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)
231 * Whiteout is represented by a char device. The following constants define the
232 * mode and device number to use.
234 #define WHITEOUT_MODE 0
235 #define WHITEOUT_DEV 0
238 * This is the Inode Attributes structure, used for notify_change(). It
239 * uses the above definitions as flags, to know which values have changed.
240 * Also, in this manner, a Filesystem can look at only the values it cares
241 * about. Basically, these are the attributes that the VFS layer can
242 * request to change from the FS layer.
244 * Derek Atkins <warlord@MIT.EDU> 94-10-20
247 unsigned int ia_valid;
252 struct timespec ia_atime;
253 struct timespec ia_mtime;
254 struct timespec ia_ctime;
257 * Not an attribute, but an auxiliary info for filesystems wanting to
258 * implement an ftruncate() like method. NOTE: filesystem should
259 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
261 struct file *ia_file;
265 * Includes for diskquotas.
267 #include <linux/quota.h>
270 * Maximum number of layers of fs stack. Needs to be limited to
271 * prevent kernel stack overflow
273 #define FILESYSTEM_MAX_STACK_DEPTH 2
276 * enum positive_aop_returns - aop return codes with specific semantics
278 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
279 * completed, that the page is still locked, and
280 * should be considered active. The VM uses this hint
281 * to return the page to the active list -- it won't
282 * be a candidate for writeback again in the near
283 * future. Other callers must be careful to unlock
284 * the page if they get this return. Returned by
287 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
288 * unlocked it and the page might have been truncated.
289 * The caller should back up to acquiring a new page and
290 * trying again. The aop will be taking reasonable
291 * precautions not to livelock. If the caller held a page
292 * reference, it should drop it before retrying. Returned
295 * address_space_operation functions return these large constants to indicate
296 * special semantics to the caller. These are much larger than the bytes in a
297 * page to allow for functions that return the number of bytes operated on in a
301 enum positive_aop_returns {
302 AOP_WRITEPAGE_ACTIVATE = 0x80000,
303 AOP_TRUNCATED_PAGE = 0x80001,
306 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
307 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
308 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
309 * helper code (eg buffer layer)
310 * to clear GFP_FS from alloc */
313 * oh the beauties of C type declarations.
316 struct address_space;
317 struct writeback_control;
319 #define IOCB_EVENTFD (1 << 0)
320 #define IOCB_APPEND (1 << 1)
321 #define IOCB_DIRECT (1 << 2)
324 struct file *ki_filp;
326 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
331 static inline bool is_sync_kiocb(struct kiocb *kiocb)
333 return kiocb->ki_complete == NULL;
336 static inline int iocb_flags(struct file *file);
338 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
340 *kiocb = (struct kiocb) {
342 .ki_flags = iocb_flags(filp),
347 * "descriptor" for what we're up to with a read.
348 * This allows us to use the same read code yet
349 * have multiple different users of the data that
350 * we read from a file.
352 * The simplest case just copies the data to user
365 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
366 unsigned long, unsigned long);
368 struct address_space_operations {
369 int (*writepage)(struct page *page, struct writeback_control *wbc);
370 int (*readpage)(struct file *, struct page *);
372 /* Write back some dirty pages from this mapping. */
373 int (*writepages)(struct address_space *, struct writeback_control *);
375 /* Set a page dirty. Return true if this dirtied it */
376 int (*set_page_dirty)(struct page *page);
378 int (*readpages)(struct file *filp, struct address_space *mapping,
379 struct list_head *pages, unsigned nr_pages);
381 int (*write_begin)(struct file *, struct address_space *mapping,
382 loff_t pos, unsigned len, unsigned flags,
383 struct page **pagep, void **fsdata);
384 int (*write_end)(struct file *, struct address_space *mapping,
385 loff_t pos, unsigned len, unsigned copied,
386 struct page *page, void *fsdata);
388 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
389 sector_t (*bmap)(struct address_space *, sector_t);
390 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
391 int (*releasepage) (struct page *, gfp_t);
392 void (*freepage)(struct page *);
393 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter, loff_t offset);
395 * migrate the contents of a page to the specified target. If
396 * migrate_mode is MIGRATE_ASYNC, it must not block.
398 int (*migratepage) (struct address_space *,
399 struct page *, struct page *, enum migrate_mode);
400 int (*launder_page) (struct page *);
401 int (*is_partially_uptodate) (struct page *, unsigned long,
403 void (*is_dirty_writeback) (struct page *, bool *, bool *);
404 int (*error_remove_page)(struct address_space *, struct page *);
406 /* swapfile support */
407 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
409 void (*swap_deactivate)(struct file *file);
412 extern const struct address_space_operations empty_aops;
415 * pagecache_write_begin/pagecache_write_end must be used by general code
416 * to write into the pagecache.
418 int pagecache_write_begin(struct file *, struct address_space *mapping,
419 loff_t pos, unsigned len, unsigned flags,
420 struct page **pagep, void **fsdata);
422 int pagecache_write_end(struct file *, struct address_space *mapping,
423 loff_t pos, unsigned len, unsigned copied,
424 struct page *page, void *fsdata);
426 struct address_space {
427 struct inode *host; /* owner: inode, block_device */
428 struct radix_tree_root page_tree; /* radix tree of all pages */
429 spinlock_t tree_lock; /* and lock protecting it */
430 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
431 struct rb_root i_mmap; /* tree of private and shared mappings */
432 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
433 /* Protected by tree_lock together with the radix tree */
434 unsigned long nrpages; /* number of total pages */
435 unsigned long nrshadows; /* number of shadow entries */
436 pgoff_t writeback_index;/* writeback starts here */
437 const struct address_space_operations *a_ops; /* methods */
438 unsigned long flags; /* error bits/gfp mask */
439 spinlock_t private_lock; /* for use by the address_space */
440 struct list_head private_list; /* ditto */
441 void *private_data; /* ditto */
442 } __attribute__((aligned(sizeof(long))));
444 * On most architectures that alignment is already the case; but
445 * must be enforced here for CRIS, to let the least significant bit
446 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
448 struct request_queue;
450 struct block_device {
451 dev_t bd_dev; /* not a kdev_t - it's a search key */
453 struct inode * bd_inode; /* will die */
454 struct super_block * bd_super;
455 struct mutex bd_mutex; /* open/close mutex */
456 struct list_head bd_inodes;
460 bool bd_write_holder;
462 struct list_head bd_holder_disks;
464 struct block_device * bd_contains;
465 unsigned bd_block_size;
466 struct hd_struct * bd_part;
467 /* number of times partitions within this device have been opened. */
468 unsigned bd_part_count;
470 struct gendisk * bd_disk;
471 struct request_queue * bd_queue;
472 struct list_head bd_list;
474 * Private data. You must have bd_claim'ed the block_device
475 * to use this. NOTE: bd_claim allows an owner to claim
476 * the same device multiple times, the owner must take special
477 * care to not mess up bd_private for that case.
479 unsigned long bd_private;
481 /* The counter of freeze processes */
482 int bd_fsfreeze_count;
483 /* Mutex for freeze */
484 struct mutex bd_fsfreeze_mutex;
488 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
491 #define PAGECACHE_TAG_DIRTY 0
492 #define PAGECACHE_TAG_WRITEBACK 1
493 #define PAGECACHE_TAG_TOWRITE 2
495 int mapping_tagged(struct address_space *mapping, int tag);
497 static inline void i_mmap_lock_write(struct address_space *mapping)
499 down_write(&mapping->i_mmap_rwsem);
502 static inline void i_mmap_unlock_write(struct address_space *mapping)
504 up_write(&mapping->i_mmap_rwsem);
507 static inline void i_mmap_lock_read(struct address_space *mapping)
509 down_read(&mapping->i_mmap_rwsem);
512 static inline void i_mmap_unlock_read(struct address_space *mapping)
514 up_read(&mapping->i_mmap_rwsem);
518 * Might pages of this file be mapped into userspace?
520 static inline int mapping_mapped(struct address_space *mapping)
522 return !RB_EMPTY_ROOT(&mapping->i_mmap);
526 * Might pages of this file have been modified in userspace?
527 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
528 * marks vma as VM_SHARED if it is shared, and the file was opened for
529 * writing i.e. vma may be mprotected writable even if now readonly.
531 * If i_mmap_writable is negative, no new writable mappings are allowed. You
532 * can only deny writable mappings, if none exists right now.
534 static inline int mapping_writably_mapped(struct address_space *mapping)
536 return atomic_read(&mapping->i_mmap_writable) > 0;
539 static inline int mapping_map_writable(struct address_space *mapping)
541 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
545 static inline void mapping_unmap_writable(struct address_space *mapping)
547 atomic_dec(&mapping->i_mmap_writable);
550 static inline int mapping_deny_writable(struct address_space *mapping)
552 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
556 static inline void mapping_allow_writable(struct address_space *mapping)
558 atomic_inc(&mapping->i_mmap_writable);
562 * Use sequence counter to get consistent i_size on 32-bit processors.
