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_KILL_SUID (1 << 11)
183 #define ATTR_KILL_SGID (1 << 12)
184 #define ATTR_FILE (1 << 13)
185 #define ATTR_KILL_PRIV (1 << 14)
186 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
187 #define ATTR_TIMES_SET (1 << 16)
188 #define ATTR_TOUCH (1 << 17)
191 * Whiteout is represented by a char device. The following constants define the
192 * mode and device number to use.
194 #define WHITEOUT_MODE 0
195 #define WHITEOUT_DEV 0
198 * This is the Inode Attributes structure, used for notify_change(). It
199 * uses the above definitions as flags, to know which values have changed.
200 * Also, in this manner, a Filesystem can look at only the values it cares
201 * about. Basically, these are the attributes that the VFS layer can
202 * request to change from the FS layer.
204 * Derek Atkins <warlord@MIT.EDU> 94-10-20
207 unsigned int ia_valid;
212 struct timespec64 ia_atime;
213 struct timespec64 ia_mtime;
214 struct timespec64 ia_ctime;
217 * Not an attribute, but an auxiliary info for filesystems wanting to
218 * implement an ftruncate() like method. NOTE: filesystem should
219 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
221 struct file *ia_file;
225 * Includes for diskquotas.
227 #include <linux/quota.h>
230 * Maximum number of layers of fs stack. Needs to be limited to
231 * prevent kernel stack overflow
233 #define FILESYSTEM_MAX_STACK_DEPTH 2
236 * enum positive_aop_returns - aop return codes with specific semantics
238 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
239 * completed, that the page is still locked, and
240 * should be considered active. The VM uses this hint
241 * to return the page to the active list -- it won't
242 * be a candidate for writeback again in the near
243 * future. Other callers must be careful to unlock
244 * the page if they get this return. Returned by
247 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
248 * unlocked it and the page might have been truncated.
249 * The caller should back up to acquiring a new page and
250 * trying again. The aop will be taking reasonable
251 * precautions not to livelock. If the caller held a page
252 * reference, it should drop it before retrying. Returned
255 * address_space_operation functions return these large constants to indicate
256 * special semantics to the caller. These are much larger than the bytes in a
257 * page to allow for functions that return the number of bytes operated on in a
261 enum positive_aop_returns {
262 AOP_WRITEPAGE_ACTIVATE = 0x80000,
263 AOP_TRUNCATED_PAGE = 0x80001,
266 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
267 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
268 * helper code (eg buffer layer)
269 * to clear GFP_FS from alloc */
272 * oh the beauties of C type declarations.
275 struct address_space;
276 struct writeback_control;
279 * Write life time hint values.
280 * Stored in struct inode as u8.
283 WRITE_LIFE_NOT_SET = 0,
284 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
285 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
286 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
287 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
288 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
291 #define IOCB_EVENTFD (1 << 0)
292 #define IOCB_APPEND (1 << 1)
293 #define IOCB_DIRECT (1 << 2)
294 #define IOCB_HIPRI (1 << 3)
295 #define IOCB_DSYNC (1 << 4)
296 #define IOCB_SYNC (1 << 5)
297 #define IOCB_WRITE (1 << 6)
298 #define IOCB_NOWAIT (1 << 7)
301 struct file *ki_filp;
303 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
307 u16 ki_ioprio; /* See linux/ioprio.h */
308 } __randomize_layout;
310 static inline bool is_sync_kiocb(struct kiocb *kiocb)
312 return kiocb->ki_complete == NULL;
316 * "descriptor" for what we're up to with a read.
317 * This allows us to use the same read code yet
318 * have multiple different users of the data that
319 * we read from a file.
321 * The simplest case just copies the data to user
334 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
335 unsigned long, unsigned long);
337 struct address_space_operations {
338 int (*writepage)(struct page *page, struct writeback_control *wbc);
339 int (*readpage)(struct file *, struct page *);
341 /* Write back some dirty pages from this mapping. */
342 int (*writepages)(struct address_space *, struct writeback_control *);
344 /* Set a page dirty. Return true if this dirtied it */
345 int (*set_page_dirty)(struct page *page);
348 * Reads in the requested pages. Unlike ->readpage(), this is
349 * PURELY used for read-ahead!.
351 int (*readpages)(struct file *filp, struct address_space *mapping,
352 struct list_head *pages, unsigned nr_pages);
354 int (*write_begin)(struct file *, struct address_space *mapping,
355 loff_t pos, unsigned len, unsigned flags,
356 struct page **pagep, void **fsdata);
357 int (*write_end)(struct file *, struct address_space *mapping,
358 loff_t pos, unsigned len, unsigned copied,
359 struct page *page, void *fsdata);
361 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
362 sector_t (*bmap)(struct address_space *, sector_t);
363 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
364 int (*releasepage) (struct page *, gfp_t);
365 void (*freepage)(struct page *);
366 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
368 * migrate the contents of a page to the specified target. If
369 * migrate_mode is MIGRATE_ASYNC, it must not block.
371 int (*migratepage) (struct address_space *,
372 struct page *, struct page *, enum migrate_mode);
373 bool (*isolate_page)(struct page *, isolate_mode_t);
374 void (*putback_page)(struct page *);
375 int (*launder_page) (struct page *);
376 int (*is_partially_uptodate) (struct page *, unsigned long,
378 void (*is_dirty_writeback) (struct page *, bool *, bool *);
379 int (*error_remove_page)(struct address_space *, struct page *);
381 /* swapfile support */
382 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
384 void (*swap_deactivate)(struct file *file);
387 extern const struct address_space_operations empty_aops;
390 * pagecache_write_begin/pagecache_write_end must be used by general code
391 * to write into the pagecache.
393 int pagecache_write_begin(struct file *, struct address_space *mapping,
394 loff_t pos, unsigned len, unsigned flags,
395 struct page **pagep, void **fsdata);
397 int pagecache_write_end(struct file *, struct address_space *mapping,
398 loff_t pos, unsigned len, unsigned copied,
399 struct page *page, void *fsdata);
401 struct address_space {
402 struct inode *host; /* owner: inode, block_device */
403 struct radix_tree_root i_pages; /* cached pages */
404 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
405 struct rb_root_cached i_mmap; /* tree of private and shared mappings */
406 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
407 /* Protected by the i_pages lock */
408 unsigned long nrpages; /* number of total pages */
409 /* number of shadow or DAX exceptional entries */
410 unsigned long nrexceptional;
411 pgoff_t writeback_index;/* writeback starts here */
412 const struct address_space_operations *a_ops; /* methods */
413 unsigned long flags; /* error bits */
414 spinlock_t private_lock; /* for use by the address_space */
415 gfp_t gfp_mask; /* implicit gfp mask for allocations */
416 struct list_head private_list; /* for use by the address_space */
417 void *private_data; /* ditto */
419 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
421 * On most architectures that alignment is already the case; but
422 * must be enforced here for CRIS, to let the least significant bit
423 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
425 struct request_queue;
427 struct block_device {
428 dev_t bd_dev; /* not a kdev_t - it's a search key */
430 struct inode * bd_inode; /* will die */
431 struct super_block * bd_super;
432 struct mutex bd_mutex; /* open/close mutex */
436 bool bd_write_holder;
438 struct list_head bd_holder_disks;
440 struct block_device * bd_contains;
441 unsigned bd_block_size;
443 struct hd_struct * bd_part;
444 /* number of times partitions within this device have been opened. */
445 unsigned bd_part_count;
447 struct gendisk * bd_disk;
448 struct request_queue * bd_queue;
449 struct backing_dev_info *bd_bdi;
450 struct list_head bd_list;
452 * Private data. You must have bd_claim'ed the block_device
453 * to use this. NOTE: bd_claim allows an owner to claim
454 * the same device multiple times, the owner must take special
455 * care to not mess up bd_private for that case.
457 unsigned long bd_private;
459 /* The counter of freeze processes */
460 int bd_fsfreeze_count;
461 /* Mutex for freeze */
462 struct mutex bd_fsfreeze_mutex;
463 } __randomize_layout;
466 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
469 #define PAGECACHE_TAG_DIRTY 0
470 #define PAGECACHE_TAG_WRITEBACK 1
471 #define PAGECACHE_TAG_TOWRITE 2
473 int mapping_tagged(struct address_space *mapping, int tag);
475 static inline void i_mmap_lock_write(struct address_space *mapping)
477 down_write(&mapping->i_mmap_rwsem);
480 static inline void i_mmap_unlock_write(struct address_space *mapping)
482 up_write(&mapping->i_mmap_rwsem);
485 static inline void i_mmap_lock_read(struct address_space *mapping)
487 down_read(&mapping->i_mmap_rwsem);
490 static inline void i_mmap_unlock_read(struct address_space *mapping)
492 up_read(&mapping->i_mmap_rwsem);
496 * Might pages of this file be mapped into userspace?
498 static inline int mapping_mapped(struct address_space *mapping)
500 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
504 * Might pages of this file have been modified in userspace?
505 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
506 * marks vma as VM_SHARED if it is shared, and the file was opened for
507 * writing i.e. vma may be mprotected writable even if now readonly.
509 * If i_mmap_writable is negative, no new writable mappings are allowed. You
510 * can only deny writable mappings, if none exists right now.
512 static inline int mapping_writably_mapped(struct address_space *mapping)
514 return atomic_read(&mapping->i_mmap_writable) > 0;
517 static inline int mapping_map_writable(struct address_space *mapping)
519 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
523 static inline void mapping_unmap_writable(struct address_space *mapping)
525 atomic_dec(&mapping->i_mmap_writable);
528 static inline int mapping_deny_writable(struct address_space *mapping)
530 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
534 static inline void mapping_allow_writable(struct address_space *mapping)
536 atomic_inc(&mapping->i_mmap_writable);
540 * Use sequence counter to get consistent i_size on 32-bit processors.
