5 * This file has definitions for some important file table
9 #include <linux/limits.h>
10 #include <linux/ioctl.h>
13 * It's silly to have NR_OPEN bigger than NR_FILE, but you can change
14 * the file limit at runtime and only root can increase the per-process
15 * nr_file rlimit, so it's safe to set up a ridiculously high absolute
16 * upper limit on files-per-process.
18 * Some programs (notably those using select()) may have to be
19 * recompiled to take full advantage of the new limits..
22 /* Fixed constants first: */
24 #define NR_OPEN (1024*1024) /* Absolute upper limit on fd num */
25 #define INR_OPEN 1024 /* Initial setting for nfile rlimits */
27 #define BLOCK_SIZE_BITS 10
28 #define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
30 #define SEEK_SET 0 /* seek relative to beginning of file */
31 #define SEEK_CUR 1 /* seek relative to current file position */
32 #define SEEK_END 2 /* seek relative to end of file */
33 #define SEEK_MAX SEEK_END
35 /* And dynamically-tunable limits and defaults: */
36 struct files_stat_struct {
37 int nr_files; /* read only */
38 int nr_free_files; /* read only */
39 int max_files; /* tunable */
41 extern struct files_stat_struct files_stat;
42 extern int get_max_files(void);
44 struct inodes_stat_t {
47 int dummy[5]; /* padding for sysctl ABI compatibility */
49 extern struct inodes_stat_t inodes_stat;
51 extern int leases_enable, lease_break_time;
54 extern int dir_notify_enable;
57 #define NR_FILE 8192 /* this can well be larger on a larger system */
67 /* Internal kernel extensions */
70 #define FMODE_PWRITE FMODE_PREAD /* These go hand in hand */
72 /* File is being opened for execution. Primary users of this flag are
73 distributed filesystems that can use it to achieve correct ETXTBUSY
74 behavior for cross-node execution/opening_for_writing of files */
81 #define READA 2 /* read-ahead - don't block if no resources */
82 #define SWRITE 3 /* for ll_rw_block() - wait for buffer lock */
83 #define READ_SYNC (READ | (1 << BIO_RW_SYNC))
84 #define READ_META (READ | (1 << BIO_RW_META))
85 #define WRITE_SYNC (WRITE | (1 << BIO_RW_SYNC))
86 #define WRITE_BARRIER ((1 << BIO_RW) | (1 << BIO_RW_BARRIER))
92 /* public flags for file_system_type */
93 #define FS_REQUIRES_DEV 1
94 #define FS_BINARY_MOUNTDATA 2
95 #define FS_HAS_SUBTYPE 4
96 #define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */
97 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move()
98 * during rename() internally.
102 * These are the fs-independent mount-flags: up to 32 flags are supported
104 #define MS_RDONLY 1 /* Mount read-only */
105 #define MS_NOSUID 2 /* Ignore suid and sgid bits */
106 #define MS_NODEV 4 /* Disallow access to device special files */
107 #define MS_NOEXEC 8 /* Disallow program execution */
108 #define MS_SYNCHRONOUS 16 /* Writes are synced at once */
109 #define MS_REMOUNT 32 /* Alter flags of a mounted FS */
110 #define MS_MANDLOCK 64 /* Allow mandatory locks on an FS */
111 #define MS_DIRSYNC 128 /* Directory modifications are synchronous */
112 #define MS_NOATIME 1024 /* Do not update access times. */
113 #define MS_NODIRATIME 2048 /* Do not update directory access times */
117 #define MS_VERBOSE 32768 /* War is peace. Verbosity is silence.
118 MS_VERBOSE is deprecated. */
119 #define MS_SILENT 32768
120 #define MS_POSIXACL (1<<16) /* VFS does not apply the umask */
121 #define MS_UNBINDABLE (1<<17) /* change to unbindable */
122 #define MS_PRIVATE (1<<18) /* change to private */
123 #define MS_SLAVE (1<<19) /* change to slave */
124 #define MS_SHARED (1<<20) /* change to shared */
125 #define MS_RELATIME (1<<21) /* Update atime relative to mtime/ctime. */
126 #define MS_ACTIVE (1<<30)
127 #define MS_NOUSER (1<<31)
130 * Superblock flags that can be altered by MS_REMOUNT
132 #define MS_RMT_MASK (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK)
135 * Old magic mount flag and mask
137 #define MS_MGC_VAL 0xC0ED0000
138 #define MS_MGC_MSK 0xffff0000
140 /* Inode flags - they have nothing to superblock flags now */
142 #define S_SYNC 1 /* Writes are synced at once */
143 #define S_NOATIME 2 /* Do not update access times */
144 #define S_APPEND 4 /* Append-only file */
145 #define S_IMMUTABLE 8 /* Immutable file */
146 #define S_DEAD 16 /* removed, but still open directory */
147 #define S_NOQUOTA 32 /* Inode is not counted to quota */
148 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
149 #define S_NOCMTIME 128 /* Do not update file c/mtime */
150 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
151 #define S_PRIVATE 512 /* Inode is fs-internal */
154 * Note that nosuid etc flags are inode-specific: setting some file-system
155 * flags just means all the inodes inherit those flags by default. It might be
156 * possible to override it selectively if you really wanted to with some
157 * ioctl() that is not currently implemented.
159 * Exception: MS_RDONLY is always applied to the entire file system.
161 * Unfortunately, it is possible to change a filesystems flags with it mounted
162 * with files in use. This means that all of the inodes will not have their
163 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
164 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
166 #define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
168 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
169 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
170 ((inode)->i_flags & S_SYNC))
171 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
172 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
173 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
174 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
176 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
177 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
178 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
179 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
181 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
182 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
183 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
184 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
186 /* the read-only stuff doesn't really belong here, but any other place is
187 probably as bad and I don't want to create yet another include file. */
189 #define BLKROSET _IO(0x12,93) /* set device read-only (0 = read-write) */
190 #define BLKROGET _IO(0x12,94) /* get read-only status (0 = read_write) */
191 #define BLKRRPART _IO(0x12,95) /* re-read partition table */
192 #define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */
193 #define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */
194 #define BLKRASET _IO(0x12,98) /* set read ahead for block device */
195 #define BLKRAGET _IO(0x12,99) /* get current read ahead setting */
196 #define BLKFRASET _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
197 #define BLKFRAGET _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
198 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
199 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
200 #define BLKSSZGET _IO(0x12,104)/* get block device sector size */
202 #define BLKPG _IO(0x12,105)/* See blkpg.h */
204 /* Some people are morons. Do not use sizeof! */
206 #define BLKELVGET _IOR(0x12,106,size_t)/* elevator get */
207 #define BLKELVSET _IOW(0x12,107,size_t)/* elevator set */
208 /* This was here just to show that the number is taken -
209 probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
211 /* A jump here: 108-111 have been used for various private purposes. */
212 #define BLKBSZGET _IOR(0x12,112,size_t)
213 #define BLKBSZSET _IOW(0x12,113,size_t)
214 #define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
215 #define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
216 #define BLKTRACESTART _IO(0x12,116)
217 #define BLKTRACESTOP _IO(0x12,117)
218 #define BLKTRACETEARDOWN _IO(0x12,118)
220 #define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
221 #define FIBMAP _IO(0x00,1) /* bmap access */
222 #define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */
224 #define FS_IOC_GETFLAGS _IOR('f', 1, long)
225 #define FS_IOC_SETFLAGS _IOW('f', 2, long)
226 #define FS_IOC_GETVERSION _IOR('v', 1, long)
227 #define FS_IOC_SETVERSION _IOW('v', 2, long)
