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/rbtree.h>
17 #include <linux/init.h>
18 #include <linux/pid.h>
19 #include <linux/bug.h>
20 #include <linux/mutex.h>
21 #include <linux/rwsem.h>
22 #include <linux/mm_types.h>
23 #include <linux/capability.h>
24 #include <linux/semaphore.h>
25 #include <linux/fcntl.h>
26 #include <linux/fiemap.h>
27 #include <linux/rculist_bl.h>
28 #include <linux/atomic.h>
29 #include <linux/shrinker.h>
30 #include <linux/migrate_mode.h>
31 #include <linux/uidgid.h>
32 #include <linux/lockdep.h>
33 #include <linux/percpu-rwsem.h>
34 #include <linux/workqueue.h>
35 #include <linux/delayed_call.h>
36 #include <linux/uuid.h>
37 #include <linux/errseq.h>
39 #include <asm/byteorder.h>
40 #include <uapi/linux/fs.h>
42 struct backing_dev_info;
45 struct export_operations;
50 struct pipe_inode_info;
51 struct poll_table_struct;
53 struct vm_area_struct;
56 struct swap_info_struct;
58 struct workqueue_struct;
61 struct fscrypt_operations;
63 extern void __init inode_init(void);
64 extern void __init inode_init_early(void);
65 extern void __init files_init(void);
66 extern void __init files_maxfiles_init(void);
68 extern struct files_stat_struct files_stat;
69 extern unsigned long get_max_files(void);
70 extern unsigned int sysctl_nr_open;
71 extern struct inodes_stat_t inodes_stat;
72 extern int leases_enable, lease_break_time;
73 extern int sysctl_protected_symlinks;
74 extern int sysctl_protected_hardlinks;
76 typedef __kernel_rwf_t rwf_t;
79 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
80 struct buffer_head *bh_result, int create);
81 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
82 ssize_t bytes, void *private);
84 #define MAY_EXEC 0x00000001
85 #define MAY_WRITE 0x00000002
86 #define MAY_READ 0x00000004
87 #define MAY_APPEND 0x00000008
88 #define MAY_ACCESS 0x00000010
89 #define MAY_OPEN 0x00000020
90 #define MAY_CHDIR 0x00000040
91 /* called from RCU mode, don't block */
92 #define MAY_NOT_BLOCK 0x00000080
95 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
96 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
99 /* file is open for reading */
100 #define FMODE_READ ((__force fmode_t)0x1)
101 /* file is open for writing */
102 #define FMODE_WRITE ((__force fmode_t)0x2)
103 /* file is seekable */
104 #define FMODE_LSEEK ((__force fmode_t)0x4)
105 /* file can be accessed using pread */
106 #define FMODE_PREAD ((__force fmode_t)0x8)
107 /* file can be accessed using pwrite */
108 #define FMODE_PWRITE ((__force fmode_t)0x10)
109 /* File is opened for execution with sys_execve / sys_uselib */
110 #define FMODE_EXEC ((__force fmode_t)0x20)
111 /* File is opened with O_NDELAY (only set for block devices) */
112 #define FMODE_NDELAY ((__force fmode_t)0x40)
113 /* File is opened with O_EXCL (only set for block devices) */
114 #define FMODE_EXCL ((__force fmode_t)0x80)
115 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
116 (specialy hack for floppy.c) */
117 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
118 /* 32bit hashes as llseek() offset (for directories) */
119 #define FMODE_32BITHASH ((__force fmode_t)0x200)
120 /* 64bit hashes as llseek() offset (for directories) */
121 #define FMODE_64BITHASH ((__force fmode_t)0x400)
124 * Don't update ctime and mtime.
126 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
127 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
129 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
131 /* Expect random access pattern */
132 #define FMODE_RANDOM ((__force fmode_t)0x1000)
134 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
135 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
137 /* File is opened with O_PATH; almost nothing can be done with it */
138 #define FMODE_PATH ((__force fmode_t)0x4000)
140 /* File needs atomic accesses to f_pos */
141 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
142 /* Write access to underlying fs */
143 #define FMODE_WRITER ((__force fmode_t)0x10000)
144 /* Has read method(s) */
145 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
146 /* Has write method(s) */
147 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
149 /* File was opened by fanotify and shouldn't generate fanotify events */
150 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
152 /* File is capable of returning -EAGAIN if I/O will block */
153 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
156 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
157 * that indicates that they should check the contents of the iovec are
158 * valid, but not check the memory that the iovec elements
161 #define CHECK_IOVEC_ONLY -1
164 * Attribute flags. These should be or-ed together to figure out what
167 #define ATTR_MODE (1 << 0)
168 #define ATTR_UID (1 << 1)
169 #define ATTR_GID (1 << 2)
170 #define ATTR_SIZE (1 << 3)
171 #define ATTR_ATIME (1 << 4)
172 #define ATTR_MTIME (1 << 5)
173 #define ATTR_CTIME (1 << 6)
174 #define ATTR_ATIME_SET (1 << 7)
175 #define ATTR_MTIME_SET (1 << 8)
176 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
177 #define ATTR_ATTR_FLAG (1 << 10)
178 #define ATTR_KILL_SUID (1 << 11)
179 #define ATTR_KILL_SGID (1 << 12)
180 #define ATTR_FILE (1 << 13)
181 #define ATTR_KILL_PRIV (1 << 14)
182 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
183 #define ATTR_TIMES_SET (1 << 16)
184 #define ATTR_TOUCH (1 << 17)
187 * Whiteout is represented by a char device. The following constants define the
188 * mode and device number to use.
190 #define WHITEOUT_MODE 0
191 #define WHITEOUT_DEV 0
194 * This is the Inode Attributes structure, used for notify_change(). It
195 * uses the above definitions as flags, to know which values have changed.
196 * Also, in this manner, a Filesystem can look at only the values it cares
197 * about. Basically, these are the attributes that the VFS layer can
198 * request to change from the FS layer.
200 * Derek Atkins <warlord@MIT.EDU> 94-10-20
203 unsigned int ia_valid;
208 struct timespec ia_atime;
209 struct timespec ia_mtime;
210 struct timespec ia_ctime;
213 * Not an attribute, but an auxiliary info for filesystems wanting to
214 * implement an ftruncate() like method. NOTE: filesystem should
215 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
217 struct file *ia_file;
221 * Includes for diskquotas.
223 #include <linux/quota.h>
226 * Maximum number of layers of fs stack. Needs to be limited to
227 * prevent kernel stack overflow
229 #define FILESYSTEM_MAX_STACK_DEPTH 2
232 * enum positive_aop_returns - aop return codes with specific semantics
234 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
235 * completed, that the page is still locked, and
236 * should be considered active. The VM uses this hint
237 * to return the page to the active list -- it won't
238 * be a candidate for writeback again in the near
239 * future. Other callers must be careful to unlock
240 * the page if they get this return. Returned by
243 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
244 * unlocked it and the page might have been truncated.
245 * The caller should back up to acquiring a new page and
246 * trying again. The aop will be taking reasonable
247 * precautions not to livelock. If the caller held a page
248 * reference, it should drop it before retrying. Returned
251 * address_space_operation functions return these large constants to indicate
252 * special semantics to the caller. These are much larger than the bytes in a
253 * page to allow for functions that return the number of bytes operated on in a
257 enum positive_aop_returns {
258 AOP_WRITEPAGE_ACTIVATE = 0x80000,
259 AOP_TRUNCATED_PAGE = 0x80001,
262 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
263 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
264 * helper code (eg buffer layer)
265 * to clear GFP_FS from alloc */
268 * oh the beauties of C type declarations.
271 struct address_space;
272 struct writeback_control;
275 * Write life time hint values.
278 WRITE_LIFE_NOT_SET = 0,
279 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
280 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
281 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
282 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
283 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
286 #define IOCB_EVENTFD (1 << 0)
287 #define IOCB_APPEND (1 << 1)
288 #define IOCB_DIRECT (1 << 2)
289 #define IOCB_HIPRI (1 << 3)
290 #define IOCB_DSYNC (1 << 4)
291 #define IOCB_SYNC (1 << 5)
292 #define IOCB_WRITE (1 << 6)
293 #define IOCB_NOWAIT (1 << 7)
296 struct file *ki_filp;
298 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
301 enum rw_hint ki_hint;
302 } __randomize_layout;
304 static inline bool is_sync_kiocb(struct kiocb *kiocb)
306 return kiocb->ki_complete == NULL;
310 * "descriptor" for what we're up to with a read.
311 * This allows us to use the same read code yet
312 * have multiple different users of the data that
313 * we read from a file.
315 * The simplest case just copies the data to user
328 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
329 unsigned long, unsigned long);
331 struct address_space_operations {
332 int (*writepage)(struct page *page, struct writeback_control *wbc);
333 int (*readpage)(struct file *, struct page *);
335 /* Write back some dirty pages from this mapping. */
336 int (*writepages)(struct address_space *, struct writeback_control *);
338 /* Set a page dirty. Return true if this dirtied it */
339 int (*set_page_dirty)(struct page *page);
341 int (*readpages)(struct file *filp, struct address_space *mapping,
342 struct list_head *pages, unsigned nr_pages);
344 int (*write_begin)(struct file *, struct address_space *mapping,
345 loff_t pos, unsigned len, unsigned flags,
346 struct page **pagep, void **fsdata);
347 int (*write_end)(struct file *, struct address_space *mapping,
348 loff_t pos, unsigned len, unsigned copied,
349 struct page *page, void *fsdata);
351 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
352 sector_t (*bmap)(struct address_space *, sector_t);
353 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
354 int (*releasepage) (struct page *, gfp_t);
355 void (*freepage)(struct page *);
356 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
358 * migrate the contents of a page to the specified target. If
359 * migrate_mode is MIGRATE_ASYNC, it must not block.
361 int (*migratepage) (struct address_space *,
362 struct page *, struct page *, enum migrate_mode);
363 bool (*isolate_page)(struct page *, isolate_mode_t);
364 void (*putback_page)(struct page *);
365 int (*launder_page) (struct page *);
366 int (*is_partially_uptodate) (struct page *, unsigned long,
368 void (*is_dirty_writeback) (struct page *, bool *, bool *);
369 int (*error_remove_page)(struct address_space *, struct page *);
371 /* swapfile support */
372 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
374 void (*swap_deactivate)(struct file *file);
377 extern const struct address_space_operations empty_aops;
380 * pagecache_write_begin/pagecache_write_end must be used by general code
381 * to write into the pagecache.