564 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
565 #include <linux/seqlock.h>
566 #define __NEED_I_SIZE_ORDERED
567 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
569 #define i_size_ordered_init(inode) do { } while (0)
573 #define ACL_NOT_CACHED ((void *)(-1))
575 #define IOP_FASTPERM 0x0001
576 #define IOP_LOOKUP 0x0002
577 #define IOP_NOFOLLOW 0x0004
580 * Keep mostly read-only and often accessed (especially for
581 * the RCU path lookup and 'stat' data) fields at the beginning
582 * of the 'struct inode'
586 unsigned short i_opflags;
589 unsigned int i_flags;
591 #ifdef CONFIG_FS_POSIX_ACL
592 struct posix_acl *i_acl;
593 struct posix_acl *i_default_acl;
596 const struct inode_operations *i_op;
597 struct super_block *i_sb;
598 struct address_space *i_mapping;
600 #ifdef CONFIG_SECURITY
604 /* Stat data, not accessed from path walking */
607 * Filesystems may only read i_nlink directly. They shall use the
608 * following functions for modification:
610 * (set|clear|inc|drop)_nlink
611 * inode_(inc|dec)_link_count
614 const unsigned int i_nlink;
615 unsigned int __i_nlink;
619 struct timespec i_atime;
620 struct timespec i_mtime;
621 struct timespec i_ctime;
622 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
623 unsigned short i_bytes;
624 unsigned int i_blkbits;
627 #ifdef __NEED_I_SIZE_ORDERED
628 seqcount_t i_size_seqcount;
632 unsigned long i_state;
633 struct mutex i_mutex;
635 unsigned long dirtied_when; /* jiffies of first dirtying */
636 unsigned long dirtied_time_when;
638 struct hlist_node i_hash;
639 struct list_head i_io_list; /* backing dev IO list */
640 #ifdef CONFIG_CGROUP_WRITEBACK
641 struct bdi_writeback *i_wb; /* the associated cgroup wb */
643 /* foreign inode detection, see wbc_detach_inode() */
645 u16 i_wb_frn_avg_time;
646 u16 i_wb_frn_history;
648 struct list_head i_lru; /* inode LRU list */
649 struct list_head i_sb_list;
651 struct hlist_head i_dentry;
652 struct rcu_head i_rcu;
656 atomic_t i_dio_count;
657 atomic_t i_writecount;
659 atomic_t i_readcount; /* struct files open RO */
661 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
662 struct file_lock_context *i_flctx;
663 struct address_space i_data;
664 struct list_head i_devices;
666 struct pipe_inode_info *i_pipe;
667 struct block_device *i_bdev;
674 #ifdef CONFIG_FSNOTIFY
675 __u32 i_fsnotify_mask; /* all events this inode cares about */
676 struct hlist_head i_fsnotify_marks;
679 void *i_private; /* fs or device private pointer */
682 static inline int inode_unhashed(struct inode *inode)
684 return hlist_unhashed(&inode->i_hash);
688 * inode->i_mutex nesting subclasses for the lock validator:
690 * 0: the object of the current VFS operation
694 * 4: second non-directory
695 * 5: second parent (when locking independent directories in rename)
697 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
698 * non-directories at once.
700 * The locking order between these classes is
701 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
703 enum inode_i_mutex_lock_class
713 static inline void inode_lock(struct inode *inode)
715 mutex_lock(&inode->i_mutex);
718 static inline void inode_unlock(struct inode *inode)
720 mutex_unlock(&inode->i_mutex);
723 static inline int inode_trylock(struct inode *inode)
725 return mutex_trylock(&inode->i_mutex);
728 static inline int inode_is_locked(struct inode *inode)
730 return mutex_is_locked(&inode->i_mutex);
733 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
735 mutex_lock_nested(&inode->i_mutex, subclass);
738 void lock_two_nondirectories(struct inode *, struct inode*);
739 void unlock_two_nondirectories(struct inode *, struct inode*);
742 * NOTE: in a 32bit arch with a preemptable kernel and
743 * an UP compile the i_size_read/write must be atomic
744 * with respect to the local cpu (unlike with preempt disabled),
745 * but they don't need to be atomic with respect to other cpus like in
746 * true SMP (so they need either to either locally disable irq around
747 * the read or for example on x86 they can be still implemented as a
748 * cmpxchg8b without the need of the lock prefix). For SMP compiles
749 * and 64bit archs it makes no difference if preempt is enabled or not.
751 static inline loff_t i_size_read(const struct inode *inode)
753 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
758 seq = read_seqcount_begin(&inode->i_size_seqcount);
759 i_size = inode->i_size;
760 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
762 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
766 i_size = inode->i_size;
770 return inode->i_size;
775 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
776 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
777 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
779 static inline void i_size_write(struct inode *inode, loff_t i_size)
781 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
783 write_seqcount_begin(&inode->i_size_seqcount);
784 inode->i_size = i_size;
785 write_seqcount_end(&inode->i_size_seqcount);
787 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
789 inode->i_size = i_size;
792 inode->i_size = i_size;
796 /* Helper functions so that in most cases filesystems will
797 * not need to deal directly with kuid_t and kgid_t and can
798 * instead deal with the raw numeric values that are stored
801 static inline uid_t i_uid_read(const struct inode *inode)
803 return from_kuid(&init_user_ns, inode->i_uid);
806 static inline gid_t i_gid_read(const struct inode *inode)
808 return from_kgid(&init_user_ns, inode->i_gid);
811 static inline void i_uid_write(struct inode *inode, uid_t uid)
813 inode->i_uid = make_kuid(&init_user_ns, uid);
816 static inline void i_gid_write(struct inode *inode, gid_t gid)
818 inode->i_gid = make_kgid(&init_user_ns, gid);
821 static inline unsigned iminor(const struct inode *inode)
823 return MINOR(inode->i_rdev);
826 static inline unsigned imajor(const struct inode *inode)
828 return MAJOR(inode->i_rdev);
831 extern struct block_device *I_BDEV(struct inode *inode);
834 rwlock_t lock; /* protects pid, uid, euid fields */
835 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
836 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
837 kuid_t uid, euid; /* uid/euid of process setting the owner */
838 int signum; /* posix.1b rt signal to be delivered on IO */
842 * Track a single file's readahead state
844 struct file_ra_state {
845 pgoff_t start; /* where readahead started */
846 unsigned int size; /* # of readahead pages */
847 unsigned int async_size; /* do asynchronous readahead when
848 there are only # of pages ahead */
850 unsigned int ra_pages; /* Maximum readahead window */
851 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
852 loff_t prev_pos; /* Cache last read() position */
856 * Check if @index falls in the readahead windows.
858 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
860 return (index >= ra->start &&
861 index < ra->start + ra->size);
866 struct llist_node fu_llist;
867 struct rcu_head fu_rcuhead;
870 struct inode *f_inode; /* cached value */
871 const struct file_operations *f_op;
874 * Protects f_ep_links, f_flags.
875 * Must not be taken from IRQ context.
878 atomic_long_t f_count;
879 unsigned int f_flags;
881 struct mutex f_pos_lock;
883 struct fown_struct f_owner;
884 const struct cred *f_cred;
885 struct file_ra_state f_ra;
888 #ifdef CONFIG_SECURITY
891 /* needed for tty driver, and maybe others */
895 /* Used by fs/eventpoll.c to link all the hooks to this file */
896 struct list_head f_ep_links;
897 struct list_head f_tfile_llink;
898 #endif /* #ifdef CONFIG_EPOLL */
899 struct address_space *f_mapping;
900 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
905 /* file identifier */
906 unsigned char f_handle[0];
909 static inline struct file *get_file(struct file *f)
911 atomic_long_inc(&f->f_count);
914 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
915 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
916 #define file_count(x) atomic_long_read(&(x)->f_count)
918 #define MAX_NON_LFS ((1UL<<31) - 1)
920 /* Page cache limit. The filesystems should put that into their s_maxbytes
921 limits, otherwise bad things can happen in VM. */
922 #if BITS_PER_LONG==32
923 #define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
924 #elif BITS_PER_LONG==64
925 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
930 #define FL_DELEG 4 /* NFSv4 delegation */
931 #define FL_ACCESS 8 /* not trying to lock, just looking */
932 #define FL_EXISTS 16 /* when unlocking, test for existence */
933 #define FL_LEASE 32 /* lease held on this file */
934 #define FL_CLOSE 64 /* unlock on close */
935 #define FL_SLEEP 128 /* A blocking lock */
936 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
937 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
938 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
939 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
942 * Special return value from posix_lock_file() and vfs_lock_file() for
943 * asynchronous locking.
945 #define FILE_LOCK_DEFERRED 1
947 /* legacy typedef, should eventually be removed */
948 typedef void *fl_owner_t;
952 struct file_lock_operations {
953 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
954 void (*fl_release_private)(struct file_lock *);
957 struct lock_manager_operations {
958 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
959 unsigned long (*lm_owner_key)(struct file_lock *);
960 fl_owner_t (*lm_get_owner)(fl_owner_t);
961 void (*lm_put_owner)(fl_owner_t);
962 void (*lm_notify)(struct file_lock *); /* unblock callback */
963 int (*lm_grant)(struct file_lock *, int);
964 bool (*lm_break)(struct file_lock *);
965 int (*lm_change)(struct file_lock *, int, struct list_head *);
966 void (*lm_setup)(struct file_lock *, void **);
969 struct lock_manager {
970 struct list_head list;
972 * NFSv4 and up also want opens blocked during the grace period;
979 void locks_start_grace(struct net *, struct lock_manager *);
980 void locks_end_grace(struct lock_manager *);
981 int locks_in_grace(struct net *);
982 int opens_in_grace(struct net *);
984 /* that will die - we need it for nfs_lock_info */
985 #include <linux/nfs_fs_i.h>
988 * struct file_lock represents a generic "file lock". It's used to represent
989 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
990 * note that the same struct is used to represent both a request for a lock and
991 * the lock itself, but the same object is never used for both.
