4 #include <linux/linkage.h>
5 #include <linux/wait.h>
6 #include <linux/kdev_t.h>
7 #include <linux/dcache.h>
8 #include <linux/path.h>
9 #include <linux/stat.h>
10 #include <linux/cache.h>
11 #include <linux/list.h>
12 #include <linux/list_lru.h>
13 #include <linux/llist.h>
14 #include <linux/radix-tree.h>
15 #include <linux/rbtree.h>
16 #include <linux/init.h>
17 #include <linux/pid.h>
18 #include <linux/bug.h>
19 #include <linux/mutex.h>
20 #include <linux/rwsem.h>
21 #include <linux/capability.h>
22 #include <linux/semaphore.h>
23 #include <linux/fcntl.h>
24 #include <linux/fiemap.h>
25 #include <linux/rculist_bl.h>
26 #include <linux/atomic.h>
27 #include <linux/shrinker.h>
28 #include <linux/migrate_mode.h>
29 #include <linux/uidgid.h>
30 #include <linux/lockdep.h>
31 #include <linux/percpu-rwsem.h>
32 #include <linux/blk_types.h>
33 #include <linux/workqueue.h>
34 #include <linux/percpu-rwsem.h>
36 #include <asm/byteorder.h>
37 #include <uapi/linux/fs.h>
39 struct backing_dev_info;
41 struct export_operations;
46 struct pipe_inode_info;
47 struct poll_table_struct;
49 struct vm_area_struct;
52 struct swap_info_struct;
54 struct workqueue_struct;
57 struct fscrypt_operations;
59 extern void __init inode_init(void);
60 extern void __init inode_init_early(void);
61 extern void __init files_init(void);
62 extern void __init files_maxfiles_init(void);
64 extern struct files_stat_struct files_stat;
65 extern unsigned long get_max_files(void);
66 extern int sysctl_nr_open;
67 extern struct inodes_stat_t inodes_stat;
68 extern int leases_enable, lease_break_time;
69 extern int sysctl_protected_symlinks;
70 extern int sysctl_protected_hardlinks;
71 extern int sysctl_protected_fifos;
72 extern int sysctl_protected_regular;
75 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
76 struct buffer_head *bh_result, int create);
77 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
78 ssize_t bytes, void *private);
79 typedef void (dax_iodone_t)(struct buffer_head *bh_map, int uptodate);
81 #define MAY_EXEC 0x00000001
82 #define MAY_WRITE 0x00000002
83 #define MAY_READ 0x00000004
84 #define MAY_APPEND 0x00000008
85 #define MAY_ACCESS 0x00000010
86 #define MAY_OPEN 0x00000020
87 #define MAY_CHDIR 0x00000040
88 /* called from RCU mode, don't block */
89 #define MAY_NOT_BLOCK 0x00000080
92 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
93 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
96 /* file is open for reading */
97 #define FMODE_READ ((__force fmode_t)0x1)
98 /* file is open for writing */
99 #define FMODE_WRITE ((__force fmode_t)0x2)
100 /* file is seekable */
101 #define FMODE_LSEEK ((__force fmode_t)0x4)
102 /* file can be accessed using pread */
103 #define FMODE_PREAD ((__force fmode_t)0x8)
104 /* file can be accessed using pwrite */
105 #define FMODE_PWRITE ((__force fmode_t)0x10)
106 /* File is opened for execution with sys_execve / sys_uselib */
107 #define FMODE_EXEC ((__force fmode_t)0x20)
108 /* File is opened with O_NDELAY (only set for block devices) */
109 #define FMODE_NDELAY ((__force fmode_t)0x40)
110 /* File is opened with O_EXCL (only set for block devices) */
111 #define FMODE_EXCL ((__force fmode_t)0x80)
112 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
113 (specialy hack for floppy.c) */
114 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
115 /* 32bit hashes as llseek() offset (for directories) */
116 #define FMODE_32BITHASH ((__force fmode_t)0x200)
117 /* 64bit hashes as llseek() offset (for directories) */
118 #define FMODE_64BITHASH ((__force fmode_t)0x400)
121 * Don't update ctime and mtime.
123 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
124 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
126 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
128 /* Expect random access pattern */
129 #define FMODE_RANDOM ((__force fmode_t)0x1000)
131 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
132 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
134 /* File is opened with O_PATH; almost nothing can be done with it */
135 #define FMODE_PATH ((__force fmode_t)0x4000)
137 /* File needs atomic accesses to f_pos */
138 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
139 /* Write access to underlying fs */
140 #define FMODE_WRITER ((__force fmode_t)0x10000)
141 /* Has read method(s) */
142 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
143 /* Has write method(s) */
144 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
146 /* File is stream-like */
147 #define FMODE_STREAM ((__force fmode_t)0x200000)
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 * The below are the various read and write types that we support. Some of
165 * them include behavioral modifiers that send information down to the
166 * block layer and IO scheduler. Terminology:
168 * The block layer uses device plugging to defer IO a little bit, in
169 * the hope that we will see more IO very shortly. This increases
170 * coalescing of adjacent IO and thus reduces the number of IOs we
171 * have to send to the device. It also allows for better queuing,
172 * if the IO isn't mergeable. If the caller is going to be waiting
173 * for the IO, then he must ensure that the device is unplugged so
174 * that the IO is dispatched to the driver.
176 * All IO is handled async in Linux. This is fine for background
177 * writes, but for reads or writes that someone waits for completion
178 * on, we want to notify the block layer and IO scheduler so that they
179 * know about it. That allows them to make better scheduling
180 * decisions. So when the below references 'sync' and 'async', it
181 * is referencing this priority hint.
183 * With that in mind, the available types are:
185 * READ A normal read operation. Device will be plugged.
186 * READ_SYNC A synchronous read. Device is not plugged, caller can
187 * immediately wait on this read without caring about
189 * READA Used for read-ahead operations. Lower priority, and the
190 * block layer could (in theory) choose to ignore this
191 * request if it runs into resource problems.
192 * WRITE A normal async write. Device will be plugged.
193 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
194 * the hint that someone will be waiting on this IO
195 * shortly. The write equivalent of READ_SYNC.
196 * WRITE_ODIRECT Special case write for O_DIRECT only.
197 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
198 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
199 * non-volatile media on completion.
200 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
201 * by a cache flush and data is guaranteed to be on
202 * non-volatile media on completion.
205 #define RW_MASK REQ_WRITE
206 #define RWA_MASK REQ_RAHEAD
209 #define WRITE RW_MASK
210 #define READA RWA_MASK
212 #define READ_SYNC (READ | REQ_SYNC)
213 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
214 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
215 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
216 #define WRITE_FLUSH_BARRIER (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | \
218 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
219 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
220 #define WRITE_POST_FLUSH_BARRIER (WRITE | REQ_SYNC | REQ_NOIDLE | \
221 REQ_POST_FLUSH_BARRIER | REQ_BARRIER)
222 #define WRITE_ORDERED_FLUSH_BARRIER (WRITE | REQ_SYNC | REQ_NOIDLE | \
223 REQ_FLUSH | REQ_POST_FLUSH_BARRIER | \
227 * Attribute flags. These should be or-ed together to figure out what
230 #define ATTR_MODE (1 << 0)
231 #define ATTR_UID (1 << 1)
232 #define ATTR_GID (1 << 2)
233 #define ATTR_SIZE (1 << 3)
234 #define ATTR_ATIME (1 << 4)
235 #define ATTR_MTIME (1 << 5)
236 #define ATTR_CTIME (1 << 6)
237 #define ATTR_ATIME_SET (1 << 7)
238 #define ATTR_MTIME_SET (1 << 8)
239 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
240 #define ATTR_ATTR_FLAG (1 << 10)
241 #define ATTR_KILL_SUID (1 << 11)
242 #define ATTR_KILL_SGID (1 << 12)
243 #define ATTR_FILE (1 << 13)
244 #define ATTR_KILL_PRIV (1 << 14)
245 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
246 #define ATTR_TIMES_SET (1 << 16)
247 #define ATTR_TOUCH (1 << 17)
250 * Whiteout is represented by a char device. The following constants define the
251 * mode and device number to use.
253 #define WHITEOUT_MODE 0
254 #define WHITEOUT_DEV 0
257 * This is the Inode Attributes structure, used for notify_change(). It
258 * uses the above definitions as flags, to know which values have changed.
259 * Also, in this manner, a Filesystem can look at only the values it cares
260 * about. Basically, these are the attributes that the VFS layer can
261 * request to change from the FS layer.
263 * Derek Atkins <warlord@MIT.EDU> 94-10-20
266 unsigned int ia_valid;
271 struct timespec ia_atime;
272 struct timespec ia_mtime;
273 struct timespec ia_ctime;
276 * Not an attribute, but an auxiliary info for filesystems wanting to
277 * implement an ftruncate() like method. NOTE: filesystem should
278 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
280 struct file *ia_file;
284 * Includes for diskquotas.
286 #include <linux/quota.h>
289 * Maximum number of layers of fs stack. Needs to be limited to
290 * prevent kernel stack overflow
292 #define FILESYSTEM_MAX_STACK_DEPTH 2
295 * enum positive_aop_returns - aop return codes with specific semantics
297 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
298 * completed, that the page is still locked, and
299 * should be considered active. The VM uses this hint
300 * to return the page to the active list -- it won't
301 * be a candidate for writeback again in the near
302 * future. Other callers must be careful to unlock
303 * the page if they get this return. Returned by
306 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
307 * unlocked it and the page might have been truncated.
308 * The caller should back up to acquiring a new page and
309 * trying again. The aop will be taking reasonable
310 * precautions not to livelock. If the caller held a page
311 * reference, it should drop it before retrying. Returned
314 * address_space_operation functions return these large constants to indicate
315 * special semantics to the caller. These are much larger than the bytes in a
316 * page to allow for functions that return the number of bytes operated on in a
320 enum positive_aop_returns {
321 AOP_WRITEPAGE_ACTIVATE = 0x80000,
322 AOP_TRUNCATED_PAGE = 0x80001,
325 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
326 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
327 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
328 * helper code (eg buffer layer)
329 * to clear GFP_FS from alloc */
332 * oh the beauties of C type declarations.
335 struct address_space;
336 struct writeback_control;
339 * Write life time hint values.