383 int pagecache_write_begin(struct file *, struct address_space *mapping,
384 loff_t pos, unsigned len, unsigned flags,
385 struct page **pagep, void **fsdata);
387 int pagecache_write_end(struct file *, struct address_space *mapping,
388 loff_t pos, unsigned len, unsigned copied,
389 struct page *page, void *fsdata);
391 struct address_space {
392 struct inode *host; /* owner: inode, block_device */
393 struct radix_tree_root page_tree; /* radix tree of all pages */
394 spinlock_t tree_lock; /* and lock protecting it */
395 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
396 struct rb_root_cached i_mmap; /* tree of private and shared mappings */
397 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
398 /* Protected by tree_lock together with the radix tree */
399 unsigned long nrpages; /* number of total pages */
400 /* number of shadow or DAX exceptional entries */
401 unsigned long nrexceptional;
402 pgoff_t writeback_index;/* writeback starts here */
403 const struct address_space_operations *a_ops; /* methods */
404 unsigned long flags; /* error bits */
405 spinlock_t private_lock; /* for use by the address_space */
406 gfp_t gfp_mask; /* implicit gfp mask for allocations */
407 struct list_head private_list; /* for use by the address_space */
408 void *private_data; /* ditto */
410 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
412 * On most architectures that alignment is already the case; but
413 * must be enforced here for CRIS, to let the least significant bit
414 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
416 struct request_queue;
418 struct block_device {
419 dev_t bd_dev; /* not a kdev_t - it's a search key */
421 struct inode * bd_inode; /* will die */
422 struct super_block * bd_super;
423 struct mutex bd_mutex; /* open/close mutex */
427 bool bd_write_holder;
429 struct list_head bd_holder_disks;
431 struct block_device * bd_contains;
432 unsigned bd_block_size;
434 struct hd_struct * bd_part;
435 /* number of times partitions within this device have been opened. */
436 unsigned bd_part_count;
438 struct gendisk * bd_disk;
439 struct request_queue * bd_queue;
440 struct backing_dev_info *bd_bdi;
441 struct list_head bd_list;
443 * Private data. You must have bd_claim'ed the block_device
444 * to use this. NOTE: bd_claim allows an owner to claim
445 * the same device multiple times, the owner must take special
446 * care to not mess up bd_private for that case.
448 unsigned long bd_private;
450 /* The counter of freeze processes */
451 int bd_fsfreeze_count;
452 /* Mutex for freeze */
453 struct mutex bd_fsfreeze_mutex;
454 } __randomize_layout;
457 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
460 #define PAGECACHE_TAG_DIRTY 0
461 #define PAGECACHE_TAG_WRITEBACK 1
462 #define PAGECACHE_TAG_TOWRITE 2
464 int mapping_tagged(struct address_space *mapping, int tag);
466 static inline void i_mmap_lock_write(struct address_space *mapping)
468 down_write(&mapping->i_mmap_rwsem);
471 static inline void i_mmap_unlock_write(struct address_space *mapping)
473 up_write(&mapping->i_mmap_rwsem);
476 static inline void i_mmap_lock_read(struct address_space *mapping)
478 down_read(&mapping->i_mmap_rwsem);
481 static inline void i_mmap_unlock_read(struct address_space *mapping)
483 up_read(&mapping->i_mmap_rwsem);
487 * Might pages of this file be mapped into userspace?
489 static inline int mapping_mapped(struct address_space *mapping)
491 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
495 * Might pages of this file have been modified in userspace?
496 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
497 * marks vma as VM_SHARED if it is shared, and the file was opened for
498 * writing i.e. vma may be mprotected writable even if now readonly.
500 * If i_mmap_writable is negative, no new writable mappings are allowed. You
501 * can only deny writable mappings, if none exists right now.
503 static inline int mapping_writably_mapped(struct address_space *mapping)
505 return atomic_read(&mapping->i_mmap_writable) > 0;
508 static inline int mapping_map_writable(struct address_space *mapping)
510 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
514 static inline void mapping_unmap_writable(struct address_space *mapping)
516 atomic_dec(&mapping->i_mmap_writable);
519 static inline int mapping_deny_writable(struct address_space *mapping)
521 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
525 static inline void mapping_allow_writable(struct address_space *mapping)
527 atomic_inc(&mapping->i_mmap_writable);
531 * Use sequence counter to get consistent i_size on 32-bit processors.
533 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
534 #include <linux/seqlock.h>
535 #define __NEED_I_SIZE_ORDERED
536 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
538 #define i_size_ordered_init(inode) do { } while (0)
542 #define ACL_NOT_CACHED ((void *)(-1))
543 #define ACL_DONT_CACHE ((void *)(-3))
545 static inline struct posix_acl *
546 uncached_acl_sentinel(struct task_struct *task)
548 return (void *)task + 1;
552 is_uncached_acl(struct posix_acl *acl)
554 return (long)acl & 1;
557 #define IOP_FASTPERM 0x0001
558 #define IOP_LOOKUP 0x0002
559 #define IOP_NOFOLLOW 0x0004
560 #define IOP_XATTR 0x0008
561 #define IOP_DEFAULT_READLINK 0x0010
563 struct fsnotify_mark_connector;
566 * Keep mostly read-only and often accessed (especially for
567 * the RCU path lookup and 'stat' data) fields at the beginning
568 * of the 'struct inode'
572 unsigned short i_opflags;
575 unsigned int i_flags;
577 #ifdef CONFIG_FS_POSIX_ACL
578 struct posix_acl *i_acl;
579 struct posix_acl *i_default_acl;
582 const struct inode_operations *i_op;
583 struct super_block *i_sb;
584 struct address_space *i_mapping;
586 #ifdef CONFIG_SECURITY
590 /* Stat data, not accessed from path walking */
593 * Filesystems may only read i_nlink directly. They shall use the
594 * following functions for modification:
596 * (set|clear|inc|drop)_nlink
597 * inode_(inc|dec)_link_count
600 const unsigned int i_nlink;
601 unsigned int __i_nlink;
605 struct timespec i_atime;
606 struct timespec i_mtime;
607 struct timespec i_ctime;
608 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
609 unsigned short i_bytes;
610 unsigned int i_blkbits;
611 enum rw_hint i_write_hint;
614 #ifdef __NEED_I_SIZE_ORDERED
615 seqcount_t i_size_seqcount;
619 unsigned long i_state;
620 struct rw_semaphore i_rwsem;
622 unsigned long dirtied_when; /* jiffies of first dirtying */
623 unsigned long dirtied_time_when;
625 struct hlist_node i_hash;
626 struct list_head i_io_list; /* backing dev IO list */
627 #ifdef CONFIG_CGROUP_WRITEBACK
628 struct bdi_writeback *i_wb; /* the associated cgroup wb */
630 /* foreign inode detection, see wbc_detach_inode() */
632 u16 i_wb_frn_avg_time;
633 u16 i_wb_frn_history;
635 struct list_head i_lru; /* inode LRU list */
636 struct list_head i_sb_list;
637 struct list_head i_wb_list; /* backing dev writeback list */
639 struct hlist_head i_dentry;
640 struct rcu_head i_rcu;
644 atomic_t i_dio_count;
645 atomic_t i_writecount;
647 atomic_t i_readcount; /* struct files open RO */
649 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
650 struct file_lock_context *i_flctx;
651 struct address_space i_data;
652 struct list_head i_devices;
654 struct pipe_inode_info *i_pipe;
655 struct block_device *i_bdev;
663 #ifdef CONFIG_FSNOTIFY
664 __u32 i_fsnotify_mask; /* all events this inode cares about */
665 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
668 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
669 struct fscrypt_info *i_crypt_info;
672 void *i_private; /* fs or device private pointer */
673 } __randomize_layout;
675 static inline unsigned int i_blocksize(const struct inode *node)
677 return (1 << node->i_blkbits);
680 static inline int inode_unhashed(struct inode *inode)
682 return hlist_unhashed(&inode->i_hash);
686 * inode->i_mutex nesting subclasses for the lock validator:
688 * 0: the object of the current VFS operation
692 * 4: second non-directory
693 * 5: second parent (when locking independent directories in rename)
695 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
696 * non-directories at once.
698 * The locking order between these classes is
699 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
701 enum inode_i_mutex_lock_class
711 static inline void inode_lock(struct inode *inode)
713 down_write(&inode->i_rwsem);
716 static inline void inode_unlock(struct inode *inode)
718 up_write(&inode->i_rwsem);
721 static inline void inode_lock_shared(struct inode *inode)
723 down_read(&inode->i_rwsem);
726 static inline void inode_unlock_shared(struct inode *inode)
728 up_read(&inode->i_rwsem);
731 static inline int inode_trylock(struct inode *inode)
733 return down_write_trylock(&inode->i_rwsem);
736 static inline int inode_trylock_shared(struct inode *inode)
738 return down_read_trylock(&inode->i_rwsem);
741 static inline int inode_is_locked(struct inode *inode)
743 return rwsem_is_locked(&inode->i_rwsem);
746 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
748 down_write_nested(&inode->i_rwsem, subclass);
751 void lock_two_nondirectories(struct inode *, struct inode*);
752 void unlock_two_nondirectories(struct inode *, struct inode*);
755 * NOTE: in a 32bit arch with a preemptable kernel and
756 * an UP compile the i_size_read/write must be atomic
757 * with respect to the local cpu (unlike with preempt disabled),
758 * but they don't need to be atomic with respect to other cpus like in
759 * true SMP (so they need either to either locally disable irq around
760 * the read or for example on x86 they can be still implemented as a
761 * cmpxchg8b without the need of the lock prefix). For SMP compiles
762 * and 64bit archs it makes no difference if preempt is enabled or not.
764 static inline loff_t i_size_read(const struct inode *inode)
766 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
771 seq = read_seqcount_begin(&inode->i_size_seqcount);
772 i_size = inode->i_size;
773 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
775 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
779 i_size = inode->i_size;
783 return inode->i_size;
788 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
789 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
790 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
792 static inline void i_size_write(struct inode *inode, loff_t i_size)
794 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
796 write_seqcount_begin(&inode->i_size_seqcount);
797 inode->i_size = i_size;
798 write_seqcount_end(&inode->i_size_seqcount);
800 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
802 inode->i_size = i_size;
805 inode->i_size = i_size;
809 static inline unsigned iminor(const struct inode *inode)
811 return MINOR(inode->i_rdev);
814 static inline unsigned imajor(const struct inode *inode)
816 return MAJOR(inode->i_rdev);
819 extern struct block_device *I_BDEV(struct inode *inode);
822 rwlock_t lock; /* protects pid, uid, euid fields */
823 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
824 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
825 kuid_t uid, euid; /* uid/euid of process setting the owner */
826 int signum; /* posix.1b rt signal to be delivered on IO */
830 * Track a single file's readahead state
832 struct file_ra_state {
833 pgoff_t start; /* where readahead started */
834 unsigned int size; /* # of readahead pages */
835 unsigned int async_size; /* do asynchronous readahead when
836 there are only # of pages ahead */
838 unsigned int ra_pages; /* Maximum readahead window */
839 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
840 loff_t prev_pos; /* Cache last read() position */
844 * Check if @index falls in the readahead windows.
846 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
848 return (index >= ra->start &&
849 index < ra->start + ra->size);
854 struct llist_node fu_llist;
855 struct rcu_head fu_rcuhead;
858 struct inode *f_inode; /* cached value */
859 const struct file_operations *f_op;
862 * Protects f_ep_links, f_flags.
