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/fiemap.h>
28 #include <linux/rculist_bl.h>
29 #include <linux/atomic.h>
30 #include <linux/shrinker.h>
31 #include <linux/migrate_mode.h>
32 #include <linux/uidgid.h>
33 #include <linux/lockdep.h>
34 #include <linux/percpu-rwsem.h>
35 #include <linux/workqueue.h>
36 #include <linux/delayed_call.h>
37 #include <linux/uuid.h>
38 #include <linux/errseq.h>
40 #include <asm/byteorder.h>
41 #include <uapi/linux/fs.h>
43 struct backing_dev_info;
46 struct export_operations;
51 struct pipe_inode_info;
52 struct poll_table_struct;
54 struct vm_area_struct;
57 struct swap_info_struct;
59 struct workqueue_struct;
62 struct fscrypt_operations;
64 extern void __init inode_init(void);
65 extern void __init inode_init_early(void);
66 extern void __init files_init(void);
67 extern void __init files_maxfiles_init(void);
69 extern struct files_stat_struct files_stat;
70 extern unsigned long get_max_files(void);
71 extern unsigned int sysctl_nr_open;
72 extern struct inodes_stat_t inodes_stat;
73 extern int leases_enable, lease_break_time;
74 extern int sysctl_protected_symlinks;
75 extern int sysctl_protected_hardlinks;
77 typedef __kernel_rwf_t rwf_t;
80 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
81 struct buffer_head *bh_result, int create);
82 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
83 ssize_t bytes, void *private);
85 #define MAY_EXEC 0x00000001
86 #define MAY_WRITE 0x00000002
87 #define MAY_READ 0x00000004
88 #define MAY_APPEND 0x00000008
89 #define MAY_ACCESS 0x00000010
90 #define MAY_OPEN 0x00000020
91 #define MAY_CHDIR 0x00000040
92 /* called from RCU mode, don't block */
93 #define MAY_NOT_BLOCK 0x00000080
96 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
97 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
100 /* file is open for reading */
101 #define FMODE_READ ((__force fmode_t)0x1)
102 /* file is open for writing */
103 #define FMODE_WRITE ((__force fmode_t)0x2)
104 /* file is seekable */
105 #define FMODE_LSEEK ((__force fmode_t)0x4)
106 /* file can be accessed using pread */
107 #define FMODE_PREAD ((__force fmode_t)0x8)
108 /* file can be accessed using pwrite */
109 #define FMODE_PWRITE ((__force fmode_t)0x10)
110 /* File is opened for execution with sys_execve / sys_uselib */
111 #define FMODE_EXEC ((__force fmode_t)0x20)
112 /* File is opened with O_NDELAY (only set for block devices) */
113 #define FMODE_NDELAY ((__force fmode_t)0x40)
114 /* File is opened with O_EXCL (only set for block devices) */
115 #define FMODE_EXCL ((__force fmode_t)0x80)
116 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
117 (specialy hack for floppy.c) */
118 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
119 /* 32bit hashes as llseek() offset (for directories) */
120 #define FMODE_32BITHASH ((__force fmode_t)0x200)
121 /* 64bit hashes as llseek() offset (for directories) */
122 #define FMODE_64BITHASH ((__force fmode_t)0x400)
125 * Don't update ctime and mtime.
127 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
128 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
130 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
132 /* Expect random access pattern */
133 #define FMODE_RANDOM ((__force fmode_t)0x1000)
135 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
136 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
138 /* File is opened with O_PATH; almost nothing can be done with it */
139 #define FMODE_PATH ((__force fmode_t)0x4000)
141 /* File needs atomic accesses to f_pos */
142 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
143 /* Write access to underlying fs */
144 #define FMODE_WRITER ((__force fmode_t)0x10000)
145 /* Has read method(s) */
146 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
147 /* Has write method(s) */
148 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
150 /* File was opened by fanotify and shouldn't generate fanotify events */
151 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
153 /* File is capable of returning -EAGAIN if I/O will block */
154 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
157 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
158 * that indicates that they should check the contents of the iovec are
159 * valid, but not check the memory that the iovec elements
162 #define CHECK_IOVEC_ONLY -1
165 * Attribute flags. These should be or-ed together to figure out what
168 #define ATTR_MODE (1 << 0)
169 #define ATTR_UID (1 << 1)
170 #define ATTR_GID (1 << 2)
171 #define ATTR_SIZE (1 << 3)
172 #define ATTR_ATIME (1 << 4)
173 #define ATTR_MTIME (1 << 5)
174 #define ATTR_CTIME (1 << 6)
175 #define ATTR_ATIME_SET (1 << 7)
176 #define ATTR_MTIME_SET (1 << 8)
177 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
178 #define ATTR_ATTR_FLAG (1 << 10)
179 #define ATTR_KILL_SUID (1 << 11)
180 #define ATTR_KILL_SGID (1 << 12)
181 #define ATTR_FILE (1 << 13)
182 #define ATTR_KILL_PRIV (1 << 14)
183 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
184 #define ATTR_TIMES_SET (1 << 16)
185 #define ATTR_TOUCH (1 << 17)
188 * Whiteout is represented by a char device. The following constants define the
189 * mode and device number to use.
191 #define WHITEOUT_MODE 0
192 #define WHITEOUT_DEV 0
195 * This is the Inode Attributes structure, used for notify_change(). It
196 * uses the above definitions as flags, to know which values have changed.
197 * Also, in this manner, a Filesystem can look at only the values it cares
198 * about. Basically, these are the attributes that the VFS layer can
199 * request to change from the FS layer.
201 * Derek Atkins <warlord@MIT.EDU> 94-10-20
204 unsigned int ia_valid;
209 struct timespec64 ia_atime;
210 struct timespec64 ia_mtime;
211 struct timespec64 ia_ctime;
214 * Not an attribute, but an auxiliary info for filesystems wanting to
215 * implement an ftruncate() like method. NOTE: filesystem should
216 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
218 struct file *ia_file;
222 * Includes for diskquotas.
224 #include <linux/quota.h>
227 * Maximum number of layers of fs stack. Needs to be limited to
228 * prevent kernel stack overflow
230 #define FILESYSTEM_MAX_STACK_DEPTH 2
233 * enum positive_aop_returns - aop return codes with specific semantics
235 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
236 * completed, that the page is still locked, and
237 * should be considered active. The VM uses this hint
238 * to return the page to the active list -- it won't
239 * be a candidate for writeback again in the near
240 * future. Other callers must be careful to unlock
241 * the page if they get this return. Returned by
244 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
245 * unlocked it and the page might have been truncated.
246 * The caller should back up to acquiring a new page and
247 * trying again. The aop will be taking reasonable
248 * precautions not to livelock. If the caller held a page
249 * reference, it should drop it before retrying. Returned
252 * address_space_operation functions return these large constants to indicate
253 * special semantics to the caller. These are much larger than the bytes in a
254 * page to allow for functions that return the number of bytes operated on in a
258 enum positive_aop_returns {
259 AOP_WRITEPAGE_ACTIVATE = 0x80000,
260 AOP_TRUNCATED_PAGE = 0x80001,
263 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
264 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
265 * helper code (eg buffer layer)
266 * to clear GFP_FS from alloc */
269 * oh the beauties of C type declarations.
272 struct address_space;
273 struct writeback_control;
276 * Write life time hint values.
279 WRITE_LIFE_NOT_SET = 0,
280 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
281 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
282 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
283 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
284 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
287 #define IOCB_EVENTFD (1 << 0)
288 #define IOCB_APPEND (1 << 1)
289 #define IOCB_DIRECT (1 << 2)
290 #define IOCB_HIPRI (1 << 3)
291 #define IOCB_DSYNC (1 << 4)
292 #define IOCB_SYNC (1 << 5)
293 #define IOCB_WRITE (1 << 6)
294 #define IOCB_NOWAIT (1 << 7)
297 struct file *ki_filp;
299 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
302 enum rw_hint ki_hint;
303 } __randomize_layout;
305 static inline bool is_sync_kiocb(struct kiocb *kiocb)
307 return kiocb->ki_complete == NULL;
311 * "descriptor" for what we're up to with a read.
312 * This allows us to use the same read code yet
313 * have multiple different users of the data that
314 * we read from a file.
316 * The simplest case just copies the data to user
329 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
330 unsigned long, unsigned long);
332 struct address_space_operations {
333 int (*writepage)(struct page *page, struct writeback_control *wbc);
334 int (*readpage)(struct file *, struct page *);
336 /* Write back some dirty pages from this mapping. */
337 int (*writepages)(struct address_space *, struct writeback_control *);
339 /* Set a page dirty. Return true if this dirtied it */
340 int (*set_page_dirty)(struct page *page);
342 int (*readpages)(struct file *filp, struct address_space *mapping,
343 struct list_head *pages, unsigned nr_pages);
345 int (*write_begin)(struct file *, struct address_space *mapping,
346 loff_t pos, unsigned len, unsigned flags,
347 struct page **pagep, void **fsdata);
348 int (*write_end)(struct file *, struct address_space *mapping,
349 loff_t pos, unsigned len, unsigned copied,
350 struct page *page, void *fsdata);
352 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
353 sector_t (*bmap)(struct address_space *, sector_t);
354 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
355 int (*releasepage) (struct page *, gfp_t);
356 void (*freepage)(struct page *);
357 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
359 * migrate the contents of a page to the specified target. If
360 * migrate_mode is MIGRATE_ASYNC, it must not block.
362 int (*migratepage) (struct address_space *,
363 struct page *, struct page *, enum migrate_mode);
364 bool (*isolate_page)(struct page *, isolate_mode_t);
365 void (*putback_page)(struct page *);
366 int (*launder_page) (struct page *);
367 int (*is_partially_uptodate) (struct page *, unsigned long,
369 void (*is_dirty_writeback) (struct page *, bool *, bool *);
370 int (*error_remove_page)(struct address_space *, struct page *);
372 /* swapfile support */
373 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
375 void (*swap_deactivate)(struct file *file);
378 extern const struct address_space_operations empty_aops;
381 * pagecache_write_begin/pagecache_write_end must be used by general code
382 * to write into the pagecache.