342 WRITE_LIFE_NOT_SET = 0,
343 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
344 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
345 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
346 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
347 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
350 #define IOCB_EVENTFD (1 << 0)
351 #define IOCB_APPEND (1 << 1)
352 #define IOCB_DIRECT (1 << 2)
353 #define IOCB_NOWAIT (1 << 7)
356 struct file *ki_filp;
358 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
361 enum rw_hint ki_hint;
364 static inline bool is_sync_kiocb(struct kiocb *kiocb)
366 return kiocb->ki_complete == NULL;
369 static inline int iocb_flags(struct file *file);
371 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
373 *kiocb = (struct kiocb) {
375 .ki_flags = iocb_flags(filp),
380 * "descriptor" for what we're up to with a read.
381 * This allows us to use the same read code yet
382 * have multiple different users of the data that
383 * we read from a file.
385 * The simplest case just copies the data to user
398 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
399 unsigned long, unsigned long);
401 struct address_space_operations {
402 int (*writepage)(struct page *page, struct writeback_control *wbc);
403 int (*readpage)(struct file *, struct page *);
405 /* Write back some dirty pages from this mapping. */
406 int (*writepages)(struct address_space *, struct writeback_control *);
408 /* Set a page dirty. Return true if this dirtied it */
409 int (*set_page_dirty)(struct page *page);
411 int (*readpages)(struct file *filp, struct address_space *mapping,
412 struct list_head *pages, unsigned nr_pages);
414 int (*write_begin)(struct file *, struct address_space *mapping,
415 loff_t pos, unsigned len, unsigned flags,
416 struct page **pagep, void **fsdata);
417 int (*write_end)(struct file *, struct address_space *mapping,
418 loff_t pos, unsigned len, unsigned copied,
419 struct page *page, void *fsdata);
421 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
422 sector_t (*bmap)(struct address_space *, sector_t);
423 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
424 int (*releasepage) (struct page *, gfp_t);
425 void (*freepage)(struct page *);
426 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter, loff_t offset);
428 * migrate the contents of a page to the specified target. If
429 * migrate_mode is MIGRATE_ASYNC, it must not block.
431 int (*migratepage) (struct address_space *,
432 struct page *, struct page *, enum migrate_mode);
433 bool (*isolate_page)(struct page *, isolate_mode_t);
434 void (*putback_page)(struct page *);
435 int (*launder_page) (struct page *);
436 int (*is_partially_uptodate) (struct page *, unsigned long,
438 void (*is_dirty_writeback) (struct page *, bool *, bool *);
439 int (*error_remove_page)(struct address_space *, struct page *);
441 /* swapfile support */
442 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
444 void (*swap_deactivate)(struct file *file);
447 extern const struct address_space_operations empty_aops;
450 * pagecache_write_begin/pagecache_write_end must be used by general code
451 * to write into the pagecache.
453 int pagecache_write_begin(struct file *, struct address_space *mapping,
454 loff_t pos, unsigned len, unsigned flags,
455 struct page **pagep, void **fsdata);
457 int pagecache_write_end(struct file *, struct address_space *mapping,
458 loff_t pos, unsigned len, unsigned copied,
459 struct page *page, void *fsdata);
461 struct address_space {
462 struct inode *host; /* owner: inode, block_device */
463 struct radix_tree_root page_tree; /* radix tree of all pages */
464 spinlock_t tree_lock; /* and lock protecting it */
465 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
466 struct rb_root i_mmap; /* tree of private and shared mappings */
467 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
468 /* Protected by tree_lock together with the radix tree */
469 unsigned long nrpages; /* number of total pages */
470 unsigned long nrshadows; /* number of shadow entries */
471 pgoff_t writeback_index;/* writeback starts here */
472 const struct address_space_operations *a_ops; /* methods */
473 unsigned long flags; /* error bits/gfp mask */
474 spinlock_t private_lock; /* for use by the address_space */
475 struct list_head private_list; /* ditto */
476 void *private_data; /* ditto */
477 } __attribute__((aligned(sizeof(long))));
479 * On most architectures that alignment is already the case; but
480 * must be enforced here for CRIS, to let the least significant bit
481 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
483 struct request_queue;
485 struct block_device {
486 dev_t bd_dev; /* not a kdev_t - it's a search key */
488 struct inode * bd_inode; /* will die */
489 struct super_block * bd_super;
490 struct mutex bd_mutex; /* open/close mutex */
491 struct list_head bd_inodes;
495 bool bd_write_holder;
497 struct list_head bd_holder_disks;
499 struct block_device * bd_contains;
500 unsigned bd_block_size;
501 struct hd_struct * bd_part;
502 /* number of times partitions within this device have been opened. */
503 unsigned bd_part_count;
505 struct gendisk * bd_disk;
506 struct request_queue * bd_queue;
507 struct backing_dev_info *bd_bdi;
508 struct list_head bd_list;
510 * Private data. You must have bd_claim'ed the block_device
511 * to use this. NOTE: bd_claim allows an owner to claim
512 * the same device multiple times, the owner must take special
513 * care to not mess up bd_private for that case.
515 unsigned long bd_private;
517 /* The counter of freeze processes */
518 int bd_fsfreeze_count;
519 /* Mutex for freeze */
520 struct mutex bd_fsfreeze_mutex;
524 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
527 #define PAGECACHE_TAG_DIRTY 0
528 #define PAGECACHE_TAG_WRITEBACK 1
529 #define PAGECACHE_TAG_TOWRITE 2
531 int mapping_tagged(struct address_space *mapping, int tag);
533 static inline void i_mmap_lock_write(struct address_space *mapping)
535 down_write(&mapping->i_mmap_rwsem);
538 static inline void i_mmap_unlock_write(struct address_space *mapping)
540 up_write(&mapping->i_mmap_rwsem);
543 static inline void i_mmap_lock_read(struct address_space *mapping)
545 down_read(&mapping->i_mmap_rwsem);
548 static inline void i_mmap_unlock_read(struct address_space *mapping)
550 up_read(&mapping->i_mmap_rwsem);
554 * Might pages of this file be mapped into userspace?
556 static inline int mapping_mapped(struct address_space *mapping)
558 return !RB_EMPTY_ROOT(&mapping->i_mmap);
562 * Might pages of this file have been modified in userspace?
563 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
564 * marks vma as VM_SHARED if it is shared, and the file was opened for
565 * writing i.e. vma may be mprotected writable even if now readonly.
567 * If i_mmap_writable is negative, no new writable mappings are allowed. You
568 * can only deny writable mappings, if none exists right now.
570 static inline int mapping_writably_mapped(struct address_space *mapping)
572 return atomic_read(&mapping->i_mmap_writable) > 0;
575 static inline int mapping_map_writable(struct address_space *mapping)
577 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
581 static inline void mapping_unmap_writable(struct address_space *mapping)
583 atomic_dec(&mapping->i_mmap_writable);
586 static inline int mapping_deny_writable(struct address_space *mapping)
588 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
592 static inline void mapping_allow_writable(struct address_space *mapping)
594 atomic_inc(&mapping->i_mmap_writable);
598 * Use sequence counter to get consistent i_size on 32-bit processors.
600 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
601 #include <linux/seqlock.h>
602 #define __NEED_I_SIZE_ORDERED
603 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
605 #define i_size_ordered_init(inode) do { } while (0)
609 #define ACL_NOT_CACHED ((void *)(-1))
611 #define IOP_FASTPERM 0x0001
612 #define IOP_LOOKUP 0x0002
613 #define IOP_NOFOLLOW 0x0004
616 * Keep mostly read-only and often accessed (especially for
617 * the RCU path lookup and 'stat' data) fields at the beginning
618 * of the 'struct inode'
622 unsigned short i_opflags;
625 unsigned int i_flags;
627 #ifdef CONFIG_FS_POSIX_ACL
628 struct posix_acl *i_acl;
629 struct posix_acl *i_default_acl;
632 const struct inode_operations *i_op;
633 struct super_block *i_sb;
634 struct address_space *i_mapping;
636 #ifdef CONFIG_SECURITY
640 /* Stat data, not accessed from path walking */
643 * Filesystems may only read i_nlink directly. They shall use the
644 * following functions for modification:
646 * (set|clear|inc|drop)_nlink
647 * inode_(inc|dec)_link_count
650 const unsigned int i_nlink;
651 unsigned int __i_nlink;
655 struct timespec i_atime;
656 struct timespec i_mtime;
657 struct timespec i_ctime;
658 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
659 unsigned short i_bytes;
660 unsigned int i_blkbits;
661 enum rw_hint i_write_hint;
664 #ifdef __NEED_I_SIZE_ORDERED
665 seqcount_t i_size_seqcount;
669 unsigned long i_state;
670 struct mutex i_mutex;
672 unsigned long dirtied_when; /* jiffies of first dirtying */
673 unsigned long dirtied_time_when;
675 struct hlist_node i_hash;
676 struct list_head i_io_list; /* backing dev IO list */
677 #ifdef CONFIG_CGROUP_WRITEBACK
678 struct bdi_writeback *i_wb; /* the associated cgroup wb */
680 /* foreign inode detection, see wbc_detach_inode() */
682 u16 i_wb_frn_avg_time;
683 u16 i_wb_frn_history;
685 struct list_head i_lru; /* inode LRU list */
686 struct list_head i_sb_list;
688 struct hlist_head i_dentry;
689 struct rcu_head i_rcu;
693 atomic_t i_dio_count;
694 atomic_t i_writecount;
696 atomic_t i_readcount; /* struct files open RO */
698 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
699 struct file_lock_context *i_flctx;
700 struct address_space i_data;
701 struct list_head i_devices;
703 struct pipe_inode_info *i_pipe;
704 struct block_device *i_bdev;
711 #ifdef CONFIG_FSNOTIFY
712 __u32 i_fsnotify_mask; /* all events this inode cares about */
713 struct hlist_head i_fsnotify_marks;
716 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
717 struct fscrypt_info *i_crypt_info;
719 void *i_private; /* fs or device private pointer */
722 static inline unsigned int i_blocksize(const struct inode *node)
724 return (1 << node->i_blkbits);
727 static inline int inode_unhashed(struct inode *inode)
729 return hlist_unhashed(&inode->i_hash);
733 * inode->i_mutex nesting subclasses for the lock validator:
735 * 0: the object of the current VFS operation
739 * 4: second non-directory
740 * 5: second parent (when locking independent directories in rename)
742 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
743 * non-directories at once.
