xref: /linux-6.15/include/linux/fs.h (revision 42fda663)
1 #ifndef _LINUX_FS_H
2 #define _LINUX_FS_H
3 
4 /*
5  * This file has definitions for some important file table
6  * structures etc.
7  */
8 
9 #include <linux/limits.h>
10 #include <linux/ioctl.h>
11 
12 /*
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.
17  *
18  * Some programs (notably those using select()) may have to be
19  * recompiled to take full advantage of the new limits..
20  */
21 
22 /* Fixed constants first: */
23 #undef NR_OPEN
24 #define NR_OPEN (1024*1024)	/* Absolute upper limit on fd num */
25 #define INR_OPEN 1024		/* Initial setting for nfile rlimits */
26 
27 #define BLOCK_SIZE_BITS 10
28 #define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
29 
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
34 
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 */
40 };
41 extern struct files_stat_struct files_stat;
42 extern int get_max_files(void);
43 
44 struct inodes_stat_t {
45 	int nr_inodes;
46 	int nr_unused;
47 	int dummy[5];		/* padding for sysctl ABI compatibility */
48 };
49 extern struct inodes_stat_t inodes_stat;
50 
51 extern int leases_enable, lease_break_time;
52 
53 #ifdef CONFIG_DNOTIFY
54 extern int dir_notify_enable;
55 #endif
56 
57 #define NR_FILE  8192	/* this can well be larger on a larger system */
58 
59 #define MAY_EXEC 1
60 #define MAY_WRITE 2
61 #define MAY_READ 4
62 #define MAY_APPEND 8
63 
64 #define FMODE_READ 1
65 #define FMODE_WRITE 2
66 
67 /* Internal kernel extensions */
68 #define FMODE_LSEEK	4
69 #define FMODE_PREAD	8
70 #define FMODE_PWRITE	FMODE_PREAD	/* These go hand in hand */
71 
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 */
75 #define FMODE_EXEC	16
76 
77 #define RW_MASK		1
78 #define RWA_MASK	2
79 #define READ 0
80 #define WRITE 1
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))
87 
88 #define SEL_IN		1
89 #define SEL_OUT		2
90 #define SEL_EX		4
91 
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.
99 					 */
100 
101 /*
102  * These are the fs-independent mount-flags: up to 32 flags are supported
103  */
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 */
114 #define MS_BIND		4096
115 #define MS_MOVE		8192
116 #define MS_REC		16384
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)
128 
129 /*
130  * Superblock flags that can be altered by MS_REMOUNT
131  */
132 #define MS_RMT_MASK	(MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK)
133 
134 /*
135  * Old magic mount flag and mask
136  */
137 #define MS_MGC_VAL 0xC0ED0000
138 #define MS_MGC_MSK 0xffff0000
139 
140 /* Inode flags - they have nothing to superblock flags now */
141 
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 */
152 
153 /*
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.
158  *
159  * Exception: MS_RDONLY is always applied to the entire file system.
160  *
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. -- [email protected]
165  */
166 #define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
167 
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)
175 
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)
180 
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)
185 
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. */
188 
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 */
201 #if 0
202 #define BLKPG      _IO(0x12,105)/* See blkpg.h */
203 
204 /* Some people are morons.  Do not use sizeof! */
205 
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. */
210 #endif
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)
219 
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 */
223 
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)
232 
233 /*
234  * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS)
235  */
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 */
260 
261 #define FS_FL_USER_VISIBLE		0x0003DFFF /* User visible flags */
262 #define FS_FL_USER_MODIFIABLE		0x000380FF /* User modifiable flags */
263 
264 
265 #define SYNC_FILE_RANGE_WAIT_BEFORE	1
266 #define SYNC_FILE_RANGE_WRITE		2
267 #define SYNC_FILE_RANGE_WAIT_AFTER	4
268 
269 #ifdef __KERNEL__
270 
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/sysctl.h>
287 #include <linux/capability.h>
288 
289 #include <asm/atomic.h>
290 #include <asm/semaphore.h>
291 #include <asm/byteorder.h>
292 
293 struct export_operations;
294 struct hd_geometry;
295 struct iovec;
296 struct nameidata;
297 struct kiocb;
298 struct pipe_inode_info;
299 struct poll_table_struct;
300 struct kstatfs;
301 struct vm_area_struct;
302 struct vfsmount;
303 
304 extern void __init inode_init(unsigned long);
305 extern void __init inode_init_early(void);
306 extern void __init mnt_init(unsigned long);
307 extern void __init files_init(unsigned long);
308 
309 struct buffer_head;
310 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
311 			struct buffer_head *bh_result, int create);
312 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
313 			ssize_t bytes, void *private);
314 
315 /*
316  * Attribute flags.  These should be or-ed together to figure out what
317  * has been changed!
318  */
319 #define ATTR_MODE	1
320 #define ATTR_UID	2
321 #define ATTR_GID	4
322 #define ATTR_SIZE	8
323 #define ATTR_ATIME	16
324 #define ATTR_MTIME	32
325 #define ATTR_CTIME	64
326 #define ATTR_ATIME_SET	128
327 #define ATTR_MTIME_SET	256
328 #define ATTR_FORCE	512	/* Not a change, but a change it */
329 #define ATTR_ATTR_FLAG	1024
330 #define ATTR_KILL_SUID	2048
331 #define ATTR_KILL_SGID	4096
332 #define ATTR_FILE	8192
333 
334 /*
335  * This is the Inode Attributes structure, used for notify_change().  It
336  * uses the above definitions as flags, to know which values have changed.
337  * Also, in this manner, a Filesystem can look at only the values it cares
338  * about.  Basically, these are the attributes that the VFS layer can
339  * request to change from the FS layer.
340  *
341  * Derek Atkins <[email protected]> 94-10-20
342  */
343 struct iattr {
344 	unsigned int	ia_valid;
345 	umode_t		ia_mode;
346 	uid_t		ia_uid;
347 	gid_t		ia_gid;
348 	loff_t		ia_size;
349 	struct timespec	ia_atime;
350 	struct timespec	ia_mtime;
351 	struct timespec	ia_ctime;
352 
353 	/*
354 	 * Not an attribute, but an auxilary info for filesystems wanting to
355 	 * implement an ftruncate() like method.  NOTE: filesystem should
356 	 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
357 	 */
358 	struct file	*ia_file;
359 };
360 
361 /*
362  * Includes for diskquotas.
363  */
364 #include <linux/quota.h>
365 
366 /**
367  * enum positive_aop_returns - aop return codes with specific semantics
368  *
369  * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
370  * 			    completed, that the page is still locked, and
371  * 			    should be considered active.  The VM uses this hint
372  * 			    to return the page to the active list -- it won't
373  * 			    be a candidate for writeback again in the near
374  * 			    future.  Other callers must be careful to unlock
375  * 			    the page if they get this return.  Returned by
376  * 			    writepage();
377  *
378  * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
379  *  			unlocked it and the page might have been truncated.
380  *  			The caller should back up to acquiring a new page and
381  *  			trying again.  The aop will be taking reasonable
382  *  			precautions not to livelock.  If the caller held a page
383  *  			reference, it should drop it before retrying.  Returned
384  *  			by readpage().
385  *
386  * address_space_operation functions return these large constants to indicate
387  * special semantics to the caller.  These are much larger than the bytes in a
388  * page to allow for functions that return the number of bytes operated on in a
389  * given page.
390  */
391 
392 enum positive_aop_returns {
393 	AOP_WRITEPAGE_ACTIVATE	= 0x80000,
394 	AOP_TRUNCATED_PAGE	= 0x80001,
395 };
396 
397 #define AOP_FLAG_UNINTERRUPTIBLE	0x0001 /* will not do a short write */
398 #define AOP_FLAG_CONT_EXPAND		0x0002 /* called from cont_expand */
399 
400 /*
401  * oh the beauties of C type declarations.