384 int pagecache_write_begin(struct file *, struct address_space *mapping,
385 loff_t pos, unsigned len, unsigned flags,
386 struct page **pagep, void **fsdata);
388 int pagecache_write_end(struct file *, struct address_space *mapping,
389 loff_t pos, unsigned len, unsigned copied,
390 struct page *page, void *fsdata);
392 struct address_space {
393 struct inode *host; /* owner: inode, block_device */
394 struct radix_tree_root i_pages; /* cached pages */
395 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
396 struct rb_root_cached i_mmap; /* tree of private and shared mappings */
397 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
398 /* Protected by the i_pages lock */
399 unsigned long nrpages; /* number of total pages */
400 /* number of shadow or DAX exceptional entries */
401 unsigned long nrexceptional;
402 pgoff_t writeback_index;/* writeback starts here */
403 const struct address_space_operations *a_ops; /* methods */
404 unsigned long flags; /* error bits */
405 spinlock_t private_lock; /* for use by the address_space */
406 gfp_t gfp_mask; /* implicit gfp mask for allocations */
407 struct list_head private_list; /* for use by the address_space */
408 void *private_data; /* ditto */
410 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
412 * On most architectures that alignment is already the case; but
413 * must be enforced here for CRIS, to let the least significant bit
414 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
416 struct request_queue;
418 struct block_device {
419 dev_t bd_dev; /* not a kdev_t - it's a search key */
421 struct inode * bd_inode; /* will die */
422 struct super_block * bd_super;
423 struct mutex bd_mutex; /* open/close mutex */
427 bool bd_write_holder;
429 struct list_head bd_holder_disks;
431 struct block_device * bd_contains;
432 unsigned bd_block_size;
434 struct hd_struct * bd_part;
435 /* number of times partitions within this device have been opened. */
436 unsigned bd_part_count;
438 struct gendisk * bd_disk;
439 struct request_queue * bd_queue;
440 struct backing_dev_info *bd_bdi;
441 struct list_head bd_list;
443 * Private data. You must have bd_claim'ed the block_device
444 * to use this. NOTE: bd_claim allows an owner to claim
445 * the same device multiple times, the owner must take special
446 * care to not mess up bd_private for that case.
448 unsigned long bd_private;
450 /* The counter of freeze processes */
451 int bd_fsfreeze_count;
452 /* Mutex for freeze */
453 struct mutex bd_fsfreeze_mutex;
454 } __randomize_layout;
457 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
460 #define PAGECACHE_TAG_DIRTY 0
461 #define PAGECACHE_TAG_WRITEBACK 1
462 #define PAGECACHE_TAG_TOWRITE 2
464 int mapping_tagged(struct address_space *mapping, int tag);
466 static inline void i_mmap_lock_write(struct address_space *mapping)
468 down_write(&mapping->i_mmap_rwsem);
471 static inline void i_mmap_unlock_write(struct address_space *mapping)
473 up_write(&mapping->i_mmap_rwsem);
476 static inline void i_mmap_lock_read(struct address_space *mapping)
478 down_read(&mapping->i_mmap_rwsem);
481 static inline void i_mmap_unlock_read(struct address_space *mapping)
483 up_read(&mapping->i_mmap_rwsem);
487 * Might pages of this file be mapped into userspace?
489 static inline int mapping_mapped(struct address_space *mapping)
491 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
495 * Might pages of this file have been modified in userspace?
496 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
497 * marks vma as VM_SHARED if it is shared, and the file was opened for
498 * writing i.e. vma may be mprotected writable even if now readonly.
500 * If i_mmap_writable is negative, no new writable mappings are allowed. You
501 * can only deny writable mappings, if none exists right now.
503 static inline int mapping_writably_mapped(struct address_space *mapping)
505 return atomic_read(&mapping->i_mmap_writable) > 0;
508 static inline int mapping_map_writable(struct address_space *mapping)
510 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
514 static inline void mapping_unmap_writable(struct address_space *mapping)
516 atomic_dec(&mapping->i_mmap_writable);
519 static inline int mapping_deny_writable(struct address_space *mapping)
521 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
525 static inline void mapping_allow_writable(struct address_space *mapping)
527 atomic_inc(&mapping->i_mmap_writable);
531 * Use sequence counter to get consistent i_size on 32-bit processors.
533 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
534 #include <linux/seqlock.h>
535 #define __NEED_I_SIZE_ORDERED
536 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
538 #define i_size_ordered_init(inode) do { } while (0)
542 #define ACL_NOT_CACHED ((void *)(-1))
543 #define ACL_DONT_CACHE ((void *)(-3))
545 static inline struct posix_acl *
546 uncached_acl_sentinel(struct task_struct *task)
548 return (void *)task + 1;
552 is_uncached_acl(struct posix_acl *acl)
554 return (long)acl & 1;
557 #define IOP_FASTPERM 0x0001
558 #define IOP_LOOKUP 0x0002
559 #define IOP_NOFOLLOW 0x0004
560 #define IOP_XATTR 0x0008
561 #define IOP_DEFAULT_READLINK 0x0010
563 struct fsnotify_mark_connector;
566 * Keep mostly read-only and often accessed (especially for
567 * the RCU path lookup and 'stat' data) fields at the beginning
568 * of the 'struct inode'
572 unsigned short i_opflags;
575 unsigned int i_flags;
577 #ifdef CONFIG_FS_POSIX_ACL
578 struct posix_acl *i_acl;
579 struct posix_acl *i_default_acl;
582 const struct inode_operations *i_op;
583 struct super_block *i_sb;
584 struct address_space *i_mapping;
586 #ifdef CONFIG_SECURITY
590 /* Stat data, not accessed from path walking */
593 * Filesystems may only read i_nlink directly. They shall use the
594 * following functions for modification:
596 * (set|clear|inc|drop)_nlink
597 * inode_(inc|dec)_link_count
600 const unsigned int i_nlink;
601 unsigned int __i_nlink;
605 struct timespec64 i_atime;
606 struct timespec64 i_mtime;
607 struct timespec64 i_ctime;
608 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
609 unsigned short i_bytes;
610 unsigned int i_blkbits;
611 enum rw_hint i_write_hint;
614 #ifdef __NEED_I_SIZE_ORDERED
615 seqcount_t i_size_seqcount;
619 unsigned long i_state;
620 struct rw_semaphore i_rwsem;
622 unsigned long dirtied_when; /* jiffies of first dirtying */
623 unsigned long dirtied_time_when;
625 struct hlist_node i_hash;
626 struct list_head i_io_list; /* backing dev IO list */
627 #ifdef CONFIG_CGROUP_WRITEBACK
628 struct bdi_writeback *i_wb; /* the associated cgroup wb */
630 /* foreign inode detection, see wbc_detach_inode() */
632 u16 i_wb_frn_avg_time;
633 u16 i_wb_frn_history;
635 struct list_head i_lru; /* inode LRU list */
636 struct list_head i_sb_list;
637 struct list_head i_wb_list; /* backing dev writeback list */
639 struct hlist_head i_dentry;
640 struct rcu_head i_rcu;
642 atomic64_t i_version;
644 atomic_t i_dio_count;
645 atomic_t i_writecount;
647 atomic_t i_readcount; /* struct files open RO */
649 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
650 struct file_lock_context *i_flctx;
651 struct address_space i_data;
652 struct list_head i_devices;
654 struct pipe_inode_info *i_pipe;
655 struct block_device *i_bdev;
663 #ifdef CONFIG_FSNOTIFY
664 __u32 i_fsnotify_mask; /* all events this inode cares about */
665 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
668 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
669 struct fscrypt_info *i_crypt_info;
672 void *i_private; /* fs or device private pointer */
673 } __randomize_layout;
675 static inline unsigned int i_blocksize(const struct inode *node)
677 return (1 << node->i_blkbits);
680 static inline int inode_unhashed(struct inode *inode)
682 return hlist_unhashed(&inode->i_hash);
686 * inode->i_mutex nesting subclasses for the lock validator:
688 * 0: the object of the current VFS operation
692 * 4: second non-directory
693 * 5: second parent (when locking independent directories in rename)
695 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
696 * non-directories at once.
698 * The locking order between these classes is
699 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
701 enum inode_i_mutex_lock_class
711 static inline void inode_lock(struct inode *inode)
713 down_write(&inode->i_rwsem);
716 static inline void inode_unlock(struct inode *inode)
718 up_write(&inode->i_rwsem);
721 static inline void inode_lock_shared(struct inode *inode)
723 down_read(&inode->i_rwsem);
726 static inline void inode_unlock_shared(struct inode *inode)
728 up_read(&inode->i_rwsem);
731 static inline int inode_trylock(struct inode *inode)
733 return down_write_trylock(&inode->i_rwsem);
736 static inline int inode_trylock_shared(struct inode *inode)
738 return down_read_trylock(&inode->i_rwsem);
741 static inline int inode_is_locked(struct inode *inode)
743 return rwsem_is_locked(&inode->i_rwsem);
746 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
748 down_write_nested(&inode->i_rwsem, subclass);
751 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
753 down_read_nested(&inode->i_rwsem, subclass);
756 void lock_two_nondirectories(struct inode *, struct inode*);
757 void unlock_two_nondirectories(struct inode *, struct inode*);
760 * NOTE: in a 32bit arch with a preemptable kernel and
761 * an UP compile the i_size_read/write must be atomic
762 * with respect to the local cpu (unlike with preempt disabled),
763 * but they don't need to be atomic with respect to other cpus like in
764 * true SMP (so they need either to either locally disable irq around
765 * the read or for example on x86 they can be still implemented as a
766 * cmpxchg8b without the need of the lock prefix). For SMP compiles
767 * and 64bit archs it makes no difference if preempt is enabled or not.
769 static inline loff_t i_size_read(const struct inode *inode)
771 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
776 seq = read_seqcount_begin(&inode->i_size_seqcount);
777 i_size = inode->i_size;
778 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
780 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
784 i_size = inode->i_size;
788 return inode->i_size;
793 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
794 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
795 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
797 static inline void i_size_write(struct inode *inode, loff_t i_size)
799 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
801 write_seqcount_begin(&inode->i_size_seqcount);
802 inode->i_size = i_size;
803 write_seqcount_end(&inode->i_size_seqcount);
805 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
807 inode->i_size = i_size;
810 inode->i_size = i_size;
814 static inline unsigned iminor(const struct inode *inode)
816 return MINOR(inode->i_rdev);
819 static inline unsigned imajor(const struct inode *inode)
821 return MAJOR(inode->i_rdev);
824 extern struct block_device *I_BDEV(struct inode *inode);
827 rwlock_t lock; /* protects pid, uid, euid fields */
828 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
829 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
830 kuid_t uid, euid; /* uid/euid of process setting the owner */
831 int signum; /* posix.1b rt signal to be delivered on IO */
835 * Track a single file's readahead state
837 struct file_ra_state {
838 pgoff_t start; /* where readahead started */
839 unsigned int size; /* # of readahead pages */
840 unsigned int async_size; /* do asynchronous readahead when
841 there are only # of pages ahead */
843 unsigned int ra_pages; /* Maximum readahead window */
844 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
845 loff_t prev_pos; /* Cache last read() position */
849 * Check if @index falls in the readahead windows.
851 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
853 return (index >= ra->start &&
854 index < ra->start + ra->size);
859 struct llist_node fu_llist;
860 struct rcu_head fu_rcuhead;
863 struct inode *f_inode; /* cached value */
864 const struct file_operations *f_op;
867 * Protects f_ep_links, f_flags.
868 * Must not be taken from IRQ context.