745 * The locking order between these classes is
746 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
748 enum inode_i_mutex_lock_class
758 static inline void inode_lock(struct inode *inode)
760 mutex_lock(&inode->i_mutex);
763 static inline void inode_unlock(struct inode *inode)
765 mutex_unlock(&inode->i_mutex);
768 static inline int inode_trylock(struct inode *inode)
770 return mutex_trylock(&inode->i_mutex);
773 static inline int inode_is_locked(struct inode *inode)
775 return mutex_is_locked(&inode->i_mutex);
778 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
780 mutex_lock_nested(&inode->i_mutex, subclass);
783 void lock_two_nondirectories(struct inode *, struct inode*);
784 void unlock_two_nondirectories(struct inode *, struct inode*);
787 * NOTE: in a 32bit arch with a preemptable kernel and
788 * an UP compile the i_size_read/write must be atomic
789 * with respect to the local cpu (unlike with preempt disabled),
790 * but they don't need to be atomic with respect to other cpus like in
791 * true SMP (so they need either to either locally disable irq around
792 * the read or for example on x86 they can be still implemented as a
793 * cmpxchg8b without the need of the lock prefix). For SMP compiles
794 * and 64bit archs it makes no difference if preempt is enabled or not.
796 static inline loff_t i_size_read(const struct inode *inode)
798 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
803 seq = read_seqcount_begin(&inode->i_size_seqcount);
804 i_size = inode->i_size;
805 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
807 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
811 i_size = inode->i_size;
815 return inode->i_size;
820 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
821 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
822 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
824 static inline void i_size_write(struct inode *inode, loff_t i_size)
826 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
828 write_seqcount_begin(&inode->i_size_seqcount);
829 inode->i_size = i_size;
830 write_seqcount_end(&inode->i_size_seqcount);
832 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
834 inode->i_size = i_size;
837 inode->i_size = i_size;
841 /* Helper functions so that in most cases filesystems will
842 * not need to deal directly with kuid_t and kgid_t and can
843 * instead deal with the raw numeric values that are stored
846 static inline uid_t i_uid_read(const struct inode *inode)
848 return from_kuid(&init_user_ns, inode->i_uid);
851 static inline gid_t i_gid_read(const struct inode *inode)
853 return from_kgid(&init_user_ns, inode->i_gid);
856 static inline void i_uid_write(struct inode *inode, uid_t uid)
858 inode->i_uid = make_kuid(&init_user_ns, uid);
861 static inline void i_gid_write(struct inode *inode, gid_t gid)
863 inode->i_gid = make_kgid(&init_user_ns, gid);
866 static inline unsigned iminor(const struct inode *inode)
868 return MINOR(inode->i_rdev);
871 static inline unsigned imajor(const struct inode *inode)
873 return MAJOR(inode->i_rdev);
876 extern struct block_device *I_BDEV(struct inode *inode);
879 rwlock_t lock; /* protects pid, uid, euid fields */
880 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
881 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
882 kuid_t uid, euid; /* uid/euid of process setting the owner */
883 int signum; /* posix.1b rt signal to be delivered on IO */
887 * Track a single file's readahead state
889 struct file_ra_state {
890 pgoff_t start; /* where readahead started */
891 unsigned int size; /* # of readahead pages */
892 unsigned int async_size; /* do asynchronous readahead when
893 there are only # of pages ahead */
895 unsigned int ra_pages; /* Maximum readahead window */
896 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
897 loff_t prev_pos; /* Cache last read() position */
901 * Check if @index falls in the readahead windows.
903 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
905 return (index >= ra->start &&
906 index < ra->start + ra->size);
911 struct llist_node fu_llist;
912 struct rcu_head fu_rcuhead;
915 struct inode *f_inode; /* cached value */
916 const struct file_operations *f_op;
919 * Protects f_ep_links, f_flags.
920 * Must not be taken from IRQ context.
923 atomic_long_t f_count;
924 unsigned int f_flags;
926 struct mutex f_pos_lock;
928 struct fown_struct f_owner;
929 const struct cred *f_cred;
930 struct file_ra_state f_ra;
933 #ifdef CONFIG_SECURITY
936 /* needed for tty driver, and maybe others */
940 /* Used by fs/eventpoll.c to link all the hooks to this file */
941 struct list_head f_ep_links;
942 struct list_head f_tfile_llink;
943 #endif /* #ifdef CONFIG_EPOLL */
944 struct address_space *f_mapping;
946 #ifdef CONFIG_FILE_TABLE_DEBUG
947 struct hlist_node f_hash;
948 #endif /* #ifdef CONFIG_FILE_TABLE_DEBUG */
949 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
954 /* file identifier */
955 unsigned char f_handle[0];
958 static inline struct file *get_file(struct file *f)
960 atomic_long_inc(&f->f_count);
963 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
964 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
965 #define file_count(x) atomic_long_read(&(x)->f_count)
967 #define MAX_NON_LFS ((1UL<<31) - 1)
969 /* Page cache limit. The filesystems should put that into their s_maxbytes
970 limits, otherwise bad things can happen in VM. */
971 #if BITS_PER_LONG==32
972 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
973 #elif BITS_PER_LONG==64
974 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
979 #define FL_DELEG 4 /* NFSv4 delegation */
980 #define FL_ACCESS 8 /* not trying to lock, just looking */
981 #define FL_EXISTS 16 /* when unlocking, test for existence */
982 #define FL_LEASE 32 /* lease held on this file */
983 #define FL_CLOSE 64 /* unlock on close */
984 #define FL_SLEEP 128 /* A blocking lock */
985 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
986 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
987 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
988 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
991 * Special return value from posix_lock_file() and vfs_lock_file() for
992 * asynchronous locking.
994 #define FILE_LOCK_DEFERRED 1
996 /* legacy typedef, should eventually be removed */
997 typedef void *fl_owner_t;
1001 struct file_lock_operations {
1002 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1003 void (*fl_release_private)(struct file_lock *);
1006 struct lock_manager_operations {
1007 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
1008 unsigned long (*lm_owner_key)(struct file_lock *);
1009 fl_owner_t (*lm_get_owner)(fl_owner_t);
1010 void (*lm_put_owner)(fl_owner_t);
1011 void (*lm_notify)(struct file_lock *); /* unblock callback */
1012 int (*lm_grant)(struct file_lock *, int);
1013 bool (*lm_break)(struct file_lock *);
1014 int (*lm_change)(struct file_lock *, int, struct list_head *);
1015 void (*lm_setup)(struct file_lock *, void **);
1018 struct lock_manager {
1019 struct list_head list;
1021 * NFSv4 and up also want opens blocked during the grace period;
1028 void locks_start_grace(struct net *, struct lock_manager *);
1029 void locks_end_grace(struct lock_manager *);
1030 int locks_in_grace(struct net *);
1031 int opens_in_grace(struct net *);
1033 /* that will die - we need it for nfs_lock_info */
1034 #include <linux/nfs_fs_i.h>
1037 * struct file_lock represents a generic "file lock". It's used to represent
1038 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1039 * note that the same struct is used to represent both a request for a lock and
1040 * the lock itself, but the same object is never used for both.
1042 * FIXME: should we create a separate "struct lock_request" to help distinguish
1045 * The varous i_flctx lists are ordered by:
1048 * 2) lock range start
1051 * Obviously, the last two criteria only matter for POSIX locks.
1054 struct file_lock *fl_next; /* singly linked list for this inode */
1055 struct list_head fl_list; /* link into file_lock_context */
1056 struct hlist_node fl_link; /* node in global lists */
1057 struct list_head fl_block; /* circular list of blocked processes */
1058 fl_owner_t fl_owner;
1059 unsigned int fl_flags;
1060 unsigned char fl_type;
1061 unsigned int fl_pid;
1062 int fl_link_cpu; /* what cpu's list is this on? */
1063 struct pid *fl_nspid;
1064 wait_queue_head_t fl_wait;
1065 struct file *fl_file;
1069 struct fasync_struct * fl_fasync; /* for lease break notifications */
1070 /* for lease breaks: */
1071 unsigned long fl_break_time;
1072 unsigned long fl_downgrade_time;
1074 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1075 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1077 struct nfs_lock_info nfs_fl;
1078 struct nfs4_lock_info nfs4_fl;
1080 struct list_head link; /* link in AFS vnode's pending_locks list */
1081 int state; /* state of grant or error if -ve */
1086 struct file_lock_context {
1087 spinlock_t flc_lock;
1088 struct list_head flc_flock;
1089 struct list_head flc_posix;
1090 struct list_head flc_lease;
1093 /* The following constant reflects the upper bound of the file/locking space */
1095 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1096 #define OFFSET_MAX INT_LIMIT(loff_t)
1097 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1100 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1102 #ifdef CONFIG_FILE_LOCKING
1103 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
1104 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1105 struct flock __user *);
1107 #if BITS_PER_LONG == 32
1108 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
1109 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1110 struct flock64 __user *);
1113 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1114 extern int fcntl_getlease(struct file *filp);
1117 void locks_free_lock_context(struct file_lock_context *ctx);
1118 void locks_free_lock(struct file_lock *fl);
1119 extern void locks_init_lock(struct file_lock *);
1120 extern struct file_lock * locks_alloc_lock(void);
1121 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1122 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1123 extern void locks_remove_posix(struct file *, fl_owner_t);
1124 extern void locks_remove_file(struct file *);
1125 extern void locks_release_private(struct file_lock *);
1126 extern void posix_test_lock(struct file *, struct file_lock *);
1127 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1128 extern int posix_unblock_lock(struct file_lock *);
1129 extern int vfs_test_lock(struct file *, struct file_lock *);
1130 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1131 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1132 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1133 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1134 extern void lease_get_mtime(struct inode *, struct timespec *time);
1135 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1136 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1137 extern int lease_modify(struct file_lock *, int, struct list_head *);
1138 struct files_struct;
1139 extern void show_fd_locks(struct seq_file *f,
1140 struct file *filp, struct files_struct *files);
1141 #else /* !CONFIG_FILE_LOCKING */
1142 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1143 struct flock __user *user)
1148 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1149 unsigned int cmd, struct flock __user *user)
1154 #if BITS_PER_LONG == 32
1155 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1156 struct flock64 __user *user)
1161 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1162 unsigned int cmd, struct flock64 __user *user)
1167 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1172 static inline int fcntl_getlease(struct file *filp)
1178 locks_free_lock_context(struct file_lock_context *ctx)
1182 static inline void locks_init_lock(struct file_lock *fl)
1187 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1192 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1197 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1202 static inline void locks_remove_file(struct file *filp)
1207 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1212 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1213 struct file_lock *conflock)
1218 static inline int posix_unblock_lock(struct file_lock *waiter)
1223 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1228 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1229 struct file_lock *fl, struct file_lock *conf)
