xref: /linux-6.15/include/linux/fs.h (revision 4f3865fb)
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 /* And dynamically-tunable limits and defaults: */
31 struct files_stat_struct {
32 	int nr_files;		/* read only */
33 	int nr_free_files;	/* read only */
34 	int max_files;		/* tunable */
35 };
36 extern struct files_stat_struct files_stat;
37 extern int get_max_files(void);
38 
39 struct inodes_stat_t {
40 	int nr_inodes;
41 	int nr_unused;
42 	int dummy[5];
43 };
44 extern struct inodes_stat_t inodes_stat;
45 
46 extern int leases_enable, lease_break_time;
47 
48 #ifdef CONFIG_DNOTIFY
49 extern int dir_notify_enable;
50 #endif
51 
52 #define NR_FILE  8192	/* this can well be larger on a larger system */
53 
54 #define MAY_EXEC 1
55 #define MAY_WRITE 2
56 #define MAY_READ 4
57 #define MAY_APPEND 8
58 
59 #define FMODE_READ 1
60 #define FMODE_WRITE 2
61 
62 /* Internal kernel extensions */
63 #define FMODE_LSEEK	4
64 #define FMODE_PREAD	8
65 #define FMODE_PWRITE	FMODE_PREAD	/* These go hand in hand */
66 
67 /* File is being opened for execution. Primary users of this flag are
68    distributed filesystems that can use it to achieve correct ETXTBUSY
69    behavior for cross-node execution/opening_for_writing of files */
70 #define FMODE_EXEC	16
71 
72 #define RW_MASK		1
73 #define RWA_MASK	2
74 #define READ 0
75 #define WRITE 1
76 #define READA 2		/* read-ahead  - don't block if no resources */
77 #define SWRITE 3	/* for ll_rw_block() - wait for buffer lock */
78 #define SPECIAL 4	/* For non-blockdevice requests in request queue */
79 #define READ_SYNC	(READ | (1 << BIO_RW_SYNC))
80 #define WRITE_SYNC	(WRITE | (1 << BIO_RW_SYNC))
81 #define WRITE_BARRIER	((1 << BIO_RW) | (1 << BIO_RW_BARRIER))
82 
83 #define SEL_IN		1
84 #define SEL_OUT		2
85 #define SEL_EX		4
86 
87 /* public flags for file_system_type */
88 #define FS_REQUIRES_DEV 1
89 #define FS_BINARY_MOUNTDATA 2
90 #define FS_REVAL_DOT	16384	/* Check the paths ".", ".." for staleness */
91 #define FS_ODD_RENAME	32768	/* Temporary stuff; will go away as soon
92 				  * as nfs_rename() will be cleaned up
93 				  */
94 /*
95  * These are the fs-independent mount-flags: up to 32 flags are supported
96  */
97 #define MS_RDONLY	 1	/* Mount read-only */
98 #define MS_NOSUID	 2	/* Ignore suid and sgid bits */
99 #define MS_NODEV	 4	/* Disallow access to device special files */
100 #define MS_NOEXEC	 8	/* Disallow program execution */
101 #define MS_SYNCHRONOUS	16	/* Writes are synced at once */
102 #define MS_REMOUNT	32	/* Alter flags of a mounted FS */
103 #define MS_MANDLOCK	64	/* Allow mandatory locks on an FS */
104 #define MS_DIRSYNC	128	/* Directory modifications are synchronous */
105 #define MS_NOATIME	1024	/* Do not update access times. */
106 #define MS_NODIRATIME	2048	/* Do not update directory access times */
107 #define MS_BIND		4096
108 #define MS_MOVE		8192
109 #define MS_REC		16384
110 #define MS_VERBOSE	32768	/* War is peace. Verbosity is silence.