871 enum rw_hint f_write_hint;
872 atomic_long_t f_count;
873 unsigned int f_flags;
875 struct mutex f_pos_lock;
877 struct fown_struct f_owner;
878 const struct cred *f_cred;
879 struct file_ra_state f_ra;
882 #ifdef CONFIG_SECURITY
885 /* needed for tty driver, and maybe others */
889 /* Used by fs/eventpoll.c to link all the hooks to this file */
890 struct list_head f_ep_links;
891 struct list_head f_tfile_llink;
892 #endif /* #ifdef CONFIG_EPOLL */
893 struct address_space *f_mapping;
896 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
901 /* file identifier */
902 unsigned char f_handle[0];
905 static inline struct file *get_file(struct file *f)
907 atomic_long_inc(&f->f_count);
910 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
911 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
912 #define file_count(x) atomic_long_read(&(x)->f_count)
914 #define MAX_NON_LFS ((1UL<<31) - 1)
916 /* Page cache limit. The filesystems should put that into their s_maxbytes
917 limits, otherwise bad things can happen in VM. */
918 #if BITS_PER_LONG==32
919 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
920 #elif BITS_PER_LONG==64
921 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
926 #define FL_DELEG 4 /* NFSv4 delegation */
927 #define FL_ACCESS 8 /* not trying to lock, just looking */
928 #define FL_EXISTS 16 /* when unlocking, test for existence */
929 #define FL_LEASE 32 /* lease held on this file */
930 #define FL_CLOSE 64 /* unlock on close */
931 #define FL_SLEEP 128 /* A blocking lock */
932 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
933 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
934 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
935 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
937 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
940 * Special return value from posix_lock_file() and vfs_lock_file() for
941 * asynchronous locking.
943 #define FILE_LOCK_DEFERRED 1
945 /* legacy typedef, should eventually be removed */
946 typedef void *fl_owner_t;
950 struct file_lock_operations {
951 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
952 void (*fl_release_private)(struct file_lock *);
955 struct lock_manager_operations {
956 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
957 unsigned long (*lm_owner_key)(struct file_lock *);
958 fl_owner_t (*lm_get_owner)(fl_owner_t);
959 void (*lm_put_owner)(fl_owner_t);
960 void (*lm_notify)(struct file_lock *); /* unblock callback */
961 int (*lm_grant)(struct file_lock *, int);
962 bool (*lm_break)(struct file_lock *);
963 int (*lm_change)(struct file_lock *, int, struct list_head *);
964 void (*lm_setup)(struct file_lock *, void **);
967 struct lock_manager {
968 struct list_head list;
970 * NFSv4 and up also want opens blocked during the grace period;
977 void locks_start_grace(struct net *, struct lock_manager *);
978 void locks_end_grace(struct lock_manager *);
979 bool locks_in_grace(struct net *);
980 bool opens_in_grace(struct net *);
982 /* that will die - we need it for nfs_lock_info */
983 #include <linux/nfs_fs_i.h>
986 * struct file_lock represents a generic "file lock". It's used to represent
987 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
988 * note that the same struct is used to represent both a request for a lock and
989 * the lock itself, but the same object is never used for both.
991 * FIXME: should we create a separate "struct lock_request" to help distinguish
994 * The varous i_flctx lists are ordered by:
997 * 2) lock range start
1000 * Obviously, the last two criteria only matter for POSIX locks.
1003 struct file_lock *fl_next; /* singly linked list for this inode */
1004 struct list_head fl_list; /* link into file_lock_context */
1005 struct hlist_node fl_link; /* node in global lists */
1006 struct list_head fl_block; /* circular list of blocked processes */
1007 fl_owner_t fl_owner;
1008 unsigned int fl_flags;
1009 unsigned char fl_type;
1010 unsigned int fl_pid;
1011 int fl_link_cpu; /* what cpu's list is this on? */
1012 wait_queue_head_t fl_wait;
1013 struct file *fl_file;
1017 struct fasync_struct * fl_fasync; /* for lease break notifications */
1018 /* for lease breaks: */
1019 unsigned long fl_break_time;
1020 unsigned long fl_downgrade_time;
1022 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1023 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1025 struct nfs_lock_info nfs_fl;
1026 struct nfs4_lock_info nfs4_fl;
1028 struct list_head link; /* link in AFS vnode's pending_locks list */
1029 int state; /* state of grant or error if -ve */
1032 } __randomize_layout;
1034 struct file_lock_context {
1035 spinlock_t flc_lock;
1036 struct list_head flc_flock;
1037 struct list_head flc_posix;
1038 struct list_head flc_lease;
1041 /* The following constant reflects the upper bound of the file/locking space */
1043 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1044 #define OFFSET_MAX INT_LIMIT(loff_t)
1045 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1048 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1051 * Return the inode to use for locking
1053 * For overlayfs this should be the overlay inode, not the real inode returned
1054 * by file_inode(). For any other fs file_inode(filp) and locks_inode(filp) are
1057 static inline struct inode *locks_inode(const struct file *f)
1059 return f->f_path.dentry->d_inode;
1062 #ifdef CONFIG_FILE_LOCKING
1063 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1064 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1067 #if BITS_PER_LONG == 32
1068 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1069 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1073 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1074 extern int fcntl_getlease(struct file *filp);
1077 void locks_free_lock_context(struct inode *inode);
1078 void locks_free_lock(struct file_lock *fl);
1079 extern void locks_init_lock(struct file_lock *);
1080 extern struct file_lock * locks_alloc_lock(void);
1081 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1082 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1083 extern void locks_remove_posix(struct file *, fl_owner_t);
1084 extern void locks_remove_file(struct file *);
1085 extern void locks_release_private(struct file_lock *);
1086 extern void posix_test_lock(struct file *, struct file_lock *);
1087 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1088 extern int posix_unblock_lock(struct file_lock *);
1089 extern int vfs_test_lock(struct file *, struct file_lock *);
1090 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1091 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1092 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1093 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1094 extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1095 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1096 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1097 extern int lease_modify(struct file_lock *, int, struct list_head *);
1098 struct files_struct;
1099 extern void show_fd_locks(struct seq_file *f,
1100 struct file *filp, struct files_struct *files);
1101 #else /* !CONFIG_FILE_LOCKING */
1102 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1103 struct flock __user *user)
1108 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1109 unsigned int cmd, struct flock __user *user)
1114 #if BITS_PER_LONG == 32
1115 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1116 struct flock64 __user *user)
1121 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1122 unsigned int cmd, struct flock64 __user *user)
1127 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1132 static inline int fcntl_getlease(struct file *filp)
1138 locks_free_lock_context(struct inode *inode)
1142 static inline void locks_init_lock(struct file_lock *fl)
1147 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1152 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1157 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1162 static inline void locks_remove_file(struct file *filp)
1167 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1172 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1173 struct file_lock *conflock)
1178 static inline int posix_unblock_lock(struct file_lock *waiter)
1183 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1188 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1189 struct file_lock *fl, struct file_lock *conf)
1194 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1199 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1204 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1209 static inline void lease_get_mtime(struct inode *inode,
1210 struct timespec64 *time)
1215 static inline int generic_setlease(struct file *filp, long arg,
1216 struct file_lock **flp, void **priv)
1221 static inline int vfs_setlease(struct file *filp, long arg,
1222 struct file_lock **lease, void **priv)
1227 static inline int lease_modify(struct file_lock *fl, int arg,
1228 struct list_head *dispose)
1233 struct files_struct;
1234 static inline void show_fd_locks(struct seq_file *f,
1235 struct file *filp, struct files_struct *files) {}
1236 #endif /* !CONFIG_FILE_LOCKING */
1238 static inline struct inode *file_inode(const struct file *f)
1243 static inline struct dentry *file_dentry(const struct file *file)
1245 return d_real(file->f_path.dentry, file_inode(file), 0, 0);
1248 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1250 return locks_lock_inode_wait(locks_inode(filp), fl);
1253 struct fasync_struct {
1257 struct fasync_struct *fa_next; /* singly linked list */
1258 struct file *fa_file;
1259 struct rcu_head fa_rcu;
1262 #define FASYNC_MAGIC 0x4601
1264 /* SMP safe fasync helpers: */
1265 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1266 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1267 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1268 extern struct fasync_struct *fasync_alloc(void);
1269 extern void fasync_free(struct fasync_struct *);
1271 /* can be called from interrupts */
1272 extern void kill_fasync(struct fasync_struct **, int, int);
1274 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1275 extern int f_setown(struct file *filp, unsigned long arg, int force);
1276 extern void f_delown(struct file *filp);
1277 extern pid_t f_getown(struct file *filp);
1278 extern int send_sigurg(struct fown_struct *fown);
1281 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1282 * represented in both.
1284 #define SB_RDONLY 1 /* Mount read-only */
1285 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1286 #define SB_NODEV 4 /* Disallow access to device special files */
1287 #define SB_NOEXEC 8 /* Disallow program execution */
1288 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1289 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1290 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1291 #define SB_NOATIME 1024 /* Do not update access times. */
1292 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1293 #define SB_SILENT 32768
1294 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1295 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1296 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1297 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1299 /* These sb flags are internal to the kernel */
1300 #define SB_SUBMOUNT (1<<26)
1301 #define SB_NOREMOTELOCK (1<<27)
1302 #define SB_NOSEC (1<<28)
1303 #define SB_BORN (1<<29)
1304 #define SB_ACTIVE (1<<30)
1305 #define SB_NOUSER (1<<31)
1311 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1312 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1313 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1314 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1315 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1318 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1319 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1320 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1321 #define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */
1323 /* sb->s_iflags to limit user namespace mounts */
1324 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1325 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1326 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1328 /* Possible states of 'frozen' field */
1330 SB_UNFROZEN = 0, /* FS is unfrozen */
1331 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1332 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1333 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1334 * internal threads if needed) */
1335 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1338 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1341 int frozen; /* Is sb frozen? */
1342 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1343 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1346 struct super_block {
1347 struct list_head s_list; /* Keep this first */
1348 dev_t s_dev; /* search index; _not_ kdev_t */
1349 unsigned char s_blocksize_bits;
1350 unsigned long s_blocksize;
1351 loff_t s_maxbytes; /* Max file size */
1352 struct file_system_type *s_type;
1353 const struct super_operations *s_op;
1354 const struct dquot_operations *dq_op;
1355 const struct quotactl_ops *s_qcop;
1356 const struct export_operations *s_export_op;
1357 unsigned long s_flags;
1358 unsigned long s_iflags; /* internal SB_I_* flags */
1359 unsigned long s_magic;
1360 struct dentry *s_root;
1361 struct rw_semaphore s_umount;
1364 #ifdef CONFIG_SECURITY
1367 const struct xattr_handler **s_xattr;
1369 const struct fscrypt_operations *s_cop;
1371 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1372 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1373 struct block_device *s_bdev;
1374 struct backing_dev_info *s_bdi;
1375 struct mtd_info *s_mtd;
1376 struct hlist_node s_instances;
1377 unsigned int s_quota_types; /* Bitmask of supported quota types */
1378 struct quota_info s_dquot; /* Diskquota specific options */
1380 struct sb_writers s_writers;
1382 char s_id[32]; /* Informational name */
1383 uuid_t s_uuid; /* UUID */
1385 void *s_fs_info; /* Filesystem private info */
1386 unsigned int s_max_links;
1389 /* Granularity of c/m/atime in ns.
