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/mm_types.h>
22 #include <linux/capability.h>
23 #include <linux/semaphore.h>
24 #include <linux/fiemap.h>
25 #include <linux/rculist_bl.h>
26 #include <linux/atomic.h>
27 #include <linux/shrinker.h>
28 #include <linux/migrate_mode.h>
29 #include <linux/uidgid.h>
30 #include <linux/lockdep.h>
31 #include <linux/percpu-rwsem.h>
32 #include <linux/blk_types.h>
33 #include <linux/workqueue.h>
34 #include <linux/percpu-rwsem.h>
35 #include <linux/delayed_call.h>
37 #include <asm/byteorder.h>
38 #include <uapi/linux/fs.h>
40 struct backing_dev_info;
42 struct export_operations;
47 struct pipe_inode_info;
48 struct poll_table_struct;
50 struct vm_area_struct;
53 struct swap_info_struct;
55 struct workqueue_struct;
58 struct fscrypt_operations;
60 extern void __init inode_init(void);
61 extern void __init inode_init_early(void);
62 extern void __init files_init(void);
63 extern void __init files_maxfiles_init(void);
65 extern struct files_stat_struct files_stat;
66 extern unsigned long get_max_files(void);
67 extern unsigned int sysctl_nr_open;
68 extern struct inodes_stat_t inodes_stat;
69 extern int leases_enable, lease_break_time;
70 extern int sysctl_protected_symlinks;
71 extern int sysctl_protected_hardlinks;
74 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
75 struct buffer_head *bh_result, int create);
76 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
77 ssize_t bytes, void *private);
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 flags that we support. Some of
157 * them include behavioral modifiers that send information down to the
158 * block layer and IO scheduler. They should be used along with a req_op.
161 * The block layer uses device plugging to defer IO a little bit, in
162 * the hope that we will see more IO very shortly. This increases
163 * coalescing of adjacent IO and thus reduces the number of IOs we
164 * have to send to the device. It also allows for better queuing,
165 * if the IO isn't mergeable. If the caller is going to be waiting
166 * for the IO, then he must ensure that the device is unplugged so
167 * that the IO is dispatched to the driver.
169 * All IO is handled async in Linux. This is fine for background
170 * writes, but for reads or writes that someone waits for completion
171 * on, we want to notify the block layer and IO scheduler so that they
172 * know about it. That allows them to make better scheduling
173 * decisions. So when the below references 'sync' and 'async', it
174 * is referencing this priority hint.
176 * With that in mind, the available types are:
178 * READ A normal read operation. Device will be plugged.
179 * READ_SYNC A synchronous read. Device is not plugged, caller can
180 * immediately wait on this read without caring about
182 * WRITE A normal async write. Device will be plugged.
183 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
184 * the hint that someone will be waiting on this IO
185 * shortly. The write equivalent of READ_SYNC.
186 * WRITE_ODIRECT Special case write for O_DIRECT only.
187 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
188 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
189 * non-volatile media on completion.
190 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
191 * by a cache flush and data is guaranteed to be on
192 * non-volatile media on completion.
195 #define RW_MASK REQ_OP_WRITE
197 #define READ REQ_OP_READ
198 #define WRITE REQ_OP_WRITE
200 #define READ_SYNC REQ_SYNC
201 #define WRITE_SYNC (REQ_SYNC | REQ_NOIDLE)
202 #define WRITE_ODIRECT REQ_SYNC
203 #define WRITE_FLUSH (REQ_SYNC | REQ_NOIDLE | REQ_PREFLUSH)
204 #define WRITE_FUA (REQ_SYNC | REQ_NOIDLE | REQ_FUA)
205 #define WRITE_FLUSH_FUA (REQ_SYNC | REQ_NOIDLE | REQ_PREFLUSH | REQ_FUA)
208 * Attribute flags. These should be or-ed together to figure out what
211 #define ATTR_MODE (1 << 0)
212 #define ATTR_UID (1 << 1)
213 #define ATTR_GID (1 << 2)
214 #define ATTR_SIZE (1 << 3)
215 #define ATTR_ATIME (1 << 4)
216 #define ATTR_MTIME (1 << 5)
217 #define ATTR_CTIME (1 << 6)
218 #define ATTR_ATIME_SET (1 << 7)
219 #define ATTR_MTIME_SET (1 << 8)
220 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
221 #define ATTR_ATTR_FLAG (1 << 10)
222 #define ATTR_KILL_SUID (1 << 11)
223 #define ATTR_KILL_SGID (1 << 12)
224 #define ATTR_FILE (1 << 13)
225 #define ATTR_KILL_PRIV (1 << 14)
226 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
227 #define ATTR_TIMES_SET (1 << 16)
228 #define ATTR_TOUCH (1 << 17)
231 * Whiteout is represented by a char device. The following constants define the
232 * mode and device number to use.
234 #define WHITEOUT_MODE 0
235 #define WHITEOUT_DEV 0
238 * This is the Inode Attributes structure, used for notify_change(). It
239 * uses the above definitions as flags, to know which values have changed.
240 * Also, in this manner, a Filesystem can look at only the values it cares
241 * about. Basically, these are the attributes that the VFS layer can
242 * request to change from the FS layer.
244 * Derek Atkins <warlord@MIT.EDU> 94-10-20
247 unsigned int ia_valid;
252 struct timespec ia_atime;
253 struct timespec ia_mtime;
254 struct timespec ia_ctime;
257 * Not an attribute, but an auxiliary info for filesystems wanting to
258 * implement an ftruncate() like method. NOTE: filesystem should
259 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
261 struct file *ia_file;
265 * Includes for diskquotas.
267 #include <linux/quota.h>
270 * Maximum number of layers of fs stack. Needs to be limited to
271 * prevent kernel stack overflow
273 #define FILESYSTEM_MAX_STACK_DEPTH 2
276 * enum positive_aop_returns - aop return codes with specific semantics
278 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
279 * completed, that the page is still locked, and
280 * should be considered active. The VM uses this hint
281 * to return the page to the active list -- it won't
282 * be a candidate for writeback again in the near
283 * future. Other callers must be careful to unlock
284 * the page if they get this return. Returned by
287 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
288 * unlocked it and the page might have been truncated.
289 * The caller should back up to acquiring a new page and
290 * trying again. The aop will be taking reasonable
291 * precautions not to livelock. If the caller held a page
292 * reference, it should drop it before retrying. Returned
295 * address_space_operation functions return these large constants to indicate
296 * special semantics to the caller. These are much larger than the bytes in a
297 * page to allow for functions that return the number of bytes operated on in a
301 enum positive_aop_returns {
302 AOP_WRITEPAGE_ACTIVATE = 0x80000,
303 AOP_TRUNCATED_PAGE = 0x80001,
306 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
307 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
308 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
309 * helper code (eg buffer layer)
310 * to clear GFP_FS from alloc */
313 * oh the beauties of C type declarations.
316 struct address_space;
317 struct writeback_control;
319 #define IOCB_EVENTFD (1 << 0)
320 #define IOCB_APPEND (1 << 1)
321 #define IOCB_DIRECT (1 << 2)
322 #define IOCB_HIPRI (1 << 3)
323 #define IOCB_DSYNC (1 << 4)
324 #define IOCB_SYNC (1 << 5)
325 #define IOCB_WRITE (1 << 6)
328 struct file *ki_filp;
330 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
335 static inline bool is_sync_kiocb(struct kiocb *kiocb)
337 return kiocb->ki_complete == NULL;
340 static inline int iocb_flags(struct file *file);
342 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
344 *kiocb = (struct kiocb) {
346 .ki_flags = iocb_flags(filp),
351 * "descriptor" for what we're up to with a read.
352 * This allows us to use the same read code yet
353 * have multiple different users of the data that
354 * we read from a file.
356 * The simplest case just copies the data to user
369 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
370 unsigned long, unsigned long);
372 struct address_space_operations {
373 int (*writepage)(struct page *page, struct writeback_control *wbc);
374 int (*readpage)(struct file *, struct page *);
376 /* Write back some dirty pages from this mapping. */
377 int (*writepages)(struct address_space *, struct writeback_control *);
379 /* Set a page dirty. Return true if this dirtied it */
380 int (*set_page_dirty)(struct page *page);
382 int (*readpages)(struct file *filp, struct address_space *mapping,
383 struct list_head *pages, unsigned nr_pages);
385 int (*write_begin)(struct file *, struct address_space *mapping,
386 loff_t pos, unsigned len, unsigned flags,
387 struct page **pagep, void **fsdata);
388 int (*write_end)(struct file *, struct address_space *mapping,
389 loff_t pos, unsigned len, unsigned copied,
390 struct page *page, void *fsdata);
392 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
393 sector_t (*bmap)(struct address_space *, sector_t);
394 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
395 int (*releasepage) (struct page *, gfp_t);
396 void (*freepage)(struct page *);
397 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
399 * migrate the contents of a page to the specified target. If
400 * migrate_mode is MIGRATE_ASYNC, it must not block.
