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 was opened by fanotify and shouldn't generate fanotify events */
147 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
149 /* File is capable of returning -EAGAIN if I/O will block */
150 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
153 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
154 * that indicates that they should check the contents of the iovec are
155 * valid, but not check the memory that the iovec elements
158 #define CHECK_IOVEC_ONLY -1
161 * The below are the various read and write types that we support. Some of
162 * them include behavioral modifiers that send information down to the
163 * block layer and IO scheduler. Terminology:
165 * The block layer uses device plugging to defer IO a little bit, in
166 * the hope that we will see more IO very shortly. This increases
167 * coalescing of adjacent IO and thus reduces the number of IOs we
168 * have to send to the device. It also allows for better queuing,
169 * if the IO isn't mergeable. If the caller is going to be waiting
170 * for the IO, then he must ensure that the device is unplugged so
171 * that the IO is dispatched to the driver.
173 * All IO is handled async in Linux. This is fine for background
174 * writes, but for reads or writes that someone waits for completion
175 * on, we want to notify the block layer and IO scheduler so that they
176 * know about it. That allows them to make better scheduling
177 * decisions. So when the below references 'sync' and 'async', it
178 * is referencing this priority hint.
180 * With that in mind, the available types are:
182 * READ A normal read operation. Device will be plugged.
183 * READ_SYNC A synchronous read. Device is not plugged, caller can
184 * immediately wait on this read without caring about
186 * READA Used for read-ahead operations. Lower priority, and the
187 * block layer could (in theory) choose to ignore this
188 * request if it runs into resource problems.
189 * WRITE A normal async write. Device will be plugged.
190 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
191 * the hint that someone will be waiting on this IO
192 * shortly. The write equivalent of READ_SYNC.
193 * WRITE_ODIRECT Special case write for O_DIRECT only.
194 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
195 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
196 * non-volatile media on completion.
197 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
198 * by a cache flush and data is guaranteed to be on
199 * non-volatile media on completion.
202 #define RW_MASK REQ_WRITE
203 #define RWA_MASK REQ_RAHEAD
206 #define WRITE RW_MASK
207 #define READA RWA_MASK
209 #define READ_SYNC (READ | REQ_SYNC)
210 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
211 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
212 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
213 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
214 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
217 * Attribute flags. These should be or-ed together to figure out what
220 #define ATTR_MODE (1 << 0)
221 #define ATTR_UID (1 << 1)
222 #define ATTR_GID (1 << 2)
223 #define ATTR_SIZE (1 << 3)
224 #define ATTR_ATIME (1 << 4)
225 #define ATTR_MTIME (1 << 5)
226 #define ATTR_CTIME (1 << 6)
227 #define ATTR_ATIME_SET (1 << 7)
228 #define ATTR_MTIME_SET (1 << 8)
229 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
230 #define ATTR_ATTR_FLAG (1 << 10)
231 #define ATTR_KILL_SUID (1 << 11)
232 #define ATTR_KILL_SGID (1 << 12)
233 #define ATTR_FILE (1 << 13)
234 #define ATTR_KILL_PRIV (1 << 14)
235 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
236 #define ATTR_TIMES_SET (1 << 16)
237 #define ATTR_TOUCH (1 << 17)
240 * Whiteout is represented by a char device. The following constants define the
241 * mode and device number to use.
243 #define WHITEOUT_MODE 0
244 #define WHITEOUT_DEV 0
247 * This is the Inode Attributes structure, used for notify_change(). It
248 * uses the above definitions as flags, to know which values have changed.
249 * Also, in this manner, a Filesystem can look at only the values it cares
250 * about. Basically, these are the attributes that the VFS layer can
251 * request to change from the FS layer.
253 * Derek Atkins <warlord@MIT.EDU> 94-10-20
256 unsigned int ia_valid;
261 struct timespec ia_atime;
262 struct timespec ia_mtime;
263 struct timespec ia_ctime;
266 * Not an attribute, but an auxiliary info for filesystems wanting to
267 * implement an ftruncate() like method. NOTE: filesystem should
268 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
270 struct file *ia_file;
274 * Includes for diskquotas.
276 #include <linux/quota.h>
279 * Maximum number of layers of fs stack. Needs to be limited to
280 * prevent kernel stack overflow
282 #define FILESYSTEM_MAX_STACK_DEPTH 2
285 * enum positive_aop_returns - aop return codes with specific semantics
287 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
288 * completed, that the page is still locked, and
289 * should be considered active. The VM uses this hint
290 * to return the page to the active list -- it won't
291 * be a candidate for writeback again in the near
292 * future. Other callers must be careful to unlock
293 * the page if they get this return. Returned by
296 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
297 * unlocked it and the page might have been truncated.
298 * The caller should back up to acquiring a new page and
299 * trying again. The aop will be taking reasonable
300 * precautions not to livelock. If the caller held a page
301 * reference, it should drop it before retrying. Returned
304 * address_space_operation functions return these large constants to indicate
305 * special semantics to the caller. These are much larger than the bytes in a
306 * page to allow for functions that return the number of bytes operated on in a
310 enum positive_aop_returns {
311 AOP_WRITEPAGE_ACTIVATE = 0x80000,
312 AOP_TRUNCATED_PAGE = 0x80001,
315 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
316 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
317 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
318 * helper code (eg buffer layer)
319 * to clear GFP_FS from alloc */
322 * oh the beauties of C type declarations.
325 struct address_space;
326 struct writeback_control;
329 * Write life time hint values.
332 WRITE_LIFE_NOT_SET = 0,
333 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
334 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
335 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
336 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
337 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
340 #define IOCB_EVENTFD (1 << 0)
341 #define IOCB_APPEND (1 << 1)
342 #define IOCB_DIRECT (1 << 2)
343 #define IOCB_NOWAIT (1 << 7)
346 struct file *ki_filp;
348 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
351 enum rw_hint ki_hint;
354 static inline bool is_sync_kiocb(struct kiocb *kiocb)
356 return kiocb->ki_complete == NULL;
359 static inline int iocb_flags(struct file *file);
361 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
363 *kiocb = (struct kiocb) {
365 .ki_flags = iocb_flags(filp),
370 * "descriptor" for what we're up to with a read.
371 * This allows us to use the same read code yet
372 * have multiple different users of the data that
373 * we read from a file.
375 * The simplest case just copies the data to user
388 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
389 unsigned long, unsigned long);
391 struct address_space_operations {
392 int (*writepage)(struct page *page, struct writeback_control *wbc);
393 int (*readpage)(struct file *, struct page *);
395 /* Write back some dirty pages from this mapping. */
396 int (*writepages)(struct address_space *, struct writeback_control *);
398 /* Set a page dirty. Return true if this dirtied it */
399 int (*set_page_dirty)(struct page *page);
401 int (*readpages)(struct file *filp, struct address_space *mapping,
402 struct list_head *pages, unsigned nr_pages);
404 int (*write_begin)(struct file *, struct address_space *mapping,
405 loff_t pos, unsigned len, unsigned flags,
406 struct page **pagep, void **fsdata);
407 int (*write_end)(struct file *, struct address_space *mapping,
408 loff_t pos, unsigned len, unsigned copied,
409 struct page *page, void *fsdata);
411 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
412 sector_t (*bmap)(struct address_space *, sector_t);
413 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
414 int (*releasepage) (struct page *, gfp_t);
415 void (*freepage)(struct page *);
416 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter, loff_t offset);
418 * migrate the contents of a page to the specified target. If
419 * migrate_mode is MIGRATE_ASYNC, it must not block.
421 int (*migratepage) (struct address_space *,
422 struct page *, struct page *, enum migrate_mode);
423 int (*launder_page) (struct page *);
424 int (*is_partially_uptodate) (struct page *, unsigned long,
426 void (*is_dirty_writeback) (struct page *, bool *, bool *);
427 int (*error_remove_page)(struct address_space *, struct page *);
429 /* swapfile support */
430 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
432 void (*swap_deactivate)(struct file *file);
435 extern const struct address_space_operations empty_aops;
438 * pagecache_write_begin/pagecache_write_end must be used by general code
439 * to write into the pagecache.
441 int pagecache_write_begin(struct file *, struct address_space *mapping,
442 loff_t pos, unsigned len, unsigned flags,
443 struct page **pagep, void **fsdata);
445 int pagecache_write_end(struct file *, struct address_space *mapping,
446 loff_t pos, unsigned len, unsigned copied,
447 struct page *page, void *fsdata);
449 struct address_space {
450 struct inode *host; /* owner: inode, block_device */
451 struct radix_tree_root page_tree; /* radix tree of all pages */
452 spinlock_t tree_lock; /* and lock protecting it */
453 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
454 struct rb_root i_mmap; /* tree of private and shared mappings */
455 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
456 /* Protected by tree_lock together with the radix tree */
457 unsigned long nrpages; /* number of total pages */
458 unsigned long nrshadows; /* number of shadow entries */
459 pgoff_t writeback_index;/* writeback starts here */
460 const struct address_space_operations *a_ops; /* methods */
461 unsigned long flags; /* error bits/gfp mask */
462 spinlock_t private_lock; /* for use by the address_space */
463 struct list_head private_list; /* ditto */
464 void *private_data; /* ditto */
465 } __attribute__((aligned(sizeof(long))));
467 * On most architectures that alignment is already the case; but
468 * must be enforced here for CRIS, to let the least significant bit
469 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
471 struct request_queue;
473 struct block_device {
474 dev_t bd_dev; /* not a kdev_t - it's a search key */
476 struct inode * bd_inode; /* will die */
477 struct super_block * bd_super;
478 struct mutex bd_mutex; /* open/close mutex */
479 struct list_head bd_inodes;
483 bool bd_write_holder;
485 struct list_head bd_holder_disks;
487 struct block_device * bd_contains;
488 unsigned bd_block_size;
489 struct hd_struct * bd_part;
490 /* number of times partitions within this device have been opened. */
491 unsigned bd_part_count;
493 struct gendisk * bd_disk;
494 struct request_queue * bd_queue;
495 struct list_head bd_list;
497 * Private data. You must have bd_claim'ed the block_device
498 * to use this. NOTE: bd_claim allows an owner to claim
499 * the same device multiple times, the owner must take special
500 * care to not mess up bd_private for that case.
