdev_t */
1435 unsigned char s_blocksize_bits;
1436 unsigned long s_blocksize;
1437 loff_t s_maxbytes; /* Max file size */
1438 struct file_system_type *s_type;
1439 const struct super_operations *s_op;
1440 const struct dquot_operations *dq_op;
1441 const struct quotactl_ops *s_qcop;
1442 const struct export_operations *s_export_op;
1443 unsigned long s_flags;
1444 unsigned long s_iflags; /* internal SB_I_* flags */
1445 unsigned long s_magic;
1446 struct dentry *s_root;
1447 struct rw_semaphore s_umount;
1450 #ifdef CONFIG_SECURITY
1453 const struct xattr_handler **s_xattr;
1454 #ifdef CONFIG_FS_ENCRYPTION
1455 const struct fscrypt_operations *s_cop;
1456 struct key *s_master_keys; /* master crypto keys in use */
1458 #ifdef CONFIG_FS_VERITY
1459 const struct fsverity_operations *s_vop;
1461 #if IS_ENABLED(CONFIG_UNICODE)
1462 struct unicode_map *s_encoding;
1463 __u16 s_encoding_flags;
1465 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1466 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1467 struct block_device *s_bdev;
1468 struct backing_dev_info *s_bdi;
1469 struct mtd_info *s_mtd;
1470 struct hlist_node s_instances;
1471 unsigned int s_quota_types; /* Bitmask of supported quota types */
1472 struct quota_info s_dquot; /* Diskquota specific options */
1474 struct sb_writers s_writers;
1477 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1478 * s_fsnotify_marks together for cache efficiency. They are frequently
1479 * accessed and rarely modified.
1481 void *s_fs_info; /* Filesystem private info */
1483 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1485 /* Time limits for c/m/atime in seconds */
1486 time64_t s_time_min;
1487 time64_t s_time_max;
1488 #ifdef CONFIG_FSNOTIFY
1489 __u32 s_fsnotify_mask;
1490 struct fsnotify_mark_connector __rcu *s_fsnotify_marks;
1493 char s_id[32]; /* Informational name */
1494 uuid_t s_uuid; /* UUID */
1496 unsigned int s_max_links;
1500 * The next field is for VFS *only*. No filesystems have any business
1501 * even looking at it. You had been warned.
1503 struct mutex s_vfs_rename_mutex; /* Kludge */
1506 * Filesystem subtype. If non-empty the filesystem type field
1507 * in /proc/mounts will be "type.subtype"
1509 const char *s_subtype;
1511 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1513 struct shrinker s_shrink; /* per-sb shrinker handle */
1515 /* Number of inodes with nlink == 0 but still referenced */
1516 atomic_long_t s_remove_count;
1519 * Number of inode/mount/sb objects that are being watched, note that
1520 * inodes objects are currently double-accounted.
1522 atomic_long_t s_fsnotify_connectors;
1524 /* Being remounted read-only */
1525 int s_readonly_remount;
1527 /* per-sb errseq_t for reporting writeback errors via syncfs */
1530 /* AIO completions deferred from interrupt context */
1531 struct workqueue_struct *s_dio_done_wq;
1532 struct hlist_head s_pins;
1535 * Owning user namespace and default context in which to
1536 * interpret filesystem uids, gids, quotas, device nodes,
1537 * xattrs and security labels.
1539 struct user_namespace *s_user_ns;
1542 * The list_lru structure is essentially just a pointer to a table
1543 * of per-node lru lists, each of which has its own spinlock.
1544 * There is no need to put them into separate cachelines.
1546 struct list_lru s_dentry_lru;
1547 struct list_lru s_inode_lru;
1548 struct rcu_head rcu;
1549 struct work_struct destroy_work;
1551 struct mutex s_sync_lock; /* sync serialisation lock */
1554 * Indicates how deep in a filesystem stack this SB is
1558 /* s_inode_list_lock protects s_inodes */
1559 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1560 struct list_head s_inodes; /* all inodes */
1562 spinlock_t s_inode_wblist_lock;
1563 struct list_head s_inodes_wb; /* writeback inodes */
1564 } __randomize_layout;
1566 static inline struct user_namespace *i_user_ns(const struct inode *inode)
1568 return inode->i_sb->s_user_ns;
1571 /* Helper functions so that in most cases filesystems will
1572 * not need to deal directly with kuid_t and kgid_t and can
1573 * instead deal with the raw numeric values that are stored
1574 * in the filesystem.
1576 static inline uid_t i_uid_read(const struct inode *inode)
1578 return from_kuid(i_user_ns(inode), inode->i_uid);
1581 static inline gid_t i_gid_read(const struct inode *inode)
1583 return from_kgid(i_user_ns(inode), inode->i_gid);
1586 static inline void i_uid_write(struct inode *inode, uid_t uid)
1588 inode->i_uid = make_kuid(i_user_ns(inode), uid);
1591 static inline void i_gid_write(struct inode *inode, gid_t gid)
1593 inode->i_gid = make_kgid(i_user_ns(inode), gid);
1597 * i_uid_into_mnt - map an inode's i_uid down into a mnt_userns
1598 * @mnt_userns: user namespace of the mount the inode was found from
1599 * @inode: inode to map
1601 * Return: the inode's i_uid mapped down according to @mnt_userns.
1602 * If the inode's i_uid has no mapping INVALID_UID is returned.
1604 static inline kuid_t i_uid_into_mnt(struct user_namespace *mnt_userns,
1605 const struct inode *inode)
1607 return mapped_kuid_fs(mnt_userns, i_user_ns(inode), inode->i_uid);
1611 * i_gid_into_mnt - map an inode's i_gid down into a mnt_userns
1612 * @mnt_userns: user namespace of the mount the inode was found from
1613 * @inode: inode to map
1615 * Return: the inode's i_gid mapped down according to @mnt_userns.
1616 * If the inode's i_gid has no mapping INVALID_GID is returned.
1618 static inline kgid_t i_gid_into_mnt(struct user_namespace *mnt_userns,
1619 const struct inode *inode)
1621 return mapped_kgid_fs(mnt_userns, i_user_ns(inode), inode->i_gid);
1625 * inode_fsuid_set - initialize inode's i_uid field with callers fsuid
1626 * @inode: inode to initialize
1627 * @mnt_userns: user namespace of the mount the inode was found from
1629 * Initialize the i_uid field of @inode. If the inode was found/created via
1630 * an idmapped mount map the caller's fsuid according to @mnt_users.
1632 static inline void inode_fsuid_set(struct inode *inode,
1633 struct user_namespace *mnt_userns)
1635 inode->i_uid = mapped_fsuid(mnt_userns, i_user_ns(inode));
1639 * inode_fsgid_set - initialize inode's i_gid field with callers fsgid
1640 * @inode: inode to initialize
1641 * @mnt_userns: user namespace of the mount the inode was found from
1643 * Initialize the i_gid field of @inode. If the inode was found/created via
1644 * an idmapped mount map the caller's fsgid according to @mnt_users.
1646 static inline void inode_fsgid_set(struct inode *inode,
1647 struct user_namespace *mnt_userns)
1649 inode->i_gid = mapped_fsgid(mnt_userns, i_user_ns(inode));
1653 * fsuidgid_has_mapping() - check whether caller's fsuid/fsgid is mapped
1654 * @sb: the superblock we want a mapping in
1655 * @mnt_userns: user namespace of the relevant mount
1657 * Check whether the caller's fsuid and fsgid have a valid mapping in the
1658 * s_user_ns of the superblock @sb. If the caller is on an idmapped mount map
1659 * the caller's fsuid and fsgid according to the @mnt_userns first.
1661 * Return: true if fsuid and fsgid is mapped, false if not.
1663 static inline bool fsuidgid_has_mapping(struct super_block *sb,
1664 struct user_namespace *mnt_userns)
1666 struct user_namespace *fs_userns = sb->s_user_ns;
1670 kuid = mapped_fsuid(mnt_userns, fs_userns);
1671 if (!uid_valid(kuid))
1673 kgid = mapped_fsgid(mnt_userns, fs_userns);
1674 if (!gid_valid(kgid))
1676 return kuid_has_mapping(fs_userns, kuid) &&
1677 kgid_has_mapping(fs_userns, kgid);
1680 extern struct timespec64 current_time(struct inode *inode);
1683 * Snapshotting support.
