5 #include <linux/linkage.h>
6 #include <linux/wait.h>
7 #include <linux/kdev_t.h>
8 #include <linux/dcache.h>
9 #include <linux/path.h>
10 #include <linux/stat.h>
11 #include <linux/cache.h>
12 #include <linux/list.h>
13 #include <linux/list_lru.h>
14 #include <linux/llist.h>
15 #include <linux/radix-tree.h>
16 #include <linux/rbtree.h>
17 #include <linux/init.h>
18 #include <linux/pid.h>
19 #include <linux/bug.h>
20 #include <linux/mutex.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>
33 #include <asm/byteorder.h>
34 #include <uapi/linux/fs.h>
36 struct export_operations;
42 struct pipe_inode_info;
43 struct poll_table_struct;
45 struct vm_area_struct;
48 struct swap_info_struct;
50 struct workqueue_struct;
52 extern void __init inode_init(void);
53 extern void __init inode_init_early(void);
54 extern void __init files_init(unsigned long);
56 extern struct files_stat_struct files_stat;
57 extern unsigned long get_max_files(void);
58 extern int sysctl_nr_open;
59 extern struct inodes_stat_t inodes_stat;
60 extern int leases_enable, lease_break_time;
61 extern int sysctl_protected_symlinks;
62 extern int sysctl_protected_hardlinks;
65 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
66 struct buffer_head *bh_result, int create);
67 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
68 ssize_t bytes, void *private);
70 #define MAY_EXEC 0x00000001
71 #define MAY_WRITE 0x00000002
72 #define MAY_READ 0x00000004
73 #define MAY_APPEND 0x00000008
74 #define MAY_ACCESS 0x00000010
75 #define MAY_OPEN 0x00000020
76 #define MAY_CHDIR 0x00000040
77 /* called from RCU mode, don't block */
78 #define MAY_NOT_BLOCK 0x00000080
81 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
82 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
85 /* file is open for reading */
86 #define FMODE_READ ((__force fmode_t)0x1)
87 /* file is open for writing */
88 #define FMODE_WRITE ((__force fmode_t)0x2)
89 /* file is seekable */
90 #define FMODE_LSEEK ((__force fmode_t)0x4)
91 /* file can be accessed using pread */
92 #define FMODE_PREAD ((__force fmode_t)0x8)
93 /* file can be accessed using pwrite */
94 #define FMODE_PWRITE ((__force fmode_t)0x10)
95 /* File is opened for execution with sys_execve / sys_uselib */
96 #define FMODE_EXEC ((__force fmode_t)0x20)
97 /* File is opened with O_NDELAY (only set for block devices) */
98 #define FMODE_NDELAY ((__force fmode_t)0x40)
99 /* File is opened with O_EXCL (only set for block devices) */
100 #define FMODE_EXCL ((__force fmode_t)0x80)
101 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
102 (specialy hack for floppy.c) */
103 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
104 /* 32bit hashes as llseek() offset (for directories) */
105 #define FMODE_32BITHASH ((__force fmode_t)0x200)
106 /* 64bit hashes as llseek() offset (for directories) */
107 #define FMODE_64BITHASH ((__force fmode_t)0x400)
110 * Don't update ctime and mtime.
112 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
113 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
115 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
117 /* Expect random access pattern */
118 #define FMODE_RANDOM ((__force fmode_t)0x1000)
120 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
121 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
123 /* File is opened with O_PATH; almost nothing can be done with it */
124 #define FMODE_PATH ((__force fmode_t)0x4000)
126 /* File was opened by fanotify and shouldn't generate fanotify events */
127 #define FMODE_NONOTIFY ((__force fmode_t)0x1000000)
130 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
131 * that indicates that they should check the contents of the iovec are
132 * valid, but not check the memory that the iovec elements
135 #define CHECK_IOVEC_ONLY -1
138 * The below are the various read and write types that we support. Some of
139 * them include behavioral modifiers that send information down to the
140 * block layer and IO scheduler. Terminology:
142 * The block layer uses device plugging to defer IO a little bit, in
143 * the hope that we will see more IO very shortly. This increases
144 * coalescing of adjacent IO and thus reduces the number of IOs we
145 * have to send to the device. It also allows for better queuing,
146 * if the IO isn't mergeable. If the caller is going to be waiting
147 * for the IO, then he must ensure that the device is unplugged so
148 * that the IO is dispatched to the driver.
150 * All IO is handled async in Linux. This is fine for background
151 * writes, but for reads or writes that someone waits for completion
152 * on, we want to notify the block layer and IO scheduler so that they
153 * know about it. That allows them to make better scheduling
154 * decisions. So when the below references 'sync' and 'async', it
155 * is referencing this priority hint.
157 * With that in mind, the available types are:
159 * READ A normal read operation. Device will be plugged.
160 * READ_SYNC A synchronous read. Device is not plugged, caller can
161 * immediately wait on this read without caring about
163 * READA Used for read-ahead operations. Lower priority, and the
164 * block layer could (in theory) choose to ignore this
165 * request if it runs into resource problems.
166 * WRITE A normal async write. Device will be plugged.
167 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
168 * the hint that someone will be waiting on this IO
169 * shortly. The write equivalent of READ_SYNC.
170 * WRITE_ODIRECT Special case write for O_DIRECT only.
171 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
172 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
173 * non-volatile media on completion.
174 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
175 * by a cache flush and data is guaranteed to be on
176 * non-volatile media on completion.
179 #define RW_MASK REQ_WRITE
180 #define RWA_MASK REQ_RAHEAD
183 #define WRITE RW_MASK
184 #define READA RWA_MASK
186 #define READ_SYNC (READ | REQ_SYNC)
187 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
188 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
189 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
190 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
191 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
194 * Attribute flags. These should be or-ed together to figure out what
197 #define ATTR_MODE (1 << 0)
198 #define ATTR_UID (1 << 1)
199 #define ATTR_GID (1 << 2)
200 #define ATTR_SIZE (1 << 3)
201 #define ATTR_ATIME (1 << 4)
202 #define ATTR_MTIME (1 << 5)
203 #define ATTR_CTIME (1 << 6)
204 #define ATTR_ATIME_SET (1 << 7)
205 #define ATTR_MTIME_SET (1 << 8)
206 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
207 #define ATTR_ATTR_FLAG (1 << 10)
208 #define ATTR_KILL_SUID (1 << 11)
209 #define ATTR_KILL_SGID (1 << 12)
210 #define ATTR_FILE (1 << 13)
211 #define ATTR_KILL_PRIV (1 << 14)
212 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
213 #define ATTR_TIMES_SET (1 << 16)
216 * This is the Inode Attributes structure, used for notify_change(). It
217 * uses the above definitions as flags, to know which values have changed.
218 * Also, in this manner, a Filesystem can look at only the values it cares
219 * about. Basically, these are the attributes that the VFS layer can
220 * request to change from the FS layer.
222 * Derek Atkins <warlord@MIT.EDU> 94-10-20
225 unsigned int ia_valid;
230 struct timespec ia_atime;
231 struct timespec ia_mtime;
232 struct timespec ia_ctime;
235 * Not an attribute, but an auxiliary info for filesystems wanting to
236 * implement an ftruncate() like method. NOTE: filesystem should
237 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
239 struct file *ia_file;
243 * Includes for diskquotas.
245 #include <linux/quota.h>
248 * enum positive_aop_returns - aop return codes with specific semantics
250 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
251 * completed, that the page is still locked, and
252 * should be considered active. The VM uses this hint
253 * to return the page to the active list -- it won't
254 * be a candidate for writeback again in the near
255 * future. Other callers must be careful to unlock
256 * the page if they get this return. Returned by
259 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
260 * unlocked it and the page might have been truncated.
261 * The caller should back up to acquiring a new page and
262 * trying again. The aop will be taking reasonable
263 * precautions not to livelock. If the caller held a page
264 * reference, it should drop it before retrying. Returned
267 * address_space_operation functions return these large constants to indicate
268 * special semantics to the caller. These are much larger than the bytes in a
269 * page to allow for functions that return the number of bytes operated on in a
273 enum positive_aop_returns {
274 AOP_WRITEPAGE_ACTIVATE = 0x80000,
275 AOP_TRUNCATED_PAGE = 0x80001,
278 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
279 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
280 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
281 * helper code (eg buffer layer)
282 * to clear GFP_FS from alloc */
285 * oh the beauties of C type declarations.