542 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
543 #include <linux/seqlock.h>
544 #define __NEED_I_SIZE_ORDERED
545 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
547 #define i_size_ordered_init(inode) do { } while (0)
551 #define ACL_NOT_CACHED ((void *)(-1))
552 #define ACL_DONT_CACHE ((void *)(-3))
554 static inline struct posix_acl *
555 uncached_acl_sentinel(struct task_struct *task)
557 return (void *)task + 1;
561 is_uncached_acl(struct posix_acl *acl)
563 return (long)acl & 1;
566 #define IOP_FASTPERM 0x0001
567 #define IOP_LOOKUP 0x0002
568 #define IOP_NOFOLLOW 0x0004
569 #define IOP_XATTR 0x0008
570 #define IOP_DEFAULT_READLINK 0x0010
572 struct fsnotify_mark_connector;
575 * Keep mostly read-only and often accessed (especially for
576 * the RCU path lookup and 'stat' data) fields at the beginning
577 * of the 'struct inode'
581 unsigned short i_opflags;
584 unsigned int i_flags;
586 #ifdef CONFIG_FS_POSIX_ACL
587 struct posix_acl *i_acl;
588 struct posix_acl *i_default_acl;
591 const struct inode_operations *i_op;
592 struct super_block *i_sb;
593 struct address_space *i_mapping;
595 #ifdef CONFIG_SECURITY
599 /* Stat data, not accessed from path walking */
602 * Filesystems may only read i_nlink directly. They shall use the
603 * following functions for modification:
605 * (set|clear|inc|drop)_nlink
606 * inode_(inc|dec)_link_count
609 const unsigned int i_nlink;
610 unsigned int __i_nlink;
614 struct timespec64 i_atime;
615 struct timespec64 i_mtime;
616 struct timespec64 i_ctime;
617 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
618 unsigned short i_bytes;
623 #ifdef __NEED_I_SIZE_ORDERED
624 seqcount_t i_size_seqcount;
628 unsigned long i_state;
629 struct rw_semaphore i_rwsem;
631 unsigned long dirtied_when; /* jiffies of first dirtying */
632 unsigned long dirtied_time_when;
634 struct hlist_node i_hash;
635 struct list_head i_io_list; /* backing dev IO list */
636 #ifdef CONFIG_CGROUP_WRITEBACK
637 struct bdi_writeback *i_wb; /* the associated cgroup wb */
639 /* foreign inode detection, see wbc_detach_inode() */
641 u16 i_wb_frn_avg_time;
642 u16 i_wb_frn_history;
644 struct list_head i_lru; /* inode LRU list */
645 struct list_head i_sb_list;
646 struct list_head i_wb_list; /* backing dev writeback list */
648 struct hlist_head i_dentry;
649 struct rcu_head i_rcu;
651 atomic64_t i_version;
653 atomic_t i_dio_count;
654 atomic_t i_writecount;
656 atomic_t i_readcount; /* struct files open RO */
658 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
659 struct file_lock_context *i_flctx;
660 struct address_space i_data;
661 struct list_head i_devices;
663 struct pipe_inode_info *i_pipe;
664 struct block_device *i_bdev;
672 #ifdef CONFIG_FSNOTIFY
673 __u32 i_fsnotify_mask; /* all events this inode cares about */
674 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
677 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
678 struct fscrypt_info *i_crypt_info;
681 void *i_private; /* fs or device private pointer */
682 } __randomize_layout;
684 static inline unsigned int i_blocksize(const struct inode *node)
686 return (1 << node->i_blkbits);
689 static inline int inode_unhashed(struct inode *inode)
691 return hlist_unhashed(&inode->i_hash);
695 * __mark_inode_dirty expects inodes to be hashed. Since we don't
696 * want special inodes in the fileset inode space, we make them
697 * appear hashed, but do not put on any lists. hlist_del()
698 * will work fine and require no locking.
700 static inline void inode_fake_hash(struct inode *inode)
702 hlist_add_fake(&inode->i_hash);
706 * inode->i_mutex nesting subclasses for the lock validator:
708 * 0: the object of the current VFS operation
712 * 4: second non-directory
713 * 5: second parent (when locking independent directories in rename)
715 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
716 * non-directories at once.
718 * The locking order between these classes is
719 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
721 enum inode_i_mutex_lock_class
731 static inline void inode_lock(struct inode *inode)
733 down_write(&inode->i_rwsem);
736 static inline void inode_unlock(struct inode *inode)
738 up_write(&inode->i_rwsem);
741 static inline void inode_lock_shared(struct inode *inode)
743 down_read(&inode->i_rwsem);
746 static inline void inode_unlock_shared(struct inode *inode)
748 up_read(&inode->i_rwsem);
751 static inline int inode_trylock(struct inode *inode)
753 return down_write_trylock(&inode->i_rwsem);
756 static inline int inode_trylock_shared(struct inode *inode)
758 return down_read_trylock(&inode->i_rwsem);
761 static inline int inode_is_locked(struct inode *inode)
763 return rwsem_is_locked(&inode->i_rwsem);
766 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
768 down_write_nested(&inode->i_rwsem, subclass);
771 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
773 down_read_nested(&inode->i_rwsem, subclass);
776 void lock_two_nondirectories(struct inode *, struct inode*);
777 void unlock_two_nondirectories(struct inode *, struct inode*);
780 * NOTE: in a 32bit arch with a preemptable kernel and
781 * an UP compile the i_size_read/write must be atomic
782 * with respect to the local cpu (unlike with preempt disabled),
783 * but they don't need to be atomic with respect to other cpus like in
784 * true SMP (so they need either to either locally disable irq around
785 * the read or for example on x86 they can be still implemented as a
786 * cmpxchg8b without the need of the lock prefix). For SMP compiles
787 * and 64bit archs it makes no difference if preempt is enabled or not.
789 static inline loff_t i_size_read(const struct inode *inode)
791 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
796 seq = read_seqcount_begin(&inode->i_size_seqcount);
797 i_size = inode->i_size;
798 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
800 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
804 i_size = inode->i_size;
808 return inode->i_size;
813 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
814 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
815 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
817 static inline void i_size_write(struct inode *inode, loff_t i_size)
819 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
821 write_seqcount_begin(&inode->i_size_seqcount);
822 inode->i_size = i_size;
823 write_seqcount_end(&inode->i_size_seqcount);
825 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
827 inode->i_size = i_size;
830 inode->i_size = i_size;
834 static inline unsigned iminor(const struct inode *inode)
836 return MINOR(inode->i_rdev);
839 static inline unsigned imajor(const struct inode *inode)
841 return MAJOR(inode->i_rdev);
844 extern struct block_device *I_BDEV(struct inode *inode);
847 rwlock_t lock; /* protects pid, uid, euid fields */
848 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
849 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
850 kuid_t uid, euid; /* uid/euid of process setting the owner */
851 int signum; /* posix.1b rt signal to be delivered on IO */
855 * Track a single file's readahead state
857 struct file_ra_state {
858 pgoff_t start; /* where readahead started */
859 unsigned int size; /* # of readahead pages */
860 unsigned int async_size; /* do asynchronous readahead when
861 there are only # of pages ahead */
863 unsigned int ra_pages; /* Maximum readahead window */
864 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
865 loff_t prev_pos; /* Cache last read() position */
869 * Check if @index falls in the readahead windows.
871 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
873 return (index >= ra->start &&
874 index < ra->start + ra->size);
879 struct llist_node fu_llist;
880 struct rcu_head fu_rcuhead;
883 struct inode *f_inode; /* cached value */
884 const struct file_operations *f_op;
887 * Protects f_ep_links, f_flags.
888 * Must not be taken from IRQ context.
891 enum rw_hint f_write_hint;
892 atomic_long_t f_count;
893 unsigned int f_flags;
895 struct mutex f_pos_lock;
897 struct fown_struct f_owner;
898 const struct cred *f_cred;
899 struct file_ra_state f_ra;
902 #ifdef CONFIG_SECURITY
905 /* needed for tty driver, and maybe others */
909 /* Used by fs/eventpoll.c to link all the hooks to this file */
910 struct list_head f_ep_links;
911 struct list_head f_tfile_llink;
912 #endif /* #ifdef CONFIG_EPOLL */
913 struct address_space *f_mapping;
916 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
921 /* file identifier */
922 unsigned char f_handle[0];
925 static inline struct file *get_file(struct file *f)
927 atomic_long_inc(&f->f_count);
930 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
931 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
932 #define file_count(x) atomic_long_read(&(x)->f_count)
934 #define MAX_NON_LFS ((1UL<<31) - 1)
936 /* Page cache limit. The filesystems should put that into their s_maxbytes
937 limits, otherwise bad things can happen in VM. */
938 #if BITS_PER_LONG==32
939 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
940 #elif BITS_PER_LONG==64
941 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
946 #define FL_DELEG 4 /* NFSv4 delegation */
947 #define FL_ACCESS 8 /* not trying to lock, just looking */
948 #define FL_EXISTS 16 /* when unlocking, test for existence */
949 #define FL_LEASE 32 /* lease held on this file */
950 #define FL_CLOSE 64 /* unlock on close */
951 #define FL_SLEEP 128 /* A blocking lock */
952 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
953 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
954 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
955 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
957 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
960 * Special return value from posix_lock_file() and vfs_lock_file() for
961 * asynchronous locking.
963 #define FILE_LOCK_DEFERRED 1
965 /* legacy typedef, should eventually be removed */
966 typedef void *fl_owner_t;
970 struct file_lock_operations {
971 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
972 void (*fl_release_private)(struct file_lock *);
975 struct lock_manager_operations {
976 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
977 unsigned long (*lm_owner_key)(struct file_lock *);
978 fl_owner_t (*lm_get_owner)(fl_owner_t);
979 void (*lm_put_owner)(fl_owner_t);
980 void (*lm_notify)(struct file_lock *); /* unblock callback */
981 int (*lm_grant)(struct file_lock *, int);
982 bool (*lm_break)(struct file_lock *);
983 int (*lm_change)(struct file_lock *, int, struct list_head *);
984 void (*lm_setup)(struct file_lock *, void **);
987 struct lock_manager {
988 struct list_head list;
990 * NFSv4 and up also want opens blocked during the grace period;
997 void locks_start_grace(struct net *, struct lock_manager *);
998 void locks_end_grace(struct lock_manager *);
999 bool locks_in_grace(struct net *);
1000 bool opens_in_grace(struct net *);
1002 /* that will die - we need it for nfs_lock_info */
1003 #include <linux/nfs_fs_i.h>
1006 * struct file_lock represents a generic "file lock". It's used to represent
1007 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1008 * note that the same struct is used to represent both a request for a lock and
1009 * the lock itself, but the same object is never used for both.
1011 * FIXME: should we create a separate "struct lock_request" to help distinguish
1014 * The varous i_flctx lists are ordered by:
1017 * 2) lock range start
1020 * Obviously, the last two criteria only matter for POSIX locks.
