4 #include <linux/linkage.h>
5 #include <linux/wait.h>
6 #include <linux/kdev_t.h>
7 #include <linux/dcache.h>
8 #include <linux/path.h>
9 #include <linux/stat.h>
10 #include <linux/cache.h>
11 #include <linux/list.h>
12 #include <linux/list_lru.h>
13 #include <linux/llist.h>
14 #include <linux/radix-tree.h>
15 #include <linux/rbtree.h>
16 #include <linux/init.h>
17 #include <linux/pid.h>
18 #include <linux/bug.h>
19 #include <linux/mutex.h>
20 #include <linux/rwsem.h>
21 #include <linux/capability.h>
22 #include <linux/semaphore.h>
23 #include <linux/fiemap.h>
24 #include <linux/rculist_bl.h>
25 #include <linux/atomic.h>
26 #include <linux/shrinker.h>
27 #include <linux/migrate_mode.h>
28 #include <linux/uidgid.h>
29 #include <linux/lockdep.h>
30 #include <linux/percpu-rwsem.h>
31 #include <linux/blk_types.h>
32 #include <linux/workqueue.h>
33 #include <linux/percpu-rwsem.h>
34 #include <linux/delayed_call.h>
36 #include <asm/byteorder.h>
37 #include <uapi/linux/fs.h>
39 struct backing_dev_info;
41 struct export_operations;
46 struct pipe_inode_info;
47 struct poll_table_struct;
49 struct vm_area_struct;
52 struct swap_info_struct;
54 struct workqueue_struct;
57 struct fscrypt_operations;
59 extern void __init inode_init(void);
60 extern void __init inode_init_early(void);
61 extern void __init files_init(void);
62 extern void __init files_maxfiles_init(void);
64 extern struct files_stat_struct files_stat;
65 extern unsigned long get_max_files(void);
66 extern int sysctl_nr_open;
67 extern struct inodes_stat_t inodes_stat;
68 extern int leases_enable, lease_break_time;
69 extern int sysctl_protected_symlinks;
70 extern int sysctl_protected_hardlinks;
73 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
74 struct buffer_head *bh_result, int create);
75 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
76 ssize_t bytes, void *private);
78 #define MAY_EXEC 0x00000001
79 #define MAY_WRITE 0x00000002
80 #define MAY_READ 0x00000004
81 #define MAY_APPEND 0x00000008
82 #define MAY_ACCESS 0x00000010
83 #define MAY_OPEN 0x00000020
84 #define MAY_CHDIR 0x00000040
85 /* called from RCU mode, don't block */
86 #define MAY_NOT_BLOCK 0x00000080
89 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
90 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
93 /* file is open for reading */
94 #define FMODE_READ ((__force fmode_t)0x1)
95 /* file is open for writing */
96 #define FMODE_WRITE ((__force fmode_t)0x2)
97 /* file is seekable */
98 #define FMODE_LSEEK ((__force fmode_t)0x4)
99 /* file can be accessed using pread */
100 #define FMODE_PREAD ((__force fmode_t)0x8)
101 /* file can be accessed using pwrite */
102 #define FMODE_PWRITE ((__force fmode_t)0x10)
103 /* File is opened for execution with sys_execve / sys_uselib */
104 #define FMODE_EXEC ((__force fmode_t)0x20)
105 /* File is opened with O_NDELAY (only set for block devices) */
106 #define FMODE_NDELAY ((__force fmode_t)0x40)
107 /* File is opened with O_EXCL (only set for block devices) */
108 #define FMODE_EXCL ((__force fmode_t)0x80)
109 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
110 (specialy hack for floppy.c) */
111 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
112 /* 32bit hashes as llseek() offset (for directories) */
113 #define FMODE_32BITHASH ((__force fmode_t)0x200)
114 /* 64bit hashes as llseek() offset (for directories) */
115 #define FMODE_64BITHASH ((__force fmode_t)0x400)
118 * Don't update ctime and mtime.
120 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
121 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
123 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
125 /* Expect random access pattern */
126 #define FMODE_RANDOM ((__force fmode_t)0x1000)
128 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
129 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
131 /* File is opened with O_PATH; almost nothing can be done with it */
132 #define FMODE_PATH ((__force fmode_t)0x4000)
134 /* File needs atomic accesses to f_pos */
135 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
136 /* Write access to underlying fs */
137 #define FMODE_WRITER ((__force fmode_t)0x10000)
138 /* Has read method(s) */
139 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
140 /* Has write method(s) */
141 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
143 /* File was opened by fanotify and shouldn't generate fanotify events */
144 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
147 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
148 * that indicates that they should check the contents of the iovec are
149 * valid, but not check the memory that the iovec elements
152 #define CHECK_IOVEC_ONLY -1
155 * The below are the various read and write types that we support. Some of
156 * them include behavioral modifiers that send information down to the
157 * block layer and IO scheduler. Terminology:
159 * The block layer uses device plugging to defer IO a little bit, in
160 * the hope that we will see more IO very shortly. This increases
161 * coalescing of adjacent IO and thus reduces the number of IOs we
162 * have to send to the device. It also allows for better queuing,
163 * if the IO isn't mergeable. If the caller is going to be waiting
164 * for the IO, then he must ensure that the device is unplugged so
165 * that the IO is dispatched to the driver.
167 * All IO is handled async in Linux. This is fine for background
168 * writes, but for reads or writes that someone waits for completion
169 * on, we want to notify the block layer and IO scheduler so that they
170 * know about it. That allows them to make better scheduling
171 * decisions. So when the below references 'sync' and 'async', it
172 * is referencing this priority hint.
174 * With that in mind, the available types are:
176 * READ A normal read operation. Device will be plugged.
177 * READ_SYNC A synchronous read. Device is not plugged, caller can
178 * immediately wait on this read without caring about
180 * READA Used for read-ahead operations. Lower priority, and the
181 * block layer could (in theory) choose to ignore this
182 * request if it runs into resource problems.
183 * WRITE A normal async write. Device will be plugged.
184 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
185 * the hint that someone will be waiting on this IO
186 * shortly. The write equivalent of READ_SYNC.
187 * WRITE_ODIRECT Special case write for O_DIRECT only.
188 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
189 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
190 * non-volatile media on completion.
191 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
192 * by a cache flush and data is guaranteed to be on
193 * non-volatile media on completion.
196 #define RW_MASK REQ_WRITE
197 #define RWA_MASK REQ_RAHEAD
200 #define WRITE RW_MASK
201 #define READA RWA_MASK
203 #define READ_SYNC (READ | REQ_SYNC)
204 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
205 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
206 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
207 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
208 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
211 * Attribute flags. These should be or-ed together to figure out what
214 #define ATTR_MODE (1 << 0)
215 #define ATTR_UID (1 << 1)
216 #define ATTR_GID (1 << 2)
217 #define ATTR_SIZE (1 << 3)
218 #define ATTR_ATIME (1 << 4)
219 #define ATTR_MTIME (1 << 5)
220 #define ATTR_CTIME (1 << 6)
221 #define ATTR_ATIME_SET (1 << 7)
222 #define ATTR_MTIME_SET (1 << 8)
223 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
224 #define ATTR_ATTR_FLAG (1 << 10)
225 #define ATTR_KILL_SUID (1 << 11)
226 #define ATTR_KILL_SGID (1 << 12)
227 #define ATTR_FILE (1 << 13)
228 #define ATTR_KILL_PRIV (1 << 14)
229 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
230 #define ATTR_TIMES_SET (1 << 16)
233 * Whiteout is represented by a char device. The following constants define the
234 * mode and device number to use.
236 #define WHITEOUT_MODE 0
237 #define WHITEOUT_DEV 0
240 * This is the Inode Attributes structure, used for notify_change(). It
241 * uses the above definitions as flags, to know which values have changed.
242 * Also, in this manner, a Filesystem can look at only the values it cares
243 * about. Basically, these are the attributes that the VFS layer can
244 * request to change from the FS layer.
246 * Derek Atkins <warlord@MIT.EDU> 94-10-20
249 unsigned int ia_valid;
254 struct timespec ia_atime;
255 struct timespec ia_mtime;
256 struct timespec ia_ctime;
259 * Not an attribute, but an auxiliary info for filesystems wanting to
260 * implement an ftruncate() like method. NOTE: filesystem should
261 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
263 struct file *ia_file;
267 * Includes for diskquotas.
269 #include <linux/quota.h>
272 * Maximum number of layers of fs stack. Needs to be limited to
273 * prevent kernel stack overflow
275 #define FILESYSTEM_MAX_STACK_DEPTH 2
278 * enum positive_aop_returns - aop return codes with specific semantics
280 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
281 * completed, that the page is still locked, and
282 * should be considered active. The VM uses this hint
283 * to return the page to the active list -- it won't
284 * be a candidate for writeback again in the near
285 * future. Other callers must be careful to unlock
286 * the page if they get this return. Returned by
289 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
290 * unlocked it and the page might have been truncated.
291 * The caller should back up to acquiring a new page and
292 * trying again. The aop will be taking reasonable
293 * precautions not to livelock. If the caller held a page
294 * reference, it should drop it before retrying. Returned
297 * address_space_operation functions return these large constants to indicate
298 * special semantics to the caller. These are much larger than the bytes in a
299 * page to allow for functions that return the number of bytes operated on in a
303 enum positive_aop_returns {
304 AOP_WRITEPAGE_ACTIVATE = 0x80000,
305 AOP_TRUNCATED_PAGE = 0x80001,
308 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
309 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
310 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
311 * helper code (eg buffer layer)
312 * to clear GFP_FS from alloc */
315 * oh the beauties of C type declarations.
318 struct address_space;
319 struct writeback_control;
321 #define IOCB_EVENTFD (1 << 0)
322 #define IOCB_APPEND (1 << 1)
323 #define IOCB_DIRECT (1 << 2)
324 #define IOCB_HIPRI (1 << 3)
327 struct file *ki_filp;
329 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
334 static inline bool is_sync_kiocb(struct kiocb *kiocb)
336 return kiocb->ki_complete == NULL;
339 static inline int iocb_flags(struct file *file);
341 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
343 *kiocb = (struct kiocb) {
345 .ki_flags = iocb_flags(filp),
350 * "descriptor" for what we're up to with a read.
