1 /* SPDX-License-Identifier: GPL-2.0 */
5 #include <linux/linkage.h>
6 #include <linux/wait_bit.h>
7 #include <linux/kdev_t.h>
8 #include <linux/dcache.h>
9 #include <linux/path.h>
10 #include <linux/stat.h>
11 #include <linux/cache.h>
12 #include <linux/list.h>
13 #include <linux/list_lru.h>
14 #include <linux/llist.h>
15 #include <linux/radix-tree.h>
16 #include <linux/xarray.h>
17 #include <linux/rbtree.h>
18 #include <linux/init.h>
19 #include <linux/pid.h>
20 #include <linux/bug.h>
21 #include <linux/mutex.h>
22 #include <linux/rwsem.h>
23 #include <linux/mm_types.h>
24 #include <linux/capability.h>
25 #include <linux/semaphore.h>
26 #include <linux/fcntl.h>
27 #include <linux/fiemap.h>
28 #include <linux/rculist_bl.h>
29 #include <linux/atomic.h>
30 #include <linux/shrinker.h>
31 #include <linux/migrate_mode.h>
32 #include <linux/uidgid.h>
33 #include <linux/lockdep.h>
34 #include <linux/percpu-rwsem.h>
35 #include <linux/workqueue.h>
36 #include <linux/delayed_call.h>
37 #include <linux/uuid.h>
38 #include <linux/errseq.h>
39 #include <linux/ioprio.h>
41 #include <asm/byteorder.h>
42 #include <uapi/linux/fs.h>
44 struct backing_dev_info;
47 struct export_operations;
52 struct pipe_inode_info;
53 struct poll_table_struct;
55 struct vm_area_struct;
58 struct swap_info_struct;
60 struct workqueue_struct;
63 struct fscrypt_operations;
65 extern void __init inode_init(void);
66 extern void __init inode_init_early(void);
67 extern void __init files_init(void);
68 extern void __init files_maxfiles_init(void);
70 extern struct files_stat_struct files_stat;
71 extern unsigned long get_max_files(void);
72 extern unsigned int sysctl_nr_open;
73 extern struct inodes_stat_t inodes_stat;
74 extern int leases_enable, lease_break_time;
75 extern int sysctl_protected_symlinks;
76 extern int sysctl_protected_hardlinks;
78 typedef __kernel_rwf_t rwf_t;
81 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
82 struct buffer_head *bh_result, int create);
83 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
84 ssize_t bytes, void *private);
86 #define MAY_EXEC 0x00000001
87 #define MAY_WRITE 0x00000002
88 #define MAY_READ 0x00000004
89 #define MAY_APPEND 0x00000008
90 #define MAY_ACCESS 0x00000010
91 #define MAY_OPEN 0x00000020
92 #define MAY_CHDIR 0x00000040
93 /* called from RCU mode, don't block */
94 #define MAY_NOT_BLOCK 0x00000080
97 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
98 * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
101 /* file is open for reading */
102 #define FMODE_READ ((__force fmode_t)0x1)
103 /* file is open for writing */
104 #define FMODE_WRITE ((__force fmode_t)0x2)
105 /* file is seekable */
106 #define FMODE_LSEEK ((__force fmode_t)0x4)
107 /* file can be accessed using pread */
108 #define FMODE_PREAD ((__force fmode_t)0x8)
109 /* file can be accessed using pwrite */
110 #define FMODE_PWRITE ((__force fmode_t)0x10)
111 /* File is opened for execution with sys_execve / sys_uselib */
112 #define FMODE_EXEC ((__force fmode_t)0x20)
113 /* File is opened with O_NDELAY (only set for block devices) */
114 #define FMODE_NDELAY ((__force fmode_t)0x40)
115 /* File is opened with O_EXCL (only set for block devices) */
116 #define FMODE_EXCL ((__force fmode_t)0x80)
117 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
118 (specialy hack for floppy.c) */
119 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
120 /* 32bit hashes as llseek() offset (for directories) */
121 #define FMODE_32BITHASH ((__force fmode_t)0x200)
122 /* 64bit hashes as llseek() offset (for directories) */
123 #define FMODE_64BITHASH ((__force fmode_t)0x400)
126 * Don't update ctime and mtime.
128 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
129 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
131 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
133 /* Expect random access pattern */
134 #define FMODE_RANDOM ((__force fmode_t)0x1000)
136 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
137 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
139 /* File is opened with O_PATH; almost nothing can be done with it */
140 #define FMODE_PATH ((__force fmode_t)0x4000)
142 /* File needs atomic accesses to f_pos */
143 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
144 /* Write access to underlying fs */
145 #define FMODE_WRITER ((__force fmode_t)0x10000)
146 /* Has read method(s) */
147 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
148 /* Has write method(s) */
149 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
151 #define FMODE_OPENED ((__force fmode_t)0x80000)
152 #define FMODE_CREATED ((__force fmode_t)0x100000)
154 /* File was opened by fanotify and shouldn't generate fanotify events */
155 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
157 /* File is capable of returning -EAGAIN if I/O will block */
158 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
161 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
162 * that indicates that they should check the contents of the iovec are
163 * valid, but not check the memory that the iovec elements
166 #define CHECK_IOVEC_ONLY -1
169 * Attribute flags. These should be or-ed together to figure out what
172 #define ATTR_MODE (1 << 0)
173 #define ATTR_UID (1 << 1)
174 #define ATTR_GID (1 << 2)
175 #define ATTR_SIZE (1 << 3)
176 #define ATTR_ATIME (1 << 4)
177 #define ATTR_MTIME (1 << 5)
178 #define ATTR_CTIME (1 << 6)
179 #define ATTR_ATIME_SET (1 << 7)
180 #define ATTR_MTIME_SET (1 << 8)
181 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
182 #define ATTR_ATTR_FLAG (1 << 10)
183 #define ATTR_KILL_SUID (1 << 11)
184 #define ATTR_KILL_SGID (1 << 12)
185 #define ATTR_FILE (1 << 13)
186 #define ATTR_KILL_PRIV (1 << 14)
187 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
188 #define ATTR_TIMES_SET (1 << 16)
189 #define ATTR_TOUCH (1 << 17)
192 * Whiteout is represented by a char device. The following constants define the
193 * mode and device number to use.
195 #define WHITEOUT_MODE 0
196 #define WHITEOUT_DEV 0
199 * This is the Inode Attributes structure, used for notify_change(). It
200 * uses the above definitions as flags, to know which values have changed.
201 * Also, in this manner, a Filesystem can look at only the values it cares
202 * about. Basically, these are the attributes that the VFS layer can
203 * request to change from the FS layer.
205 * Derek Atkins <warlord@MIT.EDU> 94-10-20
208 unsigned int ia_valid;
213 struct timespec64 ia_atime;
214 struct timespec64 ia_mtime;
215 struct timespec64 ia_ctime;
218 * Not an attribute, but an auxiliary info for filesystems wanting to
219 * implement an ftruncate() like method. NOTE: filesystem should
220 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
222 struct file *ia_file;
226 * Includes for diskquotas.
228 #include <linux/quota.h>
231 * Maximum number of layers of fs stack. Needs to be limited to
232 * prevent kernel stack overflow
234 #define FILESYSTEM_MAX_STACK_DEPTH 2
237 * enum positive_aop_returns - aop return codes with specific semantics
239 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
240 * completed, that the page is still locked, and
241 * should be considered active. The VM uses this hint
242 * to return the page to the active list -- it won't
243 * be a candidate for writeback again in the near
244 * future. Other callers must be careful to unlock
245 * the page if they get this return. Returned by
248 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
249 * unlocked it and the page might have been truncated.
250 * The caller should back up to acquiring a new page and
251 * trying again. The aop will be taking reasonable
252 * precautions not to livelock. If the caller held a page
253 * reference, it should drop it before retrying. Returned
256 * address_space_operation functions return these large constants to indicate
257 * special semantics to the caller. These are much larger than the bytes in a
258 * page to allow for functions that return the number of bytes operated on in a
262 enum positive_aop_returns {
263 AOP_WRITEPAGE_ACTIVATE = 0x80000,
264 AOP_TRUNCATED_PAGE = 0x80001,
267 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
268 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
269 * helper code (eg buffer layer)
270 * to clear GFP_FS from alloc */
273 * oh the beauties of C type declarations.
276 struct address_space;
277 struct writeback_control;
280 * Write life time hint values.
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);
347 int (*readpages)(struct file *filp, struct address_space *mapping,
348 struct list_head *pages, unsigned nr_pages);
350 int (*write_begin)(struct file *, struct address_space *mapping,
351 loff_t pos, unsigned len, unsigned flags,
352 struct page **pagep, void **fsdata);
353 int (*write_end)(struct file *, struct address_space *mapping,
354 loff_t pos, unsigned len, unsigned copied,
355 struct page *page, void *fsdata);
357 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
358 sector_t (*bmap)(struct address_space *, sector_t);
359 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
360 int (*releasepage) (struct page *, gfp_t);
361 void (*freepage)(struct page *);
362 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
364 * migrate the contents of a page to the specified target. If
365 * migrate_mode is MIGRATE_ASYNC, it must not block.
367 int (*migratepage) (struct address_space *,
368 struct page *, struct page *, enum migrate_mode);
369 bool (*isolate_page)(struct page *, isolate_mode_t);
370 void (*putback_page)(struct page *);
371 int (*launder_page) (struct page *);
372 int (*is_partially_uptodate) (struct page *, unsigned long,
374 void (*is_dirty_writeback) (struct page *, bool *, bool *);
375 int (*error_remove_page)(struct address_space *, struct page *);
377 /* swapfile support */
378 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
380 void (*swap_deactivate)(struct file *file);
383 extern const struct address_space_operations empty_aops;
386 * pagecache_write_begin/pagecache_write_end must be used by general code
387 * to write into the pagecache.
389 int pagecache_write_begin(struct file *, struct address_space *mapping,
390 loff_t pos, unsigned len, unsigned flags,
391 struct page **pagep, void **fsdata);
393 int pagecache_write_end(struct file *, struct address_space *mapping,
394 loff_t pos, unsigned len, unsigned copied,
395 struct page *page, void *fsdata);
397 struct address_space {
398 struct inode *host; /* owner: inode, block_device */
399 struct radix_tree_root i_pages; /* cached pages */
400 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
401 struct rb_root_cached i_mmap; /* tree of private and shared mappings */
402 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
403 /* Protected by the i_pages lock */
404 unsigned long nrpages; /* number of total pages */
405 /* number of shadow or DAX exceptional entries */
406 unsigned long nrexceptional;
407 pgoff_t writeback_index;/* writeback starts here */
408 const struct address_space_operations *a_ops; /* methods */
409 unsigned long flags; /* error bits */
410 spinlock_t private_lock; /* for use by the address_space */
411 gfp_t gfp_mask; /* implicit gfp mask for allocations */
412 struct list_head private_list; /* for use by the address_space */
413 void *private_data; /* ditto */
415 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
417 * On most architectures that alignment is already the case; but
418 * must be enforced here for CRIS, to let the least significant bit
419 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
421 struct request_queue;
423 struct block_device {
424 dev_t bd_dev; /* not a kdev_t - it's a search key */
426 struct inode * bd_inode; /* will die */
427 struct super_block * bd_super;
428 struct mutex bd_mutex; /* open/close mutex */
432 bool bd_write_holder;
434 struct list_head bd_holder_disks;
436 struct block_device * bd_contains;
437 unsigned bd_block_size;
439 struct hd_struct * bd_part;
440 /* number of times partitions within this device have been opened. */
441 unsigned bd_part_count;
443 struct gendisk * bd_disk;
444 struct request_queue * bd_queue;
445 struct backing_dev_info *bd_bdi;
446 struct list_head bd_list;
448 * Private data. You must have bd_claim'ed the block_device
449 * to use this. NOTE: bd_claim allows an owner to claim
450 * the same device multiple times, the owner must take special
451 * care to not mess up bd_private for that case.