863 * Must not be taken from IRQ context.
866 enum rw_hint f_write_hint;
867 atomic_long_t f_count;
868 unsigned int f_flags;
870 struct mutex f_pos_lock;
872 struct fown_struct f_owner;
873 const struct cred *f_cred;
874 struct file_ra_state f_ra;
877 #ifdef CONFIG_SECURITY
880 /* needed for tty driver, and maybe others */
884 /* Used by fs/eventpoll.c to link all the hooks to this file */
885 struct list_head f_ep_links;
886 struct list_head f_tfile_llink;
887 #endif /* #ifdef CONFIG_EPOLL */
888 struct address_space *f_mapping;
891 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
896 /* file identifier */
897 unsigned char f_handle[0];
900 static inline struct file *get_file(struct file *f)
902 atomic_long_inc(&f->f_count);
905 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
906 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
907 #define file_count(x) atomic_long_read(&(x)->f_count)
909 #define MAX_NON_LFS ((1UL<<31) - 1)
911 /* Page cache limit. The filesystems should put that into their s_maxbytes
912 limits, otherwise bad things can happen in VM. */
913 #if BITS_PER_LONG==32
914 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
915 #elif BITS_PER_LONG==64
916 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
921 #define FL_DELEG 4 /* NFSv4 delegation */
922 #define FL_ACCESS 8 /* not trying to lock, just looking */
923 #define FL_EXISTS 16 /* when unlocking, test for existence */
924 #define FL_LEASE 32 /* lease held on this file */
925 #define FL_CLOSE 64 /* unlock on close */
926 #define FL_SLEEP 128 /* A blocking lock */
927 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
928 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
929 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
930 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
932 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
935 * Special return value from posix_lock_file() and vfs_lock_file() for
936 * asynchronous locking.
938 #define FILE_LOCK_DEFERRED 1
940 /* legacy typedef, should eventually be removed */
941 typedef void *fl_owner_t;
945 struct file_lock_operations {
946 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
947 void (*fl_release_private)(struct file_lock *);
950 struct lock_manager_operations {
951 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
952 unsigned long (*lm_owner_key)(struct file_lock *);
953 fl_owner_t (*lm_get_owner)(fl_owner_t);
954 void (*lm_put_owner)(fl_owner_t);
955 void (*lm_notify)(struct file_lock *); /* unblock callback */
956 int (*lm_grant)(struct file_lock *, int);
957 bool (*lm_break)(struct file_lock *);
958 int (*lm_change)(struct file_lock *, int, struct list_head *);
959 void (*lm_setup)(struct file_lock *, void **);
962 struct lock_manager {
963 struct list_head list;
965 * NFSv4 and up also want opens blocked during the grace period;
972 void locks_start_grace(struct net *, struct lock_manager *);
973 void locks_end_grace(struct lock_manager *);
974 int locks_in_grace(struct net *);
975 int opens_in_grace(struct net *);
977 /* that will die - we need it for nfs_lock_info */
978 #include <linux/nfs_fs_i.h>
981 * struct file_lock represents a generic "file lock". It's used to represent
982 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
983 * note that the same struct is used to represent both a request for a lock and
984 * the lock itself, but the same object is never used for both.
986 * FIXME: should we create a separate "struct lock_request" to help distinguish
989 * The varous i_flctx lists are ordered by:
992 * 2) lock range start
995 * Obviously, the last two criteria only matter for POSIX locks.
998 struct file_lock *fl_next; /* singly linked list for this inode */
999 struct list_head fl_list; /* link into file_lock_context */
1000 struct hlist_node fl_link; /* node in global lists */
1001 struct list_head fl_block; /* circular list of blocked processes */
1002 fl_owner_t fl_owner;
1003 unsigned int fl_flags;
1004 unsigned char fl_type;
1005 unsigned int fl_pid;
1006 int fl_link_cpu; /* what cpu's list is this on? */
1007 wait_queue_head_t fl_wait;
1008 struct file *fl_file;
1012 struct fasync_struct * fl_fasync; /* for lease break notifications */
1013 /* for lease breaks: */
1014 unsigned long fl_break_time;
1015 unsigned long fl_downgrade_time;
1017 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1018 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1020 struct nfs_lock_info nfs_fl;
1021 struct nfs4_lock_info nfs4_fl;
1023 struct list_head link; /* link in AFS vnode's pending_locks list */
1024 int state; /* state of grant or error if -ve */
1027 } __randomize_layout;
1029 struct file_lock_context {
1030 spinlock_t flc_lock;
1031 struct list_head flc_flock;
1032 struct list_head flc_posix;
1033 struct list_head flc_lease;
1036 /* The following constant reflects the upper bound of the file/locking space */
1038 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1039 #define OFFSET_MAX INT_LIMIT(loff_t)
1040 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1043 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1046 * Return the inode to use for locking
1048 * For overlayfs this should be the overlay inode, not the real inode returned
1049 * by file_inode(). For any other fs file_inode(filp) and locks_inode(filp) are
1052 static inline struct inode *locks_inode(const struct file *f)
1054 return f->f_path.dentry->d_inode;
1057 #ifdef CONFIG_FILE_LOCKING
1058 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1059 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1062 #if BITS_PER_LONG == 32
1063 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1064 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1068 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1069 extern int fcntl_getlease(struct file *filp);
1072 void locks_free_lock_context(struct inode *inode);
1073 void locks_free_lock(struct file_lock *fl);
1074 extern void locks_init_lock(struct file_lock *);
1075 extern struct file_lock * locks_alloc_lock(void);
1076 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1077 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1078 extern void locks_remove_posix(struct file *, fl_owner_t);
1079 extern void locks_remove_file(struct file *);
1080 extern void locks_release_private(struct file_lock *);
1081 extern void posix_test_lock(struct file *, struct file_lock *);
1082 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1083 extern int posix_unblock_lock(struct file_lock *);
1084 extern int vfs_test_lock(struct file *, struct file_lock *);
1085 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1086 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1087 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1088 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1089 extern void lease_get_mtime(struct inode *, struct timespec *time);
1090 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1091 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1092 extern int lease_modify(struct file_lock *, int, struct list_head *);
1093 struct files_struct;
1094 extern void show_fd_locks(struct seq_file *f,
1095 struct file *filp, struct files_struct *files);
1096 #else /* !CONFIG_FILE_LOCKING */
1097 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1098 struct flock __user *user)
1103 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1104 unsigned int cmd, struct flock __user *user)
1109 #if BITS_PER_LONG == 32
1110 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1111 struct flock64 __user *user)
1116 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1117 unsigned int cmd, struct flock64 __user *user)
1122 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1127 static inline int fcntl_getlease(struct file *filp)
1133 locks_free_lock_context(struct inode *inode)
1137 static inline void locks_init_lock(struct file_lock *fl)
1142 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1147 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1152 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1157 static inline void locks_remove_file(struct file *filp)
1162 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1167 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1168 struct file_lock *conflock)
1173 static inline int posix_unblock_lock(struct file_lock *waiter)
1178 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1183 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1184 struct file_lock *fl, struct file_lock *conf)
1189 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1194 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1199 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1204 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1209 static inline int generic_setlease(struct file *filp, long arg,
1210 struct file_lock **flp, void **priv)
1215 static inline int vfs_setlease(struct file *filp, long arg,
1216 struct file_lock **lease, void **priv)
1221 static inline int lease_modify(struct file_lock *fl, int arg,
1222 struct list_head *dispose)
1227 struct files_struct;
1228 static inline void show_fd_locks(struct seq_file *f,
1229 struct file *filp, struct files_struct *files) {}
1230 #endif /* !CONFIG_FILE_LOCKING */
1232 static inline struct inode *file_inode(const struct file *f)
1237 static inline struct dentry *file_dentry(const struct file *file)
1239 return d_real(file->f_path.dentry, file_inode(file), 0, 0);
1242 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1244 return locks_lock_inode_wait(locks_inode(filp), fl);
1247 struct fasync_struct {
1251 struct fasync_struct *fa_next; /* singly linked list */
1252 struct file *fa_file;
1253 struct rcu_head fa_rcu;
1256 #define FASYNC_MAGIC 0x4601
1258 /* SMP safe fasync helpers: */
1259 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1260 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1261 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1262 extern struct fasync_struct *fasync_alloc(void);
1263 extern void fasync_free(struct fasync_struct *);
1265 /* can be called from interrupts */
1266 extern void kill_fasync(struct fasync_struct **, int, int);
1268 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1269 extern int f_setown(struct file *filp, unsigned long arg, int force);
1270 extern void f_delown(struct file *filp);
1271 extern pid_t f_getown(struct file *filp);
1272 extern int send_sigurg(struct fown_struct *fown);
1275 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1276 * represented in both.
1278 #define SB_RDONLY 1 /* Mount read-only */
1279 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1280 #define SB_NODEV 4 /* Disallow access to device special files */
1281 #define SB_NOEXEC 8 /* Disallow program execution */
1282 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1283 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1284 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1285 #define SB_NOATIME 1024 /* Do not update access times. */
1286 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1287 #define SB_SILENT 32768
1288 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1289 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1290 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1291 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1293 /* These sb flags are internal to the kernel */
1294 #define SB_SUBMOUNT (1<<26)
1295 #define SB_NOREMOTELOCK (1<<27)
1296 #define SB_NOSEC (1<<28)
1297 #define SB_BORN (1<<29)
1298 #define SB_ACTIVE (1<<30)
1299 #define SB_NOUSER (1<<31)
1305 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1306 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1307 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1308 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1309 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1312 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1313 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1314 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1316 /* sb->s_iflags to limit user namespace mounts */
1317 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1319 /* Possible states of 'frozen' field */
1321 SB_UNFROZEN = 0, /* FS is unfrozen */
1322 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1323 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1324 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1325 * internal threads if needed) */
1326 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1329 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1332 int frozen; /* Is sb frozen? */
1333 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1334 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1337 struct super_block {
1338 struct list_head s_list; /* Keep this first */
1339 dev_t s_dev; /* search index; _not_ kdev_t */
1340 unsigned char s_blocksize_bits;
1341 unsigned long s_blocksize;
1342 loff_t s_maxbytes; /* Max file size */
1343 struct file_system_type *s_type;
1344 const struct super_operations *s_op;
1345 const struct dquot_operations *dq_op;
1346 const struct quotactl_ops *s_qcop;
1347 const struct export_operations *s_export_op;
1348 unsigned long s_flags;
1349 unsigned long s_iflags; /* internal SB_I_* flags */
1350 unsigned long s_magic;
1351 struct dentry *s_root;
1352 struct rw_semaphore s_umount;
1355 #ifdef CONFIG_SECURITY
1358 const struct xattr_handler **s_xattr;
1360 const struct fscrypt_operations *s_cop;
1362 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1363 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1364 struct block_device *s_bdev;
1365 struct backing_dev_info *s_bdi;
1366 struct mtd_info *s_mtd;
1367 struct hlist_node s_instances;
1368 unsigned int s_quota_types; /* Bitmask of supported quota types */
1369 struct quota_info s_dquot; /* Diskquota specific options */
1371 struct sb_writers s_writers;
1373 char s_id[32]; /* Informational name */
1374 uuid_t s_uuid; /* UUID */
1376 void *s_fs_info; /* Filesystem private info */
1377 unsigned int s_max_links;
1380 /* Granularity of c/m/atime in ns.