351 * This allows us to use the same read code yet
352 * have multiple different users of the data that
353 * we read from a file.
355 * The simplest case just copies the data to user
368 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
369 unsigned long, unsigned long);
371 struct address_space_operations {
372 int (*writepage)(struct page *page, struct writeback_control *wbc);
373 int (*readpage)(struct file *, struct page *);
375 /* Write back some dirty pages from this mapping. */
376 int (*writepages)(struct address_space *, struct writeback_control *);
378 /* Set a page dirty. Return true if this dirtied it */
379 int (*set_page_dirty)(struct page *page);
381 int (*readpages)(struct file *filp, struct address_space *mapping,
382 struct list_head *pages, unsigned nr_pages);
384 int (*write_begin)(struct file *, struct address_space *mapping,
385 loff_t pos, unsigned len, unsigned flags,
386 struct page **pagep, void **fsdata);
387 int (*write_end)(struct file *, struct address_space *mapping,
388 loff_t pos, unsigned len, unsigned copied,
389 struct page *page, void *fsdata);
391 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
392 sector_t (*bmap)(struct address_space *, sector_t);
393 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
394 int (*releasepage) (struct page *, gfp_t);
395 void (*freepage)(struct page *);
396 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter, loff_t offset);
398 * migrate the contents of a page to the specified target. If
399 * migrate_mode is MIGRATE_ASYNC, it must not block.
401 int (*migratepage) (struct address_space *,
402 struct page *, struct page *, enum migrate_mode);
403 int (*launder_page) (struct page *);
404 int (*is_partially_uptodate) (struct page *, unsigned long,
406 void (*is_dirty_writeback) (struct page *, bool *, bool *);
407 int (*error_remove_page)(struct address_space *, struct page *);
409 /* swapfile support */
410 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
412 void (*swap_deactivate)(struct file *file);
415 extern const struct address_space_operations empty_aops;
418 * pagecache_write_begin/pagecache_write_end must be used by general code
419 * to write into the pagecache.
421 int pagecache_write_begin(struct file *, struct address_space *mapping,
422 loff_t pos, unsigned len, unsigned flags,
423 struct page **pagep, void **fsdata);
425 int pagecache_write_end(struct file *, struct address_space *mapping,
426 loff_t pos, unsigned len, unsigned copied,
427 struct page *page, void *fsdata);
429 struct address_space {
430 struct inode *host; /* owner: inode, block_device */
431 struct radix_tree_root page_tree; /* radix tree of all pages */
432 spinlock_t tree_lock; /* and lock protecting it */
433 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
434 struct rb_root i_mmap; /* tree of private and shared mappings */
435 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
436 /* Protected by tree_lock together with the radix tree */
437 unsigned long nrpages; /* number of total pages */
438 /* number of shadow or DAX exceptional entries */
439 unsigned long nrexceptional;
440 pgoff_t writeback_index;/* writeback starts here */
441 const struct address_space_operations *a_ops; /* methods */
442 unsigned long flags; /* error bits/gfp mask */
443 spinlock_t private_lock; /* for use by the address_space */
444 struct list_head private_list; /* ditto */
445 void *private_data; /* ditto */
446 } __attribute__((aligned(sizeof(long))));
448 * On most architectures that alignment is already the case; but
449 * must be enforced here for CRIS, to let the least significant bit
450 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
452 struct request_queue;
454 struct block_device {
455 dev_t bd_dev; /* not a kdev_t - it's a search key */
457 struct inode * bd_inode; /* will die */
458 struct super_block * bd_super;
459 struct mutex bd_mutex; /* open/close mutex */
460 struct list_head bd_inodes;
464 bool bd_write_holder;
466 struct list_head bd_holder_disks;
468 struct block_device * bd_contains;
469 unsigned bd_block_size;
470 struct hd_struct * bd_part;
471 /* number of times partitions within this device have been opened. */
472 unsigned bd_part_count;
474 struct gendisk * bd_disk;
475 struct request_queue * bd_queue;
476 struct list_head bd_list;
478 * Private data. You must have bd_claim'ed the block_device
479 * to use this. NOTE: bd_claim allows an owner to claim
480 * the same device multiple times, the owner must take special
481 * care to not mess up bd_private for that case.
483 unsigned long bd_private;
485 /* The counter of freeze processes */
486 int bd_fsfreeze_count;
487 /* Mutex for freeze */
488 struct mutex bd_fsfreeze_mutex;
492 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
495 #define PAGECACHE_TAG_DIRTY 0
496 #define PAGECACHE_TAG_WRITEBACK 1
497 #define PAGECACHE_TAG_TOWRITE 2
499 int mapping_tagged(struct address_space *mapping, int tag);
501 static inline void i_mmap_lock_write(struct address_space *mapping)
503 down_write(&mapping->i_mmap_rwsem);
506 static inline void i_mmap_unlock_write(struct address_space *mapping)
508 up_write(&mapping->i_mmap_rwsem);
511 static inline void i_mmap_lock_read(struct address_space *mapping)
513 down_read(&mapping->i_mmap_rwsem);
516 static inline void i_mmap_unlock_read(struct address_space *mapping)
518 up_read(&mapping->i_mmap_rwsem);
522 * Might pages of this file be mapped into userspace?
524 static inline int mapping_mapped(struct address_space *mapping)
526 return !RB_EMPTY_ROOT(&mapping->i_mmap);
530 * Might pages of this file have been modified in userspace?
531 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
532 * marks vma as VM_SHARED if it is shared, and the file was opened for
533 * writing i.e. vma may be mprotected writable even if now readonly.
535 * If i_mmap_writable is negative, no new writable mappings are allowed. You
536 * can only deny writable mappings, if none exists right now.
538 static inline int mapping_writably_mapped(struct address_space *mapping)
540 return atomic_read(&mapping->i_mmap_writable) > 0;
543 static inline int mapping_map_writable(struct address_space *mapping)
545 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
549 static inline void mapping_unmap_writable(struct address_space *mapping)
551 atomic_dec(&mapping->i_mmap_writable);
554 static inline int mapping_deny_writable(struct address_space *mapping)
556 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
560 static inline void mapping_allow_writable(struct address_space *mapping)
562 atomic_inc(&mapping->i_mmap_writable);
566 * Use sequence counter to get consistent i_size on 32-bit processors.
568 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
569 #include <linux/seqlock.h>
570 #define __NEED_I_SIZE_ORDERED
571 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
573 #define i_size_ordered_init(inode) do { } while (0)
577 #define ACL_NOT_CACHED ((void *)(-1))
579 #define IOP_FASTPERM 0x0001
580 #define IOP_LOOKUP 0x0002
581 #define IOP_NOFOLLOW 0x0004
584 * Keep mostly read-only and often accessed (especially for
585 * the RCU path lookup and 'stat' data) fields at the beginning
586 * of the 'struct inode'
590 unsigned short i_opflags;
593 unsigned int i_flags;
595 #ifdef CONFIG_FS_POSIX_ACL
596 struct posix_acl *i_acl;
597 struct posix_acl *i_default_acl;
600 const struct inode_operations *i_op;
601 struct super_block *i_sb;
602 struct address_space *i_mapping;
604 #ifdef CONFIG_SECURITY
608 /* Stat data, not accessed from path walking */
611 * Filesystems may only read i_nlink directly. They shall use the
612 * following functions for modification:
614 * (set|clear|inc|drop)_nlink
615 * inode_(inc|dec)_link_count
618 const unsigned int i_nlink;
619 unsigned int __i_nlink;
623 struct timespec i_atime;
624 struct timespec i_mtime;
625 struct timespec i_ctime;
626 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
627 unsigned short i_bytes;
628 unsigned int i_blkbits;
631 #ifdef __NEED_I_SIZE_ORDERED
632 seqcount_t i_size_seqcount;
636 unsigned long i_state;
637 struct mutex i_mutex;
639 unsigned long dirtied_when; /* jiffies of first dirtying */
640 unsigned long dirtied_time_when;
642 struct hlist_node i_hash;
643 struct list_head i_io_list; /* backing dev IO list */
644 #ifdef CONFIG_CGROUP_WRITEBACK
645 struct bdi_writeback *i_wb; /* the associated cgroup wb */
647 /* foreign inode detection, see wbc_detach_inode() */
649 u16 i_wb_frn_avg_time;
650 u16 i_wb_frn_history;
652 struct list_head i_lru; /* inode LRU list */
653 struct list_head i_sb_list;
655 struct hlist_head i_dentry;
656 struct rcu_head i_rcu;
660 atomic_t i_dio_count;
661 atomic_t i_writecount;
663 atomic_t i_readcount; /* struct files open RO */
665 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
666 struct file_lock_context *i_flctx;
667 struct address_space i_data;
668 struct list_head i_devices;
670 struct pipe_inode_info *i_pipe;
671 struct block_device *i_bdev;
678 #ifdef CONFIG_FSNOTIFY
679 __u32 i_fsnotify_mask; /* all events this inode cares about */
680 struct hlist_head i_fsnotify_marks;
683 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
684 struct fscrypt_info *i_crypt_info;
687 void *i_private; /* fs or device private pointer */
690 static inline int inode_unhashed(struct inode *inode)
692 return hlist_unhashed(&inode->i_hash);
696 * inode->i_mutex nesting subclasses for the lock validator:
698 * 0: the object of the current VFS operation
702 * 4: second non-directory
703 * 5: second parent (when locking independent directories in rename)
705 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
706 * non-directories at once.
708 * The locking order between these classes is
709 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
711 enum inode_i_mutex_lock_class
721 static inline void inode_lock(struct inode *inode)
723 mutex_lock(&inode->i_mutex);
726 static inline void inode_unlock(struct inode *inode)
728 mutex_unlock(&inode->i_mutex);
731 static inline int inode_trylock(struct inode *inode)
733 return mutex_trylock(&inode->i_mutex);
736 static inline int inode_is_locked(struct inode *inode)
738 return mutex_is_locked(&inode->i_mutex);
741 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
743 mutex_lock_nested(&inode->i_mutex, subclass);
746 void lock_two_nondirectories(struct inode *, struct inode*);
747 void unlock_two_nondirectories(struct inode *, struct inode*);
750 * NOTE: in a 32bit arch with a preemptable kernel and
751 * an UP compile the i_size_read/write must be atomic
752 * with respect to the local cpu (unlike with preempt disabled),
753 * but they don't need to be atomic with respect to other cpus like in
754 * true SMP (so they need either to either locally disable irq around
755 * the read or for example on x86 they can be still implemented as a
756 * cmpxchg8b without the need of the lock prefix). For SMP compiles
757 * and 64bit archs it makes no difference if preempt is enabled or not.