502 unsigned long bd_private;
504 /* The counter of freeze processes */
505 int bd_fsfreeze_count;
506 /* Mutex for freeze */
507 struct mutex bd_fsfreeze_mutex;
511 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
514 #define PAGECACHE_TAG_DIRTY 0
515 #define PAGECACHE_TAG_WRITEBACK 1
516 #define PAGECACHE_TAG_TOWRITE 2
518 int mapping_tagged(struct address_space *mapping, int tag);
520 static inline void i_mmap_lock_write(struct address_space *mapping)
522 down_write(&mapping->i_mmap_rwsem);
525 static inline void i_mmap_unlock_write(struct address_space *mapping)
527 up_write(&mapping->i_mmap_rwsem);
530 static inline void i_mmap_lock_read(struct address_space *mapping)
532 down_read(&mapping->i_mmap_rwsem);
535 static inline void i_mmap_unlock_read(struct address_space *mapping)
537 up_read(&mapping->i_mmap_rwsem);
541 * Might pages of this file be mapped into userspace?
543 static inline int mapping_mapped(struct address_space *mapping)
545 return !RB_EMPTY_ROOT(&mapping->i_mmap);
549 * Might pages of this file have been modified in userspace?
550 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
551 * marks vma as VM_SHARED if it is shared, and the file was opened for
552 * writing i.e. vma may be mprotected writable even if now readonly.
554 * If i_mmap_writable is negative, no new writable mappings are allowed. You
555 * can only deny writable mappings, if none exists right now.
557 static inline int mapping_writably_mapped(struct address_space *mapping)
559 return atomic_read(&mapping->i_mmap_writable) > 0;
562 static inline int mapping_map_writable(struct address_space *mapping)
564 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
568 static inline void mapping_unmap_writable(struct address_space *mapping)
570 atomic_dec(&mapping->i_mmap_writable);
573 static inline int mapping_deny_writable(struct address_space *mapping)
575 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
579 static inline void mapping_allow_writable(struct address_space *mapping)
581 atomic_inc(&mapping->i_mmap_writable);
585 * Use sequence counter to get consistent i_size on 32-bit processors.
587 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
588 #include <linux/seqlock.h>
589 #define __NEED_I_SIZE_ORDERED
590 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
592 #define i_size_ordered_init(inode) do { } while (0)
596 #define ACL_NOT_CACHED ((void *)(-1))
598 #define IOP_FASTPERM 0x0001
599 #define IOP_LOOKUP 0x0002
600 #define IOP_NOFOLLOW 0x0004
603 * Keep mostly read-only and often accessed (especially for
604 * the RCU path lookup and 'stat' data) fields at the beginning
605 * of the 'struct inode'
609 unsigned short i_opflags;
612 unsigned int i_flags;
614 #ifdef CONFIG_FS_POSIX_ACL
615 struct posix_acl *i_acl;
616 struct posix_acl *i_default_acl;
619 const struct inode_operations *i_op;
620 struct super_block *i_sb;
621 struct address_space *i_mapping;
623 #ifdef CONFIG_SECURITY
627 /* Stat data, not accessed from path walking */
630 * Filesystems may only read i_nlink directly. They shall use the
631 * following functions for modification:
633 * (set|clear|inc|drop)_nlink
634 * inode_(inc|dec)_link_count
637 const unsigned int i_nlink;
638 unsigned int __i_nlink;
642 struct timespec i_atime;
643 struct timespec i_mtime;
644 struct timespec i_ctime;
645 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
646 unsigned short i_bytes;
647 unsigned int i_blkbits;
648 enum rw_hint i_write_hint;
651 #ifdef __NEED_I_SIZE_ORDERED
652 seqcount_t i_size_seqcount;
656 unsigned long i_state;
657 struct mutex i_mutex;
659 unsigned long dirtied_when; /* jiffies of first dirtying */
660 unsigned long dirtied_time_when;
662 struct hlist_node i_hash;
663 struct list_head i_io_list; /* backing dev IO list */
664 #ifdef CONFIG_CGROUP_WRITEBACK
665 struct bdi_writeback *i_wb; /* the associated cgroup wb */
667 /* foreign inode detection, see wbc_detach_inode() */
669 u16 i_wb_frn_avg_time;
670 u16 i_wb_frn_history;
672 struct list_head i_lru; /* inode LRU list */
673 struct list_head i_sb_list;
675 struct hlist_head i_dentry;
676 struct rcu_head i_rcu;
680 atomic_t i_dio_count;
681 atomic_t i_writecount;
683 atomic_t i_readcount; /* struct files open RO */
685 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
686 struct file_lock_context *i_flctx;
687 struct address_space i_data;
688 struct list_head i_devices;
690 struct pipe_inode_info *i_pipe;
691 struct block_device *i_bdev;
698 #ifdef CONFIG_FSNOTIFY
699 __u32 i_fsnotify_mask; /* all events this inode cares about */
700 struct hlist_head i_fsnotify_marks;
703 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
704 struct fscrypt_info *i_crypt_info;
706 void *i_private; /* fs or device private pointer */
709 static inline unsigned int i_blocksize(const struct inode *node)
711 return (1 << node->i_blkbits);
714 static inline int inode_unhashed(struct inode *inode)
716 return hlist_unhashed(&inode->i_hash);
720 * inode->i_mutex nesting subclasses for the lock validator:
722 * 0: the object of the current VFS operation
726 * 4: second non-directory
727 * 5: second parent (when locking independent directories in rename)
729 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
730 * non-directories at once.
732 * The locking order between these classes is
733 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
735 enum inode_i_mutex_lock_class
745 static inline void inode_lock(struct inode *inode)
747 mutex_lock(&inode->i_mutex);
750 static inline void inode_unlock(struct inode *inode)
752 mutex_unlock(&inode->i_mutex);
755 static inline int inode_trylock(struct inode *inode)
757 return mutex_trylock(&inode->i_mutex);
760 static inline int inode_is_locked(struct inode *inode)
762 return mutex_is_locked(&inode->i_mutex);
765 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
767 mutex_lock_nested(&inode->i_mutex, subclass);
770 void lock_two_nondirectories(struct inode *, struct inode*);
771 void unlock_two_nondirectories(struct inode *, struct inode*);
774 * NOTE: in a 32bit arch with a preemptable kernel and
775 * an UP compile the i_size_read/write must be atomic
776 * with respect to the local cpu (unlike with preempt disabled),
777 * but they don't need to be atomic with respect to other cpus like in
778 * true SMP (so they need either to either locally disable irq around
779 * the read or for example on x86 they can be still implemented as a
780 * cmpxchg8b without the need of the lock prefix). For SMP compiles
781 * and 64bit archs it makes no difference if preempt is enabled or not.
783 static inline loff_t i_size_read(const struct inode *inode)
785 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
790 seq = read_seqcount_begin(&inode->i_size_seqcount);
791 i_size = inode->i_size;
792 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
794 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
798 i_size = inode->i_size;
802 return inode->i_size;
807 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
808 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
809 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
811 static inline void i_size_write(struct inode *inode, loff_t i_size)
813 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
815 write_seqcount_begin(&inode->i_size_seqcount);
816 inode->i_size = i_size;
817 write_seqcount_end(&inode->i_size_seqcount);
819 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
821 inode->i_size = i_size;
824 inode->i_size = i_size;
828 /* Helper functions so that in most cases filesystems will
829 * not need to deal directly with kuid_t and kgid_t and can
830 * instead deal with the raw numeric values that are stored
833 static inline uid_t i_uid_read(const struct inode *inode)
835 return from_kuid(&init_user_ns, inode->i_uid);
838 static inline gid_t i_gid_read(const struct inode *inode)
840 return from_kgid(&init_user_ns, inode->i_gid);
843 static inline void i_uid_write(struct inode *inode, uid_t uid)
845 inode->i_uid = make_kuid(&init_user_ns, uid);
848 static inline void i_gid_write(struct inode *inode, gid_t gid)
850 inode->i_gid = make_kgid(&init_user_ns, gid);
853 static inline unsigned iminor(const struct inode *inode)
855 return MINOR(inode->i_rdev);
858 static inline unsigned imajor(const struct inode *inode)
860 return MAJOR(inode->i_rdev);
863 extern struct block_device *I_BDEV(struct inode *inode);
866 rwlock_t lock; /* protects pid, uid, euid fields */
867 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
868 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
869 kuid_t uid, euid; /* uid/euid of process setting the owner */
870 int signum; /* posix.1b rt signal to be delivered on IO */
874 * Track a single file's readahead state
876 struct file_ra_state {
877 pgoff_t start; /* where readahead started */
878 unsigned int size; /* # of readahead pages */
879 unsigned int async_size; /* do asynchronous readahead when
880 there are only # of pages ahead */
882 unsigned int ra_pages; /* Maximum readahead window */
883 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
884 loff_t prev_pos; /* Cache last read() position */
888 * Check if @index falls in the readahead windows.
890 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
892 return (index >= ra->start &&
893 index < ra->start + ra->size);
898 struct llist_node fu_llist;
899 struct rcu_head fu_rcuhead;
902 struct inode *f_inode; /* cached value */
903 const struct file_operations *f_op;
906 * Protects f_ep_links, f_flags.
907 * Must not be taken from IRQ context.