228 #define FS_IOC32_GETFLAGS _IOR('f', 1, int)
229 #define FS_IOC32_SETFLAGS _IOW('f', 2, int)
230 #define FS_IOC32_GETVERSION _IOR('v', 1, int)
231 #define FS_IOC32_SETVERSION _IOW('v', 2, int)
234 * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS)
236 #define FS_SECRM_FL 0x00000001 /* Secure deletion */
237 #define FS_UNRM_FL 0x00000002 /* Undelete */
238 #define FS_COMPR_FL 0x00000004 /* Compress file */
239 #define FS_SYNC_FL 0x00000008 /* Synchronous updates */
240 #define FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
241 #define FS_APPEND_FL 0x00000020 /* writes to file may only append */
242 #define FS_NODUMP_FL 0x00000040 /* do not dump file */
243 #define FS_NOATIME_FL 0x00000080 /* do not update atime */
244 /* Reserved for compression usage... */
245 #define FS_DIRTY_FL 0x00000100
246 #define FS_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
247 #define FS_NOCOMP_FL 0x00000400 /* Don't compress */
248 #define FS_ECOMPR_FL 0x00000800 /* Compression error */
249 /* End compression flags --- maybe not all used */
250 #define FS_BTREE_FL 0x00001000 /* btree format dir */
251 #define FS_INDEX_FL 0x00001000 /* hash-indexed directory */
252 #define FS_IMAGIC_FL 0x00002000 /* AFS directory */
253 #define FS_JOURNAL_DATA_FL 0x00004000 /* Reserved for ext3 */
254 #define FS_NOTAIL_FL 0x00008000 /* file tail should not be merged */
255 #define FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
256 #define FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
257 #define FS_EXTENT_FL 0x00080000 /* Extents */
258 #define FS_DIRECTIO_FL 0x00100000 /* Use direct i/o */
259 #define FS_RESERVED_FL 0x80000000 /* reserved for ext2 lib */
261 #define FS_FL_USER_VISIBLE 0x0003DFFF /* User visible flags */
262 #define FS_FL_USER_MODIFIABLE 0x000380FF /* User modifiable flags */
265 #define SYNC_FILE_RANGE_WAIT_BEFORE 1
266 #define SYNC_FILE_RANGE_WRITE 2
267 #define SYNC_FILE_RANGE_WAIT_AFTER 4
271 #include <linux/linkage.h>
272 #include <linux/wait.h>
273 #include <linux/types.h>
274 #include <linux/kdev_t.h>
275 #include <linux/dcache.h>
276 #include <linux/namei.h>
277 #include <linux/stat.h>
278 #include <linux/cache.h>
279 #include <linux/kobject.h>
280 #include <linux/list.h>
281 #include <linux/radix-tree.h>
282 #include <linux/prio_tree.h>
283 #include <linux/init.h>
284 #include <linux/pid.h>
285 #include <linux/mutex.h>
286 #include <linux/capability.h>
288 #include <asm/atomic.h>
289 #include <asm/semaphore.h>
290 #include <asm/byteorder.h>
292 struct export_operations;
297 struct pipe_inode_info;
298 struct poll_table_struct;
300 struct vm_area_struct;
303 extern void __init inode_init(void);
304 extern void __init inode_init_early(void);
305 extern void __init mnt_init(void);
306 extern void __init files_init(unsigned long);
309 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
310 struct buffer_head *bh_result, int create);
311 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
312 ssize_t bytes, void *private);
315 * Attribute flags. These should be or-ed together to figure out what
322 #define ATTR_ATIME 16
323 #define ATTR_MTIME 32
324 #define ATTR_CTIME 64
325 #define ATTR_ATIME_SET 128
326 #define ATTR_MTIME_SET 256
327 #define ATTR_FORCE 512 /* Not a change, but a change it */
328 #define ATTR_ATTR_FLAG 1024
329 #define ATTR_KILL_SUID 2048
330 #define ATTR_KILL_SGID 4096
331 #define ATTR_FILE 8192
332 #define ATTR_KILL_PRIV 16384
333 #define ATTR_OPEN 32768 /* Truncating from open(O_TRUNC) */
336 * This is the Inode Attributes structure, used for notify_change(). It
337 * uses the above definitions as flags, to know which values have changed.
338 * Also, in this manner, a Filesystem can look at only the values it cares
339 * about. Basically, these are the attributes that the VFS layer can
340 * request to change from the FS layer.
342 * Derek Atkins <warlord@MIT.EDU> 94-10-20
345 unsigned int ia_valid;
350 struct timespec ia_atime;
351 struct timespec ia_mtime;
352 struct timespec ia_ctime;
355 * Not an attribute, but an auxilary info for filesystems wanting to
356 * implement an ftruncate() like method. NOTE: filesystem should
357 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
359 struct file *ia_file;
363 * Includes for diskquotas.
365 #include <linux/quota.h>
368 * enum positive_aop_returns - aop return codes with specific semantics
370 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
371 * completed, that the page is still locked, and
372 * should be considered active. The VM uses this hint
373 * to return the page to the active list -- it won't
374 * be a candidate for writeback again in the near
375 * future. Other callers must be careful to unlock
376 * the page if they get this return. Returned by
379 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
380 * unlocked it and the page might have been truncated.
381 * The caller should back up to acquiring a new page and
382 * trying again. The aop will be taking reasonable
383 * precautions not to livelock. If the caller held a page
384 * reference, it should drop it before retrying. Returned
387 * address_space_operation functions return these large constants to indicate
388 * special semantics to the caller. These are much larger than the bytes in a
389 * page to allow for functions that return the number of bytes operated on in a
393 enum positive_aop_returns {
394 AOP_WRITEPAGE_ACTIVATE = 0x80000,
395 AOP_TRUNCATED_PAGE = 0x80001,
398 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
399 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
402 * oh the beauties of C type declarations.
405 struct address_space;
406 struct writeback_control;
409 const struct iovec *iov;
410 unsigned long nr_segs;
415 size_t iov_iter_copy_from_user_atomic(struct page *page,
416 struct iov_iter *i, unsigned long offset, size_t bytes);
417 size_t iov_iter_copy_from_user(struct page *page,
418 struct iov_iter *i, unsigned long offset, size_t bytes);
419 void iov_iter_advance(struct iov_iter *i, size_t bytes);
420 int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
421 size_t iov_iter_single_seg_count(struct iov_iter *i);
423 static inline void iov_iter_init(struct iov_iter *i,
424 const struct iovec *iov, unsigned long nr_segs,
425 size_t count, size_t written)
428 i->nr_segs = nr_segs;
430 i->count = count + written;
432 iov_iter_advance(i, written);
435 static inline size_t iov_iter_count(struct iov_iter *i)
441 struct address_space_operations {
442 int (*writepage)(struct page *page, struct writeback_control *wbc);
443 int (*readpage)(struct file *, struct page *);
444 void (*sync_page)(struct page *);
446 /* Write back some dirty pages from this mapping. */
447 int (*writepages)(struct address_space *, struct writeback_control *);
449 /* Set a page dirty. Return true if this dirtied it */
450 int (*set_page_dirty)(struct page *page);
452 int (*readpages)(struct file *filp, struct address_space *mapping,
453 struct list_head *pages, unsigned nr_pages);
456 * ext3 requires that a successful prepare_write() call be followed
457 * by a commit_write() call - they must be balanced
459 int (*prepare_write)(struct file *, struct page *, unsigned, unsigned);
460 int (*commit_write)(struct file *, struct page *, unsigned, unsigned);
462 int (*write_begin)(struct file *, struct address_space *mapping,
463 loff_t pos, unsigned len, unsigned flags,
464 struct page **pagep, void **fsdata);
465 int (*write_end)(struct file *, struct address_space *mapping,
466 loff_t pos, unsigned len, unsigned copied,
467 struct page *page, void *fsdata);
469 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
470 sector_t (*bmap)(struct address_space *, sector_t);
471 void (*invalidatepage) (struct page *, unsigned long);
472 int (*releasepage) (struct page *, gfp_t);
473 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
474 loff_t offset, unsigned long nr_segs);
475 struct page* (*get_xip_page)(struct address_space *, sector_t,
477 /* migrate the contents of a page to the specified target */
478 int (*migratepage) (struct address_space *,
479 struct page *, struct page *);
480 int (*launder_page) (struct page *);
484 * pagecache_write_begin/pagecache_write_end must be used by general code
485 * to write into the pagecache.