759 static inline loff_t i_size_read(const struct inode *inode)
761 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
766 seq = read_seqcount_begin(&inode->i_size_seqcount);
767 i_size = inode->i_size;
768 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
770 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
774 i_size = inode->i_size;
778 return inode->i_size;
783 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
784 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
785 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
787 static inline void i_size_write(struct inode *inode, loff_t i_size)
789 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
791 write_seqcount_begin(&inode->i_size_seqcount);
792 inode->i_size = i_size;
793 write_seqcount_end(&inode->i_size_seqcount);
795 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
797 inode->i_size = i_size;
800 inode->i_size = i_size;
804 /* Helper functions so that in most cases filesystems will
805 * not need to deal directly with kuid_t and kgid_t and can
806 * instead deal with the raw numeric values that are stored
809 static inline uid_t i_uid_read(const struct inode *inode)
811 return from_kuid(&init_user_ns, inode->i_uid);
814 static inline gid_t i_gid_read(const struct inode *inode)
816 return from_kgid(&init_user_ns, inode->i_gid);
819 static inline void i_uid_write(struct inode *inode, uid_t uid)
821 inode->i_uid = make_kuid(&init_user_ns, uid);
824 static inline void i_gid_write(struct inode *inode, gid_t gid)
826 inode->i_gid = make_kgid(&init_user_ns, gid);
829 static inline unsigned iminor(const struct inode *inode)
831 return MINOR(inode->i_rdev);
834 static inline unsigned imajor(const struct inode *inode)
836 return MAJOR(inode->i_rdev);
839 extern struct block_device *I_BDEV(struct inode *inode);
842 rwlock_t lock; /* protects pid, uid, euid fields */
843 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
844 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
845 kuid_t uid, euid; /* uid/euid of process setting the owner */
846 int signum; /* posix.1b rt signal to be delivered on IO */
850 * Track a single file's readahead state
852 struct file_ra_state {
853 pgoff_t start; /* where readahead started */
854 unsigned int size; /* # of readahead pages */
855 unsigned int async_size; /* do asynchronous readahead when
856 there are only # of pages ahead */
858 unsigned int ra_pages; /* Maximum readahead window */
859 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
860 loff_t prev_pos; /* Cache last read() position */
864 * Check if @index falls in the readahead windows.
866 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
868 return (index >= ra->start &&
869 index < ra->start + ra->size);
874 struct llist_node fu_llist;
875 struct rcu_head fu_rcuhead;
878 struct inode *f_inode; /* cached value */
879 const struct file_operations *f_op;
882 * Protects f_ep_links, f_flags.
883 * Must not be taken from IRQ context.
886 atomic_long_t f_count;
887 unsigned int f_flags;
889 struct mutex f_pos_lock;
891 struct fown_struct f_owner;
892 const struct cred *f_cred;
893 struct file_ra_state f_ra;
896 #ifdef CONFIG_SECURITY
899 /* needed for tty driver, and maybe others */
903 /* Used by fs/eventpoll.c to link all the hooks to this file */
904 struct list_head f_ep_links;
905 struct list_head f_tfile_llink;
906 #endif /* #ifdef CONFIG_EPOLL */
907 struct address_space *f_mapping;
908 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
913 /* file identifier */
914 unsigned char f_handle[0];
917 static inline struct file *get_file(struct file *f)
919 atomic_long_inc(&f->f_count);
922 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
923 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
924 #define file_count(x) atomic_long_read(&(x)->f_count)
926 #define MAX_NON_LFS ((1UL<<31) - 1)
928 /* Page cache limit. The filesystems should put that into their s_maxbytes
929 limits, otherwise bad things can happen in VM. */
930 #if BITS_PER_LONG==32
931 #define MAX_LFS_FILESIZE (((loff_t)PAGE_SIZE << (BITS_PER_LONG-1))-1)
932 #elif BITS_PER_LONG==64
933 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
938 #define FL_DELEG 4 /* NFSv4 delegation */
939 #define FL_ACCESS 8 /* not trying to lock, just looking */
940 #define FL_EXISTS 16 /* when unlocking, test for existence */
941 #define FL_LEASE 32 /* lease held on this file */
942 #define FL_CLOSE 64 /* unlock on close */
943 #define FL_SLEEP 128 /* A blocking lock */
944 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
945 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
946 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
947 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
950 * Special return value from posix_lock_file() and vfs_lock_file() for
951 * asynchronous locking.
953 #define FILE_LOCK_DEFERRED 1
955 /* legacy typedef, should eventually be removed */
956 typedef void *fl_owner_t;
960 struct file_lock_operations {
961 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
962 void (*fl_release_private)(struct file_lock *);
965 struct lock_manager_operations {
966 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
967 unsigned long (*lm_owner_key)(struct file_lock *);
968 fl_owner_t (*lm_get_owner)(fl_owner_t);
969 void (*lm_put_owner)(fl_owner_t);
970 void (*lm_notify)(struct file_lock *); /* unblock callback */
971 int (*lm_grant)(struct file_lock *, int);
972 bool (*lm_break)(struct file_lock *);
973 int (*lm_change)(struct file_lock *, int, struct list_head *);
974 void (*lm_setup)(struct file_lock *, void **);
977 struct lock_manager {
978 struct list_head list;
980 * NFSv4 and up also want opens blocked during the grace period;
987 void locks_start_grace(struct net *, struct lock_manager *);
988 void locks_end_grace(struct lock_manager *);
989 int locks_in_grace(struct net *);
990 int opens_in_grace(struct net *);
992 /* that will die - we need it for nfs_lock_info */
993 #include <linux/nfs_fs_i.h>
996 * struct file_lock represents a generic "file lock". It's used to represent
997 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
998 * note that the same struct is used to represent both a request for a lock and
999 * the lock itself, but the same object is never used for both.
1001 * FIXME: should we create a separate "struct lock_request" to help distinguish
1004 * The varous i_flctx lists are ordered by:
1007 * 2) lock range start
1010 * Obviously, the last two criteria only matter for POSIX locks.
1013 struct file_lock *fl_next; /* singly linked list for this inode */
1014 struct list_head fl_list; /* link into file_lock_context */
1015 struct hlist_node fl_link; /* node in global lists */
1016 struct list_head fl_block; /* circular list of blocked processes */
1017 fl_owner_t fl_owner;
1018 unsigned int fl_flags;
1019 unsigned char fl_type;
1020 unsigned int fl_pid;
1021 int fl_link_cpu; /* what cpu's list is this on? */
1022 struct pid *fl_nspid;
1023 wait_queue_head_t fl_wait;
1024 struct file *fl_file;
1028 struct fasync_struct * fl_fasync; /* for lease break notifications */
1029 /* for lease breaks: */
1030 unsigned long fl_break_time;
1031 unsigned long fl_downgrade_time;
1033 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1034 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1036 struct nfs_lock_info nfs_fl;
1037 struct nfs4_lock_info nfs4_fl;
1039 struct list_head link; /* link in AFS vnode's pending_locks list */
1040 int state; /* state of grant or error if -ve */
1045 struct file_lock_context {
1046 spinlock_t flc_lock;
1047 struct list_head flc_flock;
1048 struct list_head flc_posix;
1049 struct list_head flc_lease;
1052 /* The following constant reflects the upper bound of the file/locking space */
1054 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1055 #define OFFSET_MAX INT_LIMIT(loff_t)
1056 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1059 #include <linux/fcntl.h>
1061 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1063 #ifdef CONFIG_FILE_LOCKING
1064 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
1065 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1066 struct flock __user *);
1068 #if BITS_PER_LONG == 32
1069 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
1070 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1071 struct flock64 __user *);
1074 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1075 extern int fcntl_getlease(struct file *filp);
1078 void locks_free_lock_context(struct inode *inode);
1079 void locks_free_lock(struct file_lock *fl);
1080 extern void locks_init_lock(struct file_lock *);
1081 extern struct file_lock * locks_alloc_lock(void);
1082 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1083 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1084 extern void locks_remove_posix(struct file *, fl_owner_t);
1085 extern void locks_remove_file(struct file *);
1086 extern void locks_release_private(struct file_lock *);
1087 extern void posix_test_lock(struct file *, struct file_lock *);
1088 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1089 extern int posix_unblock_lock(struct file_lock *);
1090 extern int vfs_test_lock(struct file *, struct file_lock *);
1091 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1092 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1093 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1094 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1095 extern void lease_get_mtime(struct inode *, struct timespec *time);
1096 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1097 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1098 extern int lease_modify(struct file_lock *, int, struct list_head *);
1099 struct files_struct;
1100 extern void show_fd_locks(struct seq_file *f,
1101 struct file *filp, struct files_struct *files);
1102 #else /* !CONFIG_FILE_LOCKING */
1103 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1104 struct flock __user *user)
1109 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1110 unsigned int cmd, struct flock __user *user)
1115 #if BITS_PER_LONG == 32
1116 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1117 struct flock64 __user *user)
1122 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1123 unsigned int cmd, struct flock64 __user *user)
1128 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1133 static inline int fcntl_getlease(struct file *filp)
1139 locks_free_lock_context(struct inode *inode)
1143 static inline void locks_init_lock(struct file_lock *fl)
1148 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1153 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1158 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1163 static inline void locks_remove_file(struct file *filp)
1168 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1173 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1174 struct file_lock *conflock)
1179 static inline int posix_unblock_lock(struct file_lock *waiter)
1184 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1189 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1190 struct file_lock *fl, struct file_lock *conf)
1195 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1200 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1205 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1210 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1215 static inline int generic_setlease(struct file *filp, long arg,
