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/fiemap.h>
24 #include <linux/rculist_bl.h>
25 #include <linux/atomic.h>
26 #include <linux/shrinker.h>
27 #include <linux/migrate_mode.h>
28 #include <linux/uidgid.h>
29 #include <linux/lockdep.h>
30 #include <linux/percpu-rwsem.h>
31 #include <linux/blk_types.h>
32 #include <linux/workqueue.h>
33 #include <linux/percpu-rwsem.h>
34 #include <linux/delayed_call.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;
73 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
74 struct buffer_head *bh_result, int create);
75 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
76 ssize_t bytes, void *private);
77 typedef void (dax_iodone_t)(struct buffer_head *bh_map, int uptodate);
79 #define MAY_EXEC 0x00000001
80 #define MAY_WRITE 0x00000002
81 #define MAY_READ 0x00000004
82 #define MAY_APPEND 0x00000008
83 #define MAY_ACCESS 0x00000010
84 #define MAY_OPEN 0x00000020
85 #define MAY_CHDIR 0x00000040
86 /* called from RCU mode, don't block */
87 #define MAY_NOT_BLOCK 0x00000080
90 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
91 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
94 /* file is open for reading */
95 #define FMODE_READ ((__force fmode_t)0x1)
96 /* file is open for writing */
97 #define FMODE_WRITE ((__force fmode_t)0x2)
98 /* file is seekable */
99 #define FMODE_LSEEK ((__force fmode_t)0x4)
100 /* file can be accessed using pread */
101 #define FMODE_PREAD ((__force fmode_t)0x8)
102 /* file can be accessed using pwrite */
103 #define FMODE_PWRITE ((__force fmode_t)0x10)
104 /* File is opened for execution with sys_execve / sys_uselib */
105 #define FMODE_EXEC ((__force fmode_t)0x20)
106 /* File is opened with O_NDELAY (only set for block devices) */
107 #define FMODE_NDELAY ((__force fmode_t)0x40)
108 /* File is opened with O_EXCL (only set for block devices) */
109 #define FMODE_EXCL ((__force fmode_t)0x80)
110 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
111 (specialy hack for floppy.c) */
112 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
113 /* 32bit hashes as llseek() offset (for directories) */
114 #define FMODE_32BITHASH ((__force fmode_t)0x200)
115 /* 64bit hashes as llseek() offset (for directories) */
116 #define FMODE_64BITHASH ((__force fmode_t)0x400)
119 * Don't update ctime and mtime.
121 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
122 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
124 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
126 /* Expect random access pattern */
127 #define FMODE_RANDOM ((__force fmode_t)0x1000)
129 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
130 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
132 /* File is opened with O_PATH; almost nothing can be done with it */
133 #define FMODE_PATH ((__force fmode_t)0x4000)
135 /* File needs atomic accesses to f_pos */
136 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
137 /* Write access to underlying fs */
138 #define FMODE_WRITER ((__force fmode_t)0x10000)
139 /* Has read method(s) */
140 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
141 /* Has write method(s) */
142 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
144 /* File was opened by fanotify and shouldn't generate fanotify events */
145 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
148 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
149 * that indicates that they should check the contents of the iovec are
150 * valid, but not check the memory that the iovec elements
153 #define CHECK_IOVEC_ONLY -1
156 * The below are the various read and write types that we support. Some of
157 * them include behavioral modifiers that send information down to the
158 * block layer and IO scheduler. Terminology:
160 * The block layer uses device plugging to defer IO a little bit, in
161 * the hope that we will see more IO very shortly. This increases
162 * coalescing of adjacent IO and thus reduces the number of IOs we
163 * have to send to the device. It also allows for better queuing,
164 * if the IO isn't mergeable. If the caller is going to be waiting
165 * for the IO, then he must ensure that the device is unplugged so
166 * that the IO is dispatched to the driver.
168 * All IO is handled async in Linux. This is fine for background
169 * writes, but for reads or writes that someone waits for completion
170 * on, we want to notify the block layer and IO scheduler so that they
171 * know about it. That allows them to make better scheduling
172 * decisions. So when the below references 'sync' and 'async', it
173 * is referencing this priority hint.
175 * With that in mind, the available types are:
177 * READ A normal read operation. Device will be plugged.
178 * READ_SYNC A synchronous read. Device is not plugged, caller can
179 * immediately wait on this read without caring about
181 * READA Used for read-ahead operations. Lower priority, and the
182 * block layer could (in theory) choose to ignore this
183 * request if it runs into resource problems.
184 * WRITE A normal async write. Device will be plugged.
185 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
186 * the hint that someone will be waiting on this IO
187 * shortly. The write equivalent of READ_SYNC.
188 * WRITE_ODIRECT Special case write for O_DIRECT only.
189 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
190 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
191 * non-volatile media on completion.
192 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
193 * by a cache flush and data is guaranteed to be on
194 * non-volatile media on completion.
197 #define RW_MASK REQ_WRITE
198 #define RWA_MASK REQ_RAHEAD
201 #define WRITE RW_MASK
202 #define READA RWA_MASK
204 #define READ_SYNC (READ | REQ_SYNC)
205 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
206 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
207 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
208 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
209 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
212 * Attribute flags. These should be or-ed together to figure out what
215 #define ATTR_MODE (1 << 0)
216 #define ATTR_UID (1 << 1)
217 #define ATTR_GID (1 << 2)
218 #define ATTR_SIZE (1 << 3)
219 #define ATTR_ATIME (1 << 4)
220 #define ATTR_MTIME (1 << 5)
221 #define ATTR_CTIME (1 << 6)
222 #define ATTR_ATIME_SET (1 << 7)
223 #define ATTR_MTIME_SET (1 << 8)
224 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
225 #define ATTR_ATTR_FLAG (1 << 10)
226 #define ATTR_KILL_SUID (1 << 11)
227 #define ATTR_KILL_SGID (1 << 12)
228 #define ATTR_FILE (1 << 13)
229 #define ATTR_KILL_PRIV (1 << 14)
230 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
231 #define ATTR_TIMES_SET (1 << 16)
234 * Whiteout is represented by a char device. The following constants define the
235 * mode and device number to use.
237 #define WHITEOUT_MODE 0
238 #define WHITEOUT_DEV 0
241 * This is the Inode Attributes structure, used for notify_change(). It
242 * uses the above definitions as flags, to know which values have changed.
243 * Also, in this manner, a Filesystem can look at only the values it cares
244 * about. Basically, these are the attributes that the VFS layer can
245 * request to change from the FS layer.
247 * Derek Atkins <warlord@MIT.EDU> 94-10-20
250 unsigned int ia_valid;
255 struct timespec ia_atime;
256 struct timespec ia_mtime;
257 struct timespec ia_ctime;
260 * Not an attribute, but an auxiliary info for filesystems wanting to
261 * implement an ftruncate() like method. NOTE: filesystem should
262 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
264 struct file *ia_file;
268 * Includes for diskquotas.
270 #include <linux/quota.h>
273 * Maximum number of layers of fs stack. Needs to be limited to
274 * prevent kernel stack overflow
276 #define FILESYSTEM_MAX_STACK_DEPTH 2
279 * enum positive_aop_returns - aop return codes with specific semantics
281 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
282 * completed, that the page is still locked, and
283 * should be considered active. The VM uses this hint
284 * to return the page to the active list -- it won't
285 * be a candidate for writeback again in the near
286 * future. Other callers must be careful to unlock
287 * the page if they get this return. Returned by
290 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
291 * unlocked it and the page might have been truncated.
292 * The caller should back up to acquiring a new page and
293 * trying again. The aop will be taking reasonable
294 * precautions not to livelock. If the caller held a page
295 * reference, it should drop it before retrying. Returned
298 * address_space_operation functions return these large constants to indicate
299 * special semantics to the caller. These are much larger than the bytes in a
300 * page to allow for functions that return the number of bytes operated on in a
304 enum positive_aop_returns {
305 AOP_WRITEPAGE_ACTIVATE = 0x80000,
306 AOP_TRUNCATED_PAGE = 0x80001,
309 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
310 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
311 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
312 * helper code (eg buffer layer)
313 * to clear GFP_FS from alloc */
316 * oh the beauties of C type declarations.
319 struct address_space;
320 struct writeback_control;
322 #define IOCB_EVENTFD (1 << 0)
323 #define IOCB_APPEND (1 << 1)
324 #define IOCB_DIRECT (1 << 2)
325 #define IOCB_HIPRI (1 << 3)
326 #define IOCB_DSYNC (1 << 4)
327 #define IOCB_SYNC (1 << 5)
330 struct file *ki_filp;
332 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
337 static inline bool is_sync_kiocb(struct kiocb *kiocb)
339 return kiocb->ki_complete == NULL;
342 static inline int iocb_flags(struct file *file);
344 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
346 *kiocb = (struct kiocb) {
348 .ki_flags = iocb_flags(filp),
353 * "descriptor" for what we're up to with a read.
354 * This allows us to use the same read code yet
355 * have multiple different users of the data that
356 * we read from a file.
358 * The simplest case just copies the data to user
371 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
372 unsigned long, unsigned long);
374 struct address_space_operations {
375 int (*writepage)(struct page *page, struct writeback_control *wbc);
376 int (*readpage)(struct file *, struct page *);
378 /* Write back some dirty pages from this mapping. */
379 int (*writepages)(struct address_space *, struct writeback_control *);
381 /* Set a page dirty. Return true if this dirtied it */
382 int (*set_page_dirty)(struct page *page);
384 int (*readpages)(struct file *filp, struct address_space *mapping,
385 struct list_head *pages, unsigned nr_pages);
387 int (*write_begin)(struct file *, struct address_space *mapping,
388 loff_t pos, unsigned len, unsigned flags,
389 struct page **pagep, void **fsdata);
390 int (*write_end)(struct file *, struct address_space *mapping,
391 loff_t pos, unsigned len, unsigned copied,
392 struct page *page, void *fsdata);
394 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
395 sector_t (*bmap)(struct address_space *, sector_t);
396 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
397 int (*releasepage) (struct page *, gfp_t);
398 void (*freepage)(struct page *);
399 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
401 * migrate the contents of a page to the specified target. If
402 * migrate_mode is MIGRATE_ASYNC, it must not block.