1687 * These are internal functions, please use sb_start_{write,pagefault,intwrite}
1690 static inline void __sb_end_write(struct super_block *sb, int level)
1692 percpu_up_read(sb->s_writers.rw_sem + level-1);
1695 static inline void __sb_start_write(struct super_block *sb, int level)
1697 percpu_down_read(sb->s_writers.rw_sem + level - 1);
1700 static inline bool __sb_start_write_trylock(struct super_block *sb, int level)
1702 return percpu_down_read_trylock(sb->s_writers.rw_sem + level - 1);
1705 #define __sb_writers_acquired(sb, lev) \
1706 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1707 #define __sb_writers_release(sb, lev) \
1708 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1710 static inline bool sb_write_started(const struct super_block *sb)
1712 return lockdep_is_held_type(sb->s_writers.rw_sem + SB_FREEZE_WRITE - 1, 1);
1716 * sb_end_write - drop write access to a superblock
1717 * @sb: the super we wrote to
1719 * Decrement number of writers to the filesystem. Wake up possible waiters
1720 * wanting to freeze the filesystem.
1722 static inline void sb_end_write(struct super_block *sb)
1724 __sb_end_write(sb, SB_FREEZE_WRITE);
1728 * sb_end_pagefault - drop write access to a superblock from a page fault
1729 * @sb: the super we wrote to
1731 * Decrement number of processes handling write page fault to the filesystem.
1732 * Wake up possible waiters wanting to freeze the filesystem.
1734 static inline void sb_end_pagefault(struct super_block *sb)
1736 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1740 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1741 * @sb: the super we wrote to
1743 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1744 * waiters wanting to freeze the filesystem.
1746 static inline void sb_end_intwrite(struct super_block *sb)
1748 __sb_end_write(sb, SB_FREEZE_FS);
1752 * sb_start_write - get write access to a superblock
1753 * @sb: the super we write to
1755 * When a process wants to write data or metadata to a file system (i.e. dirty
1756 * a page or an inode), it should embed the operation in a sb_start_write() -
1757 * sb_end_write() pair to get exclusion against file system freezing. This
1758 * function increments number of writers preventing freezing. If the file
1759 * system is already frozen, the function waits until the file system is
1762 * Since freeze protection behaves as a lock, users have to preserve
1763 * ordering of freeze protection and other filesystem locks. Generally,
1764 * freeze protection should be the outermost lock. In particular, we have:
1767 * -> i_mutex (write path, truncate, directory ops, ...)
1768 * -> s_umount (freeze_super, thaw_super)
1770 static inline void sb_start_write(struct super_block *sb)
1772 __sb_start_write(sb, SB_FREEZE_WRITE);
1775 static inline bool sb_start_write_trylock(struct super_block *sb)
1777 return __sb_start_write_trylock(sb, SB_FREEZE_WRITE);
1781 * sb_start_pagefault - get write access to a superblock from a page fault
1782 * @sb: the super we write to
1784 * When a process starts handling write page fault, it should embed the
1785 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1786 * exclusion against file system freezing. This is needed since the page fault
1787 * is going to dirty a page. This function increments number of running page
1788 * faults preventing freezing. If the file system is already frozen, the
1789 * function waits until the file system is thawed.
1791 * Since page fault freeze protection behaves as a lock, users have to preserve
1792 * ordering of freeze protection and other filesystem locks. It is advised to
1793 * put sb_start_pagefault() close to mmap_lock in lock ordering. Page fault
1794 * handling code implies lock dependency:
1797 * -> sb_start_pagefault
1799 static inline void sb_start_pagefault(struct super_block *sb)
1801 __sb_start_write(sb, SB_FREEZE_PAGEFAULT);
1805 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1806 * @sb: the super we write to
1808 * This is the third level of protection against filesystem freezing. It is
1809 * free for use by a filesystem. The only requirement is that it must rank
1810 * below sb_start_pagefault.
1812 * For example filesystem can call sb_start_intwrite() when starting a
1813 * transaction which somewhat eases handling of freezing for internal sources
1814 * of filesystem changes (internal fs threads, discarding preallocation on file
1817 static inline void sb_start_intwrite(struct super_block *sb)
1819 __sb_start_write(sb, SB_FREEZE_FS);
1822 static inline bool sb_start_intwrite_trylock(struct super_block *sb)
1824 return __sb_start_write_trylock(sb, SB_FREEZE_FS);
1827 bool inode_owner_or_capable(struct user_namespace *mnt_userns,
1828 const struct inode *inode);
1831 * VFS helper functions..
1833 int vfs_create(struct user_namespace *, struct inode *,
1834 struct dentry *, umode_t, bool);
1835 int vfs_mkdir(struct user_namespace *, struct inode *,
1836 struct dentry *, umode_t);
1837 int vfs_mknod(struct user_namespace *, struct inode *, struct dentry *,
1839 int vfs_symlink(struct user_namespace *, struct inode *,
1840 struct dentry *, const char *);
1841 int vfs_link(struct dentry *, struct user_namespace *, struct inode *,
1842 struct dentry *, struct inode **);
1843 int vfs_rmdir(struct user_namespace *, struct inode *, struct dentry *);
1844 int vfs_unlink(struct user_namespace *, struct inode *, struct dentry *,
1848 * struct renamedata - contains all information required for renaming
1849 * @old_mnt_userns: old user namespace of the mount the inode was found from
1850 * @old_dir: parent of source
1851 * @old_dentry: source
1852 * @new_mnt_userns: new user namespace of the mount the inode was found from
1853 * @new_dir: parent of destination
1854 * @new_dentry: destination
1855 * @delegated_inode: returns an inode needing a delegation break
1856 * @flags: rename flags
1859 struct user_namespace *old_mnt_userns;
1860 struct inode *old_dir;
1861 struct dentry *old_dentry;
1862 struct user_namespace *new_mnt_userns;
1863 struct inode *new_dir;
1864 struct dentry *new_dentry;
1865 struct inode **delegated_inode;
1867 } __randomize_layout;
1869 int vfs_rename(struct renamedata *);
1871 static inline int vfs_whiteout(struct user_namespace *mnt_userns,
1872 struct inode *dir, struct dentry *dentry)
1874 return vfs_mknod(mnt_userns, dir, dentry, S_IFCHR | WHITEOUT_MODE,
1878 struct dentry *vfs_tmpfile(struct user_namespace *mnt_userns,
1879 struct dentry *dentry, umode_t mode, int open_flag);
1881 int vfs_mkobj(struct dentry *, umode_t,
1882 int (*f)(struct dentry *, umode_t, void *),
1885 int vfs_fchown(struct file *file, uid_t user, gid_t group);
1886 int vfs_fchmod(struct file *file, umode_t mode);
1887 int vfs_utimes(const struct path *path, struct timespec64 *times);
1889 extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1891 #ifdef CONFIG_COMPAT
1892 extern long compat_ptr_ioctl(struct file *file, unsigned int cmd,
1895 #define compat_ptr_ioctl NULL
1899 * VFS file helper functions.
1901 void inode_init_owner(struct user_namespace *mnt_userns, struct inode *inode,
1902 const struct inode *dir, umode_t mode);
1903 extern bool may_open_dev(const struct path *path);
1906 * This is the "filldir" function type, used by readdir() to let
1907 * the kernel specify what kind of dirent layout it wants to have.
1908 * This allows the kernel to read directories into kernel space or
1909 * to have different dirent layouts depending on the binary type.
1912 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1915 struct dir_context {
1921 * These flags let !MMU mmap() govern direct device mapping vs immediate
1922 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1924 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1925 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1926 * NOMMU_MAP_READ: Can be mapped for reading
1927 * NOMMU_MAP_WRITE: Can be mapped for writing
1928 * NOMMU_MAP_EXEC: Can be mapped for execution
1930 #define NOMMU_MAP_COPY 0x00000001
1931 #define NOMMU_MAP_DIRECT 0x00000008
1932 #define NOMMU_MAP_READ VM_MAYREAD
1933 #define NOMMU_MAP_WRITE VM_MAYWRITE
1934 #define NOMMU_MAP_EXEC VM_MAYEXEC
1936 #define NOMMU_VMFLAGS \
1937 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1940 * These flags control the behavior of the remap_file_range function pointer.