1381 Cannot be worse than a second */
1385 * The next field is for VFS *only*. No filesystems have any business
1386 * even looking at it. You had been warned.
1388 struct mutex s_vfs_rename_mutex; /* Kludge */
1391 * Filesystem subtype. If non-empty the filesystem type field
1392 * in /proc/mounts will be "type.subtype"
1396 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1399 * Saved pool identifier for cleancache (-1 means none)
1401 int cleancache_poolid;
1403 struct shrinker s_shrink; /* per-sb shrinker handle */
1405 /* Number of inodes with nlink == 0 but still referenced */
1406 atomic_long_t s_remove_count;
1408 /* Being remounted read-only */
1409 int s_readonly_remount;
1411 /* AIO completions deferred from interrupt context */
1412 struct workqueue_struct *s_dio_done_wq;
1413 struct hlist_head s_pins;
1416 * Owning user namespace and default context in which to
1417 * interpret filesystem uids, gids, quotas, device nodes,
1418 * xattrs and security labels.
1420 struct user_namespace *s_user_ns;
1423 * Keep the lru lists last in the structure so they always sit on their
1424 * own individual cachelines.
1426 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1427 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1428 struct rcu_head rcu;
1429 struct work_struct destroy_work;
1431 struct mutex s_sync_lock; /* sync serialisation lock */
1434 * Indicates how deep in a filesystem stack this SB is
1438 /* s_inode_list_lock protects s_inodes */
1439 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1440 struct list_head s_inodes; /* all inodes */
1442 spinlock_t s_inode_wblist_lock;
1443 struct list_head s_inodes_wb; /* writeback inodes */
1444 } __randomize_layout;
1446 /* Helper functions so that in most cases filesystems will
1447 * not need to deal directly with kuid_t and kgid_t and can
1448 * instead deal with the raw numeric values that are stored
1449 * in the filesystem.
1451 static inline uid_t i_uid_read(const struct inode *inode)
1453 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1456 static inline gid_t i_gid_read(const struct inode *inode)
1458 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1461 static inline void i_uid_write(struct inode *inode, uid_t uid)
1463 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1466 static inline void i_gid_write(struct inode *inode, gid_t gid)
1468 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1471 extern struct timespec current_time(struct inode *inode);
1474 * Snapshotting support.
1477 void __sb_end_write(struct super_block *sb, int level);
1478 int __sb_start_write(struct super_block *sb, int level, bool wait);
1480 #define __sb_writers_acquired(sb, lev) \
1481 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1482 #define __sb_writers_release(sb, lev) \
1483 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1486 * sb_end_write - drop write access to a superblock
1487 * @sb: the super we wrote to
1489 * Decrement number of writers to the filesystem. Wake up possible waiters
1490 * wanting to freeze the filesystem.
1492 static inline void sb_end_write(struct super_block *sb)
1494 __sb_end_write(sb, SB_FREEZE_WRITE);
1498 * sb_end_pagefault - drop write access to a superblock from a page fault
1499 * @sb: the super we wrote to
1501 * Decrement number of processes handling write page fault to the filesystem.
1502 * Wake up possible waiters wanting to freeze the filesystem.
1504 static inline void sb_end_pagefault(struct super_block *sb)
1506 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1510 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1511 * @sb: the super we wrote to
1513 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1514 * waiters wanting to freeze the filesystem.
1516 static inline void sb_end_intwrite(struct super_block *sb)
1518 __sb_end_write(sb, SB_FREEZE_FS);
1522 * sb_start_write - get write access to a superblock
1523 * @sb: the super we write to
1525 * When a process wants to write data or metadata to a file system (i.e. dirty
1526 * a page or an inode), it should embed the operation in a sb_start_write() -
1527 * sb_end_write() pair to get exclusion against file system freezing. This
1528 * function increments number of writers preventing freezing. If the file
1529 * system is already frozen, the function waits until the file system is
1532 * Since freeze protection behaves as a lock, users have to preserve
1533 * ordering of freeze protection and other filesystem locks. Generally,
1534 * freeze protection should be the outermost lock. In particular, we have:
1537 * -> i_mutex (write path, truncate, directory ops, ...)
1538 * -> s_umount (freeze_super, thaw_super)
1540 static inline void sb_start_write(struct super_block *sb)
1542 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1545 static inline int sb_start_write_trylock(struct super_block *sb)
1547 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1551 * sb_start_pagefault - get write access to a superblock from a page fault
1552 * @sb: the super we write to
1554 * When a process starts handling write page fault, it should embed the
1555 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1556 * exclusion against file system freezing. This is needed since the page fault
1557 * is going to dirty a page. This function increments number of running page
1558 * faults preventing freezing. If the file system is already frozen, the
1559 * function waits until the file system is thawed.
1561 * Since page fault freeze protection behaves as a lock, users have to preserve
1562 * ordering of freeze protection and other filesystem locks. It is advised to
1563 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1564 * handling code implies lock dependency:
1567 * -> sb_start_pagefault
1569 static inline void sb_start_pagefault(struct super_block *sb)
1571 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1575 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1576 * @sb: the super we write to
1578 * This is the third level of protection against filesystem freezing. It is
1579 * free for use by a filesystem. The only requirement is that it must rank
1580 * below sb_start_pagefault.
1582 * For example filesystem can call sb_start_intwrite() when starting a
1583 * transaction which somewhat eases handling of freezing for internal sources
1584 * of filesystem changes (internal fs threads, discarding preallocation on file
1587 static inline void sb_start_intwrite(struct super_block *sb)
1589 __sb_start_write(sb, SB_FREEZE_FS, true);
1593 extern bool inode_owner_or_capable(const struct inode *inode);
1596 * VFS helper functions..
1598 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1599 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1600 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1601 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1602 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1603 extern int vfs_rmdir(struct inode *, struct dentry *);
1604 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1605 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1606 extern int vfs_whiteout(struct inode *, struct dentry *);
1608 extern struct dentry *vfs_tmpfile(struct dentry *dentry, umode_t mode,
1612 * VFS file helper functions.
1614 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1616 extern bool may_open_dev(const struct path *path);
1618 * VFS FS_IOC_FIEMAP helper definitions.
1620 struct fiemap_extent_info {
1621 unsigned int fi_flags; /* Flags as passed from user */
1622 unsigned int fi_extents_mapped; /* Number of mapped extents */
1623 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1624 struct fiemap_extent __user *fi_extents_start; /* Start of
1625 fiemap_extent array */
1627 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1628 u64 phys, u64 len, u32 flags);
1629 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1634 * NOTE! These match bits 12..15 of stat.st_mode
1635 * (ie "(i_mode >> 12) & 15").
1637 #define DT_UNKNOWN 0
1648 * This is the "filldir" function type, used by readdir() to let
1649 * the kernel specify what kind of dirent layout it wants to have.
1650 * This allows the kernel to read directories into kernel space or
1651 * to have different dirent layouts depending on the binary type.
1654 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1657 struct dir_context {
1658 const filldir_t actor;
1662 struct block_device_operations;
1664 /* These macros are for out of kernel modules to test that
1665 * the kernel supports the unlocked_ioctl and compat_ioctl
1666 * fields in struct file_operations. */
1667 #define HAVE_COMPAT_IOCTL 1
1668 #define HAVE_UNLOCKED_IOCTL 1
1671 * These flags let !MMU mmap() govern direct device mapping vs immediate
1672 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1674 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1675 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1676 * NOMMU_MAP_READ: Can be mapped for reading
1677 * NOMMU_MAP_WRITE: Can be mapped for writing
1678 * NOMMU_MAP_EXEC: Can be mapped for execution
1680 #define NOMMU_MAP_COPY 0x00000001
1681 #define NOMMU_MAP_DIRECT 0x00000008
1682 #define NOMMU_MAP_READ VM_MAYREAD
1683 #define NOMMU_MAP_WRITE VM_MAYWRITE
1684 #define NOMMU_MAP_EXEC VM_MAYEXEC
1686 #define NOMMU_VMFLAGS \
1687 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1692 struct file_operations {
1693 struct module *owner;
1694 loff_t (*llseek) (struct file *, loff_t, int);
1695 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1696 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1697 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1698 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1699 int (*iterate) (struct file *, struct dir_context *);
1700 int (*iterate_shared) (struct file *, struct dir_context *);
1701 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1702 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1703 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1704 int (*mmap) (struct file *, struct vm_area_struct *);
1705 int (*open) (struct inode *, struct file *);
1706 int (*flush) (struct file *, fl_owner_t id);
1707 int (*release) (struct inode *, struct file *);
1708 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1709 int (*fasync) (int, struct file *, int);
1710 int (*lock) (struct file *, int, struct file_lock *);
1711 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1712 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1713 int (*check_flags)(int);
1714 int (*flock) (struct file *, int, struct file_lock *);
1715 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1716 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1717 int (*setlease)(struct file *, long, struct file_lock **, void **);
1718 long (*fallocate)(struct file *file, int mode, loff_t offset,
1720 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1722 unsigned (*mmap_capabilities)(struct file *);
1724 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1725 loff_t, size_t, unsigned int);
1726 int (*clone_file_range)(struct file *, loff_t, struct file *, loff_t,
1728 ssize_t (*dedupe_file_range)(struct file *, u64, u64, struct file *,
1730 } __randomize_layout;
1732 struct inode_operations {
1733 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1734 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1735 int (*permission) (struct inode *, int);
1736 struct posix_acl * (*get_acl)(struct inode *, int);
1738 int (*readlink) (struct dentry *, char __user *,int);
1740 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1741 int (*link) (struct dentry *,struct inode *,struct dentry *);
1742 int (*unlink) (struct inode *,struct dentry *);
1743 int (*symlink) (struct inode *,struct dentry *,const char *);
1744 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1745 int (*rmdir) (struct inode *,struct dentry *);
1746 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1747 int (*rename) (struct inode *, struct dentry *,
1748 struct inode *, struct dentry *, unsigned int);
1749 int (*setattr) (struct dentry *, struct iattr *);
1750 int (*getattr) (const struct path *, struct kstat *, u32, unsigned int);
1751 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1752 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1754 int (*update_time)(struct inode *, struct timespec *, int);
1755 int (*atomic_open)(struct inode *, struct dentry *,
1756 struct file *, unsigned open_flag,
1757 umode_t create_mode, int *opened);
1758 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1759 int (*set_acl)(struct inode *, struct posix_acl *, int);
1760 } ____cacheline_aligned;
1762 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
1763 struct iov_iter *iter)
1765 return file->f_op->read_iter(kio, iter);
1768 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
1769 struct iov_iter *iter)
1771 return file->f_op->write_iter(kio, iter);
1774 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
1776 return file->f_op->mmap(file, vma);
1779 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1780 unsigned long nr_segs, unsigned long fast_segs,
1781 struct iovec *fast_pointer,
1782 struct iovec **ret_pointer);
1784 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1785 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1786 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1787 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1788 unsigned long, loff_t *, rwf_t);
1789 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1790 loff_t, size_t, unsigned int);
1791 extern int vfs_clone_file_prep_inodes(struct inode *inode_in, loff_t pos_in,
1792 struct inode *inode_out, loff_t pos_out,
1793 u64 *len, bool is_dedupe);
1794 extern int vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1795 struct file *file_out, loff_t pos_out, u64 len);
1796 extern int vfs_dedupe_file_range_compare(struct inode *src, loff_t srcoff,
1797 struct inode *dest, loff_t destoff,
1798 loff_t len, bool *is_same);
1799 extern int vfs_dedupe_file_range(struct file *file,
1800 struct file_dedupe_range *same);
1802 struct super_operations {
1803 struct inode *(*alloc_inode)(struct super_block *sb);
1804 void (*destroy_inode)(struct inode *);
1806 void (*dirty_inode) (struct inode *, int flags);
1807 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1808 int (*drop_inode) (struct inode *);
1809 void (*evict_inode) (struct inode *);
1810 void (*put_super) (struct super_block *);
1811 int (*sync_fs)(struct super_block *sb, int wait);
1812 int (*freeze_super) (struct super_block *);
1813 int (*freeze_fs) (struct super_block *);
1814 int (*thaw_super) (struct super_block *);
1815 int (*unfreeze_fs) (struct super_block *);
1816 int (*statfs) (struct dentry *, struct kstatfs *);
1817 int (*remount_fs) (struct super_block *, int *, char *);
1818 void (*umount_begin) (struct super_block *);
1820 int (*show_options)(struct seq_file *, struct dentry *);
1821 int (*show_devname)(struct seq_file *, struct dentry *);
1822 int (*show_path)(struct seq_file *, struct dentry *);
1823 int (*show_stats)(struct seq_file *, struct dentry *);
1825 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1826 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1827 struct dquot **(*get_dquots)(struct inode *);
1829 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1830 long (*nr_cached_objects)(struct super_block *,
1831 struct shrink_control *);
1832 long (*free_cached_objects)(struct super_block *,
1833 struct shrink_control *);
1837 * Inode flags - they have no relation to superblock flags now
1839 #define S_SYNC 1 /* Writes are synced at once */
1840 #define S_NOATIME 2 /* Do not update access times */
1841 #define S_APPEND 4 /* Append-only file */
1842 #define S_IMMUTABLE 8 /* Immutable file */
1843 #define S_DEAD 16 /* removed, but still open directory */
1844 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1845 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1846 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1847 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1848 #define S_PRIVATE 512 /* Inode is fs-internal */
1849 #define S_IMA 1024 /* Inode has an associated IMA struct */
1850 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1851 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1852 #ifdef CONFIG_FS_DAX
1853 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1855 #define S_DAX 0 /* Make all the DAX code disappear */
1859 * Note that nosuid etc flags are inode-specific: setting some file-system
1860 * flags just means all the inodes inherit those flags by default. It might be
1861 * possible to override it selectively if you really wanted to with some
1862 * ioctl() that is not currently implemented.