402 int (*migratepage) (struct address_space *,
403 struct page *, struct page *, enum migrate_mode);
404 bool (*isolate_page)(struct page *, isolate_mode_t);
405 void (*putback_page)(struct page *);
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 */
446 spinlock_t private_lock; /* for use by the address_space */
447 gfp_t gfp_mask; /* implicit gfp mask for allocations */
448 struct list_head private_list; /* ditto */
449 void *private_data; /* ditto */
450 } __attribute__((aligned(sizeof(long))));
452 * On most architectures that alignment is already the case; but
453 * must be enforced here for CRIS, to let the least significant bit
454 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
456 struct request_queue;
458 struct block_device {
459 dev_t bd_dev; /* not a kdev_t - it's a search key */
461 struct inode * bd_inode; /* will die */
462 struct super_block * bd_super;
463 struct mutex bd_mutex; /* open/close mutex */
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))
581 #define ACL_DONT_CACHE ((void *)(-3))
583 static inline struct posix_acl *
584 uncached_acl_sentinel(struct task_struct *task)
586 return (void *)task + 1;
590 is_uncached_acl(struct posix_acl *acl)
592 return (long)acl & 1;
595 #define IOP_FASTPERM 0x0001
596 #define IOP_LOOKUP 0x0002
597 #define IOP_NOFOLLOW 0x0004
598 #define IOP_XATTR 0x0008
601 * Keep mostly read-only and often accessed (especially for
602 * the RCU path lookup and 'stat' data) fields at the beginning
603 * of the 'struct inode'
607 unsigned short i_opflags;
610 unsigned int i_flags;
612 #ifdef CONFIG_FS_POSIX_ACL
613 struct posix_acl *i_acl;
614 struct posix_acl *i_default_acl;
617 const struct inode_operations *i_op;
618 struct super_block *i_sb;
619 struct address_space *i_mapping;
621 #ifdef CONFIG_SECURITY
625 /* Stat data, not accessed from path walking */
628 * Filesystems may only read i_nlink directly. They shall use the
629 * following functions for modification:
631 * (set|clear|inc|drop)_nlink
632 * inode_(inc|dec)_link_count
635 const unsigned int i_nlink;
636 unsigned int __i_nlink;
640 struct timespec i_atime;
641 struct timespec i_mtime;
642 struct timespec i_ctime;
643 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
644 unsigned short i_bytes;
645 unsigned int i_blkbits;
648 #ifdef __NEED_I_SIZE_ORDERED
649 seqcount_t i_size_seqcount;
653 unsigned long i_state;
654 struct rw_semaphore i_rwsem;
656 unsigned long dirtied_when; /* jiffies of first dirtying */
657 unsigned long dirtied_time_when;
659 struct hlist_node i_hash;
660 struct list_head i_io_list; /* backing dev IO list */
661 #ifdef CONFIG_CGROUP_WRITEBACK
662 struct bdi_writeback *i_wb; /* the associated cgroup wb */
664 /* foreign inode detection, see wbc_detach_inode() */
666 u16 i_wb_frn_avg_time;
667 u16 i_wb_frn_history;
669 struct list_head i_lru; /* inode LRU list */
670 struct list_head i_sb_list;
671 struct list_head i_wb_list; /* backing dev writeback list */
673 struct hlist_head i_dentry;
674 struct rcu_head i_rcu;
678 atomic_t i_dio_count;
679 atomic_t i_writecount;
681 atomic_t i_readcount; /* struct files open RO */
683 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
684 struct file_lock_context *i_flctx;
685 struct address_space i_data;
686 struct list_head i_devices;
688 struct pipe_inode_info *i_pipe;
689 struct block_device *i_bdev;
697 #ifdef CONFIG_FSNOTIFY
698 __u32 i_fsnotify_mask; /* all events this inode cares about */
699 struct hlist_head i_fsnotify_marks;
702 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
703 struct fscrypt_info *i_crypt_info;
706 void *i_private; /* fs or device private pointer */
709 static inline unsigned int i_blocksize(const struct inode *node)
711 return (1 << node->i_blkbits);
714 static inline int inode_unhashed(struct inode *inode)
716 return hlist_unhashed(&inode->i_hash);
720 * inode->i_mutex nesting subclasses for the lock validator:
722 * 0: the object of the current VFS operation
726 * 4: second non-directory
727 * 5: second parent (when locking independent directories in rename)
729 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
730 * non-directories at once.
732 * The locking order between these classes is
733 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
735 enum inode_i_mutex_lock_class
745 static inline void inode_lock(struct inode *inode)
747 down_write(&inode->i_rwsem);
750 static inline void inode_unlock(struct inode *inode)
752 up_write(&inode->i_rwsem);
755 static inline void inode_lock_shared(struct inode *inode)
757 down_read(&inode->i_rwsem);
760 static inline void inode_unlock_shared(struct inode *inode)
762 up_read(&inode->i_rwsem);
765 static inline int inode_trylock(struct inode *inode)
767 return down_write_trylock(&inode->i_rwsem);
770 static inline int inode_trylock_shared(struct inode *inode)
772 return down_read_trylock(&inode->i_rwsem);
775 static inline int inode_is_locked(struct inode *inode)
777 return rwsem_is_locked(&inode->i_rwsem);
780 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
782 down_write_nested(&inode->i_rwsem, subclass);
785 void lock_two_nondirectories(struct inode *, struct inode*);
786 void unlock_two_nondirectories(struct inode *, struct inode*);
789 * NOTE: in a 32bit arch with a preemptable kernel and
790 * an UP compile the i_size_read/write must be atomic
791 * with respect to the local cpu (unlike with preempt disabled),
792 * but they don't need to be atomic with respect to other cpus like in
793 * true SMP (so they need either to either locally disable irq around
794 * the read or for example on x86 they can be still implemented as a
795 * cmpxchg8b without the need of the lock prefix). For SMP compiles
796 * and 64bit archs it makes no difference if preempt is enabled or not.
798 static inline loff_t i_size_read(const struct inode *inode)
800 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
805 seq = read_seqcount_begin(&inode->i_size_seqcount);
806 i_size = inode->i_size;
807 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
809 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
813 i_size = inode->i_size;
817 return inode->i_size;
822 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
823 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
824 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
826 static inline void i_size_write(struct inode *inode, loff_t i_size)
828 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
830 write_seqcount_begin(&inode->i_size_seqcount);
831 inode->i_size = i_size;
832 write_seqcount_end(&inode->i_size_seqcount);
834 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
836 inode->i_size = i_size;
839 inode->i_size = i_size;
843 static inline unsigned iminor(const struct inode *inode)
845 return MINOR(inode->i_rdev);
848 static inline unsigned imajor(const struct inode *inode)
850 return MAJOR(inode->i_rdev);
853 extern struct block_device *I_BDEV(struct inode *inode);
856 rwlock_t lock; /* protects pid, uid, euid fields */
857 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
858 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
859 kuid_t uid, euid; /* uid/euid of process setting the owner */
860 int signum; /* posix.1b rt signal to be delivered on IO */
864 * Track a single file's readahead state
866 struct file_ra_state {
867 pgoff_t start; /* where readahead started */
868 unsigned int size; /* # of readahead pages */
869 unsigned int async_size; /* do asynchronous readahead when
870 there are only # of pages ahead */
872 unsigned int ra_pages; /* Maximum readahead window */
873 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
874 loff_t prev_pos; /* Cache last read() position */
878 * Check if @index falls in the readahead windows.
880 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
882 return (index >= ra->start &&
883 index < ra->start + ra->size);
888 struct llist_node fu_llist;
889 struct rcu_head fu_rcuhead;
892 struct inode *f_inode; /* cached value */
893 const struct file_operations *f_op;
896 * Protects f_ep_links, f_flags.
897 * Must not be taken from IRQ context.
900 atomic_long_t f_count;
901 unsigned int f_flags;
903 struct mutex f_pos_lock;
905 struct fown_struct f_owner;
906 const struct cred *f_cred;
907 struct file_ra_state f_ra;
910 #ifdef CONFIG_SECURITY
913 /* needed for tty driver, and maybe others */
917 /* Used by fs/eventpoll.c to link all the hooks to this file */
918 struct list_head f_ep_links;
919 struct list_head f_tfile_llink;
920 #endif /* #ifdef CONFIG_EPOLL */
921 struct address_space *f_mapping;
922 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
927 /* file identifier */
928 unsigned char f_handle[0];
931 static inline struct file *get_file(struct file *f)
933 atomic_long_inc(&f->f_count);
936 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
937 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
938 #define file_count(x) atomic_long_read(&(x)->f_count)
940 #define MAX_NON_LFS ((1UL<<31) - 1)
942 /* Page cache limit. The filesystems should put that into their s_maxbytes
943 limits, otherwise bad things can happen in VM. */
944 #if BITS_PER_LONG==32
945 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
946 #elif BITS_PER_LONG==64
947 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
952 #define FL_DELEG 4 /* NFSv4 delegation */
953 #define FL_ACCESS 8 /* not trying to lock, just looking */
954 #define FL_EXISTS 16 /* when unlocking, test for existence */
955 #define FL_LEASE 32 /* lease held on this file */
956 #define FL_CLOSE 64 /* unlock on close */
957 #define FL_SLEEP 128 /* A blocking lock */
958 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
959 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
960 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
961 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
964 * Special return value from posix_lock_file() and vfs_lock_file() for
965 * asynchronous locking.
967 #define FILE_LOCK_DEFERRED 1
969 /* legacy typedef, should eventually be removed */
970 typedef void *fl_owner_t;
974 struct file_lock_operations {
975 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
976 void (*fl_release_private)(struct file_lock *);
979 struct lock_manager_operations {
980 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
981 unsigned long (*lm_owner_key)(struct file_lock *);
982 fl_owner_t (*lm_get_owner)(fl_owner_t);
983 void (*lm_put_owner)(fl_owner_t);
984 void (*lm_notify)(struct file_lock *); /* unblock callback */
985 int (*lm_grant)(struct file_lock *, int);
986 bool (*lm_break)(struct file_lock *);
987 int (*lm_change)(struct file_lock *, int, struct list_head *);
988 void (*lm_setup)(struct file_lock *, void **);
991 struct lock_manager {
992 struct list_head list;
994 * NFSv4 and up also want opens blocked during the grace period;
1001 void locks_start_grace(struct net *, struct lock_manager *);
1002 void locks_end_grace(struct lock_manager *);
1003 int locks_in_grace(struct net *);
1004 int opens_in_grace(struct net *);
1006 /* that will die - we need it for nfs_lock_info */
1007 #include <linux/nfs_fs_i.h>
1010 * struct file_lock represents a generic "file lock". It's used to represent
1011 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1012 * note that the same struct is used to represent both a request for a lock and
1013 * the lock itself, but the same object is never used for both.