1234 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1239 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1244 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1249 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1254 static inline int generic_setlease(struct file *filp, long arg,
1255 struct file_lock **flp, void **priv)
1260 static inline int vfs_setlease(struct file *filp, long arg,
1261 struct file_lock **lease, void **priv)
1266 static inline int lease_modify(struct file_lock *fl, int arg,
1267 struct list_head *dispose)
1272 struct files_struct;
1273 static inline void show_fd_locks(struct seq_file *f,
1274 struct file *filp, struct files_struct *files) {}
1275 #endif /* !CONFIG_FILE_LOCKING */
1277 static inline struct inode *file_inode(const struct file *f)
1282 static inline struct dentry *file_dentry(const struct file *file)
1284 struct dentry *dentry = file->f_path.dentry;
1286 if (unlikely(dentry->d_flags & DCACHE_OP_REAL))
1287 return dentry->d_op->d_real(dentry, file_inode(file));
1292 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1294 return locks_lock_inode_wait(file_inode(filp), fl);
1297 struct fasync_struct {
1301 struct fasync_struct *fa_next; /* singly linked list */
1302 struct file *fa_file;
1303 struct rcu_head fa_rcu;
1306 #define FASYNC_MAGIC 0x4601
1308 /* SMP safe fasync helpers: */
1309 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1310 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1311 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1312 extern struct fasync_struct *fasync_alloc(void);
1313 extern void fasync_free(struct fasync_struct *);
1315 /* can be called from interrupts */
1316 extern void kill_fasync(struct fasync_struct **, int, int);
1318 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1319 extern void f_setown(struct file *filp, unsigned long arg, int force);
1320 extern void f_delown(struct file *filp);
1321 extern pid_t f_getown(struct file *filp);
1322 extern int send_sigurg(struct fown_struct *fown);
1330 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1331 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1332 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1333 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1334 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1337 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1338 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1339 #define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */
1341 /* Possible states of 'frozen' field */
1343 SB_UNFROZEN = 0, /* FS is unfrozen */
1344 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1345 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1346 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1347 * internal threads if needed) */
1348 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1351 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1354 int frozen; /* Is sb frozen? */
1355 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1356 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1359 struct super_block {
1360 struct list_head s_list; /* Keep this first */
1361 dev_t s_dev; /* search index; _not_ kdev_t */
1362 unsigned char s_blocksize_bits;
1363 unsigned long s_blocksize;
1364 loff_t s_maxbytes; /* Max file size */
1365 struct file_system_type *s_type;
1366 const struct super_operations *s_op;
1367 const struct dquot_operations *dq_op;
1368 const struct quotactl_ops *s_qcop;
1369 const struct export_operations *s_export_op;
1370 unsigned long s_flags;
1371 unsigned long s_iflags; /* internal SB_I_* flags */
1372 unsigned long s_magic;
1373 struct dentry *s_root;
1374 struct rw_semaphore s_umount;
1377 #ifdef CONFIG_SECURITY
1380 const struct xattr_handler **s_xattr;
1382 const struct fscrypt_operations *s_cop;
1384 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1385 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1386 struct block_device *s_bdev;
1387 struct backing_dev_info *s_bdi;
1388 struct mtd_info *s_mtd;
1389 struct hlist_node s_instances;
1390 unsigned int s_quota_types; /* Bitmask of supported quota types */
1391 struct quota_info s_dquot; /* Diskquota specific options */
1393 struct sb_writers s_writers;
1395 char s_id[32]; /* Informational name */
1396 u8 s_uuid[16]; /* UUID */
1398 void *s_fs_info; /* Filesystem private info */
1399 unsigned int s_max_links;
1402 /* Granularity of c/m/atime in ns.
1403 Cannot be worse than a second */
1407 * The next field is for VFS *only*. No filesystems have any business
1408 * even looking at it. You had been warned.
1410 struct mutex s_vfs_rename_mutex; /* Kludge */
1413 * Filesystem subtype. If non-empty the filesystem type field
1414 * in /proc/mounts will be "type.subtype"
1419 * Saved mount options for lazy filesystems using
1420 * generic_show_options()
1422 char __rcu *s_options;
1423 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1426 * Saved pool identifier for cleancache (-1 means none)
1428 int cleancache_poolid;
1430 struct shrinker s_shrink; /* per-sb shrinker handle */
1432 /* Number of inodes with nlink == 0 but still referenced */
1433 atomic_long_t s_remove_count;
1435 /* Being remounted read-only */
1436 int s_readonly_remount;
1438 /* AIO completions deferred from interrupt context */
1439 struct workqueue_struct *s_dio_done_wq;
1440 struct hlist_head s_pins;
1443 * Keep the lru lists last in the structure so they always sit on their
1444 * own individual cachelines.
1446 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1447 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1448 struct rcu_head rcu;
1449 struct work_struct destroy_work;
1451 struct mutex s_sync_lock; /* sync serialisation lock */
1454 * Indicates how deep in a filesystem stack this SB is
1458 /* s_inode_list_lock protects s_inodes */
1459 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1460 struct list_head s_inodes; /* all inodes */
1463 extern struct timespec current_fs_time(struct super_block *sb);
1466 * Snapshotting support.
1469 void __sb_end_write(struct super_block *sb, int level);
1470 int __sb_start_write(struct super_block *sb, int level, bool wait);
1472 #define __sb_writers_acquired(sb, lev) \
1473 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1474 #define __sb_writers_release(sb, lev) \
1475 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1478 * sb_end_write - drop write access to a superblock
1479 * @sb: the super we wrote to
1481 * Decrement number of writers to the filesystem. Wake up possible waiters
1482 * wanting to freeze the filesystem.
1484 static inline void sb_end_write(struct super_block *sb)
1486 __sb_end_write(sb, SB_FREEZE_WRITE);
1490 * sb_end_pagefault - drop write access to a superblock from a page fault
1491 * @sb: the super we wrote to
1493 * Decrement number of processes handling write page fault to the filesystem.
1494 * Wake up possible waiters wanting to freeze the filesystem.
1496 static inline void sb_end_pagefault(struct super_block *sb)
1498 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1502 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1503 * @sb: the super we wrote to
1505 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1506 * waiters wanting to freeze the filesystem.
1508 static inline void sb_end_intwrite(struct super_block *sb)
1510 __sb_end_write(sb, SB_FREEZE_FS);
1514 * sb_start_write - get write access to a superblock
1515 * @sb: the super we write to
1517 * When a process wants to write data or metadata to a file system (i.e. dirty
1518 * a page or an inode), it should embed the operation in a sb_start_write() -
1519 * sb_end_write() pair to get exclusion against file system freezing. This
1520 * function increments number of writers preventing freezing. If the file
1521 * system is already frozen, the function waits until the file system is
1524 * Since freeze protection behaves as a lock, users have to preserve
1525 * ordering of freeze protection and other filesystem locks. Generally,
1526 * freeze protection should be the outermost lock. In particular, we have:
1529 * -> i_mutex (write path, truncate, directory ops, ...)
1530 * -> s_umount (freeze_super, thaw_super)
1532 static inline void sb_start_write(struct super_block *sb)
1534 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1537 static inline int sb_start_write_trylock(struct super_block *sb)
1539 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1543 * sb_start_pagefault - get write access to a superblock from a page fault
1544 * @sb: the super we write to
1546 * When a process starts handling write page fault, it should embed the
1547 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1548 * exclusion against file system freezing. This is needed since the page fault
1549 * is going to dirty a page. This function increments number of running page
1550 * faults preventing freezing. If the file system is already frozen, the
1551 * function waits until the file system is thawed.
1553 * Since page fault freeze protection behaves as a lock, users have to preserve
1554 * ordering of freeze protection and other filesystem locks. It is advised to
1555 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1556 * handling code implies lock dependency:
1559 * -> sb_start_pagefault
1561 static inline void sb_start_pagefault(struct super_block *sb)
1563 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1567 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1568 * @sb: the super we write to
1570 * This is the third level of protection against filesystem freezing. It is
1571 * free for use by a filesystem. The only requirement is that it must rank
1572 * below sb_start_pagefault.
1574 * For example filesystem can call sb_start_intwrite() when starting a
1575 * transaction which somewhat eases handling of freezing for internal sources
1576 * of filesystem changes (internal fs threads, discarding preallocation on file
1579 static inline void sb_start_intwrite(struct super_block *sb)
1581 __sb_start_write(sb, SB_FREEZE_FS, true);
1585 extern bool inode_owner_or_capable(const struct inode *inode);
1588 * VFS helper functions..
1590 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1591 extern int vfs_create2(struct vfsmount *, struct inode *, struct dentry *, umode_t, bool);
1592 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1593 extern int vfs_mkdir2(struct vfsmount *, struct inode *, struct dentry *, umode_t);
1594 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1595 extern int vfs_mknod2(struct vfsmount *, struct inode *, struct dentry *, umode_t, dev_t);
1596 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1597 extern int vfs_symlink2(struct vfsmount *, struct inode *, struct dentry *, const char *);
1598 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1599 extern int vfs_link2(struct vfsmount *, struct dentry *, struct inode *, struct dentry *, struct inode **);
1600 extern int vfs_rmdir(struct inode *, struct dentry *);
1601 extern int vfs_rmdir2(struct vfsmount *, struct inode *, struct dentry *);
1602 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1603 extern int vfs_unlink2(struct vfsmount *, struct inode *, struct dentry *, struct inode **);
1604 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1605 extern int vfs_rename2(struct vfsmount *, struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1606 extern int vfs_whiteout(struct inode *, struct dentry *);
1609 * VFS dentry helper functions.
1611 extern void dentry_unhash(struct dentry *dentry);
1614 * VFS file helper functions.
1616 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1619 * VFS FS_IOC_FIEMAP helper definitions.
1621 struct fiemap_extent_info {
1622 unsigned int fi_flags; /* Flags as passed from user */
1623 unsigned int fi_extents_mapped; /* Number of mapped extents */
1624 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1625 struct fiemap_extent __user *fi_extents_start; /* Start of
1626 fiemap_extent array */
1628 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1629 u64 phys, u64 len, u32 flags);
1630 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1635 * NOTE! These match bits 12..15 of stat.st_mode
1636 * (ie "(i_mode >> 12) & 15").
1638 #define DT_UNKNOWN 0
1649 * This is the "filldir" function type, used by readdir() to let
1650 * the kernel specify what kind of dirent layout it wants to have.
1651 * This allows the kernel to read directories into kernel space or
1652 * to have different dirent layouts depending on the binary type.