993 * FIXME: should we create a separate "struct lock_request" to help distinguish
996 * The varous i_flctx lists are ordered by:
999 * 2) lock range start
1002 * Obviously, the last two criteria only matter for POSIX locks.
1005 struct file_lock *fl_next; /* singly linked list for this inode */
1006 struct list_head fl_list; /* link into file_lock_context */
1007 struct hlist_node fl_link; /* node in global lists */
1008 struct list_head fl_block; /* circular list of blocked processes */
1009 fl_owner_t fl_owner;
1010 unsigned int fl_flags;
1011 unsigned char fl_type;
1012 unsigned int fl_pid;
1013 int fl_link_cpu; /* what cpu's list is this on? */
1014 struct pid *fl_nspid;
1015 wait_queue_head_t fl_wait;
1016 struct file *fl_file;
1020 struct fasync_struct * fl_fasync; /* for lease break notifications */
1021 /* for lease breaks: */
1022 unsigned long fl_break_time;
1023 unsigned long fl_downgrade_time;
1025 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1026 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1028 struct nfs_lock_info nfs_fl;
1029 struct nfs4_lock_info nfs4_fl;
1031 struct list_head link; /* link in AFS vnode's pending_locks list */
1032 int state; /* state of grant or error if -ve */
1037 struct file_lock_context {
1038 spinlock_t flc_lock;
1039 struct list_head flc_flock;
1040 struct list_head flc_posix;
1041 struct list_head flc_lease;
1044 /* The following constant reflects the upper bound of the file/locking space */
1046 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1047 #define OFFSET_MAX INT_LIMIT(loff_t)
1048 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1051 #include <linux/fcntl.h>
1053 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1055 #ifdef CONFIG_FILE_LOCKING
1056 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
1057 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1058 struct flock __user *);
1060 #if BITS_PER_LONG == 32
1061 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
1062 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1063 struct flock64 __user *);
1066 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1067 extern int fcntl_getlease(struct file *filp);
1070 void locks_free_lock_context(struct file_lock_context *ctx);
1071 void locks_free_lock(struct file_lock *fl);
1072 extern void locks_init_lock(struct file_lock *);
1073 extern struct file_lock * locks_alloc_lock(void);
1074 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1075 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1076 extern void locks_remove_posix(struct file *, fl_owner_t);
1077 extern void locks_remove_file(struct file *);
1078 extern void locks_release_private(struct file_lock *);
1079 extern void posix_test_lock(struct file *, struct file_lock *);
1080 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1081 extern int posix_unblock_lock(struct file_lock *);
1082 extern int vfs_test_lock(struct file *, struct file_lock *);
1083 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1084 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1085 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1086 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1087 extern void lease_get_mtime(struct inode *, struct timespec *time);
1088 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1089 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1090 extern int lease_modify(struct file_lock *, int, struct list_head *);
1091 struct files_struct;
1092 extern void show_fd_locks(struct seq_file *f,
1093 struct file *filp, struct files_struct *files);
1094 #else /* !CONFIG_FILE_LOCKING */
1095 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1096 struct flock __user *user)
1101 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1102 unsigned int cmd, struct flock __user *user)
1107 #if BITS_PER_LONG == 32
1108 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1109 struct flock64 __user *user)
1114 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1115 unsigned int cmd, struct flock64 __user *user)
1120 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1125 static inline int fcntl_getlease(struct file *filp)
1131 locks_free_lock_context(struct file_lock_context *ctx)
1135 static inline void locks_init_lock(struct file_lock *fl)
1140 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1145 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1150 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1155 static inline void locks_remove_file(struct file *filp)
1160 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1165 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1166 struct file_lock *conflock)
1171 static inline int posix_unblock_lock(struct file_lock *waiter)
1176 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1181 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1182 struct file_lock *fl, struct file_lock *conf)
1187 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1192 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1197 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1202 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1207 static inline int generic_setlease(struct file *filp, long arg,
1208 struct file_lock **flp, void **priv)
1213 static inline int vfs_setlease(struct file *filp, long arg,
1214 struct file_lock **lease, void **priv)
1219 static inline int lease_modify(struct file_lock *fl, int arg,
1220 struct list_head *dispose)
1225 struct files_struct;
1226 static inline void show_fd_locks(struct seq_file *f,
1227 struct file *filp, struct files_struct *files) {}
1228 #endif /* !CONFIG_FILE_LOCKING */
1230 static inline struct inode *file_inode(const struct file *f)
1235 static inline struct dentry *file_dentry(const struct file *file)
1237 struct dentry *dentry = file->f_path.dentry;
1239 if (unlikely(dentry->d_flags & DCACHE_OP_REAL))
1240 return dentry->d_op->d_real(dentry, file_inode(file));
1245 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1247 return locks_lock_inode_wait(file_inode(filp), fl);
1250 struct fasync_struct {
1254 struct fasync_struct *fa_next; /* singly linked list */
1255 struct file *fa_file;
1256 struct rcu_head fa_rcu;
1259 #define FASYNC_MAGIC 0x4601
1261 /* SMP safe fasync helpers: */
1262 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1263 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1264 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1265 extern struct fasync_struct *fasync_alloc(void);
1266 extern void fasync_free(struct fasync_struct *);
1268 /* can be called from interrupts */
1269 extern void kill_fasync(struct fasync_struct **, int, int);
1271 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1272 extern void f_setown(struct file *filp, unsigned long arg, int force);
1273 extern void f_delown(struct file *filp);
1274 extern pid_t f_getown(struct file *filp);
1275 extern int send_sigurg(struct fown_struct *fown);
1283 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1284 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1285 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1286 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1287 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1290 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1291 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1293 /* Possible states of 'frozen' field */
1295 SB_UNFROZEN = 0, /* FS is unfrozen */
1296 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1297 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1298 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1299 * internal threads if needed) */
1300 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1303 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1306 int frozen; /* Is sb frozen? */
1307 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1308 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1311 struct super_block {
1312 struct list_head s_list; /* Keep this first */
1313 dev_t s_dev; /* search index; _not_ kdev_t */
1314 unsigned char s_blocksize_bits;
1315 unsigned long s_blocksize;
1316 loff_t s_maxbytes; /* Max file size */
1317 struct file_system_type *s_type;
1318 const struct super_operations *s_op;
1319 const struct dquot_operations *dq_op;
1320 const struct quotactl_ops *s_qcop;
1321 const struct export_operations *s_export_op;
1322 unsigned long s_flags;
1323 unsigned long s_iflags; /* internal SB_I_* flags */
1324 unsigned long s_magic;
1325 struct dentry *s_root;
1326 struct rw_semaphore s_umount;
1329 #ifdef CONFIG_SECURITY
1332 const struct xattr_handler **s_xattr;
1334 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1335 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1336 struct block_device *s_bdev;
1337 struct backing_dev_info *s_bdi;
1338 struct mtd_info *s_mtd;
1339 struct hlist_node s_instances;
1340 unsigned int s_quota_types; /* Bitmask of supported quota types */
1341 struct quota_info s_dquot; /* Diskquota specific options */
1343 struct sb_writers s_writers;
1345 char s_id[32]; /* Informational name */
1346 u8 s_uuid[16]; /* UUID */
1348 void *s_fs_info; /* Filesystem private info */
1349 unsigned int s_max_links;
1352 /* Granularity of c/m/atime in ns.
1353 Cannot be worse than a second */
1357 * The next field is for VFS *only*. No filesystems have any business
1358 * even looking at it. You had been warned.
1360 struct mutex s_vfs_rename_mutex; /* Kludge */
1363 * Filesystem subtype. If non-empty the filesystem type field
1364 * in /proc/mounts will be "type.subtype"
1369 * Saved mount options for lazy filesystems using
1370 * generic_show_options()
1372 char __rcu *s_options;
1373 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1376 * Saved pool identifier for cleancache (-1 means none)
1378 int cleancache_poolid;
1380 struct shrinker s_shrink; /* per-sb shrinker handle */
1382 /* Number of inodes with nlink == 0 but still referenced */
1383 atomic_long_t s_remove_count;
1385 /* Being remounted read-only */
1386 int s_readonly_remount;
1388 /* AIO completions deferred from interrupt context */
1389 struct workqueue_struct *s_dio_done_wq;
1390 struct hlist_head s_pins;
1393 * Keep the lru lists last in the structure so they always sit on their
1394 * own individual cachelines.
1396 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1397 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1398 struct rcu_head rcu;
1399 struct work_struct destroy_work;
1401 struct mutex s_sync_lock; /* sync serialisation lock */
1404 * Indicates how deep in a filesystem stack this SB is
1408 /* s_inode_list_lock protects s_inodes */
1409 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1410 struct list_head s_inodes; /* all inodes */
1413 extern struct timespec current_fs_time(struct super_block *sb);
1416 * Snapshotting support.