1941 * If it is called with len == 0 that means "remap to end of source file".
1942 * See Documentation/filesystems/vfs.rst for more details about this call.
1944 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
1945 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
1947 #define REMAP_FILE_DEDUP (1 << 0)
1948 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
1951 * These flags signal that the caller is ok with altering various aspects of
1952 * the behavior of the remap operation. The changes must be made by the
1953 * implementation; the vfs remap helper functions can take advantage of them.
1954 * Flags in this category exist to preserve the quirky behavior of the hoisted
1955 * btrfs clone/dedupe ioctls.
1957 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
1960 struct io_uring_cmd;
1962 struct file_operations {
1963 struct module *owner;
1964 loff_t (*llseek) (struct file *, loff_t, int);
1965 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1966 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1967 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1968 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1969 int (*iopoll)(struct kiocb *kiocb, struct io_comp_batch *,
1970 unsigned int flags);
1971 int (*iterate) (struct file *, struct dir_context *);
1972 int (*iterate_shared) (struct file *, struct dir_context *);
1973 __poll_t (*poll) (struct file *, struct poll_table_struct *);
1974 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1975 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1976 int (*mmap) (struct file *, struct vm_area_struct *);
1977 unsigned long mmap_supported_flags;
1978 int (*open) (struct inode *, struct file *);
1979 int (*flush) (struct file *, fl_owner_t id);
1980 int (*release) (struct inode *, struct file *);
1981 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1982 int (*fasync) (int, struct file *, int);
1983 int (*lock) (struct file *, int, struct file_lock *);
1984 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1985 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1986 int (*check_flags)(int);
1987 int (*flock) (struct file *, int, struct file_lock *);
1988 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1989 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1990 int (*setlease)(struct file *, long, struct file_lock **, void **);
1991 long (*fallocate)(struct file *file, int mode, loff_t offset,
1993 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1995 unsigned (*mmap_capabilities)(struct file *);
1997 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1998 loff_t, size_t, unsigned int);
1999 loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
2000 struct file *file_out, loff_t pos_out,
2001 loff_t len, unsigned int remap_flags);
2002 int (*fadvise)(struct file *, loff_t, loff_t, int);
2003 int (*uring_cmd)(struct io_uring_cmd *ioucmd, unsigned int issue_flags);
2004 } __randomize_layout;
2006 struct inode_operations {
2007 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
2008 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
2009 int (*permission) (struct user_namespace *, struct inode *, int);
2010 struct posix_acl * (*get_acl)(struct inode *, int, bool);
2012 int (*readlink) (struct dentry *, char __user *,int);
2014 int (*create) (struct user_namespace *, struct inode *,struct dentry *,
2016 int (*link) (struct dentry *,struct inode *,struct dentry *);
2017 int (*unlink) (struct inode *,struct dentry *);
2018 int (*symlink) (struct user_namespace *, struct inode *,struct dentry *,
2020 int (*mkdir) (struct user_namespace *, struct inode *,struct dentry *,
2022 int (*rmdir) (struct inode *,struct dentry *);
2023 int (*mknod) (struct user_namespace *, struct inode *,struct dentry *,
2025 int (*rename) (struct user_namespace *, struct inode *, struct dentry *,
2026 struct inode *, struct dentry *, unsigned int);
2027 int (*setattr) (struct user_namespace *, struct dentry *,
2029 int (*getattr) (struct user_namespace *, const struct path *,
2030 struct kstat *, u32, unsigned int);
2031 ssize_t (*listxattr) (struct dentry *, char *, size_t);
2032 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
2034 int (*update_time)(struct inode *, struct timespec64 *, int);
2035 int (*atomic_open)(struct inode *, struct dentry *,
2036 struct file *, unsigned open_flag,
2037 umode_t create_mode);
2038 int (*tmpfile) (struct user_namespace *, struct inode *,
2039 struct dentry *, umode_t);
2040 int (*set_acl)(struct user_namespace *, struct inode *,
2041 struct posix_acl *, int);
2042 int (*fileattr_set)(struct user_namespace *mnt_userns,
2043 struct dentry *dentry, struct fileattr *fa);
2044 int (*fileattr_get)(struct dentry *dentry, struct fileattr *fa);
2045 } ____cacheline_aligned;
2047 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
2048 struct iov_iter *iter)
2050 return file->f_op->read_iter(kio, iter);
2053 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
2054 struct iov_iter *iter)
2056 return file->f_op->write_iter(kio, iter);
2059 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
2061 return file->f_op->mmap(file, vma);
2064 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
2065 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
2066 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
2067 loff_t, size_t, unsigned int);
2068 extern ssize_t generic_copy_file_range(struct file *file_in, loff_t pos_in,
2069 struct file *file_out, loff_t pos_out,
2070 size_t len, unsigned int flags);
2071 extern int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
2072 struct file *file_out, loff_t pos_out,
2074 unsigned int remap_flags);
2075 extern loff_t do_clone_file_range(struct file *file_in, loff_t pos_in,
2076 struct file *file_out, loff_t pos_out,
2077 loff_t len, unsigned int remap_flags);
2078 extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in,
2079 struct file *file_out, loff_t pos_out,
2080 loff_t len, unsigned int remap_flags);
2081 extern int vfs_dedupe_file_range(struct file *file,
2082 struct file_dedupe_range *same);
2083 extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
2084 struct file *dst_file, loff_t dst_pos,
2085 loff_t len, unsigned int remap_flags);
2088 struct super_operations {
2089 struct inode *(*alloc_inode)(struct super_block *sb);
2090 void (*destroy_inode)(struct inode *);
2091 void (*free_inode)(struct inode *);
2093 void (*dirty_inode) (struct inode *, int flags);
2094 int (*write_inode) (struct inode *, struct writeback_control *wbc);
2095 int (*drop_inode) (struct inode *);
2096 void (*evict_inode) (struct inode *);
2097 void (*put_super) (struct super_block *);
2098 int (*sync_fs)(struct super_block *sb, int wait);
2099 int (*freeze_super) (struct super_block *);
2100 int (*freeze_fs) (struct super_block *);
2101 int (*thaw_super) (struct super_block *);
2102 int (*unfreeze_fs) (struct super_block *);
2103 int (*statfs) (struct dentry *, struct kstatfs *);
2104 int (*remount_fs) (struct super_block *, int *, char *);
2105 void (*umount_begin) (struct super_block *);
2107 int (*show_options)(struct seq_file *, struct dentry *);
2108 int (*show_devname)(struct seq_file *, struct dentry *);
2109 int (*show_path)(struct seq_file *, struct dentry *);
2110 int (*show_stats)(struct seq_file *, struct dentry *);
2112 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
2113 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
2114 struct dquot **(*get_dquots)(struct inode *);
2116 long (*nr_cached_objects)(struct super_block *,
2117 struct shrink_control *);
2118 long (*free_cached_objects)(struct super_block *,
2119 struct shrink_control *);
2123 * Inode flags - they have no relation to superblock flags now
2125 #define S_SYNC (1 << 0) /* Writes are synced at once */
2126 #define S_NOATIME (1 << 1) /* Do not update access times */
2127 #define S_APPEND (1 << 2) /* Append-only file */
2128 #define S_IMMUTABLE (1 << 3) /* Immutable file */
2129 #define S_DEAD (1 << 4) /* removed, but still open directory */
2130 #define S_NOQUOTA (1 << 5) /* Inode is not counted to quota */
2131 #define S_DIRSYNC (1 << 6) /* Directory modifications are synchronous */
2132 #define S_NOCMTIME (1 << 7) /* Do not update file c/mtime */
2133 #define S_SWAPFILE (1 << 8) /* Do not truncate: swapon got its bmaps */
2134 #define S_PRIVATE (1 << 9) /* Inode is fs-internal */
2135 #define S_IMA (1 << 10) /* Inode has an associated IMA struct */
2136 #define S_AUTOMOUNT (1 << 11) /* Automount/referral quasi-directory */
2137 #define S_NOSEC (1 << 12) /* no suid or xattr security attributes */
2138 #ifdef CONFIG_FS_DAX
2139 #define S_DAX (1 << 13) /* Direct Access, avoiding the page cache */
2141 #define S_DAX 0 /* Make all the DAX code disappear */
2143 #define S_ENCRYPTED (1 << 14) /* Encrypted file (using fs/crypto/) */
2144 #define S_CASEFOLD (1 << 15) /* Casefolded file */
2145 #define S_VERITY (1 << 16) /* Verity file (using fs/verity/) */
2146 #define S_KERNEL_FILE (1 << 17) /* File is in use by the kernel (eg. fs/cachefiles) */
2149 * Note that nosuid etc flags are inode-specific: setting some file-system
2150 * flags just means all the inodes inherit those flags by default. It might be
2151 * possible to override it selectively if you really wanted to with some
2152 * ioctl() that is not currently implemented.
