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>
35 #include <asm/byteorder.h>
36 #include <uapi/linux/fs.h>
38 struct backing_dev_info;
40 struct export_operations;
45 struct pipe_inode_info;
46 struct poll_table_struct;
48 struct vm_area_struct;
51 struct swap_info_struct;
53 struct workqueue_struct;
56 extern void __init inode_init(void);
57 extern void __init inode_init_early(void);
58 extern void __init files_init(void);
59 extern void __init files_maxfiles_init(void);
61 extern struct files_stat_struct files_stat;
62 extern unsigned long get_max_files(void);
63 extern int sysctl_nr_open;
64 extern struct inodes_stat_t inodes_stat;
65 extern int leases_enable, lease_break_time;
66 extern int sysctl_protected_symlinks;
67 extern int sysctl_protected_hardlinks;
70 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
71 struct buffer_head *bh_result, int create);
72 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
73 ssize_t bytes, void *private);
74 typedef void (dax_iodone_t)(struct buffer_head *bh_map, int uptodate);
76 #define MAY_EXEC 0x00000001
77 #define MAY_WRITE 0x00000002
78 #define MAY_READ 0x00000004
79 #define MAY_APPEND 0x00000008
80 #define MAY_ACCESS 0x00000010
81 #define MAY_OPEN 0x00000020
82 #define MAY_CHDIR 0x00000040
83 /* called from RCU mode, don't block */
84 #define MAY_NOT_BLOCK 0x00000080
87 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
88 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
91 /* file is open for reading */
92 #define FMODE_READ ((__force fmode_t)0x1)
93 /* file is open for writing */
94 #define FMODE_WRITE ((__force fmode_t)0x2)
95 /* file is seekable */
96 #define FMODE_LSEEK ((__force fmode_t)0x4)
97 /* file can be accessed using pread */
98 #define FMODE_PREAD ((__force fmode_t)0x8)
99 /* file can be accessed using pwrite */
100 #define FMODE_PWRITE ((__force fmode_t)0x10)
101 /* File is opened for execution with sys_execve / sys_uselib */
102 #define FMODE_EXEC ((__force fmode_t)0x20)
103 /* File is opened with O_NDELAY (only set for block devices) */
104 #define FMODE_NDELAY ((__force fmode_t)0x40)
105 /* File is opened with O_EXCL (only set for block devices) */
106 #define FMODE_EXCL ((__force fmode_t)0x80)
107 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
108 (specialy hack for floppy.c) */
109 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
110 /* 32bit hashes as llseek() offset (for directories) */
111 #define FMODE_32BITHASH ((__force fmode_t)0x200)
112 /* 64bit hashes as llseek() offset (for directories) */
113 #define FMODE_64BITHASH ((__force fmode_t)0x400)
116 * Don't update ctime and mtime.
118 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
119 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
121 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
123 /* Expect random access pattern */
124 #define FMODE_RANDOM ((__force fmode_t)0x1000)
126 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
127 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
129 /* File is opened with O_PATH; almost nothing can be done with it */
130 #define FMODE_PATH ((__force fmode_t)0x4000)
132 /* File needs atomic accesses to f_pos */
133 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
134 /* Write access to underlying fs */
135 #define FMODE_WRITER ((__force fmode_t)0x10000)
136 /* Has read method(s) */
137 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
138 /* Has write method(s) */
139 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
141 /* File was opened by fanotify and shouldn't generate fanotify events */
142 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
145 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
146 * that indicates that they should check the contents of the iovec are
147 * valid, but not check the memory that the iovec elements
150 #define CHECK_IOVEC_ONLY -1
153 * The below are the various read and write types that we support. Some of
154 * them include behavioral modifiers that send information down to the
155 * block layer and IO scheduler. Terminology:
157 * The block layer uses device plugging to defer IO a little bit, in
158 * the hope that we will see more IO very shortly. This increases
159 * coalescing of adjacent IO and thus reduces the number of IOs we
160 * have to send to the device. It also allows for better queuing,
161 * if the IO isn't mergeable. If the caller is going to be waiting
162 * for the IO, then he must ensure that the device is unplugged so
163 * that the IO is dispatched to the driver.
165 * All IO is handled async in Linux. This is fine for background
166 * writes, but for reads or writes that someone waits for completion
167 * on, we want to notify the block layer and IO scheduler so that they
168 * know about it. That allows them to make better scheduling
169 * decisions. So when the below references 'sync' and 'async', it
170 * is referencing this priority hint.
172 * With that in mind, the available types are:
174 * READ A normal read operation. Device will be plugged.
175 * READ_SYNC A synchronous read. Device is not plugged, caller can
176 * immediately wait on this read without caring about
178 * READA Used for read-ahead operations. Lower priority, and the
179 * block layer could (in theory) choose to ignore this
180 * request if it runs into resource problems.
181 * WRITE A normal async write. Device will be plugged.
182 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
183 * the hint that someone will be waiting on this IO
184 * shortly. The write equivalent of READ_SYNC.
185 * WRITE_ODIRECT Special case write for O_DIRECT only.
186 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
187 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
188 * non-volatile media on completion.
189 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
190 * by a cache flush and data is guaranteed to be on
191 * non-volatile media on completion.
194 #define RW_MASK REQ_WRITE
195 #define RWA_MASK REQ_RAHEAD
198 #define WRITE RW_MASK
199 #define READA RWA_MASK
201 #define READ_SYNC (READ | REQ_SYNC)
202 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
203 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
204 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
205 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
206 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
209 * Attribute flags. These should be or-ed together to figure out what
212 #define ATTR_MODE (1 << 0)
213 #define ATTR_UID (1 << 1)
214 #define ATTR_GID (1 << 2)
215 #define ATTR_SIZE (1 << 3)
216 #define ATTR_ATIME (1 << 4)
217 #define ATTR_MTIME (1 << 5)
218 #define ATTR_CTIME (1 << 6)
219 #define ATTR_ATIME_SET (1 << 7)
220 #define ATTR_MTIME_SET (1 << 8)
221 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
222 #define ATTR_ATTR_FLAG (1 << 10)
223 #define ATTR_KILL_SUID (1 << 11)
224 #define ATTR_KILL_SGID (1 << 12)
225 #define ATTR_FILE (1 << 13)
226 #define ATTR_KILL_PRIV (1 << 14)
227 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
228 #define ATTR_TIMES_SET (1 << 16)
229 #define ATTR_TOUCH (1 << 17)
232 * Whiteout is represented by a char device. The following constants define the
233 * mode and device number to use.
235 #define WHITEOUT_MODE 0
236 #define WHITEOUT_DEV 0
239 * This is the Inode Attributes structure, used for notify_change(). It
240 * uses the above definitions as flags, to know which values have changed.
241 * Also, in this manner, a Filesystem can look at only the values it cares
242 * about. Basically, these are the attributes that the VFS layer can
243 * request to change from the FS layer.
245 * Derek Atkins <warlord@MIT.EDU> 94-10-20
248 unsigned int ia_valid;
253 struct timespec ia_atime;
254 struct timespec ia_mtime;
255 struct timespec ia_ctime;
258 * Not an attribute, but an auxiliary info for filesystems wanting to
259 * implement an ftruncate() like method. NOTE: filesystem should
260 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
262 struct file *ia_file;
266 * Includes for diskquotas.
268 #include <linux/quota.h>
271 * Maximum number of layers of fs stack. Needs to be limited to
272 * prevent kernel stack overflow
274 #define FILESYSTEM_MAX_STACK_DEPTH 2
277 * enum positive_aop_returns - aop return codes with specific semantics
279 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
280 * completed, that the page is still locked, and
281 * should be considered active. The VM uses this hint
282 * to return the page to the active list -- it won't
283 * be a candidate for writeback again in the near
284 * future. Other callers must be careful to unlock
285 * the page if they get this return. Returned by
288 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
289 * unlocked it and the page might have been truncated.
290 * The caller should back up to acquiring a new page and
291 * trying again. The aop will be taking reasonable
292 * precautions not to livelock. If the caller held a page
293 * reference, it should drop it before retrying. Returned
296 * address_space_operation functions return these large constants to indicate
297 * special semantics to the caller. These are much larger than the bytes in a
298 * page to allow for functions that return the number of bytes operated on in a
302 enum positive_aop_returns {
303 AOP_WRITEPAGE_ACTIVATE = 0x80000,
304 AOP_TRUNCATED_PAGE = 0x80001,
307 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
308 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
309 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
310 * helper code (eg buffer layer)
311 * to clear GFP_FS from alloc */
314 * oh the beauties of C type declarations.
317 struct address_space;
318 struct writeback_control;
320 #define IOCB_EVENTFD (1 << 0)
321 #define IOCB_APPEND (1 << 1)
322 #define IOCB_DIRECT (1 << 2)
325 struct file *ki_filp;
327 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
332 static inline bool is_sync_kiocb(struct kiocb *kiocb)
334 return kiocb->ki_complete == NULL;
337 static inline int iocb_flags(struct file *file);
339 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
341 *kiocb = (struct kiocb) {
343 .ki_flags = iocb_flags(filp),
348 * "descriptor" for what we're up to with a read.
349 * This allows us to use the same read code yet
350 * have multiple different users of the data that
351 * we read from a file.
353 * The simplest case just copies the data to user
366 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
367 unsigned long, unsigned long);
369 struct address_space_operations {
370 int (*writepage)(struct page *page, struct writeback_control *wbc);
371 int (*readpage)(struct file *, struct page *);
373 /* Write back some dirty pages from this mapping. */
374 int (*writepages)(struct address_space *, struct writeback_control *);
376 /* Set a page dirty. Return true if this dirtied it */
377 int (*set_page_dirty)(struct page *page);
379 int (*readpages)(struct file *filp, struct address_space *mapping,
380 struct list_head *pages, unsigned nr_pages);
382 int (*write_begin)(struct file *, struct address_space *mapping,
383 loff_t pos, unsigned len, unsigned flags,
384 struct page **pagep, void **fsdata);
385 int (*write_end)(struct file *, struct address_space *mapping,
386 loff_t pos, unsigned len, unsigned copied,
387 struct page *page, void *fsdata);
389 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
390 sector_t (*bmap)(struct address_space *, sector_t);
391 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
392 int (*releasepage) (struct page *, gfp_t);
393 void (*freepage)(struct page *);
394 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter, loff_t offset);
396 * migrate the contents of a page to the specified target. If
397 * migrate_mode is MIGRATE_ASYNC, it must not block.