1419 void __sb_end_write(struct super_block *sb, int level);
1420 int __sb_start_write(struct super_block *sb, int level, bool wait);
1422 #define __sb_writers_acquired(sb, lev) \
1423 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1424 #define __sb_writers_release(sb, lev) \
1425 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1428 * sb_end_write - drop write access to a superblock
1429 * @sb: the super we wrote to
1431 * Decrement number of writers to the filesystem. Wake up possible waiters
1432 * wanting to freeze the filesystem.
1434 static inline void sb_end_write(struct super_block *sb)
1436 __sb_end_write(sb, SB_FREEZE_WRITE);
1440 * sb_end_pagefault - drop write access to a superblock from a page fault
1441 * @sb: the super we wrote to
1443 * Decrement number of processes handling write page fault to the filesystem.
1444 * Wake up possible waiters wanting to freeze the filesystem.
1446 static inline void sb_end_pagefault(struct super_block *sb)
1448 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1452 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1453 * @sb: the super we wrote to
1455 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1456 * waiters wanting to freeze the filesystem.
1458 static inline void sb_end_intwrite(struct super_block *sb)
1460 __sb_end_write(sb, SB_FREEZE_FS);
1464 * sb_start_write - get write access to a superblock
1465 * @sb: the super we write to
1467 * When a process wants to write data or metadata to a file system (i.e. dirty
1468 * a page or an inode), it should embed the operation in a sb_start_write() -
1469 * sb_end_write() pair to get exclusion against file system freezing. This
1470 * function increments number of writers preventing freezing. If the file
1471 * system is already frozen, the function waits until the file system is
1474 * Since freeze protection behaves as a lock, users have to preserve
1475 * ordering of freeze protection and other filesystem locks. Generally,
1476 * freeze protection should be the outermost lock. In particular, we have:
1479 * -> i_mutex (write path, truncate, directory ops, ...)
1480 * -> s_umount (freeze_super, thaw_super)
1482 static inline void sb_start_write(struct super_block *sb)
1484 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1487 static inline int sb_start_write_trylock(struct super_block *sb)
1489 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1493 * sb_start_pagefault - get write access to a superblock from a page fault
1494 * @sb: the super we write to
1496 * When a process starts handling write page fault, it should embed the
1497 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1498 * exclusion against file system freezing. This is needed since the page fault
1499 * is going to dirty a page. This function increments number of running page
1500 * faults preventing freezing. If the file system is already frozen, the
1501 * function waits until the file system is thawed.
1503 * Since page fault freeze protection behaves as a lock, users have to preserve
1504 * ordering of freeze protection and other filesystem locks. It is advised to
1505 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1506 * handling code implies lock dependency:
1509 * -> sb_start_pagefault
1511 static inline void sb_start_pagefault(struct super_block *sb)
1513 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1517 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1518 * @sb: the super we write to
1520 * This is the third level of protection against filesystem freezing. It is
1521 * free for use by a filesystem. The only requirement is that it must rank
1522 * below sb_start_pagefault.
1524 * For example filesystem can call sb_start_intwrite() when starting a
1525 * transaction which somewhat eases handling of freezing for internal sources
1526 * of filesystem changes (internal fs threads, discarding preallocation on file
1529 static inline void sb_start_intwrite(struct super_block *sb)
1531 __sb_start_write(sb, SB_FREEZE_FS, true);
1535 extern bool inode_owner_or_capable(const struct inode *inode);
1538 * VFS helper functions..
1540 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1541 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1542 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1543 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1544 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1545 extern int vfs_rmdir(struct inode *, struct dentry *);
1546 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1547 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1548 extern int vfs_whiteout(struct inode *, struct dentry *);
1551 * VFS dentry helper functions.
1553 extern void dentry_unhash(struct dentry *dentry);
1556 * VFS file helper functions.
1558 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1561 * VFS FS_IOC_FIEMAP helper definitions.
1563 struct fiemap_extent_info {
1564 unsigned int fi_flags; /* Flags as passed from user */
1565 unsigned int fi_extents_mapped; /* Number of mapped extents */
1566 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1567 struct fiemap_extent __user *fi_extents_start; /* Start of
1568 fiemap_extent array */
1570 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1571 u64 phys, u64 len, u32 flags);
1572 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1577 * NOTE! These match bits 12..15 of stat.st_mode
1578 * (ie "(i_mode >> 12) & 15").
1580 #define DT_UNKNOWN 0
1591 * This is the "filldir" function type, used by readdir() to let
1592 * the kernel specify what kind of dirent layout it wants to have.
1593 * This allows the kernel to read directories into kernel space or
1594 * to have different dirent layouts depending on the binary type.
1597 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1600 struct dir_context {
1601 const filldir_t actor;
1605 struct block_device_operations;
1607 /* These macros are for out of kernel modules to test that
1608 * the kernel supports the unlocked_ioctl and compat_ioctl
1609 * fields in struct file_operations. */
1610 #define HAVE_COMPAT_IOCTL 1
1611 #define HAVE_UNLOCKED_IOCTL 1
1614 * These flags let !MMU mmap() govern direct device mapping vs immediate
1615 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1617 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1618 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1619 * NOMMU_MAP_READ: Can be mapped for reading
1620 * NOMMU_MAP_WRITE: Can be mapped for writing
1621 * NOMMU_MAP_EXEC: Can be mapped for execution
1623 #define NOMMU_MAP_COPY 0x00000001
1624 #define NOMMU_MAP_DIRECT 0x00000008
1625 #define NOMMU_MAP_READ VM_MAYREAD
1626 #define NOMMU_MAP_WRITE VM_MAYWRITE
1627 #define NOMMU_MAP_EXEC VM_MAYEXEC
1629 #define NOMMU_VMFLAGS \
1630 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1635 struct file_operations {
1636 struct module *owner;
1637 loff_t (*llseek) (struct file *, loff_t, int);
1638 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1639 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1640 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1641 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1642 int (*iterate) (struct file *, struct dir_context *);
1643 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1644 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1645 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1646 int (*mmap) (struct file *, struct vm_area_struct *);
1647 int (*open) (struct inode *, struct file *);
1648 int (*flush) (struct file *, fl_owner_t id);
1649 int (*release) (struct inode *, struct file *);
1650 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1651 int (*aio_fsync) (struct kiocb *, int datasync);
1652 int (*fasync) (int, struct file *, int);
1653 int (*lock) (struct file *, int, struct file_lock *);
1654 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1655 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1656 int (*check_flags)(int);
1657 int (*flock) (struct file *, int, struct file_lock *);
1658 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1659 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1660 int (*setlease)(struct file *, long, struct file_lock **, void **);
1661 long (*fallocate)(struct file *file, int mode, loff_t offset,
1663 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1665 unsigned (*mmap_capabilities)(struct file *);
1669 struct inode_operations {
1670 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1671 const char * (*follow_link) (struct dentry *, void **);
1672 int (*permission) (struct inode *, int);
1673 struct posix_acl * (*get_acl)(struct inode *, int);
1675 int (*readlink) (struct dentry *, char __user *,int);
1676 void (*put_link) (struct inode *, void *);
1678 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1679 int (*link) (struct dentry *,struct inode *,struct dentry *);
1680 int (*unlink) (struct inode *,struct dentry *);
1681 int (*symlink) (struct inode *,struct dentry *,const char *);
1682 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1683 int (*rmdir) (struct inode *,struct dentry *);
1684 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1685 int (*rename) (struct inode *, struct dentry *,
1686 struct inode *, struct dentry *);
1687 int (*rename2) (struct inode *, struct dentry *,
1688 struct inode *, struct dentry *, unsigned int);
1689 int (*setattr) (struct dentry *, struct iattr *);
1690 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1691 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1692 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1693 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1694 int (*removexattr) (struct dentry *, const char *);
1695 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1697 int (*update_time)(struct inode *, struct timespec *, int);
1698 int (*atomic_open)(struct inode *, struct dentry *,
1699 struct file *, unsigned open_flag,
1700 umode_t create_mode, int *opened);
1701 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1702 int (*set_acl)(struct inode *, struct posix_acl *, int);
1703 } ____cacheline_aligned;
1705 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1706 unsigned long nr_segs, unsigned long fast_segs,
1707 struct iovec *fast_pointer,
1708 struct iovec **ret_pointer);
1710 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1711 extern ssize_t __vfs_write(struct file *, const char __user *, size_t, loff_t *);
1712 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1713 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1714 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1715 unsigned long, loff_t *);
1716 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1717 unsigned long, loff_t *);
1719 struct super_operations {
1720 struct inode *(*alloc_inode)(struct super_block *sb);
1721 void (*destroy_inode)(struct inode *);
1723 void (*dirty_inode) (struct inode *, int flags);
1724 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1725 int (*drop_inode) (struct inode *);
1726 void (*evict_inode) (struct inode *);
1727 void (*put_super) (struct super_block *);
1728 int (*sync_fs)(struct super_block *sb, int wait);
1729 int (*freeze_super) (struct super_block *);
1730 int (*freeze_fs) (struct super_block *);
1731 int (*thaw_super) (struct super_block *);
1732 int (*unfreeze_fs) (struct super_block *);
1733 int (*statfs) (struct dentry *, struct kstatfs *);
1734 int (*remount_fs) (struct super_block *, int *, char *);
1735 void (*umount_begin) (struct super_block *);
1737 int (*show_options)(struct seq_file *, struct dentry *);
1738 int (*show_devname)(struct seq_file *, struct dentry *);
1739 int (*show_path)(struct seq_file *, struct dentry *);
1740 int (*show_stats)(struct seq_file *, struct dentry *);
1742 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1743 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1744 struct dquot **(*get_dquots)(struct inode *);
1746 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1747 long (*nr_cached_objects)(struct super_block *,
1748 struct shrink_control *);
1749 long (*free_cached_objects)(struct super_block *,
1750 struct shrink_control *);
1754 * Inode flags - they have no relation to superblock flags now
1756 #define S_SYNC 1 /* Writes are synced at once */
1757 #define S_NOATIME 2 /* Do not update access times */
1758 #define S_APPEND 4 /* Append-only file */
1759 #define S_IMMUTABLE 8 /* Immutable file */
1760 #define S_DEAD 16 /* removed, but still open directory */
1761 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1762 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1763 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1764 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1765 #define S_PRIVATE 512 /* Inode is fs-internal */
1766 #define S_IMA 1024 /* Inode has an associated IMA struct */
1767 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1768 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1769 #ifdef CONFIG_FS_DAX
1770 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1772 #define S_DAX 0 /* Make all the DAX code disappear */
1776 * Note that nosuid etc flags are inode-specific: setting some file-system
1777 * flags just means all the inodes inherit those flags by default. It might be
1778 * possible to override it selectively if you really wanted to with some
1779 * ioctl() that is not currently implemented.