1390 Cannot be worse than a second */
1394 * The next field is for VFS *only*. No filesystems have any business
1395 * even looking at it. You had been warned.
1397 struct mutex s_vfs_rename_mutex; /* Kludge */
1400 * Filesystem subtype. If non-empty the filesystem type field
1401 * in /proc/mounts will be "type.subtype"
1405 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1408 * Saved pool identifier for cleancache (-1 means none)
1410 int cleancache_poolid;
1412 struct shrinker s_shrink; /* per-sb shrinker handle */
1414 /* Number of inodes with nlink == 0 but still referenced */
1415 atomic_long_t s_remove_count;
1417 /* Being remounted read-only */
1418 int s_readonly_remount;
1420 /* AIO completions deferred from interrupt context */
1421 struct workqueue_struct *s_dio_done_wq;
1422 struct hlist_head s_pins;
1425 * Owning user namespace and default context in which to
1426 * interpret filesystem uids, gids, quotas, device nodes,
1427 * xattrs and security labels.
1429 struct user_namespace *s_user_ns;
1432 * Keep the lru lists last in the structure so they always sit on their
1433 * own individual cachelines.
1435 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1436 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1437 struct rcu_head rcu;
1438 struct work_struct destroy_work;
1440 struct mutex s_sync_lock; /* sync serialisation lock */
1443 * Indicates how deep in a filesystem stack this SB is
1447 /* s_inode_list_lock protects s_inodes */
1448 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1449 struct list_head s_inodes; /* all inodes */
1451 spinlock_t s_inode_wblist_lock;
1452 struct list_head s_inodes_wb; /* writeback inodes */
1453 } __randomize_layout;
1455 /* Helper functions so that in most cases filesystems will
1456 * not need to deal directly with kuid_t and kgid_t and can
1457 * instead deal with the raw numeric values that are stored
1458 * in the filesystem.
1460 static inline uid_t i_uid_read(const struct inode *inode)
1462 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1465 static inline gid_t i_gid_read(const struct inode *inode)
1467 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1470 static inline void i_uid_write(struct inode *inode, uid_t uid)
1472 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1475 static inline void i_gid_write(struct inode *inode, gid_t gid)
1477 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1480 extern struct timespec64 timespec64_trunc(struct timespec64 t, unsigned gran);
1481 extern struct timespec64 current_time(struct inode *inode);
1484 * Snapshotting support.
1487 void __sb_end_write(struct super_block *sb, int level);
1488 int __sb_start_write(struct super_block *sb, int level, bool wait);
1490 #define __sb_writers_acquired(sb, lev) \
1491 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1492 #define __sb_writers_release(sb, lev) \
1493 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1496 * sb_end_write - drop write access to a superblock
1497 * @sb: the super we wrote to
1499 * Decrement number of writers to the filesystem. Wake up possible waiters
1500 * wanting to freeze the filesystem.
1502 static inline void sb_end_write(struct super_block *sb)
1504 __sb_end_write(sb, SB_FREEZE_WRITE);
1508 * sb_end_pagefault - drop write access to a superblock from a page fault
1509 * @sb: the super we wrote to
1511 * Decrement number of processes handling write page fault to the filesystem.
1512 * Wake up possible waiters wanting to freeze the filesystem.
1514 static inline void sb_end_pagefault(struct super_block *sb)
1516 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1520 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1521 * @sb: the super we wrote to
1523 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1524 * waiters wanting to freeze the filesystem.
1526 static inline void sb_end_intwrite(struct super_block *sb)
1528 __sb_end_write(sb, SB_FREEZE_FS);
1532 * sb_start_write - get write access to a superblock
1533 * @sb: the super we write to
1535 * When a process wants to write data or metadata to a file system (i.e. dirty
1536 * a page or an inode), it should embed the operation in a sb_start_write() -
1537 * sb_end_write() pair to get exclusion against file system freezing. This
1538 * function increments number of writers preventing freezing. If the file
1539 * system is already frozen, the function waits until the file system is
1542 * Since freeze protection behaves as a lock, users have to preserve
1543 * ordering of freeze protection and other filesystem locks. Generally,
1544 * freeze protection should be the outermost lock. In particular, we have:
1547 * -> i_mutex (write path, truncate, directory ops, ...)
1548 * -> s_umount (freeze_super, thaw_super)
1550 static inline void sb_start_write(struct super_block *sb)
1552 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1555 static inline int sb_start_write_trylock(struct super_block *sb)
1557 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1561 * sb_start_pagefault - get write access to a superblock from a page fault
1562 * @sb: the super we write to
1564 * When a process starts handling write page fault, it should embed the
1565 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1566 * exclusion against file system freezing. This is needed since the page fault
1567 * is going to dirty a page. This function increments number of running page
1568 * faults preventing freezing. If the file system is already frozen, the
1569 * function waits until the file system is thawed.
1571 * Since page fault freeze protection behaves as a lock, users have to preserve
1572 * ordering of freeze protection and other filesystem locks. It is advised to
1573 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1574 * handling code implies lock dependency:
1577 * -> sb_start_pagefault
1579 static inline void sb_start_pagefault(struct super_block *sb)
1581 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1585 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1586 * @sb: the super we write to
1588 * This is the third level of protection against filesystem freezing. It is
1589 * free for use by a filesystem. The only requirement is that it must rank
1590 * below sb_start_pagefault.
1592 * For example filesystem can call sb_start_intwrite() when starting a
1593 * transaction which somewhat eases handling of freezing for internal sources
1594 * of filesystem changes (internal fs threads, discarding preallocation on file
1597 static inline void sb_start_intwrite(struct super_block *sb)
1599 __sb_start_write(sb, SB_FREEZE_FS, true);
1603 extern bool inode_owner_or_capable(const struct inode *inode);
1606 * VFS helper functions..
1608 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1609 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1610 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1611 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1612 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1613 extern int vfs_rmdir(struct inode *, struct dentry *);
1614 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1615 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1616 extern int vfs_whiteout(struct inode *, struct dentry *);
1618 extern struct dentry *vfs_tmpfile(struct dentry *dentry, umode_t mode,
1621 int vfs_mkobj(struct dentry *, umode_t,
1622 int (*f)(struct dentry *, umode_t, void *),
1626 * VFS file helper functions.
1628 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1630 extern bool may_open_dev(const struct path *path);
1632 * VFS FS_IOC_FIEMAP helper definitions.
1634 struct fiemap_extent_info {
1635 unsigned int fi_flags; /* Flags as passed from user */
1636 unsigned int fi_extents_mapped; /* Number of mapped extents */
1637 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1638 struct fiemap_extent __user *fi_extents_start; /* Start of
1639 fiemap_extent array */
1641 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1642 u64 phys, u64 len, u32 flags);
1643 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1648 * NOTE! These match bits 12..15 of stat.st_mode
1649 * (ie "(i_mode >> 12) & 15").
1651 #define DT_UNKNOWN 0
1662 * This is the "filldir" function type, used by readdir() to let
1663 * the kernel specify what kind of dirent layout it wants to have.
1664 * This allows the kernel to read directories into kernel space or
1665 * to have different dirent layouts depending on the binary type.
1668 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1671 struct dir_context {
1676 struct block_device_operations;
1678 /* These macros are for out of kernel modules to test that
1679 * the kernel supports the unlocked_ioctl and compat_ioctl
1680 * fields in struct file_operations. */
1681 #define HAVE_COMPAT_IOCTL 1
1682 #define HAVE_UNLOCKED_IOCTL 1
1685 * These flags let !MMU mmap() govern direct device mapping vs immediate
1686 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1688 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1689 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1690 * NOMMU_MAP_READ: Can be mapped for reading
1691 * NOMMU_MAP_WRITE: Can be mapped for writing
1692 * NOMMU_MAP_EXEC: Can be mapped for execution
1694 #define NOMMU_MAP_COPY 0x00000001
1695 #define NOMMU_MAP_DIRECT 0x00000008
1696 #define NOMMU_MAP_READ VM_MAYREAD
1697 #define NOMMU_MAP_WRITE VM_MAYWRITE
1698 #define NOMMU_MAP_EXEC VM_MAYEXEC
1700 #define NOMMU_VMFLAGS \
1701 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1706 struct file_operations {
1707 struct module *owner;
1708 loff_t (*llseek) (struct file *, loff_t, int);
1709 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1710 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1711 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1712 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1713 int (*iterate) (struct file *, struct dir_context *);
1714 int (*iterate_shared) (struct file *, struct dir_context *);
1715 __poll_t (*poll) (struct file *, struct poll_table_struct *);
1716 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1717 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1718 int (*mmap) (struct file *, struct vm_area_struct *);
1719 unsigned long mmap_supported_flags;
1720 int (*open) (struct inode *, struct file *);
1721 int (*flush) (struct file *, fl_owner_t id);
1722 int (*release) (struct inode *, struct file *);
1723 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1724 int (*fasync) (int, struct file *, int);
1725 int (*lock) (struct file *, int, struct file_lock *);
1726 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1727 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1728 int (*check_flags)(int);
1729 int (*flock) (struct file *, int, struct file_lock *);
1730 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1731 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1732 int (*setlease)(struct file *, long, struct file_lock **, void **);
1733 long (*fallocate)(struct file *file, int mode, loff_t offset,
1735 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1737 unsigned (*mmap_capabilities)(struct file *);
1739 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1740 loff_t, size_t, unsigned int);
1741 int (*clone_file_range)(struct file *, loff_t, struct file *, loff_t,
1743 ssize_t (*dedupe_file_range)(struct file *, u64, u64, struct file *,
1745 } __randomize_layout;
1747 struct inode_operations {
1748 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1749 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1750 int (*permission) (struct inode *, int);
1751 struct posix_acl * (*get_acl)(struct inode *, int);
1753 int (*readlink) (struct dentry *, char __user *,int);
1755 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1756 int (*link) (struct dentry *,struct inode *,struct dentry *);
1757 int (*unlink) (struct inode *,struct dentry *);
1758 int (*symlink) (struct inode *,struct dentry *,const char *);
1759 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1760 int (*rmdir) (struct inode *,struct dentry *);
1761 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1762 int (*rename) (struct inode *, struct dentry *,
1763 struct inode *, struct dentry *, unsigned int);
1764 int (*setattr) (struct dentry *, struct iattr *);
1765 int (*getattr) (const struct path *, struct kstat *, u32, unsigned int);
1766 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1767 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1769 int (*update_time)(struct inode *, struct timespec64 *, int);
1770 int (*atomic_open)(struct inode *, struct dentry *,
1771 struct file *, unsigned open_flag,
1772 umode_t create_mode, int *opened);
1773 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1774 int (*set_acl)(struct inode *, struct posix_acl *, int);
1775 } ____cacheline_aligned;
1777 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
1778 struct iov_iter *iter)
1780 return file->f_op->read_iter(kio, iter);
1783 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
1784 struct iov_iter *iter)
1786 return file->f_op->write_iter(kio, iter);
1789 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
1791 return file->f_op->mmap(file, vma);
1794 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1795 unsigned long nr_segs, unsigned long fast_segs,
1796 struct iovec *fast_pointer,
1797 struct iovec **ret_pointer);
1799 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1800 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1801 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1802 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1803 unsigned long, loff_t *, rwf_t);
1804 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1805 loff_t, size_t, unsigned int);
1806 extern int vfs_clone_file_prep_inodes(struct inode *inode_in, loff_t pos_in,
1807 struct inode *inode_out, loff_t pos_out,
1808 u64 *len, bool is_dedupe);
1809 extern int vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1810 struct file *file_out, loff_t pos_out, u64 len);
1811 extern int vfs_dedupe_file_range_compare(struct inode *src, loff_t srcoff,
1812 struct inode *dest, loff_t destoff,
1813 loff_t len, bool *is_same);
1814 extern int vfs_dedupe_file_range(struct file *file,
1815 struct file_dedupe_range *same);
1817 struct super_operations {
1818 struct inode *(*alloc_inode)(struct super_block *sb);
1819 void (*destroy_inode)(struct inode *);
1821 void (*dirty_inode) (struct inode *, int flags);
1822 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1823 int (*drop_inode) (struct inode *);
1824 void (*evict_inode) (struct inode *);
1825 void (*put_super) (struct super_block *);
1826 int (*sync_fs)(struct super_block *sb, int wait);
1827 int (*freeze_super) (struct super_block *);
1828 int (*freeze_fs) (struct super_block *);
1829 int (*thaw_super) (struct super_block *);
1830 int (*unfreeze_fs) (struct super_block *);
1831 int (*statfs) (struct dentry *, struct kstatfs *);
1832 int (*remount_fs) (struct super_block *, int *, char *);
1833 void (*umount_begin) (struct super_block *);
1835 int (*show_options)(struct seq_file *, struct dentry *);
1836 int (*show_devname)(struct seq_file *, struct dentry *);
1837 int (*show_path)(struct seq_file *, struct dentry *);
1838 int (*show_stats)(struct seq_file *, struct dentry *);
1840 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1841 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1842 struct dquot **(*get_dquots)(struct inode *);
1844 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1845 long (*nr_cached_objects)(struct super_block *,
1846 struct shrink_control *);
1847 long (*free_cached_objects)(struct super_block *,
1848 struct shrink_control *);
1852 * Inode flags - they have no relation to superblock flags now
1854 #define S_SYNC 1 /* Writes are synced at once */
1855 #define S_NOATIME 2 /* Do not update access times */
1856 #define S_APPEND 4 /* Append-only file */
1857 #define S_IMMUTABLE 8 /* Immutable file */
1858 #define S_DEAD 16 /* removed, but still open directory */
1859 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1860 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1861 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1862 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1863 #define S_PRIVATE 512 /* Inode is fs-internal */
1864 #define S_IMA 1024 /* Inode has an associated IMA struct */
1865 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1866 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1867 #ifdef CONFIG_FS_DAX
1868 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1870 #define S_DAX 0 /* Make all the DAX code disappear */
1872 #define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */
1875 * Note that nosuid etc flags are inode-specific: setting some file-system
1876 * flags just means all the inodes inherit those flags by default. It might be
1877 * possible to override it selectively if you really wanted to with some
1878 * ioctl() that is not currently implemented.