402  */
403 struct page;
404 struct address_space;
405 struct writeback_control;
406 
407 struct iov_iter {
408 	const struct iovec *iov;
409 	unsigned long nr_segs;
410 	size_t iov_offset;
411 	size_t count;
412 };
413 
414 size_t iov_iter_copy_from_user_atomic(struct page *page,
415 		struct iov_iter *i, unsigned long offset, size_t bytes);
416 size_t iov_iter_copy_from_user(struct page *page,
417 		struct iov_iter *i, unsigned long offset, size_t bytes);
418 void iov_iter_advance(struct iov_iter *i, size_t bytes);
419 int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
420 size_t iov_iter_single_seg_count(struct iov_iter *i);
421 
422 static inline void iov_iter_init(struct iov_iter *i,
423 			const struct iovec *iov, unsigned long nr_segs,
424 			size_t count, size_t written)
425 {
426 	i->iov = iov;
427 	i->nr_segs = nr_segs;
428 	i->iov_offset = 0;
429 	i->count = count + written;
430 
431 	iov_iter_advance(i, written);
432 }
433 
434 static inline size_t iov_iter_count(struct iov_iter *i)
435 {
436 	return i->count;
437 }
438 
439 
440 struct address_space_operations {
441 	int (*writepage)(struct page *page, struct writeback_control *wbc);
442 	int (*readpage)(struct file *, struct page *);
443 	void (*sync_page)(struct page *);
444 
445 	/* Write back some dirty pages from this mapping. */
446 	int (*writepages)(struct address_space *, struct writeback_control *);
447 
448 	/* Set a page dirty.  Return true if this dirtied it */
449 	int (*set_page_dirty)(struct page *page);
450 
451 	int (*readpages)(struct file *filp, struct address_space *mapping,
452 			struct list_head *pages, unsigned nr_pages);
453 
454 	/*
455 	 * ext3 requires that a successful prepare_write() call be followed
456 	 * by a commit_write() call - they must be balanced
457 	 */
458 	int (*prepare_write)(struct file *, struct page *, unsigned, unsigned);
459 	int (*commit_write)(struct file *, struct page *, unsigned, unsigned);
460 
461 	int (*write_begin)(struct file *, struct address_space *mapping,
462 				loff_t pos, unsigned len, unsigned flags,
463 				struct page **pagep, void **fsdata);
464 	int (*write_end)(struct file *, struct address_space *mapping,
465 				loff_t pos, unsigned len, unsigned copied,
466 				struct page *page, void *fsdata);
467 
468 	/* Unfortunately this kludge is needed for FIBMAP. Don't use it */
469 	sector_t (*bmap)(struct address_space *, sector_t);
470 	void (*invalidatepage) (struct page *, unsigned long);
471 	int (*releasepage) (struct page *, gfp_t);
472 	ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
473 			loff_t offset, unsigned long nr_segs);
474 	struct page* (*get_xip_page)(struct address_space *, sector_t,
475 			int);
476 	/* migrate the contents of a page to the specified target */
477 	int (*migratepage) (struct address_space *,
478 			struct page *, struct page *);
479 	int (*launder_page) (struct page *);
480 };
481 
482 /*
483  * pagecache_write_begin/pagecache_write_end must be used by general code
484  * to write into the pagecache.
485  */
486 int pagecache_write_begin(struct file *, struct address_space *mapping,
487 				loff_t pos, unsigned len, unsigned flags,
488 				struct page **pagep, void **fsdata);
489 
490 int pagecache_write_end(struct file *, struct address_space *mapping,
491 				loff_t pos, unsigned len, unsigned copied,
492 				struct page *page, void *fsdata);
493 
494 struct backing_dev_info;
495 struct address_space {
496 	struct inode		*host;		/* owner: inode, block_device */
497 	struct radix_tree_root	page_tree;	/* radix tree of all pages */
498 	rwlock_t		tree_lock;	/* and rwlock protecting it */
499 	unsigned int		i_mmap_writable;/* count VM_SHARED mappings */
500 	struct prio_tree_root	i_mmap;		/* tree of private and shared mappings */
501 	struct list_head	i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
502 	spinlock_t		i_mmap_lock;	/* protect tree, count, list */
503 	unsigned int		truncate_count;	/* Cover race condition with truncate */
504 	unsigned long		nrpages;	/* number of total pages */
505 	pgoff_t			writeback_index;/* writeback starts here */
506 	const struct address_space_operations *a_ops;	/* methods */
507 	unsigned long		flags;		/* error bits/gfp mask */
508 	struct backing_dev_info *backing_dev_info; /* device readahead, etc */
509 	spinlock_t		private_lock;	/* for use by the address_space */
510 	struct list_head	private_list;	/* ditto */
511 	struct address_space	*assoc_mapping;	/* ditto */
512 } __attribute__((aligned(sizeof(long))));
513 	/*
514 	 * On most architectures that alignment is already the case; but
515 	 * must be enforced here for CRIS, to let the least signficant bit
516 	 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
517 	 */
518 
519 struct block_device {
520 	dev_t			bd_dev;  /* not a kdev_t - it's a search key */
521 	struct inode *		bd_inode;	/* will die */
522 	int			bd_openers;
523 	struct mutex		bd_mutex;	/* open/close mutex */
524 	struct semaphore	bd_mount_sem;
525 	struct list_head	bd_inodes;
526 	void *			bd_holder;
527 	int			bd_holders;
528 #ifdef CONFIG_SYSFS
529 	struct list_head	bd_holder_list;
530 #endif
531 	struct block_device *	bd_contains;
532 	unsigned		bd_block_size;
533 	struct hd_struct *	bd_part;
534 	/* number of times partitions within this device have been opened. */
535 	unsigned		bd_part_count;
536 	int			bd_invalidated;
537 	struct gendisk *	bd_disk;
538 	struct list_head	bd_list;
539 	struct backing_dev_info *bd_inode_backing_dev_info;
540 	/*
541 	 * Private data.  You must have bd_claim'ed the block_device
542 	 * to use this.  NOTE:  bd_claim allows an owner to claim
543 	 * the same device multiple times, the owner must take special
544 	 * care to not mess up bd_private for that case.
545 	 */
546 	unsigned long		bd_private;
547 };
548 
549 /*
550  * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
551  * radix trees
552  */
553 #define PAGECACHE_TAG_DIRTY	0
554 #define PAGECACHE_TAG_WRITEBACK	1
555 
556 int mapping_tagged(struct address_space *mapping, int tag);
557 
558 /*
559  * Might pages of this file be mapped into userspace?
560  */
561 static inline int mapping_mapped(struct address_space *mapping)
562 {
563 	return	!prio_tree_empty(&mapping->i_mmap) ||
564 		!list_empty(&mapping->i_mmap_nonlinear);
565 }
566 
567 /*
568  * Might pages of this file have been modified in userspace?
569  * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
570  * marks vma as VM_SHARED if it is shared, and the file was opened for
571  * writing i.e. vma may be mprotected writable even if now readonly.
572  */
573 static inline int mapping_writably_mapped(struct address_space *mapping)
574 {
575 	return mapping->i_mmap_writable != 0;
576 }
577 
578 /*
579  * Use sequence counter to get consistent i_size on 32-bit processors.
580  */
581 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
582 #include <linux/seqlock.h>
583 #define __NEED_I_SIZE_ORDERED
584 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
585 #else
586 #define i_size_ordered_init(inode) do { } while (0)
587 #endif
588 
589 struct inode {
590 	struct hlist_node	i_hash;
591 	struct list_head	i_list;
592 	struct list_head	i_sb_list;
593 	struct list_head	i_dentry;
594 	unsigned long		i_ino;
595 	atomic_t		i_count;
596 	unsigned int		i_nlink;
597 	uid_t			i_uid;
598 	gid_t			i_gid;
599 	dev_t			i_rdev;
600 	unsigned long		i_version;
601 	loff_t			i_size;
602 #ifdef __NEED_I_SIZE_ORDERED
603 	seqcount_t		i_size_seqcount;
604 #endif
605 	struct timespec		i_atime;
606 	struct timespec		i_mtime;
607 	struct timespec		i_ctime;
608 	unsigned int		i_blkbits;
609 	blkcnt_t		i_blocks;
610 	unsigned short          i_bytes;
611 	umode_t			i_mode;
612 	spinlock_t		i_lock;	/* i_blocks, i_bytes, maybe i_size */
613 	struct mutex		i_mutex;
614 	struct rw_semaphore	i_alloc_sem;
615 	const struct inode_operations	*i_op;
616 	const struct file_operations	*i_fop;	/* former ->i_op->default_file_ops */
617 	struct super_block	*i_sb;
618 	struct file_lock	*i_flock;
619 	struct address_space	*i_mapping;
620 	struct address_space	i_data;
621 #ifdef CONFIG_QUOTA
622 	struct dquot		*i_dquot[MAXQUOTAS];
623 #endif
624 	struct list_head	i_devices;
625 	union {
626 		struct pipe_inode_info	*i_pipe;
627 		struct block_device	*i_bdev;
628 		struct cdev		*i_cdev;
629 	};
630 	int			i_cindex;
631 
632 	__u32			i_generation;
633 
634 #ifdef CONFIG_DNOTIFY
635 	unsigned long		i_dnotify_mask; /* Directory notify events */
636 	struct dnotify_struct	*i_dnotify; /* for directory notifications */
637 #endif
638 
639 #ifdef CONFIG_INOTIFY
640 	struct list_head	inotify_watches; /* watches on this inode */
641 	struct mutex		inotify_mutex;	/* protects the watches list */
642 #endif
643 
644 	unsigned long		i_state;
645 	unsigned long		dirtied_when;	/* jiffies of first dirtying */
646 
647 	unsigned int		i_flags;
648 
649 	atomic_t		i_writecount;
650 #ifdef CONFIG_SECURITY
651 	void			*i_security;
652 #endif
653 	void			*i_private; /* fs or device private pointer */
654 };
655 
656 /*
657  * inode->i_mutex nesting subclasses for the lock validator:
658  *
659  * 0: the object of the current VFS operation
660  * 1: parent
661  * 2: child/target
662  * 3: quota file
663  *
664  * The locking order between these classes is
665  * parent -> child -> normal -> xattr -> quota
666  */
667 enum inode_i_mutex_lock_class
668 {
669 	I_MUTEX_NORMAL,
670 	I_MUTEX_PARENT,
671 	I_MUTEX_CHILD,
672 	I_MUTEX_XATTR,
673 	I_MUTEX_QUOTA
674 };
675 
676 extern void inode_double_lock(struct inode *inode1, struct inode *inode2);
677 extern void inode_double_unlock(struct inode *inode1, struct inode *inode2);
678 
679 /*
680  * NOTE: in a 32bit arch with a preemptable kernel and
681  * an UP compile the i_size_read/write must be atomic
682  * with respect to the local cpu (unlike with preempt disabled),
683  * but they don't need to be atomic with respect to other cpus like in
684  * true SMP (so they need either to either locally disable irq around
685  * the read or for example on x86 they can be still implemented as a
686  * cmpxchg8b without the need of the lock prefix). For SMP compiles
687  * and 64bit archs it makes no difference if preempt is enabled or not.