453 unsigned long bd_private;
455 /* The counter of freeze processes */
456 int bd_fsfreeze_count;
457 /* Mutex for freeze */
458 struct mutex bd_fsfreeze_mutex;
459 } __randomize_layout;
462 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
465 #define PAGECACHE_TAG_DIRTY 0
466 #define PAGECACHE_TAG_WRITEBACK 1
467 #define PAGECACHE_TAG_TOWRITE 2
469 int mapping_tagged(struct address_space *mapping, int tag);
471 static inline void i_mmap_lock_write(struct address_space *mapping)
473 down_write(&mapping->i_mmap_rwsem);
476 static inline void i_mmap_unlock_write(struct address_space *mapping)
478 up_write(&mapping->i_mmap_rwsem);
481 static inline void i_mmap_lock_read(struct address_space *mapping)
483 down_read(&mapping->i_mmap_rwsem);
486 static inline void i_mmap_unlock_read(struct address_space *mapping)
488 up_read(&mapping->i_mmap_rwsem);
492 * Might pages of this file be mapped into userspace?
494 static inline int mapping_mapped(struct address_space *mapping)
496 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
500 * Might pages of this file have been modified in userspace?
501 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
502 * marks vma as VM_SHARED if it is shared, and the file was opened for
503 * writing i.e. vma may be mprotected writable even if now readonly.
505 * If i_mmap_writable is negative, no new writable mappings are allowed. You
506 * can only deny writable mappings, if none exists right now.
508 static inline int mapping_writably_mapped(struct address_space *mapping)
510 return atomic_read(&mapping->i_mmap_writable) > 0;
513 static inline int mapping_map_writable(struct address_space *mapping)
515 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
519 static inline void mapping_unmap_writable(struct address_space *mapping)
521 atomic_dec(&mapping->i_mmap_writable);
524 static inline int mapping_deny_writable(struct address_space *mapping)
526 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
530 static inline void mapping_allow_writable(struct address_space *mapping)
532 atomic_inc(&mapping->i_mmap_writable);
536 * Use sequence counter to get consistent i_size on 32-bit processors.
538 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
539 #include <linux/seqlock.h>
540 #define __NEED_I_SIZE_ORDERED
541 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
543 #define i_size_ordered_init(inode) do { } while (0)
547 #define ACL_NOT_CACHED ((void *)(-1))
548 #define ACL_DONT_CACHE ((void *)(-3))
550 static inline struct posix_acl *
551 uncached_acl_sentinel(struct task_struct *task)
553 return (void *)task + 1;
557 is_uncached_acl(struct posix_acl *acl)
559 return (long)acl & 1;
562 #define IOP_FASTPERM 0x0001
563 #define IOP_LOOKUP 0x0002
564 #define IOP_NOFOLLOW 0x0004
565 #define IOP_XATTR 0x0008
566 #define IOP_DEFAULT_READLINK 0x0010
568 struct fsnotify_mark_connector;
571 * Keep mostly read-only and often accessed (especially for
572 * the RCU path lookup and 'stat' data) fields at the beginning
573 * of the 'struct inode'
577 unsigned short i_opflags;
580 unsigned int i_flags;
582 #ifdef CONFIG_FS_POSIX_ACL
583 struct posix_acl *i_acl;
584 struct posix_acl *i_default_acl;
587 const struct inode_operations *i_op;
588 struct super_block *i_sb;
589 struct address_space *i_mapping;
591 #ifdef CONFIG_SECURITY
595 /* Stat data, not accessed from path walking */
598 * Filesystems may only read i_nlink directly. They shall use the
599 * following functions for modification:
601 * (set|clear|inc|drop)_nlink
602 * inode_(inc|dec)_link_count
605 const unsigned int i_nlink;
606 unsigned int __i_nlink;
610 struct timespec64 i_atime;
611 struct timespec64 i_mtime;
612 struct timespec64 i_ctime;
613 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
614 unsigned short i_bytes;
615 unsigned int i_blkbits;
616 enum rw_hint i_write_hint;
619 #ifdef __NEED_I_SIZE_ORDERED
620 seqcount_t i_size_seqcount;
624 unsigned long i_state;
625 struct rw_semaphore i_rwsem;
627 unsigned long dirtied_when; /* jiffies of first dirtying */
628 unsigned long dirtied_time_when;
630 struct hlist_node i_hash;
631 struct list_head i_io_list; /* backing dev IO list */
632 #ifdef CONFIG_CGROUP_WRITEBACK
633 struct bdi_writeback *i_wb; /* the associated cgroup wb */
635 /* foreign inode detection, see wbc_detach_inode() */
637 u16 i_wb_frn_avg_time;
638 u16 i_wb_frn_history;
640 struct list_head i_lru; /* inode LRU list */
641 struct list_head i_sb_list;
642 struct list_head i_wb_list; /* backing dev writeback list */
644 struct hlist_head i_dentry;
645 struct rcu_head i_rcu;
647 atomic64_t i_version;
649 atomic_t i_dio_count;
650 atomic_t i_writecount;
652 atomic_t i_readcount; /* struct files open RO */
654 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
655 struct file_lock_context *i_flctx;
656 struct address_space i_data;
657 struct list_head i_devices;
659 struct pipe_inode_info *i_pipe;
660 struct block_device *i_bdev;
668 #ifdef CONFIG_FSNOTIFY
669 __u32 i_fsnotify_mask; /* all events this inode cares about */
670 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
673 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
674 struct fscrypt_info *i_crypt_info;
677 void *i_private; /* fs or device private pointer */
678 } __randomize_layout;
680 static inline unsigned int i_blocksize(const struct inode *node)
682 return (1 << node->i_blkbits);
685 static inline int inode_unhashed(struct inode *inode)
687 return hlist_unhashed(&inode->i_hash);
691 * inode->i_mutex nesting subclasses for the lock validator:
693 * 0: the object of the current VFS operation
697 * 4: second non-directory
698 * 5: second parent (when locking independent directories in rename)
700 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
701 * non-directories at once.
703 * The locking order between these classes is
704 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
706 enum inode_i_mutex_lock_class
716 static inline void inode_lock(struct inode *inode)
718 down_write(&inode->i_rwsem);
721 static inline void inode_unlock(struct inode *inode)
723 up_write(&inode->i_rwsem);
726 static inline void inode_lock_shared(struct inode *inode)
728 down_read(&inode->i_rwsem);
731 static inline void inode_unlock_shared(struct inode *inode)
733 up_read(&inode->i_rwsem);
736 static inline int inode_trylock(struct inode *inode)
738 return down_write_trylock(&inode->i_rwsem);
741 static inline int inode_trylock_shared(struct inode *inode)
743 return down_read_trylock(&inode->i_rwsem);
746 static inline int inode_is_locked(struct inode *inode)
748 return rwsem_is_locked(&inode->i_rwsem);
751 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
753 down_write_nested(&inode->i_rwsem, subclass);
756 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
758 down_read_nested(&inode->i_rwsem, subclass);
761 void lock_two_nondirectories(struct inode *, struct inode*);
762 void unlock_two_nondirectories(struct inode *, struct inode*);
765 * NOTE: in a 32bit arch with a preemptable kernel and
766 * an UP compile the i_size_read/write must be atomic
767 * with respect to the local cpu (unlike with preempt disabled),
768 * but they don't need to be atomic with respect to other cpus like in
769 * true SMP (so they need either to either locally disable irq around
770 * the read or for example on x86 they can be still implemented as a
771 * cmpxchg8b without the need of the lock prefix). For SMP compiles
772 * and 64bit archs it makes no difference if preempt is enabled or not.
774 static inline loff_t i_size_read(const struct inode *inode)
776 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
781 seq = read_seqcount_begin(&inode->i_size_seqcount);
782 i_size = inode->i_size;
783 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
785 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
789 i_size = inode->i_size;
793 return inode->i_size;
798 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
799 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
800 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
802 static inline void i_size_write(struct inode *inode, loff_t i_size)
804 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
806 write_seqcount_begin(&inode->i_size_seqcount);
807 inode->i_size = i_size;
808 write_seqcount_end(&inode->i_size_seqcount);
810 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
812 inode->i_size = i_size;
815 inode->i_size = i_size;
819 static inline unsigned iminor(const struct inode *inode)
821 return MINOR(inode->i_rdev);
824 static inline unsigned imajor(const struct inode *inode)
826 return MAJOR(inode->i_rdev);
829 extern struct block_device *I_BDEV(struct inode *inode);
832 rwlock_t lock; /* protects pid, uid, euid fields */
833 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
834 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
835 kuid_t uid, euid; /* uid/euid of process setting the owner */
836 int signum; /* posix.1b rt signal to be delivered on IO */
840 * Track a single file's readahead state
842 struct file_ra_state {
843 pgoff_t start; /* where readahead started */
844 unsigned int size; /* # of readahead pages */
845 unsigned int async_size; /* do asynchronous readahead when
846 there are only # of pages ahead */
848 unsigned int ra_pages; /* Maximum readahead window */
849 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
850 loff_t prev_pos; /* Cache last read() position */
854 * Check if @index falls in the readahead windows.
856 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
858 return (index >= ra->start &&
859 index < ra->start + ra->size);
864 struct llist_node fu_llist;
865 struct rcu_head fu_rcuhead;
868 struct inode *f_inode; /* cached value */
869 const struct file_operations *f_op;
872 * Protects f_ep_links, f_flags.
873 * Must not be taken from IRQ context.