288 struct address_space;
289 struct writeback_control;
292 struct iov_iter_ops *ops;
294 unsigned long nr_segs;
299 struct iov_iter_ops {
300 size_t (*ii_copy_to_user_atomic)(struct page *, struct iov_iter *,
301 unsigned long, size_t);
302 size_t (*ii_copy_to_user)(struct page *, struct iov_iter *,
303 unsigned long, size_t, int);
304 size_t (*ii_copy_from_user_atomic)(struct page *, struct iov_iter *,
305 unsigned long, size_t);
306 size_t (*ii_copy_from_user)(struct page *, struct iov_iter *,
307 unsigned long, size_t);
308 void (*ii_advance)(struct iov_iter *, size_t);
309 int (*ii_fault_in_readable)(struct iov_iter *, size_t);
310 size_t (*ii_single_seg_count)(const struct iov_iter *);
311 int (*ii_shorten)(struct iov_iter *, size_t);
314 static inline size_t iov_iter_copy_to_user_atomic(struct page *page,
315 struct iov_iter *i, unsigned long offset, size_t bytes)
317 return i->ops->ii_copy_to_user_atomic(page, i, offset, bytes);
319 static inline size_t __iov_iter_copy_to_user(struct page *page,
320 struct iov_iter *i, unsigned long offset, size_t bytes)
322 return i->ops->ii_copy_to_user(page, i, offset, bytes, 0);
324 static inline size_t iov_iter_copy_to_user(struct page *page,
325 struct iov_iter *i, unsigned long offset, size_t bytes)
327 return i->ops->ii_copy_to_user(page, i, offset, bytes, 1);
329 static inline size_t iov_iter_copy_from_user_atomic(struct page *page,
330 struct iov_iter *i, unsigned long offset, size_t bytes)
332 return i->ops->ii_copy_from_user_atomic(page, i, offset, bytes);
334 static inline size_t iov_iter_copy_from_user(struct page *page,
335 struct iov_iter *i, unsigned long offset, size_t bytes)
337 return i->ops->ii_copy_from_user(page, i, offset, bytes);
339 static inline void iov_iter_advance(struct iov_iter *i, size_t bytes)
341 return i->ops->ii_advance(i, bytes);
343 static inline int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes)
345 return i->ops->ii_fault_in_readable(i, bytes);
347 static inline size_t iov_iter_single_seg_count(const struct iov_iter *i)
349 return i->ops->ii_single_seg_count(i);
351 static inline int iov_iter_shorten(struct iov_iter *i, size_t count)
353 return i->ops->ii_shorten(i, count);
357 extern struct iov_iter_ops ii_bvec_ops;
360 static inline void iov_iter_init_bvec(struct iov_iter *i,
361 struct bio_vec *bvec,
362 unsigned long nr_segs,
363 size_t count, size_t written)
365 i->ops = &ii_bvec_ops;
366 i->data = (unsigned long)bvec;
367 i->nr_segs = nr_segs;
369 i->count = count + written;
371 iov_iter_advance(i, written);
374 static inline int iov_iter_has_bvec(struct iov_iter *i)
376 return i->ops == &ii_bvec_ops;
379 static inline struct bio_vec *iov_iter_bvec(struct iov_iter *i)
381 BUG_ON(!iov_iter_has_bvec(i));
382 return (struct bio_vec *)i->data;
386 extern struct iov_iter_ops ii_iovec_ops;
388 static inline void iov_iter_init(struct iov_iter *i,
389 const struct iovec *iov, unsigned long nr_segs,
390 size_t count, size_t written)
392 i->ops = &ii_iovec_ops;
393 i->data = (unsigned long)iov;
394 i->nr_segs = nr_segs;
396 i->count = count + written;
398 iov_iter_advance(i, written);
401 static inline int iov_iter_has_iovec(struct iov_iter *i)
403 return i->ops == &ii_iovec_ops;
406 static inline struct iovec *iov_iter_iovec(struct iov_iter *i)
408 BUG_ON(!iov_iter_has_iovec(i));
409 return (struct iovec *)i->data;
412 static inline size_t iov_iter_count(struct iov_iter *i)
418 * "descriptor" for what we're up to with a read.
419 * This allows us to use the same read code yet
420 * have multiple different users of the data that
421 * we read from a file.
423 * The simplest case just copies the data to user
436 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
437 unsigned long, unsigned long);
439 struct address_space_operations {
440 int (*writepage)(struct page *page, struct writeback_control *wbc);
441 int (*readpage)(struct file *, struct page *);
443 /* Write back some dirty pages from this mapping. */
444 int (*writepages)(struct address_space *, struct writeback_control *);
446 /* Set a page dirty. Return true if this dirtied it */
447 int (*set_page_dirty)(struct page *page);
449 int (*readpages)(struct file *filp, struct address_space *mapping,
450 struct list_head *pages, unsigned nr_pages);
452 int (*write_begin)(struct file *, struct address_space *mapping,
453 loff_t pos, unsigned len, unsigned flags,
454 struct page **pagep, void **fsdata);
455 int (*write_end)(struct file *, struct address_space *mapping,
456 loff_t pos, unsigned len, unsigned copied,
457 struct page *page, void *fsdata);
459 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
460 sector_t (*bmap)(struct address_space *, sector_t);
461 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
462 int (*releasepage) (struct page *, gfp_t);
463 void (*freepage)(struct page *);
464 ssize_t (*direct_IO)(int, struct kiocb *, struct iov_iter *iter,
466 int (*get_xip_mem)(struct address_space *, pgoff_t, int,
467 void **, unsigned long *);
469 * migrate the contents of a page to the specified target. If
470 * migrate_mode is MIGRATE_ASYNC, it must not block.
472 int (*migratepage) (struct address_space *,
473 struct page *, struct page *, enum migrate_mode);
474 int (*launder_page) (struct page *);
475 int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
477 void (*is_dirty_writeback) (struct page *, bool *, bool *);
478 int (*error_remove_page)(struct address_space *, struct page *);
480 /* swapfile support */
481 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
483 void (*swap_deactivate)(struct file *file);
486 extern const struct address_space_operations empty_aops;
489 * pagecache_write_begin/pagecache_write_end must be used by general code
490 * to write into the pagecache.
492 int pagecache_write_begin(struct file *, struct address_space *mapping,
493 loff_t pos, unsigned len, unsigned flags,
494 struct page **pagep, void **fsdata);
496 int pagecache_write_end(struct file *, struct address_space *mapping,
497 loff_t pos, unsigned len, unsigned copied,
498 struct page *page, void *fsdata);
500 struct backing_dev_info;
501 struct address_space {
502 struct inode *host; /* owner: inode, block_device */
503 struct radix_tree_root page_tree; /* radix tree of all pages */
504 spinlock_t tree_lock; /* and lock protecting it */
505 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
506 struct rb_root i_mmap; /* tree of private and shared mappings */
507 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
508 struct mutex i_mmap_mutex; /* protect tree, count, list */
509 /* Protected by tree_lock together with the radix tree */
510 unsigned long nrpages; /* number of total pages */
511 pgoff_t writeback_index;/* writeback starts here */
512 const struct address_space_operations *a_ops; /* methods */
513 unsigned long flags; /* error bits/gfp mask */
514 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
515 spinlock_t private_lock; /* for use by the address_space */
516 struct list_head private_list; /* ditto */
517 void *private_data; /* ditto */
518 } __attribute__((aligned(sizeof(long))));
520 * On most architectures that alignment is already the case; but
521 * must be enforced here for CRIS, to let the least significant bit
522 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
524 struct request_queue;
526 struct block_device {
527 dev_t bd_dev; /* not a kdev_t - it's a search key */
529 struct inode * bd_inode; /* will die */
530 struct super_block * bd_super;
531 struct mutex bd_mutex; /* open/close mutex */
532 struct list_head bd_inodes;
536 bool bd_write_holder;
538 struct list_head bd_holder_disks;
540 struct block_device * bd_contains;
541 unsigned bd_block_size;
542 struct hd_struct * bd_part;
543 /* number of times partitions within this device have been opened. */
544 unsigned bd_part_count;
546 struct gendisk * bd_disk;
547 struct request_queue * bd_queue;
548 struct list_head bd_list;
550 * Private data. You must have bd_claim'ed the block_device
551 * to use this. NOTE: bd_claim allows an owner to claim
552 * the same device multiple times, the owner must take special
553 * care to not mess up bd_private for that case.
555 unsigned long bd_private;
557 /* The counter of freeze processes */
558 int bd_fsfreeze_count;
559 /* Mutex for freeze */
560 struct mutex bd_fsfreeze_mutex;
564 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
567 #define PAGECACHE_TAG_DIRTY 0
568 #define PAGECACHE_TAG_WRITEBACK 1
569 #define PAGECACHE_TAG_TOWRITE 2
571 int mapping_tagged(struct address_space *mapping, int tag);
574 * Might pages of this file be mapped into userspace?
576 static inline int mapping_mapped(struct address_space *mapping)
578 return !RB_EMPTY_ROOT(&mapping->i_mmap) ||
579 !list_empty(&mapping->i_mmap_nonlinear);
583 * Might pages of this file have been modified in userspace?
584 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
585 * marks vma as VM_SHARED if it is shared, and the file was opened for
586 * writing i.e. vma may be mprotected writable even if now readonly.
588 static inline int mapping_writably_mapped(struct address_space *mapping)
590 return mapping->i_mmap_writable != 0;
594 * Use sequence counter to get consistent i_size on 32-bit processors.