1216 struct file_lock **flp, void **priv)
1221 static inline int vfs_setlease(struct file *filp, long arg,
1222 struct file_lock **lease, void **priv)
1227 static inline int lease_modify(struct file_lock *fl, int arg,
1228 struct list_head *dispose)
1233 struct files_struct;
1234 static inline void show_fd_locks(struct seq_file *f,
1235 struct file *filp, struct files_struct *files) {}
1236 #endif /* !CONFIG_FILE_LOCKING */
1238 static inline struct inode *file_inode(const struct file *f)
1243 static inline struct dentry *file_dentry(const struct file *file)
1245 struct dentry *dentry = file->f_path.dentry;
1247 if (unlikely(dentry->d_flags & DCACHE_OP_REAL))
1248 return dentry->d_op->d_real(dentry, file_inode(file));
1253 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1255 return locks_lock_inode_wait(file_inode(filp), fl);
1258 struct fasync_struct {
1262 struct fasync_struct *fa_next; /* singly linked list */
1263 struct file *fa_file;
1264 struct rcu_head fa_rcu;
1267 #define FASYNC_MAGIC 0x4601
1269 /* SMP safe fasync helpers: */
1270 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1271 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1272 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1273 extern struct fasync_struct *fasync_alloc(void);
1274 extern void fasync_free(struct fasync_struct *);
1276 /* can be called from interrupts */
1277 extern void kill_fasync(struct fasync_struct **, int, int);
1279 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1280 extern void f_setown(struct file *filp, unsigned long arg, int force);
1281 extern void f_delown(struct file *filp);
1282 extern pid_t f_getown(struct file *filp);
1283 extern int send_sigurg(struct fown_struct *fown);
1291 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1292 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1293 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1294 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1295 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1298 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1299 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1301 /* Possible states of 'frozen' field */
1303 SB_UNFROZEN = 0, /* FS is unfrozen */
1304 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1305 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1306 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1307 * internal threads if needed) */
1308 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1311 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1314 int frozen; /* Is sb frozen? */
1315 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1316 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1319 struct super_block {
1320 struct list_head s_list; /* Keep this first */
1321 dev_t s_dev; /* search index; _not_ kdev_t */
1322 unsigned char s_blocksize_bits;
1323 unsigned long s_blocksize;
1324 loff_t s_maxbytes; /* Max file size */
1325 struct file_system_type *s_type;
1326 const struct super_operations *s_op;
1327 const struct dquot_operations *dq_op;
1328 const struct quotactl_ops *s_qcop;
1329 const struct export_operations *s_export_op;
1330 unsigned long s_flags;
1331 unsigned long s_iflags; /* internal SB_I_* flags */
1332 unsigned long s_magic;
1333 struct dentry *s_root;
1334 struct rw_semaphore s_umount;
1337 #ifdef CONFIG_SECURITY
1340 const struct xattr_handler **s_xattr;
1342 const struct fscrypt_operations *s_cop;
1344 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1345 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1346 struct block_device *s_bdev;
1347 struct backing_dev_info *s_bdi;
1348 struct mtd_info *s_mtd;
1349 struct hlist_node s_instances;
1350 unsigned int s_quota_types; /* Bitmask of supported quota types */
1351 struct quota_info s_dquot; /* Diskquota specific options */
1353 struct sb_writers s_writers;
1355 char s_id[32]; /* Informational name */
1356 u8 s_uuid[16]; /* UUID */
1358 void *s_fs_info; /* Filesystem private info */
1359 unsigned int s_max_links;
1362 /* Granularity of c/m/atime in ns.
1363 Cannot be worse than a second */
1367 * The next field is for VFS *only*. No filesystems have any business
1368 * even looking at it. You had been warned.
1370 struct mutex s_vfs_rename_mutex; /* Kludge */
1373 * Filesystem subtype. If non-empty the filesystem type field
1374 * in /proc/mounts will be "type.subtype"
1379 * Saved mount options for lazy filesystems using
1380 * generic_show_options()
1382 char __rcu *s_options;
1383 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1386 * Saved pool identifier for cleancache (-1 means none)
1388 int cleancache_poolid;
1390 struct shrinker s_shrink; /* per-sb shrinker handle */
1392 /* Number of inodes with nlink == 0 but still referenced */
1393 atomic_long_t s_remove_count;
1395 /* Being remounted read-only */
1396 int s_readonly_remount;
1398 /* AIO completions deferred from interrupt context */
1399 struct workqueue_struct *s_dio_done_wq;
1400 struct hlist_head s_pins;
1403 * Keep the lru lists last in the structure so they always sit on their
1404 * own individual cachelines.
1406 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1407 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1408 struct rcu_head rcu;
1409 struct work_struct destroy_work;
1411 struct mutex s_sync_lock; /* sync serialisation lock */
1414 * Indicates how deep in a filesystem stack this SB is
1418 /* s_inode_list_lock protects s_inodes */
1419 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1420 struct list_head s_inodes; /* all inodes */
1423 extern struct timespec current_fs_time(struct super_block *sb);
1426 * Snapshotting support.
1429 void __sb_end_write(struct super_block *sb, int level);
1430 int __sb_start_write(struct super_block *sb, int level, bool wait);
1432 #define __sb_writers_acquired(sb, lev) \
1433 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1434 #define __sb_writers_release(sb, lev) \
1435 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1438 * sb_end_write - drop write access to a superblock
1439 * @sb: the super we wrote to
1441 * Decrement number of writers to the filesystem. Wake up possible waiters
1442 * wanting to freeze the filesystem.
1444 static inline void sb_end_write(struct super_block *sb)
1446 __sb_end_write(sb, SB_FREEZE_WRITE);
1450 * sb_end_pagefault - drop write access to a superblock from a page fault
1451 * @sb: the super we wrote to
1453 * Decrement number of processes handling write page fault to the filesystem.
1454 * Wake up possible waiters wanting to freeze the filesystem.
1456 static inline void sb_end_pagefault(struct super_block *sb)
1458 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1462 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1463 * @sb: the super we wrote to
1465 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1466 * waiters wanting to freeze the filesystem.
1468 static inline void sb_end_intwrite(struct super_block *sb)
1470 __sb_end_write(sb, SB_FREEZE_FS);
1474 * sb_start_write - get write access to a superblock
1475 * @sb: the super we write to
1477 * When a process wants to write data or metadata to a file system (i.e. dirty
1478 * a page or an inode), it should embed the operation in a sb_start_write() -
1479 * sb_end_write() pair to get exclusion against file system freezing. This
1480 * function increments number of writers preventing freezing. If the file
1481 * system is already frozen, the function waits until the file system is
1484 * Since freeze protection behaves as a lock, users have to preserve
1485 * ordering of freeze protection and other filesystem locks. Generally,
1486 * freeze protection should be the outermost lock. In particular, we have:
1489 * -> i_mutex (write path, truncate, directory ops, ...)
1490 * -> s_umount (freeze_super, thaw_super)
1492 static inline void sb_start_write(struct super_block *sb)
1494 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1497 static inline int sb_start_write_trylock(struct super_block *sb)
1499 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1503 * sb_start_pagefault - get write access to a superblock from a page fault
1504 * @sb: the super we write to
1506 * When a process starts handling write page fault, it should embed the
1507 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1508 * exclusion against file system freezing. This is needed since the page fault
1509 * is going to dirty a page. This function increments number of running page
1510 * faults preventing freezing. If the file system is already frozen, the
1511 * function waits until the file system is thawed.
1513 * Since page fault freeze protection behaves as a lock, users have to preserve
1514 * ordering of freeze protection and other filesystem locks. It is advised to
1515 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1516 * handling code implies lock dependency:
1519 * -> sb_start_pagefault
1521 static inline void sb_start_pagefault(struct super_block *sb)
1523 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1527 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1528 * @sb: the super we write to
1530 * This is the third level of protection against filesystem freezing. It is
1531 * free for use by a filesystem. The only requirement is that it must rank
1532 * below sb_start_pagefault.
1534 * For example filesystem can call sb_start_intwrite() when starting a
1535 * transaction which somewhat eases handling of freezing for internal sources
1536 * of filesystem changes (internal fs threads, discarding preallocation on file
1539 static inline void sb_start_intwrite(struct super_block *sb)
1541 __sb_start_write(sb, SB_FREEZE_FS, true);
1545 extern bool inode_owner_or_capable(const struct inode *inode);
1548 * VFS helper functions..
1550 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1551 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1552 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1553 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1554 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1555 extern int vfs_rmdir(struct inode *, struct dentry *);
1556 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1557 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1558 extern int vfs_whiteout(struct inode *, struct dentry *);
1561 * VFS file helper functions.
1563 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1566 * VFS FS_IOC_FIEMAP helper definitions.
1568 struct fiemap_extent_info {
1569 unsigned int fi_flags; /* Flags as passed from user */
1570 unsigned int fi_extents_mapped; /* Number of mapped extents */
1571 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1572 struct fiemap_extent __user *fi_extents_start; /* Start of
1573 fiemap_extent array */
1575 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1576 u64 phys, u64 len, u32 flags);
1577 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1582 * NOTE! These match bits 12..15 of stat.st_mode
1583 * (ie "(i_mode >> 12) & 15").
1585 #define DT_UNKNOWN 0
1596 * This is the "filldir" function type, used by readdir() to let
1597 * the kernel specify what kind of dirent layout it wants to have.
1598 * This allows the kernel to read directories into kernel space or
1599 * to have different dirent layouts depending on the binary type.