876 enum rw_hint f_write_hint;
877 atomic_long_t f_count;
878 unsigned int f_flags;
880 struct mutex f_pos_lock;
882 struct fown_struct f_owner;
883 const struct cred *f_cred;
884 struct file_ra_state f_ra;
887 #ifdef CONFIG_SECURITY
890 /* needed for tty driver, and maybe others */
894 /* Used by fs/eventpoll.c to link all the hooks to this file */
895 struct list_head f_ep_links;
896 struct list_head f_tfile_llink;
897 #endif /* #ifdef CONFIG_EPOLL */
898 struct address_space *f_mapping;
901 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
906 /* file identifier */
907 unsigned char f_handle[0];
910 static inline struct file *get_file(struct file *f)
912 atomic_long_inc(&f->f_count);
915 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
916 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
917 #define file_count(x) atomic_long_read(&(x)->f_count)
919 #define MAX_NON_LFS ((1UL<<31) - 1)
921 /* Page cache limit. The filesystems should put that into their s_maxbytes
922 limits, otherwise bad things can happen in VM. */
923 #if BITS_PER_LONG==32
924 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
925 #elif BITS_PER_LONG==64
926 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
931 #define FL_DELEG 4 /* NFSv4 delegation */
932 #define FL_ACCESS 8 /* not trying to lock, just looking */
933 #define FL_EXISTS 16 /* when unlocking, test for existence */
934 #define FL_LEASE 32 /* lease held on this file */
935 #define FL_CLOSE 64 /* unlock on close */
936 #define FL_SLEEP 128 /* A blocking lock */
937 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
938 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
939 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
940 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
942 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
945 * Special return value from posix_lock_file() and vfs_lock_file() for
946 * asynchronous locking.
948 #define FILE_LOCK_DEFERRED 1
950 /* legacy typedef, should eventually be removed */
951 typedef void *fl_owner_t;
955 struct file_lock_operations {
956 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
957 void (*fl_release_private)(struct file_lock *);
960 struct lock_manager_operations {
961 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
962 unsigned long (*lm_owner_key)(struct file_lock *);
963 fl_owner_t (*lm_get_owner)(fl_owner_t);
964 void (*lm_put_owner)(fl_owner_t);
965 void (*lm_notify)(struct file_lock *); /* unblock callback */
966 int (*lm_grant)(struct file_lock *, int);
967 bool (*lm_break)(struct file_lock *);
968 int (*lm_change)(struct file_lock *, int, struct list_head *);
969 void (*lm_setup)(struct file_lock *, void **);
972 struct lock_manager {
973 struct list_head list;
975 * NFSv4 and up also want opens blocked during the grace period;
982 void locks_start_grace(struct net *, struct lock_manager *);
983 void locks_end_grace(struct lock_manager *);
984 bool locks_in_grace(struct net *);
985 bool opens_in_grace(struct net *);
987 /* that will die - we need it for nfs_lock_info */
988 #include <linux/nfs_fs_i.h>
991 * struct file_lock represents a generic "file lock". It's used to represent
992 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
993 * note that the same struct is used to represent both a request for a lock and
994 * the lock itself, but the same object is never used for both.
996 * FIXME: should we create a separate "struct lock_request" to help distinguish
999 * The varous i_flctx lists are ordered by:
1002 * 2) lock range start
1005 * Obviously, the last two criteria only matter for POSIX locks.
1008 struct file_lock *fl_next; /* singly linked list for this inode */
1009 struct list_head fl_list; /* link into file_lock_context */
1010 struct hlist_node fl_link; /* node in global lists */
1011 struct list_head fl_block; /* circular list of blocked processes */
1012 fl_owner_t fl_owner;
1013 unsigned int fl_flags;
1014 unsigned char fl_type;
1015 unsigned int fl_pid;
1016 int fl_link_cpu; /* what cpu's list is this on? */
1017 wait_queue_head_t fl_wait;
1018 struct file *fl_file;
1022 struct fasync_struct * fl_fasync; /* for lease break notifications */
1023 /* for lease breaks: */
1024 unsigned long fl_break_time;
1025 unsigned long fl_downgrade_time;
1027 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1028 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1030 struct nfs_lock_info nfs_fl;
1031 struct nfs4_lock_info nfs4_fl;
1033 struct list_head link; /* link in AFS vnode's pending_locks list */
1034 int state; /* state of grant or error if -ve */
1037 } __randomize_layout;
1039 struct file_lock_context {
1040 spinlock_t flc_lock;
1041 struct list_head flc_flock;
1042 struct list_head flc_posix;
1043 struct list_head flc_lease;
1046 /* The following constant reflects the upper bound of the file/locking space */
1048 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1049 #define OFFSET_MAX INT_LIMIT(loff_t)
1050 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1053 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1056 * Return the inode to use for locking
1058 * For overlayfs this should be the overlay inode, not the real inode returned
1059 * by file_inode(). For any other fs file_inode(filp) and locks_inode(filp) are
1062 static inline struct inode *locks_inode(const struct file *f)
1064 return f->f_path.dentry->d_inode;
1067 #ifdef CONFIG_FILE_LOCKING
1068 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1069 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1072 #if BITS_PER_LONG == 32
1073 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1074 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1078 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1079 extern int fcntl_getlease(struct file *filp);
1082 void locks_free_lock_context(struct inode *inode);
1083 void locks_free_lock(struct file_lock *fl);
1084 extern void locks_init_lock(struct file_lock *);
1085 extern struct file_lock * locks_alloc_lock(void);
1086 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1087 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1088 extern void locks_remove_posix(struct file *, fl_owner_t);
1089 extern void locks_remove_file(struct file *);
1090 extern void locks_release_private(struct file_lock *);
1091 extern void posix_test_lock(struct file *, struct file_lock *);
1092 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1093 extern int posix_unblock_lock(struct file_lock *);
1094 extern int vfs_test_lock(struct file *, struct file_lock *);
1095 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1096 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1097 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1098 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1099 extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1100 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1101 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1102 extern int lease_modify(struct file_lock *, int, struct list_head *);
1103 struct files_struct;
1104 extern void show_fd_locks(struct seq_file *f,
1105 struct file *filp, struct files_struct *files);
1106 #else /* !CONFIG_FILE_LOCKING */
1107 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1108 struct flock __user *user)
1113 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1114 unsigned int cmd, struct flock __user *user)
1119 #if BITS_PER_LONG == 32
1120 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1121 struct flock64 __user *user)
1126 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1127 unsigned int cmd, struct flock64 __user *user)
1132 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1137 static inline int fcntl_getlease(struct file *filp)
1143 locks_free_lock_context(struct inode *inode)
1147 static inline void locks_init_lock(struct file_lock *fl)
1152 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1157 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1162 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1167 static inline void locks_remove_file(struct file *filp)
1172 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1177 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1178 struct file_lock *conflock)
1183 static inline int posix_unblock_lock(struct file_lock *waiter)
1188 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1193 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1194 struct file_lock *fl, struct file_lock *conf)
1199 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1204 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1209 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1214 static inline void lease_get_mtime(struct inode *inode,
1215 struct timespec64 *time)
1220 static inline int generic_setlease(struct file *filp, long arg,
1221 struct file_lock **flp, void **priv)
1226 static inline int vfs_setlease(struct file *filp, long arg,
1227 struct file_lock **lease, void **priv)
1232 static inline int lease_modify(struct file_lock *fl, int arg,
1233 struct list_head *dispose)
1238 struct files_struct;
1239 static inline void show_fd_locks(struct seq_file *f,
1240 struct file *filp, struct files_struct *files) {}
1241 #endif /* !CONFIG_FILE_LOCKING */
1243 static inline struct inode *file_inode(const struct file *f)
1248 static inline struct dentry *file_dentry(const struct file *file)
1250 return d_real(file->f_path.dentry, file_inode(file), 0, 0);
1253 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1255 return locks_lock_inode_wait(locks_inode(filp), fl);
1258 struct fasync_struct {
1262 struct fasync_struct *fa_next; /* singly linked list */
1263 struct file *fa_file;
1264 struct rcu_head fa_rcu;
1267 #define FASYNC_MAGIC 0x4601
1269 /* SMP safe fasync helpers: */
1270 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1271 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1272 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1273 extern struct fasync_struct *fasync_alloc(void);
1274 extern void fasync_free(struct fasync_struct *);
1276 /* can be called from interrupts */
1277 extern void kill_fasync(struct fasync_struct **, int, int);
1279 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1280 extern int f_setown(struct file *filp, unsigned long arg, int force);
1281 extern void f_delown(struct file *filp);
1282 extern pid_t f_getown(struct file *filp);
1283 extern int send_sigurg(struct fown_struct *fown);
1286 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1287 * represented in both.
1289 #define SB_RDONLY 1 /* Mount read-only */
1290 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1291 #define SB_NODEV 4 /* Disallow access to device special files */
1292 #define SB_NOEXEC 8 /* Disallow program execution */
1293 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1294 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1295 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1296 #define SB_NOATIME 1024 /* Do not update access times. */
1297 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1298 #define SB_SILENT 32768
1299 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1300 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1301 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1302 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1304 /* These sb flags are internal to the kernel */
1305 #define SB_SUBMOUNT (1<<26)
1306 #define SB_NOREMOTELOCK (1<<27)
1307 #define SB_NOSEC (1<<28)
1308 #define SB_BORN (1<<29)
1309 #define SB_ACTIVE (1<<30)
1310 #define SB_NOUSER (1<<31)
1316 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1317 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1318 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1319 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1320 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1323 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1324 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1325 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1326 #define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */
1328 /* sb->s_iflags to limit user namespace mounts */
1329 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1330 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1331 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1333 /* Possible states of 'frozen' field */
1335 SB_UNFROZEN = 0, /* FS is unfrozen */
1336 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1337 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1338 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1339 * internal threads if needed) */
1340 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1343 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1346 int frozen; /* Is sb frozen? */
1347 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1348 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1351 struct super_block {
1352 struct list_head s_list; /* Keep this first */
1353 dev_t s_dev; /* search index; _not_ kdev_t */
1354 unsigned char s_blocksize_bits;
1355 unsigned long s_blocksize;
1356 loff_t s_maxbytes; /* Max file size */
1357 struct file_system_type *s_type;
1358 const struct super_operations *s_op;
1359 const struct dquot_operations *dq_op;
1360 const struct quotactl_ops *s_qcop;
1361 const struct export_operations *s_export_op;
1362 unsigned long s_flags;
1363 unsigned long s_iflags; /* internal SB_I_* flags */
1364 unsigned long s_magic;
1365 struct dentry *s_root;
1366 struct rw_semaphore s_umount;
1369 #ifdef CONFIG_SECURITY
1372 const struct xattr_handler **s_xattr;
1373 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
1374 const struct fscrypt_operations *s_cop;
1376 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1377 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1378 struct block_device *s_bdev;
1379 struct backing_dev_info *s_bdi;
1380 struct mtd_info *s_mtd;
1381 struct hlist_node s_instances;
1382 unsigned int s_quota_types; /* Bitmask of supported quota types */
1383 struct quota_info s_dquot; /* Diskquota specific options */
1385 struct sb_writers s_writers;
1387 char s_id[32]; /* Informational name */
1388 uuid_t s_uuid; /* UUID */
1390 void *s_fs_info; /* Filesystem private info */
1391 unsigned int s_max_links;
1394 /* Granularity of c/m/atime in ns.
1395 Cannot be worse than a second */
1399 * The next field is for VFS *only*. No filesystems have any business
1400 * even looking at it. You had been warned.