404 int (*migratepage) (struct address_space *,
405 struct page *, struct page *, enum migrate_mode);
406 int (*launder_page) (struct page *);
407 int (*is_partially_uptodate) (struct page *, unsigned long,
409 void (*is_dirty_writeback) (struct page *, bool *, bool *);
410 int (*error_remove_page)(struct address_space *, struct page *);
412 /* swapfile support */
413 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
415 void (*swap_deactivate)(struct file *file);
418 extern const struct address_space_operations empty_aops;
421 * pagecache_write_begin/pagecache_write_end must be used by general code
422 * to write into the pagecache.
424 int pagecache_write_begin(struct file *, struct address_space *mapping,
425 loff_t pos, unsigned len, unsigned flags,
426 struct page **pagep, void **fsdata);
428 int pagecache_write_end(struct file *, struct address_space *mapping,
429 loff_t pos, unsigned len, unsigned copied,
430 struct page *page, void *fsdata);
432 struct address_space {
433 struct inode *host; /* owner: inode, block_device */
434 struct radix_tree_root page_tree; /* radix tree of all pages */
435 spinlock_t tree_lock; /* and lock protecting it */
436 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
437 struct rb_root i_mmap; /* tree of private and shared mappings */
438 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
439 /* Protected by tree_lock together with the radix tree */
440 unsigned long nrpages; /* number of total pages */
441 /* number of shadow or DAX exceptional entries */
442 unsigned long nrexceptional;
443 pgoff_t writeback_index;/* writeback starts here */
444 const struct address_space_operations *a_ops; /* methods */
445 unsigned long flags; /* error bits/gfp mask */
446 spinlock_t private_lock; /* for use by the address_space */
447 struct list_head private_list; /* ditto */
448 void *private_data; /* ditto */
449 } __attribute__((aligned(sizeof(long))));
451 * On most architectures that alignment is already the case; but
452 * must be enforced here for CRIS, to let the least significant bit
453 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
455 struct request_queue;
457 struct block_device {
458 dev_t bd_dev; /* not a kdev_t - it's a search key */
460 struct inode * bd_inode; /* will die */
461 struct super_block * bd_super;
462 struct mutex bd_mutex; /* open/close mutex */
463 struct list_head bd_inodes;
467 bool bd_write_holder;
469 struct list_head bd_holder_disks;
471 struct block_device * bd_contains;
472 unsigned bd_block_size;
473 struct hd_struct * bd_part;
474 /* number of times partitions within this device have been opened. */
475 unsigned bd_part_count;
477 struct gendisk * bd_disk;
478 struct request_queue * bd_queue;
479 struct list_head bd_list;
481 * Private data. You must have bd_claim'ed the block_device
482 * to use this. NOTE: bd_claim allows an owner to claim
483 * the same device multiple times, the owner must take special
484 * care to not mess up bd_private for that case.
486 unsigned long bd_private;
488 /* The counter of freeze processes */
489 int bd_fsfreeze_count;
490 /* Mutex for freeze */
491 struct mutex bd_fsfreeze_mutex;
495 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
498 #define PAGECACHE_TAG_DIRTY 0
499 #define PAGECACHE_TAG_WRITEBACK 1
500 #define PAGECACHE_TAG_TOWRITE 2
502 int mapping_tagged(struct address_space *mapping, int tag);
504 static inline void i_mmap_lock_write(struct address_space *mapping)
506 down_write(&mapping->i_mmap_rwsem);
509 static inline void i_mmap_unlock_write(struct address_space *mapping)
511 up_write(&mapping->i_mmap_rwsem);
514 static inline void i_mmap_lock_read(struct address_space *mapping)
516 down_read(&mapping->i_mmap_rwsem);
519 static inline void i_mmap_unlock_read(struct address_space *mapping)
521 up_read(&mapping->i_mmap_rwsem);
525 * Might pages of this file be mapped into userspace?
527 static inline int mapping_mapped(struct address_space *mapping)
529 return !RB_EMPTY_ROOT(&mapping->i_mmap);
533 * Might pages of this file have been modified in userspace?
534 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
535 * marks vma as VM_SHARED if it is shared, and the file was opened for
536 * writing i.e. vma may be mprotected writable even if now readonly.
538 * If i_mmap_writable is negative, no new writable mappings are allowed. You
539 * can only deny writable mappings, if none exists right now.
541 static inline int mapping_writably_mapped(struct address_space *mapping)
543 return atomic_read(&mapping->i_mmap_writable) > 0;
546 static inline int mapping_map_writable(struct address_space *mapping)
548 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
552 static inline void mapping_unmap_writable(struct address_space *mapping)
554 atomic_dec(&mapping->i_mmap_writable);
557 static inline int mapping_deny_writable(struct address_space *mapping)
559 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
563 static inline void mapping_allow_writable(struct address_space *mapping)
565 atomic_inc(&mapping->i_mmap_writable);
569 * Use sequence counter to get consistent i_size on 32-bit processors.
571 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
572 #include <linux/seqlock.h>
573 #define __NEED_I_SIZE_ORDERED
574 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
576 #define i_size_ordered_init(inode) do { } while (0)
580 #define ACL_NOT_CACHED ((void *)(-1))
582 static inline struct posix_acl *
583 uncached_acl_sentinel(struct task_struct *task)
585 return (void *)task + 1;
589 is_uncached_acl(struct posix_acl *acl)
591 return (long)acl & 1;
594 #define IOP_FASTPERM 0x0001
595 #define IOP_LOOKUP 0x0002
596 #define IOP_NOFOLLOW 0x0004
599 * Keep mostly read-only and often accessed (especially for
600 * the RCU path lookup and 'stat' data) fields at the beginning
601 * of the 'struct inode'
605 unsigned short i_opflags;
608 unsigned int i_flags;
610 #ifdef CONFIG_FS_POSIX_ACL
611 struct posix_acl *i_acl;
612 struct posix_acl *i_default_acl;
615 const struct inode_operations *i_op;
616 struct super_block *i_sb;
617 struct address_space *i_mapping;
619 #ifdef CONFIG_SECURITY
623 /* Stat data, not accessed from path walking */
626 * Filesystems may only read i_nlink directly. They shall use the
627 * following functions for modification:
629 * (set|clear|inc|drop)_nlink
630 * inode_(inc|dec)_link_count
633 const unsigned int i_nlink;
634 unsigned int __i_nlink;
638 struct timespec i_atime;
639 struct timespec i_mtime;
640 struct timespec i_ctime;
641 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
642 unsigned short i_bytes;
643 unsigned int i_blkbits;
646 #ifdef __NEED_I_SIZE_ORDERED
647 seqcount_t i_size_seqcount;
651 unsigned long i_state;
652 struct rw_semaphore i_rwsem;
654 unsigned long dirtied_when; /* jiffies of first dirtying */
655 unsigned long dirtied_time_when;
657 struct hlist_node i_hash;
658 struct list_head i_io_list; /* backing dev IO list */
659 #ifdef CONFIG_CGROUP_WRITEBACK
660 struct bdi_writeback *i_wb; /* the associated cgroup wb */
662 /* foreign inode detection, see wbc_detach_inode() */
664 u16 i_wb_frn_avg_time;
665 u16 i_wb_frn_history;
667 struct list_head i_lru; /* inode LRU list */
668 struct list_head i_sb_list;
670 struct hlist_head i_dentry;
671 struct rcu_head i_rcu;
675 atomic_t i_dio_count;
676 atomic_t i_writecount;
678 atomic_t i_readcount; /* struct files open RO */
680 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
681 struct file_lock_context *i_flctx;
682 struct address_space i_data;
683 struct list_head i_devices;
685 struct pipe_inode_info *i_pipe;
686 struct block_device *i_bdev;
694 #ifdef CONFIG_FSNOTIFY
695 __u32 i_fsnotify_mask; /* all events this inode cares about */
696 struct hlist_head i_fsnotify_marks;
699 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
700 struct fscrypt_info *i_crypt_info;
703 void *i_private; /* fs or device private pointer */
706 static inline int inode_unhashed(struct inode *inode)
708 return hlist_unhashed(&inode->i_hash);
712 * inode->i_mutex nesting subclasses for the lock validator:
714 * 0: the object of the current VFS operation
718 * 4: second non-directory
719 * 5: second parent (when locking independent directories in rename)
721 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
722 * non-directories at once.
724 * The locking order between these classes is
725 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
727 enum inode_i_mutex_lock_class
737 static inline void inode_lock(struct inode *inode)
739 down_write(&inode->i_rwsem);
742 static inline void inode_unlock(struct inode *inode)
744 up_write(&inode->i_rwsem);
747 static inline void inode_lock_shared(struct inode *inode)
749 down_read(&inode->i_rwsem);
752 static inline void inode_unlock_shared(struct inode *inode)
754 up_read(&inode->i_rwsem);
757 static inline int inode_trylock(struct inode *inode)
759 return down_write_trylock(&inode->i_rwsem);
762 static inline int inode_trylock_shared(struct inode *inode)
764 return down_read_trylock(&inode->i_rwsem);
767 static inline int inode_is_locked(struct inode *inode)
769 return rwsem_is_locked(&inode->i_rwsem);
772 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
774 down_write_nested(&inode->i_rwsem, subclass);
777 void lock_two_nondirectories(struct inode *, struct inode*);
778 void unlock_two_nondirectories(struct inode *, struct inode*);
781 * NOTE: in a 32bit arch with a preemptable kernel and
782 * an UP compile the i_size_read/write must be atomic
783 * with respect to the local cpu (unlike with preempt disabled),
784 * but they don't need to be atomic with respect to other cpus like in
785 * true SMP (so they need either to either locally disable irq around
786 * the read or for example on x86 they can be still implemented as a
787 * cmpxchg8b without the need of the lock prefix). For SMP compiles
788 * and 64bit archs it makes no difference if preempt is enabled or not.