399 int (*migratepage) (struct address_space *,
400 struct page *, struct page *, enum migrate_mode);
401 int (*launder_page) (struct page *);
402 int (*is_partially_uptodate) (struct page *, unsigned long,
404 void (*is_dirty_writeback) (struct page *, bool *, bool *);
405 int (*error_remove_page)(struct address_space *, struct page *);
407 /* swapfile support */
408 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
410 void (*swap_deactivate)(struct file *file);
413 extern const struct address_space_operations empty_aops;
416 * pagecache_write_begin/pagecache_write_end must be used by general code
417 * to write into the pagecache.
419 int pagecache_write_begin(struct file *, struct address_space *mapping,
420 loff_t pos, unsigned len, unsigned flags,
421 struct page **pagep, void **fsdata);
423 int pagecache_write_end(struct file *, struct address_space *mapping,
424 loff_t pos, unsigned len, unsigned copied,
425 struct page *page, void *fsdata);
427 struct address_space {
428 struct inode *host; /* owner: inode, block_device */
429 struct radix_tree_root page_tree; /* radix tree of all pages */
430 spinlock_t tree_lock; /* and lock protecting it */
431 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
432 struct rb_root i_mmap; /* tree of private and shared mappings */
433 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
434 /* Protected by tree_lock together with the radix tree */
435 unsigned long nrpages; /* number of total pages */
436 unsigned long nrshadows; /* number of shadow entries */
437 pgoff_t writeback_index;/* writeback starts here */
438 const struct address_space_operations *a_ops; /* methods */
439 unsigned long flags; /* error bits/gfp mask */
440 spinlock_t private_lock; /* for use by the address_space */
441 struct list_head private_list; /* ditto */
442 void *private_data; /* ditto */
443 } __attribute__((aligned(sizeof(long))));
445 * On most architectures that alignment is already the case; but
446 * must be enforced here for CRIS, to let the least significant bit
447 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
449 struct request_queue;
451 struct block_device {
452 dev_t bd_dev; /* not a kdev_t - it's a search key */
454 struct inode * bd_inode; /* will die */
455 struct super_block * bd_super;
456 struct mutex bd_mutex; /* open/close mutex */
457 struct list_head bd_inodes;
461 bool bd_write_holder;
463 struct list_head bd_holder_disks;
465 struct block_device * bd_contains;
466 unsigned bd_block_size;
467 struct hd_struct * bd_part;
468 /* number of times partitions within this device have been opened. */
469 unsigned bd_part_count;
471 struct gendisk * bd_disk;
472 struct request_queue * bd_queue;
473 struct list_head bd_list;
475 * Private data. You must have bd_claim'ed the block_device
476 * to use this. NOTE: bd_claim allows an owner to claim
477 * the same device multiple times, the owner must take special
478 * care to not mess up bd_private for that case.
480 unsigned long bd_private;
482 /* The counter of freeze processes */
483 int bd_fsfreeze_count;
484 /* Mutex for freeze */
485 struct mutex bd_fsfreeze_mutex;
489 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
492 #define PAGECACHE_TAG_DIRTY 0
493 #define PAGECACHE_TAG_WRITEBACK 1
494 #define PAGECACHE_TAG_TOWRITE 2
496 int mapping_tagged(struct address_space *mapping, int tag);
498 static inline void i_mmap_lock_write(struct address_space *mapping)
500 down_write(&mapping->i_mmap_rwsem);
503 static inline void i_mmap_unlock_write(struct address_space *mapping)
505 up_write(&mapping->i_mmap_rwsem);
508 static inline void i_mmap_lock_read(struct address_space *mapping)
510 down_read(&mapping->i_mmap_rwsem);
513 static inline void i_mmap_unlock_read(struct address_space *mapping)
515 up_read(&mapping->i_mmap_rwsem);
519 * Might pages of this file be mapped into userspace?
521 static inline int mapping_mapped(struct address_space *mapping)
523 return !RB_EMPTY_ROOT(&mapping->i_mmap);
527 * Might pages of this file have been modified in userspace?
528 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
529 * marks vma as VM_SHARED if it is shared, and the file was opened for
530 * writing i.e. vma may be mprotected writable even if now readonly.
532 * If i_mmap_writable is negative, no new writable mappings are allowed. You
533 * can only deny writable mappings, if none exists right now.
535 static inline int mapping_writably_mapped(struct address_space *mapping)
537 return atomic_read(&mapping->i_mmap_writable) > 0;
540 static inline int mapping_map_writable(struct address_space *mapping)
542 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
546 static inline void mapping_unmap_writable(struct address_space *mapping)
548 atomic_dec(&mapping->i_mmap_writable);
551 static inline int mapping_deny_writable(struct address_space *mapping)
553 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
557 static inline void mapping_allow_writable(struct address_space *mapping)
559 atomic_inc(&mapping->i_mmap_writable);
563 * Use sequence counter to get consistent i_size on 32-bit processors.
565 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
566 #include <linux/seqlock.h>
567 #define __NEED_I_SIZE_ORDERED
568 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
570 #define i_size_ordered_init(inode) do { } while (0)
574 #define ACL_NOT_CACHED ((void *)(-1))
576 #define IOP_FASTPERM 0x0001
577 #define IOP_LOOKUP 0x0002
578 #define IOP_NOFOLLOW 0x0004
581 * Keep mostly read-only and often accessed (especially for
582 * the RCU path lookup and 'stat' data) fields at the beginning
583 * of the 'struct inode'
587 unsigned short i_opflags;
590 unsigned int i_flags;
592 #ifdef CONFIG_FS_POSIX_ACL
593 struct posix_acl *i_acl;
594 struct posix_acl *i_default_acl;
597 const struct inode_operations *i_op;
598 struct super_block *i_sb;
599 struct address_space *i_mapping;
601 #ifdef CONFIG_SECURITY
605 /* Stat data, not accessed from path walking */
608 * Filesystems may only read i_nlink directly. They shall use the
609 * following functions for modification:
611 * (set|clear|inc|drop)_nlink
612 * inode_(inc|dec)_link_count
615 const unsigned int i_nlink;
616 unsigned int __i_nlink;
620 struct timespec i_atime;
621 struct timespec i_mtime;
622 struct timespec i_ctime;
623 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
624 unsigned short i_bytes;
625 unsigned int i_blkbits;
628 #ifdef __NEED_I_SIZE_ORDERED
629 seqcount_t i_size_seqcount;
633 unsigned long i_state;
634 struct mutex i_mutex;
636 unsigned long dirtied_when; /* jiffies of first dirtying */
637 unsigned long dirtied_time_when;
639 struct hlist_node i_hash;
640 struct list_head i_io_list; /* backing dev IO list */
641 #ifdef CONFIG_CGROUP_WRITEBACK
642 struct bdi_writeback *i_wb; /* the associated cgroup wb */
644 /* foreign inode detection, see wbc_detach_inode() */
646 u16 i_wb_frn_avg_time;
647 u16 i_wb_frn_history;
649 struct list_head i_lru; /* inode LRU list */
650 struct list_head i_sb_list;
652 struct hlist_head i_dentry;
653 struct rcu_head i_rcu;
657 atomic_t i_dio_count;
658 atomic_t i_writecount;
660 atomic_t i_readcount; /* struct files open RO */
662 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
663 struct file_lock_context *i_flctx;
664 struct address_space i_data;
665 struct list_head i_devices;
667 struct pipe_inode_info *i_pipe;
668 struct block_device *i_bdev;
675 #ifdef CONFIG_FSNOTIFY
676 __u32 i_fsnotify_mask; /* all events this inode cares about */
677 struct hlist_head i_fsnotify_marks;
680 void *i_private; /* fs or device private pointer */
683 static inline int inode_unhashed(struct inode *inode)
685 return hlist_unhashed(&inode->i_hash);
689 * inode->i_mutex nesting subclasses for the lock validator:
691 * 0: the object of the current VFS operation
695 * 4: second non-directory
696 * 5: second parent (when locking independent directories in rename)
698 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
699 * non-directories at once.
701 * The locking order between these classes is
702 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
704 enum inode_i_mutex_lock_class
714 static inline void inode_lock(struct inode *inode)
716 mutex_lock(&inode->i_mutex);
719 static inline void inode_unlock(struct inode *inode)
721 mutex_unlock(&inode->i_mutex);
724 static inline int inode_trylock(struct inode *inode)
726 return mutex_trylock(&inode->i_mutex);
729 static inline int inode_is_locked(struct inode *inode)
731 return mutex_is_locked(&inode->i_mutex);
734 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
736 mutex_lock_nested(&inode->i_mutex, subclass);
739 void lock_two_nondirectories(struct inode *, struct inode*);
740 void unlock_two_nondirectories(struct inode *, struct inode*);
743 * NOTE: in a 32bit arch with a preemptable kernel and
744 * an UP compile the i_size_read/write must be atomic
745 * with respect to the local cpu (unlike with preempt disabled),
746 * but they don't need to be atomic with respect to other cpus like in
747 * true SMP (so they need either to either locally disable irq around
748 * the read or for example on x86 they can be still implemented as a
749 * cmpxchg8b without the need of the lock prefix). For SMP compiles
750 * and 64bit archs it makes no difference if preempt is enabled or not.