487 int pagecache_write_begin(struct file *, struct address_space *mapping,
488 loff_t pos, unsigned len, unsigned flags,
489 struct page **pagep, void **fsdata);
491 int pagecache_write_end(struct file *, struct address_space *mapping,
492 loff_t pos, unsigned len, unsigned copied,
493 struct page *page, void *fsdata);
495 struct backing_dev_info;
496 struct address_space {
497 struct inode *host; /* owner: inode, block_device */
498 struct radix_tree_root page_tree; /* radix tree of all pages */
499 rwlock_t tree_lock; /* and rwlock protecting it */
500 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
501 struct prio_tree_root i_mmap; /* tree of private and shared mappings */
502 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
503 spinlock_t i_mmap_lock; /* protect tree, count, list */
504 unsigned int truncate_count; /* Cover race condition with truncate */
505 unsigned long nrpages; /* number of total pages */
506 pgoff_t writeback_index;/* writeback starts here */
507 const struct address_space_operations *a_ops; /* methods */
508 unsigned long flags; /* error bits/gfp mask */
509 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
510 spinlock_t private_lock; /* for use by the address_space */
511 struct list_head private_list; /* ditto */
512 struct address_space *assoc_mapping; /* ditto */
513 } __attribute__((aligned(sizeof(long))));
515 * On most architectures that alignment is already the case; but
516 * must be enforced here for CRIS, to let the least signficant bit
517 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
520 struct block_device {
521 dev_t bd_dev; /* not a kdev_t - it's a search key */
522 struct inode * bd_inode; /* will die */
524 struct mutex bd_mutex; /* open/close mutex */
525 struct semaphore bd_mount_sem;
526 struct list_head bd_inodes;
530 struct list_head bd_holder_list;
532 struct block_device * bd_contains;
533 unsigned bd_block_size;
534 struct hd_struct * bd_part;
535 /* number of times partitions within this device have been opened. */
536 unsigned bd_part_count;
538 struct gendisk * bd_disk;
539 struct list_head bd_list;
540 struct backing_dev_info *bd_inode_backing_dev_info;
542 * Private data. You must have bd_claim'ed the block_device
543 * to use this. NOTE: bd_claim allows an owner to claim
544 * the same device multiple times, the owner must take special
545 * care to not mess up bd_private for that case.
547 unsigned long bd_private;
551 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
554 #define PAGECACHE_TAG_DIRTY 0
555 #define PAGECACHE_TAG_WRITEBACK 1
557 int mapping_tagged(struct address_space *mapping, int tag);
560 * Might pages of this file be mapped into userspace?
562 static inline int mapping_mapped(struct address_space *mapping)
564 return !prio_tree_empty(&mapping->i_mmap) ||
565 !list_empty(&mapping->i_mmap_nonlinear);
569 * Might pages of this file have been modified in userspace?
570 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
571 * marks vma as VM_SHARED if it is shared, and the file was opened for
572 * writing i.e. vma may be mprotected writable even if now readonly.
574 static inline int mapping_writably_mapped(struct address_space *mapping)
576 return mapping->i_mmap_writable != 0;
580 * Use sequence counter to get consistent i_size on 32-bit processors.
582 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
583 #include <linux/seqlock.h>
584 #define __NEED_I_SIZE_ORDERED
585 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
587 #define i_size_ordered_init(inode) do { } while (0)
591 struct hlist_node i_hash;
592 struct list_head i_list;
593 struct list_head i_sb_list;
594 struct list_head i_dentry;
597 unsigned int i_nlink;
601 unsigned long i_version;
603 #ifdef __NEED_I_SIZE_ORDERED
604 seqcount_t i_size_seqcount;
606 struct timespec i_atime;
607 struct timespec i_mtime;
608 struct timespec i_ctime;
609 unsigned int i_blkbits;
611 unsigned short i_bytes;
613 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
614 struct mutex i_mutex;
615 struct rw_semaphore i_alloc_sem;
616 const struct inode_operations *i_op;
617 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
618 struct super_block *i_sb;
619 struct file_lock *i_flock;
620 struct address_space *i_mapping;
621 struct address_space i_data;
623 struct dquot *i_dquot[MAXQUOTAS];
625 struct list_head i_devices;
627 struct pipe_inode_info *i_pipe;
628 struct block_device *i_bdev;
635 #ifdef CONFIG_DNOTIFY
636 unsigned long i_dnotify_mask; /* Directory notify events */
637 struct dnotify_struct *i_dnotify; /* for directory notifications */
640 #ifdef CONFIG_INOTIFY
641 struct list_head inotify_watches; /* watches on this inode */
642 struct mutex inotify_mutex; /* protects the watches list */
645 unsigned long i_state;
646 unsigned long dirtied_when; /* jiffies of first dirtying */
648 unsigned int i_flags;
650 atomic_t i_writecount;
651 #ifdef CONFIG_SECURITY
654 void *i_private; /* fs or device private pointer */
658 * inode->i_mutex nesting subclasses for the lock validator:
660 * 0: the object of the current VFS operation
665 * The locking order between these classes is
666 * parent -> child -> normal -> xattr -> quota
668 enum inode_i_mutex_lock_class
677 extern void inode_double_lock(struct inode *inode1, struct inode *inode2);
678 extern void inode_double_unlock(struct inode *inode1, struct inode *inode2);
681 * NOTE: in a 32bit arch with a preemptable kernel and
682 * an UP compile the i_size_read/write must be atomic
683 * with respect to the local cpu (unlike with preempt disabled),
684 * but they don't need to be atomic with respect to other cpus like in
685 * true SMP (so they need either to either locally disable irq around
686 * the read or for example on x86 they can be still implemented as a
687 * cmpxchg8b without the need of the lock prefix). For SMP compiles
688 * and 64bit archs it makes no difference if preempt is enabled or not.
690 static inline loff_t i_size_read(const struct inode *inode)
692 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
697 seq = read_seqcount_begin(&inode->i_size_seqcount);
698 i_size = inode->i_size;
699 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
701 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
705 i_size = inode->i_size;
709 return inode->i_size;
714 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
715 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
716 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
718 static inline void i_size_write(struct inode *inode, loff_t i_size)
720 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
721 write_seqcount_begin(&inode->i_size_seqcount);
722 inode->i_size = i_size;
723 write_seqcount_end(&inode->i_size_seqcount);
724 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
726 inode->i_size = i_size;
729 inode->i_size = i_size;
733 static inline unsigned iminor(const struct inode *inode)
735 return MINOR(inode->i_rdev);
738 static inline unsigned imajor(const struct inode *inode)
740 return MAJOR(inode->i_rdev);
743 extern struct block_device *I_BDEV(struct inode *inode);
746 rwlock_t lock; /* protects pid, uid, euid fields */
747 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
748 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
749 uid_t uid, euid; /* uid/euid of process setting the owner */
750 int signum; /* posix.1b rt signal to be delivered on IO */
754 * Track a single file's readahead state
756 struct file_ra_state {
757 pgoff_t start; /* where readahead started */
758 unsigned int size; /* # of readahead pages */
759 unsigned int async_size; /* do asynchronous readahead when
760 there are only # of pages ahead */
762 unsigned int ra_pages; /* Maximum readahead window */
763 int mmap_miss; /* Cache miss stat for mmap accesses */
764 loff_t prev_pos; /* Cache last read() position */
768 * Check if @index falls in the readahead windows.
770 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
772 return (index >= ra->start &&
773 index < ra->start + ra->size);
778 * fu_list becomes invalid after file_free is called and queued via
779 * fu_rcuhead for RCU freeing
782 struct list_head fu_list;
783 struct rcu_head fu_rcuhead;
786 #define f_dentry f_path.dentry
787 #define f_vfsmnt f_path.mnt
788 const struct file_operations *f_op;
790 unsigned int f_flags;
793 struct fown_struct f_owner;
794 unsigned int f_uid, f_gid;
795 struct file_ra_state f_ra;
798 #ifdef CONFIG_SECURITY
801 /* needed for tty driver, and maybe others */
805 /* Used by fs/eventpoll.c to link all the hooks to this file */
806 struct list_head f_ep_links;
807 spinlock_t f_ep_lock;
808 #endif /* #ifdef CONFIG_EPOLL */
809 struct address_space *f_mapping;
811 extern spinlock_t files_lock;
812 #define file_list_lock() spin_lock(&files_lock);
813 #define file_list_unlock() spin_unlock(&files_lock);
815 #define get_file(x) atomic_inc(&(x)->f_count)
816 #define file_count(x) atomic_read(&(x)->f_count)
818 #define MAX_NON_LFS ((1UL<<31) - 1)
820 /* Page cache limit. The filesystems should put that into their s_maxbytes
821 limits, otherwise bad things can happen in VM. */
822 #if BITS_PER_LONG==32
823 #define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
824 #elif BITS_PER_LONG==64
825 #define MAX_LFS_FILESIZE 0x7fffffffffffffffUL
830 #define FL_ACCESS 8 /* not trying to lock, just looking */
831 #define FL_EXISTS 16 /* when unlocking, test for existence */
832 #define FL_LEASE 32 /* lease held on this file */
833 #define FL_CLOSE 64 /* unlock on close */
834 #define FL_SLEEP 128 /* A blocking lock */
837 * The POSIX file lock owner is determined by
838 * the "struct files_struct" in the thread group
839 * (or NULL for no owner - BSD locks).