1602 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1605 struct dir_context {
1606 const filldir_t actor;
1610 struct block_device_operations;
1612 /* These macros are for out of kernel modules to test that
1613 * the kernel supports the unlocked_ioctl and compat_ioctl
1614 * fields in struct file_operations. */
1615 #define HAVE_COMPAT_IOCTL 1
1616 #define HAVE_UNLOCKED_IOCTL 1
1619 * These flags let !MMU mmap() govern direct device mapping vs immediate
1620 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1622 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1623 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1624 * NOMMU_MAP_READ: Can be mapped for reading
1625 * NOMMU_MAP_WRITE: Can be mapped for writing
1626 * NOMMU_MAP_EXEC: Can be mapped for execution
1628 #define NOMMU_MAP_COPY 0x00000001
1629 #define NOMMU_MAP_DIRECT 0x00000008
1630 #define NOMMU_MAP_READ VM_MAYREAD
1631 #define NOMMU_MAP_WRITE VM_MAYWRITE
1632 #define NOMMU_MAP_EXEC VM_MAYEXEC
1634 #define NOMMU_VMFLAGS \
1635 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1640 struct file_operations {
1641 struct module *owner;
1642 loff_t (*llseek) (struct file *, loff_t, int);
1643 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1644 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1645 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1646 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1647 int (*iterate) (struct file *, struct dir_context *);
1648 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1649 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1650 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1651 int (*mmap) (struct file *, struct vm_area_struct *);
1652 int (*open) (struct inode *, struct file *);
1653 int (*flush) (struct file *, fl_owner_t id);
1654 int (*release) (struct inode *, struct file *);
1655 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1656 int (*aio_fsync) (struct kiocb *, int datasync);
1657 int (*fasync) (int, struct file *, int);
1658 int (*lock) (struct file *, int, struct file_lock *);
1659 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1660 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1661 int (*check_flags)(int);
1662 int (*flock) (struct file *, int, struct file_lock *);
1663 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1664 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1665 int (*setlease)(struct file *, long, struct file_lock **, void **);
1666 long (*fallocate)(struct file *file, int mode, loff_t offset,
1668 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1670 unsigned (*mmap_capabilities)(struct file *);
1672 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1673 loff_t, size_t, unsigned int);
1674 int (*clone_file_range)(struct file *, loff_t, struct file *, loff_t,
1676 ssize_t (*dedupe_file_range)(struct file *, u64, u64, struct file *,
1680 struct inode_operations {
1681 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1682 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1683 int (*permission) (struct inode *, int);
1684 struct posix_acl * (*get_acl)(struct inode *, int);
1686 int (*readlink) (struct dentry *, char __user *,int);
1688 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1689 int (*link) (struct dentry *,struct inode *,struct dentry *);
1690 int (*unlink) (struct inode *,struct dentry *);
1691 int (*symlink) (struct inode *,struct dentry *,const char *);
1692 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1693 int (*rmdir) (struct inode *,struct dentry *);
1694 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1695 int (*rename) (struct inode *, struct dentry *,
1696 struct inode *, struct dentry *);
1697 int (*rename2) (struct inode *, struct dentry *,
1698 struct inode *, struct dentry *, unsigned int);
1699 int (*setattr) (struct dentry *, struct iattr *);
1700 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1701 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1702 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1703 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1704 int (*removexattr) (struct dentry *, const char *);
1705 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1707 int (*update_time)(struct inode *, struct timespec *, int);
1708 int (*atomic_open)(struct inode *, struct dentry *,
1709 struct file *, unsigned open_flag,
1710 umode_t create_mode, int *opened);
1711 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1712 int (*set_acl)(struct inode *, struct posix_acl *, int);
1713 } ____cacheline_aligned;
1715 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1716 unsigned long nr_segs, unsigned long fast_segs,
1717 struct iovec *fast_pointer,
1718 struct iovec **ret_pointer);
1720 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1721 extern ssize_t __vfs_write(struct file *, const char __user *, size_t, loff_t *);
1722 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1723 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1724 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1725 unsigned long, loff_t *, int);
1726 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1727 unsigned long, loff_t *, int);
1728 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1729 loff_t, size_t, unsigned int);
1730 extern int vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1731 struct file *file_out, loff_t pos_out, u64 len);
1732 extern int vfs_dedupe_file_range(struct file *file,
1733 struct file_dedupe_range *same);
1735 struct super_operations {
1736 struct inode *(*alloc_inode)(struct super_block *sb);
1737 void (*destroy_inode)(struct inode *);
1739 void (*dirty_inode) (struct inode *, int flags);
1740 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1741 int (*drop_inode) (struct inode *);
1742 void (*evict_inode) (struct inode *);
1743 void (*put_super) (struct super_block *);
1744 int (*sync_fs)(struct super_block *sb, int wait);
1745 int (*freeze_super) (struct super_block *);
1746 int (*freeze_fs) (struct super_block *);
1747 int (*thaw_super) (struct super_block *);
1748 int (*unfreeze_fs) (struct super_block *);
1749 int (*statfs) (struct dentry *, struct kstatfs *);
1750 int (*remount_fs) (struct super_block *, int *, char *);
1751 void (*umount_begin) (struct super_block *);
1753 int (*show_options)(struct seq_file *, struct dentry *);
1754 int (*show_devname)(struct seq_file *, struct dentry *);
1755 int (*show_path)(struct seq_file *, struct dentry *);
1756 int (*show_stats)(struct seq_file *, struct dentry *);
1758 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1759 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1760 struct dquot **(*get_dquots)(struct inode *);
1762 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1763 long (*nr_cached_objects)(struct super_block *,
1764 struct shrink_control *);
1765 long (*free_cached_objects)(struct super_block *,
1766 struct shrink_control *);
1770 * Inode flags - they have no relation to superblock flags now
1772 #define S_SYNC 1 /* Writes are synced at once */
1773 #define S_NOATIME 2 /* Do not update access times */
1774 #define S_APPEND 4 /* Append-only file */
1775 #define S_IMMUTABLE 8 /* Immutable file */
1776 #define S_DEAD 16 /* removed, but still open directory */
1777 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1778 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1779 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1780 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1781 #define S_PRIVATE 512 /* Inode is fs-internal */
1782 #define S_IMA 1024 /* Inode has an associated IMA struct */
1783 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1784 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1785 #ifdef CONFIG_FS_DAX
1786 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1788 #define S_DAX 0 /* Make all the DAX code disappear */
1792 * Note that nosuid etc flags are inode-specific: setting some file-system
1793 * flags just means all the inodes inherit those flags by default. It might be
1794 * possible to override it selectively if you really wanted to with some
1795 * ioctl() that is not currently implemented.
1797 * Exception: MS_RDONLY is always applied to the entire file system.
1799 * Unfortunately, it is possible to change a filesystems flags with it mounted
1800 * with files in use. This means that all of the inodes will not have their
1801 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1802 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1804 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1806 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1807 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1808 ((inode)->i_flags & S_SYNC))
1809 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1810 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1811 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1812 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1813 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1815 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1816 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1817 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1818 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1820 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1821 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1822 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1823 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1824 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1825 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1826 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1827 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1829 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1830 (inode)->i_rdev == WHITEOUT_DEV)
1833 * Inode state bits. Protected by inode->i_lock
1835 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1836 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1838 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1839 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1840 * various stages of removing an inode.
1842 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1844 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1845 * fdatasync(). i_atime is the usual cause.
1846 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1847 * these changes separately from I_DIRTY_SYNC so that we
1848 * don't have to write inode on fdatasync() when only
1849 * mtime has changed in it.
1850 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1851 * I_NEW Serves as both a mutex and completion notification.
1852 * New inodes set I_NEW. If two processes both create
1853 * the same inode, one of them will release its inode and
1854 * wait for I_NEW to be released before returning.
1855 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1856 * also cause waiting on I_NEW, without I_NEW actually
1857 * being set. find_inode() uses this to prevent returning
1858 * nearly-dead inodes.
1859 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1860 * is zero. I_FREEING must be set when I_WILL_FREE is
1862 * I_FREEING Set when inode is about to be freed but still has dirty
1863 * pages or buffers attached or the inode itself is still
1865 * I_CLEAR Added by clear_inode(). In this state the inode is
1866 * clean and can be destroyed. Inode keeps I_FREEING.
1868 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1869 * prohibited for many purposes. iget() must wait for
1870 * the inode to be completely released, then create it
1871 * anew. Other functions will just ignore such inodes,
1872 * if appropriate. I_NEW is used for waiting.
1874 * I_SYNC Writeback of inode is running. The bit is set during
1875 * data writeback, and cleared with a wakeup on the bit
1876 * address once it is done. The bit is also used to pin
1877 * the inode in memory for flusher thread.
1879 * I_REFERENCED Marks the inode as recently references on the LRU list.
1881 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1883 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1884 * synchronize competing switching instances and to tell
1885 * wb stat updates to grab mapping->tree_lock. See
1886 * inode_switch_wb_work_fn() for details.
1888 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1890 #define I_DIRTY_SYNC (1 << 0)
1891 #define I_DIRTY_DATASYNC (1 << 1)
1892 #define I_DIRTY_PAGES (1 << 2)
1894 #define I_NEW (1 << __I_NEW)
1895 #define I_WILL_FREE (1 << 4)
1896 #define I_FREEING (1 << 5)
1897 #define I_CLEAR (1 << 6)
1899 #define I_SYNC (1 << __I_SYNC)
1900 #define I_REFERENCED (1 << 8)
1901 #define __I_DIO_WAKEUP 9
1902 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1903 #define I_LINKABLE (1 << 10)
1904 #define I_DIRTY_TIME (1 << 11)
1905 #define __I_DIRTY_TIME_EXPIRED 12
1906 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1907 #define I_WB_SWITCH (1 << 13)
1909 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1910 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1912 extern void __mark_inode_dirty(struct inode *, int);
1913 static inline void mark_inode_dirty(struct inode *inode)
1915 __mark_inode_dirty(inode, I_DIRTY);
1918 static inline void mark_inode_dirty_sync(struct inode *inode)
1920 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1923 extern void inc_nlink(struct inode *inode);
1924 extern void drop_nlink(struct inode *inode);
1925 extern void clear_nlink(struct inode *inode);
1926 extern void set_nlink(struct inode *inode, unsigned int nlink);
1928 static inline void inode_inc_link_count(struct inode *inode)
1931 mark_inode_dirty(inode);
1934 static inline void inode_dec_link_count(struct inode *inode)
1937 mark_inode_dirty(inode);
1941 * inode_inc_iversion - increments i_version
1942 * @inode: inode that need to be updated
1944 * Every time the inode is modified, the i_version field will be incremented.