1015 * FIXME: should we create a separate "struct lock_request" to help distinguish
1018 * The varous i_flctx lists are ordered by:
1021 * 2) lock range start
1024 * Obviously, the last two criteria only matter for POSIX locks.
1027 struct file_lock *fl_next; /* singly linked list for this inode */
1028 struct list_head fl_list; /* link into file_lock_context */
1029 struct hlist_node fl_link; /* node in global lists */
1030 struct list_head fl_block; /* circular list of blocked processes */
1031 fl_owner_t fl_owner;
1032 unsigned int fl_flags;
1033 unsigned char fl_type;
1034 unsigned int fl_pid;
1035 int fl_link_cpu; /* what cpu's list is this on? */
1036 struct pid *fl_nspid;
1037 wait_queue_head_t fl_wait;
1038 struct file *fl_file;
1042 struct fasync_struct * fl_fasync; /* for lease break notifications */
1043 /* for lease breaks: */
1044 unsigned long fl_break_time;
1045 unsigned long fl_downgrade_time;
1047 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1048 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1050 struct nfs_lock_info nfs_fl;
1051 struct nfs4_lock_info nfs4_fl;
1053 struct list_head link; /* link in AFS vnode's pending_locks list */
1054 int state; /* state of grant or error if -ve */
1059 struct file_lock_context {
1060 spinlock_t flc_lock;
1061 struct list_head flc_flock;
1062 struct list_head flc_posix;
1063 struct list_head flc_lease;
1066 /* The following constant reflects the upper bound of the file/locking space */
1068 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1069 #define OFFSET_MAX INT_LIMIT(loff_t)
1070 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1073 #include <linux/fcntl.h>
1075 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1078 * Return the inode to use for locking
1080 * For overlayfs this should be the overlay inode, not the real inode returned
1081 * by file_inode(). For any other fs file_inode(filp) and locks_inode(filp) are
1084 static inline struct inode *locks_inode(const struct file *f)
1086 return f->f_path.dentry->d_inode;
1089 #ifdef CONFIG_FILE_LOCKING
1090 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
1091 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1092 struct flock __user *);
1094 #if BITS_PER_LONG == 32
1095 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
1096 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1097 struct flock64 __user *);
1100 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1101 extern int fcntl_getlease(struct file *filp);
1104 void locks_free_lock_context(struct inode *inode);
1105 void locks_free_lock(struct file_lock *fl);
1106 extern void locks_init_lock(struct file_lock *);
1107 extern struct file_lock * locks_alloc_lock(void);
1108 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1109 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1110 extern void locks_remove_posix(struct file *, fl_owner_t);
1111 extern void locks_remove_file(struct file *);
1112 extern void locks_release_private(struct file_lock *);
1113 extern void posix_test_lock(struct file *, struct file_lock *);
1114 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1115 extern int posix_unblock_lock(struct file_lock *);
1116 extern int vfs_test_lock(struct file *, struct file_lock *);
1117 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1118 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1119 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1120 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1121 extern void lease_get_mtime(struct inode *, struct timespec *time);
1122 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1123 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1124 extern int lease_modify(struct file_lock *, int, struct list_head *);
1125 struct files_struct;
1126 extern void show_fd_locks(struct seq_file *f,
1127 struct file *filp, struct files_struct *files);
1128 #else /* !CONFIG_FILE_LOCKING */
1129 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1130 struct flock __user *user)
1135 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1136 unsigned int cmd, struct flock __user *user)
1141 #if BITS_PER_LONG == 32
1142 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1143 struct flock64 __user *user)
1148 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1149 unsigned int cmd, struct flock64 __user *user)
1154 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1159 static inline int fcntl_getlease(struct file *filp)
1165 locks_free_lock_context(struct inode *inode)
1169 static inline void locks_init_lock(struct file_lock *fl)
1174 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1179 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1184 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1189 static inline void locks_remove_file(struct file *filp)
1194 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1199 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1200 struct file_lock *conflock)
1205 static inline int posix_unblock_lock(struct file_lock *waiter)
1210 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1215 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1216 struct file_lock *fl, struct file_lock *conf)
1221 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1226 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1231 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1236 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1241 static inline int generic_setlease(struct file *filp, long arg,
1242 struct file_lock **flp, void **priv)
1247 static inline int vfs_setlease(struct file *filp, long arg,
1248 struct file_lock **lease, void **priv)
1253 static inline int lease_modify(struct file_lock *fl, int arg,
1254 struct list_head *dispose)
1259 struct files_struct;
1260 static inline void show_fd_locks(struct seq_file *f,
1261 struct file *filp, struct files_struct *files) {}
1262 #endif /* !CONFIG_FILE_LOCKING */
1264 static inline struct inode *file_inode(const struct file *f)
1269 static inline struct dentry *file_dentry(const struct file *file)
1271 return d_real(file->f_path.dentry, file_inode(file), 0);
1274 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1276 return locks_lock_inode_wait(locks_inode(filp), fl);
1279 struct fasync_struct {
1283 struct fasync_struct *fa_next; /* singly linked list */
1284 struct file *fa_file;
1285 struct rcu_head fa_rcu;
1288 #define FASYNC_MAGIC 0x4601
1290 /* SMP safe fasync helpers: */
1291 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1292 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1293 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1294 extern struct fasync_struct *fasync_alloc(void);
1295 extern void fasync_free(struct fasync_struct *);
1297 /* can be called from interrupts */
1298 extern void kill_fasync(struct fasync_struct **, int, int);
1300 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1301 extern void f_setown(struct file *filp, unsigned long arg, int force);
1302 extern void f_delown(struct file *filp);
1303 extern pid_t f_getown(struct file *filp);
1304 extern int send_sigurg(struct fown_struct *fown);
1312 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1313 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1314 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1315 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1316 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1319 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1320 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1321 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1322 #define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */
1324 /* sb->s_iflags to limit user namespace mounts */
1325 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1327 /* Possible states of 'frozen' field */
1329 SB_UNFROZEN = 0, /* FS is unfrozen */
1330 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1331 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1332 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1333 * internal threads if needed) */
1334 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1337 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1340 int frozen; /* Is sb frozen? */
1341 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1342 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1345 struct super_block {
1346 struct list_head s_list; /* Keep this first */
1347 dev_t s_dev; /* search index; _not_ kdev_t */
1348 unsigned char s_blocksize_bits;
1349 unsigned long s_blocksize;
1350 loff_t s_maxbytes; /* Max file size */
1351 struct file_system_type *s_type;
1352 const struct super_operations *s_op;
1353 const struct dquot_operations *dq_op;
1354 const struct quotactl_ops *s_qcop;
1355 const struct export_operations *s_export_op;
1356 unsigned long s_flags;
1357 unsigned long s_iflags; /* internal SB_I_* flags */
1358 unsigned long s_magic;
1359 struct dentry *s_root;
1360 struct rw_semaphore s_umount;
1363 #ifdef CONFIG_SECURITY
1366 const struct xattr_handler **s_xattr;
1368 const struct fscrypt_operations *s_cop;
1370 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1371 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1372 struct block_device *s_bdev;
1373 struct backing_dev_info *s_bdi;
1374 struct mtd_info *s_mtd;
1375 struct hlist_node s_instances;
1376 unsigned int s_quota_types; /* Bitmask of supported quota types */
1377 struct quota_info s_dquot; /* Diskquota specific options */
1379 struct sb_writers s_writers;
1381 char s_id[32]; /* Informational name */
1382 u8 s_uuid[16]; /* UUID */
1384 void *s_fs_info; /* Filesystem private info */
1385 unsigned int s_max_links;
1388 /* Granularity of c/m/atime in ns.
1389 Cannot be worse than a second */
1393 * The next field is for VFS *only*. No filesystems have any business
1394 * even looking at it. You had been warned.
1396 struct mutex s_vfs_rename_mutex; /* Kludge */
1399 * Filesystem subtype. If non-empty the filesystem type field
1400 * in /proc/mounts will be "type.subtype"
1405 * Saved mount options for lazy filesystems using
1406 * generic_show_options()
1408 char __rcu *s_options;
1409 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1412 * Saved pool identifier for cleancache (-1 means none)
1414 int cleancache_poolid;
1416 struct shrinker s_shrink; /* per-sb shrinker handle */
1418 /* Number of inodes with nlink == 0 but still referenced */
1419 atomic_long_t s_remove_count;
1421 /* Being remounted read-only */
1422 int s_readonly_remount;
1424 /* AIO completions deferred from interrupt context */
1425 struct workqueue_struct *s_dio_done_wq;
1426 struct hlist_head s_pins;
1429 * Owning user namespace and default context in which to
1430 * interpret filesystem uids, gids, quotas, device nodes,
1431 * xattrs and security labels.
1433 struct user_namespace *s_user_ns;
1436 * Keep the lru lists last in the structure so they always sit on their
1437 * own individual cachelines.
1439 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1440 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1441 struct rcu_head rcu;
1442 struct work_struct destroy_work;
1444 struct mutex s_sync_lock; /* sync serialisation lock */
1447 * Indicates how deep in a filesystem stack this SB is
1451 /* s_inode_list_lock protects s_inodes */
1452 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1453 struct list_head s_inodes; /* all inodes */
1455 spinlock_t s_inode_wblist_lock;
1456 struct list_head s_inodes_wb; /* writeback inodes */
1459 /* Helper functions so that in most cases filesystems will
1460 * not need to deal directly with kuid_t and kgid_t and can
1461 * instead deal with the raw numeric values that are stored
1462 * in the filesystem.