841 * Lockd stuffs a "host" pointer into this.
843 typedef struct files_struct *fl_owner_t;
845 struct file_lock_operations {
846 void (*fl_insert)(struct file_lock *); /* lock insertion callback */
847 void (*fl_remove)(struct file_lock *); /* lock removal callback */
848 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
849 void (*fl_release_private)(struct file_lock *);
852 struct lock_manager_operations {
853 int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
854 void (*fl_notify)(struct file_lock *); /* unblock callback */
855 int (*fl_grant)(struct file_lock *, struct file_lock *, int);
856 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
857 void (*fl_release_private)(struct file_lock *);
858 void (*fl_break)(struct file_lock *);
859 int (*fl_mylease)(struct file_lock *, struct file_lock *);
860 int (*fl_change)(struct file_lock **, int);
863 /* that will die - we need it for nfs_lock_info */
864 #include <linux/nfs_fs_i.h>
867 struct file_lock *fl_next; /* singly linked list for this inode */
868 struct list_head fl_link; /* doubly linked list of all locks */
869 struct list_head fl_block; /* circular list of blocked processes */
872 wait_queue_head_t fl_wait;
873 struct file *fl_file;
874 unsigned char fl_flags;
875 unsigned char fl_type;
879 struct fasync_struct * fl_fasync; /* for lease break notifications */
880 unsigned long fl_break_time; /* for nonblocking lease breaks */
882 struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
883 struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
885 struct nfs_lock_info nfs_fl;
886 struct nfs4_lock_info nfs4_fl;
888 struct list_head link; /* link in AFS vnode's pending_locks list */
889 int state; /* state of grant or error if -ve */
894 /* The following constant reflects the upper bound of the file/locking space */
896 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
897 #define OFFSET_MAX INT_LIMIT(loff_t)
898 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
901 #include <linux/fcntl.h>
903 extern int fcntl_getlk(struct file *, struct flock __user *);
904 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
905 struct flock __user *);
907 #if BITS_PER_LONG == 32
908 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
909 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
910 struct flock64 __user *);
913 extern void send_sigio(struct fown_struct *fown, int fd, int band);
914 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
915 extern int fcntl_getlease(struct file *filp);
918 extern int do_sync_mapping_range(struct address_space *mapping, loff_t offset,
919 loff_t endbyte, unsigned int flags);
922 extern void locks_init_lock(struct file_lock *);
923 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
924 extern void locks_remove_posix(struct file *, fl_owner_t);
925 extern void locks_remove_flock(struct file *);
926 extern void posix_test_lock(struct file *, struct file_lock *);
927 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
928 extern int posix_lock_file_wait(struct file *, struct file_lock *);
929 extern int posix_unblock_lock(struct file *, struct file_lock *);
930 extern int vfs_test_lock(struct file *, struct file_lock *);
931 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
932 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
933 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
934 extern int __break_lease(struct inode *inode, unsigned int flags);
935 extern void lease_get_mtime(struct inode *, struct timespec *time);
936 extern int generic_setlease(struct file *, long, struct file_lock **);
937 extern int vfs_setlease(struct file *, long, struct file_lock **);
938 extern int lease_modify(struct file_lock **, int);
939 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
940 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
941 extern struct seq_operations locks_seq_operations;
943 struct fasync_struct {
946 struct fasync_struct *fa_next; /* singly linked list */
947 struct file *fa_file;
950 #define FASYNC_MAGIC 0x4601
952 /* SMP safe fasync helpers: */
953 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
954 /* can be called from interrupts */
955 extern void kill_fasync(struct fasync_struct **, int, int);
956 /* only for net: no internal synchronization */
957 extern void __kill_fasync(struct fasync_struct *, int, int);
959 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
960 extern int f_setown(struct file *filp, unsigned long arg, int force);
961 extern void f_delown(struct file *filp);
962 extern pid_t f_getown(struct file *filp);
963 extern int send_sigurg(struct fown_struct *fown);
969 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
970 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
971 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
973 extern struct list_head super_blocks;
974 extern spinlock_t sb_lock;
976 #define sb_entry(list) list_entry((list), struct super_block, s_list)
977 #define S_BIAS (1<<30)
979 struct list_head s_list; /* Keep this first */
980 dev_t s_dev; /* search index; _not_ kdev_t */
981 unsigned long s_blocksize;
982 unsigned char s_blocksize_bits;
983 unsigned char s_dirt;
984 unsigned long long s_maxbytes; /* Max file size */
985 struct file_system_type *s_type;
986 const struct super_operations *s_op;
987 struct dquot_operations *dq_op;
988 struct quotactl_ops *s_qcop;
989 struct export_operations *s_export_op;
990 unsigned long s_flags;
991 unsigned long s_magic;
992 struct dentry *s_root;
993 struct rw_semaphore s_umount;
999 #ifdef CONFIG_SECURITY
1002 struct xattr_handler **s_xattr;
1004 struct list_head s_inodes; /* all inodes */
1005 struct list_head s_dirty; /* dirty inodes */
1006 struct list_head s_io; /* parked for writeback */
1007 struct list_head s_more_io; /* parked for more writeback */
1008 struct hlist_head s_anon; /* anonymous dentries for (nfs) exporting */
1009 struct list_head s_files;
1011 struct block_device *s_bdev;
1012 struct mtd_info *s_mtd;
1013 struct list_head s_instances;
1014 struct quota_info s_dquot; /* Diskquota specific options */
1017 wait_queue_head_t s_wait_unfrozen;
1019 char s_id[32]; /* Informational name */
1021 void *s_fs_info; /* Filesystem private info */
1024 * The next field is for VFS *only*. No filesystems have any business
1025 * even looking at it. You had been warned.
1027 struct mutex s_vfs_rename_mutex; /* Kludge */
1029 /* Granularity of c/m/atime in ns.
1030 Cannot be worse than a second */
1034 * Filesystem subtype. If non-empty the filesystem type field
1035 * in /proc/mounts will be "type.subtype"
1040 extern struct timespec current_fs_time(struct super_block *sb);
1043 * Snapshotting support.
1047 SB_FREEZE_WRITE = 1,
1048 SB_FREEZE_TRANS = 2,
1051 #define vfs_check_frozen(sb, level) \
1052 wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
1054 #define get_fs_excl() atomic_inc(¤t->fs_excl)
1055 #define put_fs_excl() atomic_dec(¤t->fs_excl)
1056 #define has_fs_excl() atomic_read(¤t->fs_excl)
1058 #define is_owner_or_cap(inode) \
1059 ((current->fsuid == (inode)->i_uid) || capable(CAP_FOWNER))
1061 /* not quite ready to be deprecated, but... */
1062 extern void lock_super(struct super_block *);
1063 extern void unlock_super(struct super_block *);
1066 * VFS helper functions..
1068 extern int vfs_permission(struct nameidata *, int);
1069 extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
1070 extern int vfs_mkdir(struct inode *, struct dentry *, int);
1071 extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
1072 extern int vfs_symlink(struct inode *, struct dentry *, const char *, int);
1073 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1074 extern int vfs_rmdir(struct inode *, struct dentry *);
1075 extern int vfs_unlink(struct inode *, struct dentry *);
1076 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1079 * VFS dentry helper functions.
1081 extern void dentry_unhash(struct dentry *dentry);
1084 * VFS file helper functions.
1086 extern int file_permission(struct file *, int);
1091 * NOTE! These match bits 12..15 of stat.st_mode
1092 * (ie "(i_mode >> 12) & 15").
1094 #define DT_UNKNOWN 0
1104 #define OSYNC_METADATA (1<<0)
1105 #define OSYNC_DATA (1<<1)
1106 #define OSYNC_INODE (1<<2)
1107 int generic_osync_inode(struct inode *, struct address_space *, int);
1110 * This is the "filldir" function type, used by readdir() to let
1111 * the kernel specify what kind of dirent layout it wants to have.