1864 * Exception: SB_RDONLY is always applied to the entire file system.
1866 * Unfortunately, it is possible to change a filesystems flags with it mounted
1867 * with files in use. This means that all of the inodes will not have their
1868 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1869 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1871 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1873 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & MS_RDONLY; }
1874 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
1875 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
1876 ((inode)->i_flags & S_SYNC))
1877 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
1878 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1879 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
1880 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
1881 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
1883 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1884 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1885 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1886 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
1888 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1889 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1890 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1891 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1892 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1893 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1894 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1895 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1897 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1898 (inode)->i_rdev == WHITEOUT_DEV)
1900 static inline bool HAS_UNMAPPED_ID(struct inode *inode)
1902 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
1905 static inline enum rw_hint file_write_hint(struct file *file)
1907 if (file->f_write_hint != WRITE_LIFE_NOT_SET)
1908 return file->f_write_hint;
1910 return file_inode(file)->i_write_hint;
1913 static inline int iocb_flags(struct file *file);
1915 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
1917 *kiocb = (struct kiocb) {
1919 .ki_flags = iocb_flags(filp),
1920 .ki_hint = file_write_hint(filp),
1925 * Inode state bits. Protected by inode->i_lock
1927 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1928 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1930 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1931 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1932 * various stages of removing an inode.
1934 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1936 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1937 * fdatasync(). i_atime is the usual cause.
1938 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1939 * these changes separately from I_DIRTY_SYNC so that we
1940 * don't have to write inode on fdatasync() when only
1941 * mtime has changed in it.
1942 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1943 * I_NEW Serves as both a mutex and completion notification.
1944 * New inodes set I_NEW. If two processes both create
1945 * the same inode, one of them will release its inode and
1946 * wait for I_NEW to be released before returning.
1947 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1948 * also cause waiting on I_NEW, without I_NEW actually
1949 * being set. find_inode() uses this to prevent returning
1950 * nearly-dead inodes.
1951 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1952 * is zero. I_FREEING must be set when I_WILL_FREE is
1954 * I_FREEING Set when inode is about to be freed but still has dirty
1955 * pages or buffers attached or the inode itself is still
1957 * I_CLEAR Added by clear_inode(). In this state the inode is
1958 * clean and can be destroyed. Inode keeps I_FREEING.
1960 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1961 * prohibited for many purposes. iget() must wait for
1962 * the inode to be completely released, then create it
1963 * anew. Other functions will just ignore such inodes,
1964 * if appropriate. I_NEW is used for waiting.
1966 * I_SYNC Writeback of inode is running. The bit is set during
1967 * data writeback, and cleared with a wakeup on the bit
1968 * address once it is done. The bit is also used to pin
1969 * the inode in memory for flusher thread.
1971 * I_REFERENCED Marks the inode as recently references on the LRU list.
1973 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1975 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1976 * synchronize competing switching instances and to tell
1977 * wb stat updates to grab mapping->tree_lock. See
1978 * inode_switch_wb_work_fn() for details.
1980 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
1981 * and work dirs among overlayfs mounts.
1983 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1985 #define I_DIRTY_SYNC (1 << 0)
1986 #define I_DIRTY_DATASYNC (1 << 1)
1987 #define I_DIRTY_PAGES (1 << 2)
1989 #define I_NEW (1 << __I_NEW)
1990 #define I_WILL_FREE (1 << 4)
1991 #define I_FREEING (1 << 5)
1992 #define I_CLEAR (1 << 6)
1994 #define I_SYNC (1 << __I_SYNC)
1995 #define I_REFERENCED (1 << 8)
1996 #define __I_DIO_WAKEUP 9
1997 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1998 #define I_LINKABLE (1 << 10)
1999 #define I_DIRTY_TIME (1 << 11)
2000 #define __I_DIRTY_TIME_EXPIRED 12
2001 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
2002 #define I_WB_SWITCH (1 << 13)
2003 #define I_OVL_INUSE (1 << 14)
2005 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
2006 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2008 extern void __mark_inode_dirty(struct inode *, int);
2009 static inline void mark_inode_dirty(struct inode *inode)
2011 __mark_inode_dirty(inode, I_DIRTY);
2014 static inline void mark_inode_dirty_sync(struct inode *inode)
2016 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2019 extern void inc_nlink(struct inode *inode);
2020 extern void drop_nlink(struct inode *inode);
2021 extern void clear_nlink(struct inode *inode);
2022 extern void set_nlink(struct inode *inode, unsigned int nlink);
2024 static inline void inode_inc_link_count(struct inode *inode)
2027 mark_inode_dirty(inode);
2030 static inline void inode_dec_link_count(struct inode *inode)
2033 mark_inode_dirty(inode);
2037 * inode_inc_iversion - increments i_version
2038 * @inode: inode that need to be updated
2040 * Every time the inode is modified, the i_version field will be incremented.
2041 * The filesystem has to be mounted with i_version flag
2044 static inline void inode_inc_iversion(struct inode *inode)
2046 spin_lock(&inode->i_lock);
2048 spin_unlock(&inode->i_lock);
2051 enum file_time_flags {
2058 extern void touch_atime(const struct path *);
2059 static inline void file_accessed(struct file *file)
2061 if (!(file->f_flags & O_NOATIME))
2062 touch_atime(&file->f_path);
2065 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2066 int sync_inode_metadata(struct inode *inode, int wait);
2068 struct file_system_type {
2071 #define FS_REQUIRES_DEV 1
2072 #define FS_BINARY_MOUNTDATA 2
2073 #define FS_HAS_SUBTYPE 4
2074 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2075 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2076 struct dentry *(*mount) (struct file_system_type *, int,
2077 const char *, void *);
2078 void (*kill_sb) (struct super_block *);
2079 struct module *owner;
2080 struct file_system_type * next;
2081 struct hlist_head fs_supers;
2083 struct lock_class_key s_lock_key;
2084 struct lock_class_key s_umount_key;
2085 struct lock_class_key s_vfs_rename_key;
2086 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2088 struct lock_class_key i_lock_key;
2089 struct lock_class_key i_mutex_key;
2090 struct lock_class_key i_mutex_dir_key;
2093 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2095 extern struct dentry *mount_ns(struct file_system_type *fs_type,
2096 int flags, void *data, void *ns, struct user_namespace *user_ns,
2097 int (*fill_super)(struct super_block *, void *, int));
2098 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2099 int flags, const char *dev_name, void *data,
2100 int (*fill_super)(struct super_block *, void *, int));
2101 extern struct dentry *mount_single(struct file_system_type *fs_type,
2102 int flags, void *data,
2103 int (*fill_super)(struct super_block *, void *, int));
2104 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2105 int flags, void *data,
2106 int (*fill_super)(struct super_block *, void *, int));
2107 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2108 void generic_shutdown_super(struct super_block *sb);
2109 void kill_block_super(struct super_block *sb);
2110 void kill_anon_super(struct super_block *sb);
2111 void kill_litter_super(struct super_block *sb);
2112 void deactivate_super(struct super_block *sb);
2113 void deactivate_locked_super(struct super_block *sb);
2114 int set_anon_super(struct super_block *s, void *data);
2115 int get_anon_bdev(dev_t *);
2116 void free_anon_bdev(dev_t);
2117 struct super_block *sget_userns(struct file_system_type *type,
2118 int (*test)(struct super_block *,void *),
2119 int (*set)(struct super_block *,void *),
2120 int flags, struct user_namespace *user_ns,
2122 struct super_block *sget(struct file_system_type *type,
2123 int (*test)(struct super_block *,void *),
2124 int (*set)(struct super_block *,void *),
2125 int flags, void *data);
2126 extern struct dentry *mount_pseudo_xattr(struct file_system_type *, char *,
2127 const struct super_operations *ops,
2128 const struct xattr_handler **xattr,
2129 const struct dentry_operations *dops,
2132 static inline struct dentry *
2133 mount_pseudo(struct file_system_type *fs_type, char *name,
2134 const struct super_operations *ops,
2135 const struct dentry_operations *dops, unsigned long magic)
2137 return mount_pseudo_xattr(fs_type, name, ops, NULL, dops, magic);
2140 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2141 #define fops_get(fops) \
2142 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2143 #define fops_put(fops) \
2144 do { if (fops) module_put((fops)->owner); } while(0)
2146 * This one is to be used *ONLY* from ->open() instances.