688  */
689 static inline loff_t i_size_read(const struct inode *inode)
690 {
691 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
692 	loff_t i_size;
693 	unsigned int seq;
694 
695 	do {
696 		seq = read_seqcount_begin(&inode->i_size_seqcount);
697 		i_size = inode->i_size;
698 	} while (read_seqcount_retry(&inode->i_size_seqcount, seq));
699 	return i_size;
700 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
701 	loff_t i_size;
702 
703 	preempt_disable();
704 	i_size = inode->i_size;
705 	preempt_enable();
706 	return i_size;
707 #else
708 	return inode->i_size;
709 #endif
710 }
711 
712 /*
713  * NOTE: unlike i_size_read(), i_size_write() does need locking around it
714  * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
715  * can be lost, resulting in subsequent i_size_read() calls spinning forever.
716  */
717 static inline void i_size_write(struct inode *inode, loff_t i_size)
718 {
719 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
720 	write_seqcount_begin(&inode->i_size_seqcount);
721 	inode->i_size = i_size;
722 	write_seqcount_end(&inode->i_size_seqcount);
723 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
724 	preempt_disable();
725 	inode->i_size = i_size;
726 	preempt_enable();
727 #else
728 	inode->i_size = i_size;
729 #endif
730 }
731 
732 static inline unsigned iminor(const struct inode *inode)
733 {
734 	return MINOR(inode->i_rdev);
735 }
736 
737 static inline unsigned imajor(const struct inode *inode)
738 {
739 	return MAJOR(inode->i_rdev);
740 }
741 
742 extern struct block_device *I_BDEV(struct inode *inode);
743 
744 struct fown_struct {
745 	rwlock_t lock;          /* protects pid, uid, euid fields */
746 	struct pid *pid;	/* pid or -pgrp where SIGIO should be sent */
747 	enum pid_type pid_type;	/* Kind of process group SIGIO should be sent to */
748 	uid_t uid, euid;	/* uid/euid of process setting the owner */
749 	int signum;		/* posix.1b rt signal to be delivered on IO */
750 };
751 
752 /*
753  * Track a single file's readahead state
754  */
755 struct file_ra_state {
756 	pgoff_t start;			/* where readahead started */
757 	unsigned int size;		/* # of readahead pages */
758 	unsigned int async_size;	/* do asynchronous readahead when
759 					   there are only # of pages ahead */
760 
761 	unsigned int ra_pages;		/* Maximum readahead window */
762 	int mmap_miss;			/* Cache miss stat for mmap accesses */
763 	loff_t prev_pos;		/* Cache last read() position */
764 };
765 
766 /*
767  * Check if @index falls in the readahead windows.
768  */
769 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
770 {
771 	return (index >= ra->start &&
772 		index <  ra->start + ra->size);
773 }
774 
775 struct file {
776 	/*
777 	 * fu_list becomes invalid after file_free is called and queued via
778 	 * fu_rcuhead for RCU freeing
779 	 */
780 	union {
781 		struct list_head	fu_list;
782 		struct rcu_head 	fu_rcuhead;
783 	} f_u;
784 	struct path		f_path;
785 #define f_dentry	f_path.dentry
786 #define f_vfsmnt	f_path.mnt
787 	const struct file_operations	*f_op;
788 	atomic_t		f_count;
789 	unsigned int 		f_flags;
790 	mode_t			f_mode;
791 	loff_t			f_pos;
792 	struct fown_struct	f_owner;
793 	unsigned int		f_uid, f_gid;
794 	struct file_ra_state	f_ra;
795 
796 	unsigned long		f_version;
797 #ifdef CONFIG_SECURITY
798 	void			*f_security;
799 #endif
800 	/* needed for tty driver, and maybe others */
801 	void			*private_data;
802 
803 #ifdef CONFIG_EPOLL
804 	/* Used by fs/eventpoll.c to link all the hooks to this file */
805 	struct list_head	f_ep_links;
806 	spinlock_t		f_ep_lock;
807 #endif /* #ifdef CONFIG_EPOLL */
808 	struct address_space	*f_mapping;
809 };
810 extern spinlock_t files_lock;
811 #define file_list_lock() spin_lock(&files_lock);
812 #define file_list_unlock() spin_unlock(&files_lock);
813 
814 #define get_file(x)	atomic_inc(&(x)->f_count)
815 #define file_count(x)	atomic_read(&(x)->f_count)
816 
817 #define	MAX_NON_LFS	((1UL<<31) - 1)
818 
819 /* Page cache limit. The filesystems should put that into their s_maxbytes
820    limits, otherwise bad things can happen in VM. */
821 #if BITS_PER_LONG==32
822 #define MAX_LFS_FILESIZE	(((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
823 #elif BITS_PER_LONG==64
824 #define MAX_LFS_FILESIZE 	0x7fffffffffffffffUL
825 #endif
826 
827 #define FL_POSIX	1
828 #define FL_FLOCK	2
829 #define FL_ACCESS	8	/* not trying to lock, just looking */
830 #define FL_EXISTS	16	/* when unlocking, test for existence */
831 #define FL_LEASE	32	/* lease held on this file */
832 #define FL_CLOSE	64	/* unlock on close */
833 #define FL_SLEEP	128	/* A blocking lock */
834 
835 /*
836  * The POSIX file lock owner is determined by
837  * the "struct files_struct" in the thread group
838  * (or NULL for no owner - BSD locks).
839  *
840  * Lockd stuffs a "host" pointer into this.
841  */
842 typedef struct files_struct *fl_owner_t;
843 
844 struct file_lock_operations {
845 	void (*fl_insert)(struct file_lock *);	/* lock insertion callback */
846 	void (*fl_remove)(struct file_lock *);	/* lock removal callback */
847 	void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
848 	void (*fl_release_private)(struct file_lock *);
849 };
850 
851 struct lock_manager_operations {
852 	int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
853 	void (*fl_notify)(struct file_lock *);	/* unblock callback */
854 	int (*fl_grant)(struct file_lock *, struct file_lock *, int);
855 	void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
856 	void (*fl_release_private)(struct file_lock *);
857 	void (*fl_break)(struct file_lock *);
858 	int (*fl_mylease)(struct file_lock *, struct file_lock *);
859 	int (*fl_change)(struct file_lock **, int);
860 };
861 
862 /* that will die - we need it for nfs_lock_info */
863 #include <linux/nfs_fs_i.h>
864 
865 struct file_lock {
866 	struct file_lock *fl_next;	/* singly linked list for this inode  */
867 	struct list_head fl_link;	/* doubly linked list of all locks */
868 	struct list_head fl_block;	/* circular list of blocked processes */
869 	fl_owner_t fl_owner;
870 	unsigned int fl_pid;
871 	wait_queue_head_t fl_wait;
872 	struct file *fl_file;
873 	unsigned char fl_flags;
874 	unsigned char fl_type;
875 	loff_t fl_start;
876 	loff_t fl_end;
877 
878 	struct fasync_struct *	fl_fasync; /* for lease break notifications */
879 	unsigned long fl_break_time;	/* for nonblocking lease breaks */
880 
881 	struct file_lock_operations *fl_ops;	/* Callbacks for filesystems */
882 	struct lock_manager_operations *fl_lmops;	/* Callbacks for lockmanagers */
883 	union {
884 		struct nfs_lock_info	nfs_fl;
885 		struct nfs4_lock_info	nfs4_fl;
886 		struct {
887 			struct list_head link;	/* link in AFS vnode's pending_locks list */
888 			int state;		/* state of grant or error if -ve */
889 		} afs;
890 	} fl_u;
891 };
892 
893 /* The following constant reflects the upper bound of the file/locking space */
894 #ifndef OFFSET_MAX
895 #define INT_LIMIT(x)	(~((x)1 << (sizeof(x)*8 - 1)))
896 #define OFFSET_MAX	INT_LIMIT(loff_t)
897 #define OFFT_OFFSET_MAX	INT_LIMIT(off_t)
898 #endif
899 
900 #include <linux/fcntl.h>
901 
902 extern int fcntl_getlk(struct file *, struct flock __user *);
903 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
904 			struct flock __user *);
905 
906 #if BITS_PER_LONG == 32
907 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
908 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
909 			struct flock64 __user *);
910 #endif
911 
912 extern void send_sigio(struct fown_struct *fown, int fd, int band);
913 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
914 extern int fcntl_getlease(struct file *filp);
915 
916 /* fs/sync.c */
917 extern int do_sync_mapping_range(struct address_space *mapping, loff_t offset,
918 			loff_t endbyte, unsigned int flags);
919 
920 /* fs/locks.c */