1112 * This allows the kernel to read directories into kernel space or
1113 * to have different dirent layouts depending on the binary type.
1115 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1117 struct block_device_operations {
1118 int (*open) (struct inode *, struct file *);
1119 int (*release) (struct inode *, struct file *);
1120 int (*ioctl) (struct inode *, struct file *, unsigned, unsigned long);
1121 long (*unlocked_ioctl) (struct file *, unsigned, unsigned long);
1122 long (*compat_ioctl) (struct file *, unsigned, unsigned long);
1123 int (*direct_access) (struct block_device *, sector_t, unsigned long *);
1124 int (*media_changed) (struct gendisk *);
1125 int (*revalidate_disk) (struct gendisk *);
1126 int (*getgeo)(struct block_device *, struct hd_geometry *);
1127 struct module *owner;
1131 * "descriptor" for what we're up to with a read.
1132 * This allows us to use the same read code yet
1133 * have multiple different users of the data that
1134 * we read from a file.
1136 * The simplest case just copies the data to user
1147 } read_descriptor_t;
1149 typedef int (*read_actor_t)(read_descriptor_t *, struct page *, unsigned long, unsigned long);
1151 /* These macros are for out of kernel modules to test that
1152 * the kernel supports the unlocked_ioctl and compat_ioctl
1153 * fields in struct file_operations. */
1154 #define HAVE_COMPAT_IOCTL 1
1155 #define HAVE_UNLOCKED_IOCTL 1
1159 * read, write, poll, fsync, readv, writev, unlocked_ioctl and compat_ioctl
1160 * can be called without the big kernel lock held in all filesystems.
1162 struct file_operations {
1163 struct module *owner;
1164 loff_t (*llseek) (struct file *, loff_t, int);
1165 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1166 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1167 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1168 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1169 int (*readdir) (struct file *, void *, filldir_t);
1170 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1171 int (*ioctl) (struct inode *, struct file *, unsigned int, unsigned long);
1172 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1173 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1174 int (*mmap) (struct file *, struct vm_area_struct *);
1175 int (*open) (struct inode *, struct file *);
1176 int (*flush) (struct file *, fl_owner_t id);
1177 int (*release) (struct inode *, struct file *);
1178 int (*fsync) (struct file *, struct dentry *, int datasync);
1179 int (*aio_fsync) (struct kiocb *, int datasync);
1180 int (*fasync) (int, struct file *, int);
1181 int (*lock) (struct file *, int, struct file_lock *);
1182 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1183 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1184 int (*check_flags)(int);
1185 int (*dir_notify)(struct file *filp, unsigned long arg);
1186 int (*flock) (struct file *, int, struct file_lock *);
1187 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1188 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1189 int (*setlease)(struct file *, long, struct file_lock **);
1192 struct inode_operations {
1193 int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1194 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1195 int (*link) (struct dentry *,struct inode *,struct dentry *);
1196 int (*unlink) (struct inode *,struct dentry *);
1197 int (*symlink) (struct inode *,struct dentry *,const char *);
1198 int (*mkdir) (struct inode *,struct dentry *,int);
1199 int (*rmdir) (struct inode *,struct dentry *);
1200 int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1201 int (*rename) (struct inode *, struct dentry *,
1202 struct inode *, struct dentry *);
1203 int (*readlink) (struct dentry *, char __user *,int);
1204 void * (*follow_link) (struct dentry *, struct nameidata *);
1205 void (*put_link) (struct dentry *, struct nameidata *, void *);
1206 void (*truncate) (struct inode *);
1207 int (*permission) (struct inode *, int, struct nameidata *);
1208 int (*setattr) (struct dentry *, struct iattr *);
1209 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1210 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1211 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1212 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1213 int (*removexattr) (struct dentry *, const char *);
1214 void (*truncate_range)(struct inode *, loff_t, loff_t);
1215 long (*fallocate)(struct inode *inode, int mode, loff_t offset,
1221 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1222 unsigned long nr_segs, unsigned long fast_segs,
1223 struct iovec *fast_pointer,
1224 struct iovec **ret_pointer);
1226 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1227 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1228 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1229 unsigned long, loff_t *);
1230 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1231 unsigned long, loff_t *);
1234 * NOTE: write_inode, delete_inode, clear_inode, put_inode can be called
1235 * without the big kernel lock held in all filesystems.
1237 struct super_operations {
1238 struct inode *(*alloc_inode)(struct super_block *sb);
1239 void (*destroy_inode)(struct inode *);
1241 void (*read_inode) (struct inode *);
1243 void (*dirty_inode) (struct inode *);
1244 int (*write_inode) (struct inode *, int);
1245 void (*put_inode) (struct inode *);
1246 void (*drop_inode) (struct inode *);
1247 void (*delete_inode) (struct inode *);
1248 void (*put_super) (struct super_block *);
1249 void (*write_super) (struct super_block *);
1250 int (*sync_fs)(struct super_block *sb, int wait);
1251 void (*write_super_lockfs) (struct super_block *);
1252 void (*unlockfs) (struct super_block *);
1253 int (*statfs) (struct dentry *, struct kstatfs *);
1254 int (*remount_fs) (struct super_block *, int *, char *);
1255 void (*clear_inode) (struct inode *);
1256 void (*umount_begin) (struct vfsmount *, int);
1258 int (*show_options)(struct seq_file *, struct vfsmount *);
1259 int (*show_stats)(struct seq_file *, struct vfsmount *);
1261 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1262 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1267 * Inode state bits. Protected by inode_lock.
1269 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1270 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1272 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1273 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1274 * various stages of removing an inode.
1276 * Two bits are used for locking and completion notification, I_LOCK and I_SYNC.
1278 * I_DIRTY_SYNC Inode itself is dirty.
1279 * I_DIRTY_DATASYNC Data-related inode changes pending
1280 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1281 * I_NEW get_new_inode() sets i_state to I_LOCK|I_NEW. Both
1282 * are cleared by unlock_new_inode(), called from iget().
1283 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1284 * is zero. I_FREEING must be set when I_WILL_FREE is
1286 * I_FREEING Set when inode is about to be freed but still has dirty
1287 * pages or buffers attached or the inode itself is still
1289 * I_CLEAR Set by clear_inode(). In this state the inode is clean
1290 * and can be destroyed.
1292 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1293 * prohibited for many purposes. iget() must wait for
1294 * the inode to be completely released, then create it
1295 * anew. Other functions will just ignore such inodes,
1296 * if appropriate. I_LOCK is used for waiting.
1298 * I_LOCK Serves as both a mutex and completion notification.
1299 * New inodes set I_LOCK. If two processes both create
1300 * the same inode, one of them will release its inode and
1301 * wait for I_LOCK to be released before returning.
1302 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1303 * also cause waiting on I_LOCK, without I_LOCK actually
1304 * being set. find_inode() uses this to prevent returning
1305 * nearly-dead inodes.
1306 * I_SYNC Similar to I_LOCK, but limited in scope to writeback
1307 * of inode dirty data. Having a seperate lock for this
1308 * purpose reduces latency and prevents some filesystem-
1309 * specific deadlocks.
1311 * Q: Why does I_DIRTY_DATASYNC exist? It appears as if it could be replaced
1312 * by (I_DIRTY_SYNC|I_DIRTY_PAGES).
1313 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1314 * Q: igrab() only checks on (I_FREEING|I_WILL_FREE). Should it also check on
1315 * I_CLEAR? If not, why?
1317 #define I_DIRTY_SYNC 1
1318 #define I_DIRTY_DATASYNC 2
1319 #define I_DIRTY_PAGES 4
1321 #define I_WILL_FREE 16
1322 #define I_FREEING 32
1325 #define I_LOCK (1 << __I_LOCK)
1327 #define I_SYNC (1 << __I_SYNC)
1329 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1331 extern void __mark_inode_dirty(struct inode *, int);
1332 static inline void mark_inode_dirty(struct inode *inode)
1334 __mark_inode_dirty(inode, I_DIRTY);
1337 static inline void mark_inode_dirty_sync(struct inode *inode)
1339 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1343 * inc_nlink - directly increment an inode's link count
1346 * This is a low-level filesystem helper to replace any
1347 * direct filesystem manipulation of i_nlink. Currently,
1348 * it is only here for parity with dec_nlink().