1023 struct file_lock *fl_next; /* singly linked list for this inode */
1024 struct list_head fl_list; /* link into file_lock_context */
1025 struct hlist_node fl_link; /* node in global lists */
1026 struct list_head fl_block; /* circular list of blocked processes */
1027 fl_owner_t fl_owner;
1028 unsigned int fl_flags;
1029 unsigned char fl_type;
1030 unsigned int fl_pid;
1031 int fl_link_cpu; /* what cpu's list is this on? */
1032 wait_queue_head_t fl_wait;
1033 struct file *fl_file;
1037 struct fasync_struct * fl_fasync; /* for lease break notifications */
1038 /* for lease breaks: */
1039 unsigned long fl_break_time;
1040 unsigned long fl_downgrade_time;
1042 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1043 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1045 struct nfs_lock_info nfs_fl;
1046 struct nfs4_lock_info nfs4_fl;
1048 struct list_head link; /* link in AFS vnode's pending_locks list */
1049 int state; /* state of grant or error if -ve */
1052 } __randomize_layout;
1054 struct file_lock_context {
1055 spinlock_t flc_lock;
1056 struct list_head flc_flock;
1057 struct list_head flc_posix;
1058 struct list_head flc_lease;
1061 /* The following constant reflects the upper bound of the file/locking space */
1063 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1064 #define OFFSET_MAX INT_LIMIT(loff_t)
1065 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1068 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1071 * Return the inode to use for locking
1073 * For overlayfs this should be the overlay inode, not the real inode returned
1074 * by file_inode(). For any other fs file_inode(filp) and locks_inode(filp) are
1077 static inline struct inode *locks_inode(const struct file *f)
1079 return f->f_path.dentry->d_inode;
1082 #ifdef CONFIG_FILE_LOCKING
1083 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1084 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1087 #if BITS_PER_LONG == 32
1088 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1089 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1093 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1094 extern int fcntl_getlease(struct file *filp);
1097 void locks_free_lock_context(struct inode *inode);
1098 void locks_free_lock(struct file_lock *fl);
1099 extern void locks_init_lock(struct file_lock *);
1100 extern struct file_lock * locks_alloc_lock(void);
1101 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1102 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1103 extern void locks_remove_posix(struct file *, fl_owner_t);
1104 extern void locks_remove_file(struct file *);
1105 extern void locks_release_private(struct file_lock *);
1106 extern void posix_test_lock(struct file *, struct file_lock *);
1107 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1108 extern int posix_unblock_lock(struct file_lock *);
1109 extern int vfs_test_lock(struct file *, struct file_lock *);
1110 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1111 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1112 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1113 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1114 extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1115 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1116 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1117 extern int lease_modify(struct file_lock *, int, struct list_head *);
1118 struct files_struct;
1119 extern void show_fd_locks(struct seq_file *f,
1120 struct file *filp, struct files_struct *files);
1121 #else /* !CONFIG_FILE_LOCKING */
1122 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1123 struct flock __user *user)
1128 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1129 unsigned int cmd, struct flock __user *user)
1134 #if BITS_PER_LONG == 32
1135 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1136 struct flock64 __user *user)
1141 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1142 unsigned int cmd, struct flock64 __user *user)
1147 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1152 static inline int fcntl_getlease(struct file *filp)
1158 locks_free_lock_context(struct inode *inode)
1162 static inline void locks_init_lock(struct file_lock *fl)
1167 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1172 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1177 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1182 static inline void locks_remove_file(struct file *filp)
1187 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1192 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1193 struct file_lock *conflock)
1198 static inline int posix_unblock_lock(struct file_lock *waiter)
1203 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1208 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1209 struct file_lock *fl, struct file_lock *conf)
1214 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1219 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1224 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1229 static inline void lease_get_mtime(struct inode *inode,
1230 struct timespec64 *time)
1235 static inline int generic_setlease(struct file *filp, long arg,
1236 struct file_lock **flp, void **priv)
1241 static inline int vfs_setlease(struct file *filp, long arg,
1242 struct file_lock **lease, void **priv)
1247 static inline int lease_modify(struct file_lock *fl, int arg,
1248 struct list_head *dispose)
1253 struct files_struct;
1254 static inline void show_fd_locks(struct seq_file *f,
1255 struct file *filp, struct files_struct *files) {}
1256 #endif /* !CONFIG_FILE_LOCKING */
1258 static inline struct inode *file_inode(const struct file *f)
1263 static inline struct dentry *file_dentry(const struct file *file)
1265 return d_real(file->f_path.dentry, file_inode(file), 0, 0);
1268 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1270 return locks_lock_inode_wait(locks_inode(filp), fl);
1273 struct fasync_struct {
1277 struct fasync_struct *fa_next; /* singly linked list */
1278 struct file *fa_file;
1279 struct rcu_head fa_rcu;
1282 #define FASYNC_MAGIC 0x4601
1284 /* SMP safe fasync helpers: */
1285 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1286 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1287 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1288 extern struct fasync_struct *fasync_alloc(void);
1289 extern void fasync_free(struct fasync_struct *);
1291 /* can be called from interrupts */
1292 extern void kill_fasync(struct fasync_struct **, int, int);
1294 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1295 extern int f_setown(struct file *filp, unsigned long arg, int force);
1296 extern void f_delown(struct file *filp);
1297 extern pid_t f_getown(struct file *filp);
1298 extern int send_sigurg(struct fown_struct *fown);
1301 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1302 * represented in both.
1304 #define SB_RDONLY 1 /* Mount read-only */
1305 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1306 #define SB_NODEV 4 /* Disallow access to device special files */
1307 #define SB_NOEXEC 8 /* Disallow program execution */
1308 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1309 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1310 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1311 #define SB_NOATIME 1024 /* Do not update access times. */
1312 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1313 #define SB_SILENT 32768
1314 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1315 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1316 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1317 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1319 /* These sb flags are internal to the kernel */
1320 #define SB_SUBMOUNT (1<<26)
1321 #define SB_NOREMOTELOCK (1<<27)
1322 #define SB_NOSEC (1<<28)
1323 #define SB_BORN (1<<29)
1324 #define SB_ACTIVE (1<<30)
1325 #define SB_NOUSER (1<<31)
1331 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1332 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1333 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1334 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1335 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1338 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1339 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1340 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1341 #define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */
1343 /* sb->s_iflags to limit user namespace mounts */
1344 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1345 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1346 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1348 /* Possible states of 'frozen' field */
1350 SB_UNFROZEN = 0, /* FS is unfrozen */
1351 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1352 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1353 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1354 * internal threads if needed) */
1355 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1358 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1361 int frozen; /* Is sb frozen? */
1362 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1363 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1366 struct super_block {
1367 struct list_head s_list; /* Keep this first */
1368 dev_t s_dev; /* search index; _not_ kdev_t */
1369 unsigned char s_blocksize_bits;
1370 unsigned long s_blocksize;
1371 loff_t s_maxbytes; /* Max file size */
1372 struct file_system_type *s_type;
1373 const struct super_operations *s_op;
1374 const struct dquot_operations *dq_op;
1375 const struct quotactl_ops *s_qcop;
1376 const struct export_operations *s_export_op;
1377 unsigned long s_flags;
1378 unsigned long s_iflags; /* internal SB_I_* flags */
1379 unsigned long s_magic;
1380 struct dentry *s_root;
1381 struct rw_semaphore s_umount;
1384 #ifdef CONFIG_SECURITY
1387 const struct xattr_handler **s_xattr;
1388 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
1389 const struct fscrypt_operations *s_cop;
1391 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1392 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1393 struct block_device *s_bdev;
1394 struct backing_dev_info *s_bdi;
1395 struct mtd_info *s_mtd;
1396 struct hlist_node s_instances;
1397 unsigned int s_quota_types; /* Bitmask of supported quota types */
1398 struct quota_info s_dquot; /* Diskquota specific options */
1400 struct sb_writers s_writers;
1402 char s_id[32]; /* Informational name */
1403 uuid_t s_uuid; /* UUID */
1405 void *s_fs_info; /* Filesystem private info */
1406 unsigned int s_max_links;
1409 /* Granularity of c/m/atime in ns.
1410 Cannot be worse than a second */
1414 * The next field is for VFS *only*. No filesystems have any business
1415 * even looking at it. You had been warned.
1417 struct mutex s_vfs_rename_mutex; /* Kludge */
1420 * Filesystem subtype. If non-empty the filesystem type field
1421 * in /proc/mounts will be "type.subtype"
1425 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1428 * Saved pool identifier for cleancache (-1 means none)
1430 int cleancache_poolid;
1432 struct shrinker s_shrink; /* per-sb shrinker handle */
1434 /* Number of inodes with nlink == 0 but still referenced */
1435 atomic_long_t s_remove_count;
1437 /* Being remounted read-only */
1438 int s_readonly_remount;
1440 /* AIO completions deferred from interrupt context */
1441 struct workqueue_struct *s_dio_done_wq;
1442 struct hlist_head s_pins;
1445 * Owning user namespace and default context in which to
1446 * interpret filesystem uids, gids, quotas, device nodes,
1447 * xattrs and security labels.
1449 struct user_namespace *s_user_ns;
1452 * Keep the lru lists last in the structure so they always sit on their
1453 * own individual cachelines.
1455 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1456 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1457 struct rcu_head rcu;
1458 struct work_struct destroy_work;
1460 struct mutex s_sync_lock; /* sync serialisation lock */
1463 * Indicates how deep in a filesystem stack this SB is
1467 /* s_inode_list_lock protects s_inodes */
1468 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1469 struct list_head s_inodes; /* all inodes */
1471 spinlock_t s_inode_wblist_lock;
1472 struct list_head s_inodes_wb; /* writeback inodes */
1473 } __randomize_layout;
1475 /* Helper functions so that in most cases filesystems will
1476 * not need to deal directly with kuid_t and kgid_t and can
1477 * instead deal with the raw numeric values that are stored
1478 * in the filesystem.
1480 static inline uid_t i_uid_read(const struct inode *inode)
1482 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1485 static inline gid_t i_gid_read(const struct inode *inode)
1487 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1490 static inline void i_uid_write(struct inode *inode, uid_t uid)
1492 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1495 static inline void i_gid_write(struct inode *inode, gid_t gid)
1497 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1500 extern struct timespec64 timespec64_trunc(struct timespec64 t, unsigned gran);
1501 extern struct timespec64 current_time(struct inode *inode);
1504 * Snapshotting support.
1507 void __sb_end_write(struct super_block *sb, int level);
1508 int __sb_start_write(struct super_block *sb, int level, bool wait);
1510 #define __sb_writers_acquired(sb, lev) \
1511 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1512 #define __sb_writers_release(sb, lev) \
1513 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1516 * sb_end_write - drop write access to a superblock
1517 * @sb: the super we wrote to
1519 * Decrement number of writers to the filesystem. Wake up possible waiters
1520 * wanting to freeze the filesystem.