596 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
597 #include <linux/seqlock.h>
598 #define __NEED_I_SIZE_ORDERED
599 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
601 #define i_size_ordered_init(inode) do { } while (0)
605 #define ACL_NOT_CACHED ((void *)(-1))
607 #define IOP_FASTPERM 0x0001
608 #define IOP_LOOKUP 0x0002
609 #define IOP_NOFOLLOW 0x0004
612 * Keep mostly read-only and often accessed (especially for
613 * the RCU path lookup and 'stat' data) fields at the beginning
614 * of the 'struct inode'
618 unsigned short i_opflags;
621 unsigned int i_flags;
623 #ifdef CONFIG_FS_POSIX_ACL
624 struct posix_acl *i_acl;
625 struct posix_acl *i_default_acl;
628 const struct inode_operations *i_op;
629 struct super_block *i_sb;
630 struct address_space *i_mapping;
632 #ifdef CONFIG_SECURITY
636 /* Stat data, not accessed from path walking */
639 * Filesystems may only read i_nlink directly. They shall use the
640 * following functions for modification:
642 * (set|clear|inc|drop)_nlink
643 * inode_(inc|dec)_link_count
646 const unsigned int i_nlink;
647 unsigned int __i_nlink;
651 struct timespec i_atime;
652 struct timespec i_mtime;
653 struct timespec i_ctime;
654 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
655 unsigned short i_bytes;
656 unsigned int i_blkbits;
659 #ifdef __NEED_I_SIZE_ORDERED
660 seqcount_t i_size_seqcount;
664 unsigned long i_state;
665 struct mutex i_mutex;
667 unsigned long dirtied_when; /* jiffies of first dirtying */
669 struct hlist_node i_hash;
670 struct list_head i_wb_list; /* backing dev IO list */
671 struct list_head i_lru; /* inode LRU list */
672 struct list_head i_sb_list;
674 struct hlist_head i_dentry;
675 struct rcu_head i_rcu;
679 atomic_t i_dio_count;
680 atomic_t i_writecount;
681 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
682 struct file_lock *i_flock;
683 struct address_space i_data;
685 struct dquot *i_dquot[MAXQUOTAS];
687 struct list_head i_devices;
689 struct pipe_inode_info *i_pipe;
690 struct block_device *i_bdev;
696 #ifdef CONFIG_FSNOTIFY
697 __u32 i_fsnotify_mask; /* all events this inode cares about */
698 struct hlist_head i_fsnotify_marks;
702 atomic_t i_readcount; /* struct files open RO */
704 void *i_private; /* fs or device private pointer */
707 static inline int inode_unhashed(struct inode *inode)
709 return hlist_unhashed(&inode->i_hash);
713 * inode->i_mutex nesting subclasses for the lock validator:
715 * 0: the object of the current VFS operation
720 * The locking order between these classes is
721 * parent -> child -> normal -> xattr -> quota
723 enum inode_i_mutex_lock_class
733 * NOTE: in a 32bit arch with a preemptable kernel and
734 * an UP compile the i_size_read/write must be atomic
735 * with respect to the local cpu (unlike with preempt disabled),
736 * but they don't need to be atomic with respect to other cpus like in
737 * true SMP (so they need either to either locally disable irq around
738 * the read or for example on x86 they can be still implemented as a
739 * cmpxchg8b without the need of the lock prefix). For SMP compiles
740 * and 64bit archs it makes no difference if preempt is enabled or not.
742 static inline loff_t i_size_read(const struct inode *inode)
744 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
749 seq = read_seqcount_begin(&inode->i_size_seqcount);
750 i_size = inode->i_size;
751 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
753 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
757 i_size = inode->i_size;
761 return inode->i_size;
766 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
767 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
768 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
770 static inline void i_size_write(struct inode *inode, loff_t i_size)
772 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
774 write_seqcount_begin(&inode->i_size_seqcount);
775 inode->i_size = i_size;
776 write_seqcount_end(&inode->i_size_seqcount);
778 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
780 inode->i_size = i_size;
783 inode->i_size = i_size;
787 /* Helper functions so that in most cases filesystems will
788 * not need to deal directly with kuid_t and kgid_t and can
789 * instead deal with the raw numeric values that are stored
792 static inline uid_t i_uid_read(const struct inode *inode)
794 return from_kuid(&init_user_ns, inode->i_uid);
797 static inline gid_t i_gid_read(const struct inode *inode)
799 return from_kgid(&init_user_ns, inode->i_gid);
802 static inline void i_uid_write(struct inode *inode, uid_t uid)
804 inode->i_uid = make_kuid(&init_user_ns, uid);
807 static inline void i_gid_write(struct inode *inode, gid_t gid)
809 inode->i_gid = make_kgid(&init_user_ns, gid);
812 static inline unsigned iminor(const struct inode *inode)
814 return MINOR(inode->i_rdev);
817 static inline unsigned imajor(const struct inode *inode)
819 return MAJOR(inode->i_rdev);
822 extern struct block_device *I_BDEV(struct inode *inode);
825 rwlock_t lock; /* protects pid, uid, euid fields */
826 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
827 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
828 kuid_t uid, euid; /* uid/euid of process setting the owner */
829 int signum; /* posix.1b rt signal to be delivered on IO */
833 * Track a single file's readahead state
835 struct file_ra_state {
836 pgoff_t start; /* where readahead started */
837 unsigned int size; /* # of readahead pages */
838 unsigned int async_size; /* do asynchronous readahead when
839 there are only # of pages ahead */
841 unsigned int ra_pages; /* Maximum readahead window */
842 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
843 loff_t prev_pos; /* Cache last read() position */
847 * Check if @index falls in the readahead windows.
849 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
851 return (index >= ra->start &&
852 index < ra->start + ra->size);
855 #define FILE_MNT_WRITE_TAKEN 1
856 #define FILE_MNT_WRITE_RELEASED 2
860 * fu_list becomes invalid after file_free is called and queued via
861 * fu_rcuhead for RCU freeing
864 struct list_head fu_list;
865 struct llist_node fu_llist;
866 struct rcu_head fu_rcuhead;
869 #define f_dentry f_path.dentry
870 struct inode *f_inode; /* cached value */
871 const struct file_operations *f_op;
874 * Protects f_ep_links, f_flags, f_pos vs i_size in lseek SEEK_CUR.
875 * Must not be taken from IRQ context.
881 atomic_long_t f_count;
882 unsigned int f_flags;
885 struct fown_struct f_owner;
886 const struct cred *f_cred;
887 struct file_ra_state f_ra;
890 #ifdef CONFIG_SECURITY
893 /* needed for tty driver, and maybe others */
897 /* Used by fs/eventpoll.c to link all the hooks to this file */
898 struct list_head f_ep_links;
899 struct list_head f_tfile_llink;
900 #endif /* #ifdef CONFIG_EPOLL */
901 struct address_space *f_mapping;
902 #ifdef CONFIG_DEBUG_WRITECOUNT
903 unsigned long f_mnt_write_state;
910 /* file identifier */
911 unsigned char f_handle[0];
914 static inline struct file *get_file(struct file *f)
916 atomic_long_inc(&f->f_count);
919 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
920 #define file_count(x) atomic_long_read(&(x)->f_count)
922 #ifdef CONFIG_DEBUG_WRITECOUNT
923 static inline void file_take_write(struct file *f)
925 WARN_ON(f->f_mnt_write_state != 0);
926 f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
928 static inline void file_release_write(struct file *f)
930 f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
932 static inline void file_reset_write(struct file *f)
934 f->f_mnt_write_state = 0;
936 static inline void file_check_state(struct file *f)
939 * At this point, either both or neither of these bits
942 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
943 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
945 static inline int file_check_writeable(struct file *f)
947 if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
949 printk(KERN_WARNING "writeable file with no "
950 "mnt_want_write()\n");
954 #else /* !CONFIG_DEBUG_WRITECOUNT */
955 static inline void file_take_write(struct file *filp) {}
956 static inline void file_release_write(struct file *filp) {}
957 static inline void file_reset_write(struct file *filp) {}
958 static inline void file_check_state(struct file *filp) {}
959 static inline int file_check_writeable(struct file *filp)
963 #endif /* CONFIG_DEBUG_WRITECOUNT */
965 #define MAX_NON_LFS ((1UL<<31) - 1)
967 /* Page cache limit. The filesystems should put that into their s_maxbytes
968 limits, otherwise bad things can happen in VM. */
969 #if BITS_PER_LONG==32
970 #define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
971 #elif BITS_PER_LONG==64
972 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
977 #define FL_ACCESS 8 /* not trying to lock, just looking */
978 #define FL_EXISTS 16 /* when unlocking, test for existence */
979 #define FL_LEASE 32 /* lease held on this file */
980 #define FL_CLOSE 64 /* unlock on close */
981 #define FL_SLEEP 128 /* A blocking lock */
982 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
983 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
986 * Special return value from posix_lock_file() and vfs_lock_file() for
987 * asynchronous locking.
989 #define FILE_LOCK_DEFERRED 1
992 * The POSIX file lock owner is determined by
993 * the "struct files_struct" in the thread group
994 * (or NULL for no owner - BSD locks).
996 * Lockd stuffs a "host" pointer into this.
998 typedef struct files_struct *fl_owner_t;
1000 struct file_lock_operations {
1001 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1002 void (*fl_release_private)(struct file_lock *);
1005 struct lock_manager_operations {
1006 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
1007 unsigned long (*lm_owner_key)(struct file_lock *);
1008 void (*lm_notify)(struct file_lock *); /* unblock callback */
1009 int (*lm_grant)(struct file_lock *, struct file_lock *, int);
1010 void (*lm_break)(struct file_lock *);
1011 int (*lm_change)(struct file_lock **, int);
1014 struct lock_manager {
1015 struct list_head list;
1019 void locks_start_grace(struct net *, struct lock_manager *);
1020 void locks_end_grace(struct lock_manager *);
1021 int locks_in_grace(struct net *);
1023 /* that will die - we need it for nfs_lock_info */
1024 #include <linux/nfs_fs_i.h>
1027 * struct file_lock represents a generic "file lock". It's used to represent
1028 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1029 * note that the same struct is used to represent both a request for a lock and
1030 * the lock itself, but the same object is never used for both.
1032 * FIXME: should we create a separate "struct lock_request" to help distinguish
1035 * The i_flock list is ordered by:
1037 * 1) lock type -- FL_LEASEs first, then FL_FLOCK, and finally FL_POSIX
1039 * 3) lock range start
1042 * Obviously, the last two criteria only matter for POSIX locks.