1350 static inline void inc_nlink(struct inode *inode)
1355 static inline void inode_inc_link_count(struct inode *inode)
1358 mark_inode_dirty(inode);
1362 * drop_nlink - directly drop an inode's link count
1365 * This is a low-level filesystem helper to replace any
1366 * direct filesystem manipulation of i_nlink. In cases
1367 * where we are attempting to track writes to the
1368 * filesystem, a decrement to zero means an imminent
1369 * write when the file is truncated and actually unlinked
1370 * on the filesystem.
1372 static inline void drop_nlink(struct inode *inode)
1378 * clear_nlink - directly zero an inode's link count
1381 * This is a low-level filesystem helper to replace any
1382 * direct filesystem manipulation of i_nlink. See
1383 * drop_nlink() for why we care about i_nlink hitting zero.
1385 static inline void clear_nlink(struct inode *inode)
1390 static inline void inode_dec_link_count(struct inode *inode)
1393 mark_inode_dirty(inode);
1396 extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
1397 static inline void file_accessed(struct file *file)
1399 if (!(file->f_flags & O_NOATIME))
1400 touch_atime(file->f_path.mnt, file->f_path.dentry);
1403 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1405 struct file_system_type {
1408 int (*get_sb) (struct file_system_type *, int,
1409 const char *, void *, struct vfsmount *);
1410 void (*kill_sb) (struct super_block *);
1411 struct module *owner;
1412 struct file_system_type * next;
1413 struct list_head fs_supers;
1415 struct lock_class_key s_lock_key;
1416 struct lock_class_key s_umount_key;
1418 struct lock_class_key i_lock_key;
1419 struct lock_class_key i_mutex_key;
1420 struct lock_class_key i_mutex_dir_key;
1421 struct lock_class_key i_alloc_sem_key;
1424 extern int get_sb_bdev(struct file_system_type *fs_type,
1425 int flags, const char *dev_name, void *data,
1426 int (*fill_super)(struct super_block *, void *, int),
1427 struct vfsmount *mnt);
1428 extern int get_sb_single(struct file_system_type *fs_type,
1429 int flags, void *data,
1430 int (*fill_super)(struct super_block *, void *, int),
1431 struct vfsmount *mnt);
1432 extern int get_sb_nodev(struct file_system_type *fs_type,
1433 int flags, void *data,
1434 int (*fill_super)(struct super_block *, void *, int),
1435 struct vfsmount *mnt);
1436 void generic_shutdown_super(struct super_block *sb);
1437 void kill_block_super(struct super_block *sb);
1438 void kill_anon_super(struct super_block *sb);
1439 void kill_litter_super(struct super_block *sb);
1440 void deactivate_super(struct super_block *sb);
1441 int set_anon_super(struct super_block *s, void *data);
1442 struct super_block *sget(struct file_system_type *type,
1443 int (*test)(struct super_block *,void *),
1444 int (*set)(struct super_block *,void *),
1446 extern int get_sb_pseudo(struct file_system_type *, char *,
1447 const struct super_operations *ops, unsigned long,
1448 struct vfsmount *mnt);
1449 extern int simple_set_mnt(struct vfsmount *mnt, struct super_block *sb);
1450 int __put_super(struct super_block *sb);
1451 int __put_super_and_need_restart(struct super_block *sb);
1452 void unnamed_dev_init(void);
1454 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1455 #define fops_get(fops) \
1456 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1457 #define fops_put(fops) \
1458 do { if (fops) module_put((fops)->owner); } while(0)
1460 extern int register_filesystem(struct file_system_type *);
1461 extern int unregister_filesystem(struct file_system_type *);
1462 extern struct vfsmount *kern_mount(struct file_system_type *);
1463 extern int may_umount_tree(struct vfsmount *);
1464 extern int may_umount(struct vfsmount *);
1465 extern void umount_tree(struct vfsmount *, int, struct list_head *);
1466 extern void release_mounts(struct list_head *);
1467 extern long do_mount(char *, char *, char *, unsigned long, void *);
1468 extern struct vfsmount *copy_tree(struct vfsmount *, struct dentry *, int);
1469 extern void mnt_set_mountpoint(struct vfsmount *, struct dentry *,
1472 extern int vfs_statfs(struct dentry *, struct kstatfs *);
1475 extern struct kset fs_subsys;
1477 #define FLOCK_VERIFY_READ 1
1478 #define FLOCK_VERIFY_WRITE 2
1480 extern int locks_mandatory_locked(struct inode *);
1481 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1484 * Candidates for mandatory locking have the setgid bit set
1485 * but no group execute bit - an otherwise meaningless combination.
1488 static inline int __mandatory_lock(struct inode *ino)
1490 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1494 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1495 * otherwise these will be advisory locks
1498 static inline int mandatory_lock(struct inode *ino)
1500 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1503 static inline int locks_verify_locked(struct inode *inode)
1505 if (mandatory_lock(inode))
1506 return locks_mandatory_locked(inode);
1510 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1512 static inline int locks_verify_truncate(struct inode *inode,
1516 if (inode->i_flock && mandatory_lock(inode))
1517 return locks_mandatory_area(
1518 FLOCK_VERIFY_WRITE, inode, filp,
1519 size < inode->i_size ? size : inode->i_size,
1520 (size < inode->i_size ? inode->i_size - size
1521 : size - inode->i_size)
1526 static inline int break_lease(struct inode *inode, unsigned int mode)
1529 return __break_lease(inode, mode);
1535 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
1537 extern long do_sys_open(int fdf, const char __user *filename, int flags,
1539 extern struct file *filp_open(const char *, int, int);
1540 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int);
1541 extern int filp_close(struct file *, fl_owner_t id);
1542 extern char * getname(const char __user *);
1545 extern void __init vfs_caches_init_early(void);
1546 extern void __init vfs_caches_init(unsigned long);
1548 extern struct kmem_cache *names_cachep;
1550 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
1551 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
1552 #ifndef CONFIG_AUDITSYSCALL
1553 #define putname(name) __putname(name)
1555 extern void putname(const char *name);
1559 extern int register_blkdev(unsigned int, const char *);
1560 extern void unregister_blkdev(unsigned int, const char *);
1561 extern struct block_device *bdget(dev_t);
1562 extern void bd_set_size(struct block_device *, loff_t size);
1563 extern void bd_forget(struct inode *inode);
1564 extern void bdput(struct block_device *);
1565 extern struct block_device *open_by_devnum(dev_t, unsigned);
1566 extern const struct address_space_operations def_blk_aops;
1568 static inline void bd_forget(struct inode *inode) {}
1570 extern const struct file_operations def_blk_fops;
1571 extern const struct file_operations def_chr_fops;
1572 extern const struct file_operations bad_sock_fops;
1573 extern const struct file_operations def_fifo_fops;
1575 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
1576 extern int blkdev_ioctl(struct inode *, struct file *, unsigned, unsigned long);
1577 extern int blkdev_driver_ioctl(struct inode *inode, struct file *file,
1578 struct gendisk *disk, unsigned cmd,
1580 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
1581 extern int blkdev_get(struct block_device *, mode_t, unsigned);
1582 extern int blkdev_put(struct block_device *);
1583 extern int bd_claim(struct block_device *, void *);
1584 extern void bd_release(struct block_device *);
1586 extern int bd_claim_by_disk(struct block_device *, void *, struct gendisk *);
1587 extern void bd_release_from_disk(struct block_device *, struct gendisk *);
1589 #define bd_claim_by_disk(bdev, holder, disk) bd_claim(bdev, holder)
1590 #define bd_release_from_disk(bdev, disk) bd_release(bdev)
1595 #define CHRDEV_MAJOR_HASH_SIZE 255
1596 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
1597 extern int register_chrdev_region(dev_t, unsigned, const char *);
1598 extern int register_chrdev(unsigned int, const char *,
1599 const struct file_operations *);
1600 extern void unregister_chrdev(unsigned int, const char *);
1601 extern void unregister_chrdev_region(dev_t, unsigned);
1602 extern int chrdev_open(struct inode *, struct file *);
1603 extern void chrdev_show(struct seq_file *,off_t);
1605 /* fs/block_dev.c */
1606 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
1609 #define BLKDEV_MAJOR_HASH_SIZE 255
1610 extern const char *__bdevname(dev_t, char *buffer);
1611 extern const char *bdevname(struct block_device *bdev, char *buffer);
1612 extern struct block_device *lookup_bdev(const char *);
1613 extern struct block_device *open_bdev_excl(const char *, int, void *);
1614 extern void close_bdev_excl(struct block_device *);
1615 extern void blkdev_show(struct seq_file *,off_t);
1617 #define BLKDEV_MAJOR_HASH_SIZE 0