1781 * Exception: MS_RDONLY is always applied to the entire file system.
1783 * Unfortunately, it is possible to change a filesystems flags with it mounted
1784 * with files in use. This means that all of the inodes will not have their
1785 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1786 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1788 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1790 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1791 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1792 ((inode)->i_flags & S_SYNC))
1793 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1794 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1795 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1796 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1797 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1799 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1800 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1801 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1802 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1804 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1805 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1806 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1807 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1808 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1809 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1810 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1811 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1813 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1814 (inode)->i_rdev == WHITEOUT_DEV)
1817 * Inode state bits. Protected by inode->i_lock
1819 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1820 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1822 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1823 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1824 * various stages of removing an inode.
1826 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1828 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1829 * fdatasync(). i_atime is the usual cause.
1830 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1831 * these changes separately from I_DIRTY_SYNC so that we
1832 * don't have to write inode on fdatasync() when only
1833 * mtime has changed in it.
1834 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1835 * I_NEW Serves as both a mutex and completion notification.
1836 * New inodes set I_NEW. If two processes both create
1837 * the same inode, one of them will release its inode and
1838 * wait for I_NEW to be released before returning.
1839 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1840 * also cause waiting on I_NEW, without I_NEW actually
1841 * being set. find_inode() uses this to prevent returning
1842 * nearly-dead inodes.
1843 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1844 * is zero. I_FREEING must be set when I_WILL_FREE is
1846 * I_FREEING Set when inode is about to be freed but still has dirty
1847 * pages or buffers attached or the inode itself is still
1849 * I_CLEAR Added by clear_inode(). In this state the inode is
1850 * clean and can be destroyed. Inode keeps I_FREEING.
1852 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1853 * prohibited for many purposes. iget() must wait for
1854 * the inode to be completely released, then create it
1855 * anew. Other functions will just ignore such inodes,
1856 * if appropriate. I_NEW is used for waiting.
1858 * I_SYNC Writeback of inode is running. The bit is set during
1859 * data writeback, and cleared with a wakeup on the bit
1860 * address once it is done. The bit is also used to pin
1861 * the inode in memory for flusher thread.
1863 * I_REFERENCED Marks the inode as recently references on the LRU list.
1865 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1867 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1868 * synchronize competing switching instances and to tell
1869 * wb stat updates to grab mapping->tree_lock. See
1870 * inode_switch_wb_work_fn() for details.
1872 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1874 #define I_DIRTY_SYNC (1 << 0)
1875 #define I_DIRTY_DATASYNC (1 << 1)
1876 #define I_DIRTY_PAGES (1 << 2)
1878 #define I_NEW (1 << __I_NEW)
1879 #define I_WILL_FREE (1 << 4)
1880 #define I_FREEING (1 << 5)
1881 #define I_CLEAR (1 << 6)
1883 #define I_SYNC (1 << __I_SYNC)
1884 #define I_REFERENCED (1 << 8)
1885 #define __I_DIO_WAKEUP 9
1886 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1887 #define I_LINKABLE (1 << 10)
1888 #define I_DIRTY_TIME (1 << 11)
1889 #define __I_DIRTY_TIME_EXPIRED 12
1890 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1891 #define I_WB_SWITCH (1 << 13)
1893 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1894 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1896 extern void __mark_inode_dirty(struct inode *, int);
1897 static inline void mark_inode_dirty(struct inode *inode)
1899 __mark_inode_dirty(inode, I_DIRTY);
1902 static inline void mark_inode_dirty_sync(struct inode *inode)
1904 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1907 extern void inc_nlink(struct inode *inode);
1908 extern void drop_nlink(struct inode *inode);
1909 extern void clear_nlink(struct inode *inode);
1910 extern void set_nlink(struct inode *inode, unsigned int nlink);
1912 static inline void inode_inc_link_count(struct inode *inode)
1915 mark_inode_dirty(inode);
1918 static inline void inode_dec_link_count(struct inode *inode)
1921 mark_inode_dirty(inode);
1925 * inode_inc_iversion - increments i_version
1926 * @inode: inode that need to be updated
1928 * Every time the inode is modified, the i_version field will be incremented.
1929 * The filesystem has to be mounted with i_version flag
1932 static inline void inode_inc_iversion(struct inode *inode)
1934 spin_lock(&inode->i_lock);
1936 spin_unlock(&inode->i_lock);
1939 enum file_time_flags {
1946 extern bool atime_needs_update(const struct path *, struct inode *);
1947 extern void touch_atime(const struct path *);
1948 static inline void file_accessed(struct file *file)
1950 if (!(file->f_flags & O_NOATIME))
1951 touch_atime(&file->f_path);
1954 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1955 int sync_inode_metadata(struct inode *inode, int wait);
1957 struct file_system_type {
1960 #define FS_REQUIRES_DEV 1
1961 #define FS_BINARY_MOUNTDATA 2
1962 #define FS_HAS_SUBTYPE 4
1963 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
1964 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
1965 #define FS_USERNS_VISIBLE 32 /* FS must already be visible */
1966 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
1967 struct dentry *(*mount) (struct file_system_type *, int,
1968 const char *, void *);
1969 void (*kill_sb) (struct super_block *);
1970 struct module *owner;
1971 struct file_system_type * next;
1972 struct hlist_head fs_supers;
1974 struct lock_class_key s_lock_key;
1975 struct lock_class_key s_umount_key;
1976 struct lock_class_key s_vfs_rename_key;
1977 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
1979 struct lock_class_key i_lock_key;
1980 struct lock_class_key i_mutex_key;
1981 struct lock_class_key i_mutex_dir_key;
1984 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1986 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1987 void *data, int (*fill_super)(struct super_block *, void *, int));
1988 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1989 int flags, const char *dev_name, void *data,
1990 int (*fill_super)(struct super_block *, void *, int));
1991 extern struct dentry *mount_single(struct file_system_type *fs_type,
1992 int flags, void *data,
1993 int (*fill_super)(struct super_block *, void *, int));
1994 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1995 int flags, void *data,
1996 int (*fill_super)(struct super_block *, void *, int));
1997 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
1998 void generic_shutdown_super(struct super_block *sb);
1999 void kill_block_super(struct super_block *sb);
2000 void kill_anon_super(struct super_block *sb);
2001 void kill_litter_super(struct super_block *sb);
2002 void deactivate_super(struct super_block *sb);
2003 void deactivate_locked_super(struct super_block *sb);
2004 int set_anon_super(struct super_block *s, void *data);
2005 int get_anon_bdev(dev_t *);
2006 void free_anon_bdev(dev_t);
2007 struct super_block *sget(struct file_system_type *type,
2008 int (*test)(struct super_block *,void *),
2009 int (*set)(struct super_block *,void *),
2010 int flags, void *data);
2011 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
2012 const struct super_operations *ops,
2013 const struct dentry_operations *dops,
2016 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2017 #define fops_get(fops) \
2018 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2019 #define fops_put(fops) \
2020 do { if (fops) module_put((fops)->owner); } while(0)
2022 * This one is to be used *ONLY* from ->open() instances.
2023 * fops must be non-NULL, pinned down *and* module dependencies
2024 * should be sufficient to pin the caller down as well.