921 extern void locks_init_lock(struct file_lock *);
922 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
923 extern void locks_remove_posix(struct file *, fl_owner_t);
924 extern void locks_remove_flock(struct file *);
925 extern void posix_test_lock(struct file *, struct file_lock *);
926 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
927 extern int posix_lock_file_wait(struct file *, struct file_lock *);
928 extern int posix_unblock_lock(struct file *, struct file_lock *);
929 extern int vfs_test_lock(struct file *, struct file_lock *);
930 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
931 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
932 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
933 extern int __break_lease(struct inode *inode, unsigned int flags);
934 extern void lease_get_mtime(struct inode *, struct timespec *time);
935 extern int generic_setlease(struct file *, long, struct file_lock **);
936 extern int vfs_setlease(struct file *, long, struct file_lock **);
937 extern int lease_modify(struct file_lock **, int);
938 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
939 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
940 extern struct seq_operations locks_seq_operations;
941 
942 struct fasync_struct {
943 	int	magic;
944 	int	fa_fd;
945 	struct	fasync_struct	*fa_next; /* singly linked list */
946 	struct	file 		*fa_file;
947 };
948 
949 #define FASYNC_MAGIC 0x4601
950 
951 /* SMP safe fasync helpers: */
952 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
953 /* can be called from interrupts */
954 extern void kill_fasync(struct fasync_struct **, int, int);
955 /* only for net: no internal synchronization */
956 extern void __kill_fasync(struct fasync_struct *, int, int);
957 
958 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
959 extern int f_setown(struct file *filp, unsigned long arg, int force);
960 extern void f_delown(struct file *filp);
961 extern pid_t f_getown(struct file *filp);
962 extern int send_sigurg(struct fown_struct *fown);
963 
964 /*
965  *	Umount options
966  */
967 
968 #define MNT_FORCE	0x00000001	/* Attempt to forcibily umount */
969 #define MNT_DETACH	0x00000002	/* Just detach from the tree */
970 #define MNT_EXPIRE	0x00000004	/* Mark for expiry */
971 
972 extern struct list_head super_blocks;
973 extern spinlock_t sb_lock;
974 
975 #define sb_entry(list)	list_entry((list), struct super_block, s_list)
976 #define S_BIAS (1<<30)
977 struct super_block {
978 	struct list_head	s_list;		/* Keep this first */
979 	dev_t			s_dev;		/* search index; _not_ kdev_t */
980 	unsigned long		s_blocksize;
981 	unsigned char		s_blocksize_bits;
982 	unsigned char		s_dirt;
983 	unsigned long long	s_maxbytes;	/* Max file size */
984 	struct file_system_type	*s_type;
985 	const struct super_operations	*s_op;
986 	struct dquot_operations	*dq_op;
987  	struct quotactl_ops	*s_qcop;
988 	struct export_operations *s_export_op;
989 	unsigned long		s_flags;
990 	unsigned long		s_magic;
991 	struct dentry		*s_root;
992 	struct rw_semaphore	s_umount;
993 	struct mutex		s_lock;
994 	int			s_count;
995 	int			s_syncing;
996 	int			s_need_sync_fs;
997 	atomic_t		s_active;
998 #ifdef CONFIG_SECURITY
999 	void                    *s_security;
1000 #endif
1001 	struct xattr_handler	**s_xattr;
1002 
1003 	struct list_head	s_inodes;	/* all inodes */
1004 	struct list_head	s_dirty;	/* dirty inodes */
1005 	struct list_head	s_io;		/* parked for writeback */
1006 	struct hlist_head	s_anon;		/* anonymous dentries for (nfs) exporting */
1007 	struct list_head	s_files;
1008 
1009 	struct block_device	*s_bdev;
1010 	struct mtd_info		*s_mtd;
1011 	struct list_head	s_instances;
1012 	struct quota_info	s_dquot;	/* Diskquota specific options */
1013 
1014 	int			s_frozen;
1015 	wait_queue_head_t	s_wait_unfrozen;
1016 
1017 	char s_id[32];				/* Informational name */
1018 
1019 	void 			*s_fs_info;	/* Filesystem private info */
1020 
1021 	/*
1022 	 * The next field is for VFS *only*. No filesystems have any business
1023 	 * even looking at it. You had been warned.
1024 	 */
1025 	struct mutex s_vfs_rename_mutex;	/* Kludge */
1026 
1027 	/* Granularity of c/m/atime in ns.
1028 	   Cannot be worse than a second */
1029 	u32		   s_time_gran;
1030 
1031 	/*
1032 	 * Filesystem subtype.  If non-empty the filesystem type field
1033 	 * in /proc/mounts will be "type.subtype"
1034 	 */
1035 	char *s_subtype;
1036 };
1037 
1038 extern struct timespec current_fs_time(struct super_block *sb);
1039 
1040 /*
1041  * Snapshotting support.
1042  */
1043 enum {
1044 	SB_UNFROZEN = 0,
1045 	SB_FREEZE_WRITE	= 1,
1046 	SB_FREEZE_TRANS = 2,
1047 };
1048 
1049 #define vfs_check_frozen(sb, level) \
1050 	wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
1051 
1052 #define get_fs_excl() atomic_inc(&current->fs_excl)
1053 #define put_fs_excl() atomic_dec(&current->fs_excl)
1054 #define has_fs_excl() atomic_read(&current->fs_excl)
1055 
1056 #define is_owner_or_cap(inode)	\
1057 	((current->fsuid == (inode)->i_uid) || capable(CAP_FOWNER))
1058 
1059 /* not quite ready to be deprecated, but... */
1060 extern void lock_super(struct super_block *);
1061 extern void unlock_super(struct super_block *);
1062 
1063 /*
1064  * VFS helper functions..
1065  */
1066 extern int vfs_permission(struct nameidata *, int);
1067 extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
1068 extern int vfs_mkdir(struct inode *, struct dentry *, int);
1069 extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
1070 extern int vfs_symlink(struct inode *, struct dentry *, const char *, int);
1071 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1072 extern int vfs_rmdir(struct inode *, struct dentry *);
1073 extern int vfs_unlink(struct inode *, struct dentry *);
1074 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1075 
1076 /*
1077  * VFS dentry helper functions.
1078  */
1079 extern void dentry_unhash(struct dentry *dentry);
1080 
1081 /*
1082  * VFS file helper functions.
1083  */
1084 extern int file_permission(struct file *, int);
1085 
1086 /*
1087  * File types
1088  *
1089  * NOTE! These match bits 12..15 of stat.st_mode
1090  * (ie "(i_mode >> 12) & 15").
1091  */
1092 #define DT_UNKNOWN	0
1093 #define DT_FIFO		1
1094 #define DT_CHR		2
1095 #define DT_DIR		4
1096 #define DT_BLK		6
1097 #define DT_REG		8
1098 #define DT_LNK		10
1099 #define DT_SOCK		12
1100 #define DT_WHT		14
1101 
1102 #define OSYNC_METADATA	(1<<0)
1103 #define OSYNC_DATA	(1<<1)
1104 #define OSYNC_INODE	(1<<2)
1105 int generic_osync_inode(struct inode *, struct address_space *, int);
1106 
1107 /*
1108  * This is the "filldir" function type, used by readdir() to let
1109  * the kernel specify what kind of dirent layout it wants to have.
1110  * This allows the kernel to read directories into kernel space or
1111  * to have different dirent layouts depending on the binary type.
1112  */
1113 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1114 
1115 struct block_device_operations {
1116 	int (*open) (struct inode *, struct file *);
1117 	int (*release) (struct inode *, struct file *);
1118 	int (*ioctl) (struct inode *, struct file *, unsigned, unsigned long);
1119 	long (*unlocked_ioctl) (struct file *, unsigned, unsigned long);
1120 	long (*compat_ioctl) (struct file *, unsigned, unsigned long);
1121 	int (*direct_access) (struct block_device *, sector_t, unsigned long *);
1122 	int (*media_changed) (struct gendisk *);
1123 	int (*revalidate_disk) (struct gendisk *);
1124 	int (*getgeo)(struct block_device *, struct hd_geometry *);
1125 	struct module *owner;
1126 };
1127 
1128 /*
1129  * "descriptor" for what we're up to with a read.
1130  * This allows us to use the same read code yet
1131  * have multiple different users of the data that
1132  * we read from a file.
1133  *
1134  * The simplest case just copies the data to user
1135  * mode.