1620 extern void init_special_inode(struct inode *, umode_t, dev_t);
1622 /* Invalid inode operations -- fs/bad_inode.c */
1623 extern void make_bad_inode(struct inode *);
1624 extern int is_bad_inode(struct inode *);
1626 extern const struct file_operations read_fifo_fops;
1627 extern const struct file_operations write_fifo_fops;
1628 extern const struct file_operations rdwr_fifo_fops;
1630 extern int fs_may_remount_ro(struct super_block *);
1634 * return READ, READA, or WRITE
1636 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
1639 * return data direction, READ or WRITE
1641 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
1643 extern int check_disk_change(struct block_device *);
1644 extern int __invalidate_device(struct block_device *);
1645 extern int invalidate_partition(struct gendisk *, int);
1647 extern int invalidate_inodes(struct super_block *);
1648 unsigned long __invalidate_mapping_pages(struct address_space *mapping,
1649 pgoff_t start, pgoff_t end,
1651 unsigned long invalidate_mapping_pages(struct address_space *mapping,
1652 pgoff_t start, pgoff_t end);
1654 static inline unsigned long __deprecated
1655 invalidate_inode_pages(struct address_space *mapping)
1657 return invalidate_mapping_pages(mapping, 0, ~0UL);
1660 static inline void invalidate_remote_inode(struct inode *inode)
1662 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
1663 S_ISLNK(inode->i_mode))
1664 invalidate_mapping_pages(inode->i_mapping, 0, -1);
1666 extern int invalidate_inode_pages2(struct address_space *mapping);
1667 extern int invalidate_inode_pages2_range(struct address_space *mapping,
1668 pgoff_t start, pgoff_t end);
1669 extern int write_inode_now(struct inode *, int);
1670 extern int filemap_fdatawrite(struct address_space *);
1671 extern int filemap_flush(struct address_space *);
1672 extern int filemap_fdatawait(struct address_space *);
1673 extern int filemap_write_and_wait(struct address_space *mapping);
1674 extern int filemap_write_and_wait_range(struct address_space *mapping,
1675 loff_t lstart, loff_t lend);
1676 extern int wait_on_page_writeback_range(struct address_space *mapping,
1677 pgoff_t start, pgoff_t end);
1678 extern int __filemap_fdatawrite_range(struct address_space *mapping,
1679 loff_t start, loff_t end, int sync_mode);
1681 extern long do_fsync(struct file *file, int datasync);
1682 extern void sync_supers(void);
1683 extern void sync_filesystems(int wait);
1684 extern void __fsync_super(struct super_block *sb);
1685 extern void emergency_sync(void);
1686 extern void emergency_remount(void);
1687 extern int do_remount_sb(struct super_block *sb, int flags,
1688 void *data, int force);
1690 extern sector_t bmap(struct inode *, sector_t);
1692 extern int notify_change(struct dentry *, struct iattr *);
1693 extern int permission(struct inode *, int, struct nameidata *);
1694 extern int generic_permission(struct inode *, int,
1695 int (*check_acl)(struct inode *, int));
1697 extern int get_write_access(struct inode *);
1698 extern int deny_write_access(struct file *);
1699 static inline void put_write_access(struct inode * inode)
1701 atomic_dec(&inode->i_writecount);
1703 static inline void allow_write_access(struct file *file)
1706 atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
1708 extern int do_pipe(int *);
1709 extern struct file *create_read_pipe(struct file *f);
1710 extern struct file *create_write_pipe(void);
1711 extern void free_write_pipe(struct file *);
1713 extern int open_namei(int dfd, const char *, int, int, struct nameidata *);
1714 extern int may_open(struct nameidata *, int, int);
1716 extern int kernel_read(struct file *, unsigned long, char *, unsigned long);
1717 extern struct file * open_exec(const char *);
1719 /* fs/dcache.c -- generic fs support functions */
1720 extern int is_subdir(struct dentry *, struct dentry *);
1721 extern ino_t find_inode_number(struct dentry *, struct qstr *);
1723 #include <linux/err.h>
1725 /* needed for stackable file system support */
1726 extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
1728 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
1730 extern void inode_init_once(struct inode *);
1731 extern void iput(struct inode *);
1732 extern struct inode * igrab(struct inode *);
1733 extern ino_t iunique(struct super_block *, ino_t);
1734 extern int inode_needs_sync(struct inode *inode);
1735 extern void generic_delete_inode(struct inode *inode);
1736 extern void generic_drop_inode(struct inode *inode);
1738 extern struct inode *ilookup5_nowait(struct super_block *sb,
1739 unsigned long hashval, int (*test)(struct inode *, void *),
1741 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
1742 int (*test)(struct inode *, void *), void *data);
1743 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
1745 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
1746 extern struct inode * iget_locked(struct super_block *, unsigned long);
1747 extern void unlock_new_inode(struct inode *);
1749 static inline struct inode *iget(struct super_block *sb, unsigned long ino)
1751 struct inode *inode = iget_locked(sb, ino);
1753 if (inode && (inode->i_state & I_NEW)) {
1754 sb->s_op->read_inode(inode);
1755 unlock_new_inode(inode);
1761 extern void __iget(struct inode * inode);
1762 extern void clear_inode(struct inode *);
1763 extern void destroy_inode(struct inode *);
1764 extern struct inode *new_inode(struct super_block *);
1765 extern int __remove_suid(struct dentry *, int);
1766 extern int should_remove_suid(struct dentry *);
1767 extern int remove_suid(struct dentry *);
1769 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
1770 extern void remove_inode_hash(struct inode *);
1771 static inline void insert_inode_hash(struct inode *inode) {
1772 __insert_inode_hash(inode, inode->i_ino);
1775 extern struct file * get_empty_filp(void);
1776 extern void file_move(struct file *f, struct list_head *list);
1777 extern void file_kill(struct file *f);
1780 extern void submit_bio(int, struct bio *);
1781 extern int bdev_read_only(struct block_device *);
1783 extern int set_blocksize(struct block_device *, int);
1784 extern int sb_set_blocksize(struct super_block *, int);
1785 extern int sb_min_blocksize(struct super_block *, int);
1786 extern int sb_has_dirty_inodes(struct super_block *);
1788 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
1789 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
1790 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1791 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
1792 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
1793 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
1794 extern ssize_t generic_file_aio_write_nolock(struct kiocb *, const struct iovec *,
1795 unsigned long, loff_t);
1796 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
1797 unsigned long *, loff_t, loff_t *, size_t, size_t);
1798 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
1799 unsigned long, loff_t, loff_t *, size_t, ssize_t);
1800 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
1801 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
1802 extern void do_generic_mapping_read(struct address_space *mapping,
1803 struct file_ra_state *, struct file *,
1804 loff_t *, read_descriptor_t *, read_actor_t);
1805 extern int generic_segment_checks(const struct iovec *iov,
1806 unsigned long *nr_segs, size_t *count, int access_flags);
1809 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
1810 struct pipe_inode_info *, size_t, unsigned int);
1811 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
1812 struct file *, loff_t *, size_t, unsigned int);
1813 extern ssize_t generic_file_splice_write_nolock(struct pipe_inode_info *,
1814 struct file *, loff_t *, size_t, unsigned int);
1815 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
1816 struct file *out, loff_t *, size_t len, unsigned int flags);
1817 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
1818 size_t len, unsigned int flags);
1821 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
1822 extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
1823 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
1824 extern loff_t remote_llseek(struct file *file, loff_t offset, int origin);
1825 extern int generic_file_open(struct inode * inode, struct file * filp);
1826 extern int nonseekable_open(struct inode * inode, struct file * filp);
1828 #ifdef CONFIG_FS_XIP
1829 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
1831 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
1832 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
1833 size_t len, loff_t *ppos);
1834 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
1836 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