1945 * The filesystem has to be mounted with i_version flag
1948 static inline void inode_inc_iversion(struct inode *inode)
1950 spin_lock(&inode->i_lock);
1952 spin_unlock(&inode->i_lock);
1955 enum file_time_flags {
1962 extern bool atime_needs_update(const struct path *, struct inode *);
1963 extern void touch_atime(const struct path *);
1964 static inline void file_accessed(struct file *file)
1966 if (!(file->f_flags & O_NOATIME))
1967 touch_atime(&file->f_path);
1970 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1971 int sync_inode_metadata(struct inode *inode, int wait);
1973 struct file_system_type {
1976 #define FS_REQUIRES_DEV 1
1977 #define FS_BINARY_MOUNTDATA 2
1978 #define FS_HAS_SUBTYPE 4
1979 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
1980 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
1981 #define FS_USERNS_VISIBLE 32 /* FS must already be visible */
1982 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
1983 struct dentry *(*mount) (struct file_system_type *, int,
1984 const char *, void *);
1985 void (*kill_sb) (struct super_block *);
1986 struct module *owner;
1987 struct file_system_type * next;
1988 struct hlist_head fs_supers;
1990 struct lock_class_key s_lock_key;
1991 struct lock_class_key s_umount_key;
1992 struct lock_class_key s_vfs_rename_key;
1993 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
1995 struct lock_class_key i_lock_key;
1996 struct lock_class_key i_mutex_key;
1997 struct lock_class_key i_mutex_dir_key;
2000 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2002 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
2003 void *data, int (*fill_super)(struct super_block *, void *, int));
2004 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2005 int flags, const char *dev_name, void *data,
2006 int (*fill_super)(struct super_block *, void *, int));
2007 extern struct dentry *mount_single(struct file_system_type *fs_type,
2008 int flags, void *data,
2009 int (*fill_super)(struct super_block *, void *, int));
2010 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2011 int flags, void *data,
2012 int (*fill_super)(struct super_block *, void *, int));
2013 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2014 void generic_shutdown_super(struct super_block *sb);
2015 void kill_block_super(struct super_block *sb);
2016 void kill_anon_super(struct super_block *sb);
2017 void kill_litter_super(struct super_block *sb);
2018 void deactivate_super(struct super_block *sb);
2019 void deactivate_locked_super(struct super_block *sb);
2020 int set_anon_super(struct super_block *s, void *data);
2021 int get_anon_bdev(dev_t *);
2022 void free_anon_bdev(dev_t);
2023 struct super_block *sget(struct file_system_type *type,
2024 int (*test)(struct super_block *,void *),
2025 int (*set)(struct super_block *,void *),
2026 int flags, void *data);
2027 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
2028 const struct super_operations *ops,
2029 const struct dentry_operations *dops,
2032 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2033 #define fops_get(fops) \
2034 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2035 #define fops_put(fops) \
2036 do { if (fops) module_put((fops)->owner); } while(0)
2038 * This one is to be used *ONLY* from ->open() instances.
2039 * fops must be non-NULL, pinned down *and* module dependencies
2040 * should be sufficient to pin the caller down as well.
2042 #define replace_fops(f, fops) \
2044 struct file *__file = (f); \
2045 fops_put(__file->f_op); \
2046 BUG_ON(!(__file->f_op = (fops))); \
2049 extern int register_filesystem(struct file_system_type *);
2050 extern int unregister_filesystem(struct file_system_type *);
2051 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2052 #define kern_mount(type) kern_mount_data(type, NULL)
2053 extern void kern_unmount(struct vfsmount *mnt);
2054 extern int may_umount_tree(struct vfsmount *);
2055 extern int may_umount(struct vfsmount *);
2056 extern long do_mount(const char *, const char __user *,
2057 const char *, unsigned long, void *);
2058 extern struct vfsmount *collect_mounts(struct path *);
2059 extern void drop_collected_mounts(struct vfsmount *);
2060 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2062 extern int vfs_statfs(struct path *, struct kstatfs *);
2063 extern int user_statfs(const char __user *, struct kstatfs *);
2064 extern int fd_statfs(int, struct kstatfs *);
2065 extern int vfs_ustat(dev_t, struct kstatfs *);
2066 extern int freeze_super(struct super_block *super);
2067 extern int thaw_super(struct super_block *super);
2068 extern bool our_mnt(struct vfsmount *mnt);
2070 extern int current_umask(void);
2072 extern void ihold(struct inode * inode);
2073 extern void iput(struct inode *);
2074 extern int generic_update_time(struct inode *, struct timespec *, int);
2077 extern struct kobject *fs_kobj;
2079 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2081 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2082 extern int locks_mandatory_locked(struct file *);
2083 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2086 * Candidates for mandatory locking have the setgid bit set
2087 * but no group execute bit - an otherwise meaningless combination.
2090 static inline int __mandatory_lock(struct inode *ino)
2092 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2096 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2097 * otherwise these will be advisory locks
2100 static inline int mandatory_lock(struct inode *ino)
2102 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2105 static inline int locks_verify_locked(struct file *file)
2107 if (mandatory_lock(file_inode(file)))
2108 return locks_mandatory_locked(file);
2112 static inline int locks_verify_truncate(struct inode *inode,
2116 if (!inode->i_flctx || !mandatory_lock(inode))
2119 if (size < inode->i_size) {
2120 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2123 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2128 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2130 static inline int locks_mandatory_locked(struct file *file)
2135 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2136 loff_t start, loff_t end, unsigned char type)
2141 static inline int __mandatory_lock(struct inode *inode)
2146 static inline int mandatory_lock(struct inode *inode)
2151 static inline int locks_verify_locked(struct file *file)
2156 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2162 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2165 #ifdef CONFIG_FILE_LOCKING
2166 static inline int break_lease(struct inode *inode, unsigned int mode)
2169 * Since this check is lockless, we must ensure that any refcounts
2170 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2171 * could end up racing with tasks trying to set a new lease on this
2175 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2176 return __break_lease(inode, mode, FL_LEASE);
2180 static inline int break_deleg(struct inode *inode, unsigned int mode)
2183 * Since this check is lockless, we must ensure that any refcounts
2184 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2185 * could end up racing with tasks trying to set a new lease on this
2189 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2190 return __break_lease(inode, mode, FL_DELEG);
2194 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2198 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2199 if (ret == -EWOULDBLOCK && delegated_inode) {
2200 *delegated_inode = inode;
2206 static inline int break_deleg_wait(struct inode **delegated_inode)
2210 ret = break_deleg(*delegated_inode, O_WRONLY);
2211 iput(*delegated_inode);
2212 *delegated_inode = NULL;
2216 static inline int break_layout(struct inode *inode, bool wait)
2219 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2220 return __break_lease(inode,
2221 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2226 #else /* !CONFIG_FILE_LOCKING */
2227 static inline int break_lease(struct inode *inode, unsigned int mode)
2232 static inline int break_deleg(struct inode *inode, unsigned int mode)
2237 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2242 static inline int break_deleg_wait(struct inode **delegated_inode)
2248 static inline int break_layout(struct inode *inode, bool wait)
2253 #endif /* CONFIG_FILE_LOCKING */
2258 const char *name; /* pointer to actual string */
2259 const __user char *uptr; /* original userland pointer */
2260 struct audit_names *aname;
2265 extern long vfs_truncate(struct path *, loff_t);
2266 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2268 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2270 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2272 extern struct file *file_open_name(struct filename *, int, umode_t);
2273 extern struct file *filp_open(const char *, int, umode_t);
2274 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2275 const char *, int, umode_t);
2276 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2277 extern int filp_close(struct file *, fl_owner_t id);
2279 extern struct filename *getname_flags(const char __user *, int, int *);
2280 extern struct filename *getname(const char __user *);
2281 extern struct filename *getname_kernel(const char *);
2282 extern void putname(struct filename *name);
2288 extern int finish_open(struct file *file, struct dentry *dentry,
2289 int (*open)(struct inode *, struct file *),
2291 extern int finish_no_open(struct file *file, struct dentry *dentry);
2295 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2298 extern void __init vfs_caches_init_early(void);
2299 extern void __init vfs_caches_init(void);
2301 extern struct kmem_cache *names_cachep;
2303 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2304 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2307 extern int register_blkdev(unsigned int, const char *);
2308 extern void unregister_blkdev(unsigned int, const char *);
2309 extern struct block_device *bdget(dev_t);
2310 extern struct block_device *bdgrab(struct block_device *bdev);
2311 extern void bd_set_size(struct block_device *, loff_t size);
2312 extern void bd_forget(struct inode *inode);
2313 extern void bdput(struct block_device *);
2314 extern void invalidate_bdev(struct block_device *);
2315 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2316 extern int sync_blockdev(struct block_device *bdev);
2317 extern void kill_bdev(struct block_device *);
2318 extern struct super_block *freeze_bdev(struct block_device *);
2319 extern void emergency_thaw_all(void);
2320 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2321 extern int fsync_bdev(struct block_device *);
2322 #ifdef CONFIG_FS_DAX
2323 extern bool blkdev_dax_capable(struct block_device *bdev);
2325 static inline bool blkdev_dax_capable(struct block_device *bdev)
2331 extern struct super_block *blockdev_superblock;
2333 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2335 return sb == blockdev_superblock;
2338 static inline void bd_forget(struct inode *inode) {}
2339 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2340 static inline void kill_bdev(struct block_device *bdev) {}
2341 static inline void invalidate_bdev(struct block_device *bdev) {}
2343 static inline struct super_block *freeze_bdev(struct block_device *sb)
2348 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2353 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2357 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2362 extern int sync_filesystem(struct super_block *);
2363 extern const struct file_operations def_blk_fops;
2364 extern const struct file_operations def_chr_fops;
2366 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2367 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2368 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2369 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2370 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2372 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2374 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2375 extern int __blkdev_reread_part(struct block_device *bdev);
2376 extern int blkdev_reread_part(struct block_device *bdev);
2379 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2380 extern void bd_unlink_disk_holder(struct block_device *bdev,
2381 struct gendisk *disk);
2383 static inline int bd_link_disk_holder(struct block_device *bdev,
2384 struct gendisk *disk)
2388 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2389 struct gendisk *disk)
2396 #define CHRDEV_MAJOR_HASH_SIZE 255
2397 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2398 extern int register_chrdev_region(dev_t, unsigned, const char *);
2399 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2400 unsigned int count, const char *name,
2401 const struct file_operations *fops);
2402 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2403 unsigned int count, const char *name);
2404 extern void unregister_chrdev_region(dev_t, unsigned);
2405 extern void chrdev_show(struct seq_file *,off_t);
2407 static inline int register_chrdev(unsigned int major, const char *name,
2408 const struct file_operations *fops)