910 atomic_long_t f_count;
911 unsigned int f_flags;
913 struct mutex f_pos_lock;
915 struct fown_struct f_owner;
916 const struct cred *f_cred;
917 struct file_ra_state f_ra;
920 #ifdef CONFIG_SECURITY
923 /* needed for tty driver, and maybe others */
927 /* Used by fs/eventpoll.c to link all the hooks to this file */
928 struct list_head f_ep_links;
929 struct list_head f_tfile_llink;
930 #endif /* #ifdef CONFIG_EPOLL */
931 struct address_space *f_mapping;
932 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
937 /* file identifier */
938 unsigned char f_handle[0];
941 static inline struct file *get_file(struct file *f)
943 atomic_long_inc(&f->f_count);
946 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
947 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
948 #define file_count(x) atomic_long_read(&(x)->f_count)
950 #define MAX_NON_LFS ((1UL<<31) - 1)
952 /* Page cache limit. The filesystems should put that into their s_maxbytes
953 limits, otherwise bad things can happen in VM. */
954 #if BITS_PER_LONG==32
955 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
956 #elif BITS_PER_LONG==64
957 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
962 #define FL_DELEG 4 /* NFSv4 delegation */
963 #define FL_ACCESS 8 /* not trying to lock, just looking */
964 #define FL_EXISTS 16 /* when unlocking, test for existence */
965 #define FL_LEASE 32 /* lease held on this file */
966 #define FL_CLOSE 64 /* unlock on close */
967 #define FL_SLEEP 128 /* A blocking lock */
968 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
969 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
970 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
971 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
974 * Special return value from posix_lock_file() and vfs_lock_file() for
975 * asynchronous locking.
977 #define FILE_LOCK_DEFERRED 1
979 /* legacy typedef, should eventually be removed */
980 typedef void *fl_owner_t;
984 struct file_lock_operations {
985 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
986 void (*fl_release_private)(struct file_lock *);
989 struct lock_manager_operations {
990 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
991 unsigned long (*lm_owner_key)(struct file_lock *);
992 fl_owner_t (*lm_get_owner)(fl_owner_t);
993 void (*lm_put_owner)(fl_owner_t);
994 void (*lm_notify)(struct file_lock *); /* unblock callback */
995 int (*lm_grant)(struct file_lock *, int);
996 bool (*lm_break)(struct file_lock *);
997 int (*lm_change)(struct file_lock *, int, struct list_head *);
998 void (*lm_setup)(struct file_lock *, void **);
1001 struct lock_manager {
1002 struct list_head list;
1004 * NFSv4 and up also want opens blocked during the grace period;
1011 void locks_start_grace(struct net *, struct lock_manager *);
1012 void locks_end_grace(struct lock_manager *);
1013 int locks_in_grace(struct net *);
1014 int opens_in_grace(struct net *);
1016 /* that will die - we need it for nfs_lock_info */
1017 #include <linux/nfs_fs_i.h>
1020 * struct file_lock represents a generic "file lock". It's used to represent
1021 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1022 * note that the same struct is used to represent both a request for a lock and
1023 * the lock itself, but the same object is never used for both.
1025 * FIXME: should we create a separate "struct lock_request" to help distinguish
1028 * The varous i_flctx lists are ordered by:
1031 * 2) lock range start
1034 * Obviously, the last two criteria only matter for POSIX locks.
1037 struct file_lock *fl_next; /* singly linked list for this inode */
1038 struct list_head fl_list; /* link into file_lock_context */
1039 struct hlist_node fl_link; /* node in global lists */
1040 struct list_head fl_block; /* circular list of blocked processes */
1041 fl_owner_t fl_owner;
1042 unsigned int fl_flags;
1043 unsigned char fl_type;
1044 unsigned int fl_pid;
1045 int fl_link_cpu; /* what cpu's list is this on? */
1046 struct pid *fl_nspid;
1047 wait_queue_head_t fl_wait;
1048 struct file *fl_file;
1052 struct fasync_struct * fl_fasync; /* for lease break notifications */
1053 /* for lease breaks: */
1054 unsigned long fl_break_time;
1055 unsigned long fl_downgrade_time;
1057 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1058 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1060 struct nfs_lock_info nfs_fl;
1061 struct nfs4_lock_info nfs4_fl;
1063 struct list_head link; /* link in AFS vnode's pending_locks list */
1064 int state; /* state of grant or error if -ve */
1069 struct file_lock_context {
1070 spinlock_t flc_lock;
1071 struct list_head flc_flock;
1072 struct list_head flc_posix;
1073 struct list_head flc_lease;
1076 /* The following constant reflects the upper bound of the file/locking space */
1078 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1079 #define OFFSET_MAX INT_LIMIT(loff_t)
1080 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1083 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1085 #ifdef CONFIG_FILE_LOCKING
1086 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
1087 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1088 struct flock __user *);
1090 #if BITS_PER_LONG == 32
1091 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
1092 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1093 struct flock64 __user *);
1096 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1097 extern int fcntl_getlease(struct file *filp);
1100 void locks_free_lock_context(struct file_lock_context *ctx);
1101 void locks_free_lock(struct file_lock *fl);
1102 extern void locks_init_lock(struct file_lock *);
1103 extern struct file_lock * locks_alloc_lock(void);
1104 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1105 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1106 extern void locks_remove_posix(struct file *, fl_owner_t);
1107 extern void locks_remove_file(struct file *);
1108 extern void locks_release_private(struct file_lock *);
1109 extern void posix_test_lock(struct file *, struct file_lock *);
1110 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1111 extern int posix_unblock_lock(struct file_lock *);
1112 extern int vfs_test_lock(struct file *, struct file_lock *);
1113 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1114 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1115 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1116 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1117 extern void lease_get_mtime(struct inode *, struct timespec *time);
1118 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1119 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1120 extern int lease_modify(struct file_lock *, int, struct list_head *);
1121 struct files_struct;
1122 extern void show_fd_locks(struct seq_file *f,
1123 struct file *filp, struct files_struct *files);
1124 #else /* !CONFIG_FILE_LOCKING */
1125 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1126 struct flock __user *user)
1131 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1132 unsigned int cmd, struct flock __user *user)
1137 #if BITS_PER_LONG == 32
1138 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1139 struct flock64 __user *user)
1144 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1145 unsigned int cmd, struct flock64 __user *user)
1150 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1155 static inline int fcntl_getlease(struct file *filp)
1161 locks_free_lock_context(struct file_lock_context *ctx)
1165 static inline void locks_init_lock(struct file_lock *fl)
1170 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1175 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1180 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1185 static inline void locks_remove_file(struct file *filp)
1190 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1195 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1196 struct file_lock *conflock)
1201 static inline int posix_unblock_lock(struct file_lock *waiter)
1206 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1211 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1212 struct file_lock *fl, struct file_lock *conf)
1217 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1222 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1227 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1232 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1237 static inline int generic_setlease(struct file *filp, long arg,
1238 struct file_lock **flp, void **priv)
1243 static inline int vfs_setlease(struct file *filp, long arg,
1244 struct file_lock **lease, void **priv)
1249 static inline int lease_modify(struct file_lock *fl, int arg,
1250 struct list_head *dispose)
1255 struct files_struct;
1256 static inline void show_fd_locks(struct seq_file *f,
1257 struct file *filp, struct files_struct *files) {}
1258 #endif /* !CONFIG_FILE_LOCKING */
1260 static inline struct inode *file_inode(const struct file *f)
1265 static inline struct dentry *file_dentry(const struct file *file)
1267 struct dentry *dentry = file->f_path.dentry;
1269 if (unlikely(dentry->d_flags & DCACHE_OP_REAL))
1270 return dentry->d_op->d_real(dentry, file_inode(file));
1275 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1277 return locks_lock_inode_wait(file_inode(filp), fl);
1280 struct fasync_struct {
1284 struct fasync_struct *fa_next; /* singly linked list */
1285 struct file *fa_file;
1286 struct rcu_head fa_rcu;
1289 #define FASYNC_MAGIC 0x4601
1291 /* SMP safe fasync helpers: */
1292 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1293 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1294 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1295 extern struct fasync_struct *fasync_alloc(void);
1296 extern void fasync_free(struct fasync_struct *);
1298 /* can be called from interrupts */
1299 extern void kill_fasync(struct fasync_struct **, int, int);
1301 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1302 extern void f_setown(struct file *filp, unsigned long arg, int force);
1303 extern void f_delown(struct file *filp);
1304 extern pid_t f_getown(struct file *filp);
1305 extern int send_sigurg(struct fown_struct *fown);
1313 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1314 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1315 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1316 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1317 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1320 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1321 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1322 #define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */
1324 /* Possible states of 'frozen' field */
1326 SB_UNFROZEN = 0, /* FS is unfrozen */
1327 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1328 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1329 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1330 * internal threads if needed) */
1331 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1334 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1337 int frozen; /* Is sb frozen? */
1338 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1339 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1342 struct super_block {
1343 struct list_head s_list; /* Keep this first */
1344 dev_t s_dev; /* search index; _not_ kdev_t */
1345 unsigned char s_blocksize_bits;
1346 unsigned long s_blocksize;
1347 loff_t s_maxbytes; /* Max file size */
1348 struct file_system_type *s_type;
1349 const struct super_operations *s_op;
1350 const struct dquot_operations *dq_op;
1351 const struct quotactl_ops *s_qcop;
1352 const struct export_operations *s_export_op;
1353 unsigned long s_flags;
1354 unsigned long s_iflags; /* internal SB_I_* flags */
1355 unsigned long s_magic;
1356 struct dentry *s_root;
1357 struct rw_semaphore s_umount;
1360 #ifdef CONFIG_SECURITY
1363 const struct xattr_handler **s_xattr;
1365 const struct fscrypt_operations *s_cop;
1367 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1368 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1369 struct block_device *s_bdev;
1370 struct backing_dev_info *s_bdi;
1371 struct mtd_info *s_mtd;
1372 struct hlist_node s_instances;
1373 unsigned int s_quota_types; /* Bitmask of supported quota types */
1374 struct quota_info s_dquot; /* Diskquota specific options */
1376 struct sb_writers s_writers;
1378 char s_id[32]; /* Informational name */
1379 u8 s_uuid[16]; /* UUID */
1381 void *s_fs_info; /* Filesystem private info */
1382 unsigned int s_max_links;
1385 /* Granularity of c/m/atime in ns.