1842 static inline void do_generic_file_read(struct file * filp, loff_t *ppos,
1843 read_descriptor_t * desc,
1846 do_generic_mapping_read(filp->f_mapping,
1855 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
1856 struct block_device *bdev, const struct iovec *iov, loff_t offset,
1857 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
1861 DIO_LOCKING = 1, /* need locking between buffered and direct access */
1862 DIO_NO_LOCKING, /* bdev; no locking at all between buffered/direct */
1863 DIO_OWN_LOCKING, /* filesystem locks buffered and direct internally */
1866 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
1867 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1868 loff_t offset, unsigned long nr_segs, get_block_t get_block,
1869 dio_iodone_t end_io)
1871 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1872 nr_segs, get_block, end_io, DIO_LOCKING);
1875 static inline ssize_t blockdev_direct_IO_no_locking(int rw, struct kiocb *iocb,
1876 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1877 loff_t offset, unsigned long nr_segs, get_block_t get_block,
1878 dio_iodone_t end_io)
1880 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1881 nr_segs, get_block, end_io, DIO_NO_LOCKING);
1884 static inline ssize_t blockdev_direct_IO_own_locking(int rw, struct kiocb *iocb,
1885 struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1886 loff_t offset, unsigned long nr_segs, get_block_t get_block,
1887 dio_iodone_t end_io)
1889 return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1890 nr_segs, get_block, end_io, DIO_OWN_LOCKING);
1894 extern const struct file_operations generic_ro_fops;
1896 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
1898 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
1899 extern int vfs_follow_link(struct nameidata *, const char *);
1900 extern int page_readlink(struct dentry *, char __user *, int);
1901 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
1902 extern void page_put_link(struct dentry *, struct nameidata *, void *);
1903 extern int __page_symlink(struct inode *inode, const char *symname, int len,
1905 extern int page_symlink(struct inode *inode, const char *symname, int len);
1906 extern const struct inode_operations page_symlink_inode_operations;
1907 extern int generic_readlink(struct dentry *, char __user *, int);
1908 extern void generic_fillattr(struct inode *, struct kstat *);
1909 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1910 void inode_add_bytes(struct inode *inode, loff_t bytes);
1911 void inode_sub_bytes(struct inode *inode, loff_t bytes);
1912 loff_t inode_get_bytes(struct inode *inode);
1913 void inode_set_bytes(struct inode *inode, loff_t bytes);
1915 extern int vfs_readdir(struct file *, filldir_t, void *);
1917 extern int vfs_stat(char __user *, struct kstat *);
1918 extern int vfs_lstat(char __user *, struct kstat *);
1919 extern int vfs_stat_fd(int dfd, char __user *, struct kstat *);
1920 extern int vfs_lstat_fd(int dfd, char __user *, struct kstat *);
1921 extern int vfs_fstat(unsigned int, struct kstat *);
1923 extern int vfs_ioctl(struct file *, unsigned int, unsigned int, unsigned long);
1925 extern struct file_system_type *get_fs_type(const char *name);
1926 extern struct super_block *get_super(struct block_device *);
1927 extern struct super_block *user_get_super(dev_t);
1928 extern void drop_super(struct super_block *sb);
1930 extern int dcache_dir_open(struct inode *, struct file *);
1931 extern int dcache_dir_close(struct inode *, struct file *);
1932 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
1933 extern int dcache_readdir(struct file *, void *, filldir_t);
1934 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1935 extern int simple_statfs(struct dentry *, struct kstatfs *);
1936 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
1937 extern int simple_unlink(struct inode *, struct dentry *);
1938 extern int simple_rmdir(struct inode *, struct dentry *);
1939 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1940 extern int simple_sync_file(struct file *, struct dentry *, int);
1941 extern int simple_empty(struct dentry *);
1942 extern int simple_readpage(struct file *file, struct page *page);
1943 extern int simple_prepare_write(struct file *file, struct page *page,
1944 unsigned offset, unsigned to);
1945 extern int simple_write_begin(struct file *file, struct address_space *mapping,
1946 loff_t pos, unsigned len, unsigned flags,
1947 struct page **pagep, void **fsdata);
1948 extern int simple_write_end(struct file *file, struct address_space *mapping,
1949 loff_t pos, unsigned len, unsigned copied,
1950 struct page *page, void *fsdata);
1952 extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
1953 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
1954 extern const struct file_operations simple_dir_operations;
1955 extern const struct inode_operations simple_dir_inode_operations;
1956 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
1957 struct dentry *d_alloc_name(struct dentry *, const char *);
1958 extern int simple_fill_super(struct super_block *, int, struct tree_descr *);
1959 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
1960 extern void simple_release_fs(struct vfsmount **mount, int *count);
1962 extern ssize_t simple_read_from_buffer(void __user *, size_t, loff_t *, const void *, size_t);
1964 #ifdef CONFIG_MIGRATION
1965 extern int buffer_migrate_page(struct address_space *,
1966 struct page *, struct page *);
1968 #define buffer_migrate_page NULL
1971 extern int inode_change_ok(struct inode *, struct iattr *);
1972 extern int __must_check inode_setattr(struct inode *, struct iattr *);
1974 extern void file_update_time(struct file *file);
1976 static inline ino_t parent_ino(struct dentry *dentry)
1980 spin_lock(&dentry->d_lock);
1981 res = dentry->d_parent->d_inode->i_ino;
1982 spin_unlock(&dentry->d_lock);
1987 extern int unshare_files(void);
1989 /* Transaction based IO helpers */
1992 * An argresp is stored in an allocated page and holds the
1993 * size of the argument or response, along with its content
1995 struct simple_transaction_argresp {
2000 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2002 char *simple_transaction_get(struct file *file, const char __user *buf,
2004 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2005 size_t size, loff_t *pos);
2006 int simple_transaction_release(struct inode *inode, struct file *file);
2008 static inline void simple_transaction_set(struct file *file, size_t n)
2010 struct simple_transaction_argresp *ar = file->private_data;
2012 BUG_ON(n > SIMPLE_TRANSACTION_LIMIT);
2015 * The barrier ensures that ar->size will really remain zero until
2016 * ar->data is ready for reading.
2023 * simple attribute files
2025 * These attributes behave similar to those in sysfs:
2027 * Writing to an attribute immediately sets a value, an open file can be
2028 * written to multiple times.
2030 * Reading from an attribute creates a buffer from the value that might get
2031 * read with multiple read calls. When the attribute has been read
2032 * completely, no further read calls are possible until the file is opened
2035 * All attributes contain a text representation of a numeric value
2036 * that are accessed with the get() and set() functions.
2038 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2039 static int __fops ## _open(struct inode *inode, struct file *file) \
2041 __simple_attr_check_format(__fmt, 0ull); \
2042 return simple_attr_open(inode, file, __get, __set, __fmt); \
2044 static struct file_operations __fops = { \
2045 .owner = THIS_MODULE, \
2046 .open = __fops ## _open, \
2047 .release = simple_attr_close, \
2048 .read = simple_attr_read, \
2049 .write = simple_attr_write, \
2052 static inline void __attribute__((format(printf, 1, 2)))
2053 __simple_attr_check_format(const char *fmt, ...)
2055 /* don't do anything, just let the compiler check the arguments; */
2058 int simple_attr_open(struct inode *inode, struct file *file,
2059 u64 (*get)(void *), void (*set)(void *, u64),
2061 int simple_attr_close(struct inode *inode, struct file *file);
2062 ssize_t simple_attr_read(struct file *file, char __user *buf,
2063 size_t len, loff_t *ppos);
2064 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2065 size_t len, loff_t *ppos);
2068 #ifdef CONFIG_SECURITY
2069 static inline char *alloc_secdata(void)
2071 return (char *)get_zeroed_page(GFP_KERNEL);
2074 static inline void free_secdata(void *secdata)
2076 free_page((unsigned long)secdata);
2079 static inline char *alloc_secdata(void)
2084 static inline void free_secdata(void *secdata)
2086 #endif /* CONFIG_SECURITY */
2089 int proc_nr_files(struct ctl_table *table, int write, struct file *filp,
2090 void __user *buffer, size_t *lenp, loff_t *ppos);
2093 #endif /* __KERNEL__ */
2094 #endif /* _LINUX_FS_H */