1655 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1658 struct dir_context {
1659 const filldir_t actor;
1663 struct block_device_operations;
1665 /* These macros are for out of kernel modules to test that
1666 * the kernel supports the unlocked_ioctl and compat_ioctl
1667 * fields in struct file_operations. */
1668 #define HAVE_COMPAT_IOCTL 1
1669 #define HAVE_UNLOCKED_IOCTL 1
1672 * These flags let !MMU mmap() govern direct device mapping vs immediate
1673 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1675 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1676 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1677 * NOMMU_MAP_READ: Can be mapped for reading
1678 * NOMMU_MAP_WRITE: Can be mapped for writing
1679 * NOMMU_MAP_EXEC: Can be mapped for execution
1681 #define NOMMU_MAP_COPY 0x00000001
1682 #define NOMMU_MAP_DIRECT 0x00000008
1683 #define NOMMU_MAP_READ VM_MAYREAD
1684 #define NOMMU_MAP_WRITE VM_MAYWRITE
1685 #define NOMMU_MAP_EXEC VM_MAYEXEC
1687 #define NOMMU_VMFLAGS \
1688 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1693 struct file_operations {
1694 struct module *owner;
1695 loff_t (*llseek) (struct file *, loff_t, int);
1696 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1697 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1698 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1699 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1700 int (*iterate) (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 (*aio_fsync) (struct kiocb *, int datasync);
1710 int (*fasync) (int, struct file *, int);
1711 int (*lock) (struct file *, int, struct file_lock *);
1712 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1713 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1714 int (*check_flags)(int);
1715 int (*flock) (struct file *, int, struct file_lock *);
1716 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1717 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1718 int (*setlease)(struct file *, long, struct file_lock **, void **);
1719 long (*fallocate)(struct file *file, int mode, loff_t offset,
1721 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1723 unsigned (*mmap_capabilities)(struct file *);
1727 struct inode_operations {
1728 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1729 const char * (*follow_link) (struct dentry *, void **);
1730 int (*permission) (struct inode *, int);
1731 int (*permission2) (struct vfsmount *, struct inode *, int);
1732 struct posix_acl * (*get_acl)(struct inode *, int);
1734 int (*readlink) (struct dentry *, char __user *,int);
1735 void (*put_link) (struct inode *, void *);
1737 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1738 int (*link) (struct dentry *,struct inode *,struct dentry *);
1739 int (*unlink) (struct inode *,struct dentry *);
1740 int (*symlink) (struct inode *,struct dentry *,const char *);
1741 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1742 int (*rmdir) (struct inode *,struct dentry *);
1743 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1744 int (*rename) (struct inode *, struct dentry *,
1745 struct inode *, struct dentry *);
1746 int (*rename2) (struct inode *, struct dentry *,
1747 struct inode *, struct dentry *, unsigned int);
1748 int (*setattr) (struct dentry *, struct iattr *);
1749 int (*setattr2) (struct vfsmount *, struct dentry *, struct iattr *);
1750 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1751 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1752 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1753 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1754 int (*removexattr) (struct dentry *, const char *);
1755 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1757 int (*update_time)(struct inode *, struct timespec *, int);
1758 int (*atomic_open)(struct inode *, struct dentry *,
1759 struct file *, unsigned open_flag,
1760 umode_t create_mode, int *opened);
1761 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1762 int (*set_acl)(struct inode *, struct posix_acl *, int);
1763 } ____cacheline_aligned;
1765 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1766 unsigned long nr_segs, unsigned long fast_segs,
1767 struct iovec *fast_pointer,
1768 struct iovec **ret_pointer);
1770 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1771 extern ssize_t __vfs_write(struct file *, const char __user *, size_t, loff_t *);
1772 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1773 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1774 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1775 unsigned long, loff_t *);
1776 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1777 unsigned long, loff_t *);
1779 struct super_operations {
1780 struct inode *(*alloc_inode)(struct super_block *sb);
1781 void (*destroy_inode)(struct inode *);
1783 void (*dirty_inode) (struct inode *, int flags);
1784 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1785 int (*drop_inode) (struct inode *);
1786 void (*evict_inode) (struct inode *);
1787 void (*put_super) (struct super_block *);
1788 int (*sync_fs)(struct super_block *sb, int wait);
1789 int (*freeze_super) (struct super_block *);
1790 int (*freeze_fs) (struct super_block *);
1791 int (*thaw_super) (struct super_block *);
1792 int (*unfreeze_fs) (struct super_block *);
1793 int (*statfs) (struct dentry *, struct kstatfs *);
1794 int (*remount_fs) (struct super_block *, int *, char *);
1795 int (*remount_fs2) (struct vfsmount *, struct super_block *, int *, char *);
1796 void *(*clone_mnt_data) (void *);
1797 void (*copy_mnt_data) (void *, void *);
1798 void (*umount_begin) (struct super_block *);
1800 int (*show_options)(struct seq_file *, struct dentry *);
1801 int (*show_options2)(struct vfsmount *,struct seq_file *, struct dentry *);
1802 int (*show_devname)(struct seq_file *, struct dentry *);
1803 int (*show_path)(struct seq_file *, struct dentry *);
1804 int (*show_stats)(struct seq_file *, struct dentry *);
1806 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1807 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1808 struct dquot **(*get_dquots)(struct inode *);
1810 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1811 long (*nr_cached_objects)(struct super_block *,
1812 struct shrink_control *);
1813 long (*free_cached_objects)(struct super_block *,
1814 struct shrink_control *);
1818 * Inode flags - they have no relation to superblock flags now
1820 #define S_SYNC 1 /* Writes are synced at once */
1821 #define S_NOATIME 2 /* Do not update access times */
1822 #define S_APPEND 4 /* Append-only file */
1823 #define S_IMMUTABLE 8 /* Immutable file */
1824 #define S_DEAD 16 /* removed, but still open directory */
1825 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1826 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1827 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1828 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1829 #define S_PRIVATE 512 /* Inode is fs-internal */
1830 #define S_IMA 1024 /* Inode has an associated IMA struct */
1831 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1832 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1833 #ifdef CONFIG_FS_DAX
1834 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1836 #define S_DAX 0 /* Make all the DAX code disappear */
1838 #define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */
1841 * Note that nosuid etc flags are inode-specific: setting some file-system
1842 * flags just means all the inodes inherit those flags by default. It might be
1843 * possible to override it selectively if you really wanted to with some
1844 * ioctl() that is not currently implemented.
1846 * Exception: MS_RDONLY is always applied to the entire file system.
1848 * Unfortunately, it is possible to change a filesystems flags with it mounted
1849 * with files in use. This means that all of the inodes will not have their
1850 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1851 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1853 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1855 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1856 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1857 ((inode)->i_flags & S_SYNC))
1858 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1859 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1860 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1861 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1862 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1864 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1865 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1866 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1867 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1869 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1870 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1871 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1872 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1873 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1874 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1875 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1876 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1877 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
1879 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1880 (inode)->i_rdev == WHITEOUT_DEV)
1883 * Inode state bits. Protected by inode->i_lock
1885 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1886 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1888 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1889 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1890 * various stages of removing an inode.
1892 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1894 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1895 * fdatasync(). i_atime is the usual cause.
1896 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1897 * these changes separately from I_DIRTY_SYNC so that we
1898 * don't have to write inode on fdatasync() when only
1899 * mtime has changed in it.
1900 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1901 * I_NEW Serves as both a mutex and completion notification.
1902 * New inodes set I_NEW. If two processes both create
1903 * the same inode, one of them will release its inode and
1904 * wait for I_NEW to be released before returning.
1905 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1906 * also cause waiting on I_NEW, without I_NEW actually
1907 * being set. find_inode() uses this to prevent returning
1908 * nearly-dead inodes.
1909 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1910 * is zero. I_FREEING must be set when I_WILL_FREE is
1912 * I_FREEING Set when inode is about to be freed but still has dirty
1913 * pages or buffers attached or the inode itself is still
1915 * I_CLEAR Added by clear_inode(). In this state the inode is
1916 * clean and can be destroyed. Inode keeps I_FREEING.
1918 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1919 * prohibited for many purposes. iget() must wait for
1920 * the inode to be completely released, then create it
1921 * anew. Other functions will just ignore such inodes,
1922 * if appropriate. I_NEW is used for waiting.
1924 * I_SYNC Writeback of inode is running. The bit is set during
1925 * data writeback, and cleared with a wakeup on the bit
1926 * address once it is done. The bit is also used to pin
1927 * the inode in memory for flusher thread.
1929 * I_REFERENCED Marks the inode as recently references on the LRU list.
1931 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1933 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1934 * synchronize competing switching instances and to tell
1935 * wb stat updates to grab mapping->tree_lock. See
1936 * inode_switch_wb_work_fn() for details.
1938 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1940 #define I_DIRTY_SYNC (1 << 0)
1941 #define I_DIRTY_DATASYNC (1 << 1)
1942 #define I_DIRTY_PAGES (1 << 2)
1944 #define I_NEW (1 << __I_NEW)
1945 #define I_WILL_FREE (1 << 4)
1946 #define I_FREEING (1 << 5)
1947 #define I_CLEAR (1 << 6)
1949 #define I_SYNC (1 << __I_SYNC)
1950 #define I_REFERENCED (1 << 8)
1951 #define __I_DIO_WAKEUP 9
1952 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1953 #define I_LINKABLE (1 << 10)
1954 #define I_DIRTY_TIME (1 << 11)
1955 #define __I_DIRTY_TIME_EXPIRED 12
1956 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1957 #define I_WB_SWITCH (1 << 13)
1959 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1960 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1962 extern void __mark_inode_dirty(struct inode *, int);
1963 static inline void mark_inode_dirty(struct inode *inode)
1965 __mark_inode_dirty(inode, I_DIRTY);
1968 static inline void mark_inode_dirty_sync(struct inode *inode)
1970 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1973 extern void inc_nlink(struct inode *inode);
1974 extern void drop_nlink(struct inode *inode);
1975 extern void clear_nlink(struct inode *inode);
1976 extern void set_nlink(struct inode *inode, unsigned int nlink);
1978 static inline void inode_inc_link_count(struct inode *inode)
1981 mark_inode_dirty(inode);
1984 static inline void inode_dec_link_count(struct inode *inode)
1987 mark_inode_dirty(inode);
1991 * inode_inc_iversion - increments i_version
1992 * @inode: inode that need to be updated
1994 * Every time the inode is modified, the i_version field will be incremented.
