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>
39 #include <linux/ioprio.h>
41 #include <asm/byteorder.h>
42 #include <uapi/linux/fs.h>
44 struct backing_dev_info;
47 struct export_operations;
52 struct pipe_inode_info;
53 struct poll_table_struct;
55 struct vm_area_struct;
58 struct swap_info_struct;
60 struct workqueue_struct;
63 struct fscrypt_operations;
65 extern void __init inode_init(void);
66 extern void __init inode_init_early(void);
67 extern void __init files_init(void);
68 extern void __init files_maxfiles_init(void);
70 extern struct files_stat_struct files_stat;
71 extern unsigned long get_max_files(void);
72 extern unsigned int sysctl_nr_open;
73 extern struct inodes_stat_t inodes_stat;
74 extern int leases_enable, lease_break_time;
75 extern int sysctl_protected_symlinks;
76 extern int sysctl_protected_hardlinks;
78 typedef __kernel_rwf_t rwf_t;
81 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
82 struct buffer_head *bh_result, int create);
83 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
84 ssize_t bytes, void *private);
86 #define MAY_EXEC 0x00000001
87 #define MAY_WRITE 0x00000002
88 #define MAY_READ 0x00000004
89 #define MAY_APPEND 0x00000008
90 #define MAY_ACCESS 0x00000010
91 #define MAY_OPEN 0x00000020
92 #define MAY_CHDIR 0x00000040
93 /* called from RCU mode, don't block */
94 #define MAY_NOT_BLOCK 0x00000080
97 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
98 * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
101 /* file is open for reading */
102 #define FMODE_READ ((__force fmode_t)0x1)
103 /* file is open for writing */
104 #define FMODE_WRITE ((__force fmode_t)0x2)
105 /* file is seekable */
106 #define FMODE_LSEEK ((__force fmode_t)0x4)
107 /* file can be accessed using pread */
108 #define FMODE_PREAD ((__force fmode_t)0x8)
109 /* file can be accessed using pwrite */
110 #define FMODE_PWRITE ((__force fmode_t)0x10)
111 /* File is opened for execution with sys_execve / sys_uselib */
112 #define FMODE_EXEC ((__force fmode_t)0x20)
113 /* File is opened with O_NDELAY (only set for block devices) */
114 #define FMODE_NDELAY ((__force fmode_t)0x40)
115 /* File is opened with O_EXCL (only set for block devices) */
116 #define FMODE_EXCL ((__force fmode_t)0x80)
117 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
118 (specialy hack for floppy.c) */
119 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
120 /* 32bit hashes as llseek() offset (for directories) */
121 #define FMODE_32BITHASH ((__force fmode_t)0x200)
122 /* 64bit hashes as llseek() offset (for directories) */
123 #define FMODE_64BITHASH ((__force fmode_t)0x400)
126 * Don't update ctime and mtime.
128 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
129 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
131 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
133 /* Expect random access pattern */
134 #define FMODE_RANDOM ((__force fmode_t)0x1000)
136 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
137 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
139 /* File is opened with O_PATH; almost nothing can be done with it */
140 #define FMODE_PATH ((__force fmode_t)0x4000)
142 /* File needs atomic accesses to f_pos */
143 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
144 /* Write access to underlying fs */
145 #define FMODE_WRITER ((__force fmode_t)0x10000)
146 /* Has read method(s) */
147 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
148 /* Has write method(s) */
149 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
151 #define FMODE_OPENED ((__force fmode_t)0x80000)
152 #define FMODE_CREATED ((__force fmode_t)0x100000)
154 /* File was opened by fanotify and shouldn't generate fanotify events */
155 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
157 /* File is capable of returning -EAGAIN if I/O will block */
158 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
161 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
162 * that indicates that they should check the contents of the iovec are
163 * valid, but not check the memory that the iovec elements
166 #define CHECK_IOVEC_ONLY -1
169 * Attribute flags. These should be or-ed together to figure out what
172 #define ATTR_MODE (1 << 0)
173 #define ATTR_UID (1 << 1)
174 #define ATTR_GID (1 << 2)
175 #define ATTR_SIZE (1 << 3)
176 #define ATTR_ATIME (1 << 4)
177 #define ATTR_MTIME (1 << 5)
178 #define ATTR_CTIME (1 << 6)
179 #define ATTR_ATIME_SET (1 << 7)
180 #define ATTR_MTIME_SET (1 << 8)
181 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
182 #define ATTR_ATTR_FLAG (1 << 10)
183 #define ATTR_KILL_SUID (1 << 11)
184 #define ATTR_KILL_SGID (1 << 12)
185 #define ATTR_FILE (1 << 13)
186 #define ATTR_KILL_PRIV (1 << 14)
187 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
188 #define ATTR_TIMES_SET (1 << 16)
189 #define ATTR_TOUCH (1 << 17)
192 * Whiteout is represented by a char device. The following constants define the
193 * mode and device number to use.
195 #define WHITEOUT_MODE 0
196 #define WHITEOUT_DEV 0
199 * This is the Inode Attributes structure, used for notify_change(). It
200 * uses the above definitions as flags, to know which values have changed.
201 * Also, in this manner, a Filesystem can look at only the values it cares
202 * about. Basically, these are the attributes that the VFS layer can
203 * request to change from the FS layer.
205 * Derek Atkins <warlord@MIT.EDU> 94-10-20
208 unsigned int ia_valid;
213 struct timespec64 ia_atime;
214 struct timespec64 ia_mtime;
215 struct timespec64 ia_ctime;
218 * Not an attribute, but an auxiliary info for filesystems wanting to
219 * implement an ftruncate() like method. NOTE: filesystem should
220 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
222 struct file *ia_file;
226 * Includes for diskquotas.
228 #include <linux/quota.h>
231 * Maximum number of layers of fs stack. Needs to be limited to
232 * prevent kernel stack overflow
234 #define FILESYSTEM_MAX_STACK_DEPTH 2
237 * enum positive_aop_returns - aop return codes with specific semantics
239 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
240 * completed, that the page is still locked, and
241 * should be considered active. The VM uses this hint
242 * to return the page to the active list -- it won't
243 * be a candidate for writeback again in the near
244 * future. Other callers must be careful to unlock
245 * the page if they get this return. Returned by
248 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
249 * unlocked it and the page might have been truncated.
250 * The caller should back up to acquiring a new page and
251 * trying again. The aop will be taking reasonable
252 * precautions not to livelock. If the caller held a page
253 * reference, it should drop it before retrying. Returned
256 * address_space_operation functions return these large constants to indicate
257 * special semantics to the caller. These are much larger than the bytes in a
258 * page to allow for functions that return the number of bytes operated on in a
262 enum positive_aop_returns {
263 AOP_WRITEPAGE_ACTIVATE = 0x80000,
264 AOP_TRUNCATED_PAGE = 0x80001,
267 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
268 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
269 * helper code (eg buffer layer)
270 * to clear GFP_FS from alloc */
273 * oh the beauties of C type declarations.
276 struct address_space;
277 struct writeback_control;
280 * Write life time hint values.
281 * Stored in struct inode as u8.
284 WRITE_LIFE_NOT_SET = 0,
285 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
286 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
287 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
288 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
289 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
292 #define IOCB_EVENTFD (1 << 0)
293 #define IOCB_APPEND (1 << 1)
294 #define IOCB_DIRECT (1 << 2)
295 #define IOCB_HIPRI (1 << 3)
296 #define IOCB_DSYNC (1 << 4)
297 #define IOCB_SYNC (1 << 5)
298 #define IOCB_WRITE (1 << 6)
299 #define IOCB_NOWAIT (1 << 7)
302 struct file *ki_filp;
304 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
308 u16 ki_ioprio; /* See linux/ioprio.h */
309 } __randomize_layout;
311 static inline bool is_sync_kiocb(struct kiocb *kiocb)
313 return kiocb->ki_complete == NULL;
317 * "descriptor" for what we're up to with a read.
318 * This allows us to use the same read code yet
319 * have multiple different users of the data that
320 * we read from a file.
322 * The simplest case just copies the data to user
335 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
336 unsigned long, unsigned long);
338 struct address_space_operations {
339 int (*writepage)(struct page *page, struct writeback_control *wbc);
340 int (*readpage)(struct file *, struct page *);
342 /* Write back some dirty pages from this mapping. */
343 int (*writepages)(struct address_space *, struct writeback_control *);
345 /* Set a page dirty. Return true if this dirtied it */
346 int (*set_page_dirty)(struct page *page);
348 int (*readpages)(struct file *filp, struct address_space *mapping,
349 struct list_head *pages, unsigned nr_pages);
351 int (*write_begin)(struct file *, struct address_space *mapping,
352 loff_t pos, unsigned len, unsigned flags,
353 struct page **pagep, void **fsdata);
354 int (*write_end)(struct file *, struct address_space *mapping,
355 loff_t pos, unsigned len, unsigned copied,
356 struct page *page, void *fsdata);
358 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
359 sector_t (*bmap)(struct address_space *, sector_t);
360 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
361 int (*releasepage) (struct page *, gfp_t);
362 void (*freepage)(struct page *);
363 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
365 * migrate the contents of a page to the specified target. If
366 * migrate_mode is MIGRATE_ASYNC, it must not block.
368 int (*migratepage) (struct address_space *,
369 struct page *, struct page *, enum migrate_mode);
370 bool (*isolate_page)(struct page *, isolate_mode_t);
371 void (*putback_page)(struct page *);
372 int (*launder_page) (struct page *);
373 int (*is_partially_uptodate) (struct page *, unsigned long,
375 void (*is_dirty_writeback) (struct page *, bool *, bool *);
376 int (*error_remove_page)(struct address_space *, struct page *);
378 /* swapfile support */
379 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
381 void (*swap_deactivate)(struct file *file);
384 extern const struct address_space_operations empty_aops;
387 * pagecache_write_begin/pagecache_write_end must be used by general code
388 * to write into the pagecache.
390 int pagecache_write_begin(struct file *, struct address_space *mapping,
391 loff_t pos, unsigned len, unsigned flags,
392 struct page **pagep, void **fsdata);
394 int pagecache_write_end(struct file *, struct address_space *mapping,
395 loff_t pos, unsigned len, unsigned copied,
396 struct page *page, void *fsdata);
398 struct address_space {
399 struct inode *host; /* owner: inode, block_device */
400 struct radix_tree_root i_pages; /* cached pages */
401 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
402 struct rb_root_cached i_mmap; /* tree of private and shared mappings */
403 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
404 /* Protected by the i_pages lock */
405 unsigned long nrpages; /* number of total pages */
406 /* number of shadow or DAX exceptional entries */
407 unsigned long nrexceptional;
408 pgoff_t writeback_index;/* writeback starts here */
409 const struct address_space_operations *a_ops; /* methods */
410 unsigned long flags; /* error bits */
411 spinlock_t private_lock; /* for use by the address_space */
412 gfp_t gfp_mask; /* implicit gfp mask for allocations */
413 struct list_head private_list; /* for use by the address_space */
414 void *private_data; /* ditto */
416 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
418 * On most architectures that alignment is already the case; but
419 * must be enforced here for CRIS, to let the least significant bit
420 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
422 struct request_queue;
424 struct block_device {
425 dev_t bd_dev; /* not a kdev_t - it's a search key */
427 struct inode * bd_inode; /* will die */
428 struct super_block * bd_super;
429 struct mutex bd_mutex; /* open/close mutex */
433 bool bd_write_holder;
435 struct list_head bd_holder_disks;
437 struct block_device * bd_contains;
438 unsigned bd_block_size;
440 struct hd_struct * bd_part;
441 /* number of times partitions within this device have been opened. */
442 unsigned bd_part_count;
444 struct gendisk * bd_disk;
445 struct request_queue * bd_queue;
446 struct backing_dev_info *bd_bdi;
447 struct list_head bd_list;
449 * Private data. You must have bd_claim'ed the block_device
450 * to use this. NOTE: bd_claim allows an owner to claim
451 * the same device multiple times, the owner must take special
452 * care to not mess up bd_private for that case.