2147 * fops must be non-NULL, pinned down *and* module dependencies
2148 * should be sufficient to pin the caller down as well.
2150 #define replace_fops(f, fops) \
2152 struct file *__file = (f); \
2153 fops_put(__file->f_op); \
2154 BUG_ON(!(__file->f_op = (fops))); \
2157 extern int register_filesystem(struct file_system_type *);
2158 extern int unregister_filesystem(struct file_system_type *);
2159 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2160 #define kern_mount(type) kern_mount_data(type, NULL)
2161 extern void kern_unmount(struct vfsmount *mnt);
2162 extern int may_umount_tree(struct vfsmount *);
2163 extern int may_umount(struct vfsmount *);
2164 extern long do_mount(const char *, const char __user *,
2165 const char *, unsigned long, void *);
2166 extern struct vfsmount *collect_mounts(const struct path *);
2167 extern void drop_collected_mounts(struct vfsmount *);
2168 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2170 extern int vfs_statfs(const struct path *, struct kstatfs *);
2171 extern int user_statfs(const char __user *, struct kstatfs *);
2172 extern int fd_statfs(int, struct kstatfs *);
2173 extern int vfs_ustat(dev_t, struct kstatfs *);
2174 extern int freeze_super(struct super_block *super);
2175 extern int thaw_super(struct super_block *super);
2176 extern bool our_mnt(struct vfsmount *mnt);
2177 extern __printf(2, 3)
2178 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2179 extern int super_setup_bdi(struct super_block *sb);
2181 extern int current_umask(void);
2183 extern void ihold(struct inode * inode);
2184 extern void iput(struct inode *);
2185 extern int generic_update_time(struct inode *, struct timespec *, int);
2188 extern struct kobject *fs_kobj;
2190 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2192 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2193 extern int locks_mandatory_locked(struct file *);
2194 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2197 * Candidates for mandatory locking have the setgid bit set
2198 * but no group execute bit - an otherwise meaningless combination.
2201 static inline int __mandatory_lock(struct inode *ino)
2203 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2207 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2208 * otherwise these will be advisory locks
2211 static inline int mandatory_lock(struct inode *ino)
2213 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2216 static inline int locks_verify_locked(struct file *file)
2218 if (mandatory_lock(locks_inode(file)))
2219 return locks_mandatory_locked(file);
2223 static inline int locks_verify_truncate(struct inode *inode,
2227 if (!inode->i_flctx || !mandatory_lock(inode))
2230 if (size < inode->i_size) {
2231 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2234 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2239 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2241 static inline int locks_mandatory_locked(struct file *file)
2246 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2247 loff_t start, loff_t end, unsigned char type)
2252 static inline int __mandatory_lock(struct inode *inode)
2257 static inline int mandatory_lock(struct inode *inode)
2262 static inline int locks_verify_locked(struct file *file)
2267 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2273 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2276 #ifdef CONFIG_FILE_LOCKING
2277 static inline int break_lease(struct inode *inode, unsigned int mode)
2280 * Since this check is lockless, we must ensure that any refcounts
2281 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2282 * could end up racing with tasks trying to set a new lease on this
2286 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2287 return __break_lease(inode, mode, FL_LEASE);
2291 static inline int break_deleg(struct inode *inode, unsigned int mode)
2294 * Since this check is lockless, we must ensure that any refcounts
2295 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2296 * could end up racing with tasks trying to set a new lease on this
2300 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2301 return __break_lease(inode, mode, FL_DELEG);
2305 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2309 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2310 if (ret == -EWOULDBLOCK && delegated_inode) {
2311 *delegated_inode = inode;
2317 static inline int break_deleg_wait(struct inode **delegated_inode)
2321 ret = break_deleg(*delegated_inode, O_WRONLY);
2322 iput(*delegated_inode);
2323 *delegated_inode = NULL;
2327 static inline int break_layout(struct inode *inode, bool wait)
2330 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2331 return __break_lease(inode,
2332 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2337 #else /* !CONFIG_FILE_LOCKING */
2338 static inline int break_lease(struct inode *inode, unsigned int mode)
2343 static inline int break_deleg(struct inode *inode, unsigned int mode)
2348 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2353 static inline int break_deleg_wait(struct inode **delegated_inode)
2359 static inline int break_layout(struct inode *inode, bool wait)
2364 #endif /* CONFIG_FILE_LOCKING */
2369 const char *name; /* pointer to actual string */
2370 const __user char *uptr; /* original userland pointer */
2371 struct audit_names *aname;
2376 extern long vfs_truncate(const struct path *, loff_t);
2377 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2379 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2381 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2383 extern struct file *file_open_name(struct filename *, int, umode_t);
2384 extern struct file *filp_open(const char *, int, umode_t);
2385 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2386 const char *, int, umode_t);
2387 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2388 extern int filp_close(struct file *, fl_owner_t id);
2390 extern struct filename *getname_flags(const char __user *, int, int *);
2391 extern struct filename *getname(const char __user *);
2392 extern struct filename *getname_kernel(const char *);
2393 extern void putname(struct filename *name);
2399 extern int finish_open(struct file *file, struct dentry *dentry,
2400 int (*open)(struct inode *, struct file *),
2402 extern int finish_no_open(struct file *file, struct dentry *dentry);
2406 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2409 extern void __init vfs_caches_init_early(void);
2410 extern void __init vfs_caches_init(void);
2412 extern struct kmem_cache *names_cachep;
2414 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2415 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2418 extern int register_blkdev(unsigned int, const char *);
2419 extern void unregister_blkdev(unsigned int, const char *);
2420 extern void bdev_unhash_inode(dev_t dev);
2421 extern struct block_device *bdget(dev_t);
2422 extern struct block_device *bdgrab(struct block_device *bdev);
2423 extern void bd_set_size(struct block_device *, loff_t size);
2424 extern void bd_forget(struct inode *inode);
2425 extern void bdput(struct block_device *);
2426 extern void invalidate_bdev(struct block_device *);
2427 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2428 extern int sync_blockdev(struct block_device *bdev);
2429 extern void kill_bdev(struct block_device *);
2430 extern struct super_block *freeze_bdev(struct block_device *);
2431 extern void emergency_thaw_all(void);
2432 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2433 extern int fsync_bdev(struct block_device *);
2435 extern struct super_block *blockdev_superblock;
2437 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2439 return sb == blockdev_superblock;
2442 static inline void bd_forget(struct inode *inode) {}
2443 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2444 static inline void kill_bdev(struct block_device *bdev) {}
2445 static inline void invalidate_bdev(struct block_device *bdev) {}
2447 static inline struct super_block *freeze_bdev(struct block_device *sb)
2452 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2457 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2461 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2466 extern int sync_filesystem(struct super_block *);
2467 extern const struct file_operations def_blk_fops;
2468 extern const struct file_operations def_chr_fops;
2470 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2471 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2472 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2473 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2474 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2476 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2478 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2479 extern int __blkdev_reread_part(struct block_device *bdev);
2480 extern int blkdev_reread_part(struct block_device *bdev);
2483 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2484 extern void bd_unlink_disk_holder(struct block_device *bdev,
2485 struct gendisk *disk);
2487 static inline int bd_link_disk_holder(struct block_device *bdev,
2488 struct gendisk *disk)
2492 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2493 struct gendisk *disk)
2500 #define CHRDEV_MAJOR_MAX 512
2501 /* Marks the bottom of the first segment of free char majors */
2502 #define CHRDEV_MAJOR_DYN_END 234
2503 /* Marks the top and bottom of the second segment of free char majors */
2504 #define CHRDEV_MAJOR_DYN_EXT_START 511
2505 #define CHRDEV_MAJOR_DYN_EXT_END 384
2507 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2508 extern int register_chrdev_region(dev_t, unsigned, const char *);
2509 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2510 unsigned int count, const char *name,
2511 const struct file_operations *fops);
2512 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2513 unsigned int count, const char *name);
2514 extern void unregister_chrdev_region(dev_t, unsigned);
2515 extern void chrdev_show(struct seq_file *,off_t);
2517 static inline int register_chrdev(unsigned int major, const char *name,
2518 const struct file_operations *fops)
2520 return __register_chrdev(major, 0, 256, name, fops);
2523 static inline void unregister_chrdev(unsigned int major, const char *name)
2525 __unregister_chrdev(major, 0, 256, name);
2528 /* fs/block_dev.c */
2529 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2530 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2533 #define BLKDEV_MAJOR_MAX 512
2534 extern const char *__bdevname(dev_t, char *buffer);
2535 extern const char *bdevname(struct block_device *bdev, char *buffer);
2536 extern struct block_device *lookup_bdev(const char *);
2537 extern void blkdev_show(struct seq_file *,off_t);
2540 #define BLKDEV_MAJOR_MAX 0
2543 extern void init_special_inode(struct inode *, umode_t, dev_t);
2545 /* Invalid inode operations -- fs/bad_inode.c */
2546 extern void make_bad_inode(struct inode *);
2547 extern bool is_bad_inode(struct inode *);
2550 extern void check_disk_size_change(struct gendisk *disk,
2551 struct block_device *bdev);
2552 extern int revalidate_disk(struct gendisk *);
2553 extern int check_disk_change(struct block_device *);
2554 extern int __invalidate_device(struct block_device *, bool);
2555 extern int invalidate_partition(struct gendisk *, int);
2557 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2558 pgoff_t start, pgoff_t end);
2560 static inline void invalidate_remote_inode(struct inode *inode)
2562 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2563 S_ISLNK(inode->i_mode))
2564 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2566 extern int invalidate_inode_pages2(struct address_space *mapping);
2567 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2568 pgoff_t start, pgoff_t end);
2569 extern int write_inode_now(struct inode *, int);
2570 extern int filemap_fdatawrite(struct address_space *);
2571 extern int filemap_flush(struct address_space *);
2572 extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2573 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2576 static inline int filemap_fdatawait(struct address_space *mapping)
2578 return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2581 extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2583 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2585 extern int filemap_write_and_wait(struct address_space *mapping);
2586 extern int filemap_write_and_wait_range(struct address_space *mapping,
2587 loff_t lstart, loff_t lend);
2588 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2589 loff_t start, loff_t end, int sync_mode);
2590 extern int filemap_fdatawrite_range(struct address_space *mapping,
2591 loff_t start, loff_t end);
2592 extern int filemap_check_errors(struct address_space *mapping);
2593 extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2595 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2597 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2598 extern int __must_check file_write_and_wait_range(struct file *file,
2599 loff_t start, loff_t end);
2601 static inline int file_write_and_wait(struct file *file)
2603 return file_write_and_wait_range(file, 0, LLONG_MAX);
2607 * filemap_set_wb_err - set a writeback error on an address_space
2608 * @mapping: mapping in which to set writeback error
2609 * @err: error to be set in mapping
2611 * When writeback fails in some way, we must record that error so that
2612 * userspace can be informed when fsync and the like are called. We endeavor
2613 * to report errors on any file that was open at the time of the error. Some
2614 * internal callers also need to know when writeback errors have occurred.