1402 struct mutex s_vfs_rename_mutex; /* Kludge */
1405 * Filesystem subtype. If non-empty the filesystem type field
1406 * in /proc/mounts will be "type.subtype"
1410 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1413 * Saved pool identifier for cleancache (-1 means none)
1415 int cleancache_poolid;
1417 struct shrinker s_shrink; /* per-sb shrinker handle */
1419 /* Number of inodes with nlink == 0 but still referenced */
1420 atomic_long_t s_remove_count;
1422 /* Being remounted read-only */
1423 int s_readonly_remount;
1425 /* AIO completions deferred from interrupt context */
1426 struct workqueue_struct *s_dio_done_wq;
1427 struct hlist_head s_pins;
1430 * Owning user namespace and default context in which to
1431 * interpret filesystem uids, gids, quotas, device nodes,
1432 * xattrs and security labels.
1434 struct user_namespace *s_user_ns;
1437 * Keep the lru lists last in the structure so they always sit on their
1438 * own individual cachelines.
1440 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1441 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1442 struct rcu_head rcu;
1443 struct work_struct destroy_work;
1445 struct mutex s_sync_lock; /* sync serialisation lock */
1448 * Indicates how deep in a filesystem stack this SB is
1452 /* s_inode_list_lock protects s_inodes */
1453 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1454 struct list_head s_inodes; /* all inodes */
1456 spinlock_t s_inode_wblist_lock;
1457 struct list_head s_inodes_wb; /* writeback inodes */
1458 } __randomize_layout;
1460 /* Helper functions so that in most cases filesystems will
1461 * not need to deal directly with kuid_t and kgid_t and can
1462 * instead deal with the raw numeric values that are stored
1463 * in the filesystem.
1465 static inline uid_t i_uid_read(const struct inode *inode)
1467 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1470 static inline gid_t i_gid_read(const struct inode *inode)
1472 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1475 static inline void i_uid_write(struct inode *inode, uid_t uid)
1477 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1480 static inline void i_gid_write(struct inode *inode, gid_t gid)
1482 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1485 extern struct timespec64 timespec64_trunc(struct timespec64 t, unsigned gran);
1486 extern struct timespec64 current_time(struct inode *inode);
1489 * Snapshotting support.
1492 void __sb_end_write(struct super_block *sb, int level);
1493 int __sb_start_write(struct super_block *sb, int level, bool wait);
1495 #define __sb_writers_acquired(sb, lev) \
1496 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1497 #define __sb_writers_release(sb, lev) \
1498 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1501 * sb_end_write - drop write access to a superblock
1502 * @sb: the super we wrote to
1504 * Decrement number of writers to the filesystem. Wake up possible waiters
1505 * wanting to freeze the filesystem.
1507 static inline void sb_end_write(struct super_block *sb)
1509 __sb_end_write(sb, SB_FREEZE_WRITE);
1513 * sb_end_pagefault - drop write access to a superblock from a page fault
1514 * @sb: the super we wrote to
1516 * Decrement number of processes handling write page fault to the filesystem.
1517 * Wake up possible waiters wanting to freeze the filesystem.
1519 static inline void sb_end_pagefault(struct super_block *sb)
1521 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1525 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1526 * @sb: the super we wrote to
1528 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1529 * waiters wanting to freeze the filesystem.
1531 static inline void sb_end_intwrite(struct super_block *sb)
1533 __sb_end_write(sb, SB_FREEZE_FS);
1537 * sb_start_write - get write access to a superblock
1538 * @sb: the super we write to
1540 * When a process wants to write data or metadata to a file system (i.e. dirty
1541 * a page or an inode), it should embed the operation in a sb_start_write() -
1542 * sb_end_write() pair to get exclusion against file system freezing. This
1543 * function increments number of writers preventing freezing. If the file
1544 * system is already frozen, the function waits until the file system is
1547 * Since freeze protection behaves as a lock, users have to preserve
1548 * ordering of freeze protection and other filesystem locks. Generally,
1549 * freeze protection should be the outermost lock. In particular, we have:
1552 * -> i_mutex (write path, truncate, directory ops, ...)
1553 * -> s_umount (freeze_super, thaw_super)
1555 static inline void sb_start_write(struct super_block *sb)
1557 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1560 static inline int sb_start_write_trylock(struct super_block *sb)
1562 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1566 * sb_start_pagefault - get write access to a superblock from a page fault
1567 * @sb: the super we write to
1569 * When a process starts handling write page fault, it should embed the
1570 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1571 * exclusion against file system freezing. This is needed since the page fault
1572 * is going to dirty a page. This function increments number of running page
1573 * faults preventing freezing. If the file system is already frozen, the
1574 * function waits until the file system is thawed.
1576 * Since page fault freeze protection behaves as a lock, users have to preserve
1577 * ordering of freeze protection and other filesystem locks. It is advised to
1578 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1579 * handling code implies lock dependency:
1582 * -> sb_start_pagefault
1584 static inline void sb_start_pagefault(struct super_block *sb)
1586 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1590 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1591 * @sb: the super we write to
1593 * This is the third level of protection against filesystem freezing. It is
1594 * free for use by a filesystem. The only requirement is that it must rank
1595 * below sb_start_pagefault.
1597 * For example filesystem can call sb_start_intwrite() when starting a
1598 * transaction which somewhat eases handling of freezing for internal sources
1599 * of filesystem changes (internal fs threads, discarding preallocation on file
1602 static inline void sb_start_intwrite(struct super_block *sb)
1604 __sb_start_write(sb, SB_FREEZE_FS, true);
1607 static inline int sb_start_intwrite_trylock(struct super_block *sb)
1609 return __sb_start_write(sb, SB_FREEZE_FS, false);
1613 extern bool inode_owner_or_capable(const struct inode *inode);
1616 * VFS helper functions..
1618 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1619 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1620 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1621 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1622 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1623 extern int vfs_rmdir(struct inode *, struct dentry *);
1624 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1625 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1626 extern int vfs_whiteout(struct inode *, struct dentry *);
1628 extern struct dentry *vfs_tmpfile(struct dentry *dentry, umode_t mode,
1631 int vfs_mkobj(struct dentry *, umode_t,
1632 int (*f)(struct dentry *, umode_t, void *),
1636 * VFS file helper functions.
1638 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1640 extern bool may_open_dev(const struct path *path);
1642 * VFS FS_IOC_FIEMAP helper definitions.
1644 struct fiemap_extent_info {
1645 unsigned int fi_flags; /* Flags as passed from user */
1646 unsigned int fi_extents_mapped; /* Number of mapped extents */
1647 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1648 struct fiemap_extent __user *fi_extents_start; /* Start of
1649 fiemap_extent array */
1651 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1652 u64 phys, u64 len, u32 flags);
1653 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1658 * NOTE! These match bits 12..15 of stat.st_mode
1659 * (ie "(i_mode >> 12) & 15").
1661 #define DT_UNKNOWN 0
1672 * This is the "filldir" function type, used by readdir() to let
1673 * the kernel specify what kind of dirent layout it wants to have.
1674 * This allows the kernel to read directories into kernel space or
1675 * to have different dirent layouts depending on the binary type.
1678 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1681 struct dir_context {
1686 struct block_device_operations;
1688 /* These macros are for out of kernel modules to test that
1689 * the kernel supports the unlocked_ioctl and compat_ioctl
1690 * fields in struct file_operations. */
1691 #define HAVE_COMPAT_IOCTL 1
1692 #define HAVE_UNLOCKED_IOCTL 1
1695 * These flags let !MMU mmap() govern direct device mapping vs immediate
1696 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1698 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1699 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1700 * NOMMU_MAP_READ: Can be mapped for reading
1701 * NOMMU_MAP_WRITE: Can be mapped for writing
1702 * NOMMU_MAP_EXEC: Can be mapped for execution
1704 #define NOMMU_MAP_COPY 0x00000001
1705 #define NOMMU_MAP_DIRECT 0x00000008
1706 #define NOMMU_MAP_READ VM_MAYREAD
1707 #define NOMMU_MAP_WRITE VM_MAYWRITE
1708 #define NOMMU_MAP_EXEC VM_MAYEXEC
1710 #define NOMMU_VMFLAGS \
1711 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1716 struct file_operations {
1717 struct module *owner;
1718 loff_t (*llseek) (struct file *, loff_t, int);
1719 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1720 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1721 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1722 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1723 int (*iterate) (struct file *, struct dir_context *);
1724 int (*iterate_shared) (struct file *, struct dir_context *);
1725 __poll_t (*poll) (struct file *, struct poll_table_struct *);
1726 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1727 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1728 int (*mmap) (struct file *, struct vm_area_struct *);
1729 unsigned long mmap_supported_flags;
1730 int (*open) (struct inode *, struct file *);
1731 int (*flush) (struct file *, fl_owner_t id);
1732 int (*release) (struct inode *, struct file *);
1733 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1734 int (*fasync) (int, struct file *, int);
1735 int (*lock) (struct file *, int, struct file_lock *);
1736 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1737 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1738 int (*check_flags)(int);
1739 int (*flock) (struct file *, int, struct file_lock *);
1740 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1741 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1742 int (*setlease)(struct file *, long, struct file_lock **, void **);
1743 long (*fallocate)(struct file *file, int mode, loff_t offset,
1745 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1747 unsigned (*mmap_capabilities)(struct file *);
1749 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1750 loff_t, size_t, unsigned int);
1751 int (*clone_file_range)(struct file *, loff_t, struct file *, loff_t,
1753 ssize_t (*dedupe_file_range)(struct file *, u64, u64, struct file *,
1755 } __randomize_layout;
1757 struct inode_operations {
1758 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1759 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1760 int (*permission) (struct inode *, int);
1761 struct posix_acl * (*get_acl)(struct inode *, int);
1763 int (*readlink) (struct dentry *, char __user *,int);
1765 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1766 int (*link) (struct dentry *,struct inode *,struct dentry *);
1767 int (*unlink) (struct inode *,struct dentry *);
1768 int (*symlink) (struct inode *,struct dentry *,const char *);
1769 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1770 int (*rmdir) (struct inode *,struct dentry *);
1771 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1772 int (*rename) (struct inode *, struct dentry *,
1773 struct inode *, struct dentry *, unsigned int);
1774 int (*setattr) (struct dentry *, struct iattr *);
1775 int (*getattr) (const struct path *, struct kstat *, u32, unsigned int);
1776 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1777 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1779 int (*update_time)(struct inode *, struct timespec64 *, int);
1780 int (*atomic_open)(struct inode *, struct dentry *,
1781 struct file *, unsigned open_flag,
1782 umode_t create_mode);
1783 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1784 int (*set_acl)(struct inode *, struct posix_acl *, int);
1785 } ____cacheline_aligned;
1787 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
1788 struct iov_iter *iter)
1790 return file->f_op->read_iter(kio, iter);
1793 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
1794 struct iov_iter *iter)
1796 return file->f_op->write_iter(kio, iter);
1799 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
1801 return file->f_op->mmap(file, vma);
1804 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1805 unsigned long nr_segs, unsigned long fast_segs,
1806 struct iovec *fast_pointer,
1807 struct iovec **ret_pointer);
1809 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1810 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1811 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1812 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1813 unsigned long, loff_t *, rwf_t);
1814 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1815 loff_t, size_t, unsigned int);
1816 extern int vfs_clone_file_prep_inodes(struct inode *inode_in, loff_t pos_in,
1817 struct inode *inode_out, loff_t pos_out,
1818 u64 *len, bool is_dedupe);
1819 extern int vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1820 struct file *file_out, loff_t pos_out, u64 len);
1821 extern int vfs_dedupe_file_range_compare(struct inode *src, loff_t srcoff,
1822 struct inode *dest, loff_t destoff,
1823 loff_t len, bool *is_same);
1824 extern int vfs_dedupe_file_range(struct file *file,
1825 struct file_dedupe_range *same);
1827 struct super_operations {
1828 struct inode *(*alloc_inode)(struct super_block *sb);
1829 void (*destroy_inode)(struct inode *);
1831 void (*dirty_inode) (struct inode *, int flags);
1832 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1833 int (*drop_inode) (struct inode *);
1834 void (*evict_inode) (struct inode *);
1835 void (*put_super) (struct super_block *);
1836 int (*sync_fs)(struct super_block *sb, int wait);
1837 int (*freeze_super) (struct super_block *);
1838 int (*freeze_fs) (struct super_block *);
1839 int (*thaw_super) (struct super_block *);
1840 int (*unfreeze_fs) (struct super_block *);
1841 int (*statfs) (struct dentry *, struct kstatfs *);
1842 int (*remount_fs) (struct super_block *, int *, char *);
1843 void (*umount_begin) (struct super_block *);
1845 int (*show_options)(struct seq_file *, struct dentry *);
1846 int (*show_devname)(struct seq_file *, struct dentry *);
1847 int (*show_path)(struct seq_file *, struct dentry *);
1848 int (*show_stats)(struct seq_file *, struct dentry *);
1850 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1851 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1852 struct dquot **(*get_dquots)(struct inode *);
1854 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1855 long (*nr_cached_objects)(struct super_block *,
1856 struct shrink_control *);
1857 long (*free_cached_objects)(struct super_block *,
1858 struct shrink_control *);
1862 * Inode flags - they have no relation to superblock flags now
1864 #define S_SYNC 1 /* Writes are synced at once */
1865 #define S_NOATIME 2 /* Do not update access times */
1866 #define S_APPEND 4 /* Append-only file */
1867 #define S_IMMUTABLE 8 /* Immutable file */
1868 #define S_DEAD 16 /* removed, but still open directory */
1869 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1870 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1871 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1872 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1873 #define S_PRIVATE 512 /* Inode is fs-internal */
1874 #define S_IMA 1024 /* Inode has an associated IMA struct */
1875 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1876 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1877 #ifdef CONFIG_FS_DAX
1878 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1880 #define S_DAX 0 /* Make all the DAX code disappear */
1882 #define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */
1885 * Note that nosuid etc flags are inode-specific: setting some file-system
1886 * flags just means all the inodes inherit those flags by default. It might be
1887 * possible to override it selectively if you really wanted to with some
1888 * ioctl() that is not currently implemented.