111 				   MS_VERBOSE is deprecated. */
112 #define MS_SILENT	32768
113 #define MS_POSIXACL	(1<<16)	/* VFS does not apply the umask */
114 #define MS_UNBINDABLE	(1<<17)	/* change to unbindable */
115 #define MS_PRIVATE	(1<<18)	/* change to private */
116 #define MS_SLAVE	(1<<19)	/* change to slave */
117 #define MS_SHARED	(1<<20)	/* change to shared */
118 #define MS_ACTIVE	(1<<30)
119 #define MS_NOUSER	(1<<31)
120 
121 /*
122  * Superblock flags that can be altered by MS_REMOUNT
123  */
124 #define MS_RMT_MASK	(MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK)
125 
126 /*
127  * Old magic mount flag and mask
128  */
129 #define MS_MGC_VAL 0xC0ED0000
130 #define MS_MGC_MSK 0xffff0000
131 
132 /* Inode flags - they have nothing to superblock flags now */
133 
134 #define S_SYNC		1	/* Writes are synced at once */
135 #define S_NOATIME	2	/* Do not update access times */
136 #define S_APPEND	4	/* Append-only file */
137 #define S_IMMUTABLE	8	/* Immutable file */
138 #define S_DEAD		16	/* removed, but still open directory */
139 #define S_NOQUOTA	32	/* Inode is not counted to quota */
140 #define S_DIRSYNC	64	/* Directory modifications are synchronous */
141 #define S_NOCMTIME	128	/* Do not update file c/mtime */
142 #define S_SWAPFILE	256	/* Do not truncate: swapon got its bmaps */
143 #define S_PRIVATE	512	/* Inode is fs-internal */
144 
145 /*
146  * Note that nosuid etc flags are inode-specific: setting some file-system
147  * flags just means all the inodes inherit those flags by default. It might be
148  * possible to override it selectively if you really wanted to with some
149  * ioctl() that is not currently implemented.
150  *
151  * Exception: MS_RDONLY is always applied to the entire file system.
152  *
153  * Unfortunately, it is possible to change a filesystems flags with it mounted
154  * with files in use.  This means that all of the inodes will not have their
155  * i_flags updated.  Hence, i_flags no longer inherit the superblock mount
156  * flags, so these have to be checked separately. -- [email protected]
157  */
158 #define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
159 
160 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
161 #define IS_SYNC(inode)		(__IS_FLG(inode, MS_SYNCHRONOUS) || \
162 					((inode)->i_flags & S_SYNC))
163 #define IS_DIRSYNC(inode)	(__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
164 					((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
165 #define IS_MANDLOCK(inode)	__IS_FLG(inode, MS_MANDLOCK)
166 
167 #define IS_NOQUOTA(inode)	((inode)->i_flags & S_NOQUOTA)
168 #define IS_APPEND(inode)	((inode)->i_flags & S_APPEND)
169 #define IS_IMMUTABLE(inode)	((inode)->i_flags & S_IMMUTABLE)
170 #define IS_POSIXACL(inode)	__IS_FLG(inode, MS_POSIXACL)
171 
172 #define IS_DEADDIR(inode)	((inode)->i_flags & S_DEAD)
173 #define IS_NOCMTIME(inode)	((inode)->i_flags & S_NOCMTIME)
174 #define IS_SWAPFILE(inode)	((inode)->i_flags & S_SWAPFILE)
175 #define IS_PRIVATE(inode)	((inode)->i_flags & S_PRIVATE)
176 
177 /* the read-only stuff doesn't really belong here, but any other place is
178    probably as bad and I don't want to create yet another include file. */
179 
180 #define BLKROSET   _IO(0x12,93)	/* set device read-only (0 = read-write) */
181 #define BLKROGET   _IO(0x12,94)	/* get read-only status (0 = read_write) */
182 #define BLKRRPART  _IO(0x12,95)	/* re-read partition table */
183 #define BLKGETSIZE _IO(0x12,96)	/* return device size /512 (long *arg) */
184 #define BLKFLSBUF  _IO(0x12,97)	/* flush buffer cache */
185 #define BLKRASET   _IO(0x12,98)	/* set read ahead for block device */
186 #define BLKRAGET   _IO(0x12,99)	/* get current read ahead setting */
187 #define BLKFRASET  _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
188 #define BLKFRAGET  _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
189 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
190 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