790 static inline loff_t i_size_read(const struct inode *inode)
792 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
797 seq = read_seqcount_begin(&inode->i_size_seqcount);
798 i_size = inode->i_size;
799 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
801 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
805 i_size = inode->i_size;
809 return inode->i_size;
814 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
815 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
816 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
818 static inline void i_size_write(struct inode *inode, loff_t i_size)
820 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
822 write_seqcount_begin(&inode->i_size_seqcount);
823 inode->i_size = i_size;
824 write_seqcount_end(&inode->i_size_seqcount);
826 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
828 inode->i_size = i_size;
831 inode->i_size = i_size;
835 /* Helper functions so that in most cases filesystems will
836 * not need to deal directly with kuid_t and kgid_t and can
837 * instead deal with the raw numeric values that are stored
840 static inline uid_t i_uid_read(const struct inode *inode)
842 return from_kuid(&init_user_ns, inode->i_uid);
845 static inline gid_t i_gid_read(const struct inode *inode)
847 return from_kgid(&init_user_ns, inode->i_gid);
850 static inline void i_uid_write(struct inode *inode, uid_t uid)
852 inode->i_uid = make_kuid(&init_user_ns, uid);
855 static inline void i_gid_write(struct inode *inode, gid_t gid)
857 inode->i_gid = make_kgid(&init_user_ns, gid);
860 static inline unsigned iminor(const struct inode *inode)
862 return MINOR(inode->i_rdev);
865 static inline unsigned imajor(const struct inode *inode)
867 return MAJOR(inode->i_rdev);
870 extern struct block_device *I_BDEV(struct inode *inode);
873 rwlock_t lock; /* protects pid, uid, euid fields */
874 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
875 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
876 kuid_t uid, euid; /* uid/euid of process setting the owner */
877 int signum; /* posix.1b rt signal to be delivered on IO */
881 * Track a single file's readahead state
883 struct file_ra_state {
884 pgoff_t start; /* where readahead started */
885 unsigned int size; /* # of readahead pages */
886 unsigned int async_size; /* do asynchronous readahead when
887 there are only # of pages ahead */
889 unsigned int ra_pages; /* Maximum readahead window */
890 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
891 loff_t prev_pos; /* Cache last read() position */
895 * Check if @index falls in the readahead windows.
897 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
899 return (index >= ra->start &&
900 index < ra->start + ra->size);
905 struct llist_node fu_llist;
906 struct rcu_head fu_rcuhead;
909 struct inode *f_inode; /* cached value */
910 const struct file_operations *f_op;
913 * Protects f_ep_links, f_flags.
914 * Must not be taken from IRQ context.
917 atomic_long_t f_count;
918 unsigned int f_flags;
920 struct mutex f_pos_lock;
922 struct fown_struct f_owner;
923 const struct cred *f_cred;
924 struct file_ra_state f_ra;
927 #ifdef CONFIG_SECURITY
930 /* needed for tty driver, and maybe others */
934 /* Used by fs/eventpoll.c to link all the hooks to this file */
935 struct list_head f_ep_links;
936 struct list_head f_tfile_llink;
937 #endif /* #ifdef CONFIG_EPOLL */
938 struct address_space *f_mapping;
939 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
944 /* file identifier */
945 unsigned char f_handle[0];
948 static inline struct file *get_file(struct file *f)
950 atomic_long_inc(&f->f_count);
953 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
954 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
955 #define file_count(x) atomic_long_read(&(x)->f_count)
957 #define MAX_NON_LFS ((1UL<<31) - 1)
959 /* Page cache limit. The filesystems should put that into their s_maxbytes
960 limits, otherwise bad things can happen in VM. */
961 #if BITS_PER_LONG==32
962 #define MAX_LFS_FILESIZE (((loff_t)PAGE_SIZE << (BITS_PER_LONG-1))-1)
963 #elif BITS_PER_LONG==64
964 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
969 #define FL_DELEG 4 /* NFSv4 delegation */
970 #define FL_ACCESS 8 /* not trying to lock, just looking */
971 #define FL_EXISTS 16 /* when unlocking, test for existence */
972 #define FL_LEASE 32 /* lease held on this file */
973 #define FL_CLOSE 64 /* unlock on close */
974 #define FL_SLEEP 128 /* A blocking lock */
975 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
976 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
977 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
978 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
981 * Special return value from posix_lock_file() and vfs_lock_file() for
982 * asynchronous locking.
984 #define FILE_LOCK_DEFERRED 1
986 /* legacy typedef, should eventually be removed */
987 typedef void *fl_owner_t;
991 struct file_lock_operations {
992 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
993 void (*fl_release_private)(struct file_lock *);
996 struct lock_manager_operations {
997 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
998 unsigned long (*lm_owner_key)(struct file_lock *);
999 fl_owner_t (*lm_get_owner)(fl_owner_t);
1000 void (*lm_put_owner)(fl_owner_t);
1001 void (*lm_notify)(struct file_lock *); /* unblock callback */
1002 int (*lm_grant)(struct file_lock *, int);
1003 bool (*lm_break)(struct file_lock *);
1004 int (*lm_change)(struct file_lock *, int, struct list_head *);
1005 void (*lm_setup)(struct file_lock *, void **);
1008 struct lock_manager {
1009 struct list_head list;
1011 * NFSv4 and up also want opens blocked during the grace period;
1018 void locks_start_grace(struct net *, struct lock_manager *);
1019 void locks_end_grace(struct lock_manager *);
1020 int locks_in_grace(struct net *);
1021 int opens_in_grace(struct net *);
1023 /* that will die - we need it for nfs_lock_info */
1024 #include <linux/nfs_fs_i.h>
1027 * struct file_lock represents a generic "file lock". It's used to represent
1028 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1029 * note that the same struct is used to represent both a request for a lock and
1030 * the lock itself, but the same object is never used for both.
1032 * FIXME: should we create a separate "struct lock_request" to help distinguish
1035 * The varous i_flctx lists are ordered by:
1038 * 2) lock range start
1041 * Obviously, the last two criteria only matter for POSIX locks.
1044 struct file_lock *fl_next; /* singly linked list for this inode */
1045 struct list_head fl_list; /* link into file_lock_context */
1046 struct hlist_node fl_link; /* node in global lists */
1047 struct list_head fl_block; /* circular list of blocked processes */
1048 fl_owner_t fl_owner;
1049 unsigned int fl_flags;
1050 unsigned char fl_type;
1051 unsigned int fl_pid;
1052 int fl_link_cpu; /* what cpu's list is this on? */
1053 struct pid *fl_nspid;
1054 wait_queue_head_t fl_wait;
1055 struct file *fl_file;
1059 struct fasync_struct * fl_fasync; /* for lease break notifications */
1060 /* for lease breaks: */
1061 unsigned long fl_break_time;
1062 unsigned long fl_downgrade_time;
1064 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1065 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1067 struct nfs_lock_info nfs_fl;
1068 struct nfs4_lock_info nfs4_fl;
1070 struct list_head link; /* link in AFS vnode's pending_locks list */
1071 int state; /* state of grant or error if -ve */
1076 struct file_lock_context {
1077 spinlock_t flc_lock;
1078 struct list_head flc_flock;
1079 struct list_head flc_posix;
1080 struct list_head flc_lease;
1083 /* The following constant reflects the upper bound of the file/locking space */
1085 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1086 #define OFFSET_MAX INT_LIMIT(loff_t)
1087 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1090 #include <linux/fcntl.h>
1092 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1094 #ifdef CONFIG_FILE_LOCKING
1095 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
1096 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1097 struct flock __user *);
1099 #if BITS_PER_LONG == 32
1100 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
1101 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1102 struct flock64 __user *);
1105 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1106 extern int fcntl_getlease(struct file *filp);
1109 void locks_free_lock_context(struct inode *inode);
1110 void locks_free_lock(struct file_lock *fl);
1111 extern void locks_init_lock(struct file_lock *);
1112 extern struct file_lock * locks_alloc_lock(void);
1113 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1114 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1115 extern void locks_remove_posix(struct file *, fl_owner_t);
1116 extern void locks_remove_file(struct file *);
1117 extern void locks_release_private(struct file_lock *);
1118 extern void posix_test_lock(struct file *, struct file_lock *);
1119 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1120 extern int posix_unblock_lock(struct file_lock *);
1121 extern int vfs_test_lock(struct file *, struct file_lock *);
1122 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1123 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1124 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1125 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1126 extern void lease_get_mtime(struct inode *, struct timespec *time);
1127 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1128 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1129 extern int lease_modify(struct file_lock *, int, struct list_head *);
1130 struct files_struct;
1131 extern void show_fd_locks(struct seq_file *f,
1132 struct file *filp, struct files_struct *files);
1133 #else /* !CONFIG_FILE_LOCKING */
1134 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1135 struct flock __user *user)
1140 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1141 unsigned int cmd, struct flock __user *user)
1146 #if BITS_PER_LONG == 32
1147 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1148 struct flock64 __user *user)
1153 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1154 unsigned int cmd, struct flock64 __user *user)
1159 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1164 static inline int fcntl_getlease(struct file *filp)
1170 locks_free_lock_context(struct inode *inode)
1174 static inline void locks_init_lock(struct file_lock *fl)
1179 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1184 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1189 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1194 static inline void locks_remove_file(struct file *filp)
1199 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1204 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1205 struct file_lock *conflock)
1210 static inline int posix_unblock_lock(struct file_lock *waiter)
1215 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1220 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1221 struct file_lock *fl, struct file_lock *conf)
1226 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1231 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1236 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1241 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1246 static inline int generic_setlease(struct file *filp, long arg,