1464 static inline uid_t i_uid_read(const struct inode *inode)
1466 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1469 static inline gid_t i_gid_read(const struct inode *inode)
1471 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1474 static inline void i_uid_write(struct inode *inode, uid_t uid)
1476 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1479 static inline void i_gid_write(struct inode *inode, gid_t gid)
1481 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1484 extern struct timespec current_fs_time(struct super_block *sb);
1485 extern struct timespec current_time(struct inode *inode);
1488 * Snapshotting support.
1491 void __sb_end_write(struct super_block *sb, int level);
1492 int __sb_start_write(struct super_block *sb, int level, bool wait);
1494 #define __sb_writers_acquired(sb, lev) \
1495 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1496 #define __sb_writers_release(sb, lev) \
1497 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1500 * sb_end_write - drop write access to a superblock
1501 * @sb: the super we wrote to
1503 * Decrement number of writers to the filesystem. Wake up possible waiters
1504 * wanting to freeze the filesystem.
1506 static inline void sb_end_write(struct super_block *sb)
1508 __sb_end_write(sb, SB_FREEZE_WRITE);
1512 * sb_end_pagefault - drop write access to a superblock from a page fault
1513 * @sb: the super we wrote to
1515 * Decrement number of processes handling write page fault to the filesystem.
1516 * Wake up possible waiters wanting to freeze the filesystem.
1518 static inline void sb_end_pagefault(struct super_block *sb)
1520 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1524 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1525 * @sb: the super we wrote to
1527 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1528 * waiters wanting to freeze the filesystem.
1530 static inline void sb_end_intwrite(struct super_block *sb)
1532 __sb_end_write(sb, SB_FREEZE_FS);
1536 * sb_start_write - get write access to a superblock
1537 * @sb: the super we write to
1539 * When a process wants to write data or metadata to a file system (i.e. dirty
1540 * a page or an inode), it should embed the operation in a sb_start_write() -
1541 * sb_end_write() pair to get exclusion against file system freezing. This
1542 * function increments number of writers preventing freezing. If the file
1543 * system is already frozen, the function waits until the file system is
1546 * Since freeze protection behaves as a lock, users have to preserve
1547 * ordering of freeze protection and other filesystem locks. Generally,
1548 * freeze protection should be the outermost lock. In particular, we have:
1551 * -> i_mutex (write path, truncate, directory ops, ...)
1552 * -> s_umount (freeze_super, thaw_super)
1554 static inline void sb_start_write(struct super_block *sb)
1556 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1559 static inline int sb_start_write_trylock(struct super_block *sb)
1561 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1565 * sb_start_pagefault - get write access to a superblock from a page fault
1566 * @sb: the super we write to
1568 * When a process starts handling write page fault, it should embed the
1569 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1570 * exclusion against file system freezing. This is needed since the page fault
1571 * is going to dirty a page. This function increments number of running page
1572 * faults preventing freezing. If the file system is already frozen, the
1573 * function waits until the file system is thawed.
1575 * Since page fault freeze protection behaves as a lock, users have to preserve
1576 * ordering of freeze protection and other filesystem locks. It is advised to
1577 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1578 * handling code implies lock dependency:
1581 * -> sb_start_pagefault
1583 static inline void sb_start_pagefault(struct super_block *sb)
1585 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1589 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1590 * @sb: the super we write to
1592 * This is the third level of protection against filesystem freezing. It is
1593 * free for use by a filesystem. The only requirement is that it must rank
1594 * below sb_start_pagefault.
1596 * For example filesystem can call sb_start_intwrite() when starting a
1597 * transaction which somewhat eases handling of freezing for internal sources
1598 * of filesystem changes (internal fs threads, discarding preallocation on file
1601 static inline void sb_start_intwrite(struct super_block *sb)
1603 __sb_start_write(sb, SB_FREEZE_FS, true);
1607 extern bool inode_owner_or_capable(const struct inode *inode);
1610 * VFS helper functions..
1612 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1613 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1614 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1615 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1616 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1617 extern int vfs_rmdir(struct inode *, struct dentry *);
1618 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1619 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1620 extern int vfs_whiteout(struct inode *, struct dentry *);
1623 * VFS file helper functions.
1625 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1627 extern bool may_open_dev(const struct path *path);
1629 * VFS FS_IOC_FIEMAP helper definitions.
1631 struct fiemap_extent_info {
1632 unsigned int fi_flags; /* Flags as passed from user */
1633 unsigned int fi_extents_mapped; /* Number of mapped extents */
1634 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1635 struct fiemap_extent __user *fi_extents_start; /* Start of
1636 fiemap_extent array */
1638 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1639 u64 phys, u64 len, u32 flags);
1640 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1645 * NOTE! These match bits 12..15 of stat.st_mode
1646 * (ie "(i_mode >> 12) & 15").
1648 #define DT_UNKNOWN 0
1659 * This is the "filldir" function type, used by readdir() to let
1660 * the kernel specify what kind of dirent layout it wants to have.
1661 * This allows the kernel to read directories into kernel space or
1662 * to have different dirent layouts depending on the binary type.
1665 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1668 struct dir_context {
1669 const filldir_t actor;
1673 struct block_device_operations;
1675 /* These macros are for out of kernel modules to test that
1676 * the kernel supports the unlocked_ioctl and compat_ioctl
1677 * fields in struct file_operations. */
1678 #define HAVE_COMPAT_IOCTL 1
1679 #define HAVE_UNLOCKED_IOCTL 1
1682 * These flags let !MMU mmap() govern direct device mapping vs immediate
1683 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1685 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1686 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1687 * NOMMU_MAP_READ: Can be mapped for reading
1688 * NOMMU_MAP_WRITE: Can be mapped for writing
1689 * NOMMU_MAP_EXEC: Can be mapped for execution
1691 #define NOMMU_MAP_COPY 0x00000001
1692 #define NOMMU_MAP_DIRECT 0x00000008
1693 #define NOMMU_MAP_READ VM_MAYREAD
1694 #define NOMMU_MAP_WRITE VM_MAYWRITE
1695 #define NOMMU_MAP_EXEC VM_MAYEXEC
1697 #define NOMMU_VMFLAGS \
1698 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1703 struct file_operations {
1704 struct module *owner;
1705 loff_t (*llseek) (struct file *, loff_t, int);
1706 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1707 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1708 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1709 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1710 int (*iterate) (struct file *, struct dir_context *);
1711 int (*iterate_shared) (struct file *, struct dir_context *);
1712 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1713 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1714 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1715 int (*mmap) (struct file *, struct vm_area_struct *);
1716 int (*open) (struct inode *, struct file *);
1717 int (*flush) (struct file *, fl_owner_t id);
1718 int (*release) (struct inode *, struct file *);
1719 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1720 int (*fasync) (int, struct file *, int);
1721 int (*lock) (struct file *, int, struct file_lock *);
1722 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1723 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1724 int (*check_flags)(int);
1725 int (*flock) (struct file *, int, struct file_lock *);
1726 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1727 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1728 int (*setlease)(struct file *, long, struct file_lock **, void **);
1729 long (*fallocate)(struct file *file, int mode, loff_t offset,
1731 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1733 unsigned (*mmap_capabilities)(struct file *);
1735 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1736 loff_t, size_t, unsigned int);
1737 int (*clone_file_range)(struct file *, loff_t, struct file *, loff_t,
1739 ssize_t (*dedupe_file_range)(struct file *, u64, u64, struct file *,
1743 struct inode_operations {
1744 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1745 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1746 int (*permission) (struct inode *, int);
1747 struct posix_acl * (*get_acl)(struct inode *, int);
1749 int (*readlink) (struct dentry *, char __user *,int);
1751 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1752 int (*link) (struct dentry *,struct inode *,struct dentry *);
1753 int (*unlink) (struct inode *,struct dentry *);
1754 int (*symlink) (struct inode *,struct dentry *,const char *);
1755 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1756 int (*rmdir) (struct inode *,struct dentry *);
1757 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1758 int (*rename) (struct inode *, struct dentry *,
1759 struct inode *, struct dentry *, unsigned int);
1760 int (*setattr) (struct dentry *, struct iattr *);
1761 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1762 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1763 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1765 int (*update_time)(struct inode *, struct timespec *, int);
1766 int (*atomic_open)(struct inode *, struct dentry *,
1767 struct file *, unsigned open_flag,
1768 umode_t create_mode, int *opened);
1769 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1770 int (*set_acl)(struct inode *, struct posix_acl *, int);
1771 } ____cacheline_aligned;
1773 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1774 unsigned long nr_segs, unsigned long fast_segs,
1775 struct iovec *fast_pointer,
1776 struct iovec **ret_pointer);
1778 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1779 extern ssize_t __vfs_write(struct file *, const char __user *, size_t, loff_t *);
1780 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1781 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1782 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1783 unsigned long, loff_t *, int);
1784 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1785 unsigned long, loff_t *, int);
1786 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1787 loff_t, size_t, unsigned int);
1788 extern int vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1789 struct file *file_out, loff_t pos_out, u64 len);
1790 extern int vfs_dedupe_file_range(struct file *file,
1791 struct file_dedupe_range *same);
1793 struct super_operations {
1794 struct inode *(*alloc_inode)(struct super_block *sb);
1795 void (*destroy_inode)(struct inode *);
1797 void (*dirty_inode) (struct inode *, int flags);
1798 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1799 int (*drop_inode) (struct inode *);
1800 void (*evict_inode) (struct inode *);
1801 void (*put_super) (struct super_block *);
1802 int (*sync_fs)(struct super_block *sb, int wait);
1803 int (*freeze_super) (struct super_block *);
1804 int (*freeze_fs) (struct super_block *);
1805 int (*thaw_super) (struct super_block *);
1806 int (*unfreeze_fs) (struct super_block *);
1807 int (*statfs) (struct dentry *, struct kstatfs *);
1808 int (*remount_fs) (struct super_block *, int *, char *);
1809 void (*umount_begin) (struct super_block *);
1811 int (*show_options)(struct seq_file *, struct dentry *);
1812 int (*show_devname)(struct seq_file *, struct dentry *);
1813 int (*show_path)(struct seq_file *, struct dentry *);
1814 int (*show_stats)(struct seq_file *, struct dentry *);
1816 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1817 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1818 struct dquot **(*get_dquots)(struct inode *);
1820 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1821 long (*nr_cached_objects)(struct super_block *,
1822 struct shrink_control *);
1823 long (*free_cached_objects)(struct super_block *,
1824 struct shrink_control *);
1828 * Inode flags - they have no relation to superblock flags now
1830 #define S_SYNC 1 /* Writes are synced at once */
1831 #define S_NOATIME 2 /* Do not update access times */
1832 #define S_APPEND 4 /* Append-only file */
1833 #define S_IMMUTABLE 8 /* Immutable file */
1834 #define S_DEAD 16 /* removed, but still open directory */
1835 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1836 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1837 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1838 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1839 #define S_PRIVATE 512 /* Inode is fs-internal */
1840 #define S_IMA 1024 /* Inode has an associated IMA struct */
1841 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1842 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1843 #ifdef CONFIG_FS_DAX
1844 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1846 #define S_DAX 0 /* Make all the DAX code disappear */
1850 * Note that nosuid etc flags are inode-specific: setting some file-system
1851 * flags just means all the inodes inherit those flags by default. It might be
1852 * possible to override it selectively if you really wanted to with some
1853 * ioctl() that is not currently implemented.