1045 struct file_lock *fl_next; /* singly linked list for this inode */
1046 struct hlist_node fl_link; /* node in global lists */
1047 struct list_head fl_block; /* circular list of blocked processes */
1048 fl_owner_t fl_owner;
1049 unsigned int fl_flags;
1050 unsigned char fl_type;
1051 unsigned int fl_pid;
1052 int fl_link_cpu; /* what cpu's list is this on? */
1053 struct pid *fl_nspid;
1054 wait_queue_head_t fl_wait;
1055 struct file *fl_file;
1059 struct fasync_struct * fl_fasync; /* for lease break notifications */
1060 /* for lease breaks: */
1061 unsigned long fl_break_time;
1062 unsigned long fl_downgrade_time;
1064 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1065 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1067 struct nfs_lock_info nfs_fl;
1068 struct nfs4_lock_info nfs4_fl;
1070 struct list_head link; /* link in AFS vnode's pending_locks list */
1071 int state; /* state of grant or error if -ve */
1076 /* The following constant reflects the upper bound of the file/locking space */
1078 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1079 #define OFFSET_MAX INT_LIMIT(loff_t)
1080 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1083 #include <linux/fcntl.h>
1085 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1087 #ifdef CONFIG_FILE_LOCKING
1088 extern int fcntl_getlk(struct file *, struct flock __user *);
1089 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1090 struct flock __user *);
1092 #if BITS_PER_LONG == 32
1093 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
1094 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1095 struct flock64 __user *);
1098 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1099 extern int fcntl_getlease(struct file *filp);
1102 void locks_free_lock(struct file_lock *fl);
1103 extern void locks_init_lock(struct file_lock *);
1104 extern struct file_lock * locks_alloc_lock(void);
1105 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1106 extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
1107 extern void locks_remove_posix(struct file *, fl_owner_t);
1108 extern void locks_remove_flock(struct file *);
1109 extern void locks_release_private(struct file_lock *);
1110 extern void posix_test_lock(struct file *, struct file_lock *);
1111 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1112 extern int posix_lock_file_wait(struct file *, struct file_lock *);
1113 extern int posix_unblock_lock(struct file_lock *);
1114 extern int vfs_test_lock(struct file *, struct file_lock *);
1115 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1116 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1117 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1118 extern int __break_lease(struct inode *inode, unsigned int flags);
1119 extern void lease_get_mtime(struct inode *, struct timespec *time);
1120 extern int generic_setlease(struct file *, long, struct file_lock **);
1121 extern int vfs_setlease(struct file *, long, struct file_lock **);
1122 extern int lease_modify(struct file_lock **, int);
1123 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1124 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1125 #else /* !CONFIG_FILE_LOCKING */
1126 static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1131 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1132 unsigned int cmd, struct flock __user *user)
1137 #if BITS_PER_LONG == 32
1138 static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1143 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1144 unsigned int cmd, struct flock64 __user *user)
1149 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1154 static inline int fcntl_getlease(struct file *filp)
1159 static inline void locks_init_lock(struct file_lock *fl)
1164 static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1169 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1174 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1179 static inline void locks_remove_flock(struct file *filp)
1184 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1189 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1190 struct file_lock *conflock)
1195 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1200 static inline int posix_unblock_lock(struct file_lock *waiter)
1205 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1210 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1211 struct file_lock *fl, struct file_lock *conf)
1216 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1221 static inline int flock_lock_file_wait(struct file *filp,
1222 struct file_lock *request)
1227 static inline int __break_lease(struct inode *inode, unsigned int mode)
1232 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1237 static inline int generic_setlease(struct file *filp, long arg,
1238 struct file_lock **flp)
1243 static inline int vfs_setlease(struct file *filp, long arg,
1244 struct file_lock **lease)
1249 static inline int lease_modify(struct file_lock **before, int arg)
1254 static inline int lock_may_read(struct inode *inode, loff_t start,
1260 static inline int lock_may_write(struct inode *inode, loff_t start,
1265 #endif /* !CONFIG_FILE_LOCKING */
1268 struct fasync_struct {
1272 struct fasync_struct *fa_next; /* singly linked list */
1273 struct file *fa_file;
1274 struct rcu_head fa_rcu;
1277 #define FASYNC_MAGIC 0x4601
1279 /* SMP safe fasync helpers: */
1280 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1281 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1282 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1283 extern struct fasync_struct *fasync_alloc(void);
1284 extern void fasync_free(struct fasync_struct *);
1286 /* can be called from interrupts */
1287 extern void kill_fasync(struct fasync_struct **, int, int);
1289 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1290 extern int f_setown(struct file *filp, unsigned long arg, int force);
1291 extern void f_delown(struct file *filp);
1292 extern pid_t f_getown(struct file *filp);
1293 extern int send_sigurg(struct fown_struct *fown);
1301 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1302 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1303 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1304 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1305 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1307 extern struct list_head super_blocks;
1308 extern spinlock_t sb_lock;
1310 /* Possible states of 'frozen' field */
1312 SB_UNFROZEN = 0, /* FS is unfrozen */
1313 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1314 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1315 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1316 * internal threads if needed) */
1317 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1320 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1323 /* Counters for counting writers at each level */
1324 struct percpu_counter counter[SB_FREEZE_LEVELS];
1325 wait_queue_head_t wait; /* queue for waiting for
1326 writers / faults to finish */
1327 int frozen; /* Is sb frozen? */
1328 wait_queue_head_t wait_unfrozen; /* queue for waiting for
1330 #ifdef CONFIG_DEBUG_LOCK_ALLOC
1331 struct lockdep_map lock_map[SB_FREEZE_LEVELS];
1335 struct super_block {
1336 struct list_head s_list; /* Keep this first */
1337 dev_t s_dev; /* search index; _not_ kdev_t */
1338 unsigned char s_blocksize_bits;
1339 unsigned long s_blocksize;
1340 loff_t s_maxbytes; /* Max file size */
1341 struct file_system_type *s_type;
1342 const struct super_operations *s_op;
1343 const struct dquot_operations *dq_op;
1344 const struct quotactl_ops *s_qcop;
1345 const struct export_operations *s_export_op;
1346 unsigned long s_flags;
1347 unsigned long s_magic;
1348 struct dentry *s_root;
1349 struct rw_semaphore s_umount;
1352 #ifdef CONFIG_SECURITY
1355 const struct xattr_handler **s_xattr;
1357 struct list_head s_inodes; /* all inodes */
1358 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1360 struct list_head __percpu *s_files;
1362 struct list_head s_files;
1364 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1365 struct block_device *s_bdev;
1366 struct backing_dev_info *s_bdi;
1367 struct mtd_info *s_mtd;
1368 struct hlist_node s_instances;
1369 struct quota_info s_dquot; /* Diskquota specific options */
1371 struct sb_writers s_writers;
1373 char s_id[32]; /* Informational name */
1374 u8 s_uuid[16]; /* UUID */
1376 void *s_fs_info; /* Filesystem private info */
1377 unsigned int s_max_links;
1380 /* Granularity of c/m/atime in ns.
1381 Cannot be worse than a second */
1385 * The next field is for VFS *only*. No filesystems have any business
1386 * even looking at it. You had been warned.
1388 struct mutex s_vfs_rename_mutex; /* Kludge */
1391 * Filesystem subtype. If non-empty the filesystem type field
1392 * in /proc/mounts will be "type.subtype"
1397 * Saved mount options for lazy filesystems using
1398 * generic_show_options()
1400 char __rcu *s_options;
1401 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1404 * Saved pool identifier for cleancache (-1 means none)
1406 int cleancache_poolid;
1408 struct shrinker s_shrink; /* per-sb shrinker handle */
1410 /* Number of inodes with nlink == 0 but still referenced */
1411 atomic_long_t s_remove_count;
1413 /* Being remounted read-only */
1414 int s_readonly_remount;
1416 /* AIO completions deferred from interrupt context */
1417 struct workqueue_struct *s_dio_done_wq;
1420 * Keep the lru lists last in the structure so they always sit on their
1421 * own individual cachelines.
1423 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1424 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1427 extern struct timespec current_fs_time(struct super_block *sb);
1430 * Snapshotting support.
1433 void __sb_end_write(struct super_block *sb, int level);
1434 int __sb_start_write(struct super_block *sb, int level, bool wait);
1437 * sb_end_write - drop write access to a superblock
1438 * @sb: the super we wrote to
1440 * Decrement number of writers to the filesystem. Wake up possible waiters
1441 * wanting to freeze the filesystem.
1443 static inline void sb_end_write(struct super_block *sb)
1445 __sb_end_write(sb, SB_FREEZE_WRITE);
1449 * sb_end_pagefault - drop write access to a superblock from a page fault
1450 * @sb: the super we wrote to
1452 * Decrement number of processes handling write page fault to the filesystem.
1453 * Wake up possible waiters wanting to freeze the filesystem.
1455 static inline void sb_end_pagefault(struct super_block *sb)
1457 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1461 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1462 * @sb: the super we wrote to
1464 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1465 * waiters wanting to freeze the filesystem.
1467 static inline void sb_end_intwrite(struct super_block *sb)
1469 __sb_end_write(sb, SB_FREEZE_FS);
1473 * sb_start_write - get write access to a superblock
1474 * @sb: the super we write to
1476 * When a process wants to write data or metadata to a file system (i.e. dirty
1477 * a page or an inode), it should embed the operation in a sb_start_write() -
1478 * sb_end_write() pair to get exclusion against file system freezing. This
1479 * function increments number of writers preventing freezing. If the file
1480 * system is already frozen, the function waits until the file system is
1483 * Since freeze protection behaves as a lock, users have to preserve
1484 * ordering of freeze protection and other filesystem locks. Generally,
1485 * freeze protection should be the outermost lock. In particular, we have:
1488 * -> i_mutex (write path, truncate, directory ops, ...)