1522 static inline void sb_end_write(struct super_block *sb)
1524 __sb_end_write(sb, SB_FREEZE_WRITE);
1528 * sb_end_pagefault - drop write access to a superblock from a page fault
1529 * @sb: the super we wrote to
1531 * Decrement number of processes handling write page fault to the filesystem.
1532 * Wake up possible waiters wanting to freeze the filesystem.
1534 static inline void sb_end_pagefault(struct super_block *sb)
1536 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1540 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1541 * @sb: the super we wrote to
1543 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1544 * waiters wanting to freeze the filesystem.
1546 static inline void sb_end_intwrite(struct super_block *sb)
1548 __sb_end_write(sb, SB_FREEZE_FS);
1552 * sb_start_write - get write access to a superblock
1553 * @sb: the super we write to
1555 * When a process wants to write data or metadata to a file system (i.e. dirty
1556 * a page or an inode), it should embed the operation in a sb_start_write() -
1557 * sb_end_write() pair to get exclusion against file system freezing. This
1558 * function increments number of writers preventing freezing. If the file
1559 * system is already frozen, the function waits until the file system is
1562 * Since freeze protection behaves as a lock, users have to preserve
1563 * ordering of freeze protection and other filesystem locks. Generally,
1564 * freeze protection should be the outermost lock. In particular, we have:
1567 * -> i_mutex (write path, truncate, directory ops, ...)
1568 * -> s_umount (freeze_super, thaw_super)
1570 static inline void sb_start_write(struct super_block *sb)
1572 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1575 static inline int sb_start_write_trylock(struct super_block *sb)
1577 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1581 * sb_start_pagefault - get write access to a superblock from a page fault
1582 * @sb: the super we write to
1584 * When a process starts handling write page fault, it should embed the
1585 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1586 * exclusion against file system freezing. This is needed since the page fault
1587 * is going to dirty a page. This function increments number of running page
1588 * faults preventing freezing. If the file system is already frozen, the
1589 * function waits until the file system is thawed.
1591 * Since page fault freeze protection behaves as a lock, users have to preserve
1592 * ordering of freeze protection and other filesystem locks. It is advised to
1593 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1594 * handling code implies lock dependency:
1597 * -> sb_start_pagefault
1599 static inline void sb_start_pagefault(struct super_block *sb)
1601 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1605 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1606 * @sb: the super we write to
1608 * This is the third level of protection against filesystem freezing. It is
1609 * free for use by a filesystem. The only requirement is that it must rank
1610 * below sb_start_pagefault.
1612 * For example filesystem can call sb_start_intwrite() when starting a
1613 * transaction which somewhat eases handling of freezing for internal sources
1614 * of filesystem changes (internal fs threads, discarding preallocation on file
1617 static inline void sb_start_intwrite(struct super_block *sb)
1619 __sb_start_write(sb, SB_FREEZE_FS, true);
1622 static inline int sb_start_intwrite_trylock(struct super_block *sb)
1624 return __sb_start_write(sb, SB_FREEZE_FS, false);
1628 extern bool inode_owner_or_capable(const struct inode *inode);
1631 * VFS helper functions..
1633 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1634 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1635 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1636 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1637 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1638 extern int vfs_rmdir(struct inode *, struct dentry *);
1639 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1640 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1641 extern int vfs_whiteout(struct inode *, struct dentry *);
1643 extern struct dentry *vfs_tmpfile(struct dentry *dentry, umode_t mode,
1646 int vfs_mkobj(struct dentry *, umode_t,
1647 int (*f)(struct dentry *, umode_t, void *),
1651 * VFS file helper functions.
1653 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1655 extern bool may_open_dev(const struct path *path);
1657 * VFS FS_IOC_FIEMAP helper definitions.
1659 struct fiemap_extent_info {
1660 unsigned int fi_flags; /* Flags as passed from user */
1661 unsigned int fi_extents_mapped; /* Number of mapped extents */
1662 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1663 struct fiemap_extent __user *fi_extents_start; /* Start of
1664 fiemap_extent array */
1666 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1667 u64 phys, u64 len, u32 flags);
1668 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1673 * NOTE! These match bits 12..15 of stat.st_mode
1674 * (ie "(i_mode >> 12) & 15").
1676 #define DT_UNKNOWN 0
1687 * This is the "filldir" function type, used by readdir() to let
1688 * the kernel specify what kind of dirent layout it wants to have.
1689 * This allows the kernel to read directories into kernel space or
1690 * to have different dirent layouts depending on the binary type.
1693 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1696 struct dir_context {
1701 struct block_device_operations;
1703 /* These macros are for out of kernel modules to test that
1704 * the kernel supports the unlocked_ioctl and compat_ioctl
1705 * fields in struct file_operations. */
1706 #define HAVE_COMPAT_IOCTL 1
1707 #define HAVE_UNLOCKED_IOCTL 1
1710 * These flags let !MMU mmap() govern direct device mapping vs immediate
1711 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1713 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1714 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1715 * NOMMU_MAP_READ: Can be mapped for reading
1716 * NOMMU_MAP_WRITE: Can be mapped for writing
1717 * NOMMU_MAP_EXEC: Can be mapped for execution
1719 #define NOMMU_MAP_COPY 0x00000001
1720 #define NOMMU_MAP_DIRECT 0x00000008
1721 #define NOMMU_MAP_READ VM_MAYREAD
1722 #define NOMMU_MAP_WRITE VM_MAYWRITE
1723 #define NOMMU_MAP_EXEC VM_MAYEXEC
1725 #define NOMMU_VMFLAGS \
1726 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1731 struct file_operations {
1732 struct module *owner;
1733 loff_t (*llseek) (struct file *, loff_t, int);
1734 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1735 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1736 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1737 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1738 int (*iterate) (struct file *, struct dir_context *);
1739 int (*iterate_shared) (struct file *, struct dir_context *);
1740 __poll_t (*poll) (struct file *, struct poll_table_struct *);
1741 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1742 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1743 int (*mmap) (struct file *, struct vm_area_struct *);
1744 unsigned long mmap_supported_flags;
1745 int (*open) (struct inode *, struct file *);
1746 int (*flush) (struct file *, fl_owner_t id);
1747 int (*release) (struct inode *, struct file *);
1748 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1749 int (*fasync) (int, struct file *, int);
1750 int (*lock) (struct file *, int, struct file_lock *);
1751 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1752 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1753 int (*check_flags)(int);
1754 int (*flock) (struct file *, int, struct file_lock *);
1755 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1756 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1757 int (*setlease)(struct file *, long, struct file_lock **, void **);
1758 long (*fallocate)(struct file *file, int mode, loff_t offset,
1760 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1762 unsigned (*mmap_capabilities)(struct file *);
1764 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1765 loff_t, size_t, unsigned int);
1766 int (*clone_file_range)(struct file *, loff_t, struct file *, loff_t,
1768 ssize_t (*dedupe_file_range)(struct file *, u64, u64, struct file *,
1770 } __randomize_layout;
1772 struct inode_operations {
1773 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1774 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1775 int (*permission) (struct inode *, int);
1776 struct posix_acl * (*get_acl)(struct inode *, int);
1778 int (*readlink) (struct dentry *, char __user *,int);
1780 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1781 int (*link) (struct dentry *,struct inode *,struct dentry *);
1782 int (*unlink) (struct inode *,struct dentry *);
1783 int (*symlink) (struct inode *,struct dentry *,const char *);
1784 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1785 int (*rmdir) (struct inode *,struct dentry *);
1786 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1787 int (*rename) (struct inode *, struct dentry *,
1788 struct inode *, struct dentry *, unsigned int);
1789 int (*setattr) (struct dentry *, struct iattr *);
1790 int (*getattr) (const struct path *, struct kstat *, u32, unsigned int);
1791 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1792 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1794 int (*update_time)(struct inode *, struct timespec64 *, int);
1795 int (*atomic_open)(struct inode *, struct dentry *,
1796 struct file *, unsigned open_flag,
1797 umode_t create_mode);
1798 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1799 int (*set_acl)(struct inode *, struct posix_acl *, int);
1800 } ____cacheline_aligned;
1802 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
1803 struct iov_iter *iter)
1805 return file->f_op->read_iter(kio, iter);
1808 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
1809 struct iov_iter *iter)
1811 return file->f_op->write_iter(kio, iter);
1814 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
1816 return file->f_op->mmap(file, vma);
1819 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1820 unsigned long nr_segs, unsigned long fast_segs,
1821 struct iovec *fast_pointer,
1822 struct iovec **ret_pointer);
1824 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1825 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1826 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1827 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1828 unsigned long, loff_t *, rwf_t);
1829 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1830 loff_t, size_t, unsigned int);
1831 extern int vfs_clone_file_prep_inodes(struct inode *inode_in, loff_t pos_in,
1832 struct inode *inode_out, loff_t pos_out,
1833 u64 *len, bool is_dedupe);
1834 extern int vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1835 struct file *file_out, loff_t pos_out, u64 len);
1836 extern int vfs_dedupe_file_range_compare(struct inode *src, loff_t srcoff,
1837 struct inode *dest, loff_t destoff,
1838 loff_t len, bool *is_same);
1839 extern int vfs_dedupe_file_range(struct file *file,
1840 struct file_dedupe_range *same);
1842 struct super_operations {
1843 struct inode *(*alloc_inode)(struct super_block *sb);
1844 void (*destroy_inode)(struct inode *);
1846 void (*dirty_inode) (struct inode *, int flags);
1847 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1848 int (*drop_inode) (struct inode *);
1849 void (*evict_inode) (struct inode *);
1850 void (*put_super) (struct super_block *);
1851 int (*sync_fs)(struct super_block *sb, int wait);
1852 int (*freeze_super) (struct super_block *);
1853 int (*freeze_fs) (struct super_block *);
1854 int (*thaw_super) (struct super_block *);
1855 int (*unfreeze_fs) (struct super_block *);
1856 int (*statfs) (struct dentry *, struct kstatfs *);
1857 int (*remount_fs) (struct super_block *, int *, char *);
1858 void (*umount_begin) (struct super_block *);
1860 int (*show_options)(struct seq_file *, struct dentry *);
1861 int (*show_devname)(struct seq_file *, struct dentry *);
1862 int (*show_path)(struct seq_file *, struct dentry *);
1863 int (*show_stats)(struct seq_file *, struct dentry *);
1865 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1866 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1867 struct dquot **(*get_dquots)(struct inode *);
1869 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1870 long (*nr_cached_objects)(struct super_block *,
1871 struct shrink_control *);
1872 long (*free_cached_objects)(struct super_block *,
1873 struct shrink_control *);
1877 * Inode flags - they have no relation to superblock flags now
1879 #define S_SYNC 1 /* Writes are synced at once */
1880 #define S_NOATIME 2 /* Do not update access times */
1881 #define S_APPEND 4 /* Append-only file */
1882 #define S_IMMUTABLE 8 /* Immutable file */
1883 #define S_DEAD 16 /* removed, but still open directory */
1884 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1885 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1886 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1887 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1888 #define S_PRIVATE 512 /* Inode is fs-internal */
1889 #define S_IMA 1024 /* Inode has an associated IMA struct */
1890 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1891 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1892 #ifdef CONFIG_FS_DAX
1893 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1895 #define S_DAX 0 /* Make all the DAX code disappear */
1897 #define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */
1900 * Note that nosuid etc flags are inode-specific: setting some file-system
1901 * flags just means all the inodes inherit those flags by default. It might be
1902 * possible to override it selectively if you really wanted to with some
1903 * ioctl() that is not currently implemented.