1855 * Exception: MS_RDONLY is always applied to the entire file system.
1857 * Unfortunately, it is possible to change a filesystems flags with it mounted
1858 * with files in use. This means that all of the inodes will not have their
1859 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1860 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1862 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1864 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1865 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1866 ((inode)->i_flags & S_SYNC))
1867 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1868 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1869 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1870 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1871 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1873 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1874 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1875 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1876 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1878 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1879 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1880 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1881 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1882 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1883 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1884 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1885 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1887 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1888 (inode)->i_rdev == WHITEOUT_DEV)
1890 static inline bool HAS_UNMAPPED_ID(struct inode *inode)
1892 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
1896 * Inode state bits. Protected by inode->i_lock
1898 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1899 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1901 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1902 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1903 * various stages of removing an inode.
1905 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1907 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1908 * fdatasync(). i_atime is the usual cause.
1909 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1910 * these changes separately from I_DIRTY_SYNC so that we
1911 * don't have to write inode on fdatasync() when only
1912 * mtime has changed in it.
1913 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1914 * I_NEW Serves as both a mutex and completion notification.
1915 * New inodes set I_NEW. If two processes both create
1916 * the same inode, one of them will release its inode and
1917 * wait for I_NEW to be released before returning.
1918 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1919 * also cause waiting on I_NEW, without I_NEW actually
1920 * being set. find_inode() uses this to prevent returning
1921 * nearly-dead inodes.
1922 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1923 * is zero. I_FREEING must be set when I_WILL_FREE is
1925 * I_FREEING Set when inode is about to be freed but still has dirty
1926 * pages or buffers attached or the inode itself is still
1928 * I_CLEAR Added by clear_inode(). In this state the inode is
1929 * clean and can be destroyed. Inode keeps I_FREEING.
1931 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1932 * prohibited for many purposes. iget() must wait for
1933 * the inode to be completely released, then create it
1934 * anew. Other functions will just ignore such inodes,
1935 * if appropriate. I_NEW is used for waiting.
1937 * I_SYNC Writeback of inode is running. The bit is set during
1938 * data writeback, and cleared with a wakeup on the bit
1939 * address once it is done. The bit is also used to pin
1940 * the inode in memory for flusher thread.
1942 * I_REFERENCED Marks the inode as recently references on the LRU list.
1944 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1946 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1947 * synchronize competing switching instances and to tell
1948 * wb stat updates to grab mapping->tree_lock. See
1949 * inode_switch_wb_work_fn() for details.
1951 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1953 #define I_DIRTY_SYNC (1 << 0)
1954 #define I_DIRTY_DATASYNC (1 << 1)
1955 #define I_DIRTY_PAGES (1 << 2)
1957 #define I_NEW (1 << __I_NEW)
1958 #define I_WILL_FREE (1 << 4)
1959 #define I_FREEING (1 << 5)
1960 #define I_CLEAR (1 << 6)
1962 #define I_SYNC (1 << __I_SYNC)
1963 #define I_REFERENCED (1 << 8)
1964 #define __I_DIO_WAKEUP 9
1965 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1966 #define I_LINKABLE (1 << 10)
1967 #define I_DIRTY_TIME (1 << 11)
1968 #define __I_DIRTY_TIME_EXPIRED 12
1969 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1970 #define I_WB_SWITCH (1 << 13)
1972 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1973 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1975 extern void __mark_inode_dirty(struct inode *, int);
1976 static inline void mark_inode_dirty(struct inode *inode)
1978 __mark_inode_dirty(inode, I_DIRTY);
1981 static inline void mark_inode_dirty_sync(struct inode *inode)
1983 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1986 extern void inc_nlink(struct inode *inode);
1987 extern void drop_nlink(struct inode *inode);
1988 extern void clear_nlink(struct inode *inode);
1989 extern void set_nlink(struct inode *inode, unsigned int nlink);
1991 static inline void inode_inc_link_count(struct inode *inode)
1994 mark_inode_dirty(inode);
1997 static inline void inode_dec_link_count(struct inode *inode)
2000 mark_inode_dirty(inode);
2004 * inode_inc_iversion - increments i_version
2005 * @inode: inode that need to be updated
2007 * Every time the inode is modified, the i_version field will be incremented.
2008 * The filesystem has to be mounted with i_version flag
2011 static inline void inode_inc_iversion(struct inode *inode)
2013 spin_lock(&inode->i_lock);
2015 spin_unlock(&inode->i_lock);
2018 enum file_time_flags {
2025 extern void touch_atime(const struct path *);
2026 static inline void file_accessed(struct file *file)
2028 if (!(file->f_flags & O_NOATIME))
2029 touch_atime(&file->f_path);
2032 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2033 int sync_inode_metadata(struct inode *inode, int wait);
2035 struct file_system_type {
2038 #define FS_REQUIRES_DEV 1
2039 #define FS_BINARY_MOUNTDATA 2
2040 #define FS_HAS_SUBTYPE 4
2041 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2042 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2043 struct dentry *(*mount) (struct file_system_type *, int,
2044 const char *, void *);
2045 void (*kill_sb) (struct super_block *);
2046 struct module *owner;
2047 struct file_system_type * next;
2048 struct hlist_head fs_supers;
2050 struct lock_class_key s_lock_key;
2051 struct lock_class_key s_umount_key;
2052 struct lock_class_key s_vfs_rename_key;
2053 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2055 struct lock_class_key i_lock_key;
2056 struct lock_class_key i_mutex_key;
2057 struct lock_class_key i_mutex_dir_key;
2060 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2062 extern struct dentry *mount_ns(struct file_system_type *fs_type,
2063 int flags, void *data, void *ns, struct user_namespace *user_ns,
2064 int (*fill_super)(struct super_block *, void *, int));
2065 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2066 int flags, const char *dev_name, void *data,
2067 int (*fill_super)(struct super_block *, void *, int));
2068 extern struct dentry *mount_single(struct file_system_type *fs_type,
2069 int flags, void *data,
2070 int (*fill_super)(struct super_block *, void *, int));
2071 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2072 int flags, void *data,
2073 int (*fill_super)(struct super_block *, void *, int));
2074 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2075 void generic_shutdown_super(struct super_block *sb);
2076 void kill_block_super(struct super_block *sb);
2077 void kill_anon_super(struct super_block *sb);
2078 void kill_litter_super(struct super_block *sb);
2079 void deactivate_super(struct super_block *sb);
2080 void deactivate_locked_super(struct super_block *sb);
2081 int set_anon_super(struct super_block *s, void *data);
2082 int get_anon_bdev(dev_t *);
2083 void free_anon_bdev(dev_t);
2084 struct super_block *sget_userns(struct file_system_type *type,
2085 int (*test)(struct super_block *,void *),
2086 int (*set)(struct super_block *,void *),
2087 int flags, struct user_namespace *user_ns,
2089 struct super_block *sget(struct file_system_type *type,
2090 int (*test)(struct super_block *,void *),
2091 int (*set)(struct super_block *,void *),
2092 int flags, void *data);
2093 extern struct dentry *mount_pseudo_xattr(struct file_system_type *, char *,
2094 const struct super_operations *ops,
2095 const struct xattr_handler **xattr,
2096 const struct dentry_operations *dops,
2099 static inline struct dentry *
2100 mount_pseudo(struct file_system_type *fs_type, char *name,
2101 const struct super_operations *ops,
2102 const struct dentry_operations *dops, unsigned long magic)
2104 return mount_pseudo_xattr(fs_type, name, ops, NULL, dops, magic);
2107 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2108 #define fops_get(fops) \
2109 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2110 #define fops_put(fops) \
2111 do { if (fops) module_put((fops)->owner); } while(0)
2113 * This one is to be used *ONLY* from ->open() instances.
2114 * fops must be non-NULL, pinned down *and* module dependencies
2115 * should be sufficient to pin the caller down as well.