1136  */
1137 typedef struct {
1138 	size_t written;
1139 	size_t count;
1140 	union {
1141 		char __user * buf;
1142 		void *data;
1143 	} arg;
1144 	int error;
1145 } read_descriptor_t;
1146 
1147 typedef int (*read_actor_t)(read_descriptor_t *, struct page *, unsigned long, unsigned long);
1148 
1149 /* These macros are for out of kernel modules to test that
1150  * the kernel supports the unlocked_ioctl and compat_ioctl
1151  * fields in struct file_operations. */
1152 #define HAVE_COMPAT_IOCTL 1
1153 #define HAVE_UNLOCKED_IOCTL 1
1154 
1155 /*
1156  * NOTE:
1157  * read, write, poll, fsync, readv, writev, unlocked_ioctl and compat_ioctl
1158  * can be called without the big kernel lock held in all filesystems.
1159  */
1160 struct file_operations {
1161 	struct module *owner;
1162 	loff_t (*llseek) (struct file *, loff_t, int);
1163 	ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1164 	ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1165 	ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1166 	ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1167 	int (*readdir) (struct file *, void *, filldir_t);
1168 	unsigned int (*poll) (struct file *, struct poll_table_struct *);
1169 	int (*ioctl) (struct inode *, struct file *, unsigned int, unsigned long);
1170 	long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1171 	long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1172 	int (*mmap) (struct file *, struct vm_area_struct *);
1173 	int (*open) (struct inode *, struct file *);
1174 	int (*flush) (struct file *, fl_owner_t id);
1175 	int (*release) (struct inode *, struct file *);
1176 	int (*fsync) (struct file *, struct dentry *, int datasync);
1177 	int (*aio_fsync) (struct kiocb *, int datasync);
1178 	int (*fasync) (int, struct file *, int);
1179 	int (*lock) (struct file *, int, struct file_lock *);
1180 	ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1181 	unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1182 	int (*check_flags)(int);
1183 	int (*dir_notify)(struct file *filp, unsigned long arg);
1184 	int (*flock) (struct file *, int, struct file_lock *);
1185 	ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1186 	ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1187 	int (*setlease)(struct file *, long, struct file_lock **);
1188 };
1189 
1190 struct inode_operations {
1191 	int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1192 	struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1193 	int (*link) (struct dentry *,struct inode *,struct dentry *);
1194 	int (*unlink) (struct inode *,struct dentry *);
1195 	int (*symlink) (struct inode *,struct dentry *,const char *);
1196 	int (*mkdir) (struct inode *,struct dentry *,int);
1197 	int (*rmdir) (struct inode *,struct dentry *);
1198 	int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1199 	int (*rename) (struct inode *, struct dentry *,
1200 			struct inode *, struct dentry *);
1201 	int (*readlink) (struct dentry *, char __user *,int);
1202 	void * (*follow_link) (struct dentry *, struct nameidata *);
1203 	void (*put_link) (struct dentry *, struct nameidata *, void *);
1204 	void (*truncate) (struct inode *);
1205 	int (*permission) (struct inode *, int, struct nameidata *);
1206 	int (*setattr) (struct dentry *, struct iattr *);
1207 	int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1208 	int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1209 	ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1210 	ssize_t (*listxattr) (struct dentry *, char *, size_t);
1211 	int (*removexattr) (struct dentry *, const char *);
1212 	void (*truncate_range)(struct inode *, loff_t, loff_t);
1213 	long (*fallocate)(struct inode *inode, int mode, loff_t offset,
1214 			  loff_t len);
1215 };
1216 
1217 struct seq_file;
1218 
1219 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1220 				unsigned long nr_segs, unsigned long fast_segs,
1221 				struct iovec *fast_pointer,
1222 				struct iovec **ret_pointer);
1223 
1224 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1225 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1226 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1227 		unsigned long, loff_t *);
1228 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1229 		unsigned long, loff_t *);
1230 
1231 /*
1232  * NOTE: write_inode, delete_inode, clear_inode, put_inode can be called
1233  * without the big kernel lock held in all filesystems.
1234  */
1235 struct super_operations {
1236    	struct inode *(*alloc_inode)(struct super_block *sb);
1237 	void (*destroy_inode)(struct inode *);
1238 
1239 	void (*read_inode) (struct inode *);
1240 
1241    	void (*dirty_inode) (struct inode *);
1242 	int (*write_inode) (struct inode *, int);
1243 	void (*put_inode) (struct inode *);
1244 	void (*drop_inode) (struct inode *);
1245 	void (*delete_inode) (struct inode *);
1246 	void (*put_super) (struct super_block *);
1247 	void (*write_super) (struct super_block *);
1248 	int (*sync_fs)(struct super_block *sb, int wait);
1249 	void (*write_super_lockfs) (struct super_block *);
1250 	void (*unlockfs) (struct super_block *);
1251 	int (*statfs) (struct dentry *, struct kstatfs *);
1252 	int (*remount_fs) (struct super_block *, int *, char *);
1253 	void (*clear_inode) (struct inode *);
1254 	void (*umount_begin) (struct vfsmount *, int);
1255 
1256 	int (*show_options)(struct seq_file *, struct vfsmount *);
1257 	int (*show_stats)(struct seq_file *, struct vfsmount *);
1258 #ifdef CONFIG_QUOTA
1259 	ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1260 	ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1261 #endif
1262 };
1263 
1264 /* Inode state bits.  Protected by inode_lock. */
1265 #define I_DIRTY_SYNC		1 /* Not dirty enough for O_DATASYNC */
1266 #define I_DIRTY_DATASYNC	2 /* Data-related inode changes pending */
1267 #define I_DIRTY_PAGES		4 /* Data-related inode changes pending */
1268 #define __I_LOCK		3
1269 #define I_LOCK			(1 << __I_LOCK)
1270 #define I_FREEING		16
1271 #define I_CLEAR			32
1272 #define I_NEW			64
1273 #define I_WILL_FREE		128
1274 
1275 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1276 
1277 extern void __mark_inode_dirty(struct inode *, int);
1278 static inline void mark_inode_dirty(struct inode *inode)
1279 {
1280 	__mark_inode_dirty(inode, I_DIRTY);
1281 }
1282 
1283 static inline void mark_inode_dirty_sync(struct inode *inode)
1284 {
1285 	__mark_inode_dirty(inode, I_DIRTY_SYNC);
1286 }
1287 
1288 /**
1289  * inc_nlink - directly increment an inode's link count
1290  * @inode: inode
1291  *
1292  * This is a low-level filesystem helper to replace any
1293  * direct filesystem manipulation of i_nlink.  Currently,
1294  * it is only here for parity with dec_nlink().
1295  */
1296 static inline void inc_nlink(struct inode *inode)
1297 {
1298 	inode->i_nlink++;
1299 }
1300 
1301 static inline void inode_inc_link_count(struct inode *inode)
1302 {
1303 	inc_nlink(inode);
1304 	mark_inode_dirty(inode);
1305 }
1306 
1307 /**
1308  * drop_nlink - directly drop an inode's link count
1309  * @inode: inode
1310  *
1311  * This is a low-level filesystem helper to replace any
1312  * direct filesystem manipulation of i_nlink.  In cases
1313  * where we are attempting to track writes to the
1314  * filesystem, a decrement to zero means an imminent
1315  * write when the file is truncated and actually unlinked
1316  * on the filesystem.
1317  */
1318 static inline void drop_nlink(struct inode *inode)
1319 {
1320 	inode->i_nlink--;
1321 }
1322 
1323 /**
1324  * clear_nlink - directly zero an inode's link count
1325  * @inode: inode
1326  *
1327  * This is a low-level filesystem helper to replace any
1328  * direct filesystem manipulation of i_nlink.  See
1329  * drop_nlink() for why we care about i_nlink hitting zero.