1386 Cannot be worse than a second */
1390 * The next field is for VFS *only*. No filesystems have any business
1391 * even looking at it. You had been warned.
1393 struct mutex s_vfs_rename_mutex; /* Kludge */
1396 * Filesystem subtype. If non-empty the filesystem type field
1397 * in /proc/mounts will be "type.subtype"
1402 * Saved mount options for lazy filesystems using
1403 * generic_show_options()
1405 char __rcu *s_options;
1406 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1409 * Saved pool identifier for cleancache (-1 means none)
1411 int cleancache_poolid;
1413 struct shrinker s_shrink; /* per-sb shrinker handle */
1415 /* Number of inodes with nlink == 0 but still referenced */
1416 atomic_long_t s_remove_count;
1418 /* Being remounted read-only */
1419 int s_readonly_remount;
1421 /* AIO completions deferred from interrupt context */
1422 struct workqueue_struct *s_dio_done_wq;
1423 struct hlist_head s_pins;
1426 * Keep the lru lists last in the structure so they always sit on their
1427 * own individual cachelines.
1429 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1430 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1431 struct rcu_head rcu;
1432 struct work_struct destroy_work;
1434 struct mutex s_sync_lock; /* sync serialisation lock */
1437 * Indicates how deep in a filesystem stack this SB is
1441 /* s_inode_list_lock protects s_inodes */
1442 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1443 struct list_head s_inodes; /* all inodes */
1446 extern struct timespec current_fs_time(struct super_block *sb);
1449 * Snapshotting support.
1452 void __sb_end_write(struct super_block *sb, int level);
1453 int __sb_start_write(struct super_block *sb, int level, bool wait);
1455 #define __sb_writers_acquired(sb, lev) \
1456 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1457 #define __sb_writers_release(sb, lev) \
1458 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1461 * sb_end_write - drop write access to a superblock
1462 * @sb: the super we wrote to
1464 * Decrement number of writers to the filesystem. Wake up possible waiters
1465 * wanting to freeze the filesystem.
1467 static inline void sb_end_write(struct super_block *sb)
1469 __sb_end_write(sb, SB_FREEZE_WRITE);
1473 * sb_end_pagefault - drop write access to a superblock from a page fault
1474 * @sb: the super we wrote to
1476 * Decrement number of processes handling write page fault to the filesystem.
1477 * Wake up possible waiters wanting to freeze the filesystem.
1479 static inline void sb_end_pagefault(struct super_block *sb)
1481 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1485 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1486 * @sb: the super we wrote to
1488 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1489 * waiters wanting to freeze the filesystem.
1491 static inline void sb_end_intwrite(struct super_block *sb)
1493 __sb_end_write(sb, SB_FREEZE_FS);
1497 * sb_start_write - get write access to a superblock
1498 * @sb: the super we write to
1500 * When a process wants to write data or metadata to a file system (i.e. dirty
1501 * a page or an inode), it should embed the operation in a sb_start_write() -
1502 * sb_end_write() pair to get exclusion against file system freezing. This
1503 * function increments number of writers preventing freezing. If the file
1504 * system is already frozen, the function waits until the file system is
1507 * Since freeze protection behaves as a lock, users have to preserve
1508 * ordering of freeze protection and other filesystem locks. Generally,
1509 * freeze protection should be the outermost lock. In particular, we have:
1512 * -> i_mutex (write path, truncate, directory ops, ...)
1513 * -> s_umount (freeze_super, thaw_super)
1515 static inline void sb_start_write(struct super_block *sb)
1517 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1520 static inline int sb_start_write_trylock(struct super_block *sb)
1522 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1526 * sb_start_pagefault - get write access to a superblock from a page fault
1527 * @sb: the super we write to
1529 * When a process starts handling write page fault, it should embed the
1530 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1531 * exclusion against file system freezing. This is needed since the page fault
1532 * is going to dirty a page. This function increments number of running page
1533 * faults preventing freezing. If the file system is already frozen, the
1534 * function waits until the file system is thawed.
1536 * Since page fault freeze protection behaves as a lock, users have to preserve
1537 * ordering of freeze protection and other filesystem locks. It is advised to
1538 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1539 * handling code implies lock dependency:
1542 * -> sb_start_pagefault
1544 static inline void sb_start_pagefault(struct super_block *sb)
1546 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1550 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1551 * @sb: the super we write to
1553 * This is the third level of protection against filesystem freezing. It is
1554 * free for use by a filesystem. The only requirement is that it must rank
1555 * below sb_start_pagefault.
1557 * For example filesystem can call sb_start_intwrite() when starting a
1558 * transaction which somewhat eases handling of freezing for internal sources
1559 * of filesystem changes (internal fs threads, discarding preallocation on file
1562 static inline void sb_start_intwrite(struct super_block *sb)
1564 __sb_start_write(sb, SB_FREEZE_FS, true);
1568 extern bool inode_owner_or_capable(const struct inode *inode);
1571 * VFS helper functions..
1573 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1574 extern int vfs_create2(struct vfsmount *, struct inode *, struct dentry *, umode_t, bool);
1575 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1576 extern int vfs_mkdir2(struct vfsmount *, struct inode *, struct dentry *, umode_t);
1577 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1578 extern int vfs_mknod2(struct vfsmount *, struct inode *, struct dentry *, umode_t, dev_t);
1579 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1580 extern int vfs_symlink2(struct vfsmount *, struct inode *, struct dentry *, const char *);
1581 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1582 extern int vfs_link2(struct vfsmount *, struct dentry *, struct inode *, struct dentry *, struct inode **);
1583 extern int vfs_rmdir(struct inode *, struct dentry *);
1584 extern int vfs_rmdir2(struct vfsmount *, struct inode *, struct dentry *);
1585 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1586 extern int vfs_unlink2(struct vfsmount *, struct inode *, struct dentry *, struct inode **);
1587 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1588 extern int vfs_rename2(struct vfsmount *, struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1589 extern int vfs_whiteout(struct inode *, struct dentry *);
1592 * VFS dentry helper functions.
1594 extern void dentry_unhash(struct dentry *dentry);
1597 * VFS file helper functions.
1599 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1602 * VFS FS_IOC_FIEMAP helper definitions.
1604 struct fiemap_extent_info {
1605 unsigned int fi_flags; /* Flags as passed from user */
1606 unsigned int fi_extents_mapped; /* Number of mapped extents */
1607 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1608 struct fiemap_extent __user *fi_extents_start; /* Start of
1609 fiemap_extent array */
1611 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1612 u64 phys, u64 len, u32 flags);
1613 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1618 * NOTE! These match bits 12..15 of stat.st_mode
1619 * (ie "(i_mode >> 12) & 15").
1621 #define DT_UNKNOWN 0
1632 * This is the "filldir" function type, used by readdir() to let
1633 * the kernel specify what kind of dirent layout it wants to have.
1634 * This allows the kernel to read directories into kernel space or
1635 * to have different dirent layouts depending on the binary type.