752 static inline loff_t i_size_read(const struct inode *inode)
754 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
759 seq = read_seqcount_begin(&inode->i_size_seqcount);
760 i_size = inode->i_size;
761 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
763 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
767 i_size = inode->i_size;
771 return inode->i_size;
776 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
777 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
778 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
780 static inline void i_size_write(struct inode *inode, loff_t i_size)
782 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
784 write_seqcount_begin(&inode->i_size_seqcount);
785 inode->i_size = i_size;
786 write_seqcount_end(&inode->i_size_seqcount);
788 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
790 inode->i_size = i_size;
793 inode->i_size = i_size;
797 /* Helper functions so that in most cases filesystems will
798 * not need to deal directly with kuid_t and kgid_t and can
799 * instead deal with the raw numeric values that are stored
802 static inline uid_t i_uid_read(const struct inode *inode)
804 return from_kuid(&init_user_ns, inode->i_uid);
807 static inline gid_t i_gid_read(const struct inode *inode)
809 return from_kgid(&init_user_ns, inode->i_gid);
812 static inline void i_uid_write(struct inode *inode, uid_t uid)
814 inode->i_uid = make_kuid(&init_user_ns, uid);
817 static inline void i_gid_write(struct inode *inode, gid_t gid)
819 inode->i_gid = make_kgid(&init_user_ns, gid);
822 static inline unsigned iminor(const struct inode *inode)
824 return MINOR(inode->i_rdev);
827 static inline unsigned imajor(const struct inode *inode)
829 return MAJOR(inode->i_rdev);
832 extern struct block_device *I_BDEV(struct inode *inode);
835 rwlock_t lock; /* protects pid, uid, euid fields */
836 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
837 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
838 kuid_t uid, euid; /* uid/euid of process setting the owner */
839 int signum; /* posix.1b rt signal to be delivered on IO */
843 * Track a single file's readahead state
845 struct file_ra_state {
846 pgoff_t start; /* where readahead started */
847 unsigned int size; /* # of readahead pages */
848 unsigned int async_size; /* do asynchronous readahead when
849 there are only # of pages ahead */
851 unsigned int ra_pages; /* Maximum readahead window */
852 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
853 loff_t prev_pos; /* Cache last read() position */
857 * Check if @index falls in the readahead windows.
859 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
861 return (index >= ra->start &&
862 index < ra->start + ra->size);
867 struct llist_node fu_llist;
868 struct rcu_head fu_rcuhead;
871 struct inode *f_inode; /* cached value */
872 const struct file_operations *f_op;
875 * Protects f_ep_links, f_flags.
876 * Must not be taken from IRQ context.
879 atomic_long_t f_count;
880 unsigned int f_flags;
882 struct mutex f_pos_lock;
884 struct fown_struct f_owner;
885 const struct cred *f_cred;
886 struct file_ra_state f_ra;
889 #ifdef CONFIG_SECURITY
892 /* needed for tty driver, and maybe others */
896 /* Used by fs/eventpoll.c to link all the hooks to this file */
897 struct list_head f_ep_links;
898 struct list_head f_tfile_llink;
899 #endif /* #ifdef CONFIG_EPOLL */
900 struct address_space *f_mapping;
901 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
906 /* file identifier */
907 unsigned char f_handle[0];
910 static inline struct file *get_file(struct file *f)
912 atomic_long_inc(&f->f_count);
915 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
916 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
917 #define file_count(x) atomic_long_read(&(x)->f_count)
919 #define MAX_NON_LFS ((1UL<<31) - 1)
921 /* Page cache limit. The filesystems should put that into their s_maxbytes
922 limits, otherwise bad things can happen in VM. */
923 #if BITS_PER_LONG==32
924 #define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
925 #elif BITS_PER_LONG==64
926 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
931 #define FL_DELEG 4 /* NFSv4 delegation */
932 #define FL_ACCESS 8 /* not trying to lock, just looking */
933 #define FL_EXISTS 16 /* when unlocking, test for existence */
934 #define FL_LEASE 32 /* lease held on this file */
935 #define FL_CLOSE 64 /* unlock on close */
936 #define FL_SLEEP 128 /* A blocking lock */
937 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
938 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
939 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
940 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
943 * Special return value from posix_lock_file() and vfs_lock_file() for
944 * asynchronous locking.
946 #define FILE_LOCK_DEFERRED 1
948 /* legacy typedef, should eventually be removed */
949 typedef void *fl_owner_t;
953 struct file_lock_operations {
954 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
955 void (*fl_release_private)(struct file_lock *);
958 struct lock_manager_operations {
959 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
960 unsigned long (*lm_owner_key)(struct file_lock *);
961 fl_owner_t (*lm_get_owner)(fl_owner_t);
962 void (*lm_put_owner)(fl_owner_t);
963 void (*lm_notify)(struct file_lock *); /* unblock callback */
964 int (*lm_grant)(struct file_lock *, int);
965 bool (*lm_break)(struct file_lock *);
966 int (*lm_change)(struct file_lock *, int, struct list_head *);
967 void (*lm_setup)(struct file_lock *, void **);
970 struct lock_manager {
971 struct list_head list;
973 * NFSv4 and up also want opens blocked during the grace period;
980 void locks_start_grace(struct net *, struct lock_manager *);
981 void locks_end_grace(struct lock_manager *);
982 int locks_in_grace(struct net *);
983 int opens_in_grace(struct net *);
985 /* that will die - we need it for nfs_lock_info */
986 #include <linux/nfs_fs_i.h>
989 * struct file_lock represents a generic "file lock". It's used to represent
990 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
991 * note that the same struct is used to represent both a request for a lock and
992 * the lock itself, but the same object is never used for both.
994 * FIXME: should we create a separate "struct lock_request" to help distinguish
997 * The varous i_flctx lists are ordered by:
1000 * 2) lock range start
1003 * Obviously, the last two criteria only matter for POSIX locks.
1006 struct file_lock *fl_next; /* singly linked list for this inode */
1007 struct list_head fl_list; /* link into file_lock_context */
1008 struct hlist_node fl_link; /* node in global lists */
1009 struct list_head fl_block; /* circular list of blocked processes */
1010 fl_owner_t fl_owner;
1011 unsigned int fl_flags;
1012 unsigned char fl_type;
1013 unsigned int fl_pid;
1014 int fl_link_cpu; /* what cpu's list is this on? */
1015 struct pid *fl_nspid;
1016 wait_queue_head_t fl_wait;
1017 struct file *fl_file;
1021 struct fasync_struct * fl_fasync; /* for lease break notifications */
1022 /* for lease breaks: */
1023 unsigned long fl_break_time;
1024 unsigned long fl_downgrade_time;
1026 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1027 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1029 struct nfs_lock_info nfs_fl;
1030 struct nfs4_lock_info nfs4_fl;
1032 struct list_head link; /* link in AFS vnode's pending_locks list */
1033 int state; /* state of grant or error if -ve */
1038 struct file_lock_context {
1039 spinlock_t flc_lock;
1040 struct list_head flc_flock;
1041 struct list_head flc_posix;
1042 struct list_head flc_lease;
1045 /* The following constant reflects the upper bound of the file/locking space */
1047 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1048 #define OFFSET_MAX INT_LIMIT(loff_t)
1049 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1052 #include <linux/fcntl.h>
1054 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1056 #ifdef CONFIG_FILE_LOCKING
1057 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
1058 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1059 struct flock __user *);
1061 #if BITS_PER_LONG == 32
1062 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
1063 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1064 struct flock64 __user *);
1067 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1068 extern int fcntl_getlease(struct file *filp);
1071 void locks_free_lock_context(struct file_lock_context *ctx);
1072 void locks_free_lock(struct file_lock *fl);
1073 extern void locks_init_lock(struct file_lock *);
1074 extern struct file_lock * locks_alloc_lock(void);
1075 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1076 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1077 extern void locks_remove_posix(struct file *, fl_owner_t);
1078 extern void locks_remove_file(struct file *);
1079 extern void locks_release_private(struct file_lock *);
1080 extern void posix_test_lock(struct file *, struct file_lock *);
1081 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1082 extern int posix_unblock_lock(struct file_lock *);
1083 extern int vfs_test_lock(struct file *, struct file_lock *);
1084 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1085 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1086 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1087 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1088 extern void lease_get_mtime(struct inode *, struct timespec *time);
1089 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1090 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1091 extern int lease_modify(struct file_lock *, int, struct list_head *);
1092 struct files_struct;
1093 extern void show_fd_locks(struct seq_file *f,
1094 struct file *filp, struct files_struct *files);
1095 #else /* !CONFIG_FILE_LOCKING */
1096 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1097 struct flock __user *user)
1102 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1103 unsigned int cmd, struct flock __user *user)
1108 #if BITS_PER_LONG == 32
1109 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1110 struct flock64 __user *user)
1115 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1116 unsigned int cmd, struct flock64 __user *user)
1121 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1126 static inline int fcntl_getlease(struct file *filp)
1132 locks_free_lock_context(struct file_lock_context *ctx)
1136 static inline void locks_init_lock(struct file_lock *fl)
1141 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1146 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1151 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1156 static inline void locks_remove_file(struct file *filp)
1161 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1166 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1167 struct file_lock *conflock)
1172 static inline int posix_unblock_lock(struct file_lock *waiter)
1177 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1182 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1183 struct file_lock *fl, struct file_lock *conf)
1188 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1193 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1198 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1203 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1208 static inline int generic_setlease(struct file *filp, long arg,
1209 struct file_lock **flp, void **priv)
1214 static inline int vfs_setlease(struct file *filp, long arg,
1215 struct file_lock **lease, void **priv)
1220 static inline int lease_modify(struct file_lock *fl, int arg,
1221 struct list_head *dispose)
1226 struct files_struct;
1227 static inline void show_fd_locks(struct seq_file *f,
1228 struct file *filp, struct files_struct *files) {}
1229 #endif /* !CONFIG_FILE_LOCKING */
1231 static inline struct inode *file_inode(const struct file *f)
1236 static inline struct dentry *file_dentry(const struct file *file)
1238 struct dentry *dentry = file->f_path.dentry;
1240 if (unlikely(dentry->d_flags & DCACHE_OP_REAL))
1241 return dentry->d_op->d_real(dentry, file_inode(file));
1246 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1248 return locks_lock_inode_wait(file_inode(filp), fl);
1251 struct fasync_struct {
1255 struct fasync_struct *fa_next; /* singly linked list */
1256 struct file *fa_file;
1257 struct rcu_head fa_rcu;
1260 #define FASYNC_MAGIC 0x4601
1262 /* SMP safe fasync helpers: */
1263 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1264 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1265 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1266 extern struct fasync_struct *fasync_alloc(void);
1267 extern void fasync_free(struct fasync_struct *);
1269 /* can be called from interrupts */
1270 extern void kill_fasync(struct fasync_struct **, int, int);
1272 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1273 extern void f_setown(struct file *filp, unsigned long arg, int force);
1274 extern void f_delown(struct file *filp);
1275 extern pid_t f_getown(struct file *filp);
1276 extern int send_sigurg(struct fown_struct *fown);
1284 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1285 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1286 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1287 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1288 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1291 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1292 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1294 /* Possible states of 'frozen' field */
1296 SB_UNFROZEN = 0, /* FS is unfrozen */
1297 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1298 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1299 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1300 * internal threads if needed) */
1301 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1304 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1307 int frozen; /* Is sb frozen? */
1308 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1309 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1312 struct super_block {
1313 struct list_head s_list; /* Keep this first */
1314 dev_t s_dev; /* search index; _not_ kdev_t */
1315 unsigned char s_blocksize_bits;
1316 unsigned long s_blocksize;
1317 loff_t s_maxbytes; /* Max file size */
1318 struct file_system_type *s_type;
1319 const struct super_operations *s_op;
1320 const struct dquot_operations *dq_op;
1321 const struct quotactl_ops *s_qcop;
1322 const struct export_operations *s_export_op;
1323 unsigned long s_flags;
1324 unsigned long s_iflags; /* internal SB_I_* flags */
1325 unsigned long s_magic;
1326 struct dentry *s_root;
1327 struct rw_semaphore s_umount;
1330 #ifdef CONFIG_SECURITY
1333 const struct xattr_handler **s_xattr;
1335 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1336 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1337 struct block_device *s_bdev;
1338 struct backing_dev_info *s_bdi;
1339 struct mtd_info *s_mtd;
1340 struct hlist_node s_instances;
1341 unsigned int s_quota_types; /* Bitmask of supported quota types */
1342 struct quota_info s_dquot; /* Diskquota specific options */
1344 struct sb_writers s_writers;
1346 char s_id[32]; /* Informational name */
1347 u8 s_uuid[16]; /* UUID */
1349 void *s_fs_info; /* Filesystem private info */
1350 unsigned int s_max_links;
1353 /* Granularity of c/m/atime in ns.