1489 * -> s_umount (freeze_super, thaw_super)
1491 static inline void sb_start_write(struct super_block *sb)
1493 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1496 static inline int sb_start_write_trylock(struct super_block *sb)
1498 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1502 * sb_start_pagefault - get write access to a superblock from a page fault
1503 * @sb: the super we write to
1505 * When a process starts handling write page fault, it should embed the
1506 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1507 * exclusion against file system freezing. This is needed since the page fault
1508 * is going to dirty a page. This function increments number of running page
1509 * faults preventing freezing. If the file system is already frozen, the
1510 * function waits until the file system is thawed.
1512 * Since page fault freeze protection behaves as a lock, users have to preserve
1513 * ordering of freeze protection and other filesystem locks. It is advised to
1514 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1515 * handling code implies lock dependency:
1518 * -> sb_start_pagefault
1520 static inline void sb_start_pagefault(struct super_block *sb)
1522 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1526 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1527 * @sb: the super we write to
1529 * This is the third level of protection against filesystem freezing. It is
1530 * free for use by a filesystem. The only requirement is that it must rank
1531 * below sb_start_pagefault.
1533 * For example filesystem can call sb_start_intwrite() when starting a
1534 * transaction which somewhat eases handling of freezing for internal sources
1535 * of filesystem changes (internal fs threads, discarding preallocation on file
1538 static inline void sb_start_intwrite(struct super_block *sb)
1540 __sb_start_write(sb, SB_FREEZE_FS, true);
1544 extern bool inode_owner_or_capable(const struct inode *inode);
1547 * VFS helper functions..
1549 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1550 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1551 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1552 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1553 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1554 extern int vfs_rmdir(struct inode *, struct dentry *);
1555 extern int vfs_unlink(struct inode *, struct dentry *);
1556 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1559 * VFS dentry helper functions.
1561 extern void dentry_unhash(struct dentry *dentry);
1564 * VFS file helper functions.
1566 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1569 * VFS FS_IOC_FIEMAP helper definitions.
1571 struct fiemap_extent_info {
1572 unsigned int fi_flags; /* Flags as passed from user */
1573 unsigned int fi_extents_mapped; /* Number of mapped extents */
1574 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1575 struct fiemap_extent __user *fi_extents_start; /* Start of
1576 fiemap_extent array */
1578 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1579 u64 phys, u64 len, u32 flags);
1580 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1585 * NOTE! These match bits 12..15 of stat.st_mode
1586 * (ie "(i_mode >> 12) & 15").
1588 #define DT_UNKNOWN 0
1599 * This is the "filldir" function type, used by readdir() to let
1600 * the kernel specify what kind of dirent layout it wants to have.
1601 * This allows the kernel to read directories into kernel space or
1602 * to have different dirent layouts depending on the binary type.
1604 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1605 struct dir_context {
1606 const filldir_t actor;
1610 struct block_device_operations;
1612 /* These macros are for out of kernel modules to test that
1613 * the kernel supports the unlocked_ioctl and compat_ioctl
1614 * fields in struct file_operations. */
1615 #define HAVE_COMPAT_IOCTL 1
1616 #define HAVE_UNLOCKED_IOCTL 1
1618 struct file_operations {
1619 struct module *owner;
1620 loff_t (*llseek) (struct file *, loff_t, int);
1621 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1622 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1623 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1624 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *, loff_t);
1625 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1626 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *, loff_t);
1627 int (*iterate) (struct file *, struct dir_context *);
1628 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1629 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1630 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1631 int (*mmap) (struct file *, struct vm_area_struct *);
1632 int (*open) (struct inode *, struct file *);
1633 int (*flush) (struct file *, fl_owner_t id);
1634 int (*release) (struct inode *, struct file *);
1635 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1636 int (*aio_fsync) (struct kiocb *, int datasync);
1637 int (*fasync) (int, struct file *, int);
1638 int (*lock) (struct file *, int, struct file_lock *);
1639 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1640 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1641 int (*check_flags)(int);
1642 int (*flock) (struct file *, int, struct file_lock *);
1643 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1644 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1645 int (*setlease)(struct file *, long, struct file_lock **);
1646 long (*fallocate)(struct file *file, int mode, loff_t offset,
1648 int (*show_fdinfo)(struct seq_file *m, struct file *f);
1651 static inline int file_readable(struct file *filp)
1653 return filp && (filp->f_op->read || filp->f_op->aio_read ||
1654 filp->f_op->read_iter);
1657 static inline int file_writable(struct file *filp)
1659 return filp && (filp->f_op->write || filp->f_op->aio_write ||
1660 filp->f_op->write_iter);
1663 struct inode_operations {
1664 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1665 void * (*follow_link) (struct dentry *, struct nameidata *);
1666 int (*permission) (struct inode *, int);
1667 struct posix_acl * (*get_acl)(struct inode *, int);
1669 int (*readlink) (struct dentry *, char __user *,int);
1670 void (*put_link) (struct dentry *, struct nameidata *, void *);
1672 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1673 int (*link) (struct dentry *,struct inode *,struct dentry *);
1674 int (*unlink) (struct inode *,struct dentry *);
1675 int (*symlink) (struct inode *,struct dentry *,const char *);
1676 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1677 int (*rmdir) (struct inode *,struct dentry *);
1678 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1679 int (*rename) (struct inode *, struct dentry *,
1680 struct inode *, struct dentry *);
1681 int (*setattr) (struct dentry *, struct iattr *);
1682 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1683 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1684 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1685 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1686 int (*removexattr) (struct dentry *, const char *);
1687 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1689 int (*update_time)(struct inode *, struct timespec *, int);
1690 int (*atomic_open)(struct inode *, struct dentry *,
1691 struct file *, unsigned open_flag,
1692 umode_t create_mode, int *opened);
1693 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1694 } ____cacheline_aligned;
1696 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1697 unsigned long nr_segs, unsigned long fast_segs,
1698 struct iovec *fast_pointer,
1699 struct iovec **ret_pointer);
1701 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1702 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1703 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1704 unsigned long, loff_t *);
1705 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1706 unsigned long, loff_t *);
1708 struct super_operations {
1709 struct inode *(*alloc_inode)(struct super_block *sb);
1710 void (*destroy_inode)(struct inode *);
1712 void (*dirty_inode) (struct inode *, int flags);
1713 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1714 int (*drop_inode) (struct inode *);
1715 void (*evict_inode) (struct inode *);
1716 void (*put_super) (struct super_block *);
1717 int (*sync_fs)(struct super_block *sb, int wait);
1718 int (*freeze_fs) (struct super_block *);
1719 int (*unfreeze_fs) (struct super_block *);
1720 int (*statfs) (struct dentry *, struct kstatfs *);
1721 int (*remount_fs) (struct super_block *, int *, char *);
1722 void (*umount_begin) (struct super_block *);
1724 int (*show_options)(struct seq_file *, struct dentry *);
1725 int (*show_devname)(struct seq_file *, struct dentry *);
1726 int (*show_path)(struct seq_file *, struct dentry *);
1727 int (*show_stats)(struct seq_file *, struct dentry *);
1729 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1730 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1732 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1733 long (*nr_cached_objects)(struct super_block *, int);
1734 long (*free_cached_objects)(struct super_block *, long, int);
1738 * Inode flags - they have no relation to superblock flags now
1740 #define S_SYNC 1 /* Writes are synced at once */
1741 #define S_NOATIME 2 /* Do not update access times */
1742 #define S_APPEND 4 /* Append-only file */
1743 #define S_IMMUTABLE 8 /* Immutable file */
1744 #define S_DEAD 16 /* removed, but still open directory */
1745 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1746 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1747 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1748 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1749 #define S_PRIVATE 512 /* Inode is fs-internal */
1750 #define S_IMA 1024 /* Inode has an associated IMA struct */
1751 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1752 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1755 * Note that nosuid etc flags are inode-specific: setting some file-system
1756 * flags just means all the inodes inherit those flags by default. It might be
1757 * possible to override it selectively if you really wanted to with some
1758 * ioctl() that is not currently implemented.
1760 * Exception: MS_RDONLY is always applied to the entire file system.
1762 * Unfortunately, it is possible to change a filesystems flags with it mounted
1763 * with files in use. This means that all of the inodes will not have their
1764 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1765 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1767 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1769 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1770 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1771 ((inode)->i_flags & S_SYNC))
1772 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1773 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1774 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1775 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1776 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1778 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1779 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1780 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1781 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1783 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1784 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1785 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1786 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1787 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1788 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1789 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1792 * Inode state bits. Protected by inode->i_lock
1794 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1795 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1797 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1798 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1799 * various stages of removing an inode.
1801 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1803 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1804 * fdatasync(). i_atime is the usual cause.
1805 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1806 * these changes separately from I_DIRTY_SYNC so that we
1807 * don't have to write inode on fdatasync() when only
1808 * mtime has changed in it.
1809 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1810 * I_NEW Serves as both a mutex and completion notification.
1811 * New inodes set I_NEW. If two processes both create
1812 * the same inode, one of them will release its inode and
1813 * wait for I_NEW to be released before returning.
1814 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1815 * also cause waiting on I_NEW, without I_NEW actually
1816 * being set. find_inode() uses this to prevent returning
1817 * nearly-dead inodes.
1818 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1819 * is zero. I_FREEING must be set when I_WILL_FREE is
1821 * I_FREEING Set when inode is about to be freed but still has dirty
1822 * pages or buffers attached or the inode itself is still
1824 * I_CLEAR Added by clear_inode(). In this state the inode is
1825 * clean and can be destroyed. Inode keeps I_FREEING.
1827 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1828 * prohibited for many purposes. iget() must wait for
1829 * the inode to be completely released, then create it
1830 * anew. Other functions will just ignore such inodes,
1831 * if appropriate. I_NEW is used for waiting.
1833 * I_SYNC Writeback of inode is running. The bit is set during
1834 * data writeback, and cleared with a wakeup on the bit
1835 * address once it is done. The bit is also used to pin
1836 * the inode in memory for flusher thread.