1905 * Exception: SB_RDONLY is always applied to the entire file system.
1907 * Unfortunately, it is possible to change a filesystems flags with it mounted
1908 * with files in use. This means that all of the inodes will not have their
1909 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1910 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1912 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1914 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
1915 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
1916 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
1917 ((inode)->i_flags & S_SYNC))
1918 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
1919 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1920 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
1921 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
1922 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
1924 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1925 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1926 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1927 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
1929 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1930 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1931 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1932 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1933 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1934 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1935 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1936 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1937 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
1939 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1940 (inode)->i_rdev == WHITEOUT_DEV)
1942 static inline bool HAS_UNMAPPED_ID(struct inode *inode)
1944 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
1947 static inline enum rw_hint file_write_hint(struct file *file)
1949 if (file->f_write_hint != WRITE_LIFE_NOT_SET)
1950 return file->f_write_hint;
1952 return file_inode(file)->i_write_hint;
1955 static inline int iocb_flags(struct file *file);
1957 static inline u16 ki_hint_validate(enum rw_hint hint)
1959 typeof(((struct kiocb *)0)->ki_hint) max_hint = -1;
1961 if (hint <= max_hint)
1966 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
1968 *kiocb = (struct kiocb) {
1970 .ki_flags = iocb_flags(filp),
1971 .ki_hint = ki_hint_validate(file_write_hint(filp)),
1972 .ki_ioprio = IOPRIO_PRIO_VALUE(IOPRIO_CLASS_NONE, 0),
1977 * Inode state bits. Protected by inode->i_lock
1979 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1980 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1982 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1983 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1984 * various stages of removing an inode.
1986 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1988 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1989 * fdatasync(). i_atime is the usual cause.
1990 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1991 * these changes separately from I_DIRTY_SYNC so that we
1992 * don't have to write inode on fdatasync() when only
1993 * mtime has changed in it.
1994 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1995 * I_NEW Serves as both a mutex and completion notification.
1996 * New inodes set I_NEW. If two processes both create
1997 * the same inode, one of them will release its inode and
1998 * wait for I_NEW to be released before returning.
1999 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2000 * also cause waiting on I_NEW, without I_NEW actually
2001 * being set. find_inode() uses this to prevent returning
2002 * nearly-dead inodes.
2003 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2004 * is zero. I_FREEING must be set when I_WILL_FREE is
2006 * I_FREEING Set when inode is about to be freed but still has dirty
2007 * pages or buffers attached or the inode itself is still
2009 * I_CLEAR Added by clear_inode(). In this state the inode is
2010 * clean and can be destroyed. Inode keeps I_FREEING.
2012 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2013 * prohibited for many purposes. iget() must wait for
2014 * the inode to be completely released, then create it
2015 * anew. Other functions will just ignore such inodes,
2016 * if appropriate. I_NEW is used for waiting.
2018 * I_SYNC Writeback of inode is running. The bit is set during
2019 * data writeback, and cleared with a wakeup on the bit
2020 * address once it is done. The bit is also used to pin
2021 * the inode in memory for flusher thread.
2023 * I_REFERENCED Marks the inode as recently references on the LRU list.
2025 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2027 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2028 * synchronize competing switching instances and to tell
2029 * wb stat updates to grab the i_pages lock. See
2030 * inode_switch_wb_work_fn() for details.
2032 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2033 * and work dirs among overlayfs mounts.
2035 * I_CREATING New object's inode in the middle of setting up.
2037 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2039 #define I_DIRTY_SYNC (1 << 0)
2040 #define I_DIRTY_DATASYNC (1 << 1)
2041 #define I_DIRTY_PAGES (1 << 2)
2043 #define I_NEW (1 << __I_NEW)
2044 #define I_WILL_FREE (1 << 4)
2045 #define I_FREEING (1 << 5)
2046 #define I_CLEAR (1 << 6)
2048 #define I_SYNC (1 << __I_SYNC)
2049 #define I_REFERENCED (1 << 8)
2050 #define __I_DIO_WAKEUP 9
2051 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2052 #define I_LINKABLE (1 << 10)
2053 #define I_DIRTY_TIME (1 << 11)
2054 #define __I_DIRTY_TIME_EXPIRED 12
2055 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
2056 #define I_WB_SWITCH (1 << 13)
2057 #define I_OVL_INUSE (1 << 14)
2058 #define I_CREATING (1 << 15)
2060 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2061 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2062 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2064 extern void __mark_inode_dirty(struct inode *, int);
2065 static inline void mark_inode_dirty(struct inode *inode)
2067 __mark_inode_dirty(inode, I_DIRTY);
2070 static inline void mark_inode_dirty_sync(struct inode *inode)
2072 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2075 extern void inc_nlink(struct inode *inode);
2076 extern void drop_nlink(struct inode *inode);
2077 extern void clear_nlink(struct inode *inode);
2078 extern void set_nlink(struct inode *inode, unsigned int nlink);
2080 static inline void inode_inc_link_count(struct inode *inode)
2083 mark_inode_dirty(inode);
2086 static inline void inode_dec_link_count(struct inode *inode)
2089 mark_inode_dirty(inode);
2092 enum file_time_flags {
2099 extern void touch_atime(const struct path *);
2100 static inline void file_accessed(struct file *file)
2102 if (!(file->f_flags & O_NOATIME))
2103 touch_atime(&file->f_path);
2106 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2107 int sync_inode_metadata(struct inode *inode, int wait);
2109 struct file_system_type {
2112 #define FS_REQUIRES_DEV 1
2113 #define FS_BINARY_MOUNTDATA 2
2114 #define FS_HAS_SUBTYPE 4
2115 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2116 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2117 struct dentry *(*mount) (struct file_system_type *, int,
2118 const char *, void *);
2119 void (*kill_sb) (struct super_block *);
2120 struct module *owner;
2121 struct file_system_type * next;
2122 struct hlist_head fs_supers;
2124 struct lock_class_key s_lock_key;
2125 struct lock_class_key s_umount_key;
2126 struct lock_class_key s_vfs_rename_key;
2127 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2129 struct lock_class_key i_lock_key;
2130 struct lock_class_key i_mutex_key;
2131 struct lock_class_key i_mutex_dir_key;
2134 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2136 extern struct dentry *mount_ns(struct file_system_type *fs_type,
2137 int flags, void *data, void *ns, struct user_namespace *user_ns,
2138 int (*fill_super)(struct super_block *, void *, int));
2140 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2141 int flags, const char *dev_name, void *data,
2142 int (*fill_super)(struct super_block *, void *, int));
2144 static inline struct dentry *mount_bdev(struct file_system_type *fs_type,
2145 int flags, const char *dev_name, void *data,
2146 int (*fill_super)(struct super_block *, void *, int))
2148 return ERR_PTR(-ENODEV);
2151 extern struct dentry *mount_single(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_nodev(struct file_system_type *fs_type,
2155 int flags, void *data,
2156 int (*fill_super)(struct super_block *, void *, int));
2157 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2158 void generic_shutdown_super(struct super_block *sb);
2160 void kill_block_super(struct super_block *sb);
2162 static inline void kill_block_super(struct super_block *sb)
2167 void kill_anon_super(struct super_block *sb);
2168 void kill_litter_super(struct super_block *sb);
2169 void deactivate_super(struct super_block *sb);
2170 void deactivate_locked_super(struct super_block *sb);
2171 int set_anon_super(struct super_block *s, void *data);
2172 int get_anon_bdev(dev_t *);
2173 void free_anon_bdev(dev_t);
2174 struct super_block *sget_userns(struct file_system_type *type,
2175 int (*test)(struct super_block *,void *),
2176 int (*set)(struct super_block *,void *),
2177 int flags, struct user_namespace *user_ns,
2179 struct super_block *sget(struct file_system_type *type,
2180 int (*test)(struct super_block *,void *),
2181 int (*set)(struct super_block *,void *),
2182 int flags, void *data);
2183 extern struct dentry *mount_pseudo_xattr(struct file_system_type *, char *,
2184 const struct super_operations *ops,
2185 const struct xattr_handler **xattr,
2186 const struct dentry_operations *dops,
2189 static inline struct dentry *
2190 mount_pseudo(struct file_system_type *fs_type, char *name,
2191 const struct super_operations *ops,
2192 const struct dentry_operations *dops, unsigned long magic)
2194 return mount_pseudo_xattr(fs_type, name, ops, NULL, dops, magic);
2197 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2198 #define fops_get(fops) \
2199 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2200 #define fops_put(fops) \
2201 do { if (fops) module_put((fops)->owner); } while(0)
2203 * This one is to be used *ONLY* from ->open() instances.
2204 * fops must be non-NULL, pinned down *and* module dependencies
2205 * should be sufficient to pin the caller down as well.
2207 #define replace_fops(f, fops) \
2209 struct file *__file = (f); \
2210 fops_put(__file->f_op); \
2211 BUG_ON(!(__file->f_op = (fops))); \
2214 extern int register_filesystem(struct file_system_type *);
2215 extern int unregister_filesystem(struct file_system_type *);
2216 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2217 #define kern_mount(type) kern_mount_data(type, NULL)
2218 extern void kern_unmount(struct vfsmount *mnt);
2219 extern int may_umount_tree(struct vfsmount *);
2220 extern int may_umount(struct vfsmount *);
2221 extern long do_mount(const char *, const char __user *,
2222 const char *, unsigned long, void *);
2223 extern struct vfsmount *collect_mounts(const struct path *);
2224 extern void drop_collected_mounts(struct vfsmount *);
2225 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2227 extern int vfs_statfs(const struct path *, struct kstatfs *);
2228 extern int user_statfs(const char __user *, struct kstatfs *);
2229 extern int fd_statfs(int, struct kstatfs *);
2230 extern int freeze_super(struct super_block *super);
2231 extern int thaw_super(struct super_block *super);
2232 extern bool our_mnt(struct vfsmount *mnt);
2233 extern __printf(2, 3)
2234 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2235 extern int super_setup_bdi(struct super_block *sb);
2237 extern int current_umask(void);
2239 extern void ihold(struct inode * inode);
2240 extern void iput(struct inode *);
2241 extern int generic_update_time(struct inode *, struct timespec64 *, int);
2244 extern struct kobject *fs_kobj;
2246 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2248 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2249 extern int locks_mandatory_locked(struct file *);
2250 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2253 * Candidates for mandatory locking have the setgid bit set
2254 * but no group execute bit - an otherwise meaningless combination.