1354 Cannot be worse than a second */
1358 * The next field is for VFS *only*. No filesystems have any business
1359 * even looking at it. You had been warned.
1361 struct mutex s_vfs_rename_mutex; /* Kludge */
1364 * Filesystem subtype. If non-empty the filesystem type field
1365 * in /proc/mounts will be "type.subtype"
1370 * Saved mount options for lazy filesystems using
1371 * generic_show_options()
1373 char __rcu *s_options;
1374 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1377 * Saved pool identifier for cleancache (-1 means none)
1379 int cleancache_poolid;
1381 struct shrinker s_shrink; /* per-sb shrinker handle */
1383 /* Number of inodes with nlink == 0 but still referenced */
1384 atomic_long_t s_remove_count;
1386 /* Being remounted read-only */
1387 int s_readonly_remount;
1389 /* AIO completions deferred from interrupt context */
1390 struct workqueue_struct *s_dio_done_wq;
1391 struct hlist_head s_pins;
1394 * Keep the lru lists last in the structure so they always sit on their
1395 * own individual cachelines.
1397 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1398 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1399 struct rcu_head rcu;
1400 struct work_struct destroy_work;
1402 struct mutex s_sync_lock; /* sync serialisation lock */
1405 * Indicates how deep in a filesystem stack this SB is
1409 /* s_inode_list_lock protects s_inodes */
1410 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1411 struct list_head s_inodes; /* all inodes */
1414 extern struct timespec current_fs_time(struct super_block *sb);
1417 * Snapshotting support.
1420 void __sb_end_write(struct super_block *sb, int level);
1421 int __sb_start_write(struct super_block *sb, int level, bool wait);
1423 #define __sb_writers_acquired(sb, lev) \
1424 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1425 #define __sb_writers_release(sb, lev) \
1426 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1429 * sb_end_write - drop write access to a superblock
1430 * @sb: the super we wrote to
1432 * Decrement number of writers to the filesystem. Wake up possible waiters
1433 * wanting to freeze the filesystem.
1435 static inline void sb_end_write(struct super_block *sb)
1437 __sb_end_write(sb, SB_FREEZE_WRITE);
1441 * sb_end_pagefault - drop write access to a superblock from a page fault
1442 * @sb: the super we wrote to
1444 * Decrement number of processes handling write page fault to the filesystem.
1445 * Wake up possible waiters wanting to freeze the filesystem.
1447 static inline void sb_end_pagefault(struct super_block *sb)
1449 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1453 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1454 * @sb: the super we wrote to
1456 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1457 * waiters wanting to freeze the filesystem.
1459 static inline void sb_end_intwrite(struct super_block *sb)
1461 __sb_end_write(sb, SB_FREEZE_FS);
1465 * sb_start_write - get write access to a superblock
1466 * @sb: the super we write to
1468 * When a process wants to write data or metadata to a file system (i.e. dirty
1469 * a page or an inode), it should embed the operation in a sb_start_write() -
1470 * sb_end_write() pair to get exclusion against file system freezing. This
1471 * function increments number of writers preventing freezing. If the file
1472 * system is already frozen, the function waits until the file system is
1475 * Since freeze protection behaves as a lock, users have to preserve
1476 * ordering of freeze protection and other filesystem locks. Generally,
1477 * freeze protection should be the outermost lock. In particular, we have:
1480 * -> i_mutex (write path, truncate, directory ops, ...)
1481 * -> s_umount (freeze_super, thaw_super)
1483 static inline void sb_start_write(struct super_block *sb)
1485 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1488 static inline int sb_start_write_trylock(struct super_block *sb)
1490 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1494 * sb_start_pagefault - get write access to a superblock from a page fault
1495 * @sb: the super we write to
1497 * When a process starts handling write page fault, it should embed the
1498 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1499 * exclusion against file system freezing. This is needed since the page fault
1500 * is going to dirty a page. This function increments number of running page
1501 * faults preventing freezing. If the file system is already frozen, the
1502 * function waits until the file system is thawed.
1504 * Since page fault freeze protection behaves as a lock, users have to preserve
1505 * ordering of freeze protection and other filesystem locks. It is advised to
1506 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1507 * handling code implies lock dependency:
1510 * -> sb_start_pagefault
1512 static inline void sb_start_pagefault(struct super_block *sb)
1514 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1518 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1519 * @sb: the super we write to
1521 * This is the third level of protection against filesystem freezing. It is
1522 * free for use by a filesystem. The only requirement is that it must rank
1523 * below sb_start_pagefault.
1525 * For example filesystem can call sb_start_intwrite() when starting a
1526 * transaction which somewhat eases handling of freezing for internal sources
1527 * of filesystem changes (internal fs threads, discarding preallocation on file
1530 static inline void sb_start_intwrite(struct super_block *sb)
1532 __sb_start_write(sb, SB_FREEZE_FS, true);
1536 extern bool inode_owner_or_capable(const struct inode *inode);
1539 * VFS helper functions..
1541 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1542 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1543 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1544 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1545 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1546 extern int vfs_rmdir(struct inode *, struct dentry *);
1547 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1548 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1549 extern int vfs_whiteout(struct inode *, struct dentry *);
1552 * VFS dentry helper functions.
1554 extern void dentry_unhash(struct dentry *dentry);
1557 * VFS file helper functions.
1559 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1562 * VFS FS_IOC_FIEMAP helper definitions.
1564 struct fiemap_extent_info {
1565 unsigned int fi_flags; /* Flags as passed from user */
1566 unsigned int fi_extents_mapped; /* Number of mapped extents */
1567 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1568 struct fiemap_extent __user *fi_extents_start; /* Start of
1569 fiemap_extent array */
1571 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1572 u64 phys, u64 len, u32 flags);
1573 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1578 * NOTE! These match bits 12..15 of stat.st_mode
1579 * (ie "(i_mode >> 12) & 15").
1581 #define DT_UNKNOWN 0
1592 * This is the "filldir" function type, used by readdir() to let
1593 * the kernel specify what kind of dirent layout it wants to have.
1594 * This allows the kernel to read directories into kernel space or
1595 * to have different dirent layouts depending on the binary type.