1638 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1641 struct dir_context {
1642 const filldir_t actor;
1646 struct block_device_operations;
1648 /* These macros are for out of kernel modules to test that
1649 * the kernel supports the unlocked_ioctl and compat_ioctl
1650 * fields in struct file_operations. */
1651 #define HAVE_COMPAT_IOCTL 1
1652 #define HAVE_UNLOCKED_IOCTL 1
1655 * These flags let !MMU mmap() govern direct device mapping vs immediate
1656 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1658 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1659 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1660 * NOMMU_MAP_READ: Can be mapped for reading
1661 * NOMMU_MAP_WRITE: Can be mapped for writing
1662 * NOMMU_MAP_EXEC: Can be mapped for execution
1664 #define NOMMU_MAP_COPY 0x00000001
1665 #define NOMMU_MAP_DIRECT 0x00000008
1666 #define NOMMU_MAP_READ VM_MAYREAD
1667 #define NOMMU_MAP_WRITE VM_MAYWRITE
1668 #define NOMMU_MAP_EXEC VM_MAYEXEC
1670 #define NOMMU_VMFLAGS \
1671 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1676 struct file_operations {
1677 struct module *owner;
1678 loff_t (*llseek) (struct file *, loff_t, int);
1679 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1680 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1681 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1682 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1683 int (*iterate) (struct file *, struct dir_context *);
1684 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1685 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1686 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1687 int (*mmap) (struct file *, struct vm_area_struct *);
1688 int (*open) (struct inode *, struct file *);
1689 int (*flush) (struct file *, fl_owner_t id);
1690 int (*release) (struct inode *, struct file *);
1691 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1692 int (*aio_fsync) (struct kiocb *, int datasync);
1693 int (*fasync) (int, struct file *, int);
1694 int (*lock) (struct file *, int, struct file_lock *);
1695 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1696 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1697 int (*check_flags)(int);
1698 int (*flock) (struct file *, int, struct file_lock *);
1699 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1700 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1701 int (*setlease)(struct file *, long, struct file_lock **, void **);
1702 long (*fallocate)(struct file *file, int mode, loff_t offset,
1704 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1706 unsigned (*mmap_capabilities)(struct file *);
1710 struct inode_operations {
1711 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1712 const char * (*follow_link) (struct dentry *, void **);
1713 int (*permission) (struct inode *, int);
1714 int (*permission2) (struct vfsmount *, struct inode *, int);
1715 struct posix_acl * (*get_acl)(struct inode *, int);
1717 int (*readlink) (struct dentry *, char __user *,int);
1718 void (*put_link) (struct inode *, void *);
1720 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1721 int (*link) (struct dentry *,struct inode *,struct dentry *);
1722 int (*unlink) (struct inode *,struct dentry *);
1723 int (*symlink) (struct inode *,struct dentry *,const char *);
1724 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1725 int (*rmdir) (struct inode *,struct dentry *);
1726 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1727 int (*rename) (struct inode *, struct dentry *,
1728 struct inode *, struct dentry *);
1729 int (*rename2) (struct inode *, struct dentry *,
1730 struct inode *, struct dentry *, unsigned int);
1731 int (*setattr) (struct dentry *, struct iattr *);
1732 int (*setattr2) (struct vfsmount *, struct dentry *, struct iattr *);
1733 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1734 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1735 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1736 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1737 int (*removexattr) (struct dentry *, const char *);
1738 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1740 int (*update_time)(struct inode *, struct timespec *, int);
1741 int (*atomic_open)(struct inode *, struct dentry *,
1742 struct file *, unsigned open_flag,
1743 umode_t create_mode, int *opened);
1744 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1745 int (*set_acl)(struct inode *, struct posix_acl *, int);
1746 } ____cacheline_aligned;
1748 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1749 unsigned long nr_segs, unsigned long fast_segs,
1750 struct iovec *fast_pointer,
1751 struct iovec **ret_pointer);
1753 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1754 extern ssize_t __vfs_write(struct file *, const char __user *, size_t, loff_t *);
1755 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1756 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1757 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1758 unsigned long, loff_t *);
1759 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1760 unsigned long, loff_t *);
1762 struct super_operations {
1763 struct inode *(*alloc_inode)(struct super_block *sb);
1764 void (*destroy_inode)(struct inode *);
1766 void (*dirty_inode) (struct inode *, int flags);
1767 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1768 int (*drop_inode) (struct inode *);
1769 void (*evict_inode) (struct inode *);
1770 void (*put_super) (struct super_block *);
1771 int (*sync_fs)(struct super_block *sb, int wait);
1772 int (*freeze_super) (struct super_block *);
1773 int (*freeze_fs) (struct super_block *);
1774 int (*thaw_super) (struct super_block *);
1775 int (*unfreeze_fs) (struct super_block *);
1776 int (*statfs) (struct dentry *, struct kstatfs *);
1777 int (*remount_fs) (struct super_block *, int *, char *);
1778 int (*remount_fs2) (struct vfsmount *, struct super_block *, int *, char *);
1779 void *(*clone_mnt_data) (void *);
1780 void (*copy_mnt_data) (void *, void *);
1781 void (*umount_begin) (struct super_block *);
1783 int (*show_options)(struct seq_file *, struct dentry *);
1784 int (*show_options2)(struct vfsmount *,struct seq_file *, struct dentry *);
1785 int (*show_devname)(struct seq_file *, struct dentry *);
1786 int (*show_path)(struct seq_file *, struct dentry *);
1787 int (*show_stats)(struct seq_file *, struct dentry *);
1789 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1790 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1791 struct dquot **(*get_dquots)(struct inode *);
1793 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1794 long (*nr_cached_objects)(struct super_block *,
1795 struct shrink_control *);
1796 long (*free_cached_objects)(struct super_block *,
1797 struct shrink_control *);
1801 * Inode flags - they have no relation to superblock flags now
1803 #define S_SYNC 1 /* Writes are synced at once */
1804 #define S_NOATIME 2 /* Do not update access times */
1805 #define S_APPEND 4 /* Append-only file */
1806 #define S_IMMUTABLE 8 /* Immutable file */
1807 #define S_DEAD 16 /* removed, but still open directory */
1808 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1809 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1810 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1811 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1812 #define S_PRIVATE 512 /* Inode is fs-internal */
1813 #define S_IMA 1024 /* Inode has an associated IMA struct */
1814 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1815 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1816 #ifdef CONFIG_FS_DAX
1817 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1819 #define S_DAX 0 /* Make all the DAX code disappear */
1821 #define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */
1824 * Note that nosuid etc flags are inode-specific: setting some file-system
1825 * flags just means all the inodes inherit those flags by default. It might be
1826 * possible to override it selectively if you really wanted to with some
1827 * ioctl() that is not currently implemented.
1829 * Exception: MS_RDONLY is always applied to the entire file system.
1831 * Unfortunately, it is possible to change a filesystems flags with it mounted
1832 * with files in use. This means that all of the inodes will not have their
1833 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1834 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1836 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1838 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1839 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1840 ((inode)->i_flags & S_SYNC))
1841 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1842 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1843 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1844 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1845 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1847 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1848 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1849 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1850 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1852 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1853 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1854 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1855 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1856 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1857 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1858 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1859 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1860 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
1862 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1863 (inode)->i_rdev == WHITEOUT_DEV)
1866 * Inode state bits. Protected by inode->i_lock
1868 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1869 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1871 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1872 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1873 * various stages of removing an inode.
1875 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1877 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1878 * fdatasync(). i_atime is the usual cause.
1879 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1880 * these changes separately from I_DIRTY_SYNC so that we
1881 * don't have to write inode on fdatasync() when only
1882 * mtime has changed in it.
1883 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1884 * I_NEW Serves as both a mutex and completion notification.
1885 * New inodes set I_NEW. If two processes both create
1886 * the same inode, one of them will release its inode and
1887 * wait for I_NEW to be released before returning.
1888 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1889 * also cause waiting on I_NEW, without I_NEW actually
1890 * being set. find_inode() uses this to prevent returning
1891 * nearly-dead inodes.
1892 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1893 * is zero. I_FREEING must be set when I_WILL_FREE is
1895 * I_FREEING Set when inode is about to be freed but still has dirty
1896 * pages or buffers attached or the inode itself is still
1898 * I_CLEAR Added by clear_inode(). In this state the inode is
1899 * clean and can be destroyed. Inode keeps I_FREEING.
1901 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1902 * prohibited for many purposes. iget() must wait for
1903 * the inode to be completely released, then create it
1904 * anew. Other functions will just ignore such inodes,
1905 * if appropriate. I_NEW is used for waiting.
1907 * I_SYNC Writeback of inode is running. The bit is set during
1908 * data writeback, and cleared with a wakeup on the bit
1909 * address once it is done. The bit is also used to pin
1910 * the inode in memory for flusher thread.
1912 * I_REFERENCED Marks the inode as recently references on the LRU list.
1914 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1916 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1917 * synchronize competing switching instances and to tell
1918 * wb stat updates to grab mapping->tree_lock. See
1919 * inode_switch_wb_work_fn() for details.
1921 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1923 #define I_DIRTY_SYNC (1 << 0)
1924 #define I_DIRTY_DATASYNC (1 << 1)
1925 #define I_DIRTY_PAGES (1 << 2)
1927 #define I_NEW (1 << __I_NEW)
1928 #define I_WILL_FREE (1 << 4)
1929 #define I_FREEING (1 << 5)
1930 #define I_CLEAR (1 << 6)
1932 #define I_SYNC (1 << __I_SYNC)
1933 #define I_REFERENCED (1 << 8)
1934 #define __I_DIO_WAKEUP 9
1935 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1936 #define I_LINKABLE (1 << 10)
1937 #define I_DIRTY_TIME (1 << 11)
1938 #define __I_DIRTY_TIME_EXPIRED 12
1939 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1940 #define I_WB_SWITCH (1 << 13)
1942 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1943 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1945 extern void __mark_inode_dirty(struct inode *, int);
1946 static inline void mark_inode_dirty(struct inode *inode)
1948 __mark_inode_dirty(inode, I_DIRTY);
1951 static inline void mark_inode_dirty_sync(struct inode *inode)
1953 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1956 extern void inc_nlink(struct inode *inode);
1957 extern void drop_nlink(struct inode *inode);
1958 extern void clear_nlink(struct inode *inode);
1959 extern void set_nlink(struct inode *inode, unsigned int nlink);
1961 static inline void inode_inc_link_count(struct inode *inode)
1964 mark_inode_dirty(inode);
1967 static inline void inode_dec_link_count(struct inode *inode)
1970 mark_inode_dirty(inode);
1974 * inode_inc_iversion - increments i_version
1975 * @inode: inode that need to be updated
1977 * Every time the inode is modified, the i_version field will be incremented.
1978 * The filesystem has to be mounted with i_version flag
1981 static inline void inode_inc_iversion(struct inode *inode)
1983 spin_lock(&inode->i_lock);
1985 spin_unlock(&inode->i_lock);
1988 enum file_time_flags {
1995 extern bool atime_needs_update(const struct path *, struct inode *);
1996 extern void touch_atime(const struct path *);
1997 static inline void file_accessed(struct file *file)
1999 if (!(file->f_flags & O_NOATIME))
2000 touch_atime(&file->f_path);
2003 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2004 int sync_inode_metadata(struct inode *inode, int wait);
2006 struct file_system_type {
2009 #define FS_REQUIRES_DEV 1
2010 #define FS_BINARY_MOUNTDATA 2
2011 #define FS_HAS_SUBTYPE 4
2012 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2013 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
2014 #define FS_USERNS_VISIBLE 32 /* FS must already be visible */
2015 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2016 struct dentry *(*mount) (struct file_system_type *, int,
2017 const char *, void *);
2018 struct dentry *(*mount2) (struct vfsmount *, struct file_system_type *, int,
2019 const char *, void *);
2020 void *(*alloc_mnt_data) (void);
2021 void (*kill_sb) (struct super_block *);
2022 struct module *owner;
2023 struct file_system_type * next;
2024 struct hlist_head fs_supers;
2026 struct lock_class_key s_lock_key;
2027 struct lock_class_key s_umount_key;
2028 struct lock_class_key s_vfs_rename_key;
2029 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2031 struct lock_class_key i_lock_key;
2032 struct lock_class_key i_mutex_key;
2033 struct lock_class_key i_mutex_dir_key;
2036 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2038 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
2039 void *data, int (*fill_super)(struct super_block *, void *, int));
2040 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2041 int flags, const char *dev_name, void *data,
2042 int (*fill_super)(struct super_block *, void *, int));
2043 extern struct dentry *mount_single(struct file_system_type *fs_type,
2044 int flags, void *data,
2045 int (*fill_super)(struct super_block *, void *, int));
2046 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2047 int flags, void *data,
2048 int (*fill_super)(struct super_block *, void *, int));
2049 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2050 void generic_shutdown_super(struct super_block *sb);
2051 void kill_block_super(struct super_block *sb);
2052 void kill_anon_super(struct super_block *sb);
2053 void kill_litter_super(struct super_block *sb);
2054 void deactivate_super(struct super_block *sb);
2055 void deactivate_locked_super(struct super_block *sb);
2056 int set_anon_super(struct super_block *s, void *data);
2057 int get_anon_bdev(dev_t *);
2058 void free_anon_bdev(dev_t);
2059 struct super_block *sget(struct file_system_type *type,
2060 int (*test)(struct super_block *,void *),
2061 int (*set)(struct super_block *,void *),
2062 int flags, void *data);
2063 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
2064 const struct super_operations *ops,
2065 const struct dentry_operations *dops,
2068 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2069 #define fops_get(fops) \
2070 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2071 #define fops_put(fops) \
2072 do { if (fops) module_put((fops)->owner); } while(0)
2074 * This one is to be used *ONLY* from ->open() instances.