1890 * Exception: SB_RDONLY is always applied to the entire file system.
1892 * Unfortunately, it is possible to change a filesystems flags with it mounted
1893 * with files in use. This means that all of the inodes will not have their
1894 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1895 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1897 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1899 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
1900 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
1901 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
1902 ((inode)->i_flags & S_SYNC))
1903 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
1904 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1905 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
1906 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
1907 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
1909 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1910 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1911 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1912 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
1914 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1915 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1916 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1917 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1918 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1919 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1920 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1921 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1922 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
1924 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1925 (inode)->i_rdev == WHITEOUT_DEV)
1927 static inline bool HAS_UNMAPPED_ID(struct inode *inode)
1929 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
1932 static inline enum rw_hint file_write_hint(struct file *file)
1934 if (file->f_write_hint != WRITE_LIFE_NOT_SET)
1935 return file->f_write_hint;
1937 return file_inode(file)->i_write_hint;
1940 static inline int iocb_flags(struct file *file);
1942 static inline u16 ki_hint_validate(enum rw_hint hint)
1944 typeof(((struct kiocb *)0)->ki_hint) max_hint = -1;
1946 if (hint <= max_hint)
1951 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
1953 *kiocb = (struct kiocb) {
1955 .ki_flags = iocb_flags(filp),
1956 .ki_hint = ki_hint_validate(file_write_hint(filp)),
1957 .ki_ioprio = IOPRIO_PRIO_VALUE(IOPRIO_CLASS_NONE, 0),
1962 * Inode state bits. Protected by inode->i_lock
1964 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1965 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1967 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1968 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1969 * various stages of removing an inode.
1971 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1973 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1974 * fdatasync(). i_atime is the usual cause.
1975 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1976 * these changes separately from I_DIRTY_SYNC so that we
1977 * don't have to write inode on fdatasync() when only
1978 * mtime has changed in it.
1979 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1980 * I_NEW Serves as both a mutex and completion notification.
1981 * New inodes set I_NEW. If two processes both create
1982 * the same inode, one of them will release its inode and
1983 * wait for I_NEW to be released before returning.
1984 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1985 * also cause waiting on I_NEW, without I_NEW actually
1986 * being set. find_inode() uses this to prevent returning
1987 * nearly-dead inodes.
1988 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1989 * is zero. I_FREEING must be set when I_WILL_FREE is
1991 * I_FREEING Set when inode is about to be freed but still has dirty
1992 * pages or buffers attached or the inode itself is still
1994 * I_CLEAR Added by clear_inode(). In this state the inode is
1995 * clean and can be destroyed. Inode keeps I_FREEING.
1997 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1998 * prohibited for many purposes. iget() must wait for
1999 * the inode to be completely released, then create it
2000 * anew. Other functions will just ignore such inodes,
2001 * if appropriate. I_NEW is used for waiting.
2003 * I_SYNC Writeback of inode is running. The bit is set during
2004 * data writeback, and cleared with a wakeup on the bit
2005 * address once it is done. The bit is also used to pin
2006 * the inode in memory for flusher thread.
2008 * I_REFERENCED Marks the inode as recently references on the LRU list.
2010 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2012 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2013 * synchronize competing switching instances and to tell
2014 * wb stat updates to grab the i_pages lock. See
2015 * inode_switch_wb_work_fn() for details.
2017 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2018 * and work dirs among overlayfs mounts.
2020 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2022 #define I_DIRTY_SYNC (1 << 0)
2023 #define I_DIRTY_DATASYNC (1 << 1)
2024 #define I_DIRTY_PAGES (1 << 2)
2026 #define I_NEW (1 << __I_NEW)
2027 #define I_WILL_FREE (1 << 4)
2028 #define I_FREEING (1 << 5)
2029 #define I_CLEAR (1 << 6)
2031 #define I_SYNC (1 << __I_SYNC)
2032 #define I_REFERENCED (1 << 8)
2033 #define __I_DIO_WAKEUP 9
2034 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2035 #define I_LINKABLE (1 << 10)
2036 #define I_DIRTY_TIME (1 << 11)
2037 #define __I_DIRTY_TIME_EXPIRED 12
2038 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
2039 #define I_WB_SWITCH (1 << 13)
2040 #define I_OVL_INUSE (1 << 14)
2042 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2043 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2044 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2046 extern void __mark_inode_dirty(struct inode *, int);
2047 static inline void mark_inode_dirty(struct inode *inode)
2049 __mark_inode_dirty(inode, I_DIRTY);
2052 static inline void mark_inode_dirty_sync(struct inode *inode)
2054 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2057 extern void inc_nlink(struct inode *inode);
2058 extern void drop_nlink(struct inode *inode);
2059 extern void clear_nlink(struct inode *inode);
2060 extern void set_nlink(struct inode *inode, unsigned int nlink);
2062 static inline void inode_inc_link_count(struct inode *inode)
2065 mark_inode_dirty(inode);
2068 static inline void inode_dec_link_count(struct inode *inode)
2071 mark_inode_dirty(inode);
2074 enum file_time_flags {
2081 extern void touch_atime(const struct path *);
2082 static inline void file_accessed(struct file *file)
2084 if (!(file->f_flags & O_NOATIME))
2085 touch_atime(&file->f_path);
2088 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2089 int sync_inode_metadata(struct inode *inode, int wait);
2091 struct file_system_type {
2094 #define FS_REQUIRES_DEV 1
2095 #define FS_BINARY_MOUNTDATA 2
2096 #define FS_HAS_SUBTYPE 4
2097 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2098 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2099 struct dentry *(*mount) (struct file_system_type *, int,
2100 const char *, void *);
2101 void (*kill_sb) (struct super_block *);
2102 struct module *owner;
2103 struct file_system_type * next;
2104 struct hlist_head fs_supers;
2106 struct lock_class_key s_lock_key;
2107 struct lock_class_key s_umount_key;
2108 struct lock_class_key s_vfs_rename_key;
2109 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2111 struct lock_class_key i_lock_key;
2112 struct lock_class_key i_mutex_key;
2113 struct lock_class_key i_mutex_dir_key;
2116 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2118 extern struct dentry *mount_ns(struct file_system_type *fs_type,
2119 int flags, void *data, void *ns, struct user_namespace *user_ns,
2120 int (*fill_super)(struct super_block *, void *, int));
2122 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2123 int flags, const char *dev_name, void *data,
2124 int (*fill_super)(struct super_block *, void *, int));
2126 static inline struct dentry *mount_bdev(struct file_system_type *fs_type,
2127 int flags, const char *dev_name, void *data,
2128 int (*fill_super)(struct super_block *, void *, int))
2130 return ERR_PTR(-ENODEV);
2133 extern struct dentry *mount_single(struct file_system_type *fs_type,
2134 int flags, void *data,
2135 int (*fill_super)(struct super_block *, void *, int));
2136 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2137 int flags, void *data,
2138 int (*fill_super)(struct super_block *, void *, int));
2139 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2140 void generic_shutdown_super(struct super_block *sb);
2142 void kill_block_super(struct super_block *sb);
2144 static inline void kill_block_super(struct super_block *sb)
2149 void kill_anon_super(struct super_block *sb);
2150 void kill_litter_super(struct super_block *sb);
2151 void deactivate_super(struct super_block *sb);
2152 void deactivate_locked_super(struct super_block *sb);
2153 int set_anon_super(struct super_block *s, void *data);
2154 int get_anon_bdev(dev_t *);
2155 void free_anon_bdev(dev_t);
2156 struct super_block *sget_userns(struct file_system_type *type,
2157 int (*test)(struct super_block *,void *),
2158 int (*set)(struct super_block *,void *),
2159 int flags, struct user_namespace *user_ns,
2161 struct super_block *sget(struct file_system_type *type,
2162 int (*test)(struct super_block *,void *),
2163 int (*set)(struct super_block *,void *),
2164 int flags, void *data);
2165 extern struct dentry *mount_pseudo_xattr(struct file_system_type *, char *,
2166 const struct super_operations *ops,
2167 const struct xattr_handler **xattr,
2168 const struct dentry_operations *dops,
2171 static inline struct dentry *
2172 mount_pseudo(struct file_system_type *fs_type, char *name,
2173 const struct super_operations *ops,
2174 const struct dentry_operations *dops, unsigned long magic)
2176 return mount_pseudo_xattr(fs_type, name, ops, NULL, dops, magic);
2179 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2180 #define fops_get(fops) \
2181 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2182 #define fops_put(fops) \
2183 do { if (fops) module_put((fops)->owner); } while(0)
2185 * This one is to be used *ONLY* from ->open() instances.