2117 #define replace_fops(f, fops) \
2119 struct file *__file = (f); \
2120 fops_put(__file->f_op); \
2121 BUG_ON(!(__file->f_op = (fops))); \
2124 extern int register_filesystem(struct file_system_type *);
2125 extern int unregister_filesystem(struct file_system_type *);
2126 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2127 #define kern_mount(type) kern_mount_data(type, NULL)
2128 extern void kern_unmount(struct vfsmount *mnt);
2129 extern int may_umount_tree(struct vfsmount *);
2130 extern int may_umount(struct vfsmount *);
2131 extern long do_mount(const char *, const char __user *,
2132 const char *, unsigned long, void *);
2133 extern struct vfsmount *collect_mounts(struct path *);
2134 extern void drop_collected_mounts(struct vfsmount *);
2135 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2137 extern int vfs_statfs(struct path *, struct kstatfs *);
2138 extern int user_statfs(const char __user *, struct kstatfs *);
2139 extern int fd_statfs(int, struct kstatfs *);
2140 extern int vfs_ustat(dev_t, struct kstatfs *);
2141 extern int freeze_super(struct super_block *super);
2142 extern int thaw_super(struct super_block *super);
2143 extern bool our_mnt(struct vfsmount *mnt);
2145 extern int current_umask(void);
2147 extern void ihold(struct inode * inode);
2148 extern void iput(struct inode *);
2149 extern int generic_update_time(struct inode *, struct timespec *, int);
2152 extern struct kobject *fs_kobj;
2154 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2156 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2157 extern int locks_mandatory_locked(struct file *);
2158 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2161 * Candidates for mandatory locking have the setgid bit set
2162 * but no group execute bit - an otherwise meaningless combination.
2165 static inline int __mandatory_lock(struct inode *ino)
2167 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2171 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2172 * otherwise these will be advisory locks
2175 static inline int mandatory_lock(struct inode *ino)
2177 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2180 static inline int locks_verify_locked(struct file *file)
2182 if (mandatory_lock(locks_inode(file)))
2183 return locks_mandatory_locked(file);
2187 static inline int locks_verify_truncate(struct inode *inode,
2191 if (!inode->i_flctx || !mandatory_lock(inode))
2194 if (size < inode->i_size) {
2195 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2198 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2203 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2205 static inline int locks_mandatory_locked(struct file *file)
2210 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2211 loff_t start, loff_t end, unsigned char type)
2216 static inline int __mandatory_lock(struct inode *inode)
2221 static inline int mandatory_lock(struct inode *inode)
2226 static inline int locks_verify_locked(struct file *file)
2231 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2237 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2240 #ifdef CONFIG_FILE_LOCKING
2241 static inline int break_lease(struct inode *inode, unsigned int mode)
2244 * Since this check is lockless, we must ensure that any refcounts
2245 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2246 * could end up racing with tasks trying to set a new lease on this
2250 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2251 return __break_lease(inode, mode, FL_LEASE);
2255 static inline int break_deleg(struct inode *inode, unsigned int mode)
2258 * Since this check is lockless, we must ensure that any refcounts
2259 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2260 * could end up racing with tasks trying to set a new lease on this
2264 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2265 return __break_lease(inode, mode, FL_DELEG);
2269 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2273 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2274 if (ret == -EWOULDBLOCK && delegated_inode) {
2275 *delegated_inode = inode;
2281 static inline int break_deleg_wait(struct inode **delegated_inode)
2285 ret = break_deleg(*delegated_inode, O_WRONLY);
2286 iput(*delegated_inode);
2287 *delegated_inode = NULL;
2291 static inline int break_layout(struct inode *inode, bool wait)
2294 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2295 return __break_lease(inode,
2296 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2301 #else /* !CONFIG_FILE_LOCKING */
2302 static inline int break_lease(struct inode *inode, unsigned int mode)
2307 static inline int break_deleg(struct inode *inode, unsigned int mode)
2312 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2317 static inline int break_deleg_wait(struct inode **delegated_inode)
2323 static inline int break_layout(struct inode *inode, bool wait)
2328 #endif /* CONFIG_FILE_LOCKING */
2333 const char *name; /* pointer to actual string */
2334 const __user char *uptr; /* original userland pointer */
2335 struct audit_names *aname;
2340 extern long vfs_truncate(const struct path *, loff_t);
2341 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2343 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2345 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2347 extern struct file *file_open_name(struct filename *, int, umode_t);
2348 extern struct file *filp_open(const char *, int, umode_t);
2349 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2350 const char *, int, umode_t);
2351 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2352 extern int filp_close(struct file *, fl_owner_t id);
2354 extern struct filename *getname_flags(const char __user *, int, int *);
2355 extern struct filename *getname(const char __user *);
2356 extern struct filename *getname_kernel(const char *);
2357 extern void putname(struct filename *name);
2363 extern int finish_open(struct file *file, struct dentry *dentry,
2364 int (*open)(struct inode *, struct file *),
2366 extern int finish_no_open(struct file *file, struct dentry *dentry);
2370 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2373 extern void __init vfs_caches_init_early(void);
2374 extern void __init vfs_caches_init(void);
2376 extern struct kmem_cache *names_cachep;
2378 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2379 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2382 extern int register_blkdev(unsigned int, const char *);
2383 extern void unregister_blkdev(unsigned int, const char *);
2384 extern struct block_device *bdget(dev_t);
2385 extern struct block_device *bdgrab(struct block_device *bdev);
2386 extern void bd_set_size(struct block_device *, loff_t size);
2387 extern void bd_forget(struct inode *inode);
2388 extern void bdput(struct block_device *);
2389 extern void invalidate_bdev(struct block_device *);
2390 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2391 extern int sync_blockdev(struct block_device *bdev);
2392 extern void kill_bdev(struct block_device *);
2393 extern struct super_block *freeze_bdev(struct block_device *);
2394 extern void emergency_thaw_all(void);
2395 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2396 extern int fsync_bdev(struct block_device *);
2398 extern struct super_block *blockdev_superblock;
2400 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2402 return sb == blockdev_superblock;
2405 static inline void bd_forget(struct inode *inode) {}
2406 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2407 static inline void kill_bdev(struct block_device *bdev) {}
2408 static inline void invalidate_bdev(struct block_device *bdev) {}
2410 static inline struct super_block *freeze_bdev(struct block_device *sb)
2415 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2420 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2424 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2429 extern int sync_filesystem(struct super_block *);
2430 extern const struct file_operations def_blk_fops;
2431 extern const struct file_operations def_chr_fops;
2433 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2434 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2435 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2436 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2437 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2439 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2441 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2442 extern int __blkdev_reread_part(struct block_device *bdev);
2443 extern int blkdev_reread_part(struct block_device *bdev);
2446 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2447 extern void bd_unlink_disk_holder(struct block_device *bdev,
2448 struct gendisk *disk);
2450 static inline int bd_link_disk_holder(struct block_device *bdev,
2451 struct gendisk *disk)
2455 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2456 struct gendisk *disk)
2463 #define CHRDEV_MAJOR_HASH_SIZE 255
2464 /* Marks the bottom of the first segment of free char majors */
2465 #define CHRDEV_MAJOR_DYN_END 234
2466 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2467 extern int register_chrdev_region(dev_t, unsigned, const char *);
2468 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2469 unsigned int count, const char *name,
2470 const struct file_operations *fops);
2471 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2472 unsigned int count, const char *name);
2473 extern void unregister_chrdev_region(dev_t, unsigned);
2474 extern void chrdev_show(struct seq_file *,off_t);
2476 static inline int register_chrdev(unsigned int major, const char *name,
2477 const struct file_operations *fops)
2479 return __register_chrdev(major, 0, 256, name, fops);
2482 static inline void unregister_chrdev(unsigned int major, const char *name)
2484 __unregister_chrdev(major, 0, 256, name);
2487 /* fs/block_dev.c */
2488 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2489 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2492 #define BLKDEV_MAJOR_HASH_SIZE 255
2493 extern const char *__bdevname(dev_t, char *buffer);
2494 extern const char *bdevname(struct block_device *bdev, char *buffer);
2495 extern struct block_device *lookup_bdev(const char *);
2496 extern void blkdev_show(struct seq_file *,off_t);
2499 #define BLKDEV_MAJOR_HASH_SIZE 0
2502 extern void init_special_inode(struct inode *, umode_t, dev_t);
2504 /* Invalid inode operations -- fs/bad_inode.c */
2505 extern void make_bad_inode(struct inode *);
2506 extern bool is_bad_inode(struct inode *);
2509 static inline bool op_is_write(unsigned int op)
2511 return op == REQ_OP_READ ? false : true;
2515 * return data direction, READ or WRITE
2517 static inline int bio_data_dir(struct bio *bio)
2519 return op_is_write(bio_op(bio)) ? WRITE : READ;
2522 extern void check_disk_size_change(struct gendisk *disk,
2523 struct block_device *bdev);
2524 extern int revalidate_disk(struct gendisk *);
2525 extern int check_disk_change(struct block_device *);
2526 extern int __invalidate_device(struct block_device *, bool);
2527 extern int invalidate_partition(struct gendisk *, int);
2529 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2530 pgoff_t start, pgoff_t end);
2532 static inline void invalidate_remote_inode(struct inode *inode)
2534 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2535 S_ISLNK(inode->i_mode))
2536 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2538 extern int invalidate_inode_pages2(struct address_space *mapping);
2539 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2540 pgoff_t start, pgoff_t end);
2541 extern int write_inode_now(struct inode *, int);
2542 extern int filemap_fdatawrite(struct address_space *);
2543 extern int filemap_flush(struct address_space *);
2544 extern int filemap_fdatawait(struct address_space *);
2545 extern void filemap_fdatawait_keep_errors(struct address_space *);
2546 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2548 extern int filemap_write_and_wait(struct address_space *mapping);
2549 extern int filemap_write_and_wait_range(struct address_space *mapping,
2550 loff_t lstart, loff_t lend);
2551 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2552 loff_t start, loff_t end, int sync_mode);
2553 extern int filemap_fdatawrite_range(struct address_space *mapping,
2554 loff_t start, loff_t end);
2555 extern int filemap_check_errors(struct address_space *mapping);
2557 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2559 extern int vfs_fsync(struct file *file, int datasync);
2562 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2563 * to already be updated for the write, and will return either the amount
2564 * of bytes passed in, or an error if syncing the file failed.