454 unsigned long bd_private;
456 /* The counter of freeze processes */
457 int bd_fsfreeze_count;
458 /* Mutex for freeze */
459 struct mutex bd_fsfreeze_mutex;
460 } __randomize_layout;
463 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
466 #define PAGECACHE_TAG_DIRTY 0
467 #define PAGECACHE_TAG_WRITEBACK 1
468 #define PAGECACHE_TAG_TOWRITE 2
470 int mapping_tagged(struct address_space *mapping, int tag);
472 static inline void i_mmap_lock_write(struct address_space *mapping)
474 down_write(&mapping->i_mmap_rwsem);
477 static inline void i_mmap_unlock_write(struct address_space *mapping)
479 up_write(&mapping->i_mmap_rwsem);
482 static inline void i_mmap_lock_read(struct address_space *mapping)
484 down_read(&mapping->i_mmap_rwsem);
487 static inline void i_mmap_unlock_read(struct address_space *mapping)
489 up_read(&mapping->i_mmap_rwsem);
493 * Might pages of this file be mapped into userspace?
495 static inline int mapping_mapped(struct address_space *mapping)
497 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
501 * Might pages of this file have been modified in userspace?
502 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
503 * marks vma as VM_SHARED if it is shared, and the file was opened for
504 * writing i.e. vma may be mprotected writable even if now readonly.
506 * If i_mmap_writable is negative, no new writable mappings are allowed. You
507 * can only deny writable mappings, if none exists right now.
509 static inline int mapping_writably_mapped(struct address_space *mapping)
511 return atomic_read(&mapping->i_mmap_writable) > 0;
514 static inline int mapping_map_writable(struct address_space *mapping)
516 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
520 static inline void mapping_unmap_writable(struct address_space *mapping)
522 atomic_dec(&mapping->i_mmap_writable);
525 static inline int mapping_deny_writable(struct address_space *mapping)
527 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
531 static inline void mapping_allow_writable(struct address_space *mapping)
533 atomic_inc(&mapping->i_mmap_writable);
537 * Use sequence counter to get consistent i_size on 32-bit processors.
539 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
540 #include <linux/seqlock.h>
541 #define __NEED_I_SIZE_ORDERED
542 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
544 #define i_size_ordered_init(inode) do { } while (0)
548 #define ACL_NOT_CACHED ((void *)(-1))
549 #define ACL_DONT_CACHE ((void *)(-3))
551 static inline struct posix_acl *
552 uncached_acl_sentinel(struct task_struct *task)
554 return (void *)task + 1;
558 is_uncached_acl(struct posix_acl *acl)
560 return (long)acl & 1;
563 #define IOP_FASTPERM 0x0001
564 #define IOP_LOOKUP 0x0002
565 #define IOP_NOFOLLOW 0x0004
566 #define IOP_XATTR 0x0008
567 #define IOP_DEFAULT_READLINK 0x0010
569 struct fsnotify_mark_connector;
572 * Keep mostly read-only and often accessed (especially for
573 * the RCU path lookup and 'stat' data) fields at the beginning
574 * of the 'struct inode'
578 unsigned short i_opflags;
581 unsigned int i_flags;
583 #ifdef CONFIG_FS_POSIX_ACL
584 struct posix_acl *i_acl;
585 struct posix_acl *i_default_acl;
588 const struct inode_operations *i_op;
589 struct super_block *i_sb;
590 struct address_space *i_mapping;
592 #ifdef CONFIG_SECURITY
596 /* Stat data, not accessed from path walking */
599 * Filesystems may only read i_nlink directly. They shall use the
600 * following functions for modification:
602 * (set|clear|inc|drop)_nlink
603 * inode_(inc|dec)_link_count
606 const unsigned int i_nlink;
607 unsigned int __i_nlink;
611 struct timespec64 i_atime;
612 struct timespec64 i_mtime;
613 struct timespec64 i_ctime;
614 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
615 unsigned short i_bytes;
620 #ifdef __NEED_I_SIZE_ORDERED
621 seqcount_t i_size_seqcount;
625 unsigned long i_state;
626 struct rw_semaphore i_rwsem;
628 unsigned long dirtied_when; /* jiffies of first dirtying */
629 unsigned long dirtied_time_when;
631 struct hlist_node i_hash;
632 struct list_head i_io_list; /* backing dev IO list */
633 #ifdef CONFIG_CGROUP_WRITEBACK
634 struct bdi_writeback *i_wb; /* the associated cgroup wb */
636 /* foreign inode detection, see wbc_detach_inode() */
638 u16 i_wb_frn_avg_time;
639 u16 i_wb_frn_history;
641 struct list_head i_lru; /* inode LRU list */
642 struct list_head i_sb_list;
643 struct list_head i_wb_list; /* backing dev writeback list */
645 struct hlist_head i_dentry;
646 struct rcu_head i_rcu;
648 atomic64_t i_version;
650 atomic_t i_dio_count;
651 atomic_t i_writecount;
653 atomic_t i_readcount; /* struct files open RO */
655 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
656 struct file_lock_context *i_flctx;
657 struct address_space i_data;
658 struct list_head i_devices;
660 struct pipe_inode_info *i_pipe;
661 struct block_device *i_bdev;
669 #ifdef CONFIG_FSNOTIFY
670 __u32 i_fsnotify_mask; /* all events this inode cares about */
671 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
674 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
675 struct fscrypt_info *i_crypt_info;
678 void *i_private; /* fs or device private pointer */
679 } __randomize_layout;
681 static inline unsigned int i_blocksize(const struct inode *node)
683 return (1 << node->i_blkbits);
686 static inline int inode_unhashed(struct inode *inode)
688 return hlist_unhashed(&inode->i_hash);
692 * __mark_inode_dirty expects inodes to be hashed. Since we don't
693 * want special inodes in the fileset inode space, we make them
694 * appear hashed, but do not put on any lists. hlist_del()
695 * will work fine and require no locking.
697 static inline void inode_fake_hash(struct inode *inode)
699 hlist_add_fake(&inode->i_hash);
703 * inode->i_mutex nesting subclasses for the lock validator:
705 * 0: the object of the current VFS operation
709 * 4: second non-directory
710 * 5: second parent (when locking independent directories in rename)
712 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
713 * non-directories at once.
715 * The locking order between these classes is
716 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
718 enum inode_i_mutex_lock_class
728 static inline void inode_lock(struct inode *inode)
730 down_write(&inode->i_rwsem);
733 static inline void inode_unlock(struct inode *inode)
735 up_write(&inode->i_rwsem);
738 static inline void inode_lock_shared(struct inode *inode)
740 down_read(&inode->i_rwsem);
743 static inline void inode_unlock_shared(struct inode *inode)
745 up_read(&inode->i_rwsem);
748 static inline int inode_trylock(struct inode *inode)
750 return down_write_trylock(&inode->i_rwsem);
753 static inline int inode_trylock_shared(struct inode *inode)
755 return down_read_trylock(&inode->i_rwsem);
758 static inline int inode_is_locked(struct inode *inode)
760 return rwsem_is_locked(&inode->i_rwsem);
763 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
765 down_write_nested(&inode->i_rwsem, subclass);
768 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
770 down_read_nested(&inode->i_rwsem, subclass);
773 void lock_two_nondirectories(struct inode *, struct inode*);
774 void unlock_two_nondirectories(struct inode *, struct inode*);
777 * NOTE: in a 32bit arch with a preemptable kernel and
778 * an UP compile the i_size_read/write must be atomic
779 * with respect to the local cpu (unlike with preempt disabled),
780 * but they don't need to be atomic with respect to other cpus like in
781 * true SMP (so they need either to either locally disable irq around
782 * the read or for example on x86 they can be still implemented as a
783 * cmpxchg8b without the need of the lock prefix). For SMP compiles
784 * and 64bit archs it makes no difference if preempt is enabled or not.
786 static inline loff_t i_size_read(const struct inode *inode)
788 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
793 seq = read_seqcount_begin(&inode->i_size_seqcount);
794 i_size = inode->i_size;
795 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
797 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
801 i_size = inode->i_size;
805 return inode->i_size;
810 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
811 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
812 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
814 static inline void i_size_write(struct inode *inode, loff_t i_size)
816 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
818 write_seqcount_begin(&inode->i_size_seqcount);
819 inode->i_size = i_size;
820 write_seqcount_end(&inode->i_size_seqcount);
822 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
824 inode->i_size = i_size;
827 inode->i_size = i_size;
831 static inline unsigned iminor(const struct inode *inode)
833 return MINOR(inode->i_rdev);
836 static inline unsigned imajor(const struct inode *inode)
838 return MAJOR(inode->i_rdev);
841 extern struct block_device *I_BDEV(struct inode *inode);
844 rwlock_t lock; /* protects pid, uid, euid fields */
845 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
846 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
847 kuid_t uid, euid; /* uid/euid of process setting the owner */
848 int signum; /* posix.1b rt signal to be delivered on IO */
852 * Track a single file's readahead state
854 struct file_ra_state {
855 pgoff_t start; /* where readahead started */
856 unsigned int size; /* # of readahead pages */
857 unsigned int async_size; /* do asynchronous readahead when
858 there are only # of pages ahead */
860 unsigned int ra_pages; /* Maximum readahead window */
861 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
862 loff_t prev_pos; /* Cache last read() position */
866 * Check if @index falls in the readahead windows.
868 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
870 return (index >= ra->start &&
871 index < ra->start + ra->size);
876 struct llist_node fu_llist;
877 struct rcu_head fu_rcuhead;
880 struct inode *f_inode; /* cached value */
881 const struct file_operations *f_op;
884 * Protects f_ep_links, f_flags.
885 * Must not be taken from IRQ context.