1598 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1601 struct dir_context {
1602 const filldir_t actor;
1606 struct block_device_operations;
1608 /* These macros are for out of kernel modules to test that
1609 * the kernel supports the unlocked_ioctl and compat_ioctl
1610 * fields in struct file_operations. */
1611 #define HAVE_COMPAT_IOCTL 1
1612 #define HAVE_UNLOCKED_IOCTL 1
1615 * These flags let !MMU mmap() govern direct device mapping vs immediate
1616 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1618 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1619 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1620 * NOMMU_MAP_READ: Can be mapped for reading
1621 * NOMMU_MAP_WRITE: Can be mapped for writing
1622 * NOMMU_MAP_EXEC: Can be mapped for execution
1624 #define NOMMU_MAP_COPY 0x00000001
1625 #define NOMMU_MAP_DIRECT 0x00000008
1626 #define NOMMU_MAP_READ VM_MAYREAD
1627 #define NOMMU_MAP_WRITE VM_MAYWRITE
1628 #define NOMMU_MAP_EXEC VM_MAYEXEC
1630 #define NOMMU_VMFLAGS \
1631 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1636 struct file_operations {
1637 struct module *owner;
1638 loff_t (*llseek) (struct file *, loff_t, int);
1639 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1640 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1641 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1642 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1643 int (*iterate) (struct file *, struct dir_context *);
1644 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1645 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1646 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1647 int (*mmap) (struct file *, struct vm_area_struct *);
1648 int (*open) (struct inode *, struct file *);
1649 int (*flush) (struct file *, fl_owner_t id);
1650 int (*release) (struct inode *, struct file *);
1651 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1652 int (*aio_fsync) (struct kiocb *, int datasync);
1653 int (*fasync) (int, struct file *, int);
1654 int (*lock) (struct file *, int, struct file_lock *);
1655 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1656 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1657 int (*check_flags)(int);
1658 int (*flock) (struct file *, int, struct file_lock *);
1659 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1660 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1661 int (*setlease)(struct file *, long, struct file_lock **, void **);
1662 long (*fallocate)(struct file *file, int mode, loff_t offset,
1664 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1666 unsigned (*mmap_capabilities)(struct file *);
1670 struct inode_operations {
1671 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1672 const char * (*follow_link) (struct dentry *, void **);
1673 int (*permission) (struct inode *, int);
1674 struct posix_acl * (*get_acl)(struct inode *, int);
1676 int (*readlink) (struct dentry *, char __user *,int);
1677 void (*put_link) (struct inode *, void *);
1679 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1680 int (*link) (struct dentry *,struct inode *,struct dentry *);
1681 int (*unlink) (struct inode *,struct dentry *);
1682 int (*symlink) (struct inode *,struct dentry *,const char *);
1683 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1684 int (*rmdir) (struct inode *,struct dentry *);
1685 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1686 int (*rename) (struct inode *, struct dentry *,
1687 struct inode *, struct dentry *);
1688 int (*rename2) (struct inode *, struct dentry *,
1689 struct inode *, struct dentry *, unsigned int);
1690 int (*setattr) (struct dentry *, struct iattr *);
1691 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1692 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1693 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1694 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1695 int (*removexattr) (struct dentry *, const char *);
1696 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1698 int (*update_time)(struct inode *, struct timespec *, int);
1699 int (*atomic_open)(struct inode *, struct dentry *,
1700 struct file *, unsigned open_flag,
1701 umode_t create_mode, int *opened);
1702 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1703 int (*set_acl)(struct inode *, struct posix_acl *, int);
1704 } ____cacheline_aligned;
1706 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1707 unsigned long nr_segs, unsigned long fast_segs,
1708 struct iovec *fast_pointer,
1709 struct iovec **ret_pointer);
1711 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1712 extern ssize_t __vfs_write(struct file *, const char __user *, size_t, loff_t *);
1713 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1714 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1715 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1716 unsigned long, loff_t *);
1717 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1718 unsigned long, loff_t *);
1720 struct super_operations {
1721 struct inode *(*alloc_inode)(struct super_block *sb);
1722 void (*destroy_inode)(struct inode *);
1724 void (*dirty_inode) (struct inode *, int flags);
1725 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1726 int (*drop_inode) (struct inode *);
1727 void (*evict_inode) (struct inode *);
1728 void (*put_super) (struct super_block *);
1729 int (*sync_fs)(struct super_block *sb, int wait);
1730 int (*freeze_super) (struct super_block *);
1731 int (*freeze_fs) (struct super_block *);
1732 int (*thaw_super) (struct super_block *);
1733 int (*unfreeze_fs) (struct super_block *);
1734 int (*statfs) (struct dentry *, struct kstatfs *);
1735 int (*remount_fs) (struct super_block *, int *, char *);
1736 void (*umount_begin) (struct super_block *);
1738 int (*show_options)(struct seq_file *, struct dentry *);
1739 int (*show_devname)(struct seq_file *, struct dentry *);
1740 int (*show_path)(struct seq_file *, struct dentry *);
1741 int (*show_stats)(struct seq_file *, struct dentry *);
1743 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1744 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1745 struct dquot **(*get_dquots)(struct inode *);
1747 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1748 long (*nr_cached_objects)(struct super_block *,
1749 struct shrink_control *);
1750 long (*free_cached_objects)(struct super_block *,
1751 struct shrink_control *);
1755 * Inode flags - they have no relation to superblock flags now
1757 #define S_SYNC 1 /* Writes are synced at once */
1758 #define S_NOATIME 2 /* Do not update access times */
1759 #define S_APPEND 4 /* Append-only file */
1760 #define S_IMMUTABLE 8 /* Immutable file */
1761 #define S_DEAD 16 /* removed, but still open directory */
1762 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1763 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1764 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1765 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1766 #define S_PRIVATE 512 /* Inode is fs-internal */
1767 #define S_IMA 1024 /* Inode has an associated IMA struct */
1768 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1769 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1770 #ifdef CONFIG_FS_DAX
1771 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1773 #define S_DAX 0 /* Make all the DAX code disappear */
1777 * Note that nosuid etc flags are inode-specific: setting some file-system
1778 * flags just means all the inodes inherit those flags by default. It might be
1779 * possible to override it selectively if you really wanted to with some
1780 * ioctl() that is not currently implemented.
1782 * Exception: MS_RDONLY is always applied to the entire file system.
1784 * Unfortunately, it is possible to change a filesystems flags with it mounted
1785 * with files in use. This means that all of the inodes will not have their
1786 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1787 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1789 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1791 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1792 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1793 ((inode)->i_flags & S_SYNC))
1794 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1795 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1796 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1797 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1798 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1800 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1801 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1802 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1803 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1805 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1806 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1807 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1808 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1809 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1810 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1811 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1812 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1814 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1815 (inode)->i_rdev == WHITEOUT_DEV)
1818 * Inode state bits. Protected by inode->i_lock
1820 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1821 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1823 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1824 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1825 * various stages of removing an inode.
1827 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1829 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1830 * fdatasync(). i_atime is the usual cause.
1831 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1832 * these changes separately from I_DIRTY_SYNC so that we
1833 * don't have to write inode on fdatasync() when only
1834 * mtime has changed in it.
1835 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1836 * I_NEW Serves as both a mutex and completion notification.
1837 * New inodes set I_NEW. If two processes both create
1838 * the same inode, one of them will release its inode and
1839 * wait for I_NEW to be released before returning.
1840 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1841 * also cause waiting on I_NEW, without I_NEW actually
1842 * being set. find_inode() uses this to prevent returning
1843 * nearly-dead inodes.
1844 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1845 * is zero. I_FREEING must be set when I_WILL_FREE is
1847 * I_FREEING Set when inode is about to be freed but still has dirty
1848 * pages or buffers attached or the inode itself is still
1850 * I_CLEAR Added by clear_inode(). In this state the inode is
1851 * clean and can be destroyed. Inode keeps I_FREEING.
1853 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1854 * prohibited for many purposes. iget() must wait for
1855 * the inode to be completely released, then create it
1856 * anew. Other functions will just ignore such inodes,
1857 * if appropriate. I_NEW is used for waiting.
1859 * I_SYNC Writeback of inode is running. The bit is set during
1860 * data writeback, and cleared with a wakeup on the bit
1861 * address once it is done. The bit is also used to pin
1862 * the inode in memory for flusher thread.
1864 * I_REFERENCED Marks the inode as recently references on the LRU list.
1866 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1868 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1869 * synchronize competing switching instances and to tell
1870 * wb stat updates to grab mapping->tree_lock. See
1871 * inode_switch_wb_work_fn() for details.
1873 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1875 #define I_DIRTY_SYNC (1 << 0)
1876 #define I_DIRTY_DATASYNC (1 << 1)
1877 #define I_DIRTY_PAGES (1 << 2)
1879 #define I_NEW (1 << __I_NEW)
1880 #define I_WILL_FREE (1 << 4)
1881 #define I_FREEING (1 << 5)
1882 #define I_CLEAR (1 << 6)
1884 #define I_SYNC (1 << __I_SYNC)
1885 #define I_REFERENCED (1 << 8)
1886 #define __I_DIO_WAKEUP 9
1887 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1888 #define I_LINKABLE (1 << 10)
1889 #define I_DIRTY_TIME (1 << 11)
1890 #define __I_DIRTY_TIME_EXPIRED 12
1891 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1892 #define I_WB_SWITCH (1 << 13)
1894 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1895 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1897 extern void __mark_inode_dirty(struct inode *, int);
1898 static inline void mark_inode_dirty(struct inode *inode)
1900 __mark_inode_dirty(inode, I_DIRTY);
1903 static inline void mark_inode_dirty_sync(struct inode *inode)
1905 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1908 extern void inc_nlink(struct inode *inode);
1909 extern void drop_nlink(struct inode *inode);
1910 extern void clear_nlink(struct inode *inode);
1911 extern void set_nlink(struct inode *inode, unsigned int nlink);
1913 static inline void inode_inc_link_count(struct inode *inode)
1916 mark_inode_dirty(inode);
1919 static inline void inode_dec_link_count(struct inode *inode)
1922 mark_inode_dirty(inode);
1926 * inode_inc_iversion - increments i_version
1927 * @inode: inode that need to be updated
1929 * Every time the inode is modified, the i_version field will be incremented.