1995 * The filesystem has to be mounted with i_version flag
1998 static inline void inode_inc_iversion(struct inode *inode)
2000 spin_lock(&inode->i_lock);
2002 spin_unlock(&inode->i_lock);
2005 enum file_time_flags {
2012 extern bool atime_needs_update(const struct path *, struct inode *);
2013 extern void touch_atime(const struct path *);
2014 static inline void file_accessed(struct file *file)
2016 if (!(file->f_flags & O_NOATIME))
2017 touch_atime(&file->f_path);
2020 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2021 int sync_inode_metadata(struct inode *inode, int wait);
2023 struct file_system_type {
2026 #define FS_REQUIRES_DEV 1
2027 #define FS_BINARY_MOUNTDATA 2
2028 #define FS_HAS_SUBTYPE 4
2029 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2030 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
2031 #define FS_USERNS_VISIBLE 32 /* FS must already be visible */
2032 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2033 struct dentry *(*mount) (struct file_system_type *, int,
2034 const char *, void *);
2035 struct dentry *(*mount2) (struct vfsmount *, struct file_system_type *, int,
2036 const char *, void *);
2037 void *(*alloc_mnt_data) (void);
2038 void (*kill_sb) (struct super_block *);
2039 struct module *owner;
2040 struct file_system_type * next;
2041 struct hlist_head fs_supers;
2043 struct lock_class_key s_lock_key;
2044 struct lock_class_key s_umount_key;
2045 struct lock_class_key s_vfs_rename_key;
2046 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2048 struct lock_class_key i_lock_key;
2049 struct lock_class_key i_mutex_key;
2050 struct lock_class_key i_mutex_dir_key;
2053 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2055 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
2056 void *data, int (*fill_super)(struct super_block *, void *, int));
2057 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2058 int flags, const char *dev_name, void *data,
2059 int (*fill_super)(struct super_block *, void *, int));
2060 extern struct dentry *mount_single(struct file_system_type *fs_type,
2061 int flags, void *data,
2062 int (*fill_super)(struct super_block *, void *, int));
2063 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2064 int flags, void *data,
2065 int (*fill_super)(struct super_block *, void *, int));
2066 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2067 void generic_shutdown_super(struct super_block *sb);
2068 void kill_block_super(struct super_block *sb);
2069 void kill_anon_super(struct super_block *sb);
2070 void kill_litter_super(struct super_block *sb);
2071 void deactivate_super(struct super_block *sb);
2072 void deactivate_locked_super(struct super_block *sb);
2073 int set_anon_super(struct super_block *s, void *data);
2074 int get_anon_bdev(dev_t *);
2075 void free_anon_bdev(dev_t);
2076 struct super_block *sget(struct file_system_type *type,
2077 int (*test)(struct super_block *,void *),
2078 int (*set)(struct super_block *,void *),
2079 int flags, void *data);
2080 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
2081 const struct super_operations *ops,
2082 const struct dentry_operations *dops,
2085 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2086 #define fops_get(fops) \
2087 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2088 #define fops_put(fops) \
2089 do { if (fops) module_put((fops)->owner); } while(0)
2091 * This one is to be used *ONLY* from ->open() instances.
2092 * fops must be non-NULL, pinned down *and* module dependencies
2093 * should be sufficient to pin the caller down as well.
2095 #define replace_fops(f, fops) \
2097 struct file *__file = (f); \
2098 fops_put(__file->f_op); \
2099 BUG_ON(!(__file->f_op = (fops))); \
2102 extern int register_filesystem(struct file_system_type *);
2103 extern int unregister_filesystem(struct file_system_type *);
2104 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2105 #define kern_mount(type) kern_mount_data(type, NULL)
2106 extern void kern_unmount(struct vfsmount *mnt);
2107 extern int may_umount_tree(struct vfsmount *);
2108 extern int may_umount(struct vfsmount *);
2109 extern long do_mount(const char *, const char __user *,
2110 const char *, unsigned long, void *);
2111 extern struct vfsmount *collect_mounts(struct path *);
2112 extern void drop_collected_mounts(struct vfsmount *);
2113 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2115 extern int vfs_statfs(struct path *, struct kstatfs *);
2116 extern int user_statfs(const char __user *, struct kstatfs *);
2117 extern int fd_statfs(int, struct kstatfs *);
2118 extern int vfs_ustat(dev_t, struct kstatfs *);
2119 extern int freeze_super(struct super_block *super);
2120 extern int thaw_super(struct super_block *super);
2121 extern bool our_mnt(struct vfsmount *mnt);
2123 extern int current_umask(void);
2125 extern void ihold(struct inode * inode);
2126 extern void iput(struct inode *);
2127 extern int generic_update_time(struct inode *, struct timespec *, int);
2130 extern struct kobject *fs_kobj;
2132 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
2134 #define FLOCK_VERIFY_READ 1
2135 #define FLOCK_VERIFY_WRITE 2
2137 #ifdef CONFIG_FILE_LOCKING
2138 extern int locks_mandatory_locked(struct file *);
2139 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
2142 * Candidates for mandatory locking have the setgid bit set
2143 * but no group execute bit - an otherwise meaningless combination.
2146 static inline int __mandatory_lock(struct inode *ino)
2148 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2152 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2153 * otherwise these will be advisory locks
2156 static inline int mandatory_lock(struct inode *ino)
2158 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2161 static inline int locks_verify_locked(struct file *file)
2163 if (mandatory_lock(file_inode(file)))
2164 return locks_mandatory_locked(file);
2168 static inline int locks_verify_truncate(struct inode *inode,
2172 if (inode->i_flctx && mandatory_lock(inode))
2173 return locks_mandatory_area(
2174 FLOCK_VERIFY_WRITE, inode, filp,
2175 size < inode->i_size ? size : inode->i_size,
2176 (size < inode->i_size ? inode->i_size - size
2177 : size - inode->i_size)
2182 static inline int break_lease(struct inode *inode, unsigned int mode)
2185 * Since this check is lockless, we must ensure that any refcounts
2186 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2187 * could end up racing with tasks trying to set a new lease on this
2191 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2192 return __break_lease(inode, mode, FL_LEASE);
2196 static inline int break_deleg(struct inode *inode, unsigned int mode)
2199 * Since this check is lockless, we must ensure that any refcounts
2200 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2201 * could end up racing with tasks trying to set a new lease on this
2205 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2206 return __break_lease(inode, mode, FL_DELEG);
2210 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2214 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2215 if (ret == -EWOULDBLOCK && delegated_inode) {
2216 *delegated_inode = inode;
2222 static inline int break_deleg_wait(struct inode **delegated_inode)
2226 ret = break_deleg(*delegated_inode, O_WRONLY);
2227 iput(*delegated_inode);
2228 *delegated_inode = NULL;
2232 static inline int break_layout(struct inode *inode, bool wait)
2235 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2236 return __break_lease(inode,
2237 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2242 #else /* !CONFIG_FILE_LOCKING */
2243 static inline int locks_mandatory_locked(struct file *file)
2248 static inline int locks_mandatory_area(int rw, struct inode *inode,
2249 struct file *filp, loff_t offset,
2255 static inline int __mandatory_lock(struct inode *inode)
2260 static inline int mandatory_lock(struct inode *inode)
2265 static inline int locks_verify_locked(struct file *file)
2270 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2276 static inline int break_lease(struct inode *inode, unsigned int mode)
2281 static inline int break_deleg(struct inode *inode, unsigned int mode)
2286 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2291 static inline int break_deleg_wait(struct inode **delegated_inode)
2297 static inline int break_layout(struct inode *inode, bool wait)
2302 #endif /* CONFIG_FILE_LOCKING */
2307 const char *name; /* pointer to actual string */
2308 const __user char *uptr; /* original userland pointer */
2309 struct audit_names *aname;
2314 extern long vfs_truncate(struct path *, loff_t);
2315 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2317 extern int do_truncate2(struct vfsmount *, struct dentry *, loff_t start,
2318 unsigned int time_attrs, struct file *filp);
2319 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2321 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2323 extern struct file *file_open_name(struct filename *, int, umode_t);
2324 extern struct file *filp_open(const char *, int, umode_t);
2325 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2326 const char *, int, umode_t);
2327 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2328 extern int filp_close(struct file *, fl_owner_t id);
2330 extern struct filename *getname_flags(const char __user *, int, int *);
2331 extern struct filename *getname(const char __user *);
2332 extern struct filename *getname_kernel(const char *);
2333 extern void putname(struct filename *name);
2339 extern int finish_open(struct file *file, struct dentry *dentry,
2340 int (*open)(struct inode *, struct file *),
2342 extern int finish_no_open(struct file *file, struct dentry *dentry);
2346 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2349 extern void __init vfs_caches_init_early(void);
2350 extern void __init vfs_caches_init(void);
2352 extern struct kmem_cache *names_cachep;
2354 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2355 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2358 extern int register_blkdev(unsigned int, const char *);
2359 extern void unregister_blkdev(unsigned int, const char *);
2360 extern void bdev_unhash_inode(dev_t dev);
2361 extern struct block_device *bdget(dev_t);
2362 extern struct block_device *bdgrab(struct block_device *bdev);
2363 extern void bd_set_size(struct block_device *, loff_t size);
2364 extern void bd_forget(struct inode *inode);
2365 extern void bdput(struct block_device *);
2366 extern void invalidate_bdev(struct block_device *);
2367 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2368 extern int sync_blockdev(struct block_device *bdev);
2369 extern void kill_bdev(struct block_device *);
2370 extern struct super_block *freeze_bdev(struct block_device *);
2371 extern void emergency_thaw_all(void);
2372 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2373 extern int fsync_bdev(struct block_device *);
2375 extern struct super_block *blockdev_superblock;
2377 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2379 return sb == blockdev_superblock;
2382 static inline void bd_forget(struct inode *inode) {}
2383 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2384 static inline void kill_bdev(struct block_device *bdev) {}
2385 static inline void invalidate_bdev(struct block_device *bdev) {}
2387 static inline struct super_block *freeze_bdev(struct block_device *sb)
2392 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2397 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2401 static inline int sb_is_blkdev_sb(struct super_block *sb)
2406 extern int sync_filesystem(struct super_block *);
2407 extern const struct file_operations def_blk_fops;
2408 extern const struct file_operations def_chr_fops;
2410 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2411 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2412 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2413 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2414 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2416 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2418 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2419 extern int __blkdev_reread_part(struct block_device *bdev);
2420 extern int blkdev_reread_part(struct block_device *bdev);
2423 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2424 extern void bd_unlink_disk_holder(struct block_device *bdev,
2425 struct gendisk *disk);
2427 static inline int bd_link_disk_holder(struct block_device *bdev,
2428 struct gendisk *disk)
2432 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2433 struct gendisk *disk)
2440 #define CHRDEV_MAJOR_HASH_SIZE 255
2441 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2442 extern int register_chrdev_region(dev_t, unsigned, const char *);
2443 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2444 unsigned int count, const char *name,
2445 const struct file_operations *fops);
2446 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2447 unsigned int count, const char *name);
2448 extern void unregister_chrdev_region(dev_t, unsigned);
2449 extern void chrdev_show(struct seq_file *,off_t);