2616 * When a writeback error occurs, most filesystems will want to call
2617 * filemap_set_wb_err to record the error in the mapping so that it will be
2618 * automatically reported whenever fsync is called on the file.
2620 static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2622 /* Fastpath for common case of no error */
2624 __filemap_set_wb_err(mapping, err);
2628 * filemap_check_wb_error - has an error occurred since the mark was sampled?
2629 * @mapping: mapping to check for writeback errors
2630 * @since: previously-sampled errseq_t
2632 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2633 * the given value was sampled.
2635 * If it has then report the latest error set, otherwise return 0.
2637 static inline int filemap_check_wb_err(struct address_space *mapping,
2640 return errseq_check(&mapping->wb_err, since);
2644 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2645 * @mapping: mapping to be sampled
2647 * Writeback errors are always reported relative to a particular sample point
2648 * in the past. This function provides those sample points.
2650 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2652 return errseq_sample(&mapping->wb_err);
2655 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2657 extern int vfs_fsync(struct file *file, int datasync);
2660 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2661 * to already be updated for the write, and will return either the amount
2662 * of bytes passed in, or an error if syncing the file failed.
2664 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2666 if (iocb->ki_flags & IOCB_DSYNC) {
2667 int ret = vfs_fsync_range(iocb->ki_filp,
2668 iocb->ki_pos - count, iocb->ki_pos - 1,
2669 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2677 extern void emergency_sync(void);
2678 extern void emergency_remount(void);
2680 extern sector_t bmap(struct inode *, sector_t);
2682 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2683 extern int inode_permission(struct inode *, int);
2684 extern int __inode_permission(struct inode *, int);
2685 extern int generic_permission(struct inode *, int);
2686 extern int __check_sticky(struct inode *dir, struct inode *inode);
2688 static inline bool execute_ok(struct inode *inode)
2690 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2693 static inline void file_start_write(struct file *file)
2695 if (!S_ISREG(file_inode(file)->i_mode))
2697 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2700 static inline bool file_start_write_trylock(struct file *file)
2702 if (!S_ISREG(file_inode(file)->i_mode))
2704 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2707 static inline void file_end_write(struct file *file)
2709 if (!S_ISREG(file_inode(file)->i_mode))
2711 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2714 static inline int do_clone_file_range(struct file *file_in, loff_t pos_in,
2715 struct file *file_out, loff_t pos_out,
2720 file_start_write(file_out);
2721 ret = vfs_clone_file_range(file_in, pos_in, file_out, pos_out, len);
2722 file_end_write(file_out);
2728 * get_write_access() gets write permission for a file.
2729 * put_write_access() releases this write permission.
2730 * This is used for regular files.
2731 * We cannot support write (and maybe mmap read-write shared) accesses and
2732 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2733 * can have the following values:
2734 * 0: no writers, no VM_DENYWRITE mappings
2735 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2736 * > 0: (i_writecount) users are writing to the file.
2738 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2739 * except for the cases where we don't hold i_writecount yet. Then we need to
2740 * use {get,deny}_write_access() - these functions check the sign and refuse
2741 * to do the change if sign is wrong.
2743 static inline int get_write_access(struct inode *inode)
2745 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2747 static inline int deny_write_access(struct file *file)
2749 struct inode *inode = file_inode(file);
2750 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2752 static inline void put_write_access(struct inode * inode)
2754 atomic_dec(&inode->i_writecount);
2756 static inline void allow_write_access(struct file *file)
2759 atomic_inc(&file_inode(file)->i_writecount);
2761 static inline bool inode_is_open_for_write(const struct inode *inode)
2763 return atomic_read(&inode->i_writecount) > 0;
2767 static inline void i_readcount_dec(struct inode *inode)
2769 BUG_ON(!atomic_read(&inode->i_readcount));
2770 atomic_dec(&inode->i_readcount);
2772 static inline void i_readcount_inc(struct inode *inode)
2774 atomic_inc(&inode->i_readcount);
2777 static inline void i_readcount_dec(struct inode *inode)
2781 static inline void i_readcount_inc(struct inode *inode)
2786 extern int do_pipe_flags(int *, int);
2788 #define __kernel_read_file_id(id) \
2789 id(UNKNOWN, unknown) \
2790 id(FIRMWARE, firmware) \
2791 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2792 id(MODULE, kernel-module) \
2793 id(KEXEC_IMAGE, kexec-image) \
2794 id(KEXEC_INITRAMFS, kexec-initramfs) \
2795 id(POLICY, security-policy) \
2796 id(X509_CERTIFICATE, x509-certificate) \
2799 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2800 #define __fid_stringify(dummy, str) #str,
2802 enum kernel_read_file_id {
2803 __kernel_read_file_id(__fid_enumify)
2806 static const char * const kernel_read_file_str[] = {
2807 __kernel_read_file_id(__fid_stringify)
2810 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
2812 if ((unsigned)id >= READING_MAX_ID)
2813 return kernel_read_file_str[READING_UNKNOWN];
2815 return kernel_read_file_str[id];
2818 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2819 enum kernel_read_file_id);
2820 extern int kernel_read_file_from_path(const char *, void **, loff_t *, loff_t,
2821 enum kernel_read_file_id);
2822 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2823 enum kernel_read_file_id);
2824 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
2825 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
2826 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
2827 extern struct file * open_exec(const char *);
2829 /* fs/dcache.c -- generic fs support functions */
2830 extern bool is_subdir(struct dentry *, struct dentry *);
2831 extern bool path_is_under(const struct path *, const struct path *);
2833 extern char *file_path(struct file *, char *, int);
2835 #include <linux/err.h>
2837 /* needed for stackable file system support */
2838 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2840 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2842 extern int inode_init_always(struct super_block *, struct inode *);
2843 extern void inode_init_once(struct inode *);
2844 extern void address_space_init_once(struct address_space *mapping);
2845 extern struct inode * igrab(struct inode *);
2846 extern ino_t iunique(struct super_block *, ino_t);
2847 extern int inode_needs_sync(struct inode *inode);
2848 extern int generic_delete_inode(struct inode *inode);
2849 static inline int generic_drop_inode(struct inode *inode)
2851 return !inode->i_nlink || inode_unhashed(inode);
2854 extern struct inode *ilookup5_nowait(struct super_block *sb,
2855 unsigned long hashval, int (*test)(struct inode *, void *),
2857 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2858 int (*test)(struct inode *, void *), void *data);
2859 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2861 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2862 extern struct inode * iget_locked(struct super_block *, unsigned long);
2863 extern struct inode *find_inode_nowait(struct super_block *,
2865 int (*match)(struct inode *,
2866 unsigned long, void *),
2868 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2869 extern int insert_inode_locked(struct inode *);
2870 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2871 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2873 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2875 extern void unlock_new_inode(struct inode *);
2876 extern unsigned int get_next_ino(void);
2877 extern void evict_inodes(struct super_block *sb);
2879 extern void __iget(struct inode * inode);
2880 extern void iget_failed(struct inode *);
2881 extern void clear_inode(struct inode *);
2882 extern void __destroy_inode(struct inode *);
2883 extern struct inode *new_inode_pseudo(struct super_block *sb);
2884 extern struct inode *new_inode(struct super_block *sb);
2885 extern void free_inode_nonrcu(struct inode *inode);
2886 extern int should_remove_suid(struct dentry *);
2887 extern int file_remove_privs(struct file *);
2889 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2890 static inline void insert_inode_hash(struct inode *inode)
2892 __insert_inode_hash(inode, inode->i_ino);
2895 extern void __remove_inode_hash(struct inode *);
2896 static inline void remove_inode_hash(struct inode *inode)
2898 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2899 __remove_inode_hash(inode);
2902 extern void inode_sb_list_add(struct inode *inode);
2905 extern int bdev_read_only(struct block_device *);
2907 extern int set_blocksize(struct block_device *, int);
2908 extern int sb_set_blocksize(struct super_block *, int);
2909 extern int sb_min_blocksize(struct super_block *, int);
2911 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2912 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2913 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2914 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2915 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2916 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2917 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
2918 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2920 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
2922 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
2925 /* fs/block_dev.c */
2926 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2927 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2928 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2930 extern void block_sync_page(struct page *page);
2933 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2934 struct pipe_inode_info *, size_t, unsigned int);
2935 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2936 struct file *, loff_t *, size_t, unsigned int);
2937 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2938 struct file *out, loff_t *, size_t len, unsigned int flags);
2939 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2940 loff_t *opos, size_t len, unsigned int flags);
2944 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2945 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2946 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2947 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2948 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2949 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2950 int whence, loff_t maxsize, loff_t eof);
2951 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2952 int whence, loff_t size);
2953 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
2954 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
2955 extern int generic_file_open(struct inode * inode, struct file * filp);
2956 extern int nonseekable_open(struct inode * inode, struct file * filp);
2959 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
2960 loff_t file_offset);
2963 /* need locking between buffered and direct access */
2966 /* filesystem does not support filling holes */
2967 DIO_SKIP_HOLES = 0x02,
2969 /* filesystem can handle aio writes beyond i_size */
2970 DIO_ASYNC_EXTEND = 0x04,
2972 /* inode/fs/bdev does not need truncate protection */
2973 DIO_SKIP_DIO_COUNT = 0x08,
2976 void dio_end_io(struct bio *bio);
2978 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2979 struct block_device *bdev, struct iov_iter *iter,
2980 get_block_t get_block,
2981 dio_iodone_t end_io, dio_submit_t submit_io,
2984 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2985 struct inode *inode,
2986 struct iov_iter *iter,
2987 get_block_t get_block)
2989 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
2990 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
2994 void inode_dio_wait(struct inode *inode);
2997 * inode_dio_begin - signal start of a direct I/O requests
2998 * @inode: inode the direct I/O happens on
3000 * This is called once we've finished processing a direct I/O request,
3001 * and is used to wake up callers waiting for direct I/O to be quiesced.
3003 static inline void inode_dio_begin(struct inode *inode)
3005 atomic_inc(&inode->i_dio_count);
3009 * inode_dio_end - signal finish of a direct I/O requests
3010 * @inode: inode the direct I/O happens on
3012 * This is called once we've finished processing a direct I/O request,
3013 * and is used to wake up callers waiting for direct I/O to be quiesced.