1330  */
1331 static inline void clear_nlink(struct inode *inode)
1332 {
1333 	inode->i_nlink = 0;
1334 }
1335 
1336 static inline void inode_dec_link_count(struct inode *inode)
1337 {
1338 	drop_nlink(inode);
1339 	mark_inode_dirty(inode);
1340 }
1341 
1342 extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
1343 static inline void file_accessed(struct file *file)
1344 {
1345 	if (!(file->f_flags & O_NOATIME))
1346 		touch_atime(file->f_path.mnt, file->f_path.dentry);
1347 }
1348 
1349 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1350 
1351 struct file_system_type {
1352 	const char *name;
1353 	int fs_flags;
1354 	int (*get_sb) (struct file_system_type *, int,
1355 		       const char *, void *, struct vfsmount *);
1356 	void (*kill_sb) (struct super_block *);
1357 	struct module *owner;
1358 	struct file_system_type * next;
1359 	struct list_head fs_supers;
1360 
1361 	struct lock_class_key s_lock_key;
1362 	struct lock_class_key s_umount_key;
1363 
1364 	struct lock_class_key i_lock_key;
1365 	struct lock_class_key i_mutex_key;
1366 	struct lock_class_key i_mutex_dir_key;
1367 	struct lock_class_key i_alloc_sem_key;
1368 };
1369 
1370 extern int get_sb_bdev(struct file_system_type *fs_type,
1371 	int flags, const char *dev_name, void *data,
1372 	int (*fill_super)(struct super_block *, void *, int),
1373 	struct vfsmount *mnt);
1374 extern int get_sb_single(struct file_system_type *fs_type,
1375 	int flags, void *data,
1376 	int (*fill_super)(struct super_block *, void *, int),
1377 	struct vfsmount *mnt);
1378 extern int get_sb_nodev(struct file_system_type *fs_type,
1379 	int flags, void *data,
1380 	int (*fill_super)(struct super_block *, void *, int),
1381 	struct vfsmount *mnt);
1382 void generic_shutdown_super(struct super_block *sb);
1383 void kill_block_super(struct super_block *sb);
1384 void kill_anon_super(struct super_block *sb);
1385 void kill_litter_super(struct super_block *sb);
1386 void deactivate_super(struct super_block *sb);
1387 int set_anon_super(struct super_block *s, void *data);
1388 struct super_block *sget(struct file_system_type *type,
1389 			int (*test)(struct super_block *,void *),
1390 			int (*set)(struct super_block *,void *),
1391 			void *data);
1392 extern int get_sb_pseudo(struct file_system_type *, char *,
1393 	const struct super_operations *ops, unsigned long,
1394 	struct vfsmount *mnt);
1395 extern int simple_set_mnt(struct vfsmount *mnt, struct super_block *sb);
1396 int __put_super(struct super_block *sb);
1397 int __put_super_and_need_restart(struct super_block *sb);
1398 void unnamed_dev_init(void);
1399 
1400 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1401 #define fops_get(fops) \
1402 	(((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1403 #define fops_put(fops) \
1404 	do { if (fops) module_put((fops)->owner); } while(0)
1405 
1406 extern int register_filesystem(struct file_system_type *);
1407 extern int unregister_filesystem(struct file_system_type *);
1408 extern struct vfsmount *kern_mount(struct file_system_type *);
1409 extern int may_umount_tree(struct vfsmount *);
1410 extern int may_umount(struct vfsmount *);
1411 extern void umount_tree(struct vfsmount *, int, struct list_head *);
1412 extern void release_mounts(struct list_head *);
1413 extern long do_mount(char *, char *, char *, unsigned long, void *);
1414 extern struct vfsmount *copy_tree(struct vfsmount *, struct dentry *, int);
1415 extern void mnt_set_mountpoint(struct vfsmount *, struct dentry *,
1416 				  struct vfsmount *);
1417 
1418 extern int vfs_statfs(struct dentry *, struct kstatfs *);
1419 
1420 /* /sys/fs */
1421 extern struct kset fs_subsys;
1422 
1423 #define FLOCK_VERIFY_READ  1
1424 #define FLOCK_VERIFY_WRITE 2
1425 
1426 extern int locks_mandatory_locked(struct inode *);
1427 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1428 
1429 /*
1430  * Candidates for mandatory locking have the setgid bit set
1431  * but no group execute bit -  an otherwise meaningless combination.
1432  */
1433 
1434 static inline int __mandatory_lock(struct inode *ino)
1435 {
1436 	return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1437 }
1438 
1439 /*
1440  * ... and these candidates should be on MS_MANDLOCK mounted fs,
1441  * otherwise these will be advisory locks
1442  */
1443 
1444 static inline int mandatory_lock(struct inode *ino)
1445 {
1446 	return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1447 }
1448 
1449 static inline int locks_verify_locked(struct inode *inode)
1450 {
1451 	if (mandatory_lock(inode))
1452 		return locks_mandatory_locked(inode);
1453 	return 0;
1454 }
1455 
1456 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1457 
1458 static inline int locks_verify_truncate(struct inode *inode,
1459 				    struct file *filp,
1460 				    loff_t size)
1461 {
1462 	if (inode->i_flock && mandatory_lock(inode))
1463 		return locks_mandatory_area(
1464 			FLOCK_VERIFY_WRITE, inode, filp,
1465 			size < inode->i_size ? size : inode->i_size,
1466 			(size < inode->i_size ? inode->i_size - size
1467 			 : size - inode->i_size)
1468 		);
1469 	return 0;
1470 }
1471 
1472 static inline int break_lease(struct inode *inode, unsigned int mode)
1473 {
1474 	if (inode->i_flock)
1475 		return __break_lease(inode, mode);
1476 	return 0;
1477 }
1478 
1479 /* fs/open.c */
1480 
1481 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
1482 		       struct file *filp);
1483 extern long do_sys_open(int fdf, const char __user *filename, int flags,
1484 			int mode);
1485 extern struct file *filp_open(const char *, int, int);
1486 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int);
1487 extern int filp_close(struct file *, fl_owner_t id);
1488 extern char * getname(const char __user *);
1489 
1490 /* fs/dcache.c */
1491 extern void __init vfs_caches_init_early(void);
1492 extern void __init vfs_caches_init(unsigned long);
1493 
1494 extern struct kmem_cache *names_cachep;
1495 
1496 #define __getname()	kmem_cache_alloc(names_cachep, GFP_KERNEL)
1497 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
1498 #ifndef CONFIG_AUDITSYSCALL
1499 #define putname(name)   __putname(name)
1500 #else
1501 extern void putname(const char *name);
1502 #endif
1503 
1504 #ifdef CONFIG_BLOCK
1505 extern int register_blkdev(unsigned int, const char *);
1506 extern void unregister_blkdev(unsigned int, const char *);
1507 extern struct block_device *bdget(dev_t);
1508 extern void bd_set_size(struct block_device *, loff_t size);
1509 extern void bd_forget(struct inode *inode);
1510 extern void bdput(struct block_device *);
1511 extern struct block_device *open_by_devnum(dev_t, unsigned);
1512 extern const struct address_space_operations def_blk_aops;
1513 #else
1514 static inline void bd_forget(struct inode *inode) {}
1515 #endif
1516 extern const struct file_operations def_blk_fops;
1517 extern const struct file_operations def_chr_fops;
1518 extern const struct file_operations bad_sock_fops;
1519 extern const struct file_operations def_fifo_fops;
1520 #ifdef CONFIG_BLOCK
1521 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
1522 extern int blkdev_ioctl(struct inode *, struct file *, unsigned, unsigned long);
1523 extern int blkdev_driver_ioctl(struct inode *inode, struct file *file,
1524 			       struct gendisk *disk, unsigned cmd,
1525 			       unsigned long arg);
1526 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
1527 extern int blkdev_get(struct block_device *, mode_t, unsigned);
1528 extern int blkdev_put(struct block_device *);
1529 extern int bd_claim(struct block_device *, void *);
1530 extern void bd_release(struct block_device *);
1531 #ifdef CONFIG_SYSFS
1532 extern int bd_claim_by_disk(struct block_device *, void *, struct gendisk *);
1533 extern void bd_release_from_disk(struct block_device *, struct gendisk *);
1534 #else
1535 #define bd_claim_by_disk(bdev, holder, disk)	bd_claim(bdev, holder)
1536 #define bd_release_from_disk(bdev, disk)	bd_release(bdev)
1537 #endif
1538 #endif
1539 
1540 /* fs/char_dev.c */
1541 #define CHRDEV_MAJOR_HASH_SIZE	255
1542 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
1543 extern int register_chrdev_region(dev_t, unsigned, const char *);
1544 extern int register_chrdev(unsigned int, const char *,
1545 			   const struct file_operations *);
1546 extern void unregister_chrdev(unsigned int, const char *);
1547 extern void unregister_chrdev_region(dev_t, unsigned);
1548 extern int chrdev_open(struct inode *, struct file *);
1549 extern void chrdev_show(struct seq_file *,off_t);
1550 
1551 /* fs/block_dev.c */
1552 #define BDEVNAME_SIZE	32	/* Largest string for a blockdev identifier */
1553 
1554 #ifdef CONFIG_BLOCK
1555 #define BLKDEV_MAJOR_HASH_SIZE	255
1556 extern const char *__bdevname(dev_t, char *buffer);
1557 extern const char *bdevname(struct block_device *bdev, char *buffer);
1558 extern struct block_device *lookup_bdev(const char *);
1559 extern struct block_device *open_bdev_excl(const char *, int, void *);
1560 extern void close_bdev_excl(struct block_device *);
1561 extern void blkdev_show(struct seq_file *,off_t);
1562 #else
1563 #define BLKDEV_MAJOR_HASH_SIZE	0
1564 #endif
1565 
1566 extern void init_special_inode(struct inode *, umode_t, dev_t);
1567 
1568 /* Invalid inode operations -- fs/bad_inode.c */
1569 extern void make_bad_inode(struct inode *);
1570 extern int is_bad_inode(struct inode *);
1571 
1572 extern const struct file_operations read_fifo_fops;
1573 extern const struct file_operations write_fifo_fops;
1574 extern const struct file_operations rdwr_fifo_fops;
1575 