2154 * Exception: SB_RDONLY is always applied to the entire file system.
2156 * Unfortunately, it is possible to change a filesystems flags with it mounted
2157 * with files in use. This means that all of the inodes will not have their
2158 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
2159 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
2161 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
2163 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
2164 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
2165 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
2166 ((inode)->i_flags & S_SYNC))
2167 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
2168 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
2169 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
2170 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
2171 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
2173 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
2174 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
2175 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
2176 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
2178 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
2179 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
2180 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
2181 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
2182 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
2183 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
2184 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
2185 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
2186 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
2187 #define IS_CASEFOLDED(inode) ((inode)->i_flags & S_CASEFOLD)
2188 #define IS_VERITY(inode) ((inode)->i_flags & S_VERITY)
2190 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
2191 (inode)->i_rdev == WHITEOUT_DEV)
2193 static inline bool HAS_UNMAPPED_ID(struct user_namespace *mnt_userns,
2194 struct inode *inode)
2196 return !uid_valid(i_uid_into_mnt(mnt_userns, inode)) ||
2197 !gid_valid(i_gid_into_mnt(mnt_userns, inode));
2200 static inline int iocb_flags(struct file *file);
2202 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
2204 *kiocb = (struct kiocb) {
2206 .ki_flags = iocb_flags(filp),
2207 .ki_ioprio = get_current_ioprio(),
2211 static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src,
2214 *kiocb = (struct kiocb) {
2216 .ki_flags = kiocb_src->ki_flags,
2217 .ki_ioprio = kiocb_src->ki_ioprio,
2218 .ki_pos = kiocb_src->ki_pos,
2223 * Inode state bits. Protected by inode->i_lock
2225 * Four bits determine the dirty state of the inode: I_DIRTY_SYNC,
2226 * I_DIRTY_DATASYNC, I_DIRTY_PAGES, and I_DIRTY_TIME.
2228 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2229 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2230 * various stages of removing an inode.
2232 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2234 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2235 * fdatasync() (unless I_DIRTY_DATASYNC is also set).
2236 * Timestamp updates are the usual cause.
2237 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2238 * these changes separately from I_DIRTY_SYNC so that we
2239 * don't have to write inode on fdatasync() when only
2240 * e.g. the timestamps have changed.
2241 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2242 * I_DIRTY_TIME The inode itself only has dirty timestamps, and the
2243 * lazytime mount option is enabled. We keep track of this
2244 * separately from I_DIRTY_SYNC in order to implement
2245 * lazytime. This gets cleared if I_DIRTY_INODE
2246 * (I_DIRTY_SYNC and/or I_DIRTY_DATASYNC) gets set. I.e.
2247 * either I_DIRTY_TIME *or* I_DIRTY_INODE can be set in
2248 * i_state, but not both. I_DIRTY_PAGES may still be set.
2249 * I_NEW Serves as both a mutex and completion notification.
2250 * New inodes set I_NEW. If two processes both create
2251 * the same inode, one of them will release its inode and
2252 * wait for I_NEW to be released before returning.
2253 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2254 * also cause waiting on I_NEW, without I_NEW actually
2255 * being set. find_inode() uses this to prevent returning
2256 * nearly-dead inodes.
2257 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2258 * is zero. I_FREEING must be set when I_WILL_FREE is
2260 * I_FREEING Set when inode is about to be freed but still has dirty
2261 * pages or buffers attached or the inode itself is still
2263 * I_CLEAR Added by clear_inode(). In this state the inode is
2264 * clean and can be destroyed. Inode keeps I_FREEING.
2266 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2267 * prohibited for many purposes. iget() must wait for
2268 * the inode to be completely released, then create it
2269 * anew. Other functions will just ignore such inodes,
2270 * if appropriate. I_NEW is used for waiting.
2272 * I_SYNC Writeback of inode is running. The bit is set during
2273 * data writeback, and cleared with a wakeup on the bit
2274 * address once it is done. The bit is also used to pin
2275 * the inode in memory for flusher thread.
2277 * I_REFERENCED Marks the inode as recently references on the LRU list.
2279 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2281 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2282 * synchronize competing switching instances and to tell
2283 * wb stat updates to grab the i_pages lock. See
2284 * inode_switch_wbs_work_fn() for details.
2286 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2287 * and work dirs among overlayfs mounts.
2289 * I_CREATING New object's inode in the middle of setting up.
2291 * I_DONTCACHE Evict inode as soon as it is not used anymore.
2293 * I_SYNC_QUEUED Inode is queued in b_io or b_more_io writeback lists.
2294 * Used to detect that mark_inode_dirty() should not move
2295 * inode between dirty lists.
2297 * I_PINNING_FSCACHE_WB Inode is pinning an fscache object for writeback.
2299 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2301 #define I_DIRTY_SYNC (1 << 0)
2302 #define I_DIRTY_DATASYNC (1 << 1)
2303 #define I_DIRTY_PAGES (1 << 2)
2305 #define I_NEW (1 << __I_NEW)
2306 #define I_WILL_FREE (1 << 4)
2307 #define I_FREEING (1 << 5)
2308 #define I_CLEAR (1 << 6)
2310 #define I_SYNC (1 << __I_SYNC)
2311 #define I_REFERENCED (1 << 8)
2312 #define __I_DIO_WAKEUP 9
2313 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2314 #define I_LINKABLE (1 << 10)
2315 #define I_DIRTY_TIME (1 << 11)
2316 #define I_WB_SWITCH (1 << 13)
2317 #define I_OVL_INUSE (1 << 14)
2318 #define I_CREATING (1 << 15)
2319 #define I_DONTCACHE (1 << 16)
2320 #define I_SYNC_QUEUED (1 << 17)
2321 #define I_PINNING_FSCACHE_WB (1 << 18)
2323 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2324 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2325 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2327 extern void __mark_inode_dirty(struct inode *, int);
2328 static inline void mark_inode_dirty(struct inode *inode)
2330 __mark_inode_dirty(inode, I_DIRTY);
2333 static inline void mark_inode_dirty_sync(struct inode *inode)
2335 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2339 * Returns true if the given inode itself only has dirty timestamps (its pages
2340 * may still be dirty) and isn't currently being allocated or freed.
2341 * Filesystems should call this if when writing an inode when lazytime is
2342 * enabled, they want to opportunistically write the timestamps of other inodes
2343 * located very nearby on-disk, e.g. in the same inode block. This returns true
2344 * if the given inode is in need of such an opportunistic update. Requires
2345 * i_lock, or at least later re-checking under i_lock.