1930 * The filesystem has to be mounted with i_version flag
1933 static inline void inode_inc_iversion(struct inode *inode)
1935 spin_lock(&inode->i_lock);
1937 spin_unlock(&inode->i_lock);
1940 enum file_time_flags {
1947 extern bool atime_needs_update(const struct path *, struct inode *);
1948 extern void touch_atime(const struct path *);
1949 static inline void file_accessed(struct file *file)
1951 if (!(file->f_flags & O_NOATIME))
1952 touch_atime(&file->f_path);
1955 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1956 int sync_inode_metadata(struct inode *inode, int wait);
1958 struct file_system_type {
1961 #define FS_REQUIRES_DEV 1
1962 #define FS_BINARY_MOUNTDATA 2
1963 #define FS_HAS_SUBTYPE 4
1964 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
1965 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
1966 #define FS_USERNS_VISIBLE 32 /* FS must already be visible */
1967 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
1968 struct dentry *(*mount) (struct file_system_type *, int,
1969 const char *, void *);
1970 void (*kill_sb) (struct super_block *);
1971 struct module *owner;
1972 struct file_system_type * next;
1973 struct hlist_head fs_supers;
1975 struct lock_class_key s_lock_key;
1976 struct lock_class_key s_umount_key;
1977 struct lock_class_key s_vfs_rename_key;
1978 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
1980 struct lock_class_key i_lock_key;
1981 struct lock_class_key i_mutex_key;
1982 struct lock_class_key i_mutex_dir_key;
1985 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1987 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1988 void *data, int (*fill_super)(struct super_block *, void *, int));
1989 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1990 int flags, const char *dev_name, void *data,
1991 int (*fill_super)(struct super_block *, void *, int));
1992 extern struct dentry *mount_single(struct file_system_type *fs_type,
1993 int flags, void *data,
1994 int (*fill_super)(struct super_block *, void *, int));
1995 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1996 int flags, void *data,
1997 int (*fill_super)(struct super_block *, void *, int));
1998 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
1999 void generic_shutdown_super(struct super_block *sb);
2000 void kill_block_super(struct super_block *sb);
2001 void kill_anon_super(struct super_block *sb);
2002 void kill_litter_super(struct super_block *sb);
2003 void deactivate_super(struct super_block *sb);
2004 void deactivate_locked_super(struct super_block *sb);
2005 int set_anon_super(struct super_block *s, void *data);
2006 int get_anon_bdev(dev_t *);
2007 void free_anon_bdev(dev_t);
2008 struct super_block *sget(struct file_system_type *type,
2009 int (*test)(struct super_block *,void *),
2010 int (*set)(struct super_block *,void *),
2011 int flags, void *data);
2012 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
2013 const struct super_operations *ops,
2014 const struct dentry_operations *dops,
2017 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2018 #define fops_get(fops) \
2019 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2020 #define fops_put(fops) \
2021 do { if (fops) module_put((fops)->owner); } while(0)
2023 * This one is to be used *ONLY* from ->open() instances.
2024 * fops must be non-NULL, pinned down *and* module dependencies
2025 * should be sufficient to pin the caller down as well.
2027 #define replace_fops(f, fops) \
2029 struct file *__file = (f); \
2030 fops_put(__file->f_op); \
2031 BUG_ON(!(__file->f_op = (fops))); \
2034 extern int register_filesystem(struct file_system_type *);
2035 extern int unregister_filesystem(struct file_system_type *);
2036 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2037 #define kern_mount(type) kern_mount_data(type, NULL)
2038 extern void kern_unmount(struct vfsmount *mnt);
2039 extern int may_umount_tree(struct vfsmount *);
2040 extern int may_umount(struct vfsmount *);
2041 extern long do_mount(const char *, const char __user *,
2042 const char *, unsigned long, void *);
2043 extern struct vfsmount *collect_mounts(struct path *);
2044 extern void drop_collected_mounts(struct vfsmount *);
2045 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2047 extern int vfs_statfs(struct path *, struct kstatfs *);
2048 extern int user_statfs(const char __user *, struct kstatfs *);
2049 extern int fd_statfs(int, struct kstatfs *);
2050 extern int vfs_ustat(dev_t, struct kstatfs *);
2051 extern int freeze_super(struct super_block *super);
2052 extern int thaw_super(struct super_block *super);
2053 extern bool our_mnt(struct vfsmount *mnt);
2055 extern int current_umask(void);
2057 extern void ihold(struct inode * inode);
2058 extern void iput(struct inode *);
2059 extern int generic_update_time(struct inode *, struct timespec *, int);
2062 extern struct kobject *fs_kobj;
2064 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
2066 #define FLOCK_VERIFY_READ 1
2067 #define FLOCK_VERIFY_WRITE 2
2069 #ifdef CONFIG_FILE_LOCKING
2070 extern int locks_mandatory_locked(struct file *);
2071 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
2074 * Candidates for mandatory locking have the setgid bit set
2075 * but no group execute bit - an otherwise meaningless combination.
2078 static inline int __mandatory_lock(struct inode *ino)
2080 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2084 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2085 * otherwise these will be advisory locks
2088 static inline int mandatory_lock(struct inode *ino)
2090 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2093 static inline int locks_verify_locked(struct file *file)
2095 if (mandatory_lock(file_inode(file)))
2096 return locks_mandatory_locked(file);
2100 static inline int locks_verify_truncate(struct inode *inode,
2104 if (inode->i_flctx && mandatory_lock(inode))
2105 return locks_mandatory_area(
2106 FLOCK_VERIFY_WRITE, inode, filp,
2107 size < inode->i_size ? size : inode->i_size,
2108 (size < inode->i_size ? inode->i_size - size
2109 : size - inode->i_size)
2114 static inline int break_lease(struct inode *inode, unsigned int mode)
2117 * Since this check is lockless, we must ensure that any refcounts
2118 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2119 * could end up racing with tasks trying to set a new lease on this
2123 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2124 return __break_lease(inode, mode, FL_LEASE);
2128 static inline int break_deleg(struct inode *inode, unsigned int mode)
2131 * Since this check is lockless, we must ensure that any refcounts
2132 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2133 * could end up racing with tasks trying to set a new lease on this
2137 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2138 return __break_lease(inode, mode, FL_DELEG);
2142 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2146 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2147 if (ret == -EWOULDBLOCK && delegated_inode) {
2148 *delegated_inode = inode;
2154 static inline int break_deleg_wait(struct inode **delegated_inode)
2158 ret = break_deleg(*delegated_inode, O_WRONLY);
2159 iput(*delegated_inode);
2160 *delegated_inode = NULL;
2164 static inline int break_layout(struct inode *inode, bool wait)
2167 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2168 return __break_lease(inode,
2169 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2174 #else /* !CONFIG_FILE_LOCKING */
2175 static inline int locks_mandatory_locked(struct file *file)
2180 static inline int locks_mandatory_area(int rw, struct inode *inode,
2181 struct file *filp, loff_t offset,
2187 static inline int __mandatory_lock(struct inode *inode)
2192 static inline int mandatory_lock(struct inode *inode)
2197 static inline int locks_verify_locked(struct file *file)
2202 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2208 static inline int break_lease(struct inode *inode, unsigned int mode)
2213 static inline int break_deleg(struct inode *inode, unsigned int mode)
2218 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2223 static inline int break_deleg_wait(struct inode **delegated_inode)
2229 static inline int break_layout(struct inode *inode, bool wait)
2234 #endif /* CONFIG_FILE_LOCKING */
2239 const char *name; /* pointer to actual string */
2240 const __user char *uptr; /* original userland pointer */
2241 struct audit_names *aname;
2246 extern long vfs_truncate(struct path *, loff_t);
2247 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2249 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2251 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2253 extern struct file *file_open_name(struct filename *, int, umode_t);
2254 extern struct file *filp_open(const char *, int, umode_t);
2255 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2256 const char *, int, umode_t);
2257 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2258 extern int filp_close(struct file *, fl_owner_t id);
2260 extern struct filename *getname_flags(const char __user *, int, int *);
2261 extern struct filename *getname(const char __user *);
2262 extern struct filename *getname_kernel(const char *);
2263 extern void putname(struct filename *name);
2269 extern int finish_open(struct file *file, struct dentry *dentry,
2270 int (*open)(struct inode *, struct file *),
2272 extern int finish_no_open(struct file *file, struct dentry *dentry);
2276 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2279 extern void __init vfs_caches_init_early(void);
2280 extern void __init vfs_caches_init(void);
2282 extern struct kmem_cache *names_cachep;
2284 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2285 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2288 extern int register_blkdev(unsigned int, const char *);
2289 extern void unregister_blkdev(unsigned int, const char *);
2290 extern struct block_device *bdget(dev_t);
2291 extern struct block_device *bdgrab(struct block_device *bdev);
2292 extern void bd_set_size(struct block_device *, loff_t size);
2293 extern void bd_forget(struct inode *inode);
2294 extern void bdput(struct block_device *);
2295 extern void invalidate_bdev(struct block_device *);
2296 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2297 extern int sync_blockdev(struct block_device *bdev);
2298 extern void kill_bdev(struct block_device *);
2299 extern struct super_block *freeze_bdev(struct block_device *);
2300 extern void emergency_thaw_all(void);
2301 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2302 extern int fsync_bdev(struct block_device *);
2304 extern struct super_block *blockdev_superblock;
2306 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2308 return sb == blockdev_superblock;
2311 static inline void bd_forget(struct inode *inode) {}
2312 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2313 static inline void kill_bdev(struct block_device *bdev) {}
2314 static inline void invalidate_bdev(struct block_device *bdev) {}
2316 static inline struct super_block *freeze_bdev(struct block_device *sb)
2321 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2326 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2330 static inline int sb_is_blkdev_sb(struct super_block *sb)
2335 extern int sync_filesystem(struct super_block *);
2336 extern const struct file_operations def_blk_fops;
2337 extern const struct file_operations def_chr_fops;
2339 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2340 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2341 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2342 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2343 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2345 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2347 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2348 extern int __blkdev_reread_part(struct block_device *bdev);
2349 extern int blkdev_reread_part(struct block_device *bdev);
2352 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2353 extern void bd_unlink_disk_holder(struct block_device *bdev,
2354 struct gendisk *disk);
2356 static inline int bd_link_disk_holder(struct block_device *bdev,
2357 struct gendisk *disk)
2361 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2362 struct gendisk *disk)
2369 #define CHRDEV_MAJOR_HASH_SIZE 255
2370 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2371 extern int register_chrdev_region(dev_t, unsigned, const char *);
2372 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2373 unsigned int count, const char *name,
2374 const struct file_operations *fops);
2375 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2376 unsigned int count, const char *name);
2377 extern void unregister_chrdev_region(dev_t, unsigned);
2378 extern void chrdev_show(struct seq_file *,off_t);