191 #define BLKSSZGET  _IO(0x12,104)/* get block device sector size */
192 #if 0
193 #define BLKPG      _IO(0x12,105)/* See blkpg.h */
194 
195 /* Some people are morons.  Do not use sizeof! */
196 
197 #define BLKELVGET  _IOR(0x12,106,size_t)/* elevator get */
198 #define BLKELVSET  _IOW(0x12,107,size_t)/* elevator set */
199 /* This was here just to show that the number is taken -
200    probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
201 #endif
202 /* A jump here: 108-111 have been used for various private purposes. */
203 #define BLKBSZGET  _IOR(0x12,112,size_t)
204 #define BLKBSZSET  _IOW(0x12,113,size_t)
205 #define BLKGETSIZE64 _IOR(0x12,114,size_t)	/* return device size in bytes (u64 *arg) */
206 #define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
207 #define BLKTRACESTART _IO(0x12,116)
208 #define BLKTRACESTOP _IO(0x12,117)
209 #define BLKTRACETEARDOWN _IO(0x12,118)
210 
211 #define BMAP_IOCTL 1		/* obsolete - kept for compatibility */
212 #define FIBMAP	   _IO(0x00,1)	/* bmap access */
213 #define FIGETBSZ   _IO(0x00,2)	/* get the block size used for bmap */
214 
215 #define SYNC_FILE_RANGE_WAIT_BEFORE	1
216 #define SYNC_FILE_RANGE_WRITE		2
217 #define SYNC_FILE_RANGE_WAIT_AFTER	4
218 
219 #ifdef __KERNEL__
220 
221 #include <linux/linkage.h>
222 #include <linux/wait.h>
223 #include <linux/types.h>
224 #include <linux/kdev_t.h>
225 #include <linux/dcache.h>
226 #include <linux/stat.h>
227 #include <linux/cache.h>
228 #include <linux/kobject.h>
229 #include <linux/list.h>
230 #include <linux/radix-tree.h>
231 #include <linux/prio_tree.h>
232 #include <linux/init.h>
233 #include <linux/sched.h>
234 #include <linux/mutex.h>
235 
236 #include <asm/atomic.h>
237 #include <asm/semaphore.h>
238 #include <asm/byteorder.h>
239 
240 struct hd_geometry;
241 struct iovec;
242 struct nameidata;
243 struct kiocb;
244 struct pipe_inode_info;
245 struct poll_table_struct;
246 struct kstatfs;
247 struct vm_area_struct;
248 struct vfsmount;
249 
250 extern void __init inode_init(unsigned long);
251 extern void __init inode_init_early(void);
252 extern void __init mnt_init(unsigned long);
253 extern void __init files_init(unsigned long);
254 
255 struct buffer_head;
256 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
257 			struct buffer_head *bh_result, int create);
258 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
259 			ssize_t bytes, void *private);
260 
261 /*
262  * Attribute flags.  These should be or-ed together to figure out what
263  * has been changed!
264  */
265 #define ATTR_MODE	1
266 #define ATTR_UID	2
267 #define ATTR_GID	4
268 #define ATTR_SIZE	8
269 #define ATTR_ATIME	16
270 #define ATTR_MTIME	32
271 #define ATTR_CTIME	64
272 #define ATTR_ATIME_SET	128
273 #define ATTR_MTIME_SET	256
274 #define ATTR_FORCE	512	/* Not a change, but a change it */
275 #define ATTR_ATTR_FLAG	1024
276 #define ATTR_KILL_SUID	2048
277 #define ATTR_KILL_SGID	4096
278 #define ATTR_FILE	8192
279 
280 /*
281  * This is the Inode Attributes structure, used for notify_change().  It
282  * uses the above definitions as flags, to know which values have changed.
283  * Also, in this manner, a Filesystem can look at only the values it cares
284  * about.  Basically, these are the attributes that the VFS layer can
285  * request to change from the FS layer.
286  *
287  * Derek Atkins <[email protected]> 94-10-20
288  */
289 struct iattr {
290 	unsigned int	ia_valid;
291 	umode_t		ia_mode;
292 	uid_t		ia_uid;
293 	gid_t		ia_gid;
294 	loff_t		ia_size;
295 	struct timespec	ia_atime;
296 	struct timespec	ia_mtime;
297 	struct timespec	ia_ctime;
298 
299 	/*
300 	 * Not an attribute, but an auxilary info for filesystems wanting to
301 	 * implement an ftruncate() like method.  NOTE: filesystem should
302 	 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
303 	 */
304 	struct file	*ia_file;
305 };
306 
307 /*
308  * Includes for diskquotas.