2075 * fops must be non-NULL, pinned down *and* module dependencies
2076 * should be sufficient to pin the caller down as well.
2078 #define replace_fops(f, fops) \
2080 struct file *__file = (f); \
2081 fops_put(__file->f_op); \
2082 BUG_ON(!(__file->f_op = (fops))); \
2085 extern int register_filesystem(struct file_system_type *);
2086 extern int unregister_filesystem(struct file_system_type *);
2087 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2088 #define kern_mount(type) kern_mount_data(type, NULL)
2089 extern void kern_unmount(struct vfsmount *mnt);
2090 extern int may_umount_tree(struct vfsmount *);
2091 extern int may_umount(struct vfsmount *);
2092 extern long do_mount(const char *, const char __user *,
2093 const char *, unsigned long, void *);
2094 extern struct vfsmount *collect_mounts(struct path *);
2095 extern void drop_collected_mounts(struct vfsmount *);
2096 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2098 extern int vfs_statfs(struct path *, struct kstatfs *);
2099 extern int user_statfs(const char __user *, struct kstatfs *);
2100 extern int fd_statfs(int, struct kstatfs *);
2101 extern int vfs_ustat(dev_t, struct kstatfs *);
2102 extern int freeze_super(struct super_block *super);
2103 extern int thaw_super(struct super_block *super);
2104 extern bool our_mnt(struct vfsmount *mnt);
2106 extern int current_umask(void);
2108 extern void ihold(struct inode * inode);
2109 extern void iput(struct inode *);
2110 extern int generic_update_time(struct inode *, struct timespec *, int);
2113 extern struct kobject *fs_kobj;
2115 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
2117 #define FLOCK_VERIFY_READ 1
2118 #define FLOCK_VERIFY_WRITE 2
2120 #ifdef CONFIG_FILE_LOCKING
2121 extern int locks_mandatory_locked(struct file *);
2122 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
2125 * Candidates for mandatory locking have the setgid bit set
2126 * but no group execute bit - an otherwise meaningless combination.
2129 static inline int __mandatory_lock(struct inode *ino)
2131 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2135 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2136 * otherwise these will be advisory locks
2139 static inline int mandatory_lock(struct inode *ino)
2141 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2144 static inline int locks_verify_locked(struct file *file)
2146 if (mandatory_lock(file_inode(file)))
2147 return locks_mandatory_locked(file);
2151 static inline int locks_verify_truncate(struct inode *inode,
2155 if (inode->i_flctx && mandatory_lock(inode))
2156 return locks_mandatory_area(
2157 FLOCK_VERIFY_WRITE, inode, filp,
2158 size < inode->i_size ? size : inode->i_size,
2159 (size < inode->i_size ? inode->i_size - size
2160 : size - inode->i_size)
2165 static inline int break_lease(struct inode *inode, unsigned int mode)
2168 * Since this check is lockless, we must ensure that any refcounts
2169 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2170 * could end up racing with tasks trying to set a new lease on this
2174 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2175 return __break_lease(inode, mode, FL_LEASE);
2179 static inline int break_deleg(struct inode *inode, unsigned int mode)
2182 * Since this check is lockless, we must ensure that any refcounts
2183 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2184 * could end up racing with tasks trying to set a new lease on this
2188 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2189 return __break_lease(inode, mode, FL_DELEG);
2193 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2197 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2198 if (ret == -EWOULDBLOCK && delegated_inode) {
2199 *delegated_inode = inode;
2205 static inline int break_deleg_wait(struct inode **delegated_inode)
2209 ret = break_deleg(*delegated_inode, O_WRONLY);
2210 iput(*delegated_inode);
2211 *delegated_inode = NULL;
2215 static inline int break_layout(struct inode *inode, bool wait)
2218 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2219 return __break_lease(inode,
2220 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2225 #else /* !CONFIG_FILE_LOCKING */
2226 static inline int locks_mandatory_locked(struct file *file)
2231 static inline int locks_mandatory_area(int rw, struct inode *inode,
2232 struct file *filp, loff_t offset,
2238 static inline int __mandatory_lock(struct inode *inode)
2243 static inline int mandatory_lock(struct inode *inode)
2248 static inline int locks_verify_locked(struct file *file)
2253 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2259 static inline int break_lease(struct inode *inode, unsigned int mode)
2264 static inline int break_deleg(struct inode *inode, unsigned int mode)
2269 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2274 static inline int break_deleg_wait(struct inode **delegated_inode)
2280 static inline int break_layout(struct inode *inode, bool wait)
2285 #endif /* CONFIG_FILE_LOCKING */
2290 const char *name; /* pointer to actual string */
2291 const __user char *uptr; /* original userland pointer */
2292 struct audit_names *aname;
2297 extern long vfs_truncate(struct path *, loff_t);
2298 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2300 extern int do_truncate2(struct vfsmount *, struct dentry *, loff_t start,
2301 unsigned int time_attrs, struct file *filp);
2302 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2304 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2306 extern struct file *file_open_name(struct filename *, int, umode_t);
2307 extern struct file *filp_open(const char *, int, umode_t);
2308 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2309 const char *, int, umode_t);
2310 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2311 extern int filp_close(struct file *, fl_owner_t id);
2313 extern struct filename *getname_flags(const char __user *, int, int *);
2314 extern struct filename *getname(const char __user *);
2315 extern struct filename *getname_kernel(const char *);
2316 extern void putname(struct filename *name);
2322 extern int finish_open(struct file *file, struct dentry *dentry,
2323 int (*open)(struct inode *, struct file *),
2325 extern int finish_no_open(struct file *file, struct dentry *dentry);
2329 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2332 extern void __init vfs_caches_init_early(void);
2333 extern void __init vfs_caches_init(void);
2335 extern struct kmem_cache *names_cachep;
2337 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2338 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2341 extern int register_blkdev(unsigned int, const char *);
2342 extern void unregister_blkdev(unsigned int, const char *);
2343 extern struct block_device *bdget(dev_t);
2344 extern struct block_device *bdgrab(struct block_device *bdev);
2345 extern void bd_set_size(struct block_device *, loff_t size);
2346 extern void bd_forget(struct inode *inode);
2347 extern void bdput(struct block_device *);
2348 extern void invalidate_bdev(struct block_device *);
2349 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2350 extern int sync_blockdev(struct block_device *bdev);
2351 extern void kill_bdev(struct block_device *);
2352 extern struct super_block *freeze_bdev(struct block_device *);
2353 extern void emergency_thaw_all(void);
2354 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2355 extern int fsync_bdev(struct block_device *);
2357 extern struct super_block *blockdev_superblock;
2359 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2361 return sb == blockdev_superblock;
2364 static inline void bd_forget(struct inode *inode) {}
2365 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2366 static inline void kill_bdev(struct block_device *bdev) {}
2367 static inline void invalidate_bdev(struct block_device *bdev) {}
2369 static inline struct super_block *freeze_bdev(struct block_device *sb)
2374 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2379 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2383 static inline int sb_is_blkdev_sb(struct super_block *sb)
2388 extern int sync_filesystem(struct super_block *);
2389 extern const struct file_operations def_blk_fops;
2390 extern const struct file_operations def_chr_fops;
2392 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2393 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2394 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2395 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2396 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2398 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2400 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2401 extern int __blkdev_reread_part(struct block_device *bdev);
2402 extern int blkdev_reread_part(struct block_device *bdev);
2405 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2406 extern void bd_unlink_disk_holder(struct block_device *bdev,
2407 struct gendisk *disk);
2409 static inline int bd_link_disk_holder(struct block_device *bdev,
2410 struct gendisk *disk)
2414 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2415 struct gendisk *disk)
2422 #define CHRDEV_MAJOR_HASH_SIZE 255
2423 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2424 extern int register_chrdev_region(dev_t, unsigned, const char *);
2425 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2426 unsigned int count, const char *name,
2427 const struct file_operations *fops);
2428 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2429 unsigned int count, const char *name);
2430 extern void unregister_chrdev_region(dev_t, unsigned);
2431 extern void chrdev_show(struct seq_file *,off_t);
2433 static inline int register_chrdev(unsigned int major, const char *name,
2434 const struct file_operations *fops)
2436 return __register_chrdev(major, 0, 256, name, fops);
2439 static inline void unregister_chrdev(unsigned int major, const char *name)
2441 __unregister_chrdev(major, 0, 256, name);
2444 /* fs/block_dev.c */
2445 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2446 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2449 #define BLKDEV_MAJOR_HASH_SIZE 255
2450 extern const char *__bdevname(dev_t, char *buffer);
2451 extern const char *bdevname(struct block_device *bdev, char *buffer);
2452 extern struct block_device *lookup_bdev(const char *);
2453 extern void blkdev_show(struct seq_file *,off_t);
2456 #define BLKDEV_MAJOR_HASH_SIZE 0
2459 extern void init_special_inode(struct inode *, umode_t, dev_t);
2461 /* Invalid inode operations -- fs/bad_inode.c */
2462 extern void make_bad_inode(struct inode *);
2463 extern int is_bad_inode(struct inode *);
2467 * return READ, READA, or WRITE
2469 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2472 * return data direction, READ or WRITE
2474 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2476 extern void check_disk_size_change(struct gendisk *disk,
2477 struct block_device *bdev);
2478 extern int revalidate_disk(struct gendisk *);
2479 extern int check_disk_change(struct block_device *);
2480 extern int __invalidate_device(struct block_device *, bool);
2481 extern int invalidate_partition(struct gendisk *, int);
2483 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2484 pgoff_t start, pgoff_t end);
2486 static inline void invalidate_remote_inode(struct inode *inode)