2186 * fops must be non-NULL, pinned down *and* module dependencies
2187 * should be sufficient to pin the caller down as well.
2189 #define replace_fops(f, fops) \
2191 struct file *__file = (f); \
2192 fops_put(__file->f_op); \
2193 BUG_ON(!(__file->f_op = (fops))); \
2196 extern int register_filesystem(struct file_system_type *);
2197 extern int unregister_filesystem(struct file_system_type *);
2198 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2199 #define kern_mount(type) kern_mount_data(type, NULL)
2200 extern void kern_unmount(struct vfsmount *mnt);
2201 extern int may_umount_tree(struct vfsmount *);
2202 extern int may_umount(struct vfsmount *);
2203 extern long do_mount(const char *, const char __user *,
2204 const char *, unsigned long, void *);
2205 extern struct vfsmount *collect_mounts(const struct path *);
2206 extern void drop_collected_mounts(struct vfsmount *);
2207 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2209 extern int vfs_statfs(const struct path *, struct kstatfs *);
2210 extern int user_statfs(const char __user *, struct kstatfs *);
2211 extern int fd_statfs(int, struct kstatfs *);
2212 extern int freeze_super(struct super_block *super);
2213 extern int thaw_super(struct super_block *super);
2214 extern bool our_mnt(struct vfsmount *mnt);
2215 extern __printf(2, 3)
2216 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2217 extern int super_setup_bdi(struct super_block *sb);
2219 extern int current_umask(void);
2221 extern void ihold(struct inode * inode);
2222 extern void iput(struct inode *);
2223 extern int generic_update_time(struct inode *, struct timespec64 *, int);
2226 extern struct kobject *fs_kobj;
2228 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2230 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2231 extern int locks_mandatory_locked(struct file *);
2232 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2235 * Candidates for mandatory locking have the setgid bit set
2236 * but no group execute bit - an otherwise meaningless combination.
2239 static inline int __mandatory_lock(struct inode *ino)
2241 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2245 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2246 * otherwise these will be advisory locks
2249 static inline int mandatory_lock(struct inode *ino)
2251 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2254 static inline int locks_verify_locked(struct file *file)
2256 if (mandatory_lock(locks_inode(file)))
2257 return locks_mandatory_locked(file);
2261 static inline int locks_verify_truncate(struct inode *inode,
2265 if (!inode->i_flctx || !mandatory_lock(inode))
2268 if (size < inode->i_size) {
2269 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2272 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2277 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2279 static inline int locks_mandatory_locked(struct file *file)
2284 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2285 loff_t start, loff_t end, unsigned char type)
2290 static inline int __mandatory_lock(struct inode *inode)
2295 static inline int mandatory_lock(struct inode *inode)
2300 static inline int locks_verify_locked(struct file *file)
2305 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2311 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2314 #ifdef CONFIG_FILE_LOCKING
2315 static inline int break_lease(struct inode *inode, unsigned int mode)
2318 * Since this check is lockless, we must ensure that any refcounts
2319 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2320 * could end up racing with tasks trying to set a new lease on this
2324 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2325 return __break_lease(inode, mode, FL_LEASE);
2329 static inline int break_deleg(struct inode *inode, unsigned int mode)
2332 * Since this check is lockless, we must ensure that any refcounts
2333 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2334 * could end up racing with tasks trying to set a new lease on this
2338 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2339 return __break_lease(inode, mode, FL_DELEG);
2343 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2347 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2348 if (ret == -EWOULDBLOCK && delegated_inode) {
2349 *delegated_inode = inode;
2355 static inline int break_deleg_wait(struct inode **delegated_inode)
2359 ret = break_deleg(*delegated_inode, O_WRONLY);
2360 iput(*delegated_inode);
2361 *delegated_inode = NULL;
2365 static inline int break_layout(struct inode *inode, bool wait)
2368 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2369 return __break_lease(inode,
2370 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2375 #else /* !CONFIG_FILE_LOCKING */
2376 static inline int break_lease(struct inode *inode, unsigned int mode)
2381 static inline int break_deleg(struct inode *inode, unsigned int mode)
2386 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2391 static inline int break_deleg_wait(struct inode **delegated_inode)
2397 static inline int break_layout(struct inode *inode, bool wait)
2402 #endif /* CONFIG_FILE_LOCKING */
2407 const char *name; /* pointer to actual string */
2408 const __user char *uptr; /* original userland pointer */
2410 struct audit_names *aname;
2414 extern long vfs_truncate(const struct path *, loff_t);
2415 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2417 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2419 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2421 extern struct file *file_open_name(struct filename *, int, umode_t);
2422 extern struct file *filp_open(const char *, int, umode_t);
2423 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2424 const char *, int, umode_t);
2425 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2426 static inline struct file *file_clone_open(struct file *file)
2428 return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2430 extern int filp_close(struct file *, fl_owner_t id);
2432 extern struct filename *getname_flags(const char __user *, int, int *);
2433 extern struct filename *getname(const char __user *);
2434 extern struct filename *getname_kernel(const char *);
2435 extern void putname(struct filename *name);
2437 extern int finish_open(struct file *file, struct dentry *dentry,
2438 int (*open)(struct inode *, struct file *));
2439 extern int finish_no_open(struct file *file, struct dentry *dentry);
2443 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2446 extern void __init vfs_caches_init_early(void);
2447 extern void __init vfs_caches_init(void);
2449 extern struct kmem_cache *names_cachep;
2451 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2452 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2455 extern int register_blkdev(unsigned int, const char *);
2456 extern void unregister_blkdev(unsigned int, const char *);
2457 extern void bdev_unhash_inode(dev_t dev);
2458 extern struct block_device *bdget(dev_t);
2459 extern struct block_device *bdgrab(struct block_device *bdev);
2460 extern void bd_set_size(struct block_device *, loff_t size);
2461 extern void bd_forget(struct inode *inode);
2462 extern void bdput(struct block_device *);
2463 extern void invalidate_bdev(struct block_device *);
2464 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2465 extern int sync_blockdev(struct block_device *bdev);
2466 extern void kill_bdev(struct block_device *);
2467 extern struct super_block *freeze_bdev(struct block_device *);
2468 extern void emergency_thaw_all(void);
2469 extern void emergency_thaw_bdev(struct super_block *sb);
2470 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2471 extern int fsync_bdev(struct block_device *);
2473 extern struct super_block *blockdev_superblock;
2475 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2477 return sb == blockdev_superblock;
2480 static inline void bd_forget(struct inode *inode) {}
2481 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2482 static inline void kill_bdev(struct block_device *bdev) {}
2483 static inline void invalidate_bdev(struct block_device *bdev) {}
2485 static inline struct super_block *freeze_bdev(struct block_device *sb)
2490 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2495 static inline int emergency_thaw_bdev(struct super_block *sb)
2500 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2504 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2509 extern int sync_filesystem(struct super_block *);
2510 extern const struct file_operations def_blk_fops;
2511 extern const struct file_operations def_chr_fops;
2513 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2514 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2515 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2516 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2517 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2519 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2521 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2522 extern int __blkdev_reread_part(struct block_device *bdev);
2523 extern int blkdev_reread_part(struct block_device *bdev);
2526 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2527 extern void bd_unlink_disk_holder(struct block_device *bdev,
2528 struct gendisk *disk);
2530 static inline int bd_link_disk_holder(struct block_device *bdev,
2531 struct gendisk *disk)
2535 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2536 struct gendisk *disk)
2543 #define CHRDEV_MAJOR_MAX 512
2544 /* Marks the bottom of the first segment of free char majors */
2545 #define CHRDEV_MAJOR_DYN_END 234
2546 /* Marks the top and bottom of the second segment of free char majors */
2547 #define CHRDEV_MAJOR_DYN_EXT_START 511
2548 #define CHRDEV_MAJOR_DYN_EXT_END 384
2550 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2551 extern int register_chrdev_region(dev_t, unsigned, const char *);
2552 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2553 unsigned int count, const char *name,
2554 const struct file_operations *fops);
2555 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2556 unsigned int count, const char *name);
2557 extern void unregister_chrdev_region(dev_t, unsigned);
2558 extern void chrdev_show(struct seq_file *,off_t);
2560 static inline int register_chrdev(unsigned int major, const char *name,
2561 const struct file_operations *fops)
2563 return __register_chrdev(major, 0, 256, name, fops);
2566 static inline void unregister_chrdev(unsigned int major, const char *name)
2568 __unregister_chrdev(major, 0, 256, name);
2571 /* fs/block_dev.c */
2572 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2573 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2576 #define BLKDEV_MAJOR_MAX 512
2577 extern const char *__bdevname(dev_t, char *buffer);
2578 extern const char *bdevname(struct block_device *bdev, char *buffer);
2579 extern struct block_device *lookup_bdev(const char *);
2580 extern void blkdev_show(struct seq_file *,off_t);
2583 #define BLKDEV_MAJOR_MAX 0
2586 extern void init_special_inode(struct inode *, umode_t, dev_t);
2588 /* Invalid inode operations -- fs/bad_inode.c */
2589 extern void make_bad_inode(struct inode *);
2590 extern bool is_bad_inode(struct inode *);
2593 extern void check_disk_size_change(struct gendisk *disk,
2594 struct block_device *bdev, bool verbose);
2595 extern int revalidate_disk(struct gendisk *);
2596 extern int check_disk_change(struct block_device *);
2597 extern int __invalidate_device(struct block_device *, bool);
2598 extern int invalidate_partition(struct gendisk *, int);
2600 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2601 pgoff_t start, pgoff_t end);
2603 static inline void invalidate_remote_inode(struct inode *inode)
2605 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2606 S_ISLNK(inode->i_mode))
2607 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2609 extern int invalidate_inode_pages2(struct address_space *mapping);
2610 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2611 pgoff_t start, pgoff_t end);
2612 extern int write_inode_now(struct inode *, int);
2613 extern int filemap_fdatawrite(struct address_space *);
2614 extern int filemap_flush(struct address_space *);
2615 extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2616 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2619 static inline int filemap_fdatawait(struct address_space *mapping)
2621 return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2624 extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2626 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2628 extern int filemap_write_and_wait(struct address_space *mapping);
2629 extern int filemap_write_and_wait_range(struct address_space *mapping,
2630 loff_t lstart, loff_t lend);
2631 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2632 loff_t start, loff_t end, int sync_mode);
2633 extern int filemap_fdatawrite_range(struct address_space *mapping,
2634 loff_t start, loff_t end);
2635 extern int filemap_check_errors(struct address_space *mapping);
2636 extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2638 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2640 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2641 extern int __must_check file_write_and_wait_range(struct file *file,
2642 loff_t start, loff_t end);
2644 static inline int file_write_and_wait(struct file *file)
2646 return file_write_and_wait_range(file, 0, LLONG_MAX);
2650 * filemap_set_wb_err - set a writeback error on an address_space
2651 * @mapping: mapping in which to set writeback error
2652 * @err: error to be set in mapping
2654 * When writeback fails in some way, we must record that error so that
2655 * userspace can be informed when fsync and the like are called. We endeavor
2656 * to report errors on any file that was open at the time of the error. Some
2657 * internal callers also need to know when writeback errors have occurred.