2257 static inline int __mandatory_lock(struct inode *ino)
2259 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2263 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2264 * otherwise these will be advisory locks
2267 static inline int mandatory_lock(struct inode *ino)
2269 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2272 static inline int locks_verify_locked(struct file *file)
2274 if (mandatory_lock(locks_inode(file)))
2275 return locks_mandatory_locked(file);
2279 static inline int locks_verify_truncate(struct inode *inode,
2283 if (!inode->i_flctx || !mandatory_lock(inode))
2286 if (size < inode->i_size) {
2287 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2290 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2295 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2297 static inline int locks_mandatory_locked(struct file *file)
2302 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2303 loff_t start, loff_t end, unsigned char type)
2308 static inline int __mandatory_lock(struct inode *inode)
2313 static inline int mandatory_lock(struct inode *inode)
2318 static inline int locks_verify_locked(struct file *file)
2323 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2329 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2332 #ifdef CONFIG_FILE_LOCKING
2333 static inline int break_lease(struct inode *inode, unsigned int mode)
2336 * Since this check is lockless, we must ensure that any refcounts
2337 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2338 * could end up racing with tasks trying to set a new lease on this
2342 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2343 return __break_lease(inode, mode, FL_LEASE);
2347 static inline int break_deleg(struct inode *inode, unsigned int mode)
2350 * Since this check is lockless, we must ensure that any refcounts
2351 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2352 * could end up racing with tasks trying to set a new lease on this
2356 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2357 return __break_lease(inode, mode, FL_DELEG);
2361 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2365 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2366 if (ret == -EWOULDBLOCK && delegated_inode) {
2367 *delegated_inode = inode;
2373 static inline int break_deleg_wait(struct inode **delegated_inode)
2377 ret = break_deleg(*delegated_inode, O_WRONLY);
2378 iput(*delegated_inode);
2379 *delegated_inode = NULL;
2383 static inline int break_layout(struct inode *inode, bool wait)
2386 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2387 return __break_lease(inode,
2388 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2393 #else /* !CONFIG_FILE_LOCKING */
2394 static inline int break_lease(struct inode *inode, unsigned int mode)
2399 static inline int break_deleg(struct inode *inode, unsigned int mode)
2404 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2409 static inline int break_deleg_wait(struct inode **delegated_inode)
2415 static inline int break_layout(struct inode *inode, bool wait)
2420 #endif /* CONFIG_FILE_LOCKING */
2425 const char *name; /* pointer to actual string */
2426 const __user char *uptr; /* original userland pointer */
2428 struct audit_names *aname;
2432 extern long vfs_truncate(const struct path *, loff_t);
2433 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2435 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2437 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2439 extern struct file *file_open_name(struct filename *, int, umode_t);
2440 extern struct file *filp_open(const char *, int, umode_t);
2441 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2442 const char *, int, umode_t);
2443 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2444 static inline struct file *file_clone_open(struct file *file)
2446 return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2448 extern int filp_close(struct file *, fl_owner_t id);
2450 extern struct filename *getname_flags(const char __user *, int, int *);
2451 extern struct filename *getname(const char __user *);
2452 extern struct filename *getname_kernel(const char *);
2453 extern void putname(struct filename *name);
2455 extern int finish_open(struct file *file, struct dentry *dentry,
2456 int (*open)(struct inode *, struct file *));
2457 extern int finish_no_open(struct file *file, struct dentry *dentry);
2461 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2464 extern void __init vfs_caches_init_early(void);
2465 extern void __init vfs_caches_init(void);
2467 extern struct kmem_cache *names_cachep;
2469 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2470 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2473 extern int register_blkdev(unsigned int, const char *);
2474 extern void unregister_blkdev(unsigned int, const char *);
2475 extern void bdev_unhash_inode(dev_t dev);
2476 extern struct block_device *bdget(dev_t);
2477 extern struct block_device *bdgrab(struct block_device *bdev);
2478 extern void bd_set_size(struct block_device *, loff_t size);
2479 extern void bd_forget(struct inode *inode);
2480 extern void bdput(struct block_device *);
2481 extern void invalidate_bdev(struct block_device *);
2482 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2483 extern int sync_blockdev(struct block_device *bdev);
2484 extern void kill_bdev(struct block_device *);
2485 extern struct super_block *freeze_bdev(struct block_device *);
2486 extern void emergency_thaw_all(void);
2487 extern void emergency_thaw_bdev(struct super_block *sb);
2488 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2489 extern int fsync_bdev(struct block_device *);
2491 extern struct super_block *blockdev_superblock;
2493 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2495 return sb == blockdev_superblock;
2498 static inline void bd_forget(struct inode *inode) {}
2499 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2500 static inline void kill_bdev(struct block_device *bdev) {}
2501 static inline void invalidate_bdev(struct block_device *bdev) {}
2503 static inline struct super_block *freeze_bdev(struct block_device *sb)
2508 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2513 static inline int emergency_thaw_bdev(struct super_block *sb)
2518 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2522 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2527 extern int sync_filesystem(struct super_block *);
2528 extern const struct file_operations def_blk_fops;
2529 extern const struct file_operations def_chr_fops;
2531 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2532 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2533 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2534 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2535 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2537 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2539 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2540 extern int __blkdev_reread_part(struct block_device *bdev);
2541 extern int blkdev_reread_part(struct block_device *bdev);
2544 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2545 extern void bd_unlink_disk_holder(struct block_device *bdev,
2546 struct gendisk *disk);
2548 static inline int bd_link_disk_holder(struct block_device *bdev,
2549 struct gendisk *disk)
2553 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2554 struct gendisk *disk)
2561 #define CHRDEV_MAJOR_MAX 512
2562 /* Marks the bottom of the first segment of free char majors */
2563 #define CHRDEV_MAJOR_DYN_END 234
2564 /* Marks the top and bottom of the second segment of free char majors */
2565 #define CHRDEV_MAJOR_DYN_EXT_START 511
2566 #define CHRDEV_MAJOR_DYN_EXT_END 384
2568 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2569 extern int register_chrdev_region(dev_t, unsigned, const char *);
2570 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2571 unsigned int count, const char *name,
2572 const struct file_operations *fops);
2573 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2574 unsigned int count, const char *name);
2575 extern void unregister_chrdev_region(dev_t, unsigned);
2576 extern void chrdev_show(struct seq_file *,off_t);
2578 static inline int register_chrdev(unsigned int major, const char *name,
2579 const struct file_operations *fops)
2581 return __register_chrdev(major, 0, 256, name, fops);
2584 static inline void unregister_chrdev(unsigned int major, const char *name)
2586 __unregister_chrdev(major, 0, 256, name);
2589 /* fs/block_dev.c */
2590 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2591 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2594 #define BLKDEV_MAJOR_MAX 512
2595 extern const char *__bdevname(dev_t, char *buffer);
2596 extern const char *bdevname(struct block_device *bdev, char *buffer);
2597 extern struct block_device *lookup_bdev(const char *);
2598 extern void blkdev_show(struct seq_file *,off_t);
2601 #define BLKDEV_MAJOR_MAX 0
2604 extern void init_special_inode(struct inode *, umode_t, dev_t);
2606 /* Invalid inode operations -- fs/bad_inode.c */
2607 extern void make_bad_inode(struct inode *);
2608 extern bool is_bad_inode(struct inode *);
2611 extern void check_disk_size_change(struct gendisk *disk,
2612 struct block_device *bdev, bool verbose);
2613 extern int revalidate_disk(struct gendisk *);
2614 extern int check_disk_change(struct block_device *);
2615 extern int __invalidate_device(struct block_device *, bool);
2616 extern int invalidate_partition(struct gendisk *, int);
2618 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2619 pgoff_t start, pgoff_t end);
2621 static inline void invalidate_remote_inode(struct inode *inode)
2623 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2624 S_ISLNK(inode->i_mode))
2625 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2627 extern int invalidate_inode_pages2(struct address_space *mapping);
2628 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2629 pgoff_t start, pgoff_t end);
2630 extern int write_inode_now(struct inode *, int);
2631 extern int filemap_fdatawrite(struct address_space *);
2632 extern int filemap_flush(struct address_space *);
2633 extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2634 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2637 static inline int filemap_fdatawait(struct address_space *mapping)
2639 return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2642 extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2644 extern int filemap_write_and_wait(struct address_space *mapping);
2645 extern int filemap_write_and_wait_range(struct address_space *mapping,
2646 loff_t lstart, loff_t lend);
2647 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2648 loff_t start, loff_t end, int sync_mode);
2649 extern int filemap_fdatawrite_range(struct address_space *mapping,
2650 loff_t start, loff_t end);
2651 extern int filemap_check_errors(struct address_space *mapping);
2652 extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2654 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2656 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2657 extern int __must_check file_write_and_wait_range(struct file *file,
2658 loff_t start, loff_t end);
2660 static inline int file_write_and_wait(struct file *file)
2662 return file_write_and_wait_range(file, 0, LLONG_MAX);
2666 * filemap_set_wb_err - set a writeback error on an address_space
2667 * @mapping: mapping in which to set writeback error
2668 * @err: error to be set in mapping
2670 * When writeback fails in some way, we must record that error so that
2671 * userspace can be informed when fsync and the like are called. We endeavor
2672 * to report errors on any file that was open at the time of the error. Some
2673 * internal callers also need to know when writeback errors have occurred.
2675 * When a writeback error occurs, most filesystems will want to call
2676 * filemap_set_wb_err to record the error in the mapping so that it will be
2677 * automatically reported whenever fsync is called on the file.
2679 static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2681 /* Fastpath for common case of no error */
2683 __filemap_set_wb_err(mapping, err);
2687 * filemap_check_wb_error - has an error occurred since the mark was sampled?
2688 * @mapping: mapping to check for writeback errors
2689 * @since: previously-sampled errseq_t
2691 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2692 * the given value was sampled.
2694 * If it has then report the latest error set, otherwise return 0.