1838 * I_REFERENCED Marks the inode as recently references on the LRU list.
1840 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1842 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1844 #define I_DIRTY_SYNC (1 << 0)
1845 #define I_DIRTY_DATASYNC (1 << 1)
1846 #define I_DIRTY_PAGES (1 << 2)
1848 #define I_NEW (1 << __I_NEW)
1849 #define I_WILL_FREE (1 << 4)
1850 #define I_FREEING (1 << 5)
1851 #define I_CLEAR (1 << 6)
1853 #define I_SYNC (1 << __I_SYNC)
1854 #define I_REFERENCED (1 << 8)
1855 #define __I_DIO_WAKEUP 9
1856 #define I_DIO_WAKEUP (1 << I_DIO_WAKEUP)
1857 #define I_LINKABLE (1 << 10)
1859 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1861 extern void __mark_inode_dirty(struct inode *, int);
1862 static inline void mark_inode_dirty(struct inode *inode)
1864 __mark_inode_dirty(inode, I_DIRTY);
1867 static inline void mark_inode_dirty_sync(struct inode *inode)
1869 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1872 extern void inc_nlink(struct inode *inode);
1873 extern void drop_nlink(struct inode *inode);
1874 extern void clear_nlink(struct inode *inode);
1875 extern void set_nlink(struct inode *inode, unsigned int nlink);
1877 static inline void inode_inc_link_count(struct inode *inode)
1880 mark_inode_dirty(inode);
1883 static inline void inode_dec_link_count(struct inode *inode)
1886 mark_inode_dirty(inode);
1890 * inode_inc_iversion - increments i_version
1891 * @inode: inode that need to be updated
1893 * Every time the inode is modified, the i_version field will be incremented.
1894 * The filesystem has to be mounted with i_version flag
1897 static inline void inode_inc_iversion(struct inode *inode)
1899 spin_lock(&inode->i_lock);
1901 spin_unlock(&inode->i_lock);
1904 enum file_time_flags {
1911 extern void touch_atime(const struct path *);
1912 static inline void file_accessed(struct file *file)
1914 if (!(file->f_flags & O_NOATIME))
1915 touch_atime(&file->f_path);
1918 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1919 int sync_inode_metadata(struct inode *inode, int wait);
1921 struct file_system_type {
1924 #define FS_REQUIRES_DEV 1
1925 #define FS_BINARY_MOUNTDATA 2
1926 #define FS_HAS_SUBTYPE 4
1927 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
1928 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
1929 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
1930 struct dentry *(*mount) (struct file_system_type *, int,
1931 const char *, void *);
1932 void (*kill_sb) (struct super_block *);
1933 struct module *owner;
1934 struct file_system_type * next;
1935 struct hlist_head fs_supers;
1937 struct lock_class_key s_lock_key;
1938 struct lock_class_key s_umount_key;
1939 struct lock_class_key s_vfs_rename_key;
1940 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
1942 struct lock_class_key i_lock_key;
1943 struct lock_class_key i_mutex_key;
1944 struct lock_class_key i_mutex_dir_key;
1947 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1949 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1950 void *data, int (*fill_super)(struct super_block *, void *, int));
1951 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1952 int flags, const char *dev_name, void *data,
1953 int (*fill_super)(struct super_block *, void *, int));
1954 extern struct dentry *mount_single(struct file_system_type *fs_type,
1955 int flags, void *data,
1956 int (*fill_super)(struct super_block *, void *, int));
1957 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1958 int flags, void *data,
1959 int (*fill_super)(struct super_block *, void *, int));
1960 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
1961 void generic_shutdown_super(struct super_block *sb);
1962 void kill_block_super(struct super_block *sb);
1963 void kill_anon_super(struct super_block *sb);
1964 void kill_litter_super(struct super_block *sb);
1965 void deactivate_super(struct super_block *sb);
1966 void deactivate_locked_super(struct super_block *sb);
1967 int set_anon_super(struct super_block *s, void *data);
1968 int get_anon_bdev(dev_t *);
1969 void free_anon_bdev(dev_t);
1970 struct super_block *sget(struct file_system_type *type,
1971 int (*test)(struct super_block *,void *),
1972 int (*set)(struct super_block *,void *),
1973 int flags, void *data);
1974 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1975 const struct super_operations *ops,
1976 const struct dentry_operations *dops,
1979 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1980 #define fops_get(fops) \
1981 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1982 #define fops_put(fops) \
1983 do { if (fops) module_put((fops)->owner); } while(0)
1985 extern int register_filesystem(struct file_system_type *);
1986 extern int unregister_filesystem(struct file_system_type *);
1987 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1988 #define kern_mount(type) kern_mount_data(type, NULL)
1989 extern void kern_unmount(struct vfsmount *mnt);
1990 extern int may_umount_tree(struct vfsmount *);
1991 extern int may_umount(struct vfsmount *);
1992 extern long do_mount(const char *, const char *, const char *, unsigned long, void *);
1993 extern struct vfsmount *collect_mounts(struct path *);
1994 extern void drop_collected_mounts(struct vfsmount *);
1995 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1997 extern int vfs_statfs(struct path *, struct kstatfs *);
1998 extern int user_statfs(const char __user *, struct kstatfs *);
1999 extern int fd_statfs(int, struct kstatfs *);
2000 extern int vfs_ustat(dev_t, struct kstatfs *);
2001 extern int freeze_super(struct super_block *super);
2002 extern int thaw_super(struct super_block *super);
2003 extern bool our_mnt(struct vfsmount *mnt);
2004 extern bool fs_fully_visible(struct file_system_type *);
2006 extern int current_umask(void);
2009 extern struct kobject *fs_kobj;
2011 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
2013 #define FLOCK_VERIFY_READ 1
2014 #define FLOCK_VERIFY_WRITE 2
2016 #ifdef CONFIG_FILE_LOCKING
2017 extern int locks_mandatory_locked(struct inode *);
2018 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
2021 * Candidates for mandatory locking have the setgid bit set
2022 * but no group execute bit - an otherwise meaningless combination.
2025 static inline int __mandatory_lock(struct inode *ino)
2027 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2031 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2032 * otherwise these will be advisory locks
2035 static inline int mandatory_lock(struct inode *ino)
2037 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2040 static inline int locks_verify_locked(struct inode *inode)
2042 if (mandatory_lock(inode))
2043 return locks_mandatory_locked(inode);
2047 static inline int locks_verify_truncate(struct inode *inode,
2051 if (inode->i_flock && mandatory_lock(inode))
2052 return locks_mandatory_area(
2053 FLOCK_VERIFY_WRITE, inode, filp,
2054 size < inode->i_size ? size : inode->i_size,
2055 (size < inode->i_size ? inode->i_size - size
2056 : size - inode->i_size)
2061 static inline int break_lease(struct inode *inode, unsigned int mode)
2064 return __break_lease(inode, mode);
2067 #else /* !CONFIG_FILE_LOCKING */
2068 static inline int locks_mandatory_locked(struct inode *inode)
2073 static inline int locks_mandatory_area(int rw, struct inode *inode,
2074 struct file *filp, loff_t offset,
2080 static inline int __mandatory_lock(struct inode *inode)
2085 static inline int mandatory_lock(struct inode *inode)
2090 static inline int locks_verify_locked(struct inode *inode)
2095 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2101 static inline int break_lease(struct inode *inode, unsigned int mode)
2106 #endif /* CONFIG_FILE_LOCKING */
2111 const char *name; /* pointer to actual string */
2112 const __user char *uptr; /* original userland pointer */
2113 struct audit_names *aname;
2114 bool separate; /* should "name" be freed? */
2117 extern long vfs_truncate(struct path *, loff_t);
2118 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2120 extern int do_fallocate(struct file *file, int mode, loff_t offset,
2122 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2124 extern struct file *file_open_name(struct filename *, int, umode_t);
2125 extern struct file *filp_open(const char *, int, umode_t);
2126 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2128 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2129 extern int filp_close(struct file *, fl_owner_t id);
2131 extern struct filename *getname(const char __user *);
2137 extern int finish_open(struct file *file, struct dentry *dentry,
2138 int (*open)(struct inode *, struct file *),
2140 extern int finish_no_open(struct file *file, struct dentry *dentry);
2144 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2147 extern void __init vfs_caches_init_early(void);
2148 extern void __init vfs_caches_init(unsigned long);
2150 extern struct kmem_cache *names_cachep;
2152 extern void final_putname(struct filename *name);
2154 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2155 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2156 #ifndef CONFIG_AUDITSYSCALL
2157 #define putname(name) final_putname(name)
2159 extern void putname(struct filename *name);
2163 extern int register_blkdev(unsigned int, const char *);
2164 extern void unregister_blkdev(unsigned int, const char *);
2165 extern struct block_device *bdget(dev_t);
2166 extern struct block_device *bdgrab(struct block_device *bdev);
2167 extern void bd_set_size(struct block_device *, loff_t size);
2168 extern void bd_forget(struct inode *inode);
2169 extern void bdput(struct block_device *);
2170 extern void invalidate_bdev(struct block_device *);
2171 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2172 extern int sync_blockdev(struct block_device *bdev);
2173 extern void kill_bdev(struct block_device *);
2174 extern struct super_block *freeze_bdev(struct block_device *);
2175 extern void emergency_thaw_all(void);
2176 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2177 extern int fsync_bdev(struct block_device *);
2178 extern int sb_is_blkdev_sb(struct super_block *sb);
2180 static inline void bd_forget(struct inode *inode) {}
2181 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2182 static inline void kill_bdev(struct block_device *bdev) {}
2183 static inline void invalidate_bdev(struct block_device *bdev) {}
2185 static inline struct super_block *freeze_bdev(struct block_device *sb)
2190 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2195 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2199 static inline int sb_is_blkdev_sb(struct super_block *sb)
2204 extern int sync_filesystem(struct super_block *);
2205 extern const struct file_operations def_blk_fops;
2206 extern const struct file_operations def_chr_fops;
2207 extern const struct file_operations bad_sock_fops;
2209 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2210 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2211 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2212 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2213 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2215 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2217 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2219 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2220 extern void bd_unlink_disk_holder(struct block_device *bdev,
2221 struct gendisk *disk);