2488 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2489 S_ISLNK(inode->i_mode))
2490 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2492 extern int invalidate_inode_pages2(struct address_space *mapping);
2493 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2494 pgoff_t start, pgoff_t end);
2495 extern int write_inode_now(struct inode *, int);
2496 extern int filemap_fdatawrite(struct address_space *);
2497 extern int filemap_flush(struct address_space *);
2498 extern int filemap_fdatawait(struct address_space *);
2499 extern void filemap_fdatawait_keep_errors(struct address_space *);
2500 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2502 extern int filemap_write_and_wait(struct address_space *mapping);
2503 extern int filemap_write_and_wait_range(struct address_space *mapping,
2504 loff_t lstart, loff_t lend);
2505 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2506 loff_t start, loff_t end, int sync_mode);
2507 extern int filemap_fdatawrite_range(struct address_space *mapping,
2508 loff_t start, loff_t end);
2510 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2512 extern int vfs_fsync(struct file *file, int datasync);
2513 static inline int generic_write_sync(struct file *file, loff_t pos, loff_t count)
2515 if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host))
2517 return vfs_fsync_range(file, pos, pos + count - 1,
2518 (file->f_flags & __O_SYNC) ? 0 : 1);
2520 extern void emergency_sync(void);
2521 extern void emergency_remount(void);
2523 extern sector_t bmap(struct inode *, sector_t);
2525 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2526 extern int notify_change2(struct vfsmount *, struct dentry *, struct iattr *, struct inode **);
2527 extern int inode_permission(struct inode *, int);
2528 extern int inode_permission2(struct vfsmount *, struct inode *, int);
2529 extern int __inode_permission(struct inode *, int);
2530 extern int __inode_permission2(struct vfsmount *, struct inode *, int);
2531 extern int generic_permission(struct inode *, int);
2532 extern int __check_sticky(struct inode *dir, struct inode *inode);
2534 static inline bool execute_ok(struct inode *inode)
2536 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2539 static inline void file_start_write(struct file *file)
2541 if (!S_ISREG(file_inode(file)->i_mode))
2543 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2546 static inline bool file_start_write_trylock(struct file *file)
2548 if (!S_ISREG(file_inode(file)->i_mode))
2550 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2553 static inline void file_end_write(struct file *file)
2555 if (!S_ISREG(file_inode(file)->i_mode))
2557 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2561 * get_write_access() gets write permission for a file.
2562 * put_write_access() releases this write permission.
2563 * This is used for regular files.
2564 * We cannot support write (and maybe mmap read-write shared) accesses and
2565 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2566 * can have the following values:
2567 * 0: no writers, no VM_DENYWRITE mappings
2568 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2569 * > 0: (i_writecount) users are writing to the file.
2571 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2572 * except for the cases where we don't hold i_writecount yet. Then we need to
2573 * use {get,deny}_write_access() - these functions check the sign and refuse
2574 * to do the change if sign is wrong.
2576 static inline int get_write_access(struct inode *inode)
2578 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2580 static inline int deny_write_access(struct file *file)
2582 struct inode *inode = file_inode(file);
2583 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2585 static inline void put_write_access(struct inode * inode)
2587 atomic_dec(&inode->i_writecount);
2589 static inline void allow_write_access(struct file *file)
2592 atomic_inc(&file_inode(file)->i_writecount);
2594 static inline bool inode_is_open_for_write(const struct inode *inode)
2596 return atomic_read(&inode->i_writecount) > 0;
2600 static inline void i_readcount_dec(struct inode *inode)
2602 BUG_ON(!atomic_read(&inode->i_readcount));
2603 atomic_dec(&inode->i_readcount);
2605 static inline void i_readcount_inc(struct inode *inode)
2607 atomic_inc(&inode->i_readcount);
2610 static inline void i_readcount_dec(struct inode *inode)
2614 static inline void i_readcount_inc(struct inode *inode)
2619 extern int do_pipe_flags(int *, int);
2621 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2622 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2623 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
2624 extern struct file * open_exec(const char *);
2626 /* fs/dcache.c -- generic fs support functions */
2627 extern int is_subdir(struct dentry *, struct dentry *);
2628 extern int path_is_under(struct path *, struct path *);
2630 extern char *file_path(struct file *, char *, int);
2632 #include <linux/err.h>
2634 /* needed for stackable file system support */
2635 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2637 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2639 extern int inode_init_always(struct super_block *, struct inode *);
2640 extern void inode_init_once(struct inode *);
2641 extern void address_space_init_once(struct address_space *mapping);
2642 extern struct inode * igrab(struct inode *);
2643 extern ino_t iunique(struct super_block *, ino_t);
2644 extern int inode_needs_sync(struct inode *inode);
2645 extern int generic_delete_inode(struct inode *inode);
2646 static inline int generic_drop_inode(struct inode *inode)
2648 return !inode->i_nlink || inode_unhashed(inode);
2651 extern struct inode *ilookup5_nowait(struct super_block *sb,
2652 unsigned long hashval, int (*test)(struct inode *, void *),
2654 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2655 int (*test)(struct inode *, void *), void *data);
2656 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2658 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2659 extern struct inode * iget_locked(struct super_block *, unsigned long);
2660 extern struct inode *find_inode_nowait(struct super_block *,
2662 int (*match)(struct inode *,
2663 unsigned long, void *),
2665 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2666 extern int insert_inode_locked(struct inode *);
2667 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2668 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2670 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2672 extern void unlock_new_inode(struct inode *);
2673 extern unsigned int get_next_ino(void);
2674 extern void evict_inodes(struct super_block *sb);
2676 extern void __iget(struct inode * inode);
2677 extern void iget_failed(struct inode *);
2678 extern void clear_inode(struct inode *);
2679 extern void __destroy_inode(struct inode *);
2680 extern struct inode *new_inode_pseudo(struct super_block *sb);
2681 extern struct inode *new_inode(struct super_block *sb);
2682 extern void free_inode_nonrcu(struct inode *inode);
2683 extern int should_remove_suid(struct dentry *);
2684 extern int file_remove_privs(struct file *);
2685 extern int dentry_needs_remove_privs(struct dentry *dentry);
2686 static inline int file_needs_remove_privs(struct file *file)
2688 return dentry_needs_remove_privs(file->f_path.dentry);
2691 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2692 static inline void insert_inode_hash(struct inode *inode)
2694 __insert_inode_hash(inode, inode->i_ino);
2697 extern void __remove_inode_hash(struct inode *);
2698 static inline void remove_inode_hash(struct inode *inode)
2700 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2701 __remove_inode_hash(inode);
2704 extern void inode_sb_list_add(struct inode *inode);
2707 extern blk_qc_t submit_bio(int, struct bio *);
2708 extern int bdev_read_only(struct block_device *);
2710 extern int set_blocksize(struct block_device *, int);
2711 extern int sb_set_blocksize(struct super_block *, int);
2712 extern int sb_min_blocksize(struct super_block *, int);
2714 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2715 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2716 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2717 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2718 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2719 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2720 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *, loff_t);
2721 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2723 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos);
2724 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos);
2726 /* fs/block_dev.c */
2727 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2728 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2729 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2731 extern void block_sync_page(struct page *page);
2734 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2735 struct pipe_inode_info *, size_t, unsigned int);
2736 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2737 struct pipe_inode_info *, size_t, unsigned int);
2738 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2739 struct file *, loff_t *, size_t, unsigned int);
2740 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2741 struct file *out, loff_t *, size_t len, unsigned int flags);
2742 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2743 loff_t *opos, size_t len, unsigned int flags);
2747 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2748 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2749 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2750 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2751 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2752 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2753 int whence, loff_t maxsize, loff_t eof);
2754 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2755 int whence, loff_t size);
2756 extern int generic_file_open(struct inode * inode, struct file * filp);
2757 extern int nonseekable_open(struct inode * inode, struct file * filp);
2760 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2761 loff_t file_offset);
2764 /* need locking between buffered and direct access */
2767 /* filesystem does not support filling holes */
2768 DIO_SKIP_HOLES = 0x02,
2770 /* filesystem can handle aio writes beyond i_size */
2771 DIO_ASYNC_EXTEND = 0x04,
2773 /* inode/fs/bdev does not need truncate protection */
2774 DIO_SKIP_DIO_COUNT = 0x08,
2777 void dio_end_io(struct bio *bio, int error);
2779 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2780 struct block_device *bdev, struct iov_iter *iter,
2781 loff_t offset, get_block_t get_block,
2782 dio_iodone_t end_io, dio_submit_t submit_io,
2785 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2786 struct inode *inode,
2787 struct iov_iter *iter, loff_t offset,
2788 get_block_t get_block)
2790 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
2791 offset, get_block, NULL, NULL,
2792 DIO_LOCKING | DIO_SKIP_HOLES);
2796 void inode_dio_wait(struct inode *inode);
2799 * inode_dio_begin - signal start of a direct I/O requests
2800 * @inode: inode the direct I/O happens on
2802 * This is called once we've finished processing a direct I/O request,
2803 * and is used to wake up callers waiting for direct I/O to be quiesced.