1880 * Exception: SB_RDONLY is always applied to the entire file system.
1882 * Unfortunately, it is possible to change a filesystems flags with it mounted
1883 * with files in use. This means that all of the inodes will not have their
1884 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1885 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1887 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1889 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
1890 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
1891 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
1892 ((inode)->i_flags & S_SYNC))
1893 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
1894 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1895 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
1896 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
1897 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
1899 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1900 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1901 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1902 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
1904 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1905 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1906 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1907 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1908 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1909 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1910 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1911 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1912 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
1914 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1915 (inode)->i_rdev == WHITEOUT_DEV)
1917 static inline bool HAS_UNMAPPED_ID(struct inode *inode)
1919 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
1922 static inline enum rw_hint file_write_hint(struct file *file)
1924 if (file->f_write_hint != WRITE_LIFE_NOT_SET)
1925 return file->f_write_hint;
1927 return file_inode(file)->i_write_hint;
1930 static inline int iocb_flags(struct file *file);
1932 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
1934 *kiocb = (struct kiocb) {
1936 .ki_flags = iocb_flags(filp),
1937 .ki_hint = file_write_hint(filp),
1942 * Inode state bits. Protected by inode->i_lock
1944 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1945 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1947 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1948 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1949 * various stages of removing an inode.
1951 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1953 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1954 * fdatasync(). i_atime is the usual cause.
1955 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1956 * these changes separately from I_DIRTY_SYNC so that we
1957 * don't have to write inode on fdatasync() when only
1958 * mtime has changed in it.
1959 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1960 * I_NEW Serves as both a mutex and completion notification.
1961 * New inodes set I_NEW. If two processes both create
1962 * the same inode, one of them will release its inode and
1963 * wait for I_NEW to be released before returning.
1964 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1965 * also cause waiting on I_NEW, without I_NEW actually
1966 * being set. find_inode() uses this to prevent returning
1967 * nearly-dead inodes.
1968 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1969 * is zero. I_FREEING must be set when I_WILL_FREE is
1971 * I_FREEING Set when inode is about to be freed but still has dirty
1972 * pages or buffers attached or the inode itself is still
1974 * I_CLEAR Added by clear_inode(). In this state the inode is
1975 * clean and can be destroyed. Inode keeps I_FREEING.
1977 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1978 * prohibited for many purposes. iget() must wait for
1979 * the inode to be completely released, then create it
1980 * anew. Other functions will just ignore such inodes,
1981 * if appropriate. I_NEW is used for waiting.
1983 * I_SYNC Writeback of inode is running. The bit is set during
1984 * data writeback, and cleared with a wakeup on the bit
1985 * address once it is done. The bit is also used to pin
1986 * the inode in memory for flusher thread.
1988 * I_REFERENCED Marks the inode as recently references on the LRU list.
1990 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1992 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1993 * synchronize competing switching instances and to tell
1994 * wb stat updates to grab the i_pages lock. See
1995 * inode_switch_wb_work_fn() for details.
1997 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
1998 * and work dirs among overlayfs mounts.
2000 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2002 #define I_DIRTY_SYNC (1 << 0)
2003 #define I_DIRTY_DATASYNC (1 << 1)
2004 #define I_DIRTY_PAGES (1 << 2)
2006 #define I_NEW (1 << __I_NEW)
2007 #define I_WILL_FREE (1 << 4)
2008 #define I_FREEING (1 << 5)
2009 #define I_CLEAR (1 << 6)
2011 #define I_SYNC (1 << __I_SYNC)
2012 #define I_REFERENCED (1 << 8)
2013 #define __I_DIO_WAKEUP 9
2014 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2015 #define I_LINKABLE (1 << 10)
2016 #define I_DIRTY_TIME (1 << 11)
2017 #define __I_DIRTY_TIME_EXPIRED 12
2018 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
2019 #define I_WB_SWITCH (1 << 13)
2020 #define I_OVL_INUSE (1 << 14)
2022 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2023 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2024 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2026 extern void __mark_inode_dirty(struct inode *, int);
2027 static inline void mark_inode_dirty(struct inode *inode)
2029 __mark_inode_dirty(inode, I_DIRTY);
2032 static inline void mark_inode_dirty_sync(struct inode *inode)
2034 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2037 extern void inc_nlink(struct inode *inode);
2038 extern void drop_nlink(struct inode *inode);
2039 extern void clear_nlink(struct inode *inode);
2040 extern void set_nlink(struct inode *inode, unsigned int nlink);
2042 static inline void inode_inc_link_count(struct inode *inode)
2045 mark_inode_dirty(inode);
2048 static inline void inode_dec_link_count(struct inode *inode)
2051 mark_inode_dirty(inode);
2054 enum file_time_flags {
2061 extern void touch_atime(const struct path *);
2062 static inline void file_accessed(struct file *file)
2064 if (!(file->f_flags & O_NOATIME))
2065 touch_atime(&file->f_path);
2068 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2069 int sync_inode_metadata(struct inode *inode, int wait);
2071 struct file_system_type {
2074 #define FS_REQUIRES_DEV 1
2075 #define FS_BINARY_MOUNTDATA 2
2076 #define FS_HAS_SUBTYPE 4
2077 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2078 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2079 struct dentry *(*mount) (struct file_system_type *, int,
2080 const char *, void *);
2081 void (*kill_sb) (struct super_block *);
2082 struct module *owner;
2083 struct file_system_type * next;
2084 struct hlist_head fs_supers;
2086 struct lock_class_key s_lock_key;
2087 struct lock_class_key s_umount_key;
2088 struct lock_class_key s_vfs_rename_key;
2089 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2091 struct lock_class_key i_lock_key;
2092 struct lock_class_key i_mutex_key;
2093 struct lock_class_key i_mutex_dir_key;
2096 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2098 extern struct dentry *mount_ns(struct file_system_type *fs_type,
2099 int flags, void *data, void *ns, struct user_namespace *user_ns,
2100 int (*fill_super)(struct super_block *, void *, int));
2102 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2103 int flags, const char *dev_name, void *data,
2104 int (*fill_super)(struct super_block *, void *, int));
2106 static inline struct dentry *mount_bdev(struct file_system_type *fs_type,
2107 int flags, const char *dev_name, void *data,
2108 int (*fill_super)(struct super_block *, void *, int))
2110 return ERR_PTR(-ENODEV);
2113 extern struct dentry *mount_single(struct file_system_type *fs_type,
2114 int flags, void *data,
2115 int (*fill_super)(struct super_block *, void *, int));
2116 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2117 int flags, void *data,
2118 int (*fill_super)(struct super_block *, void *, int));
2119 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2120 void generic_shutdown_super(struct super_block *sb);
2122 void kill_block_super(struct super_block *sb);
2124 static inline void kill_block_super(struct super_block *sb)
2129 void kill_anon_super(struct super_block *sb);
2130 void kill_litter_super(struct super_block *sb);
2131 void deactivate_super(struct super_block *sb);
2132 void deactivate_locked_super(struct super_block *sb);
2133 int set_anon_super(struct super_block *s, void *data);
2134 int get_anon_bdev(dev_t *);
2135 void free_anon_bdev(dev_t);
2136 struct super_block *sget_userns(struct file_system_type *type,
2137 int (*test)(struct super_block *,void *),
2138 int (*set)(struct super_block *,void *),
2139 int flags, struct user_namespace *user_ns,
2141 struct super_block *sget(struct file_system_type *type,
2142 int (*test)(struct super_block *,void *),
2143 int (*set)(struct super_block *,void *),
2144 int flags, void *data);
2145 extern struct dentry *mount_pseudo_xattr(struct file_system_type *, char *,
2146 const struct super_operations *ops,
2147 const struct xattr_handler **xattr,
2148 const struct dentry_operations *dops,
2151 static inline struct dentry *
2152 mount_pseudo(struct file_system_type *fs_type, char *name,
2153 const struct super_operations *ops,
2154 const struct dentry_operations *dops, unsigned long magic)
2156 return mount_pseudo_xattr(fs_type, name, ops, NULL, dops, magic);
2159 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2160 #define fops_get(fops) \
2161 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2162 #define fops_put(fops) \
2163 do { if (fops) module_put((fops)->owner); } while(0)
2165 * This one is to be used *ONLY* from ->open() instances.