2347 static inline bool inode_is_dirtytime_only(struct inode *inode)
2349 return (inode->i_state & (I_DIRTY_TIME | I_NEW |
2350 I_FREEING | I_WILL_FREE)) == I_DIRTY_TIME;
2353 extern void inc_nlink(struct inode *inode);
2354 extern void drop_nlink(struct inode *inode);
2355 extern void clear_nlink(struct inode *inode);
2356 extern void set_nlink(struct inode *inode, unsigned int nlink);
2358 static inline void inode_inc_link_count(struct inode *inode)
2361 mark_inode_dirty(inode);
2364 static inline void inode_dec_link_count(struct inode *inode)
2367 mark_inode_dirty(inode);
2370 enum file_time_flags {
2377 extern bool atime_needs_update(const struct path *, struct inode *);
2378 extern void touch_atime(const struct path *);
2379 int inode_update_time(struct inode *inode, struct timespec64 *time, int flags);
2381 static inline void file_accessed(struct file *file)
2383 if (!(file->f_flags & O_NOATIME))
2384 touch_atime(&file->f_path);
2387 extern int file_modified(struct file *file);
2389 int sync_inode_metadata(struct inode *inode, int wait);
2391 struct file_system_type {
2394 #define FS_REQUIRES_DEV 1
2395 #define FS_BINARY_MOUNTDATA 2
2396 #define FS_HAS_SUBTYPE 4
2397 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2398 #define FS_DISALLOW_NOTIFY_PERM 16 /* Disable fanotify permission events */
2399 #define FS_ALLOW_IDMAP 32 /* FS has been updated to handle vfs idmappings. */
2400 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2401 int (*init_fs_context)(struct fs_context *);
2402 const struct fs_parameter_spec *parameters;
2403 struct dentry *(*mount) (struct file_system_type *, int,
2404 const char *, void *);
2405 void (*kill_sb) (struct super_block *);
2406 struct module *owner;
2407 struct file_system_type * next;
2408 struct hlist_head fs_supers;
2410 struct lock_class_key s_lock_key;
2411 struct lock_class_key s_umount_key;
2412 struct lock_class_key s_vfs_rename_key;
2413 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2415 struct lock_class_key i_lock_key;
2416 struct lock_class_key i_mutex_key;
2417 struct lock_class_key invalidate_lock_key;
2418 struct lock_class_key i_mutex_dir_key;
2421 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2423 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2424 int flags, const char *dev_name, void *data,
2425 int (*fill_super)(struct super_block *, void *, int));
2426 extern struct dentry *mount_single(struct file_system_type *fs_type,
2427 int flags, void *data,
2428 int (*fill_super)(struct super_block *, void *, int));
2429 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2430 int flags, void *data,
2431 int (*fill_super)(struct super_block *, void *, int));
2432 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2433 void generic_shutdown_super(struct super_block *sb);
2434 void kill_block_super(struct super_block *sb);
2435 void kill_anon_super(struct super_block *sb);
2436 void kill_litter_super(struct super_block *sb);
2437 void deactivate_super(struct super_block *sb);
2438 void deactivate_locked_super(struct super_block *sb);
2439 int set_anon_super(struct super_block *s, void *data);
2440 int set_anon_super_fc(struct super_block *s, struct fs_context *fc);
2441 int get_anon_bdev(dev_t *);
2442 void free_anon_bdev(dev_t);
2443 struct super_block *sget_fc(struct fs_context *fc,
2444 int (*test)(struct super_block *, struct fs_context *),
2445 int (*set)(struct super_block *, struct fs_context *));
2446 struct super_block *sget(struct file_system_type *type,
2447 int (*test)(struct super_block *,void *),
2448 int (*set)(struct super_block *,void *),
2449 int flags, void *data);
2451 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2452 #define fops_get(fops) \
2453 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2454 #define fops_put(fops) \
2455 do { if (fops) module_put((fops)->owner); } while(0)
2457 * This one is to be used *ONLY* from ->open() instances.
2458 * fops must be non-NULL, pinned down *and* module dependencies
2459 * should be sufficient to pin the caller down as well.
2461 #define replace_fops(f, fops) \
2463 struct file *__file = (f); \
2464 fops_put(__file->f_op); \
2465 BUG_ON(!(__file->f_op = (fops))); \
2468 extern int register_filesystem(struct file_system_type *);
2469 extern int unregister_filesystem(struct file_system_type *);
2470 extern int vfs_statfs(const struct path *, struct kstatfs *);
2471 extern int user_statfs(const char __user *, struct kstatfs *);
2472 extern int fd_statfs(int, struct kstatfs *);
2473 extern int freeze_super(struct super_block *super);
2474 extern int thaw_super(struct super_block *super);
2475 extern __printf(2, 3)
2476 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2477 extern int super_setup_bdi(struct super_block *sb);
2479 extern int current_umask(void);
2481 extern void ihold(struct inode * inode);
2482 extern void iput(struct inode *);
2483 extern int generic_update_time(struct inode *, struct timespec64 *, int);
2486 extern struct kobject *fs_kobj;
2488 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2490 #ifdef CONFIG_FILE_LOCKING
2491 static inline int break_lease(struct inode *inode, unsigned int mode)
2494 * Since this check is lockless, we must ensure that any refcounts
2495 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2496 * could end up racing with tasks trying to set a new lease on this
2500 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2501 return __break_lease(inode, mode, FL_LEASE);
2505 static inline int break_deleg(struct inode *inode, unsigned int mode)
2508 * Since this check is lockless, we must ensure that any refcounts
2509 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2510 * could end up racing with tasks trying to set a new lease on this
2514 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2515 return __break_lease(inode, mode, FL_DELEG);
2519 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2523 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2524 if (ret == -EWOULDBLOCK && delegated_inode) {
2525 *delegated_inode = inode;
2531 static inline int break_deleg_wait(struct inode **delegated_inode)
2535 ret = break_deleg(*delegated_inode, O_WRONLY);
2536 iput(*delegated_inode);
2537 *delegated_inode = NULL;
2541 static inline int break_layout(struct inode *inode, bool wait)
2544 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2545 return __break_lease(inode,
2546 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2551 #else /* !CONFIG_FILE_LOCKING */
2552 static inline int break_lease(struct inode *inode, unsigned int mode)
2557 static inline int break_deleg(struct inode *inode, unsigned int mode)
2562 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2567 static inline int break_deleg_wait(struct inode **delegated_inode)
2573 static inline int break_layout(struct inode *inode, bool wait)
2578 #endif /* CONFIG_FILE_LOCKING */
2583 const char *name; /* pointer to actual string */
2584 const __user char *uptr; /* original userland pointer */
2586 struct audit_names *aname;
2589 static_assert(offsetof(struct filename, iname) % sizeof(long) == 0);
2591 static inline struct user_namespace *file_mnt_user_ns(struct file *file)
2593 return mnt_user_ns(file->f_path.mnt);
2597 * is_idmapped_mnt - check whether a mount is mapped
2598 * @mnt: the mount to check
2600 * If @mnt has an idmapping attached different from the
2601 * filesystem's idmapping then @mnt is mapped.
2603 * Return: true if mount is mapped, false if not.