1247 struct file_lock **flp, void **priv)
1252 static inline int vfs_setlease(struct file *filp, long arg,
1253 struct file_lock **lease, void **priv)
1258 static inline int lease_modify(struct file_lock *fl, int arg,
1259 struct list_head *dispose)
1264 struct files_struct;
1265 static inline void show_fd_locks(struct seq_file *f,
1266 struct file *filp, struct files_struct *files) {}
1267 #endif /* !CONFIG_FILE_LOCKING */
1269 static inline struct inode *file_inode(const struct file *f)
1274 static inline struct dentry *file_dentry(const struct file *file)
1276 struct dentry *dentry = file->f_path.dentry;
1278 if (unlikely(dentry->d_flags & DCACHE_OP_REAL))
1279 return dentry->d_op->d_real(dentry, file_inode(file));
1284 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1286 return locks_lock_inode_wait(file_inode(filp), fl);
1289 struct fasync_struct {
1293 struct fasync_struct *fa_next; /* singly linked list */
1294 struct file *fa_file;
1295 struct rcu_head fa_rcu;
1298 #define FASYNC_MAGIC 0x4601
1300 /* SMP safe fasync helpers: */
1301 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1302 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1303 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1304 extern struct fasync_struct *fasync_alloc(void);
1305 extern void fasync_free(struct fasync_struct *);
1307 /* can be called from interrupts */
1308 extern void kill_fasync(struct fasync_struct **, int, int);
1310 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1311 extern void f_setown(struct file *filp, unsigned long arg, int force);
1312 extern void f_delown(struct file *filp);
1313 extern pid_t f_getown(struct file *filp);
1314 extern int send_sigurg(struct fown_struct *fown);
1322 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1323 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1324 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1325 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1326 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1329 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1330 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1332 /* Possible states of 'frozen' field */
1334 SB_UNFROZEN = 0, /* FS is unfrozen */
1335 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1336 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1337 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1338 * internal threads if needed) */
1339 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1342 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1345 int frozen; /* Is sb frozen? */
1346 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1347 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1350 struct super_block {
1351 struct list_head s_list; /* Keep this first */
1352 dev_t s_dev; /* search index; _not_ kdev_t */
1353 unsigned char s_blocksize_bits;
1354 unsigned long s_blocksize;
1355 loff_t s_maxbytes; /* Max file size */
1356 struct file_system_type *s_type;
1357 const struct super_operations *s_op;
1358 const struct dquot_operations *dq_op;
1359 const struct quotactl_ops *s_qcop;
1360 const struct export_operations *s_export_op;
1361 unsigned long s_flags;
1362 unsigned long s_iflags; /* internal SB_I_* flags */
1363 unsigned long s_magic;
1364 struct dentry *s_root;
1365 struct rw_semaphore s_umount;
1368 #ifdef CONFIG_SECURITY
1371 const struct xattr_handler **s_xattr;
1373 const struct fscrypt_operations *s_cop;
1375 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1376 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1377 struct block_device *s_bdev;
1378 struct backing_dev_info *s_bdi;
1379 struct mtd_info *s_mtd;
1380 struct hlist_node s_instances;
1381 unsigned int s_quota_types; /* Bitmask of supported quota types */
1382 struct quota_info s_dquot; /* Diskquota specific options */
1384 struct sb_writers s_writers;
1386 char s_id[32]; /* Informational name */
1387 u8 s_uuid[16]; /* UUID */
1389 void *s_fs_info; /* Filesystem private info */
1390 unsigned int s_max_links;
1393 /* Granularity of c/m/atime in ns.
1394 Cannot be worse than a second */
1398 * The next field is for VFS *only*. No filesystems have any business
1399 * even looking at it. You had been warned.
1401 struct mutex s_vfs_rename_mutex; /* Kludge */
1404 * Filesystem subtype. If non-empty the filesystem type field
1405 * in /proc/mounts will be "type.subtype"
1410 * Saved mount options for lazy filesystems using
1411 * generic_show_options()
1413 char __rcu *s_options;
1414 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1417 * Saved pool identifier for cleancache (-1 means none)
1419 int cleancache_poolid;
1421 struct shrinker s_shrink; /* per-sb shrinker handle */
1423 /* Number of inodes with nlink == 0 but still referenced */
1424 atomic_long_t s_remove_count;
1426 /* Being remounted read-only */
1427 int s_readonly_remount;
1429 /* AIO completions deferred from interrupt context */
1430 struct workqueue_struct *s_dio_done_wq;
1431 struct hlist_head s_pins;
1434 * Keep the lru lists last in the structure so they always sit on their
1435 * own individual cachelines.
1437 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1438 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1439 struct rcu_head rcu;
1440 struct work_struct destroy_work;
1442 struct mutex s_sync_lock; /* sync serialisation lock */
1445 * Indicates how deep in a filesystem stack this SB is
1449 /* s_inode_list_lock protects s_inodes */
1450 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1451 struct list_head s_inodes; /* all inodes */
1454 extern struct timespec current_fs_time(struct super_block *sb);
1457 * Snapshotting support.
1460 void __sb_end_write(struct super_block *sb, int level);
1461 int __sb_start_write(struct super_block *sb, int level, bool wait);
1463 #define __sb_writers_acquired(sb, lev) \
1464 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1465 #define __sb_writers_release(sb, lev) \
1466 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1469 * sb_end_write - drop write access to a superblock
1470 * @sb: the super we wrote to
1472 * Decrement number of writers to the filesystem. Wake up possible waiters
1473 * wanting to freeze the filesystem.
1475 static inline void sb_end_write(struct super_block *sb)
1477 __sb_end_write(sb, SB_FREEZE_WRITE);
1481 * sb_end_pagefault - drop write access to a superblock from a page fault
1482 * @sb: the super we wrote to
1484 * Decrement number of processes handling write page fault to the filesystem.
1485 * Wake up possible waiters wanting to freeze the filesystem.
1487 static inline void sb_end_pagefault(struct super_block *sb)
1489 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1493 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1494 * @sb: the super we wrote to
1496 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1497 * waiters wanting to freeze the filesystem.
1499 static inline void sb_end_intwrite(struct super_block *sb)
1501 __sb_end_write(sb, SB_FREEZE_FS);
1505 * sb_start_write - get write access to a superblock
1506 * @sb: the super we write to
1508 * When a process wants to write data or metadata to a file system (i.e. dirty
1509 * a page or an inode), it should embed the operation in a sb_start_write() -
1510 * sb_end_write() pair to get exclusion against file system freezing. This
1511 * function increments number of writers preventing freezing. If the file
1512 * system is already frozen, the function waits until the file system is
1515 * Since freeze protection behaves as a lock, users have to preserve
1516 * ordering of freeze protection and other filesystem locks. Generally,
1517 * freeze protection should be the outermost lock. In particular, we have:
1520 * -> i_mutex (write path, truncate, directory ops, ...)
1521 * -> s_umount (freeze_super, thaw_super)
1523 static inline void sb_start_write(struct super_block *sb)
1525 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1528 static inline int sb_start_write_trylock(struct super_block *sb)
1530 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1534 * sb_start_pagefault - get write access to a superblock from a page fault
1535 * @sb: the super we write to
1537 * When a process starts handling write page fault, it should embed the
1538 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1539 * exclusion against file system freezing. This is needed since the page fault
1540 * is going to dirty a page. This function increments number of running page
1541 * faults preventing freezing. If the file system is already frozen, the
1542 * function waits until the file system is thawed.
1544 * Since page fault freeze protection behaves as a lock, users have to preserve
1545 * ordering of freeze protection and other filesystem locks. It is advised to
1546 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1547 * handling code implies lock dependency:
1550 * -> sb_start_pagefault
1552 static inline void sb_start_pagefault(struct super_block *sb)
1554 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1558 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1559 * @sb: the super we write to
1561 * This is the third level of protection against filesystem freezing. It is
1562 * free for use by a filesystem. The only requirement is that it must rank
1563 * below sb_start_pagefault.
1565 * For example filesystem can call sb_start_intwrite() when starting a
1566 * transaction which somewhat eases handling of freezing for internal sources
1567 * of filesystem changes (internal fs threads, discarding preallocation on file
1570 static inline void sb_start_intwrite(struct super_block *sb)
1572 __sb_start_write(sb, SB_FREEZE_FS, true);
1576 extern bool inode_owner_or_capable(const struct inode *inode);
1579 * VFS helper functions..
1581 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1582 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1583 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1584 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1585 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1586 extern int vfs_rmdir(struct inode *, struct dentry *);
1587 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1588 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1589 extern int vfs_whiteout(struct inode *, struct dentry *);
1592 * VFS file helper functions.
1594 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1597 * VFS FS_IOC_FIEMAP helper definitions.
1599 struct fiemap_extent_info {
1600 unsigned int fi_flags; /* Flags as passed from user */
1601 unsigned int fi_extents_mapped; /* Number of mapped extents */
1602 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1603 struct fiemap_extent __user *fi_extents_start; /* Start of
1604 fiemap_extent array */
1606 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1607 u64 phys, u64 len, u32 flags);
1608 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1613 * NOTE! These match bits 12..15 of stat.st_mode
1614 * (ie "(i_mode >> 12) & 15").
1616 #define DT_UNKNOWN 0
1627 * This is the "filldir" function type, used by readdir() to let
1628 * the kernel specify what kind of dirent layout it wants to have.
1629 * This allows the kernel to read directories into kernel space or
1630 * to have different dirent layouts depending on the binary type.