2451 static inline int register_chrdev(unsigned int major, const char *name,
2452 const struct file_operations *fops)
2454 return __register_chrdev(major, 0, 256, name, fops);
2457 static inline void unregister_chrdev(unsigned int major, const char *name)
2459 __unregister_chrdev(major, 0, 256, name);
2462 /* fs/block_dev.c */
2463 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2464 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2467 #define BLKDEV_MAJOR_HASH_SIZE 255
2468 extern const char *__bdevname(dev_t, char *buffer);
2469 extern const char *bdevname(struct block_device *bdev, char *buffer);
2470 extern struct block_device *lookup_bdev(const char *);
2471 extern void blkdev_show(struct seq_file *,off_t);
2474 #define BLKDEV_MAJOR_HASH_SIZE 0
2477 extern void init_special_inode(struct inode *, umode_t, dev_t);
2479 /* Invalid inode operations -- fs/bad_inode.c */
2480 extern void make_bad_inode(struct inode *);
2481 extern int is_bad_inode(struct inode *);
2485 * return READ, READA, or WRITE
2487 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2490 * return data direction, READ or WRITE
2492 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2494 extern void check_disk_size_change(struct gendisk *disk,
2495 struct block_device *bdev);
2496 extern int revalidate_disk(struct gendisk *);
2497 extern int check_disk_change(struct block_device *);
2498 extern int __invalidate_device(struct block_device *, bool);
2499 extern int invalidate_partition(struct gendisk *, int);
2501 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2502 pgoff_t start, pgoff_t end);
2504 static inline void invalidate_remote_inode(struct inode *inode)
2506 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2507 S_ISLNK(inode->i_mode))
2508 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2510 extern int invalidate_inode_pages2(struct address_space *mapping);
2511 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2512 pgoff_t start, pgoff_t end);
2513 extern int write_inode_now(struct inode *, int);
2514 extern int filemap_fdatawrite(struct address_space *);
2515 extern int filemap_flush(struct address_space *);
2516 extern int filemap_fdatawait(struct address_space *);
2517 extern void filemap_fdatawait_keep_errors(struct address_space *);
2518 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2520 extern int filemap_write_and_wait(struct address_space *mapping);
2521 extern int filemap_write_and_wait_range(struct address_space *mapping,
2522 loff_t lstart, loff_t lend);
2523 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2524 loff_t start, loff_t end, int sync_mode);
2525 extern int filemap_fdatawrite_range(struct address_space *mapping,
2526 loff_t start, loff_t end);
2528 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2530 extern int vfs_fsync(struct file *file, int datasync);
2531 static inline int generic_write_sync(struct file *file, loff_t pos, loff_t count)
2533 if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host))
2535 return vfs_fsync_range(file, pos, pos + count - 1,
2536 (file->f_flags & __O_SYNC) ? 0 : 1);
2538 extern void emergency_sync(void);
2539 extern void emergency_remount(void);
2541 extern sector_t bmap(struct inode *, sector_t);
2543 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2544 extern int notify_change2(struct vfsmount *, struct dentry *, struct iattr *, struct inode **);
2545 extern int inode_permission(struct inode *, int);
2546 extern int inode_permission2(struct vfsmount *, struct inode *, int);
2547 extern int __inode_permission(struct inode *, int);
2548 extern int __inode_permission2(struct vfsmount *, struct inode *, int);
2549 extern int generic_permission(struct inode *, int);
2550 extern int __check_sticky(struct inode *dir, struct inode *inode);
2552 static inline bool execute_ok(struct inode *inode)
2554 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2557 static inline void file_start_write(struct file *file)
2559 if (!S_ISREG(file_inode(file)->i_mode))
2561 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2564 static inline bool file_start_write_trylock(struct file *file)
2566 if (!S_ISREG(file_inode(file)->i_mode))
2568 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2571 static inline void file_end_write(struct file *file)
2573 if (!S_ISREG(file_inode(file)->i_mode))
2575 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2579 * get_write_access() gets write permission for a file.
2580 * put_write_access() releases this write permission.
2581 * This is used for regular files.
2582 * We cannot support write (and maybe mmap read-write shared) accesses and
2583 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2584 * can have the following values:
2585 * 0: no writers, no VM_DENYWRITE mappings
2586 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2587 * > 0: (i_writecount) users are writing to the file.
2589 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2590 * except for the cases where we don't hold i_writecount yet. Then we need to
2591 * use {get,deny}_write_access() - these functions check the sign and refuse
2592 * to do the change if sign is wrong.
2594 static inline int get_write_access(struct inode *inode)
2596 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2598 static inline int deny_write_access(struct file *file)
2600 struct inode *inode = file_inode(file);
2601 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2603 static inline void put_write_access(struct inode * inode)
2605 atomic_dec(&inode->i_writecount);
2607 static inline void allow_write_access(struct file *file)
2610 atomic_inc(&file_inode(file)->i_writecount);
2612 static inline bool inode_is_open_for_write(const struct inode *inode)
2614 return atomic_read(&inode->i_writecount) > 0;
2618 static inline void i_readcount_dec(struct inode *inode)
2620 BUG_ON(!atomic_read(&inode->i_readcount));
2621 atomic_dec(&inode->i_readcount);
2623 static inline void i_readcount_inc(struct inode *inode)
2625 atomic_inc(&inode->i_readcount);
2628 static inline void i_readcount_dec(struct inode *inode)
2632 static inline void i_readcount_inc(struct inode *inode)
2637 extern int do_pipe_flags(int *, int);
2639 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2640 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2641 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
2642 extern struct file * open_exec(const char *);
2644 /* fs/dcache.c -- generic fs support functions */
2645 extern int is_subdir(struct dentry *, struct dentry *);
2646 extern int path_is_under(struct path *, struct path *);
2648 extern char *file_path(struct file *, char *, int);
2650 #include <linux/err.h>
2652 /* needed for stackable file system support */
2653 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2655 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2657 extern int inode_init_always(struct super_block *, struct inode *);
2658 extern void inode_init_once(struct inode *);
2659 extern void address_space_init_once(struct address_space *mapping);
2660 extern struct inode * igrab(struct inode *);
2661 extern ino_t iunique(struct super_block *, ino_t);
2662 extern int inode_needs_sync(struct inode *inode);
2663 extern int generic_delete_inode(struct inode *inode);
2664 static inline int generic_drop_inode(struct inode *inode)
2666 return !inode->i_nlink || inode_unhashed(inode);
2669 extern struct inode *ilookup5_nowait(struct super_block *sb,
2670 unsigned long hashval, int (*test)(struct inode *, void *),
2672 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2673 int (*test)(struct inode *, void *), void *data);
2674 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2676 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2677 extern struct inode * iget_locked(struct super_block *, unsigned long);
2678 extern struct inode *find_inode_nowait(struct super_block *,
2680 int (*match)(struct inode *,
2681 unsigned long, void *),
2683 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2684 extern int insert_inode_locked(struct inode *);
2685 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2686 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2688 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2690 extern void unlock_new_inode(struct inode *);
2691 extern unsigned int get_next_ino(void);
2692 extern void evict_inodes(struct super_block *sb);
2694 extern void __iget(struct inode * inode);
2695 extern void iget_failed(struct inode *);
2696 extern void clear_inode(struct inode *);
2697 extern void __destroy_inode(struct inode *);
2698 extern struct inode *new_inode_pseudo(struct super_block *sb);
2699 extern struct inode *new_inode(struct super_block *sb);
2700 extern void free_inode_nonrcu(struct inode *inode);
2701 extern int should_remove_suid(struct dentry *);
2702 extern int file_remove_privs(struct file *);
2703 extern int dentry_needs_remove_privs(struct dentry *dentry);
2704 static inline int file_needs_remove_privs(struct file *file)
2706 return dentry_needs_remove_privs(file->f_path.dentry);
2709 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2710 static inline void insert_inode_hash(struct inode *inode)
2712 __insert_inode_hash(inode, inode->i_ino);
2715 extern void __remove_inode_hash(struct inode *);
2716 static inline void remove_inode_hash(struct inode *inode)
2718 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2719 __remove_inode_hash(inode);
2722 extern void inode_sb_list_add(struct inode *inode);
2725 extern blk_qc_t submit_bio(int, struct bio *);
2726 extern int bdev_read_only(struct block_device *);
2728 extern int set_blocksize(struct block_device *, int);
2729 extern int sb_set_blocksize(struct super_block *, int);
2730 extern int sb_min_blocksize(struct super_block *, int);
2732 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2733 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2734 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2735 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2736 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2737 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2738 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *, loff_t);
2739 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2741 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos);
2742 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos);
2744 /* fs/block_dev.c */
2745 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2746 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2747 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2749 extern void block_sync_page(struct page *page);
2752 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2753 struct pipe_inode_info *, size_t, unsigned int);
2754 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2755 struct pipe_inode_info *, size_t, unsigned int);
2756 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2757 struct file *, loff_t *, size_t, unsigned int);
2758 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2759 struct file *out, loff_t *, size_t len, unsigned int flags);
2760 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2761 loff_t *opos, size_t len, unsigned int flags);
2765 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2766 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2767 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2768 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2769 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2770 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2771 int whence, loff_t maxsize, loff_t eof);
2772 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2773 int whence, loff_t size);
2774 extern int generic_file_open(struct inode * inode, struct file * filp);
2775 extern int nonseekable_open(struct inode * inode, struct file * filp);
2776 extern int stream_open(struct inode * inode, struct file * filp);
2779 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2780 loff_t file_offset);
2783 /* need locking between buffered and direct access */
2786 /* filesystem does not support filling holes */
2787 DIO_SKIP_HOLES = 0x02,
2789 /* filesystem can handle aio writes beyond i_size */
2790 DIO_ASYNC_EXTEND = 0x04,
2792 /* inode/fs/bdev does not need truncate protection */
2793 DIO_SKIP_DIO_COUNT = 0x08,
2796 void dio_end_io(struct bio *bio, int error);
2798 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2799 struct block_device *bdev, struct iov_iter *iter,
2800 loff_t offset, get_block_t get_block,
2801 dio_iodone_t end_io, dio_submit_t submit_io,
2804 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2805 struct inode *inode,
2806 struct iov_iter *iter, loff_t offset,
2807 get_block_t get_block)
2809 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
2810 offset, get_block, NULL, NULL,
2811 DIO_LOCKING | DIO_SKIP_HOLES);
2815 void inode_dio_wait(struct inode *inode);
2818 * inode_dio_begin - signal start of a direct I/O requests
2819 * @inode: inode the direct I/O happens on
2821 * This is called once we've finished processing a direct I/O request,
2822 * and is used to wake up callers waiting for direct I/O to be quiesced.