888 enum rw_hint f_write_hint;
889 atomic_long_t f_count;
890 unsigned int f_flags;
892 struct mutex f_pos_lock;
894 struct fown_struct f_owner;
895 const struct cred *f_cred;
896 struct file_ra_state f_ra;
899 #ifdef CONFIG_SECURITY
902 /* needed for tty driver, and maybe others */
906 /* Used by fs/eventpoll.c to link all the hooks to this file */
907 struct list_head f_ep_links;
908 struct list_head f_tfile_llink;
909 #endif /* #ifdef CONFIG_EPOLL */
910 struct address_space *f_mapping;
913 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
918 /* file identifier */
919 unsigned char f_handle[0];
922 static inline struct file *get_file(struct file *f)
924 atomic_long_inc(&f->f_count);
927 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
928 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
929 #define file_count(x) atomic_long_read(&(x)->f_count)
931 #define MAX_NON_LFS ((1UL<<31) - 1)
933 /* Page cache limit. The filesystems should put that into their s_maxbytes
934 limits, otherwise bad things can happen in VM. */
935 #if BITS_PER_LONG==32
936 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
937 #elif BITS_PER_LONG==64
938 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
943 #define FL_DELEG 4 /* NFSv4 delegation */
944 #define FL_ACCESS 8 /* not trying to lock, just looking */
945 #define FL_EXISTS 16 /* when unlocking, test for existence */
946 #define FL_LEASE 32 /* lease held on this file */
947 #define FL_CLOSE 64 /* unlock on close */
948 #define FL_SLEEP 128 /* A blocking lock */
949 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
950 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
951 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
952 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
954 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
957 * Special return value from posix_lock_file() and vfs_lock_file() for
958 * asynchronous locking.
960 #define FILE_LOCK_DEFERRED 1
962 /* legacy typedef, should eventually be removed */
963 typedef void *fl_owner_t;
967 struct file_lock_operations {
968 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
969 void (*fl_release_private)(struct file_lock *);
972 struct lock_manager_operations {
973 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
974 unsigned long (*lm_owner_key)(struct file_lock *);
975 fl_owner_t (*lm_get_owner)(fl_owner_t);
976 void (*lm_put_owner)(fl_owner_t);
977 void (*lm_notify)(struct file_lock *); /* unblock callback */
978 int (*lm_grant)(struct file_lock *, int);
979 bool (*lm_break)(struct file_lock *);
980 int (*lm_change)(struct file_lock *, int, struct list_head *);
981 void (*lm_setup)(struct file_lock *, void **);
984 struct lock_manager {
985 struct list_head list;
987 * NFSv4 and up also want opens blocked during the grace period;
994 void locks_start_grace(struct net *, struct lock_manager *);
995 void locks_end_grace(struct lock_manager *);
996 bool locks_in_grace(struct net *);
997 bool opens_in_grace(struct net *);
999 /* that will die - we need it for nfs_lock_info */
1000 #include <linux/nfs_fs_i.h>
1003 * struct file_lock represents a generic "file lock". It's used to represent
1004 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1005 * note that the same struct is used to represent both a request for a lock and
1006 * the lock itself, but the same object is never used for both.
1008 * FIXME: should we create a separate "struct lock_request" to help distinguish
1011 * The varous i_flctx lists are ordered by:
1014 * 2) lock range start
1017 * Obviously, the last two criteria only matter for POSIX locks.
1020 struct file_lock *fl_next; /* singly linked list for this inode */
1021 struct list_head fl_list; /* link into file_lock_context */
1022 struct hlist_node fl_link; /* node in global lists */
1023 struct list_head fl_block; /* circular list of blocked processes */
1024 fl_owner_t fl_owner;
1025 unsigned int fl_flags;
1026 unsigned char fl_type;
1027 unsigned int fl_pid;
1028 int fl_link_cpu; /* what cpu's list is this on? */
1029 wait_queue_head_t fl_wait;
1030 struct file *fl_file;
1034 struct fasync_struct * fl_fasync; /* for lease break notifications */
1035 /* for lease breaks: */
1036 unsigned long fl_break_time;
1037 unsigned long fl_downgrade_time;
1039 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1040 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1042 struct nfs_lock_info nfs_fl;
1043 struct nfs4_lock_info nfs4_fl;
1045 struct list_head link; /* link in AFS vnode's pending_locks list */
1046 int state; /* state of grant or error if -ve */
1049 } __randomize_layout;
1051 struct file_lock_context {
1052 spinlock_t flc_lock;
1053 struct list_head flc_flock;
1054 struct list_head flc_posix;
1055 struct list_head flc_lease;
1058 /* The following constant reflects the upper bound of the file/locking space */
1060 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1061 #define OFFSET_MAX INT_LIMIT(loff_t)
1062 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1065 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1068 * Return the inode to use for locking
1070 * For overlayfs this should be the overlay inode, not the real inode returned
1071 * by file_inode(). For any other fs file_inode(filp) and locks_inode(filp) are
1074 static inline struct inode *locks_inode(const struct file *f)
1076 return f->f_path.dentry->d_inode;
1079 #ifdef CONFIG_FILE_LOCKING
1080 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1081 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1084 #if BITS_PER_LONG == 32
1085 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1086 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1090 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1091 extern int fcntl_getlease(struct file *filp);
1094 void locks_free_lock_context(struct inode *inode);
1095 void locks_free_lock(struct file_lock *fl);
1096 extern void locks_init_lock(struct file_lock *);
1097 extern struct file_lock * locks_alloc_lock(void);
1098 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1099 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1100 extern void locks_remove_posix(struct file *, fl_owner_t);
1101 extern void locks_remove_file(struct file *);
1102 extern void locks_release_private(struct file_lock *);
1103 extern void posix_test_lock(struct file *, struct file_lock *);
1104 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1105 extern int posix_unblock_lock(struct file_lock *);
1106 extern int vfs_test_lock(struct file *, struct file_lock *);
1107 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1108 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1109 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1110 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1111 extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1112 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1113 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1114 extern int lease_modify(struct file_lock *, int, struct list_head *);
1115 struct files_struct;
1116 extern void show_fd_locks(struct seq_file *f,
1117 struct file *filp, struct files_struct *files);
1118 #else /* !CONFIG_FILE_LOCKING */
1119 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1120 struct flock __user *user)
1125 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1126 unsigned int cmd, struct flock __user *user)
1131 #if BITS_PER_LONG == 32
1132 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1133 struct flock64 __user *user)
1138 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1139 unsigned int cmd, struct flock64 __user *user)
1144 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1149 static inline int fcntl_getlease(struct file *filp)
1155 locks_free_lock_context(struct inode *inode)
1159 static inline void locks_init_lock(struct file_lock *fl)
1164 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1169 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1174 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1179 static inline void locks_remove_file(struct file *filp)
1184 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1189 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1190 struct file_lock *conflock)
1195 static inline int posix_unblock_lock(struct file_lock *waiter)
1200 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1205 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1206 struct file_lock *fl, struct file_lock *conf)
1211 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1216 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1221 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1226 static inline void lease_get_mtime(struct inode *inode,
1227 struct timespec64 *time)
1232 static inline int generic_setlease(struct file *filp, long arg,
1233 struct file_lock **flp, void **priv)
1238 static inline int vfs_setlease(struct file *filp, long arg,
1239 struct file_lock **lease, void **priv)
1244 static inline int lease_modify(struct file_lock *fl, int arg,
1245 struct list_head *dispose)
1250 struct files_struct;
1251 static inline void show_fd_locks(struct seq_file *f,
1252 struct file *filp, struct files_struct *files) {}
1253 #endif /* !CONFIG_FILE_LOCKING */
1255 static inline struct inode *file_inode(const struct file *f)
1260 static inline struct dentry *file_dentry(const struct file *file)
1262 return d_real(file->f_path.dentry, file_inode(file), 0, 0);
1265 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1267 return locks_lock_inode_wait(locks_inode(filp), fl);
1270 struct fasync_struct {
1274 struct fasync_struct *fa_next; /* singly linked list */
1275 struct file *fa_file;
1276 struct rcu_head fa_rcu;
1279 #define FASYNC_MAGIC 0x4601
1281 /* SMP safe fasync helpers: */
1282 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1283 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1284 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1285 extern struct fasync_struct *fasync_alloc(void);
1286 extern void fasync_free(struct fasync_struct *);
1288 /* can be called from interrupts */
1289 extern void kill_fasync(struct fasync_struct **, int, int);
1291 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1292 extern int f_setown(struct file *filp, unsigned long arg, int force);
1293 extern void f_delown(struct file *filp);
1294 extern pid_t f_getown(struct file *filp);
1295 extern int send_sigurg(struct fown_struct *fown);
1298 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1299 * represented in both.
1301 #define SB_RDONLY 1 /* Mount read-only */
1302 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1303 #define SB_NODEV 4 /* Disallow access to device special files */
1304 #define SB_NOEXEC 8 /* Disallow program execution */
1305 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1306 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1307 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1308 #define SB_NOATIME 1024 /* Do not update access times. */
1309 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1310 #define SB_SILENT 32768
1311 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1312 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1313 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1314 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1316 /* These sb flags are internal to the kernel */
1317 #define SB_SUBMOUNT (1<<26)
1318 #define SB_NOREMOTELOCK (1<<27)
1319 #define SB_NOSEC (1<<28)
1320 #define SB_BORN (1<<29)
1321 #define SB_ACTIVE (1<<30)
1322 #define SB_NOUSER (1<<31)
1328 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1329 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1330 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1331 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1332 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1335 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1336 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1337 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1338 #define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */
1340 /* sb->s_iflags to limit user namespace mounts */
1341 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1342 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1343 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1345 /* Possible states of 'frozen' field */
1347 SB_UNFROZEN = 0, /* FS is unfrozen */
1348 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1349 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1350 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1351 * internal threads if needed) */
1352 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1355 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1358 int frozen; /* Is sb frozen? */
1359 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1360 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1363 struct super_block {
1364 struct list_head s_list; /* Keep this first */
1365 dev_t s_dev; /* search index; _not_ kdev_t */
1366 unsigned char s_blocksize_bits;
1367 unsigned long s_blocksize;
1368 loff_t s_maxbytes; /* Max file size */
1369 struct file_system_type *s_type;
1370 const struct super_operations *s_op;
1371 const struct dquot_operations *dq_op;
1372 const struct quotactl_ops *s_qcop;
1373 const struct export_operations *s_export_op;
1374 unsigned long s_flags;
1375 unsigned long s_iflags; /* internal SB_I_* flags */
1376 unsigned long s_magic;
1377 struct dentry *s_root;
1378 struct rw_semaphore s_umount;
1381 #ifdef CONFIG_SECURITY
1384 const struct xattr_handler **s_xattr;
1385 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
1386 const struct fscrypt_operations *s_cop;
1388 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1389 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1390 struct block_device *s_bdev;
1391 struct backing_dev_info *s_bdi;
1392 struct mtd_info *s_mtd;
1393 struct hlist_node s_instances;
1394 unsigned int s_quota_types; /* Bitmask of supported quota types */
1395 struct quota_info s_dquot; /* Diskquota specific options */
1397 struct sb_writers s_writers;
1399 char s_id[32]; /* Informational name */
1400 uuid_t s_uuid; /* UUID */
1402 void *s_fs_info; /* Filesystem private info */
1403 unsigned int s_max_links;
1406 /* Granularity of c/m/atime in ns.
1407 Cannot be worse than a second */
1411 * The next field is for VFS *only*. No filesystems have any business
1412 * even looking at it. You had been warned.