2026 #define replace_fops(f, fops) \
2028 struct file *__file = (f); \
2029 fops_put(__file->f_op); \
2030 BUG_ON(!(__file->f_op = (fops))); \
2033 extern int register_filesystem(struct file_system_type *);
2034 extern int unregister_filesystem(struct file_system_type *);
2035 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2036 #define kern_mount(type) kern_mount_data(type, NULL)
2037 extern void kern_unmount(struct vfsmount *mnt);
2038 extern int may_umount_tree(struct vfsmount *);
2039 extern int may_umount(struct vfsmount *);
2040 extern long do_mount(const char *, const char __user *,
2041 const char *, unsigned long, void *);
2042 extern struct vfsmount *collect_mounts(struct path *);
2043 extern void drop_collected_mounts(struct vfsmount *);
2044 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2046 extern int vfs_statfs(struct path *, struct kstatfs *);
2047 extern int user_statfs(const char __user *, struct kstatfs *);
2048 extern int fd_statfs(int, struct kstatfs *);
2049 extern int vfs_ustat(dev_t, struct kstatfs *);
2050 extern int freeze_super(struct super_block *super);
2051 extern int thaw_super(struct super_block *super);
2052 extern bool our_mnt(struct vfsmount *mnt);
2054 extern int current_umask(void);
2056 extern void ihold(struct inode * inode);
2057 extern void iput(struct inode *);
2058 extern int generic_update_time(struct inode *, struct timespec *, int);
2061 extern struct kobject *fs_kobj;
2063 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
2065 #define FLOCK_VERIFY_READ 1
2066 #define FLOCK_VERIFY_WRITE 2
2068 #ifdef CONFIG_FILE_LOCKING
2069 extern int locks_mandatory_locked(struct file *);
2070 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
2073 * Candidates for mandatory locking have the setgid bit set
2074 * but no group execute bit - an otherwise meaningless combination.
2077 static inline int __mandatory_lock(struct inode *ino)
2079 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2083 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2084 * otherwise these will be advisory locks
2087 static inline int mandatory_lock(struct inode *ino)
2089 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2092 static inline int locks_verify_locked(struct file *file)
2094 if (mandatory_lock(file_inode(file)))
2095 return locks_mandatory_locked(file);
2099 static inline int locks_verify_truncate(struct inode *inode,
2103 if (inode->i_flctx && mandatory_lock(inode))
2104 return locks_mandatory_area(
2105 FLOCK_VERIFY_WRITE, inode, filp,
2106 size < inode->i_size ? size : inode->i_size,
2107 (size < inode->i_size ? inode->i_size - size
2108 : size - inode->i_size)
2113 static inline int break_lease(struct inode *inode, unsigned int mode)
2116 * Since this check is lockless, we must ensure that any refcounts
2117 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2118 * could end up racing with tasks trying to set a new lease on this
2122 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2123 return __break_lease(inode, mode, FL_LEASE);
2127 static inline int break_deleg(struct inode *inode, unsigned int mode)
2130 * Since this check is lockless, we must ensure that any refcounts
2131 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2132 * could end up racing with tasks trying to set a new lease on this
2136 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2137 return __break_lease(inode, mode, FL_DELEG);
2141 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2145 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2146 if (ret == -EWOULDBLOCK && delegated_inode) {
2147 *delegated_inode = inode;
2153 static inline int break_deleg_wait(struct inode **delegated_inode)
2157 ret = break_deleg(*delegated_inode, O_WRONLY);
2158 iput(*delegated_inode);
2159 *delegated_inode = NULL;
2163 static inline int break_layout(struct inode *inode, bool wait)
2166 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2167 return __break_lease(inode,
2168 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2173 #else /* !CONFIG_FILE_LOCKING */
2174 static inline int locks_mandatory_locked(struct file *file)
2179 static inline int locks_mandatory_area(int rw, struct inode *inode,
2180 struct file *filp, loff_t offset,
2186 static inline int __mandatory_lock(struct inode *inode)
2191 static inline int mandatory_lock(struct inode *inode)
2196 static inline int locks_verify_locked(struct file *file)
2201 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2207 static inline int break_lease(struct inode *inode, unsigned int mode)
2212 static inline int break_deleg(struct inode *inode, unsigned int mode)
2217 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2222 static inline int break_deleg_wait(struct inode **delegated_inode)
2228 static inline int break_layout(struct inode *inode, bool wait)
2233 #endif /* CONFIG_FILE_LOCKING */
2238 const char *name; /* pointer to actual string */
2239 const __user char *uptr; /* original userland pointer */
2240 struct audit_names *aname;
2245 extern long vfs_truncate(struct path *, loff_t);
2246 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2248 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2250 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2252 extern struct file *file_open_name(struct filename *, int, umode_t);
2253 extern struct file *filp_open(const char *, int, umode_t);
2254 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2255 const char *, int, umode_t);
2256 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2257 extern int filp_close(struct file *, fl_owner_t id);
2259 extern struct filename *getname_flags(const char __user *, int, int *);
2260 extern struct filename *getname(const char __user *);
2261 extern struct filename *getname_kernel(const char *);
2262 extern void putname(struct filename *name);
2268 extern int finish_open(struct file *file, struct dentry *dentry,
2269 int (*open)(struct inode *, struct file *),
2271 extern int finish_no_open(struct file *file, struct dentry *dentry);
2275 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2278 extern void __init vfs_caches_init_early(void);
2279 extern void __init vfs_caches_init(void);
2281 extern struct kmem_cache *names_cachep;
2283 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2284 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2287 extern int register_blkdev(unsigned int, const char *);
2288 extern void unregister_blkdev(unsigned int, const char *);
2289 extern struct block_device *bdget(dev_t);
2290 extern struct block_device *bdgrab(struct block_device *bdev);
2291 extern void bd_set_size(struct block_device *, loff_t size);
2292 extern void bd_forget(struct inode *inode);
2293 extern void bdput(struct block_device *);
2294 extern void invalidate_bdev(struct block_device *);
2295 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2296 extern int sync_blockdev(struct block_device *bdev);
2297 extern void kill_bdev(struct block_device *);
2298 extern struct super_block *freeze_bdev(struct block_device *);
2299 extern void emergency_thaw_all(void);
2300 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2301 extern int fsync_bdev(struct block_device *);
2303 extern struct super_block *blockdev_superblock;
2305 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2307 return sb == blockdev_superblock;
2310 static inline void bd_forget(struct inode *inode) {}
2311 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2312 static inline void kill_bdev(struct block_device *bdev) {}
2313 static inline void invalidate_bdev(struct block_device *bdev) {}
2315 static inline struct super_block *freeze_bdev(struct block_device *sb)
2320 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2325 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2329 static inline int sb_is_blkdev_sb(struct super_block *sb)
2334 extern int sync_filesystem(struct super_block *);
2335 extern const struct file_operations def_blk_fops;
2336 extern const struct file_operations def_chr_fops;
2338 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2339 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2340 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2341 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2342 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2344 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2346 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2347 extern int __blkdev_reread_part(struct block_device *bdev);
2348 extern int blkdev_reread_part(struct block_device *bdev);
2351 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2352 extern void bd_unlink_disk_holder(struct block_device *bdev,
2353 struct gendisk *disk);
2355 static inline int bd_link_disk_holder(struct block_device *bdev,
2356 struct gendisk *disk)
2360 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2361 struct gendisk *disk)
2368 #define CHRDEV_MAJOR_HASH_SIZE 255
2369 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2370 extern int register_chrdev_region(dev_t, unsigned, const char *);
2371 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2372 unsigned int count, const char *name,
2373 const struct file_operations *fops);
2374 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2375 unsigned int count, const char *name);
2376 extern void unregister_chrdev_region(dev_t, unsigned);
2377 extern void chrdev_show(struct seq_file *,off_t);
2379 static inline int register_chrdev(unsigned int major, const char *name,
2380 const struct file_operations *fops)
2382 return __register_chrdev(major, 0, 256, name, fops);
2385 static inline void unregister_chrdev(unsigned int major, const char *name)
2387 __unregister_chrdev(major, 0, 256, name);
2390 /* fs/block_dev.c */
2391 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2392 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2395 #define BLKDEV_MAJOR_HASH_SIZE 255
2396 extern const char *__bdevname(dev_t, char *buffer);
2397 extern const char *bdevname(struct block_device *bdev, char *buffer);
2398 extern struct block_device *lookup_bdev(const char *);
2399 extern void blkdev_show(struct seq_file *,off_t);
2402 #define BLKDEV_MAJOR_HASH_SIZE 0
2405 extern void init_special_inode(struct inode *, umode_t, dev_t);
2407 /* Invalid inode operations -- fs/bad_inode.c */
2408 extern void make_bad_inode(struct inode *);
2409 extern int is_bad_inode(struct inode *);
2413 * return READ, READA, or WRITE
2415 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2418 * return data direction, READ or WRITE
2420 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2422 extern void check_disk_size_change(struct gendisk *disk,
2423 struct block_device *bdev);
2424 extern int revalidate_disk(struct gendisk *);
2425 extern int check_disk_change(struct block_device *);
2426 extern int __invalidate_device(struct block_device *, bool);
2427 extern int invalidate_partition(struct gendisk *, int);
2429 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2430 pgoff_t start, pgoff_t end);
2432 static inline void invalidate_remote_inode(struct inode *inode)
2434 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2435 S_ISLNK(inode->i_mode))
2436 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2438 extern int invalidate_inode_pages2(struct address_space *mapping);
2439 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2440 pgoff_t start, pgoff_t end);
2441 extern int write_inode_now(struct inode *, int);
2442 extern int filemap_fdatawrite(struct address_space *);
2443 extern int filemap_flush(struct address_space *);
2444 extern int filemap_fdatawait(struct address_space *);
2445 extern void filemap_fdatawait_keep_errors(struct address_space *);
2446 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2448 extern int filemap_write_and_wait(struct address_space *mapping);
2449 extern int filemap_write_and_wait_range(struct address_space *mapping,
2450 loff_t lstart, loff_t lend);
2451 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2452 loff_t start, loff_t end, int sync_mode);
2453 extern int filemap_fdatawrite_range(struct address_space *mapping,
2454 loff_t start, loff_t end);
2456 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2458 extern int vfs_fsync(struct file *file, int datasync);
2459 static inline int generic_write_sync(struct file *file, loff_t pos, loff_t count)
2461 if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host))
2463 return vfs_fsync_range(file, pos, pos + count - 1,
2464 (file->f_flags & __O_SYNC) ? 0 : 1);
2466 extern void emergency_sync(void);
2467 extern void emergency_remount(void);
2469 extern sector_t bmap(struct inode *, sector_t);
2471 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2472 extern int inode_permission(struct inode *, int);
2473 extern int __inode_permission(struct inode *, int);
2474 extern int generic_permission(struct inode *, int);
2475 extern int __check_sticky(struct inode *dir, struct inode *inode);
2477 static inline bool execute_ok(struct inode *inode)
2479 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2482 static inline void file_start_write(struct file *file)
2484 if (!S_ISREG(file_inode(file)->i_mode))
2486 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2489 static inline bool file_start_write_trylock(struct file *file)
2491 if (!S_ISREG(file_inode(file)->i_mode))
2493 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2496 static inline void file_end_write(struct file *file)
2498 if (!S_ISREG(file_inode(file)->i_mode))
2500 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2504 * get_write_access() gets write permission for a file.