309  */
310 #include <linux/quota.h>
311 
312 /**
313  * enum positive_aop_returns - aop return codes with specific semantics
314  *
315  * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
316  * 			    completed, that the page is still locked, and
317  * 			    should be considered active.  The VM uses this hint
318  * 			    to return the page to the active list -- it won't
319  * 			    be a candidate for writeback again in the near
320  * 			    future.  Other callers must be careful to unlock
321  * 			    the page if they get this return.  Returned by
322  * 			    writepage();
323  *
324  * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
325  *  			unlocked it and the page might have been truncated.
326  *  			The caller should back up to acquiring a new page and
327  *  			trying again.  The aop will be taking reasonable
328  *  			precautions not to livelock.  If the caller held a page
329  *  			reference, it should drop it before retrying.  Returned
330  *  			by readpage(), prepare_write(), and commit_write().
331  *
332  * address_space_operation functions return these large constants to indicate
333  * special semantics to the caller.  These are much larger than the bytes in a
334  * page to allow for functions that return the number of bytes operated on in a
335  * given page.
336  */
337 
338 enum positive_aop_returns {
339 	AOP_WRITEPAGE_ACTIVATE	= 0x80000,
340 	AOP_TRUNCATED_PAGE	= 0x80001,
341 };
342 
343 /*
344  * oh the beauties of C type declarations.
345  */
346 struct page;
347 struct address_space;
348 struct writeback_control;
349 
350 struct address_space_operations {
351 	int (*writepage)(struct page *page, struct writeback_control *wbc);
352 	int (*readpage)(struct file *, struct page *);
353 	void (*sync_page)(struct page *);
354 
355 	/* Write back some dirty pages from this mapping. */
356 	int (*writepages)(struct address_space *, struct writeback_control *);
357 
358 	/* Set a page dirty.  Return true if this dirtied it */
359 	int (*set_page_dirty)(struct page *page);
360 
361 	int (*readpages)(struct file *filp, struct address_space *mapping,
362 			struct list_head *pages, unsigned nr_pages);
363 
364 	/*
365 	 * ext3 requires that a successful prepare_write() call be followed
366 	 * by a commit_write() call - they must be balanced
367 	 */
368 	int (*prepare_write)(struct file *, struct page *, unsigned, unsigned);
369 	int (*commit_write)(struct file *, struct page *, unsigned, unsigned);
370 	/* Unfortunately this kludge is needed for FIBMAP. Don't use it */
371 	sector_t (*bmap)(struct address_space *, sector_t);
372 	void (*invalidatepage) (struct page *, unsigned long);
373 	int (*releasepage) (struct page *, gfp_t);
374 	ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
375 			loff_t offset, unsigned long nr_segs);
376 	struct page* (*get_xip_page)(struct address_space *, sector_t,
377 			int);
378 	/* migrate the contents of a page to the specified target */
379 	int (*migratepage) (struct page *, struct page *);
380 };
381 
382 struct backing_dev_info;
383 struct address_space {
384 	struct inode		*host;		/* owner: inode, block_device */
385 	struct radix_tree_root	page_tree;	/* radix tree of all pages */
386 	rwlock_t		tree_lock;	/* and rwlock protecting it */
387 	unsigned int		i_mmap_writable;/* count VM_SHARED mappings */
388 	struct prio_tree_root	i_mmap;		/* tree of private and shared mappings */
389 	struct list_head	i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
390 	spinlock_t		i_mmap_lock;	/* protect tree, count, list */
391 	unsigned int		truncate_count;	/* Cover race condition with truncate */
392 	unsigned long		nrpages;	/* number of total pages */
393 	pgoff_t			writeback_index;/* writeback starts here */
394 	struct address_space_operations *a_ops;	/* methods */
395 	unsigned long		flags;		/* error bits/gfp mask */
396 	struct backing_dev_info *backing_dev_info; /* device readahead, etc */