2380 static inline int register_chrdev(unsigned int major, const char *name,
2381 const struct file_operations *fops)
2383 return __register_chrdev(major, 0, 256, name, fops);
2386 static inline void unregister_chrdev(unsigned int major, const char *name)
2388 __unregister_chrdev(major, 0, 256, name);
2391 /* fs/block_dev.c */
2392 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2393 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2396 #define BLKDEV_MAJOR_HASH_SIZE 255
2397 extern const char *__bdevname(dev_t, char *buffer);
2398 extern const char *bdevname(struct block_device *bdev, char *buffer);
2399 extern struct block_device *lookup_bdev(const char *);
2400 extern void blkdev_show(struct seq_file *,off_t);
2403 #define BLKDEV_MAJOR_HASH_SIZE 0
2406 extern void init_special_inode(struct inode *, umode_t, dev_t);
2408 /* Invalid inode operations -- fs/bad_inode.c */
2409 extern void make_bad_inode(struct inode *);
2410 extern int is_bad_inode(struct inode *);
2414 * return READ, READA, or WRITE
2416 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2419 * return data direction, READ or WRITE
2421 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2423 extern void check_disk_size_change(struct gendisk *disk,
2424 struct block_device *bdev);
2425 extern int revalidate_disk(struct gendisk *);
2426 extern int check_disk_change(struct block_device *);
2427 extern int __invalidate_device(struct block_device *, bool);
2428 extern int invalidate_partition(struct gendisk *, int);
2430 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2431 pgoff_t start, pgoff_t end);
2433 static inline void invalidate_remote_inode(struct inode *inode)
2435 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2436 S_ISLNK(inode->i_mode))
2437 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2439 extern int invalidate_inode_pages2(struct address_space *mapping);
2440 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2441 pgoff_t start, pgoff_t end);
2442 extern int write_inode_now(struct inode *, int);
2443 extern int filemap_fdatawrite(struct address_space *);
2444 extern int filemap_flush(struct address_space *);
2445 extern int filemap_fdatawait(struct address_space *);
2446 extern void filemap_fdatawait_keep_errors(struct address_space *);
2447 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2449 extern int filemap_write_and_wait(struct address_space *mapping);
2450 extern int filemap_write_and_wait_range(struct address_space *mapping,
2451 loff_t lstart, loff_t lend);
2452 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2453 loff_t start, loff_t end, int sync_mode);
2454 extern int filemap_fdatawrite_range(struct address_space *mapping,
2455 loff_t start, loff_t end);
2457 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2459 extern int vfs_fsync(struct file *file, int datasync);
2460 static inline int generic_write_sync(struct file *file, loff_t pos, loff_t count)
2462 if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host))
2464 return vfs_fsync_range(file, pos, pos + count - 1,
2465 (file->f_flags & __O_SYNC) ? 0 : 1);
2467 extern void emergency_sync(void);
2468 extern void emergency_remount(void);
2470 extern sector_t bmap(struct inode *, sector_t);
2472 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2473 extern int inode_permission(struct inode *, int);
2474 extern int __inode_permission(struct inode *, int);
2475 extern int generic_permission(struct inode *, int);
2476 extern int __check_sticky(struct inode *dir, struct inode *inode);
2478 static inline bool execute_ok(struct inode *inode)
2480 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2483 static inline void file_start_write(struct file *file)
2485 if (!S_ISREG(file_inode(file)->i_mode))
2487 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2490 static inline bool file_start_write_trylock(struct file *file)
2492 if (!S_ISREG(file_inode(file)->i_mode))
2494 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2497 static inline void file_end_write(struct file *file)
2499 if (!S_ISREG(file_inode(file)->i_mode))
2501 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2505 * get_write_access() gets write permission for a file.
2506 * put_write_access() releases this write permission.
2507 * This is used for regular files.
2508 * We cannot support write (and maybe mmap read-write shared) accesses and
2509 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2510 * can have the following values:
2511 * 0: no writers, no VM_DENYWRITE mappings
2512 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2513 * > 0: (i_writecount) users are writing to the file.
2515 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2516 * except for the cases where we don't hold i_writecount yet. Then we need to
2517 * use {get,deny}_write_access() - these functions check the sign and refuse
2518 * to do the change if sign is wrong.
2520 static inline int get_write_access(struct inode *inode)
2522 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2524 static inline int deny_write_access(struct file *file)
2526 struct inode *inode = file_inode(file);
2527 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2529 static inline void put_write_access(struct inode * inode)
2531 atomic_dec(&inode->i_writecount);
2533 static inline void allow_write_access(struct file *file)
2536 atomic_inc(&file_inode(file)->i_writecount);
2538 static inline bool inode_is_open_for_write(const struct inode *inode)
2540 return atomic_read(&inode->i_writecount) > 0;
2544 static inline void i_readcount_dec(struct inode *inode)
2546 BUG_ON(!atomic_read(&inode->i_readcount));
2547 atomic_dec(&inode->i_readcount);
2549 static inline void i_readcount_inc(struct inode *inode)
2551 atomic_inc(&inode->i_readcount);
2554 static inline void i_readcount_dec(struct inode *inode)
2558 static inline void i_readcount_inc(struct inode *inode)
2563 extern int do_pipe_flags(int *, int);
2565 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2566 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2567 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
2568 extern struct file * open_exec(const char *);
2570 /* fs/dcache.c -- generic fs support functions */
2571 extern int is_subdir(struct dentry *, struct dentry *);
2572 extern int path_is_under(struct path *, struct path *);
2574 extern char *file_path(struct file *, char *, int);
2576 #include <linux/err.h>
2578 /* needed for stackable file system support */
2579 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2581 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2583 extern int inode_init_always(struct super_block *, struct inode *);
2584 extern void inode_init_once(struct inode *);
2585 extern void address_space_init_once(struct address_space *mapping);
2586 extern struct inode * igrab(struct inode *);
2587 extern ino_t iunique(struct super_block *, ino_t);
2588 extern int inode_needs_sync(struct inode *inode);
2589 extern int generic_delete_inode(struct inode *inode);
2590 static inline int generic_drop_inode(struct inode *inode)
2592 return !inode->i_nlink || inode_unhashed(inode);
2595 extern struct inode *ilookup5_nowait(struct super_block *sb,
2596 unsigned long hashval, int (*test)(struct inode *, void *),
2598 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2599 int (*test)(struct inode *, void *), void *data);
2600 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2602 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2603 extern struct inode * iget_locked(struct super_block *, unsigned long);
2604 extern struct inode *find_inode_nowait(struct super_block *,
2606 int (*match)(struct inode *,
2607 unsigned long, void *),
2609 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2610 extern int insert_inode_locked(struct inode *);
2611 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2612 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2614 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2616 extern void unlock_new_inode(struct inode *);
2617 extern unsigned int get_next_ino(void);
2619 extern void __iget(struct inode * inode);
2620 extern void iget_failed(struct inode *);
2621 extern void clear_inode(struct inode *);
2622 extern void __destroy_inode(struct inode *);
2623 extern struct inode *new_inode_pseudo(struct super_block *sb);
2624 extern struct inode *new_inode(struct super_block *sb);
2625 extern void free_inode_nonrcu(struct inode *inode);
2626 extern int should_remove_suid(struct dentry *);
2627 extern int file_remove_privs(struct file *);
2628 extern int dentry_needs_remove_privs(struct dentry *dentry);
2629 static inline int file_needs_remove_privs(struct file *file)
2631 return dentry_needs_remove_privs(file->f_path.dentry);
2634 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2635 static inline void insert_inode_hash(struct inode *inode)
2637 __insert_inode_hash(inode, inode->i_ino);
2640 extern void __remove_inode_hash(struct inode *);
2641 static inline void remove_inode_hash(struct inode *inode)
2643 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2644 __remove_inode_hash(inode);
2647 extern void inode_sb_list_add(struct inode *inode);
2650 extern blk_qc_t submit_bio(int, struct bio *);
2651 extern int bdev_read_only(struct block_device *);
2653 extern int set_blocksize(struct block_device *, int);
2654 extern int sb_set_blocksize(struct super_block *, int);
2655 extern int sb_min_blocksize(struct super_block *, int);
2657 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2658 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2659 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2660 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2661 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2662 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2663 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *, loff_t);
2664 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2666 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos);
2667 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos);
2669 /* fs/block_dev.c */
2670 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2671 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2672 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2674 extern void block_sync_page(struct page *page);
2677 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2678 struct pipe_inode_info *, size_t, unsigned int);
2679 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2680 struct pipe_inode_info *, size_t, unsigned int);
2681 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2682 struct file *, loff_t *, size_t, unsigned int);
2683 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2684 struct file *out, loff_t *, size_t len, unsigned int flags);
2685 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2686 loff_t *opos, size_t len, unsigned int flags);
2690 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2691 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2692 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2693 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2694 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2695 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2696 int whence, loff_t maxsize, loff_t eof);
2697 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2698 int whence, loff_t size);
2699 extern int generic_file_open(struct inode * inode, struct file * filp);
2700 extern int nonseekable_open(struct inode * inode, struct file * filp);
2703 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2704 loff_t file_offset);
2707 /* need locking between buffered and direct access */
2710 /* filesystem does not support filling holes */
2711 DIO_SKIP_HOLES = 0x02,
2713 /* filesystem can handle aio writes beyond i_size */
2714 DIO_ASYNC_EXTEND = 0x04,
2716 /* inode/fs/bdev does not need truncate protection */
2717 DIO_SKIP_DIO_COUNT = 0x08,
2720 void dio_end_io(struct bio *bio, int error);
2722 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2723 struct block_device *bdev, struct iov_iter *iter,
2724 loff_t offset, get_block_t get_block,
2725 dio_iodone_t end_io, dio_submit_t submit_io,
2728 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2729 struct inode *inode,
2730 struct iov_iter *iter, loff_t offset,
2731 get_block_t get_block)
2733 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
2734 offset, get_block, NULL, NULL,
2735 DIO_LOCKING | DIO_SKIP_HOLES);
2739 void inode_dio_wait(struct inode *inode);
2742 * inode_dio_begin - signal start of a direct I/O requests
2743 * @inode: inode the direct I/O happens on
2745 * This is called once we've finished processing a direct I/O request,
2746 * and is used to wake up callers waiting for direct I/O to be quiesced.