2696 static inline int filemap_check_wb_err(struct address_space *mapping,
2699 return errseq_check(&mapping->wb_err, since);
2703 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2704 * @mapping: mapping to be sampled
2706 * Writeback errors are always reported relative to a particular sample point
2707 * in the past. This function provides those sample points.
2709 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2711 return errseq_sample(&mapping->wb_err);
2714 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2716 extern int vfs_fsync(struct file *file, int datasync);
2719 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2720 * to already be updated for the write, and will return either the amount
2721 * of bytes passed in, or an error if syncing the file failed.
2723 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2725 if (iocb->ki_flags & IOCB_DSYNC) {
2726 int ret = vfs_fsync_range(iocb->ki_filp,
2727 iocb->ki_pos - count, iocb->ki_pos - 1,
2728 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2736 extern void emergency_sync(void);
2737 extern void emergency_remount(void);
2739 extern sector_t bmap(struct inode *, sector_t);
2741 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2742 extern int inode_permission(struct inode *, int);
2743 extern int generic_permission(struct inode *, int);
2744 extern int __check_sticky(struct inode *dir, struct inode *inode);
2746 static inline bool execute_ok(struct inode *inode)
2748 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2751 static inline void file_start_write(struct file *file)
2753 if (!S_ISREG(file_inode(file)->i_mode))
2755 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2758 static inline bool file_start_write_trylock(struct file *file)
2760 if (!S_ISREG(file_inode(file)->i_mode))
2762 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2765 static inline void file_end_write(struct file *file)
2767 if (!S_ISREG(file_inode(file)->i_mode))
2769 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2772 static inline int do_clone_file_range(struct file *file_in, loff_t pos_in,
2773 struct file *file_out, loff_t pos_out,
2778 file_start_write(file_out);
2779 ret = vfs_clone_file_range(file_in, pos_in, file_out, pos_out, len);
2780 file_end_write(file_out);
2786 * get_write_access() gets write permission for a file.
2787 * put_write_access() releases this write permission.
2788 * This is used for regular files.
2789 * We cannot support write (and maybe mmap read-write shared) accesses and
2790 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2791 * can have the following values:
2792 * 0: no writers, no VM_DENYWRITE mappings
2793 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2794 * > 0: (i_writecount) users are writing to the file.
2796 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2797 * except for the cases where we don't hold i_writecount yet. Then we need to
2798 * use {get,deny}_write_access() - these functions check the sign and refuse
2799 * to do the change if sign is wrong.
2801 static inline int get_write_access(struct inode *inode)
2803 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2805 static inline int deny_write_access(struct file *file)
2807 struct inode *inode = file_inode(file);
2808 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2810 static inline void put_write_access(struct inode * inode)
2812 atomic_dec(&inode->i_writecount);
2814 static inline void allow_write_access(struct file *file)
2817 atomic_inc(&file_inode(file)->i_writecount);
2819 static inline bool inode_is_open_for_write(const struct inode *inode)
2821 return atomic_read(&inode->i_writecount) > 0;
2825 static inline void i_readcount_dec(struct inode *inode)
2827 BUG_ON(!atomic_read(&inode->i_readcount));
2828 atomic_dec(&inode->i_readcount);
2830 static inline void i_readcount_inc(struct inode *inode)
2832 atomic_inc(&inode->i_readcount);
2835 static inline void i_readcount_dec(struct inode *inode)
2839 static inline void i_readcount_inc(struct inode *inode)
2844 extern int do_pipe_flags(int *, int);
2846 #define __kernel_read_file_id(id) \
2847 id(UNKNOWN, unknown) \
2848 id(FIRMWARE, firmware) \
2849 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2850 id(MODULE, kernel-module) \
2851 id(KEXEC_IMAGE, kexec-image) \
2852 id(KEXEC_INITRAMFS, kexec-initramfs) \
2853 id(POLICY, security-policy) \
2854 id(X509_CERTIFICATE, x509-certificate) \
2857 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2858 #define __fid_stringify(dummy, str) #str,
2860 enum kernel_read_file_id {
2861 __kernel_read_file_id(__fid_enumify)
2864 static const char * const kernel_read_file_str[] = {
2865 __kernel_read_file_id(__fid_stringify)
2868 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
2870 if ((unsigned)id >= READING_MAX_ID)
2871 return kernel_read_file_str[READING_UNKNOWN];
2873 return kernel_read_file_str[id];
2876 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2877 enum kernel_read_file_id);
2878 extern int kernel_read_file_from_path(const char *, void **, loff_t *, loff_t,
2879 enum kernel_read_file_id);
2880 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2881 enum kernel_read_file_id);
2882 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
2883 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
2884 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
2885 extern struct file * open_exec(const char *);
2887 /* fs/dcache.c -- generic fs support functions */
2888 extern bool is_subdir(struct dentry *, struct dentry *);
2889 extern bool path_is_under(const struct path *, const struct path *);
2891 extern char *file_path(struct file *, char *, int);
2893 #include <linux/err.h>
2895 /* needed for stackable file system support */
2896 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2898 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2900 extern int inode_init_always(struct super_block *, struct inode *);
2901 extern void inode_init_once(struct inode *);
2902 extern void address_space_init_once(struct address_space *mapping);
2903 extern struct inode * igrab(struct inode *);
2904 extern ino_t iunique(struct super_block *, ino_t);
2905 extern int inode_needs_sync(struct inode *inode);
2906 extern int generic_delete_inode(struct inode *inode);
2907 static inline int generic_drop_inode(struct inode *inode)
2909 return !inode->i_nlink || inode_unhashed(inode);
2912 extern struct inode *ilookup5_nowait(struct super_block *sb,
2913 unsigned long hashval, int (*test)(struct inode *, void *),
2915 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2916 int (*test)(struct inode *, void *), void *data);
2917 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2919 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
2920 int (*test)(struct inode *, void *),
2921 int (*set)(struct inode *, void *),
2923 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2924 extern struct inode * iget_locked(struct super_block *, unsigned long);
2925 extern struct inode *find_inode_nowait(struct super_block *,
2927 int (*match)(struct inode *,
2928 unsigned long, void *),
2930 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2931 extern int insert_inode_locked(struct inode *);
2932 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2933 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2935 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2937 extern void unlock_new_inode(struct inode *);
2938 extern void discard_new_inode(struct inode *);
2939 extern unsigned int get_next_ino(void);
2940 extern void evict_inodes(struct super_block *sb);
2942 extern void __iget(struct inode * inode);
2943 extern void iget_failed(struct inode *);
2944 extern void clear_inode(struct inode *);
2945 extern void __destroy_inode(struct inode *);
2946 extern struct inode *new_inode_pseudo(struct super_block *sb);
2947 extern struct inode *new_inode(struct super_block *sb);
2948 extern void free_inode_nonrcu(struct inode *inode);
2949 extern int should_remove_suid(struct dentry *);
2950 extern int file_remove_privs(struct file *);
2952 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2953 static inline void insert_inode_hash(struct inode *inode)
2955 __insert_inode_hash(inode, inode->i_ino);
2958 extern void __remove_inode_hash(struct inode *);
2959 static inline void remove_inode_hash(struct inode *inode)
2961 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2962 __remove_inode_hash(inode);
2965 extern void inode_sb_list_add(struct inode *inode);
2968 extern int bdev_read_only(struct block_device *);
2970 extern int set_blocksize(struct block_device *, int);
2971 extern int sb_set_blocksize(struct super_block *, int);
2972 extern int sb_min_blocksize(struct super_block *, int);
2974 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2975 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2976 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2977 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2978 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2979 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2980 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
2981 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2983 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
2985 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
2988 /* fs/block_dev.c */
2989 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2990 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2991 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2993 extern void block_sync_page(struct page *page);
2996 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2997 struct pipe_inode_info *, size_t, unsigned int);
2998 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2999 struct file *, loff_t *, size_t, unsigned int);
3000 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
3001 struct file *out, loff_t *, size_t len, unsigned int flags);
3002 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
3003 loff_t *opos, size_t len, unsigned int flags);
3007 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3008 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3009 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
3010 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3011 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3012 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3013 int whence, loff_t maxsize, loff_t eof);
3014 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3015 int whence, loff_t size);
3016 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3017 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3018 extern int generic_file_open(struct inode * inode, struct file * filp);
3019 extern int nonseekable_open(struct inode * inode, struct file * filp);
3022 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3023 loff_t file_offset);
3026 /* need locking between buffered and direct access */
3029 /* filesystem does not support filling holes */
3030 DIO_SKIP_HOLES = 0x02,
3033 void dio_end_io(struct bio *bio);
3034 void dio_warn_stale_pagecache(struct file *filp);
3036 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3037 struct block_device *bdev, struct iov_iter *iter,
3038 get_block_t get_block,
3039 dio_iodone_t end_io, dio_submit_t submit_io,
3042 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3043 struct inode *inode,
3044 struct iov_iter *iter,
3045 get_block_t get_block)
3047 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3048 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3052 void inode_dio_wait(struct inode *inode);
3055 * inode_dio_begin - signal start of a direct I/O requests
3056 * @inode: inode the direct I/O happens on
3058 * This is called once we've finished processing a direct I/O request,
3059 * and is used to wake up callers waiting for direct I/O to be quiesced.
3061 static inline void inode_dio_begin(struct inode *inode)
3063 atomic_inc(&inode->i_dio_count);
3067 * inode_dio_end - signal finish of a direct I/O requests
3068 * @inode: inode the direct I/O happens on
3070 * This is called once we've finished processing a direct I/O request,
3071 * and is used to wake up callers waiting for direct I/O to be quiesced.