1414 struct mutex s_vfs_rename_mutex; /* Kludge */
1417 * Filesystem subtype. If non-empty the filesystem type field
1418 * in /proc/mounts will be "type.subtype"
1422 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1425 * Saved pool identifier for cleancache (-1 means none)
1427 int cleancache_poolid;
1429 struct shrinker s_shrink; /* per-sb shrinker handle */
1431 /* Number of inodes with nlink == 0 but still referenced */
1432 atomic_long_t s_remove_count;
1434 /* Being remounted read-only */
1435 int s_readonly_remount;
1437 /* AIO completions deferred from interrupt context */
1438 struct workqueue_struct *s_dio_done_wq;
1439 struct hlist_head s_pins;
1442 * Owning user namespace and default context in which to
1443 * interpret filesystem uids, gids, quotas, device nodes,
1444 * xattrs and security labels.
1446 struct user_namespace *s_user_ns;
1449 * Keep the lru lists last in the structure so they always sit on their
1450 * own individual cachelines.
1452 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1453 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1454 struct rcu_head rcu;
1455 struct work_struct destroy_work;
1457 struct mutex s_sync_lock; /* sync serialisation lock */
1460 * Indicates how deep in a filesystem stack this SB is
1464 /* s_inode_list_lock protects s_inodes */
1465 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1466 struct list_head s_inodes; /* all inodes */
1468 spinlock_t s_inode_wblist_lock;
1469 struct list_head s_inodes_wb; /* writeback inodes */
1470 } __randomize_layout;
1472 /* Helper functions so that in most cases filesystems will
1473 * not need to deal directly with kuid_t and kgid_t and can
1474 * instead deal with the raw numeric values that are stored
1475 * in the filesystem.
1477 static inline uid_t i_uid_read(const struct inode *inode)
1479 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1482 static inline gid_t i_gid_read(const struct inode *inode)
1484 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1487 static inline void i_uid_write(struct inode *inode, uid_t uid)
1489 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1492 static inline void i_gid_write(struct inode *inode, gid_t gid)
1494 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1497 extern struct timespec64 timespec64_trunc(struct timespec64 t, unsigned gran);
1498 extern struct timespec64 current_time(struct inode *inode);
1501 * Snapshotting support.
1504 void __sb_end_write(struct super_block *sb, int level);
1505 int __sb_start_write(struct super_block *sb, int level, bool wait);
1507 #define __sb_writers_acquired(sb, lev) \
1508 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1509 #define __sb_writers_release(sb, lev) \
1510 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1513 * sb_end_write - drop write access to a superblock
1514 * @sb: the super we wrote to
1516 * Decrement number of writers to the filesystem. Wake up possible waiters
1517 * wanting to freeze the filesystem.
1519 static inline void sb_end_write(struct super_block *sb)
1521 __sb_end_write(sb, SB_FREEZE_WRITE);
1525 * sb_end_pagefault - drop write access to a superblock from a page fault
1526 * @sb: the super we wrote to
1528 * Decrement number of processes handling write page fault to the filesystem.
1529 * Wake up possible waiters wanting to freeze the filesystem.
1531 static inline void sb_end_pagefault(struct super_block *sb)
1533 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1537 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1538 * @sb: the super we wrote to
1540 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1541 * waiters wanting to freeze the filesystem.
1543 static inline void sb_end_intwrite(struct super_block *sb)
1545 __sb_end_write(sb, SB_FREEZE_FS);
1549 * sb_start_write - get write access to a superblock
1550 * @sb: the super we write to
1552 * When a process wants to write data or metadata to a file system (i.e. dirty
1553 * a page or an inode), it should embed the operation in a sb_start_write() -
1554 * sb_end_write() pair to get exclusion against file system freezing. This
1555 * function increments number of writers preventing freezing. If the file
1556 * system is already frozen, the function waits until the file system is
1559 * Since freeze protection behaves as a lock, users have to preserve
1560 * ordering of freeze protection and other filesystem locks. Generally,
1561 * freeze protection should be the outermost lock. In particular, we have:
1564 * -> i_mutex (write path, truncate, directory ops, ...)
1565 * -> s_umount (freeze_super, thaw_super)
1567 static inline void sb_start_write(struct super_block *sb)
1569 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1572 static inline int sb_start_write_trylock(struct super_block *sb)
1574 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1578 * sb_start_pagefault - get write access to a superblock from a page fault
1579 * @sb: the super we write to
1581 * When a process starts handling write page fault, it should embed the
1582 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1583 * exclusion against file system freezing. This is needed since the page fault
1584 * is going to dirty a page. This function increments number of running page
1585 * faults preventing freezing. If the file system is already frozen, the
1586 * function waits until the file system is thawed.
1588 * Since page fault freeze protection behaves as a lock, users have to preserve
1589 * ordering of freeze protection and other filesystem locks. It is advised to
1590 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1591 * handling code implies lock dependency:
1594 * -> sb_start_pagefault
1596 static inline void sb_start_pagefault(struct super_block *sb)
1598 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1602 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1603 * @sb: the super we write to
1605 * This is the third level of protection against filesystem freezing. It is
1606 * free for use by a filesystem. The only requirement is that it must rank
1607 * below sb_start_pagefault.
1609 * For example filesystem can call sb_start_intwrite() when starting a
1610 * transaction which somewhat eases handling of freezing for internal sources
1611 * of filesystem changes (internal fs threads, discarding preallocation on file
1614 static inline void sb_start_intwrite(struct super_block *sb)
1616 __sb_start_write(sb, SB_FREEZE_FS, true);
1619 static inline int sb_start_intwrite_trylock(struct super_block *sb)
1621 return __sb_start_write(sb, SB_FREEZE_FS, false);
1625 extern bool inode_owner_or_capable(const struct inode *inode);
1628 * VFS helper functions..
1630 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1631 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1632 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1633 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1634 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1635 extern int vfs_rmdir(struct inode *, struct dentry *);
1636 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1637 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1638 extern int vfs_whiteout(struct inode *, struct dentry *);
1640 extern struct dentry *vfs_tmpfile(struct dentry *dentry, umode_t mode,
1643 int vfs_mkobj(struct dentry *, umode_t,
1644 int (*f)(struct dentry *, umode_t, void *),
1648 * VFS file helper functions.
1650 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1652 extern bool may_open_dev(const struct path *path);
1654 * VFS FS_IOC_FIEMAP helper definitions.
1656 struct fiemap_extent_info {
1657 unsigned int fi_flags; /* Flags as passed from user */
1658 unsigned int fi_extents_mapped; /* Number of mapped extents */
1659 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1660 struct fiemap_extent __user *fi_extents_start; /* Start of
1661 fiemap_extent array */
1663 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1664 u64 phys, u64 len, u32 flags);
1665 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1670 * NOTE! These match bits 12..15 of stat.st_mode
1671 * (ie "(i_mode >> 12) & 15").
1673 #define DT_UNKNOWN 0
1684 * This is the "filldir" function type, used by readdir() to let
1685 * the kernel specify what kind of dirent layout it wants to have.
1686 * This allows the kernel to read directories into kernel space or
1687 * to have different dirent layouts depending on the binary type.
1690 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1693 struct dir_context {
1698 struct block_device_operations;
1700 /* These macros are for out of kernel modules to test that
1701 * the kernel supports the unlocked_ioctl and compat_ioctl
1702 * fields in struct file_operations. */
1703 #define HAVE_COMPAT_IOCTL 1
1704 #define HAVE_UNLOCKED_IOCTL 1
1707 * These flags let !MMU mmap() govern direct device mapping vs immediate
1708 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1710 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1711 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1712 * NOMMU_MAP_READ: Can be mapped for reading
1713 * NOMMU_MAP_WRITE: Can be mapped for writing
1714 * NOMMU_MAP_EXEC: Can be mapped for execution
1716 #define NOMMU_MAP_COPY 0x00000001
1717 #define NOMMU_MAP_DIRECT 0x00000008
1718 #define NOMMU_MAP_READ VM_MAYREAD
1719 #define NOMMU_MAP_WRITE VM_MAYWRITE
1720 #define NOMMU_MAP_EXEC VM_MAYEXEC
1722 #define NOMMU_VMFLAGS \
1723 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1728 struct file_operations {
1729 struct module *owner;
1730 loff_t (*llseek) (struct file *, loff_t, int);
1731 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1732 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1733 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1734 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1735 int (*iterate) (struct file *, struct dir_context *);
1736 int (*iterate_shared) (struct file *, struct dir_context *);
1737 __poll_t (*poll) (struct file *, struct poll_table_struct *);
1738 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1739 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1740 int (*mmap) (struct file *, struct vm_area_struct *);
1741 unsigned long mmap_supported_flags;
1742 int (*open) (struct inode *, struct file *);
1743 int (*flush) (struct file *, fl_owner_t id);
1744 int (*release) (struct inode *, struct file *);
1745 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1746 int (*fasync) (int, struct file *, int);
1747 int (*lock) (struct file *, int, struct file_lock *);
1748 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1749 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1750 int (*check_flags)(int);
1751 int (*flock) (struct file *, int, struct file_lock *);
1752 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1753 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1754 int (*setlease)(struct file *, long, struct file_lock **, void **);
1755 long (*fallocate)(struct file *file, int mode, loff_t offset,
1757 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1759 unsigned (*mmap_capabilities)(struct file *);
1761 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1762 loff_t, size_t, unsigned int);
1763 int (*clone_file_range)(struct file *, loff_t, struct file *, loff_t,
1765 ssize_t (*dedupe_file_range)(struct file *, u64, u64, struct file *,
1767 } __randomize_layout;
1769 struct inode_operations {
1770 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1771 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1772 int (*permission) (struct inode *, int);
1773 struct posix_acl * (*get_acl)(struct inode *, int);
1775 int (*readlink) (struct dentry *, char __user *,int);
1777 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1778 int (*link) (struct dentry *,struct inode *,struct dentry *);
1779 int (*unlink) (struct inode *,struct dentry *);
1780 int (*symlink) (struct inode *,struct dentry *,const char *);
1781 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1782 int (*rmdir) (struct inode *,struct dentry *);
1783 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1784 int (*rename) (struct inode *, struct dentry *,
1785 struct inode *, struct dentry *, unsigned int);
1786 int (*setattr) (struct dentry *, struct iattr *);
1787 int (*getattr) (const struct path *, struct kstat *, u32, unsigned int);
1788 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1789 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1791 int (*update_time)(struct inode *, struct timespec64 *, int);
1792 int (*atomic_open)(struct inode *, struct dentry *,
1793 struct file *, unsigned open_flag,
1794 umode_t create_mode);
1795 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1796 int (*set_acl)(struct inode *, struct posix_acl *, int);
1797 } ____cacheline_aligned;
1799 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
1800 struct iov_iter *iter)
1802 return file->f_op->read_iter(kio, iter);
1805 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
1806 struct iov_iter *iter)
1808 return file->f_op->write_iter(kio, iter);
1811 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
1813 return file->f_op->mmap(file, vma);
1816 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1817 unsigned long nr_segs, unsigned long fast_segs,
1818 struct iovec *fast_pointer,
1819 struct iovec **ret_pointer);
1821 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1822 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1823 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1824 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1825 unsigned long, loff_t *, rwf_t);
1826 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1827 loff_t, size_t, unsigned int);
1828 extern int vfs_clone_file_prep_inodes(struct inode *inode_in, loff_t pos_in,
1829 struct inode *inode_out, loff_t pos_out,
1830 u64 *len, bool is_dedupe);
1831 extern int vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1832 struct file *file_out, loff_t pos_out, u64 len);
1833 extern int vfs_dedupe_file_range_compare(struct inode *src, loff_t srcoff,
1834 struct inode *dest, loff_t destoff,
1835 loff_t len, bool *is_same);
1836 extern int vfs_dedupe_file_range(struct file *file,
1837 struct file_dedupe_range *same);
1839 struct super_operations {
1840 struct inode *(*alloc_inode)(struct super_block *sb);
1841 void (*destroy_inode)(struct inode *);
1843 void (*dirty_inode) (struct inode *, int flags);
1844 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1845 int (*drop_inode) (struct inode *);
1846 void (*evict_inode) (struct inode *);
1847 void (*put_super) (struct super_block *);
1848 int (*sync_fs)(struct super_block *sb, int wait);
1849 int (*freeze_super) (struct super_block *);
1850 int (*freeze_fs) (struct super_block *);
1851 int (*thaw_super) (struct super_block *);
1852 int (*unfreeze_fs) (struct super_block *);
1853 int (*statfs) (struct dentry *, struct kstatfs *);
1854 int (*remount_fs) (struct super_block *, int *, char *);
1855 void (*umount_begin) (struct super_block *);
1857 int (*show_options)(struct seq_file *, struct dentry *);
1858 int (*show_devname)(struct seq_file *, struct dentry *);
1859 int (*show_path)(struct seq_file *, struct dentry *);
1860 int (*show_stats)(struct seq_file *, struct dentry *);
1862 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1863 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1864 struct dquot **(*get_dquots)(struct inode *);
1866 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1867 long (*nr_cached_objects)(struct super_block *,
1868 struct shrink_control *);
1869 long (*free_cached_objects)(struct super_block *,
1870 struct shrink_control *);
1874 * Inode flags - they have no relation to superblock flags now
1876 #define S_SYNC 1 /* Writes are synced at once */
1877 #define S_NOATIME 2 /* Do not update access times */
1878 #define S_APPEND 4 /* Append-only file */
1879 #define S_IMMUTABLE 8 /* Immutable file */
1880 #define S_DEAD 16 /* removed, but still open directory */
1881 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1882 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1883 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1884 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1885 #define S_PRIVATE 512 /* Inode is fs-internal */
1886 #define S_IMA 1024 /* Inode has an associated IMA struct */
1887 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1888 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1889 #ifdef CONFIG_FS_DAX
1890 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1892 #define S_DAX 0 /* Make all the DAX code disappear */
1894 #define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */
1897 * Note that nosuid etc flags are inode-specific: setting some file-system
1898 * flags just means all the inodes inherit those flags by default. It might be
1899 * possible to override it selectively if you really wanted to with some
1900 * ioctl() that is not currently implemented.