1633 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1636 struct dir_context {
1637 const filldir_t actor;
1641 struct block_device_operations;
1643 /* These macros are for out of kernel modules to test that
1644 * the kernel supports the unlocked_ioctl and compat_ioctl
1645 * fields in struct file_operations. */
1646 #define HAVE_COMPAT_IOCTL 1
1647 #define HAVE_UNLOCKED_IOCTL 1
1650 * These flags let !MMU mmap() govern direct device mapping vs immediate
1651 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1653 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1654 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1655 * NOMMU_MAP_READ: Can be mapped for reading
1656 * NOMMU_MAP_WRITE: Can be mapped for writing
1657 * NOMMU_MAP_EXEC: Can be mapped for execution
1659 #define NOMMU_MAP_COPY 0x00000001
1660 #define NOMMU_MAP_DIRECT 0x00000008
1661 #define NOMMU_MAP_READ VM_MAYREAD
1662 #define NOMMU_MAP_WRITE VM_MAYWRITE
1663 #define NOMMU_MAP_EXEC VM_MAYEXEC
1665 #define NOMMU_VMFLAGS \
1666 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1671 struct file_operations {
1672 struct module *owner;
1673 loff_t (*llseek) (struct file *, loff_t, int);
1674 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1675 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1676 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1677 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1678 int (*iterate) (struct file *, struct dir_context *);
1679 int (*iterate_shared) (struct file *, struct dir_context *);
1680 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1681 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1682 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1683 int (*mmap) (struct file *, struct vm_area_struct *);
1684 int (*open) (struct inode *, struct file *);
1685 int (*flush) (struct file *, fl_owner_t id);
1686 int (*release) (struct inode *, struct file *);
1687 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1688 int (*aio_fsync) (struct kiocb *, int datasync);
1689 int (*fasync) (int, struct file *, int);
1690 int (*lock) (struct file *, int, struct file_lock *);
1691 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1692 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1693 int (*check_flags)(int);
1694 int (*flock) (struct file *, int, struct file_lock *);
1695 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1696 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1697 int (*setlease)(struct file *, long, struct file_lock **, void **);
1698 long (*fallocate)(struct file *file, int mode, loff_t offset,
1700 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1702 unsigned (*mmap_capabilities)(struct file *);
1704 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1705 loff_t, size_t, unsigned int);
1706 int (*clone_file_range)(struct file *, loff_t, struct file *, loff_t,
1708 ssize_t (*dedupe_file_range)(struct file *, u64, u64, struct file *,
1712 struct inode_operations {
1713 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1714 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1715 int (*permission) (struct inode *, int);
1716 struct posix_acl * (*get_acl)(struct inode *, int);
1718 int (*readlink) (struct dentry *, char __user *,int);
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 (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1733 int (*setxattr) (struct dentry *, struct inode *,
1734 const char *, const void *, size_t, int);
1735 ssize_t (*getxattr) (struct dentry *, struct inode *,
1736 const char *, void *, size_t);
1737 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1738 int (*removexattr) (struct dentry *, const char *);
1739 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1741 int (*update_time)(struct inode *, struct timespec *, int);
1742 int (*atomic_open)(struct inode *, struct dentry *,
1743 struct file *, unsigned open_flag,
1744 umode_t create_mode, int *opened);
1745 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1746 int (*set_acl)(struct inode *, struct posix_acl *, int);
1747 } ____cacheline_aligned;
1749 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1750 unsigned long nr_segs, unsigned long fast_segs,
1751 struct iovec *fast_pointer,
1752 struct iovec **ret_pointer);
1754 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1755 extern ssize_t __vfs_write(struct file *, const char __user *, size_t, loff_t *);
1756 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1757 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1758 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1759 unsigned long, loff_t *, int);
1760 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1761 unsigned long, loff_t *, int);
1762 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1763 loff_t, size_t, unsigned int);
1764 extern int vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1765 struct file *file_out, loff_t pos_out, u64 len);
1766 extern int vfs_dedupe_file_range(struct file *file,
1767 struct file_dedupe_range *same);
1769 struct super_operations {
1770 struct inode *(*alloc_inode)(struct super_block *sb);
1771 void (*destroy_inode)(struct inode *);
1773 void (*dirty_inode) (struct inode *, int flags);
1774 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1775 int (*drop_inode) (struct inode *);
1776 void (*evict_inode) (struct inode *);
1777 void (*put_super) (struct super_block *);
1778 int (*sync_fs)(struct super_block *sb, int wait);
1779 int (*freeze_super) (struct super_block *);
1780 int (*freeze_fs) (struct super_block *);
1781 int (*thaw_super) (struct super_block *);
1782 int (*unfreeze_fs) (struct super_block *);
1783 int (*statfs) (struct dentry *, struct kstatfs *);
1784 int (*remount_fs) (struct super_block *, int *, char *);
1785 void (*umount_begin) (struct super_block *);
1787 int (*show_options)(struct seq_file *, struct dentry *);
1788 int (*show_devname)(struct seq_file *, struct dentry *);
1789 int (*show_path)(struct seq_file *, struct dentry *);
1790 int (*show_stats)(struct seq_file *, struct dentry *);
1792 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1793 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1794 struct dquot **(*get_dquots)(struct inode *);
1796 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1797 long (*nr_cached_objects)(struct super_block *,
1798 struct shrink_control *);
1799 long (*free_cached_objects)(struct super_block *,
1800 struct shrink_control *);
1804 * Inode flags - they have no relation to superblock flags now
1806 #define S_SYNC 1 /* Writes are synced at once */
1807 #define S_NOATIME 2 /* Do not update access times */
1808 #define S_APPEND 4 /* Append-only file */
1809 #define S_IMMUTABLE 8 /* Immutable file */
1810 #define S_DEAD 16 /* removed, but still open directory */
1811 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1812 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1813 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1814 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1815 #define S_PRIVATE 512 /* Inode is fs-internal */
1816 #define S_IMA 1024 /* Inode has an associated IMA struct */
1817 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1818 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1819 #ifdef CONFIG_FS_DAX
1820 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1822 #define S_DAX 0 /* Make all the DAX code disappear */
1826 * Note that nosuid etc flags are inode-specific: setting some file-system
1827 * flags just means all the inodes inherit those flags by default. It might be
1828 * possible to override it selectively if you really wanted to with some
1829 * ioctl() that is not currently implemented.
1831 * Exception: MS_RDONLY is always applied to the entire file system.
1833 * Unfortunately, it is possible to change a filesystems flags with it mounted
1834 * with files in use. This means that all of the inodes will not have their
1835 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1836 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1838 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1840 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1841 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1842 ((inode)->i_flags & S_SYNC))
1843 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1844 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1845 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1846 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1847 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1849 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1850 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1851 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1852 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1854 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1855 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1856 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1857 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1858 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1859 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1860 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1861 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1863 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1864 (inode)->i_rdev == WHITEOUT_DEV)
1867 * Inode state bits. Protected by inode->i_lock
1869 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1870 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1872 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1873 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1874 * various stages of removing an inode.
1876 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1878 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1879 * fdatasync(). i_atime is the usual cause.
1880 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1881 * these changes separately from I_DIRTY_SYNC so that we
1882 * don't have to write inode on fdatasync() when only
1883 * mtime has changed in it.
1884 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1885 * I_NEW Serves as both a mutex and completion notification.
1886 * New inodes set I_NEW. If two processes both create
1887 * the same inode, one of them will release its inode and
1888 * wait for I_NEW to be released before returning.
1889 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1890 * also cause waiting on I_NEW, without I_NEW actually
1891 * being set. find_inode() uses this to prevent returning
1892 * nearly-dead inodes.
1893 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1894 * is zero. I_FREEING must be set when I_WILL_FREE is
1896 * I_FREEING Set when inode is about to be freed but still has dirty
1897 * pages or buffers attached or the inode itself is still
1899 * I_CLEAR Added by clear_inode(). In this state the inode is
1900 * clean and can be destroyed. Inode keeps I_FREEING.
1902 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1903 * prohibited for many purposes. iget() must wait for
1904 * the inode to be completely released, then create it
1905 * anew. Other functions will just ignore such inodes,
1906 * if appropriate. I_NEW is used for waiting.
1908 * I_SYNC Writeback of inode is running. The bit is set during
1909 * data writeback, and cleared with a wakeup on the bit
1910 * address once it is done. The bit is also used to pin
1911 * the inode in memory for flusher thread.
1913 * I_REFERENCED Marks the inode as recently references on the LRU list.
1915 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1917 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1918 * synchronize competing switching instances and to tell
1919 * wb stat updates to grab mapping->tree_lock. See
1920 * inode_switch_wb_work_fn() for details.
1922 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1924 #define I_DIRTY_SYNC (1 << 0)
1925 #define I_DIRTY_DATASYNC (1 << 1)
1926 #define I_DIRTY_PAGES (1 << 2)
1928 #define I_NEW (1 << __I_NEW)
1929 #define I_WILL_FREE (1 << 4)
1930 #define I_FREEING (1 << 5)
1931 #define I_CLEAR (1 << 6)
1933 #define I_SYNC (1 << __I_SYNC)
1934 #define I_REFERENCED (1 << 8)
1935 #define __I_DIO_WAKEUP 9
1936 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1937 #define I_LINKABLE (1 << 10)
1938 #define I_DIRTY_TIME (1 << 11)
1939 #define __I_DIRTY_TIME_EXPIRED 12
1940 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1941 #define I_WB_SWITCH (1 << 13)
1943 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1944 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1946 extern void __mark_inode_dirty(struct inode *, int);
1947 static inline void mark_inode_dirty(struct inode *inode)
1949 __mark_inode_dirty(inode, I_DIRTY);
1952 static inline void mark_inode_dirty_sync(struct inode *inode)
1954 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1957 extern void inc_nlink(struct inode *inode);
1958 extern void drop_nlink(struct inode *inode);
1959 extern void clear_nlink(struct inode *inode);
1960 extern void set_nlink(struct inode *inode, unsigned int nlink);
1962 static inline void inode_inc_link_count(struct inode *inode)
1965 mark_inode_dirty(inode);
1968 static inline void inode_dec_link_count(struct inode *inode)
1971 mark_inode_dirty(inode);
1975 * inode_inc_iversion - increments i_version
1976 * @inode: inode that need to be updated
1978 * Every time the inode is modified, the i_version field will be incremented.