2566 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2568 if (iocb->ki_flags & IOCB_DSYNC) {
2569 int ret = vfs_fsync_range(iocb->ki_filp,
2570 iocb->ki_pos - count, iocb->ki_pos - 1,
2571 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2579 extern void emergency_sync(void);
2580 extern void emergency_remount(void);
2582 extern sector_t bmap(struct inode *, sector_t);
2584 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2585 extern int inode_permission(struct inode *, int);
2586 extern int __inode_permission(struct inode *, int);
2587 extern int generic_permission(struct inode *, int);
2588 extern int __check_sticky(struct inode *dir, struct inode *inode);
2590 static inline bool execute_ok(struct inode *inode)
2592 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2595 static inline void file_start_write(struct file *file)
2597 if (!S_ISREG(file_inode(file)->i_mode))
2599 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2602 static inline bool file_start_write_trylock(struct file *file)
2604 if (!S_ISREG(file_inode(file)->i_mode))
2606 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2609 static inline void file_end_write(struct file *file)
2611 if (!S_ISREG(file_inode(file)->i_mode))
2613 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2617 * get_write_access() gets write permission for a file.
2618 * put_write_access() releases this write permission.
2619 * This is used for regular files.
2620 * We cannot support write (and maybe mmap read-write shared) accesses and
2621 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2622 * can have the following values:
2623 * 0: no writers, no VM_DENYWRITE mappings
2624 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2625 * > 0: (i_writecount) users are writing to the file.
2627 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2628 * except for the cases where we don't hold i_writecount yet. Then we need to
2629 * use {get,deny}_write_access() - these functions check the sign and refuse
2630 * to do the change if sign is wrong.
2632 static inline int get_write_access(struct inode *inode)
2634 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2636 static inline int deny_write_access(struct file *file)
2638 struct inode *inode = file_inode(file);
2639 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2641 static inline void put_write_access(struct inode * inode)
2643 atomic_dec(&inode->i_writecount);
2645 static inline void allow_write_access(struct file *file)
2648 atomic_inc(&file_inode(file)->i_writecount);
2650 static inline bool inode_is_open_for_write(const struct inode *inode)
2652 return atomic_read(&inode->i_writecount) > 0;
2656 static inline void i_readcount_dec(struct inode *inode)
2658 BUG_ON(!atomic_read(&inode->i_readcount));
2659 atomic_dec(&inode->i_readcount);
2661 static inline void i_readcount_inc(struct inode *inode)
2663 atomic_inc(&inode->i_readcount);
2666 static inline void i_readcount_dec(struct inode *inode)
2670 static inline void i_readcount_inc(struct inode *inode)
2675 extern int do_pipe_flags(int *, int);
2677 #define __kernel_read_file_id(id) \
2678 id(UNKNOWN, unknown) \
2679 id(FIRMWARE, firmware) \
2680 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2681 id(MODULE, kernel-module) \
2682 id(KEXEC_IMAGE, kexec-image) \
2683 id(KEXEC_INITRAMFS, kexec-initramfs) \
2684 id(POLICY, security-policy) \
2687 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2688 #define __fid_stringify(dummy, str) #str,
2690 enum kernel_read_file_id {
2691 __kernel_read_file_id(__fid_enumify)
2694 static const char * const kernel_read_file_str[] = {
2695 __kernel_read_file_id(__fid_stringify)
2698 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
2700 if (id < 0 || id >= READING_MAX_ID)
2701 return kernel_read_file_str[READING_UNKNOWN];
2703 return kernel_read_file_str[id];
2706 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2707 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2708 enum kernel_read_file_id);
2709 extern int kernel_read_file_from_path(char *, void **, loff_t *, loff_t,
2710 enum kernel_read_file_id);
2711 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2712 enum kernel_read_file_id);
2713 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2714 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
2715 extern struct file * open_exec(const char *);
2717 /* fs/dcache.c -- generic fs support functions */
2718 extern bool is_subdir(struct dentry *, struct dentry *);
2719 extern bool path_is_under(struct path *, struct path *);
2721 extern char *file_path(struct file *, char *, int);
2723 #include <linux/err.h>
2725 /* needed for stackable file system support */
2726 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2728 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2730 extern int inode_init_always(struct super_block *, struct inode *);
2731 extern void inode_init_once(struct inode *);
2732 extern void address_space_init_once(struct address_space *mapping);
2733 extern struct inode * igrab(struct inode *);
2734 extern ino_t iunique(struct super_block *, ino_t);
2735 extern int inode_needs_sync(struct inode *inode);
2736 extern int generic_delete_inode(struct inode *inode);
2737 static inline int generic_drop_inode(struct inode *inode)
2739 return !inode->i_nlink || inode_unhashed(inode);
2742 extern struct inode *ilookup5_nowait(struct super_block *sb,
2743 unsigned long hashval, int (*test)(struct inode *, void *),
2745 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2746 int (*test)(struct inode *, void *), void *data);
2747 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2749 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2750 extern struct inode * iget_locked(struct super_block *, unsigned long);
2751 extern struct inode *find_inode_nowait(struct super_block *,
2753 int (*match)(struct inode *,
2754 unsigned long, void *),
2756 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2757 extern int insert_inode_locked(struct inode *);
2758 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2759 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2761 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2763 extern void unlock_new_inode(struct inode *);
2764 extern unsigned int get_next_ino(void);
2765 extern void evict_inodes(struct super_block *sb);
2767 extern void __iget(struct inode * inode);
2768 extern void iget_failed(struct inode *);
2769 extern void clear_inode(struct inode *);
2770 extern void __destroy_inode(struct inode *);
2771 extern struct inode *new_inode_pseudo(struct super_block *sb);
2772 extern struct inode *new_inode(struct super_block *sb);
2773 extern void free_inode_nonrcu(struct inode *inode);
2774 extern int should_remove_suid(struct dentry *);
2775 extern int file_remove_privs(struct file *);
2777 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2778 static inline void insert_inode_hash(struct inode *inode)
2780 __insert_inode_hash(inode, inode->i_ino);
2783 extern void __remove_inode_hash(struct inode *);
2784 static inline void remove_inode_hash(struct inode *inode)
2786 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2787 __remove_inode_hash(inode);
2790 extern void inode_sb_list_add(struct inode *inode);
2793 extern blk_qc_t submit_bio(struct bio *);
2794 extern int bdev_read_only(struct block_device *);
2796 extern int set_blocksize(struct block_device *, int);
2797 extern int sb_set_blocksize(struct super_block *, int);
2798 extern int sb_min_blocksize(struct super_block *, int);
2800 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2801 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2802 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2803 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2804 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2805 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2806 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
2807 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2809 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos);
2810 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos);
2812 /* fs/block_dev.c */
2813 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2814 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2815 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2817 extern void block_sync_page(struct page *page);
2820 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2821 struct pipe_inode_info *, size_t, unsigned int);
2822 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2823 struct file *, loff_t *, size_t, unsigned int);
2824 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2825 struct file *out, loff_t *, size_t len, unsigned int flags);
2826 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2827 loff_t *opos, size_t len, unsigned int flags);
2831 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2832 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2833 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2834 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2835 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2836 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2837 int whence, loff_t maxsize, loff_t eof);
2838 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2839 int whence, loff_t size);
2840 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
2841 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
2842 extern int generic_file_open(struct inode * inode, struct file * filp);
2843 extern int nonseekable_open(struct inode * inode, struct file * filp);
2846 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
2847 loff_t file_offset);
2850 /* need locking between buffered and direct access */
2853 /* filesystem does not support filling holes */
2854 DIO_SKIP_HOLES = 0x02,
2856 /* filesystem can handle aio writes beyond i_size */
2857 DIO_ASYNC_EXTEND = 0x04,
2859 /* inode/fs/bdev does not need truncate protection */
2860 DIO_SKIP_DIO_COUNT = 0x08,
2863 void dio_end_io(struct bio *bio, int error);
2865 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2866 struct block_device *bdev, struct iov_iter *iter,
2867 get_block_t get_block,
2868 dio_iodone_t end_io, dio_submit_t submit_io,
2871 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2872 struct inode *inode,
2873 struct iov_iter *iter,
2874 get_block_t get_block)
2876 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
2877 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
2881 void inode_dio_wait(struct inode *inode);
2884 * inode_dio_begin - signal start of a direct I/O requests
2885 * @inode: inode the direct I/O happens on
2887 * This is called once we've finished processing a direct I/O request,
2888 * and is used to wake up callers waiting for direct I/O to be quiesced.
2890 static inline void inode_dio_begin(struct inode *inode)
2892 atomic_inc(&inode->i_dio_count);
2896 * inode_dio_end - signal finish of a direct I/O requests
2897 * @inode: inode the direct I/O happens on
2899 * This is called once we've finished processing a direct I/O request,
2900 * and is used to wake up callers waiting for direct I/O to be quiesced.