1902 * Exception: SB_RDONLY is always applied to the entire file system.
1904 * Unfortunately, it is possible to change a filesystems flags with it mounted
1905 * with files in use. This means that all of the inodes will not have their
1906 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1907 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1909 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1911 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
1912 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
1913 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
1914 ((inode)->i_flags & S_SYNC))
1915 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
1916 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1917 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
1918 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
1919 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
1921 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1922 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1923 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1924 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
1926 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1927 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1928 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1929 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1930 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1931 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1932 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1933 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1934 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
1936 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1937 (inode)->i_rdev == WHITEOUT_DEV)
1939 static inline bool HAS_UNMAPPED_ID(struct inode *inode)
1941 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
1944 static inline enum rw_hint file_write_hint(struct file *file)
1946 if (file->f_write_hint != WRITE_LIFE_NOT_SET)
1947 return file->f_write_hint;
1949 return file_inode(file)->i_write_hint;
1952 static inline int iocb_flags(struct file *file);
1954 static inline u16 ki_hint_validate(enum rw_hint hint)
1956 typeof(((struct kiocb *)0)->ki_hint) max_hint = -1;
1958 if (hint <= max_hint)
1963 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
1965 *kiocb = (struct kiocb) {
1967 .ki_flags = iocb_flags(filp),
1968 .ki_hint = ki_hint_validate(file_write_hint(filp)),
1969 .ki_ioprio = IOPRIO_PRIO_VALUE(IOPRIO_CLASS_NONE, 0),
1974 * Inode state bits. Protected by inode->i_lock
1976 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1977 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1979 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1980 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1981 * various stages of removing an inode.
1983 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1985 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1986 * fdatasync(). i_atime is the usual cause.
1987 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1988 * these changes separately from I_DIRTY_SYNC so that we
1989 * don't have to write inode on fdatasync() when only
1990 * mtime has changed in it.
1991 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1992 * I_NEW Serves as both a mutex and completion notification.
1993 * New inodes set I_NEW. If two processes both create
1994 * the same inode, one of them will release its inode and
1995 * wait for I_NEW to be released before returning.
1996 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1997 * also cause waiting on I_NEW, without I_NEW actually
1998 * being set. find_inode() uses this to prevent returning
1999 * nearly-dead inodes.
2000 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2001 * is zero. I_FREEING must be set when I_WILL_FREE is
2003 * I_FREEING Set when inode is about to be freed but still has dirty
2004 * pages or buffers attached or the inode itself is still
2006 * I_CLEAR Added by clear_inode(). In this state the inode is
2007 * clean and can be destroyed. Inode keeps I_FREEING.
2009 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2010 * prohibited for many purposes. iget() must wait for
2011 * the inode to be completely released, then create it
2012 * anew. Other functions will just ignore such inodes,
2013 * if appropriate. I_NEW is used for waiting.
2015 * I_SYNC Writeback of inode is running. The bit is set during
2016 * data writeback, and cleared with a wakeup on the bit
2017 * address once it is done. The bit is also used to pin
2018 * the inode in memory for flusher thread.
2020 * I_REFERENCED Marks the inode as recently references on the LRU list.
2022 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2024 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2025 * synchronize competing switching instances and to tell
2026 * wb stat updates to grab the i_pages lock. See
2027 * inode_switch_wb_work_fn() for details.
2029 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2030 * and work dirs among overlayfs mounts.
2032 * I_CREATING New object's inode in the middle of setting up.
2034 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2036 #define I_DIRTY_SYNC (1 << 0)
2037 #define I_DIRTY_DATASYNC (1 << 1)
2038 #define I_DIRTY_PAGES (1 << 2)
2040 #define I_NEW (1 << __I_NEW)
2041 #define I_WILL_FREE (1 << 4)
2042 #define I_FREEING (1 << 5)
2043 #define I_CLEAR (1 << 6)
2045 #define I_SYNC (1 << __I_SYNC)
2046 #define I_REFERENCED (1 << 8)
2047 #define __I_DIO_WAKEUP 9
2048 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2049 #define I_LINKABLE (1 << 10)
2050 #define I_DIRTY_TIME (1 << 11)
2051 #define __I_DIRTY_TIME_EXPIRED 12
2052 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
2053 #define I_WB_SWITCH (1 << 13)
2054 #define I_OVL_INUSE (1 << 14)
2055 #define I_CREATING (1 << 15)
2057 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2058 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2059 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2061 extern void __mark_inode_dirty(struct inode *, int);
2062 static inline void mark_inode_dirty(struct inode *inode)
2064 __mark_inode_dirty(inode, I_DIRTY);
2067 static inline void mark_inode_dirty_sync(struct inode *inode)
2069 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2072 extern void inc_nlink(struct inode *inode);
2073 extern void drop_nlink(struct inode *inode);
2074 extern void clear_nlink(struct inode *inode);
2075 extern void set_nlink(struct inode *inode, unsigned int nlink);
2077 static inline void inode_inc_link_count(struct inode *inode)
2080 mark_inode_dirty(inode);
2083 static inline void inode_dec_link_count(struct inode *inode)
2086 mark_inode_dirty(inode);
2089 enum file_time_flags {
2096 extern void touch_atime(const struct path *);
2097 static inline void file_accessed(struct file *file)
2099 if (!(file->f_flags & O_NOATIME))
2100 touch_atime(&file->f_path);
2103 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2104 int sync_inode_metadata(struct inode *inode, int wait);
2106 struct file_system_type {
2109 #define FS_REQUIRES_DEV 1
2110 #define FS_BINARY_MOUNTDATA 2
2111 #define FS_HAS_SUBTYPE 4
2112 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2113 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2114 struct dentry *(*mount) (struct file_system_type *, int,
2115 const char *, void *);
2116 void (*kill_sb) (struct super_block *);
2117 struct module *owner;
2118 struct file_system_type * next;
2119 struct hlist_head fs_supers;
2121 struct lock_class_key s_lock_key;
2122 struct lock_class_key s_umount_key;
2123 struct lock_class_key s_vfs_rename_key;
2124 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2126 struct lock_class_key i_lock_key;
2127 struct lock_class_key i_mutex_key;
2128 struct lock_class_key i_mutex_dir_key;
2131 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2133 extern struct dentry *mount_ns(struct file_system_type *fs_type,
2134 int flags, void *data, void *ns, struct user_namespace *user_ns,
2135 int (*fill_super)(struct super_block *, void *, int));
2137 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2138 int flags, const char *dev_name, void *data,
2139 int (*fill_super)(struct super_block *, void *, int));
2141 static inline struct dentry *mount_bdev(struct file_system_type *fs_type,
2142 int flags, const char *dev_name, void *data,
2143 int (*fill_super)(struct super_block *, void *, int))
2145 return ERR_PTR(-ENODEV);
2148 extern struct dentry *mount_single(struct file_system_type *fs_type,
2149 int flags, void *data,
2150 int (*fill_super)(struct super_block *, void *, int));
2151 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2152 int flags, void *data,
2153 int (*fill_super)(struct super_block *, void *, int));
2154 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2155 void generic_shutdown_super(struct super_block *sb);
2157 void kill_block_super(struct super_block *sb);
2159 static inline void kill_block_super(struct super_block *sb)
2164 void kill_anon_super(struct super_block *sb);
2165 void kill_litter_super(struct super_block *sb);
2166 void deactivate_super(struct super_block *sb);
2167 void deactivate_locked_super(struct super_block *sb);
2168 int set_anon_super(struct super_block *s, void *data);
2169 int get_anon_bdev(dev_t *);
2170 void free_anon_bdev(dev_t);
2171 struct super_block *sget_userns(struct file_system_type *type,
2172 int (*test)(struct super_block *,void *),
2173 int (*set)(struct super_block *,void *),
2174 int flags, struct user_namespace *user_ns,
2176 struct super_block *sget(struct file_system_type *type,
2177 int (*test)(struct super_block *,void *),
2178 int (*set)(struct super_block *,void *),
2179 int flags, void *data);
2180 extern struct dentry *mount_pseudo_xattr(struct file_system_type *, char *,
2181 const struct super_operations *ops,
2182 const struct xattr_handler **xattr,
2183 const struct dentry_operations *dops,
2186 static inline struct dentry *
2187 mount_pseudo(struct file_system_type *fs_type, char *name,
2188 const struct super_operations *ops,
2189 const struct dentry_operations *dops, unsigned long magic)
2191 return mount_pseudo_xattr(fs_type, name, ops, NULL, dops, magic);
2194 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2195 #define fops_get(fops) \
2196 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2197 #define fops_put(fops) \
2198 do { if (fops) module_put((fops)->owner); } while(0)
2200 * This one is to be used *ONLY* from ->open() instances.