2748 static inline void inode_dio_begin(struct inode *inode)
2750 atomic_inc(&inode->i_dio_count);
2754 * inode_dio_end - signal finish of a direct I/O requests
2755 * @inode: inode the direct I/O happens on
2757 * This is called once we've finished processing a direct I/O request,
2758 * and is used to wake up callers waiting for direct I/O to be quiesced.
2760 static inline void inode_dio_end(struct inode *inode)
2762 if (atomic_dec_and_test(&inode->i_dio_count))
2763 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
2766 extern void inode_set_flags(struct inode *inode, unsigned int flags,
2769 extern const struct file_operations generic_ro_fops;
2771 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2773 extern int readlink_copy(char __user *, int, const char *);
2774 extern int page_readlink(struct dentry *, char __user *, int);
2775 extern const char *page_follow_link_light(struct dentry *, void **);
2776 extern void page_put_link(struct inode *, void *);
2777 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2779 extern int page_symlink(struct inode *inode, const char *symname, int len);
2780 extern const struct inode_operations page_symlink_inode_operations;
2781 extern void kfree_put_link(struct inode *, void *);
2782 extern void free_page_put_link(struct inode *, void *);
2783 extern int generic_readlink(struct dentry *, char __user *, int);
2784 extern void generic_fillattr(struct inode *, struct kstat *);
2785 int vfs_getattr_nosec(struct path *path, struct kstat *stat);
2786 extern int vfs_getattr(struct path *, struct kstat *);
2787 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2788 void inode_add_bytes(struct inode *inode, loff_t bytes);
2789 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2790 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2791 loff_t inode_get_bytes(struct inode *inode);
2792 void inode_set_bytes(struct inode *inode, loff_t bytes);
2793 const char *simple_follow_link(struct dentry *, void **);
2794 extern const struct inode_operations simple_symlink_inode_operations;
2796 extern int iterate_dir(struct file *, struct dir_context *);
2798 extern int vfs_stat(const char __user *, struct kstat *);
2799 extern int vfs_lstat(const char __user *, struct kstat *);
2800 extern int vfs_fstat(unsigned int, struct kstat *);
2801 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2803 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2805 extern int __generic_block_fiemap(struct inode *inode,
2806 struct fiemap_extent_info *fieinfo,
2807 loff_t start, loff_t len,
2808 get_block_t *get_block);
2809 extern int generic_block_fiemap(struct inode *inode,
2810 struct fiemap_extent_info *fieinfo, u64 start,
2811 u64 len, get_block_t *get_block);
2813 extern void get_filesystem(struct file_system_type *fs);
2814 extern void put_filesystem(struct file_system_type *fs);
2815 extern struct file_system_type *get_fs_type(const char *name);
2816 extern struct super_block *get_super(struct block_device *);
2817 extern struct super_block *get_super_thawed(struct block_device *);
2818 extern struct super_block *get_active_super(struct block_device *bdev);
2819 extern void drop_super(struct super_block *sb);
2820 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2821 extern void iterate_supers_type(struct file_system_type *,
2822 void (*)(struct super_block *, void *), void *);
2824 extern int dcache_dir_open(struct inode *, struct file *);
2825 extern int dcache_dir_close(struct inode *, struct file *);
2826 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2827 extern int dcache_readdir(struct file *, struct dir_context *);
2828 extern int simple_setattr(struct dentry *, struct iattr *);
2829 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2830 extern int simple_statfs(struct dentry *, struct kstatfs *);
2831 extern int simple_open(struct inode *inode, struct file *file);
2832 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2833 extern int simple_unlink(struct inode *, struct dentry *);
2834 extern int simple_rmdir(struct inode *, struct dentry *);
2835 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2836 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2837 extern int simple_empty(struct dentry *);
2838 extern int simple_readpage(struct file *file, struct page *page);
2839 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2840 loff_t pos, unsigned len, unsigned flags,
2841 struct page **pagep, void **fsdata);
2842 extern int simple_write_end(struct file *file, struct address_space *mapping,
2843 loff_t pos, unsigned len, unsigned copied,
2844 struct page *page, void *fsdata);
2845 extern int always_delete_dentry(const struct dentry *);
2846 extern struct inode *alloc_anon_inode(struct super_block *);
2847 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
2848 extern const struct dentry_operations simple_dentry_operations;
2850 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2851 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2852 extern const struct file_operations simple_dir_operations;
2853 extern const struct inode_operations simple_dir_inode_operations;
2854 extern void make_empty_dir_inode(struct inode *inode);
2855 extern bool is_empty_dir_inode(struct inode *inode);
2856 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2857 struct dentry *d_alloc_name(struct dentry *, const char *);
2858 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2859 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2860 extern void simple_release_fs(struct vfsmount **mount, int *count);
2862 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2863 loff_t *ppos, const void *from, size_t available);
2864 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2865 const void __user *from, size_t count);
2867 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
2868 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2870 extern int generic_check_addressable(unsigned, u64);
2872 #ifdef CONFIG_MIGRATION
2873 extern int buffer_migrate_page(struct address_space *,
2874 struct page *, struct page *,
2877 #define buffer_migrate_page NULL
2880 extern int inode_change_ok(const struct inode *, struct iattr *);
2881 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2882 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2884 extern int file_update_time(struct file *file);
2886 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2887 extern void save_mount_options(struct super_block *sb, char *options);
2888 extern void replace_mount_options(struct super_block *sb, char *options);
2890 static inline bool io_is_direct(struct file *filp)
2892 return (filp->f_flags & O_DIRECT) || IS_DAX(file_inode(filp));
2895 static inline int iocb_flags(struct file *file)
2898 if (file->f_flags & O_APPEND)
2900 if (io_is_direct(file))
2905 static inline ino_t parent_ino(struct dentry *dentry)
2910 * Don't strictly need d_lock here? If the parent ino could change
2911 * then surely we'd have a deeper race in the caller?
2913 spin_lock(&dentry->d_lock);
2914 res = dentry->d_parent->d_inode->i_ino;
2915 spin_unlock(&dentry->d_lock);
2919 /* Transaction based IO helpers */
2922 * An argresp is stored in an allocated page and holds the
2923 * size of the argument or response, along with its content
2925 struct simple_transaction_argresp {
2930 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2932 char *simple_transaction_get(struct file *file, const char __user *buf,
2934 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2935 size_t size, loff_t *pos);
2936 int simple_transaction_release(struct inode *inode, struct file *file);
2938 void simple_transaction_set(struct file *file, size_t n);
2941 * simple attribute files
2943 * These attributes behave similar to those in sysfs:
2945 * Writing to an attribute immediately sets a value, an open file can be
2946 * written to multiple times.
2948 * Reading from an attribute creates a buffer from the value that might get
2949 * read with multiple read calls. When the attribute has been read
2950 * completely, no further read calls are possible until the file is opened
2953 * All attributes contain a text representation of a numeric value
2954 * that are accessed with the get() and set() functions.
2956 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2957 static int __fops ## _open(struct inode *inode, struct file *file) \
2959 __simple_attr_check_format(__fmt, 0ull); \
2960 return simple_attr_open(inode, file, __get, __set, __fmt); \
2962 static const struct file_operations __fops = { \
2963 .owner = THIS_MODULE, \
2964 .open = __fops ## _open, \
2965 .release = simple_attr_release, \
2966 .read = simple_attr_read, \
2967 .write = simple_attr_write, \
2968 .llseek = generic_file_llseek, \
2971 static inline __printf(1, 2)
2972 void __simple_attr_check_format(const char *fmt, ...)
2974 /* don't do anything, just let the compiler check the arguments; */
2977 int simple_attr_open(struct inode *inode, struct file *file,
2978 int (*get)(void *, u64 *), int (*set)(void *, u64),
2980 int simple_attr_release(struct inode *inode, struct file *file);
2981 ssize_t simple_attr_read(struct file *file, char __user *buf,
2982 size_t len, loff_t *ppos);
2983 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2984 size_t len, loff_t *ppos);
2987 int proc_nr_files(struct ctl_table *table, int write,
2988 void __user *buffer, size_t *lenp, loff_t *ppos);
2989 int proc_nr_dentry(struct ctl_table *table, int write,
2990 void __user *buffer, size_t *lenp, loff_t *ppos);
2991 int proc_nr_inodes(struct ctl_table *table, int write,
2992 void __user *buffer, size_t *lenp, loff_t *ppos);
2993 int __init get_filesystem_list(char *buf);
2995 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2996 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2998 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2999 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3000 (flag & __FMODE_NONOTIFY)))
3002 static inline int is_sxid(umode_t mode)
3004 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3007 static inline int check_sticky(struct inode *dir, struct inode *inode)
3009 if (!(dir->i_mode & S_ISVTX))
3012 return __check_sticky(dir, inode);
3015 static inline void inode_has_no_xattr(struct inode *inode)
3017 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
3018 inode->i_flags |= S_NOSEC;
3021 static inline bool is_root_inode(struct inode *inode)
3023 return inode == inode->i_sb->s_root->d_inode;
3026 static inline bool dir_emit(struct dir_context *ctx,
3027 const char *name, int namelen,
3028 u64 ino, unsigned type)
3030 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3032 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3034 return ctx->actor(ctx, ".", 1, ctx->pos,
3035 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3037 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3039 return ctx->actor(ctx, "..", 2, ctx->pos,
3040 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3042 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3044 if (ctx->pos == 0) {
3045 if (!dir_emit_dot(file, ctx))
3049 if (ctx->pos == 1) {
3050 if (!dir_emit_dotdot(file, ctx))
3056 static inline bool dir_relax(struct inode *inode)
3058 inode_unlock(inode);
3060 return !IS_DEADDIR(inode);
3063 extern bool path_noexec(const struct path *path);
3065 #endif /* _LINUX_FS_H */