2605 static inline bool is_idmapped_mnt(const struct vfsmount *mnt)
2607 return mnt_user_ns(mnt) != mnt->mnt_sb->s_user_ns;
2610 extern long vfs_truncate(const struct path *, loff_t);
2611 int do_truncate(struct user_namespace *, struct dentry *, loff_t start,
2612 unsigned int time_attrs, struct file *filp);
2613 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2615 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2617 extern struct file *file_open_name(struct filename *, int, umode_t);
2618 extern struct file *filp_open(const char *, int, umode_t);
2619 extern struct file *file_open_root(const struct path *,
2620 const char *, int, umode_t);
2621 static inline struct file *file_open_root_mnt(struct vfsmount *mnt,
2622 const char *name, int flags, umode_t mode)
2624 return file_open_root(&(struct path){.mnt = mnt, .dentry = mnt->mnt_root},
2627 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2628 extern struct file *dentry_create(const struct path *path, int flags,
2629 umode_t mode, const struct cred *cred);
2630 extern struct file * open_with_fake_path(const struct path *, int,
2631 struct inode*, const struct cred *);
2632 static inline struct file *file_clone_open(struct file *file)
2634 return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2636 extern int filp_close(struct file *, fl_owner_t id);
2638 extern struct filename *getname_flags(const char __user *, int, int *);
2639 extern struct filename *getname_uflags(const char __user *, int);
2640 extern struct filename *getname(const char __user *);
2641 extern struct filename *getname_kernel(const char *);
2642 extern void putname(struct filename *name);
2644 extern int finish_open(struct file *file, struct dentry *dentry,
2645 int (*open)(struct inode *, struct file *));
2646 extern int finish_no_open(struct file *file, struct dentry *dentry);
2649 extern void __init vfs_caches_init_early(void);
2650 extern void __init vfs_caches_init(void);
2652 extern struct kmem_cache *names_cachep;
2654 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2655 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2657 extern struct super_block *blockdev_superblock;
2658 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2660 return IS_ENABLED(CONFIG_BLOCK) && sb == blockdev_superblock;
2663 void emergency_thaw_all(void);
2664 extern int sync_filesystem(struct super_block *);
2665 extern const struct file_operations def_blk_fops;
2666 extern const struct file_operations def_chr_fops;
2669 #define CHRDEV_MAJOR_MAX 512
2670 /* Marks the bottom of the first segment of free char majors */
2671 #define CHRDEV_MAJOR_DYN_END 234
2672 /* Marks the top and bottom of the second segment of free char majors */
2673 #define CHRDEV_MAJOR_DYN_EXT_START 511
2674 #define CHRDEV_MAJOR_DYN_EXT_END 384
2676 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2677 extern int register_chrdev_region(dev_t, unsigned, const char *);
2678 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2679 unsigned int count, const char *name,
2680 const struct file_operations *fops);
2681 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2682 unsigned int count, const char *name);
2683 extern void unregister_chrdev_region(dev_t, unsigned);
2684 extern void chrdev_show(struct seq_file *,off_t);
2686 static inline int register_chrdev(unsigned int major, const char *name,
2687 const struct file_operations *fops)
2689 return __register_chrdev(major, 0, 256, name, fops);
2692 static inline void unregister_chrdev(unsigned int major, const char *name)
2694 __unregister_chrdev(major, 0, 256, name);
2697 extern void init_special_inode(struct inode *, umode_t, dev_t);
2699 /* Invalid inode operations -- fs/bad_inode.c */
2700 extern void make_bad_inode(struct inode *);
2701 extern bool is_bad_inode(struct inode *);
2703 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2705 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2706 extern int __must_check file_write_and_wait_range(struct file *file,
2707 loff_t start, loff_t end);
2709 static inline int file_write_and_wait(struct file *file)
2711 return file_write_and_wait_range(file, 0, LLONG_MAX);
2714 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2716 extern int vfs_fsync(struct file *file, int datasync);
2718 extern int sync_file_range(struct file *file, loff_t offset, loff_t nbytes,
2719 unsigned int flags);
2722 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2723 * to already be updated for the write, and will return either the amount
2724 * of bytes passed in, or an error if syncing the file failed.
2726 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2728 if (iocb->ki_flags & IOCB_DSYNC) {
2729 int ret = vfs_fsync_range(iocb->ki_filp,
2730 iocb->ki_pos - count, iocb->ki_pos - 1,
2731 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2739 extern void emergency_sync(void);
2740 extern void emergency_remount(void);
2743 extern int bmap(struct inode *inode, sector_t *block);
2745 static inline int bmap(struct inode *inode, sector_t *block)
2751 int notify_change(struct user_namespace *, struct dentry *,
2752 struct iattr *, struct inode **);
2753 int inode_permission(struct user_namespace *, struct inode *, int);
2754 int generic_permission(struct user_namespace *, struct inode *, int);
2755 static inline int file_permission(struct file *file, int mask)
2757 return inode_permission(file_mnt_user_ns(file),
2758 file_inode(file), mask);
2760 static inline int path_permission(const struct path *path, int mask)
2762 return inode_permission(mnt_user_ns(path->mnt),
2763 d_inode(path->dentry), mask);
2765 int __check_sticky(struct user_namespace *mnt_userns, struct inode *dir,
2766 struct inode *inode);
2768 static inline bool execute_ok(struct inode *inode)
2770 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2773 static inline bool inode_wrong_type(const struct inode *inode, umode_t mode)
2775 return (inode->i_mode ^ mode) & S_IFMT;
2778 static inline void file_start_write(struct file *file)
2780 if (!S_ISREG(file_inode(file)->i_mode))
2782 sb_start_write(file_inode(file)->i_sb);
2785 static inline bool file_start_write_trylock(struct file *file)
2787 if (!S_ISREG(file_inode(file)->i_mode))
2789 return sb_start_write_trylock(file_inode(file)->i_sb);
2792 static inline void file_end_write(struct file *file)
2794 if (!S_ISREG(file_inode(file)->i_mode))
2796 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2800 * This is used for regular files where some users -- especially the
2801 * currently executed binary in a process, previously handled via
2802 * VM_DENYWRITE -- cannot handle concurrent write (and maybe mmap
2803 * read-write shared) accesses.
2805 * get_write_access() gets write permission for a file.
2806 * put_write_access() releases this write permission.
2807 * deny_write_access() denies write access to a file.
2808 * allow_write_access() re-enables write access to a file.
2810 * The i_writecount field of an inode can have the following values:
2811 * 0: no write access, no denied write access
2812 * < 0: (-i_writecount) users that denied write access to the file.
2813 * > 0: (i_writecount) users that have write access to the file.
2815 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2816 * except for the cases where we don't hold i_writecount yet. Then we need to
2817 * use {get,deny}_write_access() - these functions check the sign and refuse
2818 * to do the change if sign is wrong.
2820 static inline int get_write_access(struct inode *inode)
2822 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2824 static inline int deny_write_access(struct file *file)
2826 struct inode *inode = file_inode(file);
2827 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2829 static inline void put_write_access(struct inode * inode)
2831 atomic_dec(&inode->i_writecount);
2833 static inline void allow_write_access(struct file *file)
2836 atomic_inc(&file_inode(file)->i_writecount);
2838 static inline bool inode_is_open_for_write(const struct inode *inode)
2840 return atomic_read(&inode->i_writecount) > 0;
2843 #if defined(CONFIG_IMA) || defined(CONFIG_FILE_LOCKING)
2844 static inline void i_readcount_dec(struct inode *inode)
2846 BUG_ON(!atomic_read(&inode->i_readcount));
2847 atomic_dec(&inode->i_readcount);
2849 static inline void i_readcount_inc(struct inode *inode)
2851 atomic_inc(&inode->i_readcount);
2854 static inline void i_readcount_dec(struct inode *inode)
2858 static inline void i_readcount_inc(struct inode *inode)
2863 extern int do_pipe_flags(int *, int);
2865 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
2866 ssize_t __kernel_read(struct file *file, void *buf, size_t count, loff_t *pos);
2867 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
2868 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
2869 extern struct file * open_exec(const char *);
2871 /* fs/dcache.c -- generic fs support functions */
2872 extern bool is_subdir(struct dentry *, struct dentry *);
2873 extern bool path_is_under(const struct path *, const struct path *);
2875 extern char *file_path(struct file *, char *, int);
2877 #include <linux/err.h>
2879 /* needed for stackable file system support */
2880 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2882 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2884 extern int inode_init_always(struct super_block *, struct inode *);
2885 extern void inode_init_once(struct inode *);
2886 extern void address_space_init_once(struct address_space *mapping);
2887 extern struct inode * igrab(struct inode *);
2888 extern ino_t iunique(struct super_block *, ino_t);
2889 extern int inode_needs_sync(struct inode *inode);
2890 extern int generic_delete_inode(struct inode *inode);
2891 static inline int generic_drop_inode(struct inode *inode)
2893 return !inode->i_nlink || inode_unhashed(inode);
2895 extern void d_mark_dontcache(struct inode *inode);
2897 extern struct inode *ilookup5_nowait(struct super_block *sb,
2898 unsigned long hashval, int (*test)(struct inode *, void *),
2900 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2901 int (*test)(struct inode *, void *), void *data);
2902 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2904 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
2905 int (*test)(struct inode *, void *),
2906 int (*set)(struct inode *, void *),
2908 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2909 extern struct inode * iget_locked(struct super_block *, unsigned long);
2910 extern struct inode *find_inode_nowait(struct super_block *,
2912 int (*match)(struct inode *,
2913 unsigned long, void *),
2915 extern struct inode *find_inode_rcu(struct super_block *, unsigned long,
2916 int (*)(struct inode *, void *), void *);
2917 extern struct inode *find_inode_by_ino_rcu(struct super_block *, unsigned long);
2918 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2919 extern int insert_inode_locked(struct inode *);
2920 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2921 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2923 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2925 extern void unlock_new_inode(struct inode *);
2926 extern void discard_new_inode(struct inode *);
2927 extern unsigned int get_next_ino(void);
2928 extern void evict_inodes(struct super_block *sb);
2929 void dump_mapping(const struct address_space *);
2932 * Userspace may rely on the the inode number being non-zero. For example, glibc
2933 * simply ignores files with zero i_ino in unlink() and other places.