2505 * put_write_access() releases this write permission.
2506 * This is used for regular files.
2507 * We cannot support write (and maybe mmap read-write shared) accesses and
2508 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2509 * can have the following values:
2510 * 0: no writers, no VM_DENYWRITE mappings
2511 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2512 * > 0: (i_writecount) users are writing to the file.
2514 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2515 * except for the cases where we don't hold i_writecount yet. Then we need to
2516 * use {get,deny}_write_access() - these functions check the sign and refuse
2517 * to do the change if sign is wrong.
2519 static inline int get_write_access(struct inode *inode)
2521 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2523 static inline int deny_write_access(struct file *file)
2525 struct inode *inode = file_inode(file);
2526 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2528 static inline void put_write_access(struct inode * inode)
2530 atomic_dec(&inode->i_writecount);
2532 static inline void allow_write_access(struct file *file)
2535 atomic_inc(&file_inode(file)->i_writecount);
2537 static inline bool inode_is_open_for_write(const struct inode *inode)
2539 return atomic_read(&inode->i_writecount) > 0;
2543 static inline void i_readcount_dec(struct inode *inode)
2545 BUG_ON(!atomic_read(&inode->i_readcount));
2546 atomic_dec(&inode->i_readcount);
2548 static inline void i_readcount_inc(struct inode *inode)
2550 atomic_inc(&inode->i_readcount);
2553 static inline void i_readcount_dec(struct inode *inode)
2557 static inline void i_readcount_inc(struct inode *inode)
2562 extern int do_pipe_flags(int *, int);
2564 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2565 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2566 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
2567 extern struct file * open_exec(const char *);
2569 /* fs/dcache.c -- generic fs support functions */
2570 extern int is_subdir(struct dentry *, struct dentry *);
2571 extern int path_is_under(struct path *, struct path *);
2573 extern char *file_path(struct file *, char *, int);
2575 #include <linux/err.h>
2577 /* needed for stackable file system support */
2578 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2580 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2582 extern int inode_init_always(struct super_block *, struct inode *);
2583 extern void inode_init_once(struct inode *);
2584 extern void address_space_init_once(struct address_space *mapping);
2585 extern struct inode * igrab(struct inode *);
2586 extern ino_t iunique(struct super_block *, ino_t);
2587 extern int inode_needs_sync(struct inode *inode);
2588 extern int generic_delete_inode(struct inode *inode);
2589 static inline int generic_drop_inode(struct inode *inode)
2591 return !inode->i_nlink || inode_unhashed(inode);
2594 extern struct inode *ilookup5_nowait(struct super_block *sb,
2595 unsigned long hashval, int (*test)(struct inode *, void *),
2597 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2598 int (*test)(struct inode *, void *), void *data);
2599 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2601 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2602 extern struct inode * iget_locked(struct super_block *, unsigned long);
2603 extern struct inode *find_inode_nowait(struct super_block *,
2605 int (*match)(struct inode *,
2606 unsigned long, void *),
2608 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2609 extern int insert_inode_locked(struct inode *);
2610 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2611 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2613 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2615 extern void unlock_new_inode(struct inode *);
2616 extern unsigned int get_next_ino(void);
2618 extern void __iget(struct inode * inode);
2619 extern void iget_failed(struct inode *);
2620 extern void clear_inode(struct inode *);
2621 extern void __destroy_inode(struct inode *);
2622 extern struct inode *new_inode_pseudo(struct super_block *sb);
2623 extern struct inode *new_inode(struct super_block *sb);
2624 extern void free_inode_nonrcu(struct inode *inode);
2625 extern int should_remove_suid(struct dentry *);
2626 extern int file_remove_privs(struct file *);
2627 extern int dentry_needs_remove_privs(struct dentry *dentry);
2628 static inline int file_needs_remove_privs(struct file *file)
2630 return dentry_needs_remove_privs(file->f_path.dentry);
2633 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2634 static inline void insert_inode_hash(struct inode *inode)
2636 __insert_inode_hash(inode, inode->i_ino);
2639 extern void __remove_inode_hash(struct inode *);
2640 static inline void remove_inode_hash(struct inode *inode)
2642 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2643 __remove_inode_hash(inode);
2646 extern void inode_sb_list_add(struct inode *inode);
2649 extern blk_qc_t submit_bio(int, struct bio *);
2650 extern int bdev_read_only(struct block_device *);
2652 extern int set_blocksize(struct block_device *, int);
2653 extern int sb_set_blocksize(struct super_block *, int);
2654 extern int sb_min_blocksize(struct super_block *, int);
2656 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2657 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2658 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2659 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2660 extern ssize_t __generic_file_write_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_direct_write(struct kiocb *, struct iov_iter *, loff_t);
2663 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2665 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos);
2666 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos);
2668 /* fs/block_dev.c */
2669 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2670 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2671 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2673 extern void block_sync_page(struct page *page);
2676 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2677 struct pipe_inode_info *, size_t, unsigned int);
2678 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2679 struct pipe_inode_info *, size_t, unsigned int);
2680 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2681 struct file *, loff_t *, size_t, unsigned int);
2682 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2683 struct file *out, loff_t *, size_t len, unsigned int flags);
2684 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2685 loff_t *opos, size_t len, unsigned int flags);
2689 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2690 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2691 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2692 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2693 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2694 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2695 int whence, loff_t maxsize, loff_t eof);
2696 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2697 int whence, loff_t size);
2698 extern int generic_file_open(struct inode * inode, struct file * filp);
2699 extern int nonseekable_open(struct inode * inode, struct file * filp);
2702 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2703 loff_t file_offset);
2706 /* need locking between buffered and direct access */
2709 /* filesystem does not support filling holes */
2710 DIO_SKIP_HOLES = 0x02,
2712 /* filesystem can handle aio writes beyond i_size */
2713 DIO_ASYNC_EXTEND = 0x04,
2715 /* inode/fs/bdev does not need truncate protection */
2716 DIO_SKIP_DIO_COUNT = 0x08,
2719 void dio_end_io(struct bio *bio, int error);
2721 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2722 struct block_device *bdev, struct iov_iter *iter,
2723 loff_t offset, get_block_t get_block,
2724 dio_iodone_t end_io, dio_submit_t submit_io,
2727 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2728 struct inode *inode,
2729 struct iov_iter *iter, loff_t offset,
2730 get_block_t get_block)
2732 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
2733 offset, get_block, NULL, NULL,
2734 DIO_LOCKING | DIO_SKIP_HOLES);
2738 void inode_dio_wait(struct inode *inode);
2741 * inode_dio_begin - signal start of a direct I/O requests
2742 * @inode: inode the direct I/O happens on
2744 * This is called once we've finished processing a direct I/O request,
2745 * and is used to wake up callers waiting for direct I/O to be quiesced.
2747 static inline void inode_dio_begin(struct inode *inode)
2749 atomic_inc(&inode->i_dio_count);
2753 * inode_dio_end - signal finish of a direct I/O requests
2754 * @inode: inode the direct I/O happens on
2756 * This is called once we've finished processing a direct I/O request,
2757 * and is used to wake up callers waiting for direct I/O to be quiesced.