2902 static inline void inode_dio_end(struct inode *inode)
2904 if (atomic_dec_and_test(&inode->i_dio_count))
2905 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
2908 extern void inode_set_flags(struct inode *inode, unsigned int flags,
2911 extern const struct file_operations generic_ro_fops;
2913 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2915 extern int readlink_copy(char __user *, int, const char *);
2916 extern int page_readlink(struct dentry *, char __user *, int);
2917 extern const char *page_get_link(struct dentry *, struct inode *,
2918 struct delayed_call *);
2919 extern void page_put_link(void *);
2920 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2922 extern int page_symlink(struct inode *inode, const char *symname, int len);
2923 extern const struct inode_operations page_symlink_inode_operations;
2924 extern void kfree_link(void *);
2925 extern int generic_readlink(struct dentry *, char __user *, int);
2926 extern void generic_fillattr(struct inode *, struct kstat *);
2927 int vfs_getattr_nosec(struct path *path, struct kstat *stat);
2928 extern int vfs_getattr(struct path *, struct kstat *);
2929 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2930 void inode_add_bytes(struct inode *inode, loff_t bytes);
2931 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2932 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2933 loff_t inode_get_bytes(struct inode *inode);
2934 void inode_set_bytes(struct inode *inode, loff_t bytes);
2935 const char *simple_get_link(struct dentry *, struct inode *,
2936 struct delayed_call *);
2937 extern const struct inode_operations simple_symlink_inode_operations;
2939 extern int iterate_dir(struct file *, struct dir_context *);
2941 extern int vfs_stat(const char __user *, struct kstat *);
2942 extern int vfs_lstat(const char __user *, struct kstat *);
2943 extern int vfs_fstat(unsigned int, struct kstat *);
2944 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2945 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
2947 extern int __generic_block_fiemap(struct inode *inode,
2948 struct fiemap_extent_info *fieinfo,
2949 loff_t start, loff_t len,
2950 get_block_t *get_block);
2951 extern int generic_block_fiemap(struct inode *inode,
2952 struct fiemap_extent_info *fieinfo, u64 start,
2953 u64 len, get_block_t *get_block);
2955 extern void get_filesystem(struct file_system_type *fs);
2956 extern void put_filesystem(struct file_system_type *fs);
2957 extern struct file_system_type *get_fs_type(const char *name);
2958 extern struct super_block *get_super(struct block_device *);
2959 extern struct super_block *get_super_thawed(struct block_device *);
2960 extern struct super_block *get_active_super(struct block_device *bdev);
2961 extern void drop_super(struct super_block *sb);
2962 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2963 extern void iterate_supers_type(struct file_system_type *,
2964 void (*)(struct super_block *, void *), void *);
2966 extern int dcache_dir_open(struct inode *, struct file *);
2967 extern int dcache_dir_close(struct inode *, struct file *);
2968 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2969 extern int dcache_readdir(struct file *, struct dir_context *);
2970 extern int simple_setattr(struct dentry *, struct iattr *);
2971 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2972 extern int simple_statfs(struct dentry *, struct kstatfs *);
2973 extern int simple_open(struct inode *inode, struct file *file);
2974 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2975 extern int simple_unlink(struct inode *, struct dentry *);
2976 extern int simple_rmdir(struct inode *, struct dentry *);
2977 extern int simple_rename(struct inode *, struct dentry *,
2978 struct inode *, struct dentry *, unsigned int);
2979 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2980 extern int simple_empty(struct dentry *);
2981 extern int simple_readpage(struct file *file, struct page *page);
2982 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2983 loff_t pos, unsigned len, unsigned flags,
2984 struct page **pagep, void **fsdata);
2985 extern int simple_write_end(struct file *file, struct address_space *mapping,
2986 loff_t pos, unsigned len, unsigned copied,
2987 struct page *page, void *fsdata);
2988 extern int always_delete_dentry(const struct dentry *);
2989 extern struct inode *alloc_anon_inode(struct super_block *);
2990 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
2991 extern const struct dentry_operations simple_dentry_operations;
2993 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2994 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2995 extern const struct file_operations simple_dir_operations;
2996 extern const struct inode_operations simple_dir_inode_operations;
2997 extern void make_empty_dir_inode(struct inode *inode);
2998 extern bool is_empty_dir_inode(struct inode *inode);
2999 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
3000 struct dentry *d_alloc_name(struct dentry *, const char *);
3001 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
3002 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3003 extern void simple_release_fs(struct vfsmount **mount, int *count);
3005 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3006 loff_t *ppos, const void *from, size_t available);
3007 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3008 const void __user *from, size_t count);
3010 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3011 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3013 extern int generic_check_addressable(unsigned, u64);
3015 #ifdef CONFIG_MIGRATION
3016 extern int buffer_migrate_page(struct address_space *,
3017 struct page *, struct page *,
3020 #define buffer_migrate_page NULL
3023 extern int setattr_prepare(struct dentry *, struct iattr *);
3024 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3025 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
3027 extern int file_update_time(struct file *file);
3029 extern int generic_show_options(struct seq_file *m, struct dentry *root);
3030 extern void save_mount_options(struct super_block *sb, char *options);
3031 extern void replace_mount_options(struct super_block *sb, char *options);
3033 static inline bool io_is_direct(struct file *filp)
3035 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
3038 static inline bool vma_is_dax(struct vm_area_struct *vma)
3040 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3043 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3045 struct inode *inode;
3049 if (!vma_is_dax(vma))
3051 inode = file_inode(vma->vm_file);
3052 if (S_ISCHR(inode->i_mode))
3053 return false; /* device-dax */
3057 static inline int iocb_flags(struct file *file)
3060 if (file->f_flags & O_APPEND)
3062 if (io_is_direct(file))
3064 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3066 if (file->f_flags & __O_SYNC)
3071 static inline ino_t parent_ino(struct dentry *dentry)
3076 * Don't strictly need d_lock here? If the parent ino could change
3077 * then surely we'd have a deeper race in the caller?
3079 spin_lock(&dentry->d_lock);
3080 res = dentry->d_parent->d_inode->i_ino;
3081 spin_unlock(&dentry->d_lock);
3085 /* Transaction based IO helpers */
3088 * An argresp is stored in an allocated page and holds the
3089 * size of the argument or response, along with its content
3091 struct simple_transaction_argresp {
3096 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3098 char *simple_transaction_get(struct file *file, const char __user *buf,
3100 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3101 size_t size, loff_t *pos);
3102 int simple_transaction_release(struct inode *inode, struct file *file);
3104 void simple_transaction_set(struct file *file, size_t n);
3107 * simple attribute files
3109 * These attributes behave similar to those in sysfs:
3111 * Writing to an attribute immediately sets a value, an open file can be
3112 * written to multiple times.
3114 * Reading from an attribute creates a buffer from the value that might get
3115 * read with multiple read calls. When the attribute has been read
3116 * completely, no further read calls are possible until the file is opened
3119 * All attributes contain a text representation of a numeric value
3120 * that are accessed with the get() and set() functions.
3122 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3123 static int __fops ## _open(struct inode *inode, struct file *file) \
3125 __simple_attr_check_format(__fmt, 0ull); \
3126 return simple_attr_open(inode, file, __get, __set, __fmt); \
3128 static const struct file_operations __fops = { \
3129 .owner = THIS_MODULE, \
3130 .open = __fops ## _open, \
3131 .release = simple_attr_release, \
3132 .read = simple_attr_read, \
3133 .write = simple_attr_write, \
3134 .llseek = generic_file_llseek, \
3137 static inline __printf(1, 2)
3138 void __simple_attr_check_format(const char *fmt, ...)
3140 /* don't do anything, just let the compiler check the arguments; */
3143 int simple_attr_open(struct inode *inode, struct file *file,
3144 int (*get)(void *, u64 *), int (*set)(void *, u64),
3146 int simple_attr_release(struct inode *inode, struct file *file);
3147 ssize_t simple_attr_read(struct file *file, char __user *buf,
3148 size_t len, loff_t *ppos);
3149 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3150 size_t len, loff_t *ppos);
3153 int proc_nr_files(struct ctl_table *table, int write,
3154 void __user *buffer, size_t *lenp, loff_t *ppos);
3155 int proc_nr_dentry(struct ctl_table *table, int write,
3156 void __user *buffer, size_t *lenp, loff_t *ppos);
3157 int proc_nr_inodes(struct ctl_table *table, int write,
3158 void __user *buffer, size_t *lenp, loff_t *ppos);
3159 int __init get_filesystem_list(char *buf);
3161 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3162 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3164 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3165 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3166 (flag & __FMODE_NONOTIFY)))
3168 static inline bool is_sxid(umode_t mode)
3170 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3173 static inline int check_sticky(struct inode *dir, struct inode *inode)
3175 if (!(dir->i_mode & S_ISVTX))
3178 return __check_sticky(dir, inode);
3181 static inline void inode_has_no_xattr(struct inode *inode)
3183 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
3184 inode->i_flags |= S_NOSEC;
3187 static inline bool is_root_inode(struct inode *inode)
3189 return inode == inode->i_sb->s_root->d_inode;
3192 static inline bool dir_emit(struct dir_context *ctx,
3193 const char *name, int namelen,
3194 u64 ino, unsigned type)
3196 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3198 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3200 return ctx->actor(ctx, ".", 1, ctx->pos,
3201 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3203 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3205 return ctx->actor(ctx, "..", 2, ctx->pos,
3206 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3208 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3210 if (ctx->pos == 0) {
3211 if (!dir_emit_dot(file, ctx))
3215 if (ctx->pos == 1) {
3216 if (!dir_emit_dotdot(file, ctx))
3222 static inline bool dir_relax(struct inode *inode)
3224 inode_unlock(inode);
3226 return !IS_DEADDIR(inode);
3229 static inline bool dir_relax_shared(struct inode *inode)
3231 inode_unlock_shared(inode);
3232 inode_lock_shared(inode);
3233 return !IS_DEADDIR(inode);
3236 extern bool path_noexec(const struct path *path);
3237 extern void inode_nohighmem(struct inode *inode);
3239 #endif /* _LINUX_FS_H */