2201 * fops must be non-NULL, pinned down *and* module dependencies
2202 * should be sufficient to pin the caller down as well.
2204 #define replace_fops(f, fops) \
2206 struct file *__file = (f); \
2207 fops_put(__file->f_op); \
2208 BUG_ON(!(__file->f_op = (fops))); \
2211 extern int register_filesystem(struct file_system_type *);
2212 extern int unregister_filesystem(struct file_system_type *);
2213 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2214 #define kern_mount(type) kern_mount_data(type, NULL)
2215 extern void kern_unmount(struct vfsmount *mnt);
2216 extern int may_umount_tree(struct vfsmount *);
2217 extern int may_umount(struct vfsmount *);
2218 extern long do_mount(const char *, const char __user *,
2219 const char *, unsigned long, void *);
2220 extern struct vfsmount *collect_mounts(const struct path *);
2221 extern void drop_collected_mounts(struct vfsmount *);
2222 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2224 extern int vfs_statfs(const struct path *, struct kstatfs *);
2225 extern int user_statfs(const char __user *, struct kstatfs *);
2226 extern int fd_statfs(int, struct kstatfs *);
2227 extern int freeze_super(struct super_block *super);
2228 extern int thaw_super(struct super_block *super);
2229 extern bool our_mnt(struct vfsmount *mnt);
2230 extern __printf(2, 3)
2231 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2232 extern int super_setup_bdi(struct super_block *sb);
2234 extern int current_umask(void);
2236 extern void ihold(struct inode * inode);
2237 extern void iput(struct inode *);
2238 extern int generic_update_time(struct inode *, struct timespec64 *, int);
2241 extern struct kobject *fs_kobj;
2243 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2245 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2246 extern int locks_mandatory_locked(struct file *);
2247 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2250 * Candidates for mandatory locking have the setgid bit set
2251 * but no group execute bit - an otherwise meaningless combination.
2254 static inline int __mandatory_lock(struct inode *ino)
2256 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2260 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2261 * otherwise these will be advisory locks
2264 static inline int mandatory_lock(struct inode *ino)
2266 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2269 static inline int locks_verify_locked(struct file *file)
2271 if (mandatory_lock(locks_inode(file)))
2272 return locks_mandatory_locked(file);
2276 static inline int locks_verify_truncate(struct inode *inode,
2280 if (!inode->i_flctx || !mandatory_lock(inode))
2283 if (size < inode->i_size) {
2284 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2287 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2292 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2294 static inline int locks_mandatory_locked(struct file *file)
2299 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2300 loff_t start, loff_t end, unsigned char type)
2305 static inline int __mandatory_lock(struct inode *inode)
2310 static inline int mandatory_lock(struct inode *inode)
2315 static inline int locks_verify_locked(struct file *file)
2320 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2326 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2329 #ifdef CONFIG_FILE_LOCKING
2330 static inline int break_lease(struct inode *inode, unsigned int mode)
2333 * Since this check is lockless, we must ensure that any refcounts
2334 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2335 * could end up racing with tasks trying to set a new lease on this
2339 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2340 return __break_lease(inode, mode, FL_LEASE);
2344 static inline int break_deleg(struct inode *inode, unsigned int mode)
2347 * Since this check is lockless, we must ensure that any refcounts
2348 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2349 * could end up racing with tasks trying to set a new lease on this
2353 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2354 return __break_lease(inode, mode, FL_DELEG);
2358 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2362 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2363 if (ret == -EWOULDBLOCK && delegated_inode) {
2364 *delegated_inode = inode;
2370 static inline int break_deleg_wait(struct inode **delegated_inode)
2374 ret = break_deleg(*delegated_inode, O_WRONLY);
2375 iput(*delegated_inode);
2376 *delegated_inode = NULL;
2380 static inline int break_layout(struct inode *inode, bool wait)
2383 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2384 return __break_lease(inode,
2385 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2390 #else /* !CONFIG_FILE_LOCKING */
2391 static inline int break_lease(struct inode *inode, unsigned int mode)
2396 static inline int break_deleg(struct inode *inode, unsigned int mode)
2401 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2406 static inline int break_deleg_wait(struct inode **delegated_inode)
2412 static inline int break_layout(struct inode *inode, bool wait)
2417 #endif /* CONFIG_FILE_LOCKING */
2422 const char *name; /* pointer to actual string */
2423 const __user char *uptr; /* original userland pointer */
2425 struct audit_names *aname;
2429 extern long vfs_truncate(const struct path *, loff_t);
2430 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2432 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2434 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2436 extern struct file *file_open_name(struct filename *, int, umode_t);
2437 extern struct file *filp_open(const char *, int, umode_t);
2438 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2439 const char *, int, umode_t);
2440 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2441 static inline struct file *file_clone_open(struct file *file)
2443 return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2445 extern int filp_close(struct file *, fl_owner_t id);
2447 extern struct filename *getname_flags(const char __user *, int, int *);
2448 extern struct filename *getname(const char __user *);
2449 extern struct filename *getname_kernel(const char *);
2450 extern void putname(struct filename *name);
2452 extern int finish_open(struct file *file, struct dentry *dentry,
2453 int (*open)(struct inode *, struct file *));
2454 extern int finish_no_open(struct file *file, struct dentry *dentry);
2458 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2461 extern void __init vfs_caches_init_early(void);
2462 extern void __init vfs_caches_init(void);
2464 extern struct kmem_cache *names_cachep;
2466 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2467 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2470 extern int register_blkdev(unsigned int, const char *);
2471 extern void unregister_blkdev(unsigned int, const char *);
2472 extern void bdev_unhash_inode(dev_t dev);
2473 extern struct block_device *bdget(dev_t);
2474 extern struct block_device *bdgrab(struct block_device *bdev);
2475 extern void bd_set_size(struct block_device *, loff_t size);
2476 extern void bd_forget(struct inode *inode);
2477 extern void bdput(struct block_device *);
2478 extern void invalidate_bdev(struct block_device *);
2479 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2480 extern int sync_blockdev(struct block_device *bdev);
2481 extern void kill_bdev(struct block_device *);
2482 extern struct super_block *freeze_bdev(struct block_device *);
2483 extern void emergency_thaw_all(void);
2484 extern void emergency_thaw_bdev(struct super_block *sb);
2485 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2486 extern int fsync_bdev(struct block_device *);
2488 extern struct super_block *blockdev_superblock;
2490 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2492 return sb == blockdev_superblock;
2495 static inline void bd_forget(struct inode *inode) {}
2496 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2497 static inline void kill_bdev(struct block_device *bdev) {}
2498 static inline void invalidate_bdev(struct block_device *bdev) {}
2500 static inline struct super_block *freeze_bdev(struct block_device *sb)
2505 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2510 static inline int emergency_thaw_bdev(struct super_block *sb)
2515 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2519 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2524 extern int sync_filesystem(struct super_block *);
2525 extern const struct file_operations def_blk_fops;
2526 extern const struct file_operations def_chr_fops;
2528 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2529 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2530 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2531 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2532 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2534 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2536 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2537 extern int __blkdev_reread_part(struct block_device *bdev);
2538 extern int blkdev_reread_part(struct block_device *bdev);
2541 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2542 extern void bd_unlink_disk_holder(struct block_device *bdev,
2543 struct gendisk *disk);
2545 static inline int bd_link_disk_holder(struct block_device *bdev,
2546 struct gendisk *disk)
2550 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2551 struct gendisk *disk)
2558 #define CHRDEV_MAJOR_MAX 512
2559 /* Marks the bottom of the first segment of free char majors */
2560 #define CHRDEV_MAJOR_DYN_END 234
2561 /* Marks the top and bottom of the second segment of free char majors */
2562 #define CHRDEV_MAJOR_DYN_EXT_START 511
2563 #define CHRDEV_MAJOR_DYN_EXT_END 384
2565 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2566 extern int register_chrdev_region(dev_t, unsigned, const char *);
2567 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2568 unsigned int count, const char *name,
2569 const struct file_operations *fops);
2570 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2571 unsigned int count, const char *name);
2572 extern void unregister_chrdev_region(dev_t, unsigned);
2573 extern void chrdev_show(struct seq_file *,off_t);
2575 static inline int register_chrdev(unsigned int major, const char *name,
2576 const struct file_operations *fops)
2578 return __register_chrdev(major, 0, 256, name, fops);
2581 static inline void unregister_chrdev(unsigned int major, const char *name)
2583 __unregister_chrdev(major, 0, 256, name);
2586 /* fs/block_dev.c */
2587 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2588 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2591 #define BLKDEV_MAJOR_MAX 512
2592 extern const char *__bdevname(dev_t, char *buffer);
2593 extern const char *bdevname(struct block_device *bdev, char *buffer);
2594 extern struct block_device *lookup_bdev(const char *);
2595 extern void blkdev_show(struct seq_file *,off_t);
2598 #define BLKDEV_MAJOR_MAX 0
2601 extern void init_special_inode(struct inode *, umode_t, dev_t);
2603 /* Invalid inode operations -- fs/bad_inode.c */
2604 extern void make_bad_inode(struct inode *);
2605 extern bool is_bad_inode(struct inode *);
2608 extern void check_disk_size_change(struct gendisk *disk,
2609 struct block_device *bdev, bool verbose);
2610 extern int revalidate_disk(struct gendisk *);
2611 extern int check_disk_change(struct block_device *);
2612 extern int __invalidate_device(struct block_device *, bool);
2613 extern int invalidate_partition(struct gendisk *, int);
2615 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2616 pgoff_t start, pgoff_t end);
2618 static inline void invalidate_remote_inode(struct inode *inode)
2620 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2621 S_ISLNK(inode->i_mode))
2622 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2624 extern int invalidate_inode_pages2(struct address_space *mapping);
2625 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2626 pgoff_t start, pgoff_t end);
2627 extern int write_inode_now(struct inode *, int);
2628 extern int filemap_fdatawrite(struct address_space *);
2629 extern int filemap_flush(struct address_space *);
2630 extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2631 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2634 static inline int filemap_fdatawait(struct address_space *mapping)
2636 return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2639 extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2641 extern int filemap_write_and_wait(struct address_space *mapping);
2642 extern int filemap_write_and_wait_range(struct address_space *mapping,
2643 loff_t lstart, loff_t lend);
2644 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2645 loff_t start, loff_t end, int sync_mode);
2646 extern int filemap_fdatawrite_range(struct address_space *mapping,
2647 loff_t start, loff_t end);
2648 extern int filemap_check_errors(struct address_space *mapping);
2649 extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2651 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2653 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2654 extern int __must_check file_write_and_wait_range(struct file *file,
2655 loff_t start, loff_t end);
2657 static inline int file_write_and_wait(struct file *file)
2659 return file_write_and_wait_range(file, 0, LLONG_MAX);
2663 * filemap_set_wb_err - set a writeback error on an address_space
2664 * @mapping: mapping in which to set writeback error
2665 * @err: error to be set in mapping
2667 * When writeback fails in some way, we must record that error so that
2668 * userspace can be informed when fsync and the like are called. We endeavor
2669 * to report errors on any file that was open at the time of the error. Some
2670 * internal callers also need to know when writeback errors have occurred.