2659 * When a writeback error occurs, most filesystems will want to call
2660 * filemap_set_wb_err to record the error in the mapping so that it will be
2661 * automatically reported whenever fsync is called on the file.
2663 static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2665 /* Fastpath for common case of no error */
2667 __filemap_set_wb_err(mapping, err);
2671 * filemap_check_wb_error - has an error occurred since the mark was sampled?
2672 * @mapping: mapping to check for writeback errors
2673 * @since: previously-sampled errseq_t
2675 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2676 * the given value was sampled.
2678 * If it has then report the latest error set, otherwise return 0.
2680 static inline int filemap_check_wb_err(struct address_space *mapping,
2683 return errseq_check(&mapping->wb_err, since);
2687 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2688 * @mapping: mapping to be sampled
2690 * Writeback errors are always reported relative to a particular sample point
2691 * in the past. This function provides those sample points.
2693 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2695 return errseq_sample(&mapping->wb_err);
2698 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2700 extern int vfs_fsync(struct file *file, int datasync);
2703 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2704 * to already be updated for the write, and will return either the amount
2705 * of bytes passed in, or an error if syncing the file failed.
2707 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2709 if (iocb->ki_flags & IOCB_DSYNC) {
2710 int ret = vfs_fsync_range(iocb->ki_filp,
2711 iocb->ki_pos - count, iocb->ki_pos - 1,
2712 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2720 extern void emergency_sync(void);
2721 extern void emergency_remount(void);
2723 extern sector_t bmap(struct inode *, sector_t);
2725 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2726 extern int inode_permission(struct inode *, int);
2727 extern int generic_permission(struct inode *, int);
2728 extern int __check_sticky(struct inode *dir, struct inode *inode);
2730 static inline bool execute_ok(struct inode *inode)
2732 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2735 static inline void file_start_write(struct file *file)
2737 if (!S_ISREG(file_inode(file)->i_mode))
2739 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2742 static inline bool file_start_write_trylock(struct file *file)
2744 if (!S_ISREG(file_inode(file)->i_mode))
2746 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2749 static inline void file_end_write(struct file *file)
2751 if (!S_ISREG(file_inode(file)->i_mode))
2753 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2756 static inline int do_clone_file_range(struct file *file_in, loff_t pos_in,
2757 struct file *file_out, loff_t pos_out,
2762 file_start_write(file_out);
2763 ret = vfs_clone_file_range(file_in, pos_in, file_out, pos_out, len);
2764 file_end_write(file_out);
2770 * get_write_access() gets write permission for a file.
2771 * put_write_access() releases this write permission.
2772 * This is used for regular files.
2773 * We cannot support write (and maybe mmap read-write shared) accesses and
2774 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2775 * can have the following values:
2776 * 0: no writers, no VM_DENYWRITE mappings
2777 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2778 * > 0: (i_writecount) users are writing to the file.
2780 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2781 * except for the cases where we don't hold i_writecount yet. Then we need to
2782 * use {get,deny}_write_access() - these functions check the sign and refuse
2783 * to do the change if sign is wrong.
2785 static inline int get_write_access(struct inode *inode)
2787 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2789 static inline int deny_write_access(struct file *file)
2791 struct inode *inode = file_inode(file);
2792 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2794 static inline void put_write_access(struct inode * inode)
2796 atomic_dec(&inode->i_writecount);
2798 static inline void allow_write_access(struct file *file)
2801 atomic_inc(&file_inode(file)->i_writecount);
2803 static inline bool inode_is_open_for_write(const struct inode *inode)
2805 return atomic_read(&inode->i_writecount) > 0;
2809 static inline void i_readcount_dec(struct inode *inode)
2811 BUG_ON(!atomic_read(&inode->i_readcount));
2812 atomic_dec(&inode->i_readcount);
2814 static inline void i_readcount_inc(struct inode *inode)
2816 atomic_inc(&inode->i_readcount);
2819 static inline void i_readcount_dec(struct inode *inode)
2823 static inline void i_readcount_inc(struct inode *inode)
2828 extern int do_pipe_flags(int *, int);
2830 #define __kernel_read_file_id(id) \
2831 id(UNKNOWN, unknown) \
2832 id(FIRMWARE, firmware) \
2833 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2834 id(MODULE, kernel-module) \
2835 id(KEXEC_IMAGE, kexec-image) \
2836 id(KEXEC_INITRAMFS, kexec-initramfs) \
2837 id(POLICY, security-policy) \
2838 id(X509_CERTIFICATE, x509-certificate) \
2841 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2842 #define __fid_stringify(dummy, str) #str,
2844 enum kernel_read_file_id {
2845 __kernel_read_file_id(__fid_enumify)
2848 static const char * const kernel_read_file_str[] = {
2849 __kernel_read_file_id(__fid_stringify)
2852 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
2854 if ((unsigned)id >= READING_MAX_ID)
2855 return kernel_read_file_str[READING_UNKNOWN];
2857 return kernel_read_file_str[id];
2860 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2861 enum kernel_read_file_id);
2862 extern int kernel_read_file_from_path(const char *, void **, loff_t *, loff_t,
2863 enum kernel_read_file_id);
2864 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2865 enum kernel_read_file_id);
2866 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
2867 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
2868 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
2869 extern struct file * open_exec(const char *);
2871 /* fs/dcache.c -- generic fs support functions */
2872 extern bool is_subdir(struct dentry *, struct dentry *);
2873 extern bool path_is_under(const struct path *, const struct path *);
2875 extern char *file_path(struct file *, char *, int);
2877 #include <linux/err.h>
2879 /* needed for stackable file system support */
2880 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2882 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2884 extern int inode_init_always(struct super_block *, struct inode *);
2885 extern void inode_init_once(struct inode *);
2886 extern void address_space_init_once(struct address_space *mapping);
2887 extern struct inode * igrab(struct inode *);
2888 extern ino_t iunique(struct super_block *, ino_t);
2889 extern int inode_needs_sync(struct inode *inode);
2890 extern int generic_delete_inode(struct inode *inode);
2891 static inline int generic_drop_inode(struct inode *inode)
2893 return !inode->i_nlink || inode_unhashed(inode);
2896 extern struct inode *ilookup5_nowait(struct super_block *sb,
2897 unsigned long hashval, int (*test)(struct inode *, void *),
2899 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2900 int (*test)(struct inode *, void *), void *data);
2901 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2903 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
2904 int (*test)(struct inode *, void *),
2905 int (*set)(struct inode *, void *),
2907 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2908 extern struct inode * iget_locked(struct super_block *, unsigned long);
2909 extern struct inode *find_inode_nowait(struct super_block *,
2911 int (*match)(struct inode *,
2912 unsigned long, void *),
2914 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2915 extern int insert_inode_locked(struct inode *);
2916 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2917 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2919 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2921 extern void unlock_new_inode(struct inode *);
2922 extern unsigned int get_next_ino(void);
2923 extern void evict_inodes(struct super_block *sb);
2925 extern void __iget(struct inode * inode);
2926 extern void iget_failed(struct inode *);
2927 extern void clear_inode(struct inode *);
2928 extern void __destroy_inode(struct inode *);
2929 extern struct inode *new_inode_pseudo(struct super_block *sb);
2930 extern struct inode *new_inode(struct super_block *sb);
2931 extern void free_inode_nonrcu(struct inode *inode);
2932 extern int should_remove_suid(struct dentry *);
2933 extern int file_remove_privs(struct file *);
2935 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2936 static inline void insert_inode_hash(struct inode *inode)
2938 __insert_inode_hash(inode, inode->i_ino);
2941 extern void __remove_inode_hash(struct inode *);
2942 static inline void remove_inode_hash(struct inode *inode)
2944 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2945 __remove_inode_hash(inode);
2948 extern void inode_sb_list_add(struct inode *inode);
2951 extern int bdev_read_only(struct block_device *);
2953 extern int set_blocksize(struct block_device *, int);
2954 extern int sb_set_blocksize(struct super_block *, int);
2955 extern int sb_min_blocksize(struct super_block *, int);
2957 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2958 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2959 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2960 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2961 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2962 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2963 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
2964 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2966 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
2968 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
2971 /* fs/block_dev.c */
2972 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2973 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2974 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2976 extern void block_sync_page(struct page *page);
2979 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2980 struct pipe_inode_info *, size_t, unsigned int);
2981 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2982 struct file *, loff_t *, size_t, unsigned int);
2983 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2984 struct file *out, loff_t *, size_t len, unsigned int flags);
2985 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2986 loff_t *opos, size_t len, unsigned int flags);
2990 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2991 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2992 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2993 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2994 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2995 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2996 int whence, loff_t maxsize, loff_t eof);
2997 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2998 int whence, loff_t size);
2999 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3000 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3001 extern int generic_file_open(struct inode * inode, struct file * filp);
3002 extern int nonseekable_open(struct inode * inode, struct file * filp);
3005 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3006 loff_t file_offset);
3009 /* need locking between buffered and direct access */
3012 /* filesystem does not support filling holes */
3013 DIO_SKIP_HOLES = 0x02,
3016 void dio_end_io(struct bio *bio);
3017 void dio_warn_stale_pagecache(struct file *filp);
3019 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3020 struct block_device *bdev, struct iov_iter *iter,
3021 get_block_t get_block,
3022 dio_iodone_t end_io, dio_submit_t submit_io,
3025 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3026 struct inode *inode,
3027 struct iov_iter *iter,
3028 get_block_t get_block)
3030 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3031 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3035 void inode_dio_wait(struct inode *inode);
3038 * inode_dio_begin - signal start of a direct I/O requests
3039 * @inode: inode the direct I/O happens on
3041 * This is called once we've finished processing a direct I/O request,
3042 * and is used to wake up callers waiting for direct I/O to be quiesced.
3044 static inline void inode_dio_begin(struct inode *inode)
3046 atomic_inc(&inode->i_dio_count);
3050 * inode_dio_end - signal finish of a direct I/O requests
3051 * @inode: inode the direct I/O happens on
3053 * This is called once we've finished processing a direct I/O request,
3054 * and is used to wake up callers waiting for direct I/O to be quiesced.