1979 * The filesystem has to be mounted with i_version flag
1982 static inline void inode_inc_iversion(struct inode *inode)
1984 spin_lock(&inode->i_lock);
1986 spin_unlock(&inode->i_lock);
1989 enum file_time_flags {
1996 extern bool atime_needs_update(const struct path *, struct inode *);
1997 extern void touch_atime(const struct path *);
1998 static inline void file_accessed(struct file *file)
2000 if (!(file->f_flags & O_NOATIME))
2001 touch_atime(&file->f_path);
2004 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2005 int sync_inode_metadata(struct inode *inode, int wait);
2007 struct file_system_type {
2010 #define FS_REQUIRES_DEV 1
2011 #define FS_BINARY_MOUNTDATA 2
2012 #define FS_HAS_SUBTYPE 4
2013 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2014 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
2015 #define FS_USERNS_VISIBLE 32 /* FS must already be visible */
2016 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2017 struct dentry *(*mount) (struct file_system_type *, int,
2018 const char *, void *);
2019 void (*kill_sb) (struct super_block *);
2020 struct module *owner;
2021 struct file_system_type * next;
2022 struct hlist_head fs_supers;
2024 struct lock_class_key s_lock_key;
2025 struct lock_class_key s_umount_key;
2026 struct lock_class_key s_vfs_rename_key;
2027 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2029 struct lock_class_key i_lock_key;
2030 struct lock_class_key i_mutex_key;
2031 struct lock_class_key i_mutex_dir_key;
2034 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2036 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
2037 void *data, int (*fill_super)(struct super_block *, void *, int));
2038 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2039 int flags, const char *dev_name, void *data,
2040 int (*fill_super)(struct super_block *, void *, int));
2041 extern struct dentry *mount_single(struct file_system_type *fs_type,
2042 int flags, void *data,
2043 int (*fill_super)(struct super_block *, void *, int));
2044 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2045 int flags, void *data,
2046 int (*fill_super)(struct super_block *, void *, int));
2047 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2048 void generic_shutdown_super(struct super_block *sb);
2049 void kill_block_super(struct super_block *sb);
2050 void kill_anon_super(struct super_block *sb);
2051 void kill_litter_super(struct super_block *sb);
2052 void deactivate_super(struct super_block *sb);
2053 void deactivate_locked_super(struct super_block *sb);
2054 int set_anon_super(struct super_block *s, void *data);
2055 int get_anon_bdev(dev_t *);
2056 void free_anon_bdev(dev_t);
2057 struct super_block *sget(struct file_system_type *type,
2058 int (*test)(struct super_block *,void *),
2059 int (*set)(struct super_block *,void *),
2060 int flags, void *data);
2061 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
2062 const struct super_operations *ops,
2063 const struct dentry_operations *dops,
2066 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2067 #define fops_get(fops) \
2068 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2069 #define fops_put(fops) \
2070 do { if (fops) module_put((fops)->owner); } while(0)
2072 * This one is to be used *ONLY* from ->open() instances.
2073 * fops must be non-NULL, pinned down *and* module dependencies
2074 * should be sufficient to pin the caller down as well.
2076 #define replace_fops(f, fops) \
2078 struct file *__file = (f); \
2079 fops_put(__file->f_op); \
2080 BUG_ON(!(__file->f_op = (fops))); \
2083 extern int register_filesystem(struct file_system_type *);
2084 extern int unregister_filesystem(struct file_system_type *);
2085 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2086 #define kern_mount(type) kern_mount_data(type, NULL)
2087 extern void kern_unmount(struct vfsmount *mnt);
2088 extern int may_umount_tree(struct vfsmount *);
2089 extern int may_umount(struct vfsmount *);
2090 extern long do_mount(const char *, const char __user *,
2091 const char *, unsigned long, void *);
2092 extern struct vfsmount *collect_mounts(struct path *);
2093 extern void drop_collected_mounts(struct vfsmount *);
2094 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2096 extern int vfs_statfs(struct path *, struct kstatfs *);
2097 extern int user_statfs(const char __user *, struct kstatfs *);
2098 extern int fd_statfs(int, struct kstatfs *);
2099 extern int vfs_ustat(dev_t, struct kstatfs *);
2100 extern int freeze_super(struct super_block *super);
2101 extern int thaw_super(struct super_block *super);
2102 extern bool our_mnt(struct vfsmount *mnt);
2104 extern int current_umask(void);
2106 extern void ihold(struct inode * inode);
2107 extern void iput(struct inode *);
2108 extern int generic_update_time(struct inode *, struct timespec *, int);
2111 extern struct kobject *fs_kobj;
2113 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2115 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2116 extern int locks_mandatory_locked(struct file *);
2117 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2120 * Candidates for mandatory locking have the setgid bit set
2121 * but no group execute bit - an otherwise meaningless combination.
2124 static inline int __mandatory_lock(struct inode *ino)
2126 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2130 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2131 * otherwise these will be advisory locks
2134 static inline int mandatory_lock(struct inode *ino)
2136 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2139 static inline int locks_verify_locked(struct file *file)
2141 if (mandatory_lock(file_inode(file)))
2142 return locks_mandatory_locked(file);
2146 static inline int locks_verify_truncate(struct inode *inode,
2150 if (!inode->i_flctx || !mandatory_lock(inode))
2153 if (size < inode->i_size) {
2154 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2157 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2162 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2164 static inline int locks_mandatory_locked(struct file *file)
2169 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2170 loff_t start, loff_t end, unsigned char type)
2175 static inline int __mandatory_lock(struct inode *inode)
2180 static inline int mandatory_lock(struct inode *inode)
2185 static inline int locks_verify_locked(struct file *file)
2190 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2196 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2199 #ifdef CONFIG_FILE_LOCKING
2200 static inline int break_lease(struct inode *inode, unsigned int mode)
2203 * Since this check is lockless, we must ensure that any refcounts
2204 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2205 * could end up racing with tasks trying to set a new lease on this
2209 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2210 return __break_lease(inode, mode, FL_LEASE);
2214 static inline int break_deleg(struct inode *inode, unsigned int mode)
2217 * Since this check is lockless, we must ensure that any refcounts
2218 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2219 * could end up racing with tasks trying to set a new lease on this
2223 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2224 return __break_lease(inode, mode, FL_DELEG);
2228 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2232 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2233 if (ret == -EWOULDBLOCK && delegated_inode) {
2234 *delegated_inode = inode;
2240 static inline int break_deleg_wait(struct inode **delegated_inode)
2244 ret = break_deleg(*delegated_inode, O_WRONLY);
2245 iput(*delegated_inode);
2246 *delegated_inode = NULL;
2250 static inline int break_layout(struct inode *inode, bool wait)
2253 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2254 return __break_lease(inode,
2255 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2260 #else /* !CONFIG_FILE_LOCKING */
2261 static inline int break_lease(struct inode *inode, unsigned int mode)
2266 static inline int break_deleg(struct inode *inode, unsigned int mode)
2271 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2276 static inline int break_deleg_wait(struct inode **delegated_inode)
2282 static inline int break_layout(struct inode *inode, bool wait)
2287 #endif /* CONFIG_FILE_LOCKING */
2292 const char *name; /* pointer to actual string */
2293 const __user char *uptr; /* original userland pointer */
2294 struct audit_names *aname;
2299 extern long vfs_truncate(const struct path *, loff_t);
2300 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
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 bool 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 /* Marks the bottom of the first segment of free char majors */
2424 #define CHRDEV_MAJOR_DYN_END 234
2425 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2426 extern int register_chrdev_region(dev_t, unsigned, const char *);
2427 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2428 unsigned int count, const char *name,
2429 const struct file_operations *fops);
2430 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2431 unsigned int count, const char *name);
2432 extern void unregister_chrdev_region(dev_t, unsigned);
2433 extern void chrdev_show(struct seq_file *,off_t);
2435 static inline int register_chrdev(unsigned int major, const char *name,
2436 const struct file_operations *fops)
2438 return __register_chrdev(major, 0, 256, name, fops);
2441 static inline void unregister_chrdev(unsigned int major, const char *name)
2443 __unregister_chrdev(major, 0, 256, name);
2446 /* fs/block_dev.c */
2447 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2448 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2451 #define BLKDEV_MAJOR_HASH_SIZE 255
2452 extern const char *__bdevname(dev_t, char *buffer);
2453 extern const char *bdevname(struct block_device *bdev, char *buffer);
2454 extern struct block_device *lookup_bdev(const char *);
2455 extern void blkdev_show(struct seq_file *,off_t);
2458 #define BLKDEV_MAJOR_HASH_SIZE 0
2461 extern void init_special_inode(struct inode *, umode_t, dev_t);
2463 /* Invalid inode operations -- fs/bad_inode.c */
2464 extern void make_bad_inode(struct inode *);
2465 extern bool is_bad_inode(struct inode *);
2469 * return READ, READA, or WRITE
2471 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2474 * return data direction, READ or WRITE
2476 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2478 extern void check_disk_size_change(struct gendisk *disk,
2479 struct block_device *bdev);
2480 extern int revalidate_disk(struct gendisk *);
2481 extern int check_disk_change(struct block_device *);
2482 extern int __invalidate_device(struct block_device *, bool);
2483 extern int invalidate_partition(struct gendisk *, int);
2485 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2486 pgoff_t start, pgoff_t end);
2488 static inline void invalidate_remote_inode(struct inode *inode)
2490 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2491 S_ISLNK(inode->i_mode))
2492 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2494 extern int invalidate_inode_pages2(struct address_space *mapping);
2495 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2496 pgoff_t start, pgoff_t end);
2497 extern int write_inode_now(struct inode *, int);
2498 extern int filemap_fdatawrite(struct address_space *);
2499 extern int filemap_flush(struct address_space *);
2500 extern int filemap_fdatawait(struct address_space *);
2501 extern void filemap_fdatawait_keep_errors(struct address_space *);
2502 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2504 extern int filemap_write_and_wait(struct address_space *mapping);
2505 extern int filemap_write_and_wait_range(struct address_space *mapping,
2506 loff_t lstart, loff_t lend);
2507 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2508 loff_t start, loff_t end, int sync_mode);
2509 extern int filemap_fdatawrite_range(struct address_space *mapping,
2510 loff_t start, loff_t end);
2512 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2514 extern int vfs_fsync(struct file *file, int datasync);
2517 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2518 * to already be updated for the write, and will return either the amount
2519 * of bytes passed in, or an error if syncing the file failed.