2166 * fops must be non-NULL, pinned down *and* module dependencies
2167 * should be sufficient to pin the caller down as well.
2169 #define replace_fops(f, fops) \
2171 struct file *__file = (f); \
2172 fops_put(__file->f_op); \
2173 BUG_ON(!(__file->f_op = (fops))); \
2176 extern int register_filesystem(struct file_system_type *);
2177 extern int unregister_filesystem(struct file_system_type *);
2178 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2179 #define kern_mount(type) kern_mount_data(type, NULL)
2180 extern void kern_unmount(struct vfsmount *mnt);
2181 extern int may_umount_tree(struct vfsmount *);
2182 extern int may_umount(struct vfsmount *);
2183 extern long do_mount(const char *, const char __user *,
2184 const char *, unsigned long, void *);
2185 extern struct vfsmount *collect_mounts(const struct path *);
2186 extern void drop_collected_mounts(struct vfsmount *);
2187 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2189 extern int vfs_statfs(const struct path *, struct kstatfs *);
2190 extern int user_statfs(const char __user *, struct kstatfs *);
2191 extern int fd_statfs(int, struct kstatfs *);
2192 extern int freeze_super(struct super_block *super);
2193 extern int thaw_super(struct super_block *super);
2194 extern bool our_mnt(struct vfsmount *mnt);
2195 extern __printf(2, 3)
2196 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2197 extern int super_setup_bdi(struct super_block *sb);
2199 extern int current_umask(void);
2201 extern void ihold(struct inode * inode);
2202 extern void iput(struct inode *);
2203 extern int generic_update_time(struct inode *, struct timespec64 *, int);
2206 extern struct kobject *fs_kobj;
2208 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2210 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2211 extern int locks_mandatory_locked(struct file *);
2212 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2215 * Candidates for mandatory locking have the setgid bit set
2216 * but no group execute bit - an otherwise meaningless combination.
2219 static inline int __mandatory_lock(struct inode *ino)
2221 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2225 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2226 * otherwise these will be advisory locks
2229 static inline int mandatory_lock(struct inode *ino)
2231 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2234 static inline int locks_verify_locked(struct file *file)
2236 if (mandatory_lock(locks_inode(file)))
2237 return locks_mandatory_locked(file);
2241 static inline int locks_verify_truncate(struct inode *inode,
2245 if (!inode->i_flctx || !mandatory_lock(inode))
2248 if (size < inode->i_size) {
2249 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2252 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2257 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2259 static inline int locks_mandatory_locked(struct file *file)
2264 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2265 loff_t start, loff_t end, unsigned char type)
2270 static inline int __mandatory_lock(struct inode *inode)
2275 static inline int mandatory_lock(struct inode *inode)
2280 static inline int locks_verify_locked(struct file *file)
2285 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2291 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2294 #ifdef CONFIG_FILE_LOCKING
2295 static inline int break_lease(struct inode *inode, unsigned int mode)
2298 * Since this check is lockless, we must ensure that any refcounts
2299 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2300 * could end up racing with tasks trying to set a new lease on this
2304 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2305 return __break_lease(inode, mode, FL_LEASE);
2309 static inline int break_deleg(struct inode *inode, unsigned int mode)
2312 * Since this check is lockless, we must ensure that any refcounts
2313 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2314 * could end up racing with tasks trying to set a new lease on this
2318 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2319 return __break_lease(inode, mode, FL_DELEG);
2323 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2327 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2328 if (ret == -EWOULDBLOCK && delegated_inode) {
2329 *delegated_inode = inode;
2335 static inline int break_deleg_wait(struct inode **delegated_inode)
2339 ret = break_deleg(*delegated_inode, O_WRONLY);
2340 iput(*delegated_inode);
2341 *delegated_inode = NULL;
2345 static inline int break_layout(struct inode *inode, bool wait)
2348 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2349 return __break_lease(inode,
2350 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2355 #else /* !CONFIG_FILE_LOCKING */
2356 static inline int break_lease(struct inode *inode, unsigned int mode)
2361 static inline int break_deleg(struct inode *inode, unsigned int mode)
2366 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2371 static inline int break_deleg_wait(struct inode **delegated_inode)
2377 static inline int break_layout(struct inode *inode, bool wait)
2382 #endif /* CONFIG_FILE_LOCKING */
2387 const char *name; /* pointer to actual string */
2388 const __user char *uptr; /* original userland pointer */
2390 struct audit_names *aname;
2394 extern long vfs_truncate(const struct path *, loff_t);
2395 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2397 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2399 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2401 extern struct file *file_open_name(struct filename *, int, umode_t);
2402 extern struct file *filp_open(const char *, int, umode_t);
2403 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2404 const char *, int, umode_t);
2405 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2406 extern int filp_close(struct file *, fl_owner_t id);
2408 extern struct filename *getname_flags(const char __user *, int, int *);
2409 extern struct filename *getname(const char __user *);
2410 extern struct filename *getname_kernel(const char *);
2411 extern void putname(struct filename *name);
2417 extern int finish_open(struct file *file, struct dentry *dentry,
2418 int (*open)(struct inode *, struct file *),
2420 extern int finish_no_open(struct file *file, struct dentry *dentry);
2424 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2427 extern void __init vfs_caches_init_early(void);
2428 extern void __init vfs_caches_init(void);
2430 extern struct kmem_cache *names_cachep;
2432 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2433 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2436 extern int register_blkdev(unsigned int, const char *);
2437 extern void unregister_blkdev(unsigned int, const char *);
2438 extern void bdev_unhash_inode(dev_t dev);
2439 extern struct block_device *bdget(dev_t);
2440 extern struct block_device *bdgrab(struct block_device *bdev);
2441 extern void bd_set_size(struct block_device *, loff_t size);
2442 extern void bd_forget(struct inode *inode);
2443 extern void bdput(struct block_device *);
2444 extern void invalidate_bdev(struct block_device *);
2445 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2446 extern int sync_blockdev(struct block_device *bdev);
2447 extern void kill_bdev(struct block_device *);
2448 extern struct super_block *freeze_bdev(struct block_device *);
2449 extern void emergency_thaw_all(void);
2450 extern void emergency_thaw_bdev(struct super_block *sb);
2451 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2452 extern int fsync_bdev(struct block_device *);
2454 extern struct super_block *blockdev_superblock;
2456 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2458 return sb == blockdev_superblock;
2461 static inline void bd_forget(struct inode *inode) {}
2462 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2463 static inline void kill_bdev(struct block_device *bdev) {}
2464 static inline void invalidate_bdev(struct block_device *bdev) {}
2466 static inline struct super_block *freeze_bdev(struct block_device *sb)
2471 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2476 static inline int emergency_thaw_bdev(struct super_block *sb)
2481 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2485 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2490 extern int sync_filesystem(struct super_block *);
2491 extern const struct file_operations def_blk_fops;
2492 extern const struct file_operations def_chr_fops;
2494 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2495 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2496 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2497 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2498 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2500 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2502 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2503 extern int __blkdev_reread_part(struct block_device *bdev);
2504 extern int blkdev_reread_part(struct block_device *bdev);
2507 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2508 extern void bd_unlink_disk_holder(struct block_device *bdev,
2509 struct gendisk *disk);
2511 static inline int bd_link_disk_holder(struct block_device *bdev,
2512 struct gendisk *disk)
2516 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2517 struct gendisk *disk)
2524 #define CHRDEV_MAJOR_MAX 512
2525 /* Marks the bottom of the first segment of free char majors */
2526 #define CHRDEV_MAJOR_DYN_END 234
2527 /* Marks the top and bottom of the second segment of free char majors */
2528 #define CHRDEV_MAJOR_DYN_EXT_START 511
2529 #define CHRDEV_MAJOR_DYN_EXT_END 384
2531 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2532 extern int register_chrdev_region(dev_t, unsigned, const char *);
2533 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2534 unsigned int count, const char *name,
2535 const struct file_operations *fops);
2536 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2537 unsigned int count, const char *name);
2538 extern void unregister_chrdev_region(dev_t, unsigned);
2539 extern void chrdev_show(struct seq_file *,off_t);
2541 static inline int register_chrdev(unsigned int major, const char *name,
2542 const struct file_operations *fops)
2544 return __register_chrdev(major, 0, 256, name, fops);
2547 static inline void unregister_chrdev(unsigned int major, const char *name)
2549 __unregister_chrdev(major, 0, 256, name);
2552 /* fs/block_dev.c */
2553 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2554 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2557 #define BLKDEV_MAJOR_MAX 512
2558 extern const char *__bdevname(dev_t, char *buffer);
2559 extern const char *bdevname(struct block_device *bdev, char *buffer);
2560 extern struct block_device *lookup_bdev(const char *);
2561 extern void blkdev_show(struct seq_file *,off_t);
2564 #define BLKDEV_MAJOR_MAX 0
2567 extern void init_special_inode(struct inode *, umode_t, dev_t);
2569 /* Invalid inode operations -- fs/bad_inode.c */
2570 extern void make_bad_inode(struct inode *);
2571 extern bool is_bad_inode(struct inode *);
2574 extern void check_disk_size_change(struct gendisk *disk,
2575 struct block_device *bdev);
2576 extern int revalidate_disk(struct gendisk *);
2577 extern int check_disk_change(struct block_device *);
2578 extern int __invalidate_device(struct block_device *, bool);
2579 extern int invalidate_partition(struct gendisk *, int);
2581 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2582 pgoff_t start, pgoff_t end);
2584 static inline void invalidate_remote_inode(struct inode *inode)
2586 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2587 S_ISLNK(inode->i_mode))
2588 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2590 extern int invalidate_inode_pages2(struct address_space *mapping);
2591 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2592 pgoff_t start, pgoff_t end);
2593 extern int write_inode_now(struct inode *, int);
2594 extern int filemap_fdatawrite(struct address_space *);
2595 extern int filemap_flush(struct address_space *);
2596 extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2597 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2600 static inline int filemap_fdatawait(struct address_space *mapping)
2602 return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2605 extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2607 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2609 extern int filemap_write_and_wait(struct address_space *mapping);
2610 extern int filemap_write_and_wait_range(struct address_space *mapping,
2611 loff_t lstart, loff_t lend);
2612 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2613 loff_t start, loff_t end, int sync_mode);
2614 extern int filemap_fdatawrite_range(struct address_space *mapping,
2615 loff_t start, loff_t end);
2616 extern int filemap_check_errors(struct address_space *mapping);
2617 extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2619 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2621 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2622 extern int __must_check file_write_and_wait_range(struct file *file,
2623 loff_t start, loff_t end);
2625 static inline int file_write_and_wait(struct file *file)
2627 return file_write_and_wait_range(file, 0, LLONG_MAX);
2631 * filemap_set_wb_err - set a writeback error on an address_space
2632 * @mapping: mapping in which to set writeback error
2633 * @err: error to be set in mapping
2635 * When writeback fails in some way, we must record that error so that
2636 * userspace can be informed when fsync and the like are called. We endeavor
2637 * to report errors on any file that was open at the time of the error. Some
2638 * internal callers also need to know when writeback errors have occurred.