3056 static inline void inode_dio_end(struct inode *inode)
3058 if (atomic_dec_and_test(&inode->i_dio_count))
3059 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3062 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3065 extern const struct file_operations generic_ro_fops;
3067 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3069 extern int readlink_copy(char __user *, int, const char *);
3070 extern int page_readlink(struct dentry *, char __user *, int);
3071 extern const char *page_get_link(struct dentry *, struct inode *,
3072 struct delayed_call *);
3073 extern void page_put_link(void *);
3074 extern int __page_symlink(struct inode *inode, const char *symname, int len,
3076 extern int page_symlink(struct inode *inode, const char *symname, int len);
3077 extern const struct inode_operations page_symlink_inode_operations;
3078 extern void kfree_link(void *);
3079 extern void generic_fillattr(struct inode *, struct kstat *);
3080 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3081 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3082 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3083 void inode_add_bytes(struct inode *inode, loff_t bytes);
3084 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3085 void inode_sub_bytes(struct inode *inode, loff_t bytes);
3086 static inline loff_t __inode_get_bytes(struct inode *inode)
3088 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3090 loff_t inode_get_bytes(struct inode *inode);
3091 void inode_set_bytes(struct inode *inode, loff_t bytes);
3092 const char *simple_get_link(struct dentry *, struct inode *,
3093 struct delayed_call *);
3094 extern const struct inode_operations simple_symlink_inode_operations;
3096 extern int iterate_dir(struct file *, struct dir_context *);
3098 extern int vfs_statx(int, const char __user *, int, struct kstat *, u32);
3099 extern int vfs_statx_fd(unsigned int, struct kstat *, u32, unsigned int);
3101 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3103 return vfs_statx(AT_FDCWD, filename, AT_NO_AUTOMOUNT,
3104 stat, STATX_BASIC_STATS);
3106 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3108 return vfs_statx(AT_FDCWD, name, AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT,
3109 stat, STATX_BASIC_STATS);
3111 static inline int vfs_fstatat(int dfd, const char __user *filename,
3112 struct kstat *stat, int flags)
3114 return vfs_statx(dfd, filename, flags | AT_NO_AUTOMOUNT,
3115 stat, STATX_BASIC_STATS);
3117 static inline int vfs_fstat(int fd, struct kstat *stat)
3119 return vfs_statx_fd(fd, stat, STATX_BASIC_STATS, 0);
3123 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3124 extern int vfs_readlink(struct dentry *, char __user *, int);
3126 extern int __generic_block_fiemap(struct inode *inode,
3127 struct fiemap_extent_info *fieinfo,
3128 loff_t start, loff_t len,
3129 get_block_t *get_block);
3130 extern int generic_block_fiemap(struct inode *inode,
3131 struct fiemap_extent_info *fieinfo, u64 start,
3132 u64 len, get_block_t *get_block);
3134 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3135 extern void put_filesystem(struct file_system_type *fs);
3136 extern struct file_system_type *get_fs_type(const char *name);
3137 extern struct super_block *get_super(struct block_device *);
3138 extern struct super_block *get_super_thawed(struct block_device *);
3139 extern struct super_block *get_super_exclusive_thawed(struct block_device *bdev);
3140 extern struct super_block *get_active_super(struct block_device *bdev);
3141 extern void drop_super(struct super_block *sb);
3142 extern void drop_super_exclusive(struct super_block *sb);
3143 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3144 extern void iterate_supers_type(struct file_system_type *,
3145 void (*)(struct super_block *, void *), void *);
3147 extern int dcache_dir_open(struct inode *, struct file *);
3148 extern int dcache_dir_close(struct inode *, struct file *);
3149 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3150 extern int dcache_readdir(struct file *, struct dir_context *);
3151 extern int simple_setattr(struct dentry *, struct iattr *);
3152 extern int simple_getattr(const struct path *, struct kstat *, u32, unsigned int);
3153 extern int simple_statfs(struct dentry *, struct kstatfs *);
3154 extern int simple_open(struct inode *inode, struct file *file);
3155 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3156 extern int simple_unlink(struct inode *, struct dentry *);
3157 extern int simple_rmdir(struct inode *, struct dentry *);
3158 extern int simple_rename(struct inode *, struct dentry *,
3159 struct inode *, struct dentry *, unsigned int);
3160 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3161 extern int noop_set_page_dirty(struct page *page);
3162 extern void noop_invalidatepage(struct page *page, unsigned int offset,
3163 unsigned int length);
3164 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3165 extern int simple_empty(struct dentry *);
3166 extern int simple_readpage(struct file *file, struct page *page);
3167 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3168 loff_t pos, unsigned len, unsigned flags,
3169 struct page **pagep, void **fsdata);
3170 extern int simple_write_end(struct file *file, struct address_space *mapping,
3171 loff_t pos, unsigned len, unsigned copied,
3172 struct page *page, void *fsdata);
3173 extern int always_delete_dentry(const struct dentry *);
3174 extern struct inode *alloc_anon_inode(struct super_block *);
3175 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3176 extern const struct dentry_operations simple_dentry_operations;
3178 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3179 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3180 extern const struct file_operations simple_dir_operations;
3181 extern const struct inode_operations simple_dir_inode_operations;
3182 extern void make_empty_dir_inode(struct inode *inode);
3183 extern bool is_empty_dir_inode(struct inode *inode);
3184 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3185 struct dentry *d_alloc_name(struct dentry *, const char *);
3186 extern int simple_fill_super(struct super_block *, unsigned long,
3187 const struct tree_descr *);
3188 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3189 extern void simple_release_fs(struct vfsmount **mount, int *count);
3191 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3192 loff_t *ppos, const void *from, size_t available);
3193 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3194 const void __user *from, size_t count);
3196 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3197 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3199 extern int generic_check_addressable(unsigned, u64);
3201 #ifdef CONFIG_MIGRATION
3202 extern int buffer_migrate_page(struct address_space *,
3203 struct page *, struct page *,
3206 #define buffer_migrate_page NULL
3209 extern int setattr_prepare(struct dentry *, struct iattr *);
3210 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3211 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
3213 extern int file_update_time(struct file *file);
3215 static inline bool io_is_direct(struct file *filp)
3217 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
3220 static inline bool vma_is_dax(struct vm_area_struct *vma)
3222 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3225 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3227 struct inode *inode;
3231 if (!vma_is_dax(vma))
3233 inode = file_inode(vma->vm_file);
3234 if (S_ISCHR(inode->i_mode))
3235 return false; /* device-dax */
3239 static inline int iocb_flags(struct file *file)
3242 if (file->f_flags & O_APPEND)
3244 if (io_is_direct(file))
3246 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3248 if (file->f_flags & __O_SYNC)
3253 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3255 if (unlikely(flags & ~RWF_SUPPORTED))
3258 if (flags & RWF_NOWAIT) {
3259 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3261 ki->ki_flags |= IOCB_NOWAIT;
3263 if (flags & RWF_HIPRI)
3264 ki->ki_flags |= IOCB_HIPRI;
3265 if (flags & RWF_DSYNC)
3266 ki->ki_flags |= IOCB_DSYNC;
3267 if (flags & RWF_SYNC)
3268 ki->ki_flags |= (IOCB_DSYNC | IOCB_SYNC);
3269 if (flags & RWF_APPEND)
3270 ki->ki_flags |= IOCB_APPEND;
3274 static inline ino_t parent_ino(struct dentry *dentry)
3279 * Don't strictly need d_lock here? If the parent ino could change
3280 * then surely we'd have a deeper race in the caller?
3282 spin_lock(&dentry->d_lock);
3283 res = dentry->d_parent->d_inode->i_ino;
3284 spin_unlock(&dentry->d_lock);
3288 /* Transaction based IO helpers */
3291 * An argresp is stored in an allocated page and holds the
3292 * size of the argument or response, along with its content
3294 struct simple_transaction_argresp {
3299 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3301 char *simple_transaction_get(struct file *file, const char __user *buf,
3303 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3304 size_t size, loff_t *pos);
3305 int simple_transaction_release(struct inode *inode, struct file *file);
3307 void simple_transaction_set(struct file *file, size_t n);
3310 * simple attribute files
3312 * These attributes behave similar to those in sysfs:
3314 * Writing to an attribute immediately sets a value, an open file can be
3315 * written to multiple times.
3317 * Reading from an attribute creates a buffer from the value that might get
3318 * read with multiple read calls. When the attribute has been read
3319 * completely, no further read calls are possible until the file is opened
3322 * All attributes contain a text representation of a numeric value
3323 * that are accessed with the get() and set() functions.
3325 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3326 static int __fops ## _open(struct inode *inode, struct file *file) \
3328 __simple_attr_check_format(__fmt, 0ull); \
3329 return simple_attr_open(inode, file, __get, __set, __fmt); \
3331 static const struct file_operations __fops = { \
3332 .owner = THIS_MODULE, \
3333 .open = __fops ## _open, \
3334 .release = simple_attr_release, \
3335 .read = simple_attr_read, \
3336 .write = simple_attr_write, \
3337 .llseek = generic_file_llseek, \
3340 static inline __printf(1, 2)
3341 void __simple_attr_check_format(const char *fmt, ...)
3343 /* don't do anything, just let the compiler check the arguments; */
3346 int simple_attr_open(struct inode *inode, struct file *file,
3347 int (*get)(void *, u64 *), int (*set)(void *, u64),
3349 int simple_attr_release(struct inode *inode, struct file *file);
3350 ssize_t simple_attr_read(struct file *file, char __user *buf,
3351 size_t len, loff_t *ppos);
3352 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3353 size_t len, loff_t *ppos);
3356 int proc_nr_files(struct ctl_table *table, int write,
3357 void __user *buffer, size_t *lenp, loff_t *ppos);
3358 int proc_nr_dentry(struct ctl_table *table, int write,
3359 void __user *buffer, size_t *lenp, loff_t *ppos);
3360 int proc_nr_inodes(struct ctl_table *table, int write,
3361 void __user *buffer, size_t *lenp, loff_t *ppos);
3362 int __init get_filesystem_list(char *buf);
3364 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3365 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3367 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3368 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3369 (flag & __FMODE_NONOTIFY)))
3371 static inline bool is_sxid(umode_t mode)
3373 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3376 static inline int check_sticky(struct inode *dir, struct inode *inode)
3378 if (!(dir->i_mode & S_ISVTX))
3381 return __check_sticky(dir, inode);
3384 static inline void inode_has_no_xattr(struct inode *inode)
3386 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3387 inode->i_flags |= S_NOSEC;
3390 static inline bool is_root_inode(struct inode *inode)
3392 return inode == inode->i_sb->s_root->d_inode;
3395 static inline bool dir_emit(struct dir_context *ctx,
3396 const char *name, int namelen,
3397 u64 ino, unsigned type)
3399 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3401 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3403 return ctx->actor(ctx, ".", 1, ctx->pos,
3404 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3406 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3408 return ctx->actor(ctx, "..", 2, ctx->pos,
3409 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3411 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3413 if (ctx->pos == 0) {
3414 if (!dir_emit_dot(file, ctx))
3418 if (ctx->pos == 1) {
3419 if (!dir_emit_dotdot(file, ctx))
3425 static inline bool dir_relax(struct inode *inode)
3427 inode_unlock(inode);
3429 return !IS_DEADDIR(inode);
3432 static inline bool dir_relax_shared(struct inode *inode)
3434 inode_unlock_shared(inode);
3435 inode_lock_shared(inode);
3436 return !IS_DEADDIR(inode);
3439 extern bool path_noexec(const struct path *path);
3440 extern void inode_nohighmem(struct inode *inode);
3442 #endif /* _LINUX_FS_H */