3015 static inline void inode_dio_end(struct inode *inode)
3017 if (atomic_dec_and_test(&inode->i_dio_count))
3018 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3021 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3024 extern const struct file_operations generic_ro_fops;
3026 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3028 extern int readlink_copy(char __user *, int, const char *);
3029 extern int page_readlink(struct dentry *, char __user *, int);
3030 extern const char *page_get_link(struct dentry *, struct inode *,
3031 struct delayed_call *);
3032 extern void page_put_link(void *);
3033 extern int __page_symlink(struct inode *inode, const char *symname, int len,
3035 extern int page_symlink(struct inode *inode, const char *symname, int len);
3036 extern const struct inode_operations page_symlink_inode_operations;
3037 extern void kfree_link(void *);
3038 extern void generic_fillattr(struct inode *, struct kstat *);
3039 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3040 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3041 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3042 void inode_add_bytes(struct inode *inode, loff_t bytes);
3043 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3044 void inode_sub_bytes(struct inode *inode, loff_t bytes);
3045 static inline loff_t __inode_get_bytes(struct inode *inode)
3047 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3049 loff_t inode_get_bytes(struct inode *inode);
3050 void inode_set_bytes(struct inode *inode, loff_t bytes);
3051 const char *simple_get_link(struct dentry *, struct inode *,
3052 struct delayed_call *);
3053 extern const struct inode_operations simple_symlink_inode_operations;
3055 extern int iterate_dir(struct file *, struct dir_context *);
3057 extern int vfs_statx(int, const char __user *, int, struct kstat *, u32);
3058 extern int vfs_statx_fd(unsigned int, struct kstat *, u32, unsigned int);
3060 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3062 return vfs_statx(AT_FDCWD, filename, AT_NO_AUTOMOUNT,
3063 stat, STATX_BASIC_STATS);
3065 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3067 return vfs_statx(AT_FDCWD, name, AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT,
3068 stat, STATX_BASIC_STATS);
3070 static inline int vfs_fstatat(int dfd, const char __user *filename,
3071 struct kstat *stat, int flags)
3073 return vfs_statx(dfd, filename, flags, stat, STATX_BASIC_STATS);
3075 static inline int vfs_fstat(int fd, struct kstat *stat)
3077 return vfs_statx_fd(fd, stat, STATX_BASIC_STATS, 0);
3081 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3082 extern int vfs_readlink(struct dentry *, char __user *, int);
3084 extern int __generic_block_fiemap(struct inode *inode,
3085 struct fiemap_extent_info *fieinfo,
3086 loff_t start, loff_t len,
3087 get_block_t *get_block);
3088 extern int generic_block_fiemap(struct inode *inode,
3089 struct fiemap_extent_info *fieinfo, u64 start,
3090 u64 len, get_block_t *get_block);
3092 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3093 extern void put_filesystem(struct file_system_type *fs);
3094 extern struct file_system_type *get_fs_type(const char *name);
3095 extern struct super_block *get_super(struct block_device *);
3096 extern struct super_block *get_super_thawed(struct block_device *);
3097 extern struct super_block *get_super_exclusive_thawed(struct block_device *bdev);
3098 extern struct super_block *get_active_super(struct block_device *bdev);
3099 extern void drop_super(struct super_block *sb);
3100 extern void drop_super_exclusive(struct super_block *sb);
3101 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3102 extern void iterate_supers_type(struct file_system_type *,
3103 void (*)(struct super_block *, void *), void *);
3105 extern int dcache_dir_open(struct inode *, struct file *);
3106 extern int dcache_dir_close(struct inode *, struct file *);
3107 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3108 extern int dcache_readdir(struct file *, struct dir_context *);
3109 extern int simple_setattr(struct dentry *, struct iattr *);
3110 extern int simple_getattr(const struct path *, struct kstat *, u32, unsigned int);
3111 extern int simple_statfs(struct dentry *, struct kstatfs *);
3112 extern int simple_open(struct inode *inode, struct file *file);
3113 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3114 extern int simple_unlink(struct inode *, struct dentry *);
3115 extern int simple_rmdir(struct inode *, struct dentry *);
3116 extern int simple_rename(struct inode *, struct dentry *,
3117 struct inode *, struct dentry *, unsigned int);
3118 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3119 extern int simple_empty(struct dentry *);
3120 extern int simple_readpage(struct file *file, struct page *page);
3121 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3122 loff_t pos, unsigned len, unsigned flags,
3123 struct page **pagep, void **fsdata);
3124 extern int simple_write_end(struct file *file, struct address_space *mapping,
3125 loff_t pos, unsigned len, unsigned copied,
3126 struct page *page, void *fsdata);
3127 extern int always_delete_dentry(const struct dentry *);
3128 extern struct inode *alloc_anon_inode(struct super_block *);
3129 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3130 extern const struct dentry_operations simple_dentry_operations;
3132 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3133 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3134 extern const struct file_operations simple_dir_operations;
3135 extern const struct inode_operations simple_dir_inode_operations;
3136 extern void make_empty_dir_inode(struct inode *inode);
3137 extern bool is_empty_dir_inode(struct inode *inode);
3138 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3139 struct dentry *d_alloc_name(struct dentry *, const char *);
3140 extern int simple_fill_super(struct super_block *, unsigned long,
3141 const struct tree_descr *);
3142 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3143 extern void simple_release_fs(struct vfsmount **mount, int *count);
3145 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3146 loff_t *ppos, const void *from, size_t available);
3147 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3148 const void __user *from, size_t count);
3150 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3151 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3153 extern int generic_check_addressable(unsigned, u64);
3155 #ifdef CONFIG_MIGRATION
3156 extern int buffer_migrate_page(struct address_space *,
3157 struct page *, struct page *,
3160 #define buffer_migrate_page NULL
3163 extern int setattr_prepare(struct dentry *, struct iattr *);
3164 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3165 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
3167 extern int file_update_time(struct file *file);
3169 static inline bool io_is_direct(struct file *filp)
3171 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
3174 static inline bool vma_is_dax(struct vm_area_struct *vma)
3176 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3179 static inline int iocb_flags(struct file *file)
3182 if (file->f_flags & O_APPEND)
3184 if (io_is_direct(file))
3186 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3188 if (file->f_flags & __O_SYNC)
3193 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3195 if (unlikely(flags & ~RWF_SUPPORTED))
3198 if (flags & RWF_NOWAIT) {
3199 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3201 ki->ki_flags |= IOCB_NOWAIT;
3203 if (flags & RWF_HIPRI)
3204 ki->ki_flags |= IOCB_HIPRI;
3205 if (flags & RWF_DSYNC)
3206 ki->ki_flags |= IOCB_DSYNC;
3207 if (flags & RWF_SYNC)
3208 ki->ki_flags |= (IOCB_DSYNC | IOCB_SYNC);
3212 static inline ino_t parent_ino(struct dentry *dentry)
3217 * Don't strictly need d_lock here? If the parent ino could change
3218 * then surely we'd have a deeper race in the caller?
3220 spin_lock(&dentry->d_lock);
3221 res = dentry->d_parent->d_inode->i_ino;
3222 spin_unlock(&dentry->d_lock);
3226 /* Transaction based IO helpers */
3229 * An argresp is stored in an allocated page and holds the
3230 * size of the argument or response, along with its content
3232 struct simple_transaction_argresp {
3237 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3239 char *simple_transaction_get(struct file *file, const char __user *buf,
3241 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3242 size_t size, loff_t *pos);
3243 int simple_transaction_release(struct inode *inode, struct file *file);
3245 void simple_transaction_set(struct file *file, size_t n);
3248 * simple attribute files
3250 * These attributes behave similar to those in sysfs:
3252 * Writing to an attribute immediately sets a value, an open file can be
3253 * written to multiple times.
3255 * Reading from an attribute creates a buffer from the value that might get
3256 * read with multiple read calls. When the attribute has been read
3257 * completely, no further read calls are possible until the file is opened
3260 * All attributes contain a text representation of a numeric value
3261 * that are accessed with the get() and set() functions.
3263 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3264 static int __fops ## _open(struct inode *inode, struct file *file) \
3266 __simple_attr_check_format(__fmt, 0ull); \
3267 return simple_attr_open(inode, file, __get, __set, __fmt); \
3269 static const struct file_operations __fops = { \
3270 .owner = THIS_MODULE, \
3271 .open = __fops ## _open, \
3272 .release = simple_attr_release, \
3273 .read = simple_attr_read, \
3274 .write = simple_attr_write, \
3275 .llseek = generic_file_llseek, \
3278 static inline __printf(1, 2)
3279 void __simple_attr_check_format(const char *fmt, ...)
3281 /* don't do anything, just let the compiler check the arguments; */
3284 int simple_attr_open(struct inode *inode, struct file *file,
3285 int (*get)(void *, u64 *), int (*set)(void *, u64),
3287 int simple_attr_release(struct inode *inode, struct file *file);
3288 ssize_t simple_attr_read(struct file *file, char __user *buf,
3289 size_t len, loff_t *ppos);
3290 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3291 size_t len, loff_t *ppos);
3294 int proc_nr_files(struct ctl_table *table, int write,
3295 void __user *buffer, size_t *lenp, loff_t *ppos);
3296 int proc_nr_dentry(struct ctl_table *table, int write,
3297 void __user *buffer, size_t *lenp, loff_t *ppos);
3298 int proc_nr_inodes(struct ctl_table *table, int write,
3299 void __user *buffer, size_t *lenp, loff_t *ppos);
3300 int __init get_filesystem_list(char *buf);
3302 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3303 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3305 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3306 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3307 (flag & __FMODE_NONOTIFY)))
3309 static inline bool is_sxid(umode_t mode)
3311 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3314 static inline int check_sticky(struct inode *dir, struct inode *inode)
3316 if (!(dir->i_mode & S_ISVTX))
3319 return __check_sticky(dir, inode);
3322 static inline void inode_has_no_xattr(struct inode *inode)
3324 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3325 inode->i_flags |= S_NOSEC;
3328 static inline bool is_root_inode(struct inode *inode)
3330 return inode == inode->i_sb->s_root->d_inode;
3333 static inline bool dir_emit(struct dir_context *ctx,
3334 const char *name, int namelen,
3335 u64 ino, unsigned type)
3337 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3339 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3341 return ctx->actor(ctx, ".", 1, ctx->pos,
3342 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3344 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3346 return ctx->actor(ctx, "..", 2, ctx->pos,
3347 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3349 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3351 if (ctx->pos == 0) {
3352 if (!dir_emit_dot(file, ctx))
3356 if (ctx->pos == 1) {
3357 if (!dir_emit_dotdot(file, ctx))
3363 static inline bool dir_relax(struct inode *inode)
3365 inode_unlock(inode);
3367 return !IS_DEADDIR(inode);
3370 static inline bool dir_relax_shared(struct inode *inode)
3372 inode_unlock_shared(inode);
3373 inode_lock_shared(inode);
3374 return !IS_DEADDIR(inode);
3377 extern bool path_noexec(const struct path *path);
3378 extern void inode_nohighmem(struct inode *inode);
3380 #endif /* _LINUX_FS_H */