2805 static inline void inode_dio_begin(struct inode *inode)
2807 atomic_inc(&inode->i_dio_count);
2811 * inode_dio_end - signal finish of a direct I/O requests
2812 * @inode: inode the direct I/O happens on
2814 * This is called once we've finished processing a direct I/O request,
2815 * and is used to wake up callers waiting for direct I/O to be quiesced.
2817 static inline void inode_dio_end(struct inode *inode)
2819 if (atomic_dec_and_test(&inode->i_dio_count))
2820 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
2823 extern void inode_set_flags(struct inode *inode, unsigned int flags,
2826 extern const struct file_operations generic_ro_fops;
2828 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2830 extern int readlink_copy(char __user *, int, const char *);
2831 extern int page_readlink(struct dentry *, char __user *, int);
2832 extern const char *page_follow_link_light(struct dentry *, void **);
2833 extern void page_put_link(struct inode *, void *);
2834 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2836 extern int page_symlink(struct inode *inode, const char *symname, int len);
2837 extern const struct inode_operations page_symlink_inode_operations;
2838 extern void kfree_put_link(struct inode *, void *);
2839 extern void free_page_put_link(struct inode *, void *);
2840 extern int generic_readlink(struct dentry *, char __user *, int);
2841 extern void generic_fillattr(struct inode *, struct kstat *);
2842 int vfs_getattr_nosec(struct path *path, struct kstat *stat);
2843 extern int vfs_getattr(struct path *, struct kstat *);
2844 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2845 void inode_add_bytes(struct inode *inode, loff_t bytes);
2846 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2847 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2848 loff_t inode_get_bytes(struct inode *inode);
2849 void inode_set_bytes(struct inode *inode, loff_t bytes);
2850 const char *simple_follow_link(struct dentry *, void **);
2851 extern const struct inode_operations simple_symlink_inode_operations;
2853 extern int iterate_dir(struct file *, struct dir_context *);
2855 extern int vfs_stat(const char __user *, struct kstat *);
2856 extern int vfs_lstat(const char __user *, struct kstat *);
2857 extern int vfs_fstat(unsigned int, struct kstat *);
2858 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2860 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2862 extern int __generic_block_fiemap(struct inode *inode,
2863 struct fiemap_extent_info *fieinfo,
2864 loff_t start, loff_t len,
2865 get_block_t *get_block);
2866 extern int generic_block_fiemap(struct inode *inode,
2867 struct fiemap_extent_info *fieinfo, u64 start,
2868 u64 len, get_block_t *get_block);
2870 extern void get_filesystem(struct file_system_type *fs);
2871 extern void put_filesystem(struct file_system_type *fs);
2872 extern struct file_system_type *get_fs_type(const char *name);
2873 extern struct super_block *get_super(struct block_device *);
2874 extern struct super_block *get_super_thawed(struct block_device *);
2875 extern struct super_block *get_active_super(struct block_device *bdev);
2876 extern void drop_super(struct super_block *sb);
2877 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2878 extern void iterate_supers_type(struct file_system_type *,
2879 void (*)(struct super_block *, void *), void *);
2881 extern int dcache_dir_open(struct inode *, struct file *);
2882 extern int dcache_dir_close(struct inode *, struct file *);
2883 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2884 extern int dcache_readdir(struct file *, struct dir_context *);
2885 extern int simple_setattr(struct dentry *, struct iattr *);
2886 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2887 extern int simple_statfs(struct dentry *, struct kstatfs *);
2888 extern int simple_open(struct inode *inode, struct file *file);
2889 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2890 extern int simple_unlink(struct inode *, struct dentry *);
2891 extern int simple_rmdir(struct inode *, struct dentry *);
2892 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2893 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2894 extern int simple_empty(struct dentry *);
2895 extern int simple_readpage(struct file *file, struct page *page);
2896 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2897 loff_t pos, unsigned len, unsigned flags,
2898 struct page **pagep, void **fsdata);
2899 extern int simple_write_end(struct file *file, struct address_space *mapping,
2900 loff_t pos, unsigned len, unsigned copied,
2901 struct page *page, void *fsdata);
2902 extern int always_delete_dentry(const struct dentry *);
2903 extern struct inode *alloc_anon_inode(struct super_block *);
2904 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
2905 extern const struct dentry_operations simple_dentry_operations;
2907 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2908 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2909 extern const struct file_operations simple_dir_operations;
2910 extern const struct inode_operations simple_dir_inode_operations;
2911 extern void make_empty_dir_inode(struct inode *inode);
2912 extern bool is_empty_dir_inode(struct inode *inode);
2913 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2914 struct dentry *d_alloc_name(struct dentry *, const char *);
2915 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2916 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2917 extern void simple_release_fs(struct vfsmount **mount, int *count);
2919 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2920 loff_t *ppos, const void *from, size_t available);
2921 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2922 const void __user *from, size_t count);
2924 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
2925 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2927 extern int generic_check_addressable(unsigned, u64);
2929 #ifdef CONFIG_MIGRATION
2930 extern int buffer_migrate_page(struct address_space *,
2931 struct page *, struct page *,
2934 #define buffer_migrate_page NULL
2937 extern int inode_change_ok(const struct inode *, struct iattr *);
2938 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2939 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2941 extern int file_update_time(struct file *file);
2943 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2944 extern void save_mount_options(struct super_block *sb, char *options);
2945 extern void replace_mount_options(struct super_block *sb, char *options);
2947 static inline bool io_is_direct(struct file *filp)
2949 return (filp->f_flags & O_DIRECT) || IS_DAX(file_inode(filp));
2952 static inline int iocb_flags(struct file *file)
2955 if (file->f_flags & O_APPEND)
2957 if (io_is_direct(file))
2962 static inline ino_t parent_ino(struct dentry *dentry)
2967 * Don't strictly need d_lock here? If the parent ino could change
2968 * then surely we'd have a deeper race in the caller?
2970 spin_lock(&dentry->d_lock);
2971 res = dentry->d_parent->d_inode->i_ino;
2972 spin_unlock(&dentry->d_lock);
2976 /* Transaction based IO helpers */
2979 * An argresp is stored in an allocated page and holds the
2980 * size of the argument or response, along with its content
2982 struct simple_transaction_argresp {
2987 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2989 char *simple_transaction_get(struct file *file, const char __user *buf,
2991 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2992 size_t size, loff_t *pos);
2993 int simple_transaction_release(struct inode *inode, struct file *file);
2995 void simple_transaction_set(struct file *file, size_t n);
2998 * simple attribute files
3000 * These attributes behave similar to those in sysfs:
3002 * Writing to an attribute immediately sets a value, an open file can be
3003 * written to multiple times.
3005 * Reading from an attribute creates a buffer from the value that might get
3006 * read with multiple read calls. When the attribute has been read
3007 * completely, no further read calls are possible until the file is opened
3010 * All attributes contain a text representation of a numeric value
3011 * that are accessed with the get() and set() functions.
3013 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3014 static int __fops ## _open(struct inode *inode, struct file *file) \
3016 __simple_attr_check_format(__fmt, 0ull); \
3017 return simple_attr_open(inode, file, __get, __set, __fmt); \
3019 static const struct file_operations __fops = { \
3020 .owner = THIS_MODULE, \
3021 .open = __fops ## _open, \
3022 .release = simple_attr_release, \
3023 .read = simple_attr_read, \
3024 .write = simple_attr_write, \
3025 .llseek = generic_file_llseek, \
3028 static inline __printf(1, 2)
3029 void __simple_attr_check_format(const char *fmt, ...)
3031 /* don't do anything, just let the compiler check the arguments; */
3034 int simple_attr_open(struct inode *inode, struct file *file,
3035 int (*get)(void *, u64 *), int (*set)(void *, u64),
3037 int simple_attr_release(struct inode *inode, struct file *file);
3038 ssize_t simple_attr_read(struct file *file, char __user *buf,
3039 size_t len, loff_t *ppos);
3040 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3041 size_t len, loff_t *ppos);
3044 int proc_nr_files(struct ctl_table *table, int write,
3045 void __user *buffer, size_t *lenp, loff_t *ppos);
3046 int proc_nr_dentry(struct ctl_table *table, int write,
3047 void __user *buffer, size_t *lenp, loff_t *ppos);
3048 int proc_nr_inodes(struct ctl_table *table, int write,
3049 void __user *buffer, size_t *lenp, loff_t *ppos);
3050 int __init get_filesystem_list(char *buf);
3052 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3053 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3055 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3056 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3057 (flag & __FMODE_NONOTIFY)))
3059 static inline int is_sxid(umode_t mode)
3061 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3064 static inline int check_sticky(struct inode *dir, struct inode *inode)
3066 if (!(dir->i_mode & S_ISVTX))
3069 return __check_sticky(dir, inode);
3072 static inline void inode_has_no_xattr(struct inode *inode)
3074 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
3075 inode->i_flags |= S_NOSEC;
3078 static inline bool is_root_inode(struct inode *inode)
3080 return inode == inode->i_sb->s_root->d_inode;
3083 static inline bool dir_emit(struct dir_context *ctx,
3084 const char *name, int namelen,
3085 u64 ino, unsigned type)
3087 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3089 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3091 return ctx->actor(ctx, ".", 1, ctx->pos,
3092 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3094 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3096 return ctx->actor(ctx, "..", 2, ctx->pos,
3097 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3099 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3101 if (ctx->pos == 0) {
3102 if (!dir_emit_dot(file, ctx))
3106 if (ctx->pos == 1) {
3107 if (!dir_emit_dotdot(file, ctx))
3113 static inline bool dir_relax(struct inode *inode)
3115 inode_unlock(inode);
3117 return !IS_DEADDIR(inode);
3120 extern bool path_noexec(const struct path *path);
3121 extern void inode_nohighmem(struct inode *inode);
3123 #endif /* _LINUX_FS_H */