1576 extern int fs_may_remount_ro(struct super_block *);
1577 
1578 #ifdef CONFIG_BLOCK
1579 /*
1580  * return READ, READA, or WRITE
1581  */
1582 #define bio_rw(bio)		((bio)->bi_rw & (RW_MASK | RWA_MASK))
1583 
1584 /*
1585  * return data direction, READ or WRITE
1586  */
1587 #define bio_data_dir(bio)	((bio)->bi_rw & 1)
1588 
1589 extern int check_disk_change(struct block_device *);
1590 extern int __invalidate_device(struct block_device *);
1591 extern int invalidate_partition(struct gendisk *, int);
1592 #endif
1593 extern int invalidate_inodes(struct super_block *);
1594 unsigned long __invalidate_mapping_pages(struct address_space *mapping,
1595 					pgoff_t start, pgoff_t end,
1596 					bool be_atomic);
1597 unsigned long invalidate_mapping_pages(struct address_space *mapping,
1598 					pgoff_t start, pgoff_t end);
1599 
1600 static inline unsigned long __deprecated
1601 invalidate_inode_pages(struct address_space *mapping)
1602 {
1603 	return invalidate_mapping_pages(mapping, 0, ~0UL);
1604 }
1605 
1606 static inline void invalidate_remote_inode(struct inode *inode)
1607 {
1608 	if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
1609 	    S_ISLNK(inode->i_mode))
1610 		invalidate_mapping_pages(inode->i_mapping, 0, -1);
1611 }
1612 extern int invalidate_inode_pages2(struct address_space *mapping);
1613 extern int invalidate_inode_pages2_range(struct address_space *mapping,
1614 					 pgoff_t start, pgoff_t end);
1615 extern int write_inode_now(struct inode *, int);
1616 extern int filemap_fdatawrite(struct address_space *);
1617 extern int filemap_flush(struct address_space *);
1618 extern int filemap_fdatawait(struct address_space *);
1619 extern int filemap_write_and_wait(struct address_space *mapping);
1620 extern int filemap_write_and_wait_range(struct address_space *mapping,
1621 				        loff_t lstart, loff_t lend);
1622 extern int wait_on_page_writeback_range(struct address_space *mapping,
1623 				pgoff_t start, pgoff_t end);
1624 extern int __filemap_fdatawrite_range(struct address_space *mapping,
1625 				loff_t start, loff_t end, int sync_mode);
1626 
1627 extern long do_fsync(struct file *file, int datasync);
1628 extern void sync_supers(void);
1629 extern void sync_filesystems(int wait);
1630 extern void __fsync_super(struct super_block *sb);
1631 extern void emergency_sync(void);
1632 extern void emergency_remount(void);
1633 extern int do_remount_sb(struct super_block *sb, int flags,
1634 			 void *data, int force);
1635 #ifdef CONFIG_BLOCK
1636 extern sector_t bmap(struct inode *, sector_t);
1637 #endif
1638 extern int notify_change(struct dentry *, struct iattr *);
1639 extern int permission(struct inode *, int, struct nameidata *);
1640 extern int generic_permission(struct inode *, int,
1641 		int (*check_acl)(struct inode *, int));
1642 
1643 extern int get_write_access(struct inode *);
1644 extern int deny_write_access(struct file *);
1645 static inline void put_write_access(struct inode * inode)
1646 {
1647 	atomic_dec(&inode->i_writecount);
1648 }
1649 static inline void allow_write_access(struct file *file)
1650 {
1651 	if (file)
1652 		atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
1653 }
1654 extern int do_pipe(int *);
1655 extern struct file *create_read_pipe(struct file *f);
1656 extern struct file *create_write_pipe(void);
1657 extern void free_write_pipe(struct file *);
1658 
1659 extern int open_namei(int dfd, const char *, int, int, struct nameidata *);
1660 extern int may_open(struct nameidata *, int, int);
1661 
1662 extern int kernel_read(struct file *, unsigned long, char *, unsigned long);
1663 extern struct file * open_exec(const char *);
1664 
1665 /* fs/dcache.c -- generic fs support functions */
1666 extern int is_subdir(struct dentry *, struct dentry *);
1667 extern ino_t find_inode_number(struct dentry *, struct qstr *);
1668 
1669 #include <linux/err.h>
1670 
1671 /* needed for stackable file system support */
1672 extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
1673 
1674 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
1675 
1676 extern void inode_init_once(struct inode *);
1677 extern void iput(struct inode *);
1678 extern struct inode * igrab(struct inode *);
1679 extern ino_t iunique(struct super_block *, ino_t);
1680 extern int inode_needs_sync(struct inode *inode);
1681 extern void generic_delete_inode(struct inode *inode);
1682 extern void generic_drop_inode(struct inode *inode);
1683 
1684 extern struct inode *ilookup5_nowait(struct super_block *sb,
1685 		unsigned long hashval, int (*test)(struct inode *, void *),
1686 		void *data);
1687 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
1688 		int (*test)(struct inode *, void *), void *data);
1689 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
1690 
1691 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
1692 extern struct inode * iget_locked(struct super_block *, unsigned long);
1693 extern void unlock_new_inode(struct inode *);
1694 
1695 static inline struct inode *iget(struct super_block *sb, unsigned long ino)
1696 {
1697 	struct inode *inode = iget_locked(sb, ino);
1698 
1699 	if (inode && (inode->i_state & I_NEW)) {
1700 		sb->s_op->read_inode(inode);
1701 		unlock_new_inode(inode);
1702 	}
1703 
1704 	return inode;
1705 }
1706 
1707 extern void __iget(struct inode * inode);
1708 extern void clear_inode(struct inode *);
1709 extern void destroy_inode(struct inode *);
1710 extern struct inode *new_inode(struct super_block *);
1711 extern int __remove_suid(struct dentry *, int);
1712 extern int should_remove_suid(struct dentry *);
1713 extern int remove_suid(struct dentry *);
1714 
1715 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
1716 extern void remove_inode_hash(struct inode *);
1717 static inline void insert_inode_hash(struct inode *inode) {
1718 	__insert_inode_hash(inode, inode->i_ino);
1719 }
1720 
1721 extern struct file * get_empty_filp(void);
1722 extern void file_move(struct file *f, struct list_head *list);
1723 extern void file_kill(struct file *f);
1724 #ifdef CONFIG_BLOCK
1725 struct bio;
1726 extern void submit_bio(int, struct bio *);
1727 extern int bdev_read_only(struct block_device *);
1728 #endif
1729 extern int set_blocksize(struct block_device *, int);
1730 extern int sb_set_blocksize(struct super_block *, int);
1731 extern int sb_min_blocksize(struct super_block *, int);
1732 
1733 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
1734 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
1735 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1736 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
1737 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
1738 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
1739 extern ssize_t generic_file_aio_write_nolock(struct kiocb *, const struct iovec *,
1740 		unsigned long, loff_t);
1741 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
1742 		unsigned long *, loff_t, loff_t *, size_t, size_t);
1743 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
1744 		unsigned long, loff_t, loff_t *, size_t, ssize_t);
1745 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
1746 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
1747 extern void do_generic_mapping_read(struct address_space *mapping,
1748 				    struct file_ra_state *, struct file *,
1749 				    loff_t *, read_descriptor_t *, read_actor_t);
1750 extern int generic_segment_checks(const struct iovec *iov,
1751 		unsigned long *nr_segs, size_t *count, int access_flags);
1752 
1753 /* fs/splice.c */
1754 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
1755 		struct pipe_inode_info *, size_t, unsigned int);
1756 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
1757 		struct file *, loff_t *, size_t, unsigned int);
1758 extern ssize_t generic_file_splice_write_nolock(struct pipe_inode_info *,
1759 		struct file *, loff_t *, size_t, unsigned int);
1760 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
1761 		struct file *out, loff_t *, size_t len, unsigned int flags);
1762 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
1763 		size_t len, unsigned int flags);
1764 
1765 extern void
1766 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
1767 extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
1768 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
1769 extern loff_t remote_llseek(struct file *file, loff_t offset, int origin);
1770 extern int generic_file_open(struct inode * inode, struct file * filp);
1771 extern int nonseekable_open(struct inode * inode, struct file * filp);
1772 
1773 #ifdef CONFIG_FS_XIP
1774 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
1775 			     loff_t *ppos);
1776 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
1777 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
1778 			      size_t len, loff_t *ppos);
1779 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
1780 #else
1781 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
1782 {
1783 	return 0;
1784 }
1785 #endif
1786 
1787 static inline void do_generic_file_read(struct file * filp, loff_t *ppos,
1788 					read_descriptor_t * desc,
1789 					read_actor_t actor)
1790 {
1791 	do_generic_mapping_read(filp->f_mapping,
1792 				&filp->f_ra,
1793 				filp,
1794 				ppos,
1795 				desc,
1796 				actor);
1797 }
1798 
1799 #ifdef CONFIG_BLOCK
1800 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
1801 	struct block_device *bdev, const struct iovec *iov, loff_t offset,
1802 	unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
1803 	int lock_type);
1804 
1805 enum {