2640 * When a writeback error occurs, most filesystems will want to call
2641 * filemap_set_wb_err to record the error in the mapping so that it will be
2642 * automatically reported whenever fsync is called on the file.
2644 static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2646 /* Fastpath for common case of no error */
2648 __filemap_set_wb_err(mapping, err);
2652 * filemap_check_wb_error - has an error occurred since the mark was sampled?
2653 * @mapping: mapping to check for writeback errors
2654 * @since: previously-sampled errseq_t
2656 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2657 * the given value was sampled.
2659 * If it has then report the latest error set, otherwise return 0.
2661 static inline int filemap_check_wb_err(struct address_space *mapping,
2664 return errseq_check(&mapping->wb_err, since);
2668 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2669 * @mapping: mapping to be sampled
2671 * Writeback errors are always reported relative to a particular sample point
2672 * in the past. This function provides those sample points.
2674 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2676 return errseq_sample(&mapping->wb_err);
2679 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2681 extern int vfs_fsync(struct file *file, int datasync);
2684 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2685 * to already be updated for the write, and will return either the amount
2686 * of bytes passed in, or an error if syncing the file failed.
2688 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2690 if (iocb->ki_flags & IOCB_DSYNC) {
2691 int ret = vfs_fsync_range(iocb->ki_filp,
2692 iocb->ki_pos - count, iocb->ki_pos - 1,
2693 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2701 extern void emergency_sync(void);
2702 extern void emergency_remount(void);
2704 extern sector_t bmap(struct inode *, sector_t);
2706 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2707 extern int inode_permission(struct inode *, int);
2708 extern int generic_permission(struct inode *, int);
2709 extern int __check_sticky(struct inode *dir, struct inode *inode);
2711 static inline bool execute_ok(struct inode *inode)
2713 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2716 static inline void file_start_write(struct file *file)
2718 if (!S_ISREG(file_inode(file)->i_mode))
2720 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2723 static inline bool file_start_write_trylock(struct file *file)
2725 if (!S_ISREG(file_inode(file)->i_mode))
2727 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2730 static inline void file_end_write(struct file *file)
2732 if (!S_ISREG(file_inode(file)->i_mode))
2734 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2737 static inline int do_clone_file_range(struct file *file_in, loff_t pos_in,
2738 struct file *file_out, loff_t pos_out,
2743 file_start_write(file_out);
2744 ret = vfs_clone_file_range(file_in, pos_in, file_out, pos_out, len);
2745 file_end_write(file_out);
2751 * get_write_access() gets write permission for a file.
2752 * put_write_access() releases this write permission.
2753 * This is used for regular files.
2754 * We cannot support write (and maybe mmap read-write shared) accesses and
2755 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2756 * can have the following values:
2757 * 0: no writers, no VM_DENYWRITE mappings
2758 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2759 * > 0: (i_writecount) users are writing to the file.
2761 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2762 * except for the cases where we don't hold i_writecount yet. Then we need to
2763 * use {get,deny}_write_access() - these functions check the sign and refuse
2764 * to do the change if sign is wrong.
2766 static inline int get_write_access(struct inode *inode)
2768 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2770 static inline int deny_write_access(struct file *file)
2772 struct inode *inode = file_inode(file);
2773 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2775 static inline void put_write_access(struct inode * inode)
2777 atomic_dec(&inode->i_writecount);
2779 static inline void allow_write_access(struct file *file)
2782 atomic_inc(&file_inode(file)->i_writecount);
2784 static inline bool inode_is_open_for_write(const struct inode *inode)
2786 return atomic_read(&inode->i_writecount) > 0;
2790 static inline void i_readcount_dec(struct inode *inode)
2792 BUG_ON(!atomic_read(&inode->i_readcount));
2793 atomic_dec(&inode->i_readcount);
2795 static inline void i_readcount_inc(struct inode *inode)
2797 atomic_inc(&inode->i_readcount);
2800 static inline void i_readcount_dec(struct inode *inode)
2804 static inline void i_readcount_inc(struct inode *inode)
2809 extern int do_pipe_flags(int *, int);
2811 #define __kernel_read_file_id(id) \
2812 id(UNKNOWN, unknown) \
2813 id(FIRMWARE, firmware) \
2814 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2815 id(MODULE, kernel-module) \
2816 id(KEXEC_IMAGE, kexec-image) \
2817 id(KEXEC_INITRAMFS, kexec-initramfs) \
2818 id(POLICY, security-policy) \
2819 id(X509_CERTIFICATE, x509-certificate) \
2822 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2823 #define __fid_stringify(dummy, str) #str,
2825 enum kernel_read_file_id {
2826 __kernel_read_file_id(__fid_enumify)
2829 static const char * const kernel_read_file_str[] = {
2830 __kernel_read_file_id(__fid_stringify)
2833 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
2835 if ((unsigned)id >= READING_MAX_ID)
2836 return kernel_read_file_str[READING_UNKNOWN];
2838 return kernel_read_file_str[id];
2841 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2842 enum kernel_read_file_id);
2843 extern int kernel_read_file_from_path(const char *, void **, loff_t *, loff_t,
2844 enum kernel_read_file_id);
2845 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2846 enum kernel_read_file_id);
2847 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
2848 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
2849 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
2850 extern struct file * open_exec(const char *);
2852 /* fs/dcache.c -- generic fs support functions */
2853 extern bool is_subdir(struct dentry *, struct dentry *);
2854 extern bool path_is_under(const struct path *, const struct path *);
2856 extern char *file_path(struct file *, char *, int);
2858 #include <linux/err.h>
2860 /* needed for stackable file system support */
2861 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2863 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2865 extern int inode_init_always(struct super_block *, struct inode *);
2866 extern void inode_init_once(struct inode *);
2867 extern void address_space_init_once(struct address_space *mapping);
2868 extern struct inode * igrab(struct inode *);
2869 extern ino_t iunique(struct super_block *, ino_t);
2870 extern int inode_needs_sync(struct inode *inode);
2871 extern int generic_delete_inode(struct inode *inode);
2872 static inline int generic_drop_inode(struct inode *inode)
2874 return !inode->i_nlink || inode_unhashed(inode);
2877 extern struct inode *ilookup5_nowait(struct super_block *sb,
2878 unsigned long hashval, int (*test)(struct inode *, void *),
2880 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2881 int (*test)(struct inode *, void *), void *data);
2882 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2884 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2885 extern struct inode * iget_locked(struct super_block *, unsigned long);
2886 extern struct inode *find_inode_nowait(struct super_block *,
2888 int (*match)(struct inode *,
2889 unsigned long, void *),
2891 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2892 extern int insert_inode_locked(struct inode *);
2893 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2894 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2896 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2898 extern void unlock_new_inode(struct inode *);
2899 extern unsigned int get_next_ino(void);
2900 extern void evict_inodes(struct super_block *sb);
2902 extern void __iget(struct inode * inode);
2903 extern void iget_failed(struct inode *);
2904 extern void clear_inode(struct inode *);
2905 extern void __destroy_inode(struct inode *);
2906 extern struct inode *new_inode_pseudo(struct super_block *sb);
2907 extern struct inode *new_inode(struct super_block *sb);
2908 extern void free_inode_nonrcu(struct inode *inode);
2909 extern int should_remove_suid(struct dentry *);
2910 extern int file_remove_privs(struct file *);
2912 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2913 static inline void insert_inode_hash(struct inode *inode)
2915 __insert_inode_hash(inode, inode->i_ino);
2918 extern void __remove_inode_hash(struct inode *);
2919 static inline void remove_inode_hash(struct inode *inode)
2921 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2922 __remove_inode_hash(inode);
2925 extern void inode_sb_list_add(struct inode *inode);
2928 extern int bdev_read_only(struct block_device *);
2930 extern int set_blocksize(struct block_device *, int);
2931 extern int sb_set_blocksize(struct super_block *, int);
2932 extern int sb_min_blocksize(struct super_block *, int);
2934 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2935 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2936 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2937 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2938 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2939 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2940 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
2941 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2943 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
2945 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
2948 /* fs/block_dev.c */
2949 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2950 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2951 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2953 extern void block_sync_page(struct page *page);
2956 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2957 struct pipe_inode_info *, size_t, unsigned int);
2958 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2959 struct file *, loff_t *, size_t, unsigned int);
2960 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2961 struct file *out, loff_t *, size_t len, unsigned int flags);
2962 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2963 loff_t *opos, size_t len, unsigned int flags);
2967 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2968 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2969 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2970 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2971 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2972 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2973 int whence, loff_t maxsize, loff_t eof);
2974 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2975 int whence, loff_t size);
2976 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
2977 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
2978 extern int generic_file_open(struct inode * inode, struct file * filp);
2979 extern int nonseekable_open(struct inode * inode, struct file * filp);
2982 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
2983 loff_t file_offset);
2986 /* need locking between buffered and direct access */
2989 /* filesystem does not support filling holes */
2990 DIO_SKIP_HOLES = 0x02,
2993 void dio_end_io(struct bio *bio);
2994 void dio_warn_stale_pagecache(struct file *filp);
2996 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2997 struct block_device *bdev, struct iov_iter *iter,
2998 get_block_t get_block,
2999 dio_iodone_t end_io, dio_submit_t submit_io,
3002 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3003 struct inode *inode,
3004 struct iov_iter *iter,
3005 get_block_t get_block)
3007 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3008 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3012 void inode_dio_wait(struct inode *inode);
3015 * inode_dio_begin - signal start of a direct I/O requests
3016 * @inode: inode the direct I/O happens on
3018 * This is called once we've finished processing a direct I/O request,
3019 * and is used to wake up callers waiting for direct I/O to be quiesced.
3021 static inline void inode_dio_begin(struct inode *inode)
3023 atomic_inc(&inode->i_dio_count);
3027 * inode_dio_end - signal finish of a direct I/O requests
3028 * @inode: inode the direct I/O happens on
3030 * This is called once we've finished processing a direct I/O request,
3031 * and is used to wake up callers waiting for direct I/O to be quiesced.