2935 * As an additional complication, if userspace was compiled with
2936 * _FILE_OFFSET_BITS=32 on a 64-bit kernel we'll only end up reading out the
2937 * lower 32 bits, so we need to check that those aren't zero explicitly. With
2938 * _FILE_OFFSET_BITS=64, this may cause some harmless false-negatives, but
2939 * better safe than sorry.
2941 static inline bool is_zero_ino(ino_t ino)
2943 return (u32)ino == 0;
2946 extern void __iget(struct inode * inode);
2947 extern void iget_failed(struct inode *);
2948 extern void clear_inode(struct inode *);
2949 extern void __destroy_inode(struct inode *);
2950 extern struct inode *new_inode_pseudo(struct super_block *sb);
2951 extern struct inode *new_inode(struct super_block *sb);
2952 extern void free_inode_nonrcu(struct inode *inode);
2953 extern int should_remove_suid(struct dentry *);
2954 extern int file_remove_privs(struct file *);
2957 * This must be used for allocating filesystems specific inodes to set
2958 * up the inode reclaim context correctly.
2960 static inline void *
2961 alloc_inode_sb(struct super_block *sb, struct kmem_cache *cache, gfp_t gfp)
2963 return kmem_cache_alloc_lru(cache, &sb->s_inode_lru, gfp);
2966 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2967 static inline void insert_inode_hash(struct inode *inode)
2969 __insert_inode_hash(inode, inode->i_ino);
2972 extern void __remove_inode_hash(struct inode *);
2973 static inline void remove_inode_hash(struct inode *inode)
2975 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2976 __remove_inode_hash(inode);
2979 extern void inode_sb_list_add(struct inode *inode);
2980 extern void inode_add_lru(struct inode *inode);
2982 extern int sb_set_blocksize(struct super_block *, int);
2983 extern int sb_min_blocksize(struct super_block *, int);
2985 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2986 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2987 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2988 int generic_write_checks_count(struct kiocb *iocb, loff_t *count);
2989 extern int generic_write_check_limits(struct file *file, loff_t pos,
2991 extern int generic_file_rw_checks(struct file *file_in, struct file *file_out);
2992 ssize_t filemap_read(struct kiocb *iocb, struct iov_iter *to,
2993 ssize_t already_read);
2994 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2995 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2996 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2997 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
2998 ssize_t generic_perform_write(struct kiocb *, struct iov_iter *);
3000 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
3002 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
3004 ssize_t vfs_iocb_iter_read(struct file *file, struct kiocb *iocb,
3005 struct iov_iter *iter);
3006 ssize_t vfs_iocb_iter_write(struct file *file, struct kiocb *iocb,
3007 struct iov_iter *iter);
3010 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
3011 struct pipe_inode_info *, size_t, unsigned int);
3012 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
3013 struct file *, loff_t *, size_t, unsigned int);
3014 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
3015 struct file *out, loff_t *, size_t len, unsigned int flags);
3016 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
3017 loff_t *opos, size_t len, unsigned int flags);
3021 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3022 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3023 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
3024 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3025 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3026 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3027 int whence, loff_t maxsize, loff_t eof);
3028 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3029 int whence, loff_t size);
3030 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3031 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3032 int rw_verify_area(int, struct file *, const loff_t *, size_t);
3033 extern int generic_file_open(struct inode * inode, struct file * filp);
3034 extern int nonseekable_open(struct inode * inode, struct file * filp);
3035 extern int stream_open(struct inode * inode, struct file * filp);
3038 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3039 loff_t file_offset);
3042 /* need locking between buffered and direct access */
3045 /* filesystem does not support filling holes */
3046 DIO_SKIP_HOLES = 0x02,
3049 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3050 struct block_device *bdev, struct iov_iter *iter,
3051 get_block_t get_block,
3052 dio_iodone_t end_io, dio_submit_t submit_io,
3055 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3056 struct inode *inode,
3057 struct iov_iter *iter,
3058 get_block_t get_block)
3060 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3061 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3065 void inode_dio_wait(struct inode *inode);
3068 * inode_dio_begin - signal start of a direct I/O requests
3069 * @inode: inode the direct I/O happens on
3071 * This is called once we've finished processing a direct I/O request,
3072 * and is used to wake up callers waiting for direct I/O to be quiesced.
3074 static inline void inode_dio_begin(struct inode *inode)
3076 atomic_inc(&inode->i_dio_count);
3080 * inode_dio_end - signal finish of a direct I/O requests
3081 * @inode: inode the direct I/O happens on
3083 * This is called once we've finished processing a direct I/O request,
3084 * and is used to wake up callers waiting for direct I/O to be quiesced.
3086 static inline void inode_dio_end(struct inode *inode)
3088 if (atomic_dec_and_test(&inode->i_dio_count))
3089 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3093 * Warn about a page cache invalidation failure diring a direct I/O write.
3095 void dio_warn_stale_pagecache(struct file *filp);
3097 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3100 extern const struct file_operations generic_ro_fops;
3102 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3104 extern int readlink_copy(char __user *, int, const char *);
3105 extern int page_readlink(struct dentry *, char __user *, int);
3106 extern const char *page_get_link(struct dentry *, struct inode *,
3107 struct delayed_call *);
3108 extern void page_put_link(void *);
3109 extern int page_symlink(struct inode *inode, const char *symname, int len);
3110 extern const struct inode_operations page_symlink_inode_operations;
3111 extern void kfree_link(void *);
3112 void generic_fillattr(struct user_namespace *, struct inode *, struct kstat *);
3113 void generic_fill_statx_attr(struct inode *inode, struct kstat *stat);
3114 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3115 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3116 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3117 void inode_add_bytes(struct inode *inode, loff_t bytes);
3118 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3119 void inode_sub_bytes(struct inode *inode, loff_t bytes);
3120 static inline loff_t __inode_get_bytes(struct inode *inode)
3122 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3124 loff_t inode_get_bytes(struct inode *inode);
3125 void inode_set_bytes(struct inode *inode, loff_t bytes);
3126 const char *simple_get_link(struct dentry *, struct inode *,
3127 struct delayed_call *);
3128 extern const struct inode_operations simple_symlink_inode_operations;
3130 extern int iterate_dir(struct file *, struct dir_context *);
3132 int vfs_fstatat(int dfd, const char __user *filename, struct kstat *stat,
3134 int vfs_fstat(int fd, struct kstat *stat);
3136 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3138 return vfs_fstatat(AT_FDCWD, filename, stat, 0);
3140 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3142 return vfs_fstatat(AT_FDCWD, name, stat, AT_SYMLINK_NOFOLLOW);
3145 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3146 extern int vfs_readlink(struct dentry *, char __user *, int);
3148 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3149 extern void put_filesystem(struct file_system_type *fs);
3150 extern struct file_system_type *get_fs_type(const char *name);
3151 extern struct super_block *get_super(struct block_device *);
3152 extern struct super_block *get_active_super(struct block_device *bdev);
3153 extern void drop_super(struct super_block *sb);
3154 extern void drop_super_exclusive(struct super_block *sb);
3155 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3156 extern void iterate_supers_type(struct file_system_type *,
3157 void (*)(struct super_block *, void *), void *);
3159 extern int dcache_dir_open(struct inode *, struct file *);
3160 extern int dcache_dir_close(struct inode *, struct file *);
3161 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3162 extern int dcache_readdir(struct file *, struct dir_context *);