2410 return __register_chrdev(major, 0, 256, name, fops);
2413 static inline void unregister_chrdev(unsigned int major, const char *name)
2415 __unregister_chrdev(major, 0, 256, name);
2418 /* fs/block_dev.c */
2419 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2420 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2423 #define BLKDEV_MAJOR_HASH_SIZE 255
2424 extern const char *__bdevname(dev_t, char *buffer);
2425 extern const char *bdevname(struct block_device *bdev, char *buffer);
2426 extern struct block_device *lookup_bdev(const char *);
2427 extern void blkdev_show(struct seq_file *,off_t);
2430 #define BLKDEV_MAJOR_HASH_SIZE 0
2433 extern void init_special_inode(struct inode *, umode_t, dev_t);
2435 /* Invalid inode operations -- fs/bad_inode.c */
2436 extern void make_bad_inode(struct inode *);
2437 extern bool is_bad_inode(struct inode *);
2441 * return READ, READA, or WRITE
2443 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2446 * return data direction, READ or WRITE
2448 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2450 extern void check_disk_size_change(struct gendisk *disk,
2451 struct block_device *bdev);
2452 extern int revalidate_disk(struct gendisk *);
2453 extern int check_disk_change(struct block_device *);
2454 extern int __invalidate_device(struct block_device *, bool);
2455 extern int invalidate_partition(struct gendisk *, int);
2457 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2458 pgoff_t start, pgoff_t end);
2460 static inline void invalidate_remote_inode(struct inode *inode)
2462 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2463 S_ISLNK(inode->i_mode))
2464 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2466 extern int invalidate_inode_pages2(struct address_space *mapping);
2467 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2468 pgoff_t start, pgoff_t end);
2469 extern int write_inode_now(struct inode *, int);
2470 extern int filemap_fdatawrite(struct address_space *);
2471 extern int filemap_flush(struct address_space *);
2472 extern int filemap_fdatawait(struct address_space *);
2473 extern void filemap_fdatawait_keep_errors(struct address_space *);
2474 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2476 extern int filemap_write_and_wait(struct address_space *mapping);
2477 extern int filemap_write_and_wait_range(struct address_space *mapping,
2478 loff_t lstart, loff_t lend);
2479 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2480 loff_t start, loff_t end, int sync_mode);
2481 extern int filemap_fdatawrite_range(struct address_space *mapping,
2482 loff_t start, loff_t end);
2484 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2486 extern int vfs_fsync(struct file *file, int datasync);
2487 static inline int generic_write_sync(struct file *file, loff_t pos, loff_t count)
2489 if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host))
2491 return vfs_fsync_range(file, pos, pos + count - 1,
2492 (file->f_flags & __O_SYNC) ? 0 : 1);
2494 extern void emergency_sync(void);
2495 extern void emergency_remount(void);
2497 extern sector_t bmap(struct inode *, sector_t);
2499 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2500 extern int inode_permission(struct inode *, int);
2501 extern int __inode_permission(struct inode *, int);
2502 extern int generic_permission(struct inode *, int);
2503 extern int __check_sticky(struct inode *dir, struct inode *inode);
2505 static inline bool execute_ok(struct inode *inode)
2507 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2510 static inline void file_start_write(struct file *file)
2512 if (!S_ISREG(file_inode(file)->i_mode))
2514 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2517 static inline bool file_start_write_trylock(struct file *file)
2519 if (!S_ISREG(file_inode(file)->i_mode))
2521 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2524 static inline void file_end_write(struct file *file)
2526 if (!S_ISREG(file_inode(file)->i_mode))
2528 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2532 * get_write_access() gets write permission for a file.
2533 * put_write_access() releases this write permission.
2534 * This is used for regular files.
2535 * We cannot support write (and maybe mmap read-write shared) accesses and
2536 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2537 * can have the following values:
2538 * 0: no writers, no VM_DENYWRITE mappings
2539 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2540 * > 0: (i_writecount) users are writing to the file.
2542 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2543 * except for the cases where we don't hold i_writecount yet. Then we need to
2544 * use {get,deny}_write_access() - these functions check the sign and refuse
2545 * to do the change if sign is wrong.
2547 static inline int get_write_access(struct inode *inode)
2549 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2551 static inline int deny_write_access(struct file *file)
2553 struct inode *inode = file_inode(file);
2554 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2556 static inline void put_write_access(struct inode * inode)
2558 atomic_dec(&inode->i_writecount);
2560 static inline void allow_write_access(struct file *file)
2563 atomic_inc(&file_inode(file)->i_writecount);
2565 static inline bool inode_is_open_for_write(const struct inode *inode)
2567 return atomic_read(&inode->i_writecount) > 0;
2571 static inline void i_readcount_dec(struct inode *inode)
2573 BUG_ON(!atomic_read(&inode->i_readcount));
2574 atomic_dec(&inode->i_readcount);
2576 static inline void i_readcount_inc(struct inode *inode)
2578 atomic_inc(&inode->i_readcount);
2581 static inline void i_readcount_dec(struct inode *inode)
2585 static inline void i_readcount_inc(struct inode *inode)
2590 extern int do_pipe_flags(int *, int);
2592 enum kernel_read_file_id {
2593 READING_FIRMWARE = 1,
2595 READING_KEXEC_IMAGE,
2596 READING_KEXEC_INITRAMFS,
2601 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2602 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2603 enum kernel_read_file_id);
2604 extern int kernel_read_file_from_path(char *, void **, loff_t *, loff_t,
2605 enum kernel_read_file_id);
2606 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2607 enum kernel_read_file_id);
2608 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2609 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
2610 extern struct file * open_exec(const char *);
2612 /* fs/dcache.c -- generic fs support functions */
2613 extern bool is_subdir(struct dentry *, struct dentry *);
2614 extern bool path_is_under(struct path *, struct path *);
2616 extern char *file_path(struct file *, char *, int);
2618 #include <linux/err.h>
2620 /* needed for stackable file system support */
2621 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2623 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2625 extern int inode_init_always(struct super_block *, struct inode *);
2626 extern void inode_init_once(struct inode *);
2627 extern void address_space_init_once(struct address_space *mapping);
2628 extern struct inode * igrab(struct inode *);
2629 extern ino_t iunique(struct super_block *, ino_t);
2630 extern int inode_needs_sync(struct inode *inode);
2631 extern int generic_delete_inode(struct inode *inode);
2632 static inline int generic_drop_inode(struct inode *inode)
2634 return !inode->i_nlink || inode_unhashed(inode);
2637 extern struct inode *ilookup5_nowait(struct super_block *sb,
2638 unsigned long hashval, int (*test)(struct inode *, void *),
2640 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2641 int (*test)(struct inode *, void *), void *data);
2642 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2644 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2645 extern struct inode * iget_locked(struct super_block *, unsigned long);
2646 extern struct inode *find_inode_nowait(struct super_block *,
2648 int (*match)(struct inode *,
2649 unsigned long, void *),
2651 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2652 extern int insert_inode_locked(struct inode *);
2653 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2654 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2656 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2658 extern void unlock_new_inode(struct inode *);
2659 extern unsigned int get_next_ino(void);
2661 extern void __iget(struct inode * inode);
2662 extern void iget_failed(struct inode *);
2663 extern void clear_inode(struct inode *);
2664 extern void __destroy_inode(struct inode *);
2665 extern struct inode *new_inode_pseudo(struct super_block *sb);
2666 extern struct inode *new_inode(struct super_block *sb);
2667 extern void free_inode_nonrcu(struct inode *inode);
2668 extern int should_remove_suid(struct dentry *);
2669 extern int file_remove_privs(struct file *);
2670 extern int dentry_needs_remove_privs(struct dentry *dentry);
2671 static inline int file_needs_remove_privs(struct file *file)
2673 return dentry_needs_remove_privs(file->f_path.dentry);
2676 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2677 static inline void insert_inode_hash(struct inode *inode)
2679 __insert_inode_hash(inode, inode->i_ino);
2682 extern void __remove_inode_hash(struct inode *);
2683 static inline void remove_inode_hash(struct inode *inode)
2685 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2686 __remove_inode_hash(inode);
2689 extern void inode_sb_list_add(struct inode *inode);
2692 extern blk_qc_t submit_bio(int, struct bio *);
2693 extern int bdev_read_only(struct block_device *);
2695 extern int set_blocksize(struct block_device *, int);
2696 extern int sb_set_blocksize(struct super_block *, int);
2697 extern int sb_min_blocksize(struct super_block *, int);
2699 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2700 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2701 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2702 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2703 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2704 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2705 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *, loff_t);
2706 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2708 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos);
2709 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos);
2711 /* fs/block_dev.c */
2712 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2713 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2714 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2716 extern void block_sync_page(struct page *page);
2719 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2720 struct pipe_inode_info *, size_t, unsigned int);
2721 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2722 struct pipe_inode_info *, size_t, unsigned int);
2723 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2724 struct file *, loff_t *, size_t, unsigned int);
2725 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2726 struct file *out, loff_t *, size_t len, unsigned int flags);
2727 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2728 loff_t *opos, size_t len, unsigned int flags);
2732 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2733 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2734 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2735 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2736 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2737 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2738 int whence, loff_t maxsize, loff_t eof);
2739 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2740 int whence, loff_t size);
2741 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
2742 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
2743 extern int generic_file_open(struct inode * inode, struct file * filp);
2744 extern int nonseekable_open(struct inode * inode, struct file * filp);
2747 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2748 loff_t file_offset);
2751 /* need locking between buffered and direct access */
2754 /* filesystem does not support filling holes */
2755 DIO_SKIP_HOLES = 0x02,
2757 /* filesystem can handle aio writes beyond i_size */
2758 DIO_ASYNC_EXTEND = 0x04,
2760 /* inode/fs/bdev does not need truncate protection */
2761 DIO_SKIP_DIO_COUNT = 0x08,
2764 void dio_end_io(struct bio *bio, int error);
2766 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2767 struct block_device *bdev, struct iov_iter *iter,
2768 loff_t offset, get_block_t get_block,
2769 dio_iodone_t end_io, dio_submit_t submit_io,
2772 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2773 struct inode *inode,
2774 struct iov_iter *iter, loff_t offset,
2775 get_block_t get_block)
2777 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
2778 offset, get_block, NULL, NULL,
2779 DIO_LOCKING | DIO_SKIP_HOLES);
2783 void inode_dio_wait(struct inode *inode);
2786 * inode_dio_begin - signal start of a direct I/O requests
2787 * @inode: inode the direct I/O happens on
2789 * This is called once we've finished processing a direct I/O request,
2790 * and is used to wake up callers waiting for direct I/O to be quiesced.