3033 static inline void inode_dio_end(struct inode *inode)
3035 if (atomic_dec_and_test(&inode->i_dio_count))
3036 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3039 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3042 extern const struct file_operations generic_ro_fops;
3044 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3046 extern int readlink_copy(char __user *, int, const char *);
3047 extern int page_readlink(struct dentry *, char __user *, int);
3048 extern const char *page_get_link(struct dentry *, struct inode *,
3049 struct delayed_call *);
3050 extern void page_put_link(void *);
3051 extern int __page_symlink(struct inode *inode, const char *symname, int len,
3053 extern int page_symlink(struct inode *inode, const char *symname, int len);
3054 extern const struct inode_operations page_symlink_inode_operations;
3055 extern void kfree_link(void *);
3056 extern void generic_fillattr(struct inode *, struct kstat *);
3057 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3058 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3059 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3060 void inode_add_bytes(struct inode *inode, loff_t bytes);
3061 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3062 void inode_sub_bytes(struct inode *inode, loff_t bytes);
3063 static inline loff_t __inode_get_bytes(struct inode *inode)
3065 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3067 loff_t inode_get_bytes(struct inode *inode);
3068 void inode_set_bytes(struct inode *inode, loff_t bytes);
3069 const char *simple_get_link(struct dentry *, struct inode *,
3070 struct delayed_call *);
3071 extern const struct inode_operations simple_symlink_inode_operations;
3073 extern int iterate_dir(struct file *, struct dir_context *);
3075 extern int vfs_statx(int, const char __user *, int, struct kstat *, u32);
3076 extern int vfs_statx_fd(unsigned int, struct kstat *, u32, unsigned int);
3078 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3080 return vfs_statx(AT_FDCWD, filename, AT_NO_AUTOMOUNT,
3081 stat, STATX_BASIC_STATS);
3083 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3085 return vfs_statx(AT_FDCWD, name, AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT,
3086 stat, STATX_BASIC_STATS);
3088 static inline int vfs_fstatat(int dfd, const char __user *filename,
3089 struct kstat *stat, int flags)
3091 return vfs_statx(dfd, filename, flags | AT_NO_AUTOMOUNT,
3092 stat, STATX_BASIC_STATS);
3094 static inline int vfs_fstat(int fd, struct kstat *stat)
3096 return vfs_statx_fd(fd, stat, STATX_BASIC_STATS, 0);
3100 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3101 extern int vfs_readlink(struct dentry *, char __user *, int);
3103 extern int __generic_block_fiemap(struct inode *inode,
3104 struct fiemap_extent_info *fieinfo,
3105 loff_t start, loff_t len,
3106 get_block_t *get_block);
3107 extern int generic_block_fiemap(struct inode *inode,
3108 struct fiemap_extent_info *fieinfo, u64 start,
3109 u64 len, get_block_t *get_block);
3111 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3112 extern void put_filesystem(struct file_system_type *fs);
3113 extern struct file_system_type *get_fs_type(const char *name);
3114 extern struct super_block *get_super(struct block_device *);
3115 extern struct super_block *get_super_thawed(struct block_device *);
3116 extern struct super_block *get_super_exclusive_thawed(struct block_device *bdev);
3117 extern struct super_block *get_active_super(struct block_device *bdev);
3118 extern void drop_super(struct super_block *sb);
3119 extern void drop_super_exclusive(struct super_block *sb);
3120 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3121 extern void iterate_supers_type(struct file_system_type *,
3122 void (*)(struct super_block *, void *), void *);
3124 extern int dcache_dir_open(struct inode *, struct file *);
3125 extern int dcache_dir_close(struct inode *, struct file *);
3126 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3127 extern int dcache_readdir(struct file *, struct dir_context *);
3128 extern int simple_setattr(struct dentry *, struct iattr *);
3129 extern int simple_getattr(const struct path *, struct kstat *, u32, unsigned int);
3130 extern int simple_statfs(struct dentry *, struct kstatfs *);
3131 extern int simple_open(struct inode *inode, struct file *file);
3132 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3133 extern int simple_unlink(struct inode *, struct dentry *);
3134 extern int simple_rmdir(struct inode *, struct dentry *);
3135 extern int simple_rename(struct inode *, struct dentry *,
3136 struct inode *, struct dentry *, unsigned int);
3137 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3138 extern int noop_set_page_dirty(struct page *page);
3139 extern void noop_invalidatepage(struct page *page, unsigned int offset,
3140 unsigned int length);
3141 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3142 extern int simple_empty(struct dentry *);
3143 extern int simple_readpage(struct file *file, struct page *page);
3144 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3145 loff_t pos, unsigned len, unsigned flags,
3146 struct page **pagep, void **fsdata);
3147 extern int simple_write_end(struct file *file, struct address_space *mapping,
3148 loff_t pos, unsigned len, unsigned copied,
3149 struct page *page, void *fsdata);
3150 extern int always_delete_dentry(const struct dentry *);
3151 extern struct inode *alloc_anon_inode(struct super_block *);
3152 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3153 extern const struct dentry_operations simple_dentry_operations;
3155 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3156 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3157 extern const struct file_operations simple_dir_operations;
3158 extern const struct inode_operations simple_dir_inode_operations;
3159 extern void make_empty_dir_inode(struct inode *inode);
3160 extern bool is_empty_dir_inode(struct inode *inode);
3161 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3162 struct dentry *d_alloc_name(struct dentry *, const char *);
3163 extern int simple_fill_super(struct super_block *, unsigned long,
3164 const struct tree_descr *);
3165 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3166 extern void simple_release_fs(struct vfsmount **mount, int *count);
3168 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3169 loff_t *ppos, const void *from, size_t available);
3170 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3171 const void __user *from, size_t count);
3173 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3174 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3176 extern int generic_check_addressable(unsigned, u64);
3178 #ifdef CONFIG_MIGRATION
3179 extern int buffer_migrate_page(struct address_space *,
3180 struct page *, struct page *,
3183 #define buffer_migrate_page NULL
3186 extern int setattr_prepare(struct dentry *, struct iattr *);
3187 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3188 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
3190 extern int file_update_time(struct file *file);
3192 static inline bool io_is_direct(struct file *filp)
3194 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
3197 static inline bool vma_is_dax(struct vm_area_struct *vma)
3199 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3202 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3204 struct inode *inode;
3208 if (!vma_is_dax(vma))
3210 inode = file_inode(vma->vm_file);
3211 if (S_ISCHR(inode->i_mode))
3212 return false; /* device-dax */
3216 static inline int iocb_flags(struct file *file)
3219 if (file->f_flags & O_APPEND)
3221 if (io_is_direct(file))
3223 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3225 if (file->f_flags & __O_SYNC)
3230 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3232 if (unlikely(flags & ~RWF_SUPPORTED))
3235 if (flags & RWF_NOWAIT) {
3236 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3238 ki->ki_flags |= IOCB_NOWAIT;
3240 if (flags & RWF_HIPRI)
3241 ki->ki_flags |= IOCB_HIPRI;
3242 if (flags & RWF_DSYNC)
3243 ki->ki_flags |= IOCB_DSYNC;
3244 if (flags & RWF_SYNC)
3245 ki->ki_flags |= (IOCB_DSYNC | IOCB_SYNC);
3246 if (flags & RWF_APPEND)
3247 ki->ki_flags |= IOCB_APPEND;
3251 static inline ino_t parent_ino(struct dentry *dentry)
3256 * Don't strictly need d_lock here? If the parent ino could change
3257 * then surely we'd have a deeper race in the caller?
3259 spin_lock(&dentry->d_lock);
3260 res = dentry->d_parent->d_inode->i_ino;
3261 spin_unlock(&dentry->d_lock);
3265 /* Transaction based IO helpers */
3268 * An argresp is stored in an allocated page and holds the
3269 * size of the argument or response, along with its content
3271 struct simple_transaction_argresp {
3276 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3278 char *simple_transaction_get(struct file *file, const char __user *buf,
3280 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3281 size_t size, loff_t *pos);
3282 int simple_transaction_release(struct inode *inode, struct file *file);
3284 void simple_transaction_set(struct file *file, size_t n);
3287 * simple attribute files
3289 * These attributes behave similar to those in sysfs:
3291 * Writing to an attribute immediately sets a value, an open file can be
3292 * written to multiple times.
3294 * Reading from an attribute creates a buffer from the value that might get
3295 * read with multiple read calls. When the attribute has been read
3296 * completely, no further read calls are possible until the file is opened
3299 * All attributes contain a text representation of a numeric value
3300 * that are accessed with the get() and set() functions.
3302 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3303 static int __fops ## _open(struct inode *inode, struct file *file) \
3305 __simple_attr_check_format(__fmt, 0ull); \
3306 return simple_attr_open(inode, file, __get, __set, __fmt); \
3308 static const struct file_operations __fops = { \
3309 .owner = THIS_MODULE, \
3310 .open = __fops ## _open, \
3311 .release = simple_attr_release, \
3312 .read = simple_attr_read, \
3313 .write = simple_attr_write, \
3314 .llseek = generic_file_llseek, \
3317 static inline __printf(1, 2)
3318 void __simple_attr_check_format(const char *fmt, ...)
3320 /* don't do anything, just let the compiler check the arguments; */
3323 int simple_attr_open(struct inode *inode, struct file *file,
3324 int (*get)(void *, u64 *), int (*set)(void *, u64),
3326 int simple_attr_release(struct inode *inode, struct file *file);
3327 ssize_t simple_attr_read(struct file *file, char __user *buf,
3328 size_t len, loff_t *ppos);
3329 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3330 size_t len, loff_t *ppos);
3333 int proc_nr_files(struct ctl_table *table, int write,
3334 void __user *buffer, size_t *lenp, loff_t *ppos);
3335 int proc_nr_dentry(struct ctl_table *table, int write,
3336 void __user *buffer, size_t *lenp, loff_t *ppos);
3337 int proc_nr_inodes(struct ctl_table *table, int write,
3338 void __user *buffer, size_t *lenp, loff_t *ppos);
3339 int __init get_filesystem_list(char *buf);
3341 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3342 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3344 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3345 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3346 (flag & __FMODE_NONOTIFY)))
3348 static inline bool is_sxid(umode_t mode)
3350 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3353 static inline int check_sticky(struct inode *dir, struct inode *inode)
3355 if (!(dir->i_mode & S_ISVTX))
3358 return __check_sticky(dir, inode);
3361 static inline void inode_has_no_xattr(struct inode *inode)
3363 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3364 inode->i_flags |= S_NOSEC;
3367 static inline bool is_root_inode(struct inode *inode)
3369 return inode == inode->i_sb->s_root->d_inode;
3372 static inline bool dir_emit(struct dir_context *ctx,
3373 const char *name, int namelen,
3374 u64 ino, unsigned type)
3376 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3378 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3380 return ctx->actor(ctx, ".", 1, ctx->pos,
3381 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3383 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3385 return ctx->actor(ctx, "..", 2, ctx->pos,
3386 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3388 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3390 if (ctx->pos == 0) {
3391 if (!dir_emit_dot(file, ctx))
3395 if (ctx->pos == 1) {
3396 if (!dir_emit_dotdot(file, ctx))
3402 static inline bool dir_relax(struct inode *inode)
3404 inode_unlock(inode);
3406 return !IS_DEADDIR(inode);
3409 static inline bool dir_relax_shared(struct inode *inode)
3411 inode_unlock_shared(inode);
3412 inode_lock_shared(inode);
3413 return !IS_DEADDIR(inode);
3416 extern bool path_noexec(const struct path *path);
3417 extern void inode_nohighmem(struct inode *inode);
3419 #endif /* _LINUX_FS_H */