2521 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2523 if (iocb->ki_flags & IOCB_DSYNC) {
2524 int ret = vfs_fsync_range(iocb->ki_filp,
2525 iocb->ki_pos - count, iocb->ki_pos - 1,
2526 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2534 extern void emergency_sync(void);
2535 extern void emergency_remount(void);
2537 extern sector_t bmap(struct inode *, sector_t);
2539 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2540 extern int inode_permission(struct inode *, int);
2541 extern int __inode_permission(struct inode *, int);
2542 extern int generic_permission(struct inode *, int);
2543 extern int __check_sticky(struct inode *dir, struct inode *inode);
2545 static inline bool execute_ok(struct inode *inode)
2547 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2550 static inline void file_start_write(struct file *file)
2552 if (!S_ISREG(file_inode(file)->i_mode))
2554 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2557 static inline bool file_start_write_trylock(struct file *file)
2559 if (!S_ISREG(file_inode(file)->i_mode))
2561 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2564 static inline void file_end_write(struct file *file)
2566 if (!S_ISREG(file_inode(file)->i_mode))
2568 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2572 * get_write_access() gets write permission for a file.
2573 * put_write_access() releases this write permission.
2574 * This is used for regular files.
2575 * We cannot support write (and maybe mmap read-write shared) accesses and
2576 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2577 * can have the following values:
2578 * 0: no writers, no VM_DENYWRITE mappings
2579 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2580 * > 0: (i_writecount) users are writing to the file.
2582 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2583 * except for the cases where we don't hold i_writecount yet. Then we need to
2584 * use {get,deny}_write_access() - these functions check the sign and refuse
2585 * to do the change if sign is wrong.
2587 static inline int get_write_access(struct inode *inode)
2589 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2591 static inline int deny_write_access(struct file *file)
2593 struct inode *inode = file_inode(file);
2594 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2596 static inline void put_write_access(struct inode * inode)
2598 atomic_dec(&inode->i_writecount);
2600 static inline void allow_write_access(struct file *file)
2603 atomic_inc(&file_inode(file)->i_writecount);
2605 static inline bool inode_is_open_for_write(const struct inode *inode)
2607 return atomic_read(&inode->i_writecount) > 0;
2611 static inline void i_readcount_dec(struct inode *inode)
2613 BUG_ON(!atomic_read(&inode->i_readcount));
2614 atomic_dec(&inode->i_readcount);
2616 static inline void i_readcount_inc(struct inode *inode)
2618 atomic_inc(&inode->i_readcount);
2621 static inline void i_readcount_dec(struct inode *inode)
2625 static inline void i_readcount_inc(struct inode *inode)
2630 extern int do_pipe_flags(int *, int);
2632 #define __kernel_read_file_id(id) \
2633 id(UNKNOWN, unknown) \
2634 id(FIRMWARE, firmware) \
2635 id(MODULE, kernel-module) \
2636 id(KEXEC_IMAGE, kexec-image) \
2637 id(KEXEC_INITRAMFS, kexec-initramfs) \
2638 id(POLICY, security-policy) \
2641 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2642 #define __fid_stringify(dummy, str) #str,
2644 enum kernel_read_file_id {
2645 __kernel_read_file_id(__fid_enumify)
2648 static const char * const kernel_read_file_str[] = {
2649 __kernel_read_file_id(__fid_stringify)
2652 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
2654 if (id < 0 || id >= READING_MAX_ID)
2655 return kernel_read_file_str[READING_UNKNOWN];
2657 return kernel_read_file_str[id];
2660 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2661 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2662 enum kernel_read_file_id);
2663 extern int kernel_read_file_from_path(char *, void **, loff_t *, loff_t,
2664 enum kernel_read_file_id);
2665 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2666 enum kernel_read_file_id);
2667 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2668 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
2669 extern struct file * open_exec(const char *);
2671 /* fs/dcache.c -- generic fs support functions */
2672 extern bool is_subdir(struct dentry *, struct dentry *);
2673 extern bool path_is_under(struct path *, struct path *);
2675 extern char *file_path(struct file *, char *, int);
2677 #include <linux/err.h>
2679 /* needed for stackable file system support */
2680 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2682 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2684 extern int inode_init_always(struct super_block *, struct inode *);
2685 extern void inode_init_once(struct inode *);
2686 extern void address_space_init_once(struct address_space *mapping);
2687 extern struct inode * igrab(struct inode *);
2688 extern ino_t iunique(struct super_block *, ino_t);
2689 extern int inode_needs_sync(struct inode *inode);
2690 extern int generic_delete_inode(struct inode *inode);
2691 static inline int generic_drop_inode(struct inode *inode)
2693 return !inode->i_nlink || inode_unhashed(inode);
2696 extern struct inode *ilookup5_nowait(struct super_block *sb,
2697 unsigned long hashval, int (*test)(struct inode *, void *),
2699 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2700 int (*test)(struct inode *, void *), void *data);
2701 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2703 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2704 extern struct inode * iget_locked(struct super_block *, unsigned long);
2705 extern struct inode *find_inode_nowait(struct super_block *,
2707 int (*match)(struct inode *,
2708 unsigned long, void *),
2710 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2711 extern int insert_inode_locked(struct inode *);
2712 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2713 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2715 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2717 extern void unlock_new_inode(struct inode *);
2718 extern unsigned int get_next_ino(void);
2720 extern void __iget(struct inode * inode);
2721 extern void iget_failed(struct inode *);
2722 extern void clear_inode(struct inode *);
2723 extern void __destroy_inode(struct inode *);
2724 extern struct inode *new_inode_pseudo(struct super_block *sb);
2725 extern struct inode *new_inode(struct super_block *sb);
2726 extern void free_inode_nonrcu(struct inode *inode);
2727 extern int should_remove_suid(struct dentry *);
2728 extern int file_remove_privs(struct file *);
2729 extern int dentry_needs_remove_privs(struct dentry *dentry);
2730 static inline int file_needs_remove_privs(struct file *file)
2732 return dentry_needs_remove_privs(file->f_path.dentry);
2735 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2736 static inline void insert_inode_hash(struct inode *inode)
2738 __insert_inode_hash(inode, inode->i_ino);
2741 extern void __remove_inode_hash(struct inode *);
2742 static inline void remove_inode_hash(struct inode *inode)
2744 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2745 __remove_inode_hash(inode);
2748 extern void inode_sb_list_add(struct inode *inode);
2751 extern blk_qc_t submit_bio(int, struct bio *);
2752 extern int bdev_read_only(struct block_device *);
2754 extern int set_blocksize(struct block_device *, int);
2755 extern int sb_set_blocksize(struct super_block *, int);
2756 extern int sb_min_blocksize(struct super_block *, int);
2758 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2759 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2760 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2761 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2762 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2763 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2764 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
2765 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2767 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos);
2768 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos);
2770 /* fs/block_dev.c */
2771 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2772 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2773 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2775 extern void block_sync_page(struct page *page);
2778 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2779 struct pipe_inode_info *, size_t, unsigned int);
2780 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2781 struct pipe_inode_info *, size_t, unsigned int);
2782 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2783 struct file *, loff_t *, size_t, unsigned int);
2784 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2785 struct file *out, loff_t *, size_t len, unsigned int flags);
2786 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2787 loff_t *opos, size_t len, unsigned int flags);
2791 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2792 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2793 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2794 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2795 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2796 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2797 int whence, loff_t maxsize, loff_t eof);
2798 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2799 int whence, loff_t size);
2800 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
2801 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
2802 extern int generic_file_open(struct inode * inode, struct file * filp);
2803 extern int nonseekable_open(struct inode * inode, struct file * filp);
2806 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2807 loff_t file_offset);
2810 /* need locking between buffered and direct access */
2813 /* filesystem does not support filling holes */
2814 DIO_SKIP_HOLES = 0x02,
2816 /* filesystem can handle aio writes beyond i_size */
2817 DIO_ASYNC_EXTEND = 0x04,
2819 /* inode/fs/bdev does not need truncate protection */
2820 DIO_SKIP_DIO_COUNT = 0x08,
2823 void dio_end_io(struct bio *bio, int error);
2825 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2826 struct block_device *bdev, struct iov_iter *iter,
2827 get_block_t get_block,
2828 dio_iodone_t end_io, dio_submit_t submit_io,
2831 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2832 struct inode *inode,
2833 struct iov_iter *iter,
2834 get_block_t get_block)
2836 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
2837 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
2841 void inode_dio_wait(struct inode *inode);
2844 * inode_dio_begin - signal start of a direct I/O requests
2845 * @inode: inode the direct I/O happens on
2847 * This is called once we've finished processing a direct I/O request,
2848 * and is used to wake up callers waiting for direct I/O to be quiesced.