2223 static inline int bd_link_disk_holder(struct block_device *bdev,
2224 struct gendisk *disk)
2228 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2229 struct gendisk *disk)
2236 #define CHRDEV_MAJOR_HASH_SIZE 255
2237 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2238 extern int register_chrdev_region(dev_t, unsigned, const char *);
2239 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2240 unsigned int count, const char *name,
2241 const struct file_operations *fops);
2242 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2243 unsigned int count, const char *name);
2244 extern void unregister_chrdev_region(dev_t, unsigned);
2245 extern void chrdev_show(struct seq_file *,off_t);
2247 static inline int register_chrdev(unsigned int major, const char *name,
2248 const struct file_operations *fops)
2250 return __register_chrdev(major, 0, 256, name, fops);
2253 static inline void unregister_chrdev(unsigned int major, const char *name)
2255 __unregister_chrdev(major, 0, 256, name);
2258 /* fs/block_dev.c */
2259 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2260 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2263 #define BLKDEV_MAJOR_HASH_SIZE 255
2264 extern const char *__bdevname(dev_t, char *buffer);
2265 extern const char *bdevname(struct block_device *bdev, char *buffer);
2266 extern struct block_device *lookup_bdev(const char *);
2267 extern void blkdev_show(struct seq_file *,off_t);
2270 #define BLKDEV_MAJOR_HASH_SIZE 0
2273 extern void init_special_inode(struct inode *, umode_t, dev_t);
2275 /* Invalid inode operations -- fs/bad_inode.c */
2276 extern void make_bad_inode(struct inode *);
2277 extern int is_bad_inode(struct inode *);
2281 * return READ, READA, or WRITE
2283 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2286 * return data direction, READ or WRITE
2288 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2290 extern void check_disk_size_change(struct gendisk *disk,
2291 struct block_device *bdev);
2292 extern int revalidate_disk(struct gendisk *);
2293 extern int check_disk_change(struct block_device *);
2294 extern int __invalidate_device(struct block_device *, bool);
2295 extern int invalidate_partition(struct gendisk *, int);
2297 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2298 pgoff_t start, pgoff_t end);
2300 static inline void invalidate_remote_inode(struct inode *inode)
2302 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2303 S_ISLNK(inode->i_mode))
2304 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2306 extern int invalidate_inode_pages2(struct address_space *mapping);
2307 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2308 pgoff_t start, pgoff_t end);
2309 extern int write_inode_now(struct inode *, int);
2310 extern int filemap_fdatawrite(struct address_space *);
2311 extern int filemap_flush(struct address_space *);
2312 extern int filemap_fdatawait(struct address_space *);
2313 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2315 extern int filemap_write_and_wait(struct address_space *mapping);
2316 extern int filemap_write_and_wait_range(struct address_space *mapping,
2317 loff_t lstart, loff_t lend);
2318 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2319 loff_t start, loff_t end, int sync_mode);
2320 extern int filemap_fdatawrite_range(struct address_space *mapping,
2321 loff_t start, loff_t end);
2323 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2325 extern int vfs_fsync(struct file *file, int datasync);
2326 extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
2327 extern void emergency_sync(void);
2328 extern void emergency_remount(void);
2330 extern sector_t bmap(struct inode *, sector_t);
2332 extern int notify_change(struct dentry *, struct iattr *);
2333 extern int inode_permission(struct inode *, int);
2334 extern int generic_permission(struct inode *, int);
2336 static inline bool execute_ok(struct inode *inode)
2338 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2341 static inline struct inode *file_inode(struct file *f)
2346 static inline void file_start_write(struct file *file)
2348 if (!S_ISREG(file_inode(file)->i_mode))
2350 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2353 static inline bool file_start_write_trylock(struct file *file)
2355 if (!S_ISREG(file_inode(file)->i_mode))
2357 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2360 static inline void file_end_write(struct file *file)
2362 if (!S_ISREG(file_inode(file)->i_mode))
2364 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2368 * get_write_access() gets write permission for a file.
2369 * put_write_access() releases this write permission.
2370 * This is used for regular files.
2371 * We cannot support write (and maybe mmap read-write shared) accesses and
2372 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2373 * can have the following values:
2374 * 0: no writers, no VM_DENYWRITE mappings
2375 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2376 * > 0: (i_writecount) users are writing to the file.
2378 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2379 * except for the cases where we don't hold i_writecount yet. Then we need to
2380 * use {get,deny}_write_access() - these functions check the sign and refuse
2381 * to do the change if sign is wrong.
2383 static inline int get_write_access(struct inode *inode)
2385 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2387 static inline int deny_write_access(struct file *file)
2389 struct inode *inode = file_inode(file);
2390 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2392 static inline void put_write_access(struct inode * inode)
2394 atomic_dec(&inode->i_writecount);
2396 static inline void allow_write_access(struct file *file)
2399 atomic_inc(&file_inode(file)->i_writecount);
2402 static inline void i_readcount_dec(struct inode *inode)
2404 BUG_ON(!atomic_read(&inode->i_readcount));
2405 atomic_dec(&inode->i_readcount);
2407 static inline void i_readcount_inc(struct inode *inode)
2409 atomic_inc(&inode->i_readcount);
2412 static inline void i_readcount_dec(struct inode *inode)
2416 static inline void i_readcount_inc(struct inode *inode)
2421 extern int do_pipe_flags(int *, int);
2423 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2424 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2425 extern struct file * open_exec(const char *);
2427 /* fs/dcache.c -- generic fs support functions */
2428 extern int is_subdir(struct dentry *, struct dentry *);
2429 extern int path_is_under(struct path *, struct path *);
2431 #include <linux/err.h>
2433 /* needed for stackable file system support */
2434 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2436 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2438 extern int inode_init_always(struct super_block *, struct inode *);
2439 extern void inode_init_once(struct inode *);
2440 extern void address_space_init_once(struct address_space *mapping);
2441 extern void ihold(struct inode * inode);
2442 extern void iput(struct inode *);
2443 extern struct inode * igrab(struct inode *);
2444 extern ino_t iunique(struct super_block *, ino_t);
2445 extern int inode_needs_sync(struct inode *inode);
2446 extern int generic_delete_inode(struct inode *inode);
2447 static inline int generic_drop_inode(struct inode *inode)
2449 return !inode->i_nlink || inode_unhashed(inode);
2452 extern struct inode *ilookup5_nowait(struct super_block *sb,
2453 unsigned long hashval, int (*test)(struct inode *, void *),
2455 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2456 int (*test)(struct inode *, void *), void *data);
2457 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2459 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2460 extern struct inode * iget_locked(struct super_block *, unsigned long);
2461 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2462 extern int insert_inode_locked(struct inode *);
2463 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2464 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2466 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2468 extern void unlock_new_inode(struct inode *);
2469 extern unsigned int get_next_ino(void);
2471 extern void __iget(struct inode * inode);
2472 extern void iget_failed(struct inode *);
2473 extern void clear_inode(struct inode *);
2474 extern void __destroy_inode(struct inode *);
2475 extern struct inode *new_inode_pseudo(struct super_block *sb);
2476 extern struct inode *new_inode(struct super_block *sb);
2477 extern void free_inode_nonrcu(struct inode *inode);
2478 extern int should_remove_suid(struct dentry *);
2479 extern int file_remove_suid(struct file *);
2481 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2482 static inline void insert_inode_hash(struct inode *inode)
2484 __insert_inode_hash(inode, inode->i_ino);
2487 extern void __remove_inode_hash(struct inode *);
2488 static inline void remove_inode_hash(struct inode *inode)
2490 if (!inode_unhashed(inode))
2491 __remove_inode_hash(inode);
2494 extern void inode_sb_list_add(struct inode *inode);
2497 extern void submit_bio(int, struct bio *);
2498 extern int bdev_read_only(struct block_device *);
2500 extern int set_blocksize(struct block_device *, int);
2501 extern int sb_set_blocksize(struct super_block *, int);
2502 extern int sb_min_blocksize(struct super_block *, int);
2504 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2505 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2506 extern int generic_file_remap_pages(struct vm_area_struct *, unsigned long addr,
2507 unsigned long size, pgoff_t pgoff);
2508 extern int file_read_iter_actor(read_descriptor_t *desc, struct page *page,
2509 unsigned long offset, unsigned long size);
2510 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2511 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2512 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *,
2514 extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2516 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *,
2518 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2519 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *,
2521 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2522 unsigned long *, loff_t, loff_t *, size_t, size_t);
2523 extern ssize_t generic_file_direct_write_iter(struct kiocb *, struct iov_iter *,
2524 loff_t, loff_t *, size_t);
2525 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2526 unsigned long, loff_t, loff_t *, size_t, ssize_t);
2527 extern ssize_t generic_file_buffered_write_iter(struct kiocb *,
2528 struct iov_iter *, loff_t, loff_t *, size_t, ssize_t);
2529 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2530 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2531 extern int generic_segment_checks(const struct iovec *iov,
2532 unsigned long *nr_segs, size_t *count, int access_flags);
2534 /* fs/block_dev.c */
2535 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *iter,
2537 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2539 extern void block_sync_page(struct page *page);
2542 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2543 struct pipe_inode_info *, size_t, unsigned int);
2544 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2545 struct pipe_inode_info *, size_t, unsigned int);
2546 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
2547 struct file *, loff_t *, size_t, unsigned int);
2548 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2549 struct file *out, loff_t *, size_t len, unsigned int flags);