2792 static inline void inode_dio_begin(struct inode *inode)
2794 atomic_inc(&inode->i_dio_count);
2798 * inode_dio_end - signal finish of a direct I/O requests
2799 * @inode: inode the direct I/O happens on
2801 * This is called once we've finished processing a direct I/O request,
2802 * and is used to wake up callers waiting for direct I/O to be quiesced.
2804 static inline void inode_dio_end(struct inode *inode)
2806 if (atomic_dec_and_test(&inode->i_dio_count))
2807 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
2810 extern void inode_set_flags(struct inode *inode, unsigned int flags,
2813 extern const struct file_operations generic_ro_fops;
2815 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2817 extern int readlink_copy(char __user *, int, const char *);
2818 extern int page_readlink(struct dentry *, char __user *, int);
2819 extern const char *page_get_link(struct dentry *, struct inode *,
2820 struct delayed_call *);
2821 extern void page_put_link(void *);
2822 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2824 extern int page_symlink(struct inode *inode, const char *symname, int len);
2825 extern const struct inode_operations page_symlink_inode_operations;
2826 extern void kfree_link(void *);
2827 extern int generic_readlink(struct dentry *, char __user *, int);
2828 extern void generic_fillattr(struct inode *, struct kstat *);
2829 int vfs_getattr_nosec(struct path *path, struct kstat *stat);
2830 extern int vfs_getattr(struct path *, struct kstat *);
2831 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2832 void inode_add_bytes(struct inode *inode, loff_t bytes);
2833 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2834 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2835 loff_t inode_get_bytes(struct inode *inode);
2836 void inode_set_bytes(struct inode *inode, loff_t bytes);
2837 const char *simple_get_link(struct dentry *, struct inode *,
2838 struct delayed_call *);
2839 extern const struct inode_operations simple_symlink_inode_operations;
2841 extern int iterate_dir(struct file *, struct dir_context *);
2843 extern int vfs_stat(const char __user *, struct kstat *);
2844 extern int vfs_lstat(const char __user *, struct kstat *);
2845 extern int vfs_fstat(unsigned int, struct kstat *);
2846 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2848 extern int __generic_block_fiemap(struct inode *inode,
2849 struct fiemap_extent_info *fieinfo,
2850 loff_t start, loff_t len,
2851 get_block_t *get_block);
2852 extern int generic_block_fiemap(struct inode *inode,
2853 struct fiemap_extent_info *fieinfo, u64 start,
2854 u64 len, get_block_t *get_block);
2856 extern void get_filesystem(struct file_system_type *fs);
2857 extern void put_filesystem(struct file_system_type *fs);
2858 extern struct file_system_type *get_fs_type(const char *name);
2859 extern struct super_block *get_super(struct block_device *);
2860 extern struct super_block *get_super_thawed(struct block_device *);
2861 extern struct super_block *get_active_super(struct block_device *bdev);
2862 extern void drop_super(struct super_block *sb);
2863 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2864 extern void iterate_supers_type(struct file_system_type *,
2865 void (*)(struct super_block *, void *), void *);
2867 extern int dcache_dir_open(struct inode *, struct file *);
2868 extern int dcache_dir_close(struct inode *, struct file *);
2869 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2870 extern int dcache_readdir(struct file *, struct dir_context *);
2871 extern int simple_setattr(struct dentry *, struct iattr *);
2872 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2873 extern int simple_statfs(struct dentry *, struct kstatfs *);
2874 extern int simple_open(struct inode *inode, struct file *file);
2875 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2876 extern int simple_unlink(struct inode *, struct dentry *);
2877 extern int simple_rmdir(struct inode *, struct dentry *);
2878 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2879 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2880 extern int simple_empty(struct dentry *);
2881 extern int simple_readpage(struct file *file, struct page *page);
2882 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2883 loff_t pos, unsigned len, unsigned flags,
2884 struct page **pagep, void **fsdata);
2885 extern int simple_write_end(struct file *file, struct address_space *mapping,
2886 loff_t pos, unsigned len, unsigned copied,
2887 struct page *page, void *fsdata);
2888 extern int always_delete_dentry(const struct dentry *);
2889 extern struct inode *alloc_anon_inode(struct super_block *);
2890 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
2891 extern const struct dentry_operations simple_dentry_operations;
2893 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2894 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2895 extern const struct file_operations simple_dir_operations;
2896 extern const struct inode_operations simple_dir_inode_operations;
2897 extern void make_empty_dir_inode(struct inode *inode);
2898 extern bool is_empty_dir_inode(struct inode *inode);
2899 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2900 struct dentry *d_alloc_name(struct dentry *, const char *);
2901 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2902 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2903 extern void simple_release_fs(struct vfsmount **mount, int *count);
2905 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2906 loff_t *ppos, const void *from, size_t available);
2907 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2908 const void __user *from, size_t count);
2910 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
2911 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2913 extern int generic_check_addressable(unsigned, u64);
2915 #ifdef CONFIG_MIGRATION
2916 extern int buffer_migrate_page(struct address_space *,
2917 struct page *, struct page *,
2920 #define buffer_migrate_page NULL
2923 extern int inode_change_ok(const struct inode *, struct iattr *);
2924 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2925 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2927 extern int file_update_time(struct file *file);
2929 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2930 extern void save_mount_options(struct super_block *sb, char *options);
2931 extern void replace_mount_options(struct super_block *sb, char *options);
2933 static inline bool io_is_direct(struct file *filp)
2935 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
2938 static inline int iocb_flags(struct file *file)
2941 if (file->f_flags & O_APPEND)
2943 if (io_is_direct(file))
2948 static inline ino_t parent_ino(struct dentry *dentry)
2953 * Don't strictly need d_lock here? If the parent ino could change
2954 * then surely we'd have a deeper race in the caller?
2956 spin_lock(&dentry->d_lock);
2957 res = dentry->d_parent->d_inode->i_ino;
2958 spin_unlock(&dentry->d_lock);
2962 /* Transaction based IO helpers */
2965 * An argresp is stored in an allocated page and holds the
2966 * size of the argument or response, along with its content
2968 struct simple_transaction_argresp {
2973 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2975 char *simple_transaction_get(struct file *file, const char __user *buf,
2977 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2978 size_t size, loff_t *pos);
2979 int simple_transaction_release(struct inode *inode, struct file *file);
2981 void simple_transaction_set(struct file *file, size_t n);
2984 * simple attribute files
2986 * These attributes behave similar to those in sysfs:
2988 * Writing to an attribute immediately sets a value, an open file can be
2989 * written to multiple times.
2991 * Reading from an attribute creates a buffer from the value that might get
2992 * read with multiple read calls. When the attribute has been read
2993 * completely, no further read calls are possible until the file is opened
2996 * All attributes contain a text representation of a numeric value
2997 * that are accessed with the get() and set() functions.
2999 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3000 static int __fops ## _open(struct inode *inode, struct file *file) \
3002 __simple_attr_check_format(__fmt, 0ull); \
3003 return simple_attr_open(inode, file, __get, __set, __fmt); \
3005 static const struct file_operations __fops = { \
3006 .owner = THIS_MODULE, \
3007 .open = __fops ## _open, \
3008 .release = simple_attr_release, \
3009 .read = simple_attr_read, \
3010 .write = simple_attr_write, \
3011 .llseek = generic_file_llseek, \
3014 static inline __printf(1, 2)
3015 void __simple_attr_check_format(const char *fmt, ...)
3017 /* don't do anything, just let the compiler check the arguments; */
3020 int simple_attr_open(struct inode *inode, struct file *file,
3021 int (*get)(void *, u64 *), int (*set)(void *, u64),
3023 int simple_attr_release(struct inode *inode, struct file *file);
3024 ssize_t simple_attr_read(struct file *file, char __user *buf,
3025 size_t len, loff_t *ppos);
3026 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3027 size_t len, loff_t *ppos);
3030 int proc_nr_files(struct ctl_table *table, int write,
3031 void __user *buffer, size_t *lenp, loff_t *ppos);
3032 int proc_nr_dentry(struct ctl_table *table, int write,
3033 void __user *buffer, size_t *lenp, loff_t *ppos);
3034 int proc_nr_inodes(struct ctl_table *table, int write,
3035 void __user *buffer, size_t *lenp, loff_t *ppos);
3036 int __init get_filesystem_list(char *buf);
3038 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3039 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3041 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3042 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3043 (flag & __FMODE_NONOTIFY)))
3045 static inline bool is_sxid(umode_t mode)
3047 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3050 static inline int check_sticky(struct inode *dir, struct inode *inode)
3052 if (!(dir->i_mode & S_ISVTX))
3055 return __check_sticky(dir, inode);
3058 static inline void inode_has_no_xattr(struct inode *inode)
3060 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
3061 inode->i_flags |= S_NOSEC;
3064 static inline bool is_root_inode(struct inode *inode)
3066 return inode == inode->i_sb->s_root->d_inode;
3069 static inline bool dir_emit(struct dir_context *ctx,
3070 const char *name, int namelen,
3071 u64 ino, unsigned type)
3073 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3075 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3077 return ctx->actor(ctx, ".", 1, ctx->pos,
3078 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3080 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3082 return ctx->actor(ctx, "..", 2, ctx->pos,
3083 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3085 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3087 if (ctx->pos == 0) {
3088 if (!dir_emit_dot(file, ctx))
3092 if (ctx->pos == 1) {
3093 if (!dir_emit_dotdot(file, ctx))
3099 static inline bool dir_relax(struct inode *inode)
3101 inode_unlock(inode);
3103 return !IS_DEADDIR(inode);
3106 extern bool path_noexec(const struct path *path);
3107 extern void inode_nohighmem(struct inode *inode);
3109 #endif /* _LINUX_FS_H */