3073 static inline void inode_dio_end(struct inode *inode)
3075 if (atomic_dec_and_test(&inode->i_dio_count))
3076 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3079 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3082 extern const struct file_operations generic_ro_fops;
3084 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3086 extern int readlink_copy(char __user *, int, const char *);
3087 extern int page_readlink(struct dentry *, char __user *, int);
3088 extern const char *page_get_link(struct dentry *, struct inode *,
3089 struct delayed_call *);
3090 extern void page_put_link(void *);
3091 extern int __page_symlink(struct inode *inode, const char *symname, int len,
3093 extern int page_symlink(struct inode *inode, const char *symname, int len);
3094 extern const struct inode_operations page_symlink_inode_operations;
3095 extern void kfree_link(void *);
3096 extern void generic_fillattr(struct inode *, struct kstat *);
3097 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3098 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3099 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3100 void inode_add_bytes(struct inode *inode, loff_t bytes);
3101 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3102 void inode_sub_bytes(struct inode *inode, loff_t bytes);
3103 static inline loff_t __inode_get_bytes(struct inode *inode)
3105 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3107 loff_t inode_get_bytes(struct inode *inode);
3108 void inode_set_bytes(struct inode *inode, loff_t bytes);
3109 const char *simple_get_link(struct dentry *, struct inode *,
3110 struct delayed_call *);
3111 extern const struct inode_operations simple_symlink_inode_operations;
3113 extern int iterate_dir(struct file *, struct dir_context *);
3115 extern int vfs_statx(int, const char __user *, int, struct kstat *, u32);
3116 extern int vfs_statx_fd(unsigned int, struct kstat *, u32, unsigned int);
3118 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3120 return vfs_statx(AT_FDCWD, filename, AT_NO_AUTOMOUNT,
3121 stat, STATX_BASIC_STATS);
3123 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3125 return vfs_statx(AT_FDCWD, name, AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT,
3126 stat, STATX_BASIC_STATS);
3128 static inline int vfs_fstatat(int dfd, const char __user *filename,
3129 struct kstat *stat, int flags)
3131 return vfs_statx(dfd, filename, flags | AT_NO_AUTOMOUNT,
3132 stat, STATX_BASIC_STATS);
3134 static inline int vfs_fstat(int fd, struct kstat *stat)
3136 return vfs_statx_fd(fd, stat, STATX_BASIC_STATS, 0);
3140 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3141 extern int vfs_readlink(struct dentry *, char __user *, int);
3143 extern int __generic_block_fiemap(struct inode *inode,
3144 struct fiemap_extent_info *fieinfo,
3145 loff_t start, loff_t len,
3146 get_block_t *get_block);
3147 extern int generic_block_fiemap(struct inode *inode,
3148 struct fiemap_extent_info *fieinfo, u64 start,
3149 u64 len, get_block_t *get_block);
3151 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3152 extern void put_filesystem(struct file_system_type *fs);
3153 extern struct file_system_type *get_fs_type(const char *name);
3154 extern struct super_block *get_super(struct block_device *);
3155 extern struct super_block *get_super_thawed(struct block_device *);
3156 extern struct super_block *get_super_exclusive_thawed(struct block_device *bdev);
3157 extern struct super_block *get_active_super(struct block_device *bdev);
3158 extern void drop_super(struct super_block *sb);
3159 extern void drop_super_exclusive(struct super_block *sb);
3160 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3161 extern void iterate_supers_type(struct file_system_type *,
3162 void (*)(struct super_block *, void *), void *);
3164 extern int dcache_dir_open(struct inode *, struct file *);
3165 extern int dcache_dir_close(struct inode *, struct file *);
3166 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3167 extern int dcache_readdir(struct file *, struct dir_context *);
3168 extern int simple_setattr(struct dentry *, struct iattr *);
3169 extern int simple_getattr(const struct path *, struct kstat *, u32, unsigned int);
3170 extern int simple_statfs(struct dentry *, struct kstatfs *);
3171 extern int simple_open(struct inode *inode, struct file *file);
3172 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3173 extern int simple_unlink(struct inode *, struct dentry *);
3174 extern int simple_rmdir(struct inode *, struct dentry *);
3175 extern int simple_rename(struct inode *, struct dentry *,
3176 struct inode *, struct dentry *, unsigned int);
3177 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3178 extern int noop_set_page_dirty(struct page *page);
3179 extern void noop_invalidatepage(struct page *page, unsigned int offset,
3180 unsigned int length);
3181 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3182 extern int simple_empty(struct dentry *);
3183 extern int simple_readpage(struct file *file, struct page *page);
3184 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3185 loff_t pos, unsigned len, unsigned flags,
3186 struct page **pagep, void **fsdata);
3187 extern int simple_write_end(struct file *file, struct address_space *mapping,
3188 loff_t pos, unsigned len, unsigned copied,
3189 struct page *page, void *fsdata);
3190 extern int always_delete_dentry(const struct dentry *);
3191 extern struct inode *alloc_anon_inode(struct super_block *);
3192 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3193 extern const struct dentry_operations simple_dentry_operations;
3195 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3196 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3197 extern const struct file_operations simple_dir_operations;
3198 extern const struct inode_operations simple_dir_inode_operations;
3199 extern void make_empty_dir_inode(struct inode *inode);
3200 extern bool is_empty_dir_inode(struct inode *inode);
3201 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3202 struct dentry *d_alloc_name(struct dentry *, const char *);
3203 extern int simple_fill_super(struct super_block *, unsigned long,
3204 const struct tree_descr *);
3205 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3206 extern void simple_release_fs(struct vfsmount **mount, int *count);
3208 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3209 loff_t *ppos, const void *from, size_t available);
3210 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3211 const void __user *from, size_t count);
3213 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3214 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3216 extern int generic_check_addressable(unsigned, u64);
3218 #ifdef CONFIG_MIGRATION
3219 extern int buffer_migrate_page(struct address_space *,
3220 struct page *, struct page *,
3223 #define buffer_migrate_page NULL
3226 extern int setattr_prepare(struct dentry *, struct iattr *);
3227 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3228 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
3230 extern int file_update_time(struct file *file);
3232 static inline bool io_is_direct(struct file *filp)
3234 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
3237 static inline bool vma_is_dax(struct vm_area_struct *vma)
3239 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3242 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3244 struct inode *inode;
3248 if (!vma_is_dax(vma))
3250 inode = file_inode(vma->vm_file);
3251 if (S_ISCHR(inode->i_mode))
3252 return false; /* device-dax */
3256 static inline int iocb_flags(struct file *file)
3259 if (file->f_flags & O_APPEND)
3261 if (io_is_direct(file))
3263 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3265 if (file->f_flags & __O_SYNC)
3270 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3272 if (unlikely(flags & ~RWF_SUPPORTED))
3275 if (flags & RWF_NOWAIT) {
3276 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3278 ki->ki_flags |= IOCB_NOWAIT;
3280 if (flags & RWF_HIPRI)
3281 ki->ki_flags |= IOCB_HIPRI;
3282 if (flags & RWF_DSYNC)
3283 ki->ki_flags |= IOCB_DSYNC;
3284 if (flags & RWF_SYNC)
3285 ki->ki_flags |= (IOCB_DSYNC | IOCB_SYNC);
3286 if (flags & RWF_APPEND)
3287 ki->ki_flags |= IOCB_APPEND;
3291 static inline ino_t parent_ino(struct dentry *dentry)
3296 * Don't strictly need d_lock here? If the parent ino could change
3297 * then surely we'd have a deeper race in the caller?
3299 spin_lock(&dentry->d_lock);
3300 res = dentry->d_parent->d_inode->i_ino;
3301 spin_unlock(&dentry->d_lock);
3305 /* Transaction based IO helpers */
3308 * An argresp is stored in an allocated page and holds the
3309 * size of the argument or response, along with its content
3311 struct simple_transaction_argresp {
3316 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3318 char *simple_transaction_get(struct file *file, const char __user *buf,
3320 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3321 size_t size, loff_t *pos);
3322 int simple_transaction_release(struct inode *inode, struct file *file);
3324 void simple_transaction_set(struct file *file, size_t n);
3327 * simple attribute files
3329 * These attributes behave similar to those in sysfs:
3331 * Writing to an attribute immediately sets a value, an open file can be
3332 * written to multiple times.
3334 * Reading from an attribute creates a buffer from the value that might get
3335 * read with multiple read calls. When the attribute has been read
3336 * completely, no further read calls are possible until the file is opened
3339 * All attributes contain a text representation of a numeric value
3340 * that are accessed with the get() and set() functions.
3342 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3343 static int __fops ## _open(struct inode *inode, struct file *file) \
3345 __simple_attr_check_format(__fmt, 0ull); \
3346 return simple_attr_open(inode, file, __get, __set, __fmt); \
3348 static const struct file_operations __fops = { \
3349 .owner = THIS_MODULE, \
3350 .open = __fops ## _open, \
3351 .release = simple_attr_release, \
3352 .read = simple_attr_read, \
3353 .write = simple_attr_write, \
3354 .llseek = generic_file_llseek, \
3357 static inline __printf(1, 2)
3358 void __simple_attr_check_format(const char *fmt, ...)
3360 /* don't do anything, just let the compiler check the arguments; */
3363 int simple_attr_open(struct inode *inode, struct file *file,
3364 int (*get)(void *, u64 *), int (*set)(void *, u64),
3366 int simple_attr_release(struct inode *inode, struct file *file);
3367 ssize_t simple_attr_read(struct file *file, char __user *buf,
3368 size_t len, loff_t *ppos);
3369 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3370 size_t len, loff_t *ppos);
3373 int proc_nr_files(struct ctl_table *table, int write,
3374 void __user *buffer, size_t *lenp, loff_t *ppos);
3375 int proc_nr_dentry(struct ctl_table *table, int write,
3376 void __user *buffer, size_t *lenp, loff_t *ppos);
3377 int proc_nr_inodes(struct ctl_table *table, int write,
3378 void __user *buffer, size_t *lenp, loff_t *ppos);
3379 int __init get_filesystem_list(char *buf);
3381 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3382 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3384 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3385 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3386 (flag & __FMODE_NONOTIFY)))
3388 static inline bool is_sxid(umode_t mode)
3390 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3393 static inline int check_sticky(struct inode *dir, struct inode *inode)
3395 if (!(dir->i_mode & S_ISVTX))
3398 return __check_sticky(dir, inode);
3401 static inline void inode_has_no_xattr(struct inode *inode)
3403 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3404 inode->i_flags |= S_NOSEC;
3407 static inline bool is_root_inode(struct inode *inode)
3409 return inode == inode->i_sb->s_root->d_inode;
3412 static inline bool dir_emit(struct dir_context *ctx,
3413 const char *name, int namelen,
3414 u64 ino, unsigned type)
3416 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3418 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3420 return ctx->actor(ctx, ".", 1, ctx->pos,
3421 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3423 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3425 return ctx->actor(ctx, "..", 2, ctx->pos,
3426 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3428 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3430 if (ctx->pos == 0) {
3431 if (!dir_emit_dot(file, ctx))
3435 if (ctx->pos == 1) {
3436 if (!dir_emit_dotdot(file, ctx))
3442 static inline bool dir_relax(struct inode *inode)
3444 inode_unlock(inode);
3446 return !IS_DEADDIR(inode);
3449 static inline bool dir_relax_shared(struct inode *inode)
3451 inode_unlock_shared(inode);
3452 inode_lock_shared(inode);
3453 return !IS_DEADDIR(inode);
3456 extern bool path_noexec(const struct path *path);
3457 extern void inode_nohighmem(struct inode *inode);
3459 #endif /* _LINUX_FS_H */