2672 * When a writeback error occurs, most filesystems will want to call
2673 * filemap_set_wb_err to record the error in the mapping so that it will be
2674 * automatically reported whenever fsync is called on the file.
2676 static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2678 /* Fastpath for common case of no error */
2680 __filemap_set_wb_err(mapping, err);
2684 * filemap_check_wb_error - has an error occurred since the mark was sampled?
2685 * @mapping: mapping to check for writeback errors
2686 * @since: previously-sampled errseq_t
2688 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2689 * the given value was sampled.
2691 * If it has then report the latest error set, otherwise return 0.
2693 static inline int filemap_check_wb_err(struct address_space *mapping,
2696 return errseq_check(&mapping->wb_err, since);
2700 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2701 * @mapping: mapping to be sampled
2703 * Writeback errors are always reported relative to a particular sample point
2704 * in the past. This function provides those sample points.
2706 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2708 return errseq_sample(&mapping->wb_err);
2711 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2713 extern int vfs_fsync(struct file *file, int datasync);
2716 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2717 * to already be updated for the write, and will return either the amount
2718 * of bytes passed in, or an error if syncing the file failed.
2720 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2722 if (iocb->ki_flags & IOCB_DSYNC) {
2723 int ret = vfs_fsync_range(iocb->ki_filp,
2724 iocb->ki_pos - count, iocb->ki_pos - 1,
2725 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2733 extern void emergency_sync(void);
2734 extern void emergency_remount(void);
2736 extern sector_t bmap(struct inode *, sector_t);
2738 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2739 extern int inode_permission(struct inode *, int);
2740 extern int generic_permission(struct inode *, int);
2741 extern int __check_sticky(struct inode *dir, struct inode *inode);
2743 static inline bool execute_ok(struct inode *inode)
2745 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2748 static inline void file_start_write(struct file *file)
2750 if (!S_ISREG(file_inode(file)->i_mode))
2752 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2755 static inline bool file_start_write_trylock(struct file *file)
2757 if (!S_ISREG(file_inode(file)->i_mode))
2759 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2762 static inline void file_end_write(struct file *file)
2764 if (!S_ISREG(file_inode(file)->i_mode))
2766 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2769 static inline int do_clone_file_range(struct file *file_in, loff_t pos_in,
2770 struct file *file_out, loff_t pos_out,
2775 file_start_write(file_out);
2776 ret = vfs_clone_file_range(file_in, pos_in, file_out, pos_out, len);
2777 file_end_write(file_out);
2783 * get_write_access() gets write permission for a file.
2784 * put_write_access() releases this write permission.
2785 * This is used for regular files.
2786 * We cannot support write (and maybe mmap read-write shared) accesses and
2787 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2788 * can have the following values:
2789 * 0: no writers, no VM_DENYWRITE mappings
2790 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2791 * > 0: (i_writecount) users are writing to the file.
2793 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2794 * except for the cases where we don't hold i_writecount yet. Then we need to
2795 * use {get,deny}_write_access() - these functions check the sign and refuse
2796 * to do the change if sign is wrong.
2798 static inline int get_write_access(struct inode *inode)
2800 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2802 static inline int deny_write_access(struct file *file)
2804 struct inode *inode = file_inode(file);
2805 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2807 static inline void put_write_access(struct inode * inode)
2809 atomic_dec(&inode->i_writecount);
2811 static inline void allow_write_access(struct file *file)
2814 atomic_inc(&file_inode(file)->i_writecount);
2816 static inline bool inode_is_open_for_write(const struct inode *inode)
2818 return atomic_read(&inode->i_writecount) > 0;
2822 static inline void i_readcount_dec(struct inode *inode)
2824 BUG_ON(!atomic_read(&inode->i_readcount));
2825 atomic_dec(&inode->i_readcount);
2827 static inline void i_readcount_inc(struct inode *inode)
2829 atomic_inc(&inode->i_readcount);
2832 static inline void i_readcount_dec(struct inode *inode)
2836 static inline void i_readcount_inc(struct inode *inode)
2841 extern int do_pipe_flags(int *, int);
2843 #define __kernel_read_file_id(id) \
2844 id(UNKNOWN, unknown) \
2845 id(FIRMWARE, firmware) \
2846 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2847 id(MODULE, kernel-module) \
2848 id(KEXEC_IMAGE, kexec-image) \
2849 id(KEXEC_INITRAMFS, kexec-initramfs) \
2850 id(POLICY, security-policy) \
2851 id(X509_CERTIFICATE, x509-certificate) \
2854 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2855 #define __fid_stringify(dummy, str) #str,
2857 enum kernel_read_file_id {
2858 __kernel_read_file_id(__fid_enumify)
2861 static const char * const kernel_read_file_str[] = {
2862 __kernel_read_file_id(__fid_stringify)
2865 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
2867 if ((unsigned)id >= READING_MAX_ID)
2868 return kernel_read_file_str[READING_UNKNOWN];
2870 return kernel_read_file_str[id];
2873 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2874 enum kernel_read_file_id);
2875 extern int kernel_read_file_from_path(const char *, void **, loff_t *, loff_t,
2876 enum kernel_read_file_id);
2877 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2878 enum kernel_read_file_id);
2879 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
2880 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
2881 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
2882 extern struct file * open_exec(const char *);
2884 /* fs/dcache.c -- generic fs support functions */
2885 extern bool is_subdir(struct dentry *, struct dentry *);
2886 extern bool path_is_under(const struct path *, const struct path *);
2888 extern char *file_path(struct file *, char *, int);
2890 #include <linux/err.h>
2892 /* needed for stackable file system support */
2893 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2895 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2897 extern int inode_init_always(struct super_block *, struct inode *);
2898 extern void inode_init_once(struct inode *);
2899 extern void address_space_init_once(struct address_space *mapping);
2900 extern struct inode * igrab(struct inode *);
2901 extern ino_t iunique(struct super_block *, ino_t);
2902 extern int inode_needs_sync(struct inode *inode);
2903 extern int generic_delete_inode(struct inode *inode);
2904 static inline int generic_drop_inode(struct inode *inode)
2906 return !inode->i_nlink || inode_unhashed(inode);
2909 extern struct inode *ilookup5_nowait(struct super_block *sb,
2910 unsigned long hashval, int (*test)(struct inode *, void *),
2912 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2913 int (*test)(struct inode *, void *), void *data);
2914 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2916 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
2917 int (*test)(struct inode *, void *),
2918 int (*set)(struct inode *, void *),
2920 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2921 extern struct inode * iget_locked(struct super_block *, unsigned long);
2922 extern struct inode *find_inode_nowait(struct super_block *,
2924 int (*match)(struct inode *,
2925 unsigned long, void *),
2927 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2928 extern int insert_inode_locked(struct inode *);
2929 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2930 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2932 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2934 extern void unlock_new_inode(struct inode *);
2935 extern void discard_new_inode(struct inode *);
2936 extern unsigned int get_next_ino(void);
2937 extern void evict_inodes(struct super_block *sb);
2939 extern void __iget(struct inode * inode);
2940 extern void iget_failed(struct inode *);
2941 extern void clear_inode(struct inode *);
2942 extern void __destroy_inode(struct inode *);
2943 extern struct inode *new_inode_pseudo(struct super_block *sb);
2944 extern struct inode *new_inode(struct super_block *sb);
2945 extern void free_inode_nonrcu(struct inode *inode);
2946 extern int should_remove_suid(struct dentry *);
2947 extern int file_remove_privs(struct file *);
2949 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2950 static inline void insert_inode_hash(struct inode *inode)
2952 __insert_inode_hash(inode, inode->i_ino);
2955 extern void __remove_inode_hash(struct inode *);
2956 static inline void remove_inode_hash(struct inode *inode)
2958 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2959 __remove_inode_hash(inode);
2962 extern void inode_sb_list_add(struct inode *inode);
2965 extern int bdev_read_only(struct block_device *);
2967 extern int set_blocksize(struct block_device *, int);
2968 extern int sb_set_blocksize(struct super_block *, int);
2969 extern int sb_min_blocksize(struct super_block *, int);
2971 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2972 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2973 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2974 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2975 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2976 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2977 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
2978 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2980 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
2982 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
2985 /* fs/block_dev.c */
2986 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2987 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2988 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2990 extern void block_sync_page(struct page *page);
2993 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2994 struct pipe_inode_info *, size_t, unsigned int);
2995 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2996 struct file *, loff_t *, size_t, unsigned int);
2997 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2998 struct file *out, loff_t *, size_t len, unsigned int flags);
2999 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
3000 loff_t *opos, size_t len, unsigned int flags);
3004 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3005 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3006 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
3007 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3008 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3009 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3010 int whence, loff_t maxsize, loff_t eof);
3011 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3012 int whence, loff_t size);
3013 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3014 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3015 extern int generic_file_open(struct inode * inode, struct file * filp);
3016 extern int nonseekable_open(struct inode * inode, struct file * filp);
3019 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3020 loff_t file_offset);
3023 /* need locking between buffered and direct access */
3026 /* filesystem does not support filling holes */
3027 DIO_SKIP_HOLES = 0x02,
3030 void dio_end_io(struct bio *bio);
3031 void dio_warn_stale_pagecache(struct file *filp);
3033 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3034 struct block_device *bdev, struct iov_iter *iter,
3035 get_block_t get_block,
3036 dio_iodone_t end_io, dio_submit_t submit_io,
3039 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3040 struct inode *inode,
3041 struct iov_iter *iter,
3042 get_block_t get_block)
3044 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3045 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3049 void inode_dio_wait(struct inode *inode);
3052 * inode_dio_begin - signal start of a direct I/O requests
3053 * @inode: inode the direct I/O happens on
3055 * This is called once we've finished processing a direct I/O request,
3056 * and is used to wake up callers waiting for direct I/O to be quiesced.
3058 static inline void inode_dio_begin(struct inode *inode)
3060 atomic_inc(&inode->i_dio_count);
3064 * inode_dio_end - signal finish of a direct I/O requests
3065 * @inode: inode the direct I/O happens on
3067 * This is called once we've finished processing a direct I/O request,
3068 * and is used to wake up callers waiting for direct I/O to be quiesced.