1806 	DIO_LOCKING = 1, /* need locking between buffered and direct access */
1807 	DIO_NO_LOCKING,  /* bdev; no locking at all between buffered/direct */
1808 	DIO_OWN_LOCKING, /* filesystem locks buffered and direct internally */
1809 };
1810 
1811 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
1812 	struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1813 	loff_t offset, unsigned long nr_segs, get_block_t get_block,
1814 	dio_iodone_t end_io)
1815 {
1816 	return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1817 				nr_segs, get_block, end_io, DIO_LOCKING);
1818 }
1819 
1820 static inline ssize_t blockdev_direct_IO_no_locking(int rw, struct kiocb *iocb,
1821 	struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1822 	loff_t offset, unsigned long nr_segs, get_block_t get_block,
1823 	dio_iodone_t end_io)
1824 {
1825 	return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1826 				nr_segs, get_block, end_io, DIO_NO_LOCKING);
1827 }
1828 
1829 static inline ssize_t blockdev_direct_IO_own_locking(int rw, struct kiocb *iocb,
1830 	struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1831 	loff_t offset, unsigned long nr_segs, get_block_t get_block,
1832 	dio_iodone_t end_io)
1833 {
1834 	return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1835 				nr_segs, get_block, end_io, DIO_OWN_LOCKING);
1836 }
1837 #endif
1838 
1839 extern const struct file_operations generic_ro_fops;
1840 
1841 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
1842 
1843 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
1844 extern int vfs_follow_link(struct nameidata *, const char *);
1845 extern int page_readlink(struct dentry *, char __user *, int);
1846 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
1847 extern void page_put_link(struct dentry *, struct nameidata *, void *);
1848 extern int __page_symlink(struct inode *inode, const char *symname, int len,
1849 		gfp_t gfp_mask);
1850 extern int page_symlink(struct inode *inode, const char *symname, int len);
1851 extern const struct inode_operations page_symlink_inode_operations;
1852 extern int generic_readlink(struct dentry *, char __user *, int);
1853 extern void generic_fillattr(struct inode *, struct kstat *);
1854 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1855 void inode_add_bytes(struct inode *inode, loff_t bytes);
1856 void inode_sub_bytes(struct inode *inode, loff_t bytes);
1857 loff_t inode_get_bytes(struct inode *inode);
1858 void inode_set_bytes(struct inode *inode, loff_t bytes);
1859 
1860 extern int vfs_readdir(struct file *, filldir_t, void *);
1861 
1862 extern int vfs_stat(char __user *, struct kstat *);
1863 extern int vfs_lstat(char __user *, struct kstat *);
1864 extern int vfs_stat_fd(int dfd, char __user *, struct kstat *);
1865 extern int vfs_lstat_fd(int dfd, char __user *, struct kstat *);
1866 extern int vfs_fstat(unsigned int, struct kstat *);
1867 
1868 extern int vfs_ioctl(struct file *, unsigned int, unsigned int, unsigned long);
1869 
1870 extern struct file_system_type *get_fs_type(const char *name);
1871 extern struct super_block *get_super(struct block_device *);
1872 extern struct super_block *user_get_super(dev_t);
1873 extern void drop_super(struct super_block *sb);
1874 
1875 extern int dcache_dir_open(struct inode *, struct file *);
1876 extern int dcache_dir_close(struct inode *, struct file *);
1877 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
1878 extern int dcache_readdir(struct file *, void *, filldir_t);
1879 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1880 extern int simple_statfs(struct dentry *, struct kstatfs *);
1881 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
1882 extern int simple_unlink(struct inode *, struct dentry *);
1883 extern int simple_rmdir(struct inode *, struct dentry *);
1884 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1885 extern int simple_sync_file(struct file *, struct dentry *, int);
1886 extern int simple_empty(struct dentry *);
1887 extern int simple_readpage(struct file *file, struct page *page);
1888 extern int simple_prepare_write(struct file *file, struct page *page,
1889 			unsigned offset, unsigned to);
1890 extern int simple_commit_write(struct file *file, struct page *page,
1891 				unsigned offset, unsigned to);
1892 extern int simple_write_begin(struct file *file, struct address_space *mapping,
1893 			loff_t pos, unsigned len, unsigned flags,
1894 			struct page **pagep, void **fsdata);
1895 extern int simple_write_end(struct file *file, struct address_space *mapping,
1896 			loff_t pos, unsigned len, unsigned copied,
1897 			struct page *page, void *fsdata);
1898 
1899 extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
1900 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
1901 extern const struct file_operations simple_dir_operations;
1902 extern const struct inode_operations simple_dir_inode_operations;
1903 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
1904 struct dentry *d_alloc_name(struct dentry *, const char *);
1905 extern int simple_fill_super(struct super_block *, int, struct tree_descr *);
1906 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
1907 extern void simple_release_fs(struct vfsmount **mount, int *count);
1908 
1909 extern ssize_t simple_read_from_buffer(void __user *, size_t, loff_t *, const void *, size_t);
1910 
1911 #ifdef CONFIG_MIGRATION
1912 extern int buffer_migrate_page(struct address_space *,
1913 				struct page *, struct page *);
1914 #else
1915 #define buffer_migrate_page NULL
1916 #endif
1917 
1918 extern int inode_change_ok(struct inode *, struct iattr *);
1919 extern int __must_check inode_setattr(struct inode *, struct iattr *);
1920 
1921 extern void file_update_time(struct file *file);
1922 
1923 static inline ino_t parent_ino(struct dentry *dentry)
1924 {
1925 	ino_t res;
1926 
1927 	spin_lock(&dentry->d_lock);
1928 	res = dentry->d_parent->d_inode->i_ino;
1929 	spin_unlock(&dentry->d_lock);
1930 	return res;
1931 }
1932 
1933 /* kernel/fork.c */
1934 extern int unshare_files(void);
1935 
1936 /* Transaction based IO helpers */
1937 
1938 /*
1939  * An argresp is stored in an allocated page and holds the
1940  * size of the argument or response, along with its content
1941  */
1942 struct simple_transaction_argresp {
1943 	ssize_t size;
1944 	char data[0];
1945 };
1946 
1947 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
1948 
1949 char *simple_transaction_get(struct file *file, const char __user *buf,
1950 				size_t size);
1951 ssize_t simple_transaction_read(struct file *file, char __user *buf,
1952 				size_t size, loff_t *pos);
1953 int simple_transaction_release(struct inode *inode, struct file *file);
1954 
1955 static inline void simple_transaction_set(struct file *file, size_t n)
1956 {
1957 	struct simple_transaction_argresp *ar = file->private_data;
1958 
1959 	BUG_ON(n > SIMPLE_TRANSACTION_LIMIT);
1960 
1961 	/*
1962 	 * The barrier ensures that ar->size will really remain zero until
1963 	 * ar->data is ready for reading.
1964 	 */
1965 	smp_mb();
1966 	ar->size = n;
1967 }
1968 
1969 /*
1970  * simple attribute files
1971  *
1972  * These attributes behave similar to those in sysfs:
1973  *
1974  * Writing to an attribute immediately sets a value, an open file can be
1975  * written to multiple times.
1976  *
1977  * Reading from an attribute creates a buffer from the value that might get
1978  * read with multiple read calls. When the attribute has been read
1979  * completely, no further read calls are possible until the file is opened
1980  * again.
1981  *
1982  * All attributes contain a text representation of a numeric value
1983  * that are accessed with the get() and set() functions.
1984  */
1985 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt)		\
1986 static int __fops ## _open(struct inode *inode, struct file *file)	\
1987 {									\
1988 	__simple_attr_check_format(__fmt, 0ull);			\
1989 	return simple_attr_open(inode, file, __get, __set, __fmt);	\
1990 }									\
1991 static struct file_operations __fops = {				\
1992 	.owner	 = THIS_MODULE,						\
1993 	.open	 = __fops ## _open,					\
1994 	.release = simple_attr_close,					\
1995 	.read	 = simple_attr_read,					\
1996 	.write	 = simple_attr_write,					\
1997 };
1998 
1999 static inline void __attribute__((format(printf, 1, 2)))
2000 __simple_attr_check_format(const char *fmt, ...)
2001 {
2002 	/* don't do anything, just let the compiler check the arguments; */
2003 }
2004 
2005 int simple_attr_open(struct inode *inode, struct file *file,
2006 		     u64 (*get)(void *), void (*set)(void *, u64),
2007 		     const char *fmt);
2008 int simple_attr_close(struct inode *inode, struct file *file);
2009 ssize_t simple_attr_read(struct file *file, char __user *buf,
2010 			 size_t len, loff_t *ppos);
2011 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2012 			  size_t len, loff_t *ppos);
2013 
2014 
2015 #ifdef CONFIG_SECURITY
2016 static inline char *alloc_secdata(void)
2017 {
2018 	return (char *)get_zeroed_page(GFP_KERNEL);
2019 }
2020 
2021 static inline void free_secdata(void *secdata)
2022 {
2023 	free_page((unsigned long)secdata);
2024 }
2025 #else
2026 static inline char *alloc_secdata(void)
2027 {
2028 	return (char *)1;
2029 }
2030 
2031 static inline void free_secdata(void *secdata)
2032 { }
2033 #endif	/* CONFIG_SECURITY */
2034 
2035 int proc_nr_files(ctl_table *table, int write, struct file *filp,
2036 		  void __user *buffer, size_t *lenp, loff_t *ppos);
2037 
2038 
2039 #endif /* __KERNEL__ */
2040 #endif /* _LINUX_FS_H */
2041