2552 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2553 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2554 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2555 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2556 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2557 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2558 int whence, loff_t maxsize, loff_t eof);
2559 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2560 int whence, loff_t size);
2561 extern int generic_file_open(struct inode * inode, struct file * filp);
2562 extern int nonseekable_open(struct inode * inode, struct file * filp);
2564 #ifdef CONFIG_FS_XIP
2565 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2567 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2568 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2569 size_t len, loff_t *ppos);
2570 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2572 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2579 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2580 loff_t file_offset);
2583 /* need locking between buffered and direct access */
2586 /* filesystem does not support filling holes */
2587 DIO_SKIP_HOLES = 0x02,
2590 void dio_end_io(struct bio *bio, int error);
2592 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2593 struct block_device *bdev, struct iov_iter *iter, loff_t offset,
2594 get_block_t get_block, dio_iodone_t end_io, dio_submit_t submit_io,
2597 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2598 struct inode *inode, struct iov_iter *iter, loff_t offset,
2599 get_block_t get_block)
2601 return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iter,
2602 offset, get_block, NULL, NULL,
2603 DIO_LOCKING | DIO_SKIP_HOLES);
2607 void inode_dio_wait(struct inode *inode);
2608 void inode_dio_done(struct inode *inode);
2610 extern const struct file_operations generic_ro_fops;
2612 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2614 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2615 extern int page_readlink(struct dentry *, char __user *, int);
2616 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2617 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2618 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2620 extern int page_symlink(struct inode *inode, const char *symname, int len);
2621 extern const struct inode_operations page_symlink_inode_operations;
2622 extern int generic_readlink(struct dentry *, char __user *, int);
2623 extern void generic_fillattr(struct inode *, struct kstat *);
2624 extern int vfs_getattr(struct path *, struct kstat *);
2625 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2626 void inode_add_bytes(struct inode *inode, loff_t bytes);
2627 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2628 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2629 loff_t inode_get_bytes(struct inode *inode);
2630 void inode_set_bytes(struct inode *inode, loff_t bytes);
2632 extern int vfs_readdir(struct file *, filldir_t, void *);
2633 extern int iterate_dir(struct file *, struct dir_context *);
2635 extern int vfs_stat(const char __user *, struct kstat *);
2636 extern int vfs_lstat(const char __user *, struct kstat *);
2637 extern int vfs_fstat(unsigned int, struct kstat *);
2638 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2640 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2642 extern int __generic_block_fiemap(struct inode *inode,
2643 struct fiemap_extent_info *fieinfo,
2644 loff_t start, loff_t len,
2645 get_block_t *get_block);
2646 extern int generic_block_fiemap(struct inode *inode,
2647 struct fiemap_extent_info *fieinfo, u64 start,
2648 u64 len, get_block_t *get_block);
2650 extern void get_filesystem(struct file_system_type *fs);
2651 extern void put_filesystem(struct file_system_type *fs);
2652 extern struct file_system_type *get_fs_type(const char *name);
2653 extern struct super_block *get_super(struct block_device *);
2654 extern struct super_block *get_super_thawed(struct block_device *);
2655 extern struct super_block *get_active_super(struct block_device *bdev);
2656 extern void drop_super(struct super_block *sb);
2657 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2658 extern void iterate_supers_type(struct file_system_type *,
2659 void (*)(struct super_block *, void *), void *);
2661 extern int dcache_dir_open(struct inode *, struct file *);
2662 extern int dcache_dir_close(struct inode *, struct file *);
2663 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2664 extern int dcache_readdir(struct file *, struct dir_context *);
2665 extern int simple_setattr(struct dentry *, struct iattr *);
2666 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2667 extern int simple_statfs(struct dentry *, struct kstatfs *);
2668 extern int simple_open(struct inode *inode, struct file *file);
2669 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2670 extern int simple_unlink(struct inode *, struct dentry *);
2671 extern int simple_rmdir(struct inode *, struct dentry *);
2672 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2673 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2674 extern int simple_empty(struct dentry *);
2675 extern int simple_readpage(struct file *file, struct page *page);
2676 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2677 loff_t pos, unsigned len, unsigned flags,
2678 struct page **pagep, void **fsdata);
2679 extern int simple_write_end(struct file *file, struct address_space *mapping,
2680 loff_t pos, unsigned len, unsigned copied,
2681 struct page *page, void *fsdata);
2683 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2684 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2685 extern const struct file_operations simple_dir_operations;
2686 extern const struct inode_operations simple_dir_inode_operations;
2687 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2688 struct dentry *d_alloc_name(struct dentry *, const char *);
2689 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2690 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2691 extern void simple_release_fs(struct vfsmount **mount, int *count);
2693 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2694 loff_t *ppos, const void *from, size_t available);
2695 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2696 const void __user *from, size_t count);
2698 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2700 extern int generic_check_addressable(unsigned, u64);
2702 #ifdef CONFIG_MIGRATION
2703 extern int buffer_migrate_page(struct address_space *,
2704 struct page *, struct page *,
2707 #define buffer_migrate_page NULL
2710 extern int inode_change_ok(const struct inode *, struct iattr *);
2711 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2712 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2714 extern int file_update_time(struct file *file);
2716 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2717 extern void save_mount_options(struct super_block *sb, char *options);
2718 extern void replace_mount_options(struct super_block *sb, char *options);
2720 static inline ino_t parent_ino(struct dentry *dentry)
2725 * Don't strictly need d_lock here? If the parent ino could change
2726 * then surely we'd have a deeper race in the caller?
2728 spin_lock(&dentry->d_lock);
2729 res = dentry->d_parent->d_inode->i_ino;
2730 spin_unlock(&dentry->d_lock);
2734 /* Transaction based IO helpers */
2737 * An argresp is stored in an allocated page and holds the
2738 * size of the argument or response, along with its content
2740 struct simple_transaction_argresp {
2745 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2747 char *simple_transaction_get(struct file *file, const char __user *buf,
2749 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2750 size_t size, loff_t *pos);
2751 int simple_transaction_release(struct inode *inode, struct file *file);
2753 void simple_transaction_set(struct file *file, size_t n);
2756 * simple attribute files
2758 * These attributes behave similar to those in sysfs:
2760 * Writing to an attribute immediately sets a value, an open file can be
2761 * written to multiple times.
2763 * Reading from an attribute creates a buffer from the value that might get
2764 * read with multiple read calls. When the attribute has been read
2765 * completely, no further read calls are possible until the file is opened
2768 * All attributes contain a text representation of a numeric value
2769 * that are accessed with the get() and set() functions.
2771 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2772 static int __fops ## _open(struct inode *inode, struct file *file) \
2774 __simple_attr_check_format(__fmt, 0ull); \
2775 return simple_attr_open(inode, file, __get, __set, __fmt); \
2777 static const struct file_operations __fops = { \
2778 .owner = THIS_MODULE, \
2779 .open = __fops ## _open, \
2780 .release = simple_attr_release, \
2781 .read = simple_attr_read, \
2782 .write = simple_attr_write, \
2783 .llseek = generic_file_llseek, \
2786 static inline __printf(1, 2)
2787 void __simple_attr_check_format(const char *fmt, ...)
2789 /* don't do anything, just let the compiler check the arguments; */
2792 int simple_attr_open(struct inode *inode, struct file *file,
2793 int (*get)(void *, u64 *), int (*set)(void *, u64),
2795 int simple_attr_release(struct inode *inode, struct file *file);
2796 ssize_t simple_attr_read(struct file *file, char __user *buf,
2797 size_t len, loff_t *ppos);
2798 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2799 size_t len, loff_t *ppos);
2802 int proc_nr_files(struct ctl_table *table, int write,
2803 void __user *buffer, size_t *lenp, loff_t *ppos);
2804 int proc_nr_dentry(struct ctl_table *table, int write,
2805 void __user *buffer, size_t *lenp, loff_t *ppos);
2806 int proc_nr_inodes(struct ctl_table *table, int write,
2807 void __user *buffer, size_t *lenp, loff_t *ppos);
2808 int __init get_filesystem_list(char *buf);
2810 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2811 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2813 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2814 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2815 (flag & __FMODE_NONOTIFY)))
2817 static inline int is_sxid(umode_t mode)
2819 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2822 static inline void inode_has_no_xattr(struct inode *inode)
2824 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2825 inode->i_flags |= S_NOSEC;
2828 static inline bool dir_emit(struct dir_context *ctx,
2829 const char *name, int namelen,
2830 u64 ino, unsigned type)
2832 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
2834 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
2836 return ctx->actor(ctx, ".", 1, ctx->pos,
2837 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
2839 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
2841 return ctx->actor(ctx, "..", 2, ctx->pos,
2842 parent_ino(file->f_path.dentry), DT_DIR) == 0;
2844 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
2846 if (ctx->pos == 0) {
2847 if (!dir_emit_dot(file, ctx))
2851 if (ctx->pos == 1) {
2852 if (!dir_emit_dotdot(file, ctx))
2858 static inline bool dir_relax(struct inode *inode)
2860 mutex_unlock(&inode->i_mutex);
2861 mutex_lock(&inode->i_mutex);
2862 return !IS_DEADDIR(inode);
2865 #endif /* _LINUX_FS_H */