3163 extern int simple_setattr(struct user_namespace *, struct dentry *,
3165 extern int simple_getattr(struct user_namespace *, const struct path *,
3166 struct kstat *, u32, unsigned int);
3167 extern int simple_statfs(struct dentry *, struct kstatfs *);
3168 extern int simple_open(struct inode *inode, struct file *file);
3169 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3170 extern int simple_unlink(struct inode *, struct dentry *);
3171 extern int simple_rmdir(struct inode *, struct dentry *);
3172 extern int simple_rename_exchange(struct inode *old_dir, struct dentry *old_dentry,
3173 struct inode *new_dir, struct dentry *new_dentry);
3174 extern int simple_rename(struct user_namespace *, struct inode *,
3175 struct dentry *, struct inode *, struct dentry *,
3177 extern void simple_recursive_removal(struct dentry *,
3178 void (*callback)(struct dentry *));
3179 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3180 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3181 extern int simple_empty(struct dentry *);
3182 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3183 loff_t pos, unsigned len,
3184 struct page **pagep, void **fsdata);
3185 extern const struct address_space_operations ram_aops;
3186 extern int always_delete_dentry(const struct dentry *);
3187 extern struct inode *alloc_anon_inode(struct super_block *);
3188 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3189 extern const struct dentry_operations simple_dentry_operations;
3191 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3192 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3193 extern const struct file_operations simple_dir_operations;
3194 extern const struct inode_operations simple_dir_inode_operations;
3195 extern void make_empty_dir_inode(struct inode *inode);
3196 extern bool is_empty_dir_inode(struct inode *inode);
3197 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3198 struct dentry *d_alloc_name(struct dentry *, const char *);
3199 extern int simple_fill_super(struct super_block *, unsigned long,
3200 const struct tree_descr *);
3201 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3202 extern void simple_release_fs(struct vfsmount **mount, int *count);
3204 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3205 loff_t *ppos, const void *from, size_t available);
3206 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3207 const void __user *from, size_t count);
3209 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3210 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3212 extern int generic_check_addressable(unsigned, u64);
3214 extern void generic_set_encrypted_ci_d_ops(struct dentry *dentry);
3216 int may_setattr(struct user_namespace *mnt_userns, struct inode *inode,
3217 unsigned int ia_valid);
3218 int setattr_prepare(struct user_namespace *, struct dentry *, struct iattr *);
3219 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3220 void setattr_copy(struct user_namespace *, struct inode *inode,
3221 const struct iattr *attr);
3223 extern int file_update_time(struct file *file);
3225 static inline bool vma_is_dax(const struct vm_area_struct *vma)
3227 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3230 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3232 struct inode *inode;
3234 if (!IS_ENABLED(CONFIG_FS_DAX) || !vma->vm_file)
3236 if (!vma_is_dax(vma))
3238 inode = file_inode(vma->vm_file);
3239 if (S_ISCHR(inode->i_mode))
3240 return false; /* device-dax */
3244 static inline int iocb_flags(struct file *file)
3247 if (file->f_flags & O_APPEND)
3249 if (file->f_flags & O_DIRECT)
3251 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3253 if (file->f_flags & __O_SYNC)
3258 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3260 int kiocb_flags = 0;
3262 /* make sure there's no overlap between RWF and private IOCB flags */
3263 BUILD_BUG_ON((__force int) RWF_SUPPORTED & IOCB_EVENTFD);
3267 if (unlikely(flags & ~RWF_SUPPORTED))
3270 if (flags & RWF_NOWAIT) {
3271 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3273 kiocb_flags |= IOCB_NOIO;
3275 kiocb_flags |= (__force int) (flags & RWF_SUPPORTED);
3276 if (flags & RWF_SYNC)
3277 kiocb_flags |= IOCB_DSYNC;
3279 ki->ki_flags |= kiocb_flags;
3283 static inline ino_t parent_ino(struct dentry *dentry)
3288 * Don't strictly need d_lock here? If the parent ino could change
3289 * then surely we'd have a deeper race in the caller?
3291 spin_lock(&dentry->d_lock);
3292 res = dentry->d_parent->d_inode->i_ino;
3293 spin_unlock(&dentry->d_lock);
3297 /* Transaction based IO helpers */
3300 * An argresp is stored in an allocated page and holds the
3301 * size of the argument or response, along with its content
3303 struct simple_transaction_argresp {
3308 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3310 char *simple_transaction_get(struct file *file, const char __user *buf,
3312 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3313 size_t size, loff_t *pos);
3314 int simple_transaction_release(struct inode *inode, struct file *file);
3316 void simple_transaction_set(struct file *file, size_t n);
3319 * simple attribute files
3321 * These attributes behave similar to those in sysfs:
3323 * Writing to an attribute immediately sets a value, an open file can be
3324 * written to multiple times.
3326 * Reading from an attribute creates a buffer from the value that might get
3327 * read with multiple read calls. When the attribute has been read
3328 * completely, no further read calls are possible until the file is opened
3331 * All attributes contain a text representation of a numeric value
3332 * that are accessed with the get() and set() functions.
3334 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3335 static int __fops ## _open(struct inode *inode, struct file *file) \
3337 __simple_attr_check_format(__fmt, 0ull); \
3338 return simple_attr_open(inode, file, __get, __set, __fmt); \
3340 static const struct file_operations __fops = { \
3341 .owner = THIS_MODULE, \
3342 .open = __fops ## _open, \
3343 .release = simple_attr_release, \
3344 .read = simple_attr_read, \
3345 .write = simple_attr_write, \
3346 .llseek = generic_file_llseek, \
3349 static inline __printf(1, 2)
3350 void __simple_attr_check_format(const char *fmt, ...)
3352 /* don't do anything, just let the compiler check the arguments; */
3355 int simple_attr_open(struct inode *inode, struct file *file,
3356 int (*get)(void *, u64 *), int (*set)(void *, u64),
3358 int simple_attr_release(struct inode *inode, struct file *file);
3359 ssize_t simple_attr_read(struct file *file, char __user *buf,
3360 size_t len, loff_t *ppos);
3361 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3362 size_t len, loff_t *ppos);
3365 int __init list_bdev_fs_names(char *buf, size_t size);
3367 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3368 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3370 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3371 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3372 (flag & __FMODE_NONOTIFY)))
3374 static inline bool is_sxid(umode_t mode)
3376 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3379 static inline int check_sticky(struct user_namespace *mnt_userns,
3380 struct inode *dir, struct inode *inode)
3382 if (!(dir->i_mode & S_ISVTX))
3385 return __check_sticky(mnt_userns, dir, inode);
3388 static inline void inode_has_no_xattr(struct inode *inode)
3390 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3391 inode->i_flags |= S_NOSEC;
3394 static inline bool is_root_inode(struct inode *inode)
3396 return inode == inode->i_sb->s_root->d_inode;
3399 static inline bool dir_emit(struct dir_context *ctx,
3400 const char *name, int namelen,
3401 u64 ino, unsigned type)
3403 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3405 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3407 return ctx->actor(ctx, ".", 1, ctx->pos,
3408 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3410 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3412 return ctx->actor(ctx, "..", 2, ctx->pos,
3413 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3415 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3417 if (ctx->pos == 0) {
3418 if (!dir_emit_dot(file, ctx))
3422 if (ctx->pos == 1) {
3423 if (!dir_emit_dotdot(file, ctx))
3429 static inline bool dir_relax(struct inode *inode)
3431 inode_unlock(inode);
3433 return !IS_DEADDIR(inode);
3436 static inline bool dir_relax_shared(struct inode *inode)
3438 inode_unlock_shared(inode);
3439 inode_lock_shared(inode);
3440 return !IS_DEADDIR(inode);
3443 extern bool path_noexec(const struct path *path);
3444 extern void inode_nohighmem(struct inode *inode);
3447 extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len,
3449 extern int generic_fadvise(struct file *file, loff_t offset, loff_t len,
3452 #endif /* _LINUX_FS_H */