2850 static inline void inode_dio_begin(struct inode *inode)
2852 atomic_inc(&inode->i_dio_count);
2856 * inode_dio_end - signal finish of a direct I/O requests
2857 * @inode: inode the direct I/O happens on
2859 * This is called once we've finished processing a direct I/O request,
2860 * and is used to wake up callers waiting for direct I/O to be quiesced.
2862 static inline void inode_dio_end(struct inode *inode)
2864 if (atomic_dec_and_test(&inode->i_dio_count))
2865 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
2868 extern void inode_set_flags(struct inode *inode, unsigned int flags,
2871 extern const struct file_operations generic_ro_fops;
2873 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2875 extern int readlink_copy(char __user *, int, const char *);
2876 extern int page_readlink(struct dentry *, char __user *, int);
2877 extern const char *page_get_link(struct dentry *, struct inode *,
2878 struct delayed_call *);
2879 extern void page_put_link(void *);
2880 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2882 extern int page_symlink(struct inode *inode, const char *symname, int len);
2883 extern const struct inode_operations page_symlink_inode_operations;
2884 extern void kfree_link(void *);
2885 extern int generic_readlink(struct dentry *, char __user *, int);
2886 extern void generic_fillattr(struct inode *, struct kstat *);
2887 int vfs_getattr_nosec(struct path *path, struct kstat *stat);
2888 extern int vfs_getattr(struct path *, struct kstat *);
2889 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2890 void inode_add_bytes(struct inode *inode, loff_t bytes);
2891 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2892 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2893 loff_t inode_get_bytes(struct inode *inode);
2894 void inode_set_bytes(struct inode *inode, loff_t bytes);
2895 const char *simple_get_link(struct dentry *, struct inode *,
2896 struct delayed_call *);
2897 extern const struct inode_operations simple_symlink_inode_operations;
2899 extern int iterate_dir(struct file *, struct dir_context *);
2901 extern int vfs_stat(const char __user *, struct kstat *);
2902 extern int vfs_lstat(const char __user *, struct kstat *);
2903 extern int vfs_fstat(unsigned int, struct kstat *);
2904 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2906 extern int __generic_block_fiemap(struct inode *inode,
2907 struct fiemap_extent_info *fieinfo,
2908 loff_t start, loff_t len,
2909 get_block_t *get_block);
2910 extern int generic_block_fiemap(struct inode *inode,
2911 struct fiemap_extent_info *fieinfo, u64 start,
2912 u64 len, get_block_t *get_block);
2914 extern void get_filesystem(struct file_system_type *fs);
2915 extern void put_filesystem(struct file_system_type *fs);
2916 extern struct file_system_type *get_fs_type(const char *name);
2917 extern struct super_block *get_super(struct block_device *);
2918 extern struct super_block *get_super_thawed(struct block_device *);
2919 extern struct super_block *get_active_super(struct block_device *bdev);
2920 extern void drop_super(struct super_block *sb);
2921 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2922 extern void iterate_supers_type(struct file_system_type *,
2923 void (*)(struct super_block *, void *), void *);
2925 extern int dcache_dir_open(struct inode *, struct file *);
2926 extern int dcache_dir_close(struct inode *, struct file *);
2927 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2928 extern int dcache_readdir(struct file *, struct dir_context *);
2929 extern int simple_setattr(struct dentry *, struct iattr *);
2930 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2931 extern int simple_statfs(struct dentry *, struct kstatfs *);
2932 extern int simple_open(struct inode *inode, struct file *file);
2933 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2934 extern int simple_unlink(struct inode *, struct dentry *);
2935 extern int simple_rmdir(struct inode *, struct dentry *);
2936 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2937 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2938 extern int simple_empty(struct dentry *);
2939 extern int simple_readpage(struct file *file, struct page *page);
2940 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2941 loff_t pos, unsigned len, unsigned flags,
2942 struct page **pagep, void **fsdata);
2943 extern int simple_write_end(struct file *file, struct address_space *mapping,
2944 loff_t pos, unsigned len, unsigned copied,
2945 struct page *page, void *fsdata);
2946 extern int always_delete_dentry(const struct dentry *);
2947 extern struct inode *alloc_anon_inode(struct super_block *);
2948 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
2949 extern const struct dentry_operations simple_dentry_operations;
2951 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2952 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2953 extern const struct file_operations simple_dir_operations;
2954 extern const struct inode_operations simple_dir_inode_operations;
2955 extern void make_empty_dir_inode(struct inode *inode);
2956 extern bool is_empty_dir_inode(struct inode *inode);
2957 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2958 struct dentry *d_alloc_name(struct dentry *, const char *);
2959 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2960 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2961 extern void simple_release_fs(struct vfsmount **mount, int *count);
2963 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2964 loff_t *ppos, const void *from, size_t available);
2965 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2966 const void __user *from, size_t count);
2968 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
2969 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2971 extern int generic_check_addressable(unsigned, u64);
2973 #ifdef CONFIG_MIGRATION
2974 extern int buffer_migrate_page(struct address_space *,
2975 struct page *, struct page *,
2978 #define buffer_migrate_page NULL
2981 extern int inode_change_ok(const struct inode *, struct iattr *);
2982 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2983 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2985 extern int file_update_time(struct file *file);
2987 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2988 extern void save_mount_options(struct super_block *sb, char *options);
2989 extern void replace_mount_options(struct super_block *sb, char *options);
2991 static inline bool io_is_direct(struct file *filp)
2993 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
2996 static inline int iocb_flags(struct file *file)
2999 if (file->f_flags & O_APPEND)
3001 if (io_is_direct(file))
3003 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3005 if (file->f_flags & __O_SYNC)
3010 static inline ino_t parent_ino(struct dentry *dentry)
3015 * Don't strictly need d_lock here? If the parent ino could change
3016 * then surely we'd have a deeper race in the caller?
3018 spin_lock(&dentry->d_lock);
3019 res = dentry->d_parent->d_inode->i_ino;
3020 spin_unlock(&dentry->d_lock);
3024 /* Transaction based IO helpers */
3027 * An argresp is stored in an allocated page and holds the
3028 * size of the argument or response, along with its content
3030 struct simple_transaction_argresp {
3035 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3037 char *simple_transaction_get(struct file *file, const char __user *buf,
3039 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3040 size_t size, loff_t *pos);
3041 int simple_transaction_release(struct inode *inode, struct file *file);
3043 void simple_transaction_set(struct file *file, size_t n);
3046 * simple attribute files
3048 * These attributes behave similar to those in sysfs:
3050 * Writing to an attribute immediately sets a value, an open file can be
3051 * written to multiple times.
3053 * Reading from an attribute creates a buffer from the value that might get
3054 * read with multiple read calls. When the attribute has been read
3055 * completely, no further read calls are possible until the file is opened
3058 * All attributes contain a text representation of a numeric value
3059 * that are accessed with the get() and set() functions.
3061 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3062 static int __fops ## _open(struct inode *inode, struct file *file) \
3064 __simple_attr_check_format(__fmt, 0ull); \
3065 return simple_attr_open(inode, file, __get, __set, __fmt); \
3067 static const struct file_operations __fops = { \
3068 .owner = THIS_MODULE, \
3069 .open = __fops ## _open, \
3070 .release = simple_attr_release, \
3071 .read = simple_attr_read, \
3072 .write = simple_attr_write, \
3073 .llseek = generic_file_llseek, \
3076 static inline __printf(1, 2)
3077 void __simple_attr_check_format(const char *fmt, ...)
3079 /* don't do anything, just let the compiler check the arguments; */
3082 int simple_attr_open(struct inode *inode, struct file *file,
3083 int (*get)(void *, u64 *), int (*set)(void *, u64),
3085 int simple_attr_release(struct inode *inode, struct file *file);
3086 ssize_t simple_attr_read(struct file *file, char __user *buf,
3087 size_t len, loff_t *ppos);
3088 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3089 size_t len, loff_t *ppos);
3092 int proc_nr_files(struct ctl_table *table, int write,
3093 void __user *buffer, size_t *lenp, loff_t *ppos);
3094 int proc_nr_dentry(struct ctl_table *table, int write,
3095 void __user *buffer, size_t *lenp, loff_t *ppos);
3096 int proc_nr_inodes(struct ctl_table *table, int write,
3097 void __user *buffer, size_t *lenp, loff_t *ppos);
3098 int __init get_filesystem_list(char *buf);
3100 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3101 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3103 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3104 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3105 (flag & __FMODE_NONOTIFY)))
3107 static inline bool is_sxid(umode_t mode)
3109 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3112 static inline int check_sticky(struct inode *dir, struct inode *inode)
3114 if (!(dir->i_mode & S_ISVTX))
3117 return __check_sticky(dir, inode);
3120 static inline void inode_has_no_xattr(struct inode *inode)
3122 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
3123 inode->i_flags |= S_NOSEC;
3126 static inline bool is_root_inode(struct inode *inode)
3128 return inode == inode->i_sb->s_root->d_inode;
3131 static inline bool dir_emit(struct dir_context *ctx,
3132 const char *name, int namelen,
3133 u64 ino, unsigned type)
3135 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3137 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3139 return ctx->actor(ctx, ".", 1, ctx->pos,
3140 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3142 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3144 return ctx->actor(ctx, "..", 2, ctx->pos,
3145 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3147 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3149 if (ctx->pos == 0) {
3150 if (!dir_emit_dot(file, ctx))
3154 if (ctx->pos == 1) {
3155 if (!dir_emit_dotdot(file, ctx))
3161 static inline bool dir_relax(struct inode *inode)
3163 inode_unlock(inode);
3165 return !IS_DEADDIR(inode);
3168 static inline bool dir_relax_shared(struct inode *inode)
3170 inode_unlock_shared(inode);
3171 inode_lock_shared(inode);
3172 return !IS_DEADDIR(inode);
3175 extern bool path_noexec(const struct path *path);
3176 extern void inode_nohighmem(struct inode *inode);
3178 #endif /* _LINUX_FS_H */