2759 static inline void inode_dio_end(struct inode *inode)
2761 if (atomic_dec_and_test(&inode->i_dio_count))
2762 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
2765 extern void inode_set_flags(struct inode *inode, unsigned int flags,
2768 extern const struct file_operations generic_ro_fops;
2770 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2772 extern int readlink_copy(char __user *, int, const char *);
2773 extern int page_readlink(struct dentry *, char __user *, int);
2774 extern const char *page_follow_link_light(struct dentry *, void **);
2775 extern void page_put_link(struct inode *, void *);
2776 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2778 extern int page_symlink(struct inode *inode, const char *symname, int len);
2779 extern const struct inode_operations page_symlink_inode_operations;
2780 extern void kfree_put_link(struct inode *, void *);
2781 extern void free_page_put_link(struct inode *, void *);
2782 extern int generic_readlink(struct dentry *, char __user *, int);
2783 extern void generic_fillattr(struct inode *, struct kstat *);
2784 int vfs_getattr_nosec(struct path *path, struct kstat *stat);
2785 extern int vfs_getattr(struct path *, struct kstat *);
2786 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2787 void inode_add_bytes(struct inode *inode, loff_t bytes);
2788 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2789 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2790 loff_t inode_get_bytes(struct inode *inode);
2791 void inode_set_bytes(struct inode *inode, loff_t bytes);
2792 const char *simple_follow_link(struct dentry *, void **);
2793 extern const struct inode_operations simple_symlink_inode_operations;
2795 extern int iterate_dir(struct file *, struct dir_context *);
2797 extern int vfs_stat(const char __user *, struct kstat *);
2798 extern int vfs_lstat(const char __user *, struct kstat *);
2799 extern int vfs_fstat(unsigned int, struct kstat *);
2800 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2802 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2804 extern int __generic_block_fiemap(struct inode *inode,
2805 struct fiemap_extent_info *fieinfo,
2806 loff_t start, loff_t len,
2807 get_block_t *get_block);
2808 extern int generic_block_fiemap(struct inode *inode,
2809 struct fiemap_extent_info *fieinfo, u64 start,
2810 u64 len, get_block_t *get_block);
2812 extern void get_filesystem(struct file_system_type *fs);
2813 extern void put_filesystem(struct file_system_type *fs);
2814 extern struct file_system_type *get_fs_type(const char *name);
2815 extern struct super_block *get_super(struct block_device *);
2816 extern struct super_block *get_super_thawed(struct block_device *);
2817 extern struct super_block *get_active_super(struct block_device *bdev);
2818 extern void drop_super(struct super_block *sb);
2819 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2820 extern void iterate_supers_type(struct file_system_type *,
2821 void (*)(struct super_block *, void *), void *);
2823 extern int dcache_dir_open(struct inode *, struct file *);
2824 extern int dcache_dir_close(struct inode *, struct file *);
2825 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2826 extern int dcache_readdir(struct file *, struct dir_context *);
2827 extern int simple_setattr(struct dentry *, struct iattr *);
2828 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2829 extern int simple_statfs(struct dentry *, struct kstatfs *);
2830 extern int simple_open(struct inode *inode, struct file *file);
2831 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2832 extern int simple_unlink(struct inode *, struct dentry *);
2833 extern int simple_rmdir(struct inode *, struct dentry *);
2834 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2835 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2836 extern int simple_empty(struct dentry *);
2837 extern int simple_readpage(struct file *file, struct page *page);
2838 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2839 loff_t pos, unsigned len, unsigned flags,
2840 struct page **pagep, void **fsdata);
2841 extern int simple_write_end(struct file *file, struct address_space *mapping,
2842 loff_t pos, unsigned len, unsigned copied,
2843 struct page *page, void *fsdata);
2844 extern int always_delete_dentry(const struct dentry *);
2845 extern struct inode *alloc_anon_inode(struct super_block *);
2846 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
2847 extern const struct dentry_operations simple_dentry_operations;
2849 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2850 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2851 extern const struct file_operations simple_dir_operations;
2852 extern const struct inode_operations simple_dir_inode_operations;
2853 extern void make_empty_dir_inode(struct inode *inode);
2854 extern bool is_empty_dir_inode(struct inode *inode);
2855 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2856 struct dentry *d_alloc_name(struct dentry *, const char *);
2857 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2858 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2859 extern void simple_release_fs(struct vfsmount **mount, int *count);
2861 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2862 loff_t *ppos, const void *from, size_t available);
2863 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2864 const void __user *from, size_t count);
2866 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
2867 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2869 extern int generic_check_addressable(unsigned, u64);
2871 #ifdef CONFIG_MIGRATION
2872 extern int buffer_migrate_page(struct address_space *,
2873 struct page *, struct page *,
2876 #define buffer_migrate_page NULL
2879 extern int inode_change_ok(const struct inode *, struct iattr *);
2880 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2881 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2883 extern int file_update_time(struct file *file);
2885 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2886 extern void save_mount_options(struct super_block *sb, char *options);
2887 extern void replace_mount_options(struct super_block *sb, char *options);
2889 static inline bool io_is_direct(struct file *filp)
2891 return (filp->f_flags & O_DIRECT) || IS_DAX(file_inode(filp));
2894 static inline int iocb_flags(struct file *file)
2897 if (file->f_flags & O_APPEND)
2899 if (io_is_direct(file))
2904 static inline ino_t parent_ino(struct dentry *dentry)
2909 * Don't strictly need d_lock here? If the parent ino could change
2910 * then surely we'd have a deeper race in the caller?
2912 spin_lock(&dentry->d_lock);
2913 res = dentry->d_parent->d_inode->i_ino;
2914 spin_unlock(&dentry->d_lock);
2918 /* Transaction based IO helpers */
2921 * An argresp is stored in an allocated page and holds the
2922 * size of the argument or response, along with its content
2924 struct simple_transaction_argresp {
2929 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2931 char *simple_transaction_get(struct file *file, const char __user *buf,
2933 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2934 size_t size, loff_t *pos);
2935 int simple_transaction_release(struct inode *inode, struct file *file);
2937 void simple_transaction_set(struct file *file, size_t n);
2940 * simple attribute files
2942 * These attributes behave similar to those in sysfs:
2944 * Writing to an attribute immediately sets a value, an open file can be
2945 * written to multiple times.
2947 * Reading from an attribute creates a buffer from the value that might get
2948 * read with multiple read calls. When the attribute has been read
2949 * completely, no further read calls are possible until the file is opened
2952 * All attributes contain a text representation of a numeric value
2953 * that are accessed with the get() and set() functions.
2955 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2956 static int __fops ## _open(struct inode *inode, struct file *file) \
2958 __simple_attr_check_format(__fmt, 0ull); \
2959 return simple_attr_open(inode, file, __get, __set, __fmt); \
2961 static const struct file_operations __fops = { \
2962 .owner = THIS_MODULE, \
2963 .open = __fops ## _open, \
2964 .release = simple_attr_release, \
2965 .read = simple_attr_read, \
2966 .write = simple_attr_write, \
2967 .llseek = generic_file_llseek, \
2970 static inline __printf(1, 2)
2971 void __simple_attr_check_format(const char *fmt, ...)
2973 /* don't do anything, just let the compiler check the arguments; */
2976 int simple_attr_open(struct inode *inode, struct file *file,
2977 int (*get)(void *, u64 *), int (*set)(void *, u64),
2979 int simple_attr_release(struct inode *inode, struct file *file);
2980 ssize_t simple_attr_read(struct file *file, char __user *buf,
2981 size_t len, loff_t *ppos);
2982 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2983 size_t len, loff_t *ppos);
2986 int proc_nr_files(struct ctl_table *table, int write,
2987 void __user *buffer, size_t *lenp, loff_t *ppos);
2988 int proc_nr_dentry(struct ctl_table *table, int write,
2989 void __user *buffer, size_t *lenp, loff_t *ppos);
2990 int proc_nr_inodes(struct ctl_table *table, int write,
2991 void __user *buffer, size_t *lenp, loff_t *ppos);
2992 int __init get_filesystem_list(char *buf);
2994 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2995 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2997 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2998 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2999 (flag & __FMODE_NONOTIFY)))
3001 static inline int is_sxid(umode_t mode)
3003 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3006 static inline int check_sticky(struct inode *dir, struct inode *inode)
3008 if (!(dir->i_mode & S_ISVTX))
3011 return __check_sticky(dir, inode);
3014 static inline void inode_has_no_xattr(struct inode *inode)
3016 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
3017 inode->i_flags |= S_NOSEC;
3020 static inline bool is_root_inode(struct inode *inode)
3022 return inode == inode->i_sb->s_root->d_inode;
3025 static inline bool dir_emit(struct dir_context *ctx,
3026 const char *name, int namelen,
3027 u64 ino, unsigned type)
3029 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3031 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3033 return ctx->actor(ctx, ".", 1, ctx->pos,
3034 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3036 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3038 return ctx->actor(ctx, "..", 2, ctx->pos,
3039 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3041 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3043 if (ctx->pos == 0) {
3044 if (!dir_emit_dot(file, ctx))
3048 if (ctx->pos == 1) {
3049 if (!dir_emit_dotdot(file, ctx))
3055 static inline bool dir_relax(struct inode *inode)
3057 inode_unlock(inode);
3059 return !IS_DEADDIR(inode);
3062 extern bool path_noexec(const struct path *path);
3064 #endif /* _LINUX_FS_H */