3070 static inline void inode_dio_end(struct inode *inode)
3072 if (atomic_dec_and_test(&inode->i_dio_count))
3073 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3076 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3079 extern const struct file_operations generic_ro_fops;
3081 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3083 extern int readlink_copy(char __user *, int, const char *);
3084 extern int page_readlink(struct dentry *, char __user *, int);
3085 extern const char *page_get_link(struct dentry *, struct inode *,
3086 struct delayed_call *);
3087 extern void page_put_link(void *);
3088 extern int __page_symlink(struct inode *inode, const char *symname, int len,
3090 extern int page_symlink(struct inode *inode, const char *symname, int len);
3091 extern const struct inode_operations page_symlink_inode_operations;
3092 extern void kfree_link(void *);
3093 extern void generic_fillattr(struct inode *, struct kstat *);
3094 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3095 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3096 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3097 void inode_add_bytes(struct inode *inode, loff_t bytes);
3098 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3099 void inode_sub_bytes(struct inode *inode, loff_t bytes);
3100 static inline loff_t __inode_get_bytes(struct inode *inode)
3102 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3104 loff_t inode_get_bytes(struct inode *inode);
3105 void inode_set_bytes(struct inode *inode, loff_t bytes);
3106 const char *simple_get_link(struct dentry *, struct inode *,
3107 struct delayed_call *);
3108 extern const struct inode_operations simple_symlink_inode_operations;
3110 extern int iterate_dir(struct file *, struct dir_context *);
3112 extern int vfs_statx(int, const char __user *, int, struct kstat *, u32);
3113 extern int vfs_statx_fd(unsigned int, struct kstat *, u32, unsigned int);
3115 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3117 return vfs_statx(AT_FDCWD, filename, AT_NO_AUTOMOUNT,
3118 stat, STATX_BASIC_STATS);
3120 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3122 return vfs_statx(AT_FDCWD, name, AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT,
3123 stat, STATX_BASIC_STATS);
3125 static inline int vfs_fstatat(int dfd, const char __user *filename,
3126 struct kstat *stat, int flags)
3128 return vfs_statx(dfd, filename, flags | AT_NO_AUTOMOUNT,
3129 stat, STATX_BASIC_STATS);
3131 static inline int vfs_fstat(int fd, struct kstat *stat)
3133 return vfs_statx_fd(fd, stat, STATX_BASIC_STATS, 0);
3137 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3138 extern int vfs_readlink(struct dentry *, char __user *, int);
3140 extern int __generic_block_fiemap(struct inode *inode,
3141 struct fiemap_extent_info *fieinfo,
3142 loff_t start, loff_t len,
3143 get_block_t *get_block);
3144 extern int generic_block_fiemap(struct inode *inode,
3145 struct fiemap_extent_info *fieinfo, u64 start,
3146 u64 len, get_block_t *get_block);
3148 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3149 extern void put_filesystem(struct file_system_type *fs);
3150 extern struct file_system_type *get_fs_type(const char *name);
3151 extern struct super_block *get_super(struct block_device *);
3152 extern struct super_block *get_super_thawed(struct block_device *);
3153 extern struct super_block *get_super_exclusive_thawed(struct block_device *bdev);
3154 extern struct super_block *get_active_super(struct block_device *bdev);
3155 extern void drop_super(struct super_block *sb);
3156 extern void drop_super_exclusive(struct super_block *sb);
3157 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3158 extern void iterate_supers_type(struct file_system_type *,
3159 void (*)(struct super_block *, void *), void *);
3161 extern int dcache_dir_open(struct inode *, struct file *);
3162 extern int dcache_dir_close(struct inode *, struct file *);
3163 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3164 extern int dcache_readdir(struct file *, struct dir_context *);
3165 extern int simple_setattr(struct dentry *, struct iattr *);
3166 extern int simple_getattr(const struct path *, struct kstat *, u32, unsigned int);
3167 extern int simple_statfs(struct dentry *, struct kstatfs *);
3168 extern int simple_open(struct inode *inode, struct file *file);
3169 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3170 extern int simple_unlink(struct inode *, struct dentry *);
3171 extern int simple_rmdir(struct inode *, struct dentry *);
3172 extern int simple_rename(struct inode *, struct dentry *,
3173 struct inode *, struct dentry *, unsigned int);
3174 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3175 extern int noop_set_page_dirty(struct page *page);
3176 extern void noop_invalidatepage(struct page *page, unsigned int offset,
3177 unsigned int length);
3178 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3179 extern int simple_empty(struct dentry *);
3180 extern int simple_readpage(struct file *file, struct page *page);
3181 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3182 loff_t pos, unsigned len, unsigned flags,
3183 struct page **pagep, void **fsdata);
3184 extern int simple_write_end(struct file *file, struct address_space *mapping,
3185 loff_t pos, unsigned len, unsigned copied,
3186 struct page *page, void *fsdata);
3187 extern int always_delete_dentry(const struct dentry *);
3188 extern struct inode *alloc_anon_inode(struct super_block *);
3189 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3190 extern const struct dentry_operations simple_dentry_operations;
3192 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3193 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3194 extern const struct file_operations simple_dir_operations;
3195 extern const struct inode_operations simple_dir_inode_operations;
3196 extern void make_empty_dir_inode(struct inode *inode);
3197 extern bool is_empty_dir_inode(struct inode *inode);
3198 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3199 struct dentry *d_alloc_name(struct dentry *, const char *);
3200 extern int simple_fill_super(struct super_block *, unsigned long,
3201 const struct tree_descr *);
3202 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3203 extern void simple_release_fs(struct vfsmount **mount, int *count);
3205 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3206 loff_t *ppos, const void *from, size_t available);
3207 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3208 const void __user *from, size_t count);
3210 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3211 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3213 extern int generic_check_addressable(unsigned, u64);
3215 #ifdef CONFIG_MIGRATION
3216 extern int buffer_migrate_page(struct address_space *,
3217 struct page *, struct page *,
3220 #define buffer_migrate_page NULL
3223 extern int setattr_prepare(struct dentry *, struct iattr *);
3224 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3225 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
3227 extern int file_update_time(struct file *file);
3229 static inline bool io_is_direct(struct file *filp)
3231 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
3234 static inline bool vma_is_dax(struct vm_area_struct *vma)
3236 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3239 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3241 struct inode *inode;
3245 if (!vma_is_dax(vma))
3247 inode = file_inode(vma->vm_file);
3248 if (S_ISCHR(inode->i_mode))
3249 return false; /* device-dax */
3253 static inline int iocb_flags(struct file *file)
3256 if (file->f_flags & O_APPEND)
3258 if (io_is_direct(file))
3260 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3262 if (file->f_flags & __O_SYNC)
3267 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3269 if (unlikely(flags & ~RWF_SUPPORTED))
3272 if (flags & RWF_NOWAIT) {
3273 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3275 ki->ki_flags |= IOCB_NOWAIT;
3277 if (flags & RWF_HIPRI)
3278 ki->ki_flags |= IOCB_HIPRI;
3279 if (flags & RWF_DSYNC)
3280 ki->ki_flags |= IOCB_DSYNC;
3281 if (flags & RWF_SYNC)
3282 ki->ki_flags |= (IOCB_DSYNC | IOCB_SYNC);
3283 if (flags & RWF_APPEND)
3284 ki->ki_flags |= IOCB_APPEND;
3288 static inline ino_t parent_ino(struct dentry *dentry)
3293 * Don't strictly need d_lock here? If the parent ino could change
3294 * then surely we'd have a deeper race in the caller?
3296 spin_lock(&dentry->d_lock);
3297 res = dentry->d_parent->d_inode->i_ino;
3298 spin_unlock(&dentry->d_lock);
3302 /* Transaction based IO helpers */
3305 * An argresp is stored in an allocated page and holds the
3306 * size of the argument or response, along with its content
3308 struct simple_transaction_argresp {
3313 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3315 char *simple_transaction_get(struct file *file, const char __user *buf,
3317 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3318 size_t size, loff_t *pos);
3319 int simple_transaction_release(struct inode *inode, struct file *file);
3321 void simple_transaction_set(struct file *file, size_t n);
3324 * simple attribute files
3326 * These attributes behave similar to those in sysfs:
3328 * Writing to an attribute immediately sets a value, an open file can be
3329 * written to multiple times.
3331 * Reading from an attribute creates a buffer from the value that might get
3332 * read with multiple read calls. When the attribute has been read
3333 * completely, no further read calls are possible until the file is opened
3336 * All attributes contain a text representation of a numeric value
3337 * that are accessed with the get() and set() functions.
3339 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3340 static int __fops ## _open(struct inode *inode, struct file *file) \
3342 __simple_attr_check_format(__fmt, 0ull); \
3343 return simple_attr_open(inode, file, __get, __set, __fmt); \
3345 static const struct file_operations __fops = { \
3346 .owner = THIS_MODULE, \
3347 .open = __fops ## _open, \
3348 .release = simple_attr_release, \
3349 .read = simple_attr_read, \
3350 .write = simple_attr_write, \
3351 .llseek = generic_file_llseek, \
3354 static inline __printf(1, 2)
3355 void __simple_attr_check_format(const char *fmt, ...)
3357 /* don't do anything, just let the compiler check the arguments; */
3360 int simple_attr_open(struct inode *inode, struct file *file,
3361 int (*get)(void *, u64 *), int (*set)(void *, u64),
3363 int simple_attr_release(struct inode *inode, struct file *file);
3364 ssize_t simple_attr_read(struct file *file, char __user *buf,
3365 size_t len, loff_t *ppos);
3366 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3367 size_t len, loff_t *ppos);
3370 int proc_nr_files(struct ctl_table *table, int write,
3371 void __user *buffer, size_t *lenp, loff_t *ppos);
3372 int proc_nr_dentry(struct ctl_table *table, int write,
3373 void __user *buffer, size_t *lenp, loff_t *ppos);
3374 int proc_nr_inodes(struct ctl_table *table, int write,
3375 void __user *buffer, size_t *lenp, loff_t *ppos);
3376 int __init get_filesystem_list(char *buf);
3378 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3379 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3381 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3382 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3383 (flag & __FMODE_NONOTIFY)))
3385 static inline bool is_sxid(umode_t mode)
3387 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3390 static inline int check_sticky(struct inode *dir, struct inode *inode)
3392 if (!(dir->i_mode & S_ISVTX))
3395 return __check_sticky(dir, inode);
3398 static inline void inode_has_no_xattr(struct inode *inode)
3400 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3401 inode->i_flags |= S_NOSEC;
3404 static inline bool is_root_inode(struct inode *inode)
3406 return inode == inode->i_sb->s_root->d_inode;
3409 static inline bool dir_emit(struct dir_context *ctx,
3410 const char *name, int namelen,
3411 u64 ino, unsigned type)
3413 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3415 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3417 return ctx->actor(ctx, ".", 1, ctx->pos,
3418 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3420 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3422 return ctx->actor(ctx, "..", 2, ctx->pos,
3423 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3425 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3427 if (ctx->pos == 0) {
3428 if (!dir_emit_dot(file, ctx))
3432 if (ctx->pos == 1) {
3433 if (!dir_emit_dotdot(file, ctx))
3439 static inline bool dir_relax(struct inode *inode)
3441 inode_unlock(inode);
3443 return !IS_DEADDIR(inode);
3446 static inline bool dir_relax_shared(struct inode *inode)
3448 inode_unlock_shared(inode);
3449 inode_lock_shared(inode);
3450 return !IS_DEADDIR(inode);
3453 extern bool path_noexec(const struct path *path);
3454 extern void inode_nohighmem(struct inode *inode);
3456 #endif /* _LINUX_FS_H */