2824 static inline void inode_dio_begin(struct inode *inode)
2826 atomic_inc(&inode->i_dio_count);
2830 * inode_dio_end - signal finish of a direct I/O requests
2831 * @inode: inode the direct I/O happens on
2833 * This is called once we've finished processing a direct I/O request,
2834 * and is used to wake up callers waiting for direct I/O to be quiesced.
2836 static inline void inode_dio_end(struct inode *inode)
2838 if (atomic_dec_and_test(&inode->i_dio_count))
2839 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
2842 struct inode *dio_bio_get_inode(struct bio *bio);
2844 extern void inode_set_flags(struct inode *inode, unsigned int flags,
2847 extern const struct file_operations generic_ro_fops;
2849 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2851 extern int readlink_copy(char __user *, int, const char *);
2852 extern int page_readlink(struct dentry *, char __user *, int);
2853 extern const char *page_follow_link_light(struct dentry *, void **);
2854 extern void page_put_link(struct inode *, void *);
2855 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2857 extern int page_symlink(struct inode *inode, const char *symname, int len);
2858 extern const struct inode_operations page_symlink_inode_operations;
2859 extern void kfree_put_link(struct inode *, void *);
2860 extern void free_page_put_link(struct inode *, void *);
2861 extern int generic_readlink(struct dentry *, char __user *, int);
2862 extern void generic_fillattr(struct inode *, struct kstat *);
2863 int vfs_getattr_nosec(struct path *path, struct kstat *stat);
2864 extern int vfs_getattr(struct path *, struct kstat *);
2865 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2866 void inode_add_bytes(struct inode *inode, loff_t bytes);
2867 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2868 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2869 loff_t inode_get_bytes(struct inode *inode);
2870 void inode_set_bytes(struct inode *inode, loff_t bytes);
2871 const char *simple_follow_link(struct dentry *, void **);
2872 extern const struct inode_operations simple_symlink_inode_operations;
2874 extern int iterate_dir(struct file *, struct dir_context *);
2876 extern int vfs_stat(const char __user *, struct kstat *);
2877 extern int vfs_lstat(const char __user *, struct kstat *);
2878 extern int vfs_fstat(unsigned int, struct kstat *);
2879 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2881 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2883 extern int __generic_block_fiemap(struct inode *inode,
2884 struct fiemap_extent_info *fieinfo,
2885 loff_t start, loff_t len,
2886 get_block_t *get_block);
2887 extern int generic_block_fiemap(struct inode *inode,
2888 struct fiemap_extent_info *fieinfo, u64 start,
2889 u64 len, get_block_t *get_block);
2891 extern void get_filesystem(struct file_system_type *fs);
2892 extern void put_filesystem(struct file_system_type *fs);
2893 extern struct file_system_type *get_fs_type(const char *name);
2894 extern struct super_block *get_super(struct block_device *);
2895 extern struct super_block *get_super_thawed(struct block_device *);
2896 extern struct super_block *get_active_super(struct block_device *bdev);
2897 extern void drop_super(struct super_block *sb);
2898 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2899 extern void iterate_supers_type(struct file_system_type *,
2900 void (*)(struct super_block *, void *), void *);
2902 extern int dcache_dir_open(struct inode *, struct file *);
2903 extern int dcache_dir_close(struct inode *, struct file *);
2904 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2905 extern int dcache_readdir(struct file *, struct dir_context *);
2906 extern int simple_setattr(struct dentry *, struct iattr *);
2907 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2908 extern int simple_statfs(struct dentry *, struct kstatfs *);
2909 extern int simple_open(struct inode *inode, struct file *file);
2910 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2911 extern int simple_unlink(struct inode *, struct dentry *);
2912 extern int simple_rmdir(struct inode *, struct dentry *);
2913 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2914 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2915 extern int simple_empty(struct dentry *);
2916 extern int simple_readpage(struct file *file, struct page *page);
2917 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2918 loff_t pos, unsigned len, unsigned flags,
2919 struct page **pagep, void **fsdata);
2920 extern int simple_write_end(struct file *file, struct address_space *mapping,
2921 loff_t pos, unsigned len, unsigned copied,
2922 struct page *page, void *fsdata);
2923 extern int always_delete_dentry(const struct dentry *);
2924 extern struct inode *alloc_anon_inode(struct super_block *);
2925 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
2926 extern const struct dentry_operations simple_dentry_operations;
2928 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2929 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2930 extern const struct file_operations simple_dir_operations;
2931 extern const struct inode_operations simple_dir_inode_operations;
2932 extern void make_empty_dir_inode(struct inode *inode);
2933 extern bool is_empty_dir_inode(struct inode *inode);
2934 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2935 struct dentry *d_alloc_name(struct dentry *, const char *);
2936 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2937 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2938 extern void simple_release_fs(struct vfsmount **mount, int *count);
2940 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2941 loff_t *ppos, const void *from, size_t available);
2942 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2943 const void __user *from, size_t count);
2945 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
2946 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2948 extern int generic_check_addressable(unsigned, u64);
2950 #ifdef CONFIG_MIGRATION
2951 extern int buffer_migrate_page(struct address_space *,
2952 struct page *, struct page *,
2955 #define buffer_migrate_page NULL
2958 extern int inode_change_ok(const struct inode *, struct iattr *);
2959 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2960 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2962 extern int file_update_time(struct file *file);
2964 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2965 extern void save_mount_options(struct super_block *sb, char *options);
2966 extern void replace_mount_options(struct super_block *sb, char *options);
2968 static inline bool io_is_direct(struct file *filp)
2970 return (filp->f_flags & O_DIRECT) || IS_DAX(file_inode(filp));
2973 static inline int iocb_flags(struct file *file)
2976 if (file->f_flags & O_APPEND)
2978 if (io_is_direct(file))
2983 static inline ino_t parent_ino(struct dentry *dentry)
2988 * Don't strictly need d_lock here? If the parent ino could change
2989 * then surely we'd have a deeper race in the caller?
2991 spin_lock(&dentry->d_lock);
2992 res = dentry->d_parent->d_inode->i_ino;
2993 spin_unlock(&dentry->d_lock);
2997 /* Transaction based IO helpers */
3000 * An argresp is stored in an allocated page and holds the
3001 * size of the argument or response, along with its content
3003 struct simple_transaction_argresp {
3008 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3010 char *simple_transaction_get(struct file *file, const char __user *buf,
3012 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3013 size_t size, loff_t *pos);
3014 int simple_transaction_release(struct inode *inode, struct file *file);
3016 void simple_transaction_set(struct file *file, size_t n);
3019 * simple attribute files
3021 * These attributes behave similar to those in sysfs:
3023 * Writing to an attribute immediately sets a value, an open file can be
3024 * written to multiple times.
3026 * Reading from an attribute creates a buffer from the value that might get
3027 * read with multiple read calls. When the attribute has been read
3028 * completely, no further read calls are possible until the file is opened
3031 * All attributes contain a text representation of a numeric value
3032 * that are accessed with the get() and set() functions.
3034 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3035 static int __fops ## _open(struct inode *inode, struct file *file) \
3037 __simple_attr_check_format(__fmt, 0ull); \
3038 return simple_attr_open(inode, file, __get, __set, __fmt); \
3040 static const struct file_operations __fops = { \
3041 .owner = THIS_MODULE, \
3042 .open = __fops ## _open, \
3043 .release = simple_attr_release, \
3044 .read = simple_attr_read, \
3045 .write = simple_attr_write, \
3046 .llseek = generic_file_llseek, \
3049 static inline __printf(1, 2)
3050 void __simple_attr_check_format(const char *fmt, ...)
3052 /* don't do anything, just let the compiler check the arguments; */
3055 int simple_attr_open(struct inode *inode, struct file *file,
3056 int (*get)(void *, u64 *), int (*set)(void *, u64),
3058 int simple_attr_release(struct inode *inode, struct file *file);
3059 ssize_t simple_attr_read(struct file *file, char __user *buf,
3060 size_t len, loff_t *ppos);
3061 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3062 size_t len, loff_t *ppos);
3065 int proc_nr_files(struct ctl_table *table, int write,
3066 void __user *buffer, size_t *lenp, loff_t *ppos);
3067 int proc_nr_dentry(struct ctl_table *table, int write,
3068 void __user *buffer, size_t *lenp, loff_t *ppos);
3069 int proc_nr_inodes(struct ctl_table *table, int write,
3070 void __user *buffer, size_t *lenp, loff_t *ppos);
3071 int __init get_filesystem_list(char *buf);
3073 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3074 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3076 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3077 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3078 (flag & __FMODE_NONOTIFY)))
3080 static inline int is_sxid(umode_t mode)
3082 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3085 static inline int check_sticky(struct inode *dir, struct inode *inode)
3087 if (!(dir->i_mode & S_ISVTX))
3090 return __check_sticky(dir, inode);
3093 static inline void inode_has_no_xattr(struct inode *inode)
3095 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
3096 inode->i_flags |= S_NOSEC;
3099 static inline bool is_root_inode(struct inode *inode)
3101 return inode == inode->i_sb->s_root->d_inode;
3104 static inline bool dir_emit(struct dir_context *ctx,
3105 const char *name, int namelen,
3106 u64 ino, unsigned type)
3108 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3110 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3112 return ctx->actor(ctx, ".", 1, ctx->pos,
3113 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3115 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3117 return ctx->actor(ctx, "..", 2, ctx->pos,
3118 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3120 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3122 if (ctx->pos == 0) {
3123 if (!dir_emit_dot(file, ctx))
3127 if (ctx->pos == 1) {
3128 if (!dir_emit_dotdot(file, ctx))
3134 static inline bool dir_relax(struct inode *inode)
3136 inode_unlock(inode);
3138 return !IS_DEADDIR(inode);
3141 extern bool path_noexec(const struct path *path);
3142 extern void inode_nohighmem(struct inode *inode);
3144 int vfs_ioc_setflags_prepare(struct inode *inode, unsigned int oldflags,
3145 unsigned int flags);
3147 #endif /* _LINUX_FS_H */