397 	spinlock_t		private_lock;	/* for use by the address_space */
398 	struct list_head	private_list;	/* ditto */
399 	struct address_space	*assoc_mapping;	/* ditto */
400 } __attribute__((aligned(sizeof(long))));
401 	/*
402 	 * On most architectures that alignment is already the case; but
403 	 * must be enforced here for CRIS, to let the least signficant bit
404 	 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
405 	 */
406 
407 struct block_device {
408 	dev_t			bd_dev;  /* not a kdev_t - it's a search key */
409 	struct inode *		bd_inode;	/* will die */
410 	int			bd_openers;
411 	struct mutex		bd_mutex;	/* open/close mutex */
412 	struct mutex		bd_mount_mutex;	/* mount mutex */
413 	struct list_head	bd_inodes;
414 	void *			bd_holder;
415 	int			bd_holders;
416 #ifdef CONFIG_SYSFS
417 	struct list_head	bd_holder_list;
418 #endif
419 	struct block_device *	bd_contains;
420 	unsigned		bd_block_size;
421 	struct hd_struct *	bd_part;
422 	/* number of times partitions within this device have been opened. */
423 	unsigned		bd_part_count;
424 	int			bd_invalidated;
425 	struct gendisk *	bd_disk;
426 	struct list_head	bd_list;
427 	struct backing_dev_info *bd_inode_backing_dev_info;
428 	/*
429 	 * Private data.  You must have bd_claim'ed the block_device
430 	 * to use this.  NOTE:  bd_claim allows an owner to claim
431 	 * the same device multiple times, the owner must take special
432 	 * care to not mess up bd_private for that case.
433 	 */
434 	unsigned long		bd_private;
435 };
436 
437 /*
438  * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
439  * radix trees
440  */
441 #define PAGECACHE_TAG_DIRTY	0
442 #define PAGECACHE_TAG_WRITEBACK	1
443 
444 int mapping_tagged(struct address_space *mapping, int tag);
445 
446 /*
447  * Might pages of this file be mapped into userspace?
448  */
449 static inline int mapping_mapped(struct address_space *mapping)
450 {
451 	return	!prio_tree_empty(&mapping->i_mmap) ||
452 		!list_empty(&mapping->i_mmap_nonlinear);
453 }
454 
455 /*
456  * Might pages of this file have been modified in userspace?
457  * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
458  * marks vma as VM_SHARED if it is shared, and the file was opened for
459  * writing i.e. vma may be mprotected writable even if now readonly.
460  */
461 static inline int mapping_writably_mapped(struct address_space *mapping)
462 {
463 	return mapping->i_mmap_writable != 0;
464 }
465 
466 /*
467  * Use sequence counter to get consistent i_size on 32-bit processors.
468  */
469 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
470 #include <linux/seqlock.h>
471 #define __NEED_I_SIZE_ORDERED
472 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
473 #else
474 #define i_size_ordered_init(inode) do { } while (0)
475 #endif
476 
477 struct inode {
478 	struct hlist_node	i_hash;
479 	struct list_head	i_list;
480 	struct list_head	i_sb_list;
481 	struct list_head	i_dentry;
482 	unsigned long		i_ino;
483 	atomic_t		i_count;
484 	umode_t			i_mode;
485 	unsigned int		i_nlink;
486 	uid_t			i_uid;
487 	gid_t			i_gid;
488 	dev_t			i_rdev;
489 	loff_t			i_size;
490 	struct timespec		i_atime;
491 	struct timespec		i_mtime;
492 	struct timespec		i_ctime;
493 	unsigned int		i_blkbits;
494 	unsigned long		i_blksize;
495 	unsigned long		i_version;
496 	blkcnt_t		i_blocks;
497 	unsigned short          i_bytes;
498 	spinlock_t		i_lock;	/* i_blocks, i_bytes, maybe i_size */
499 	struct mutex		i_mutex;
500 	struct rw_semaphore	i_alloc_sem;
501 	struct inode_operations	*i_op;
502 	const struct file_operations	*i_fop;	/* former ->i_op->default_file_ops */
503 	struct super_block	*i_sb;
504 	struct file_lock	*i_flock;
505 	struct address_space	*i_mapping;
506 	struct address_space	i_data;
507 #ifdef CONFIG_QUOTA
508 	struct dquot		*i_dquot[MAXQUOTAS];
509 #endif
510 	/* These three should probably be a union */
511 	struct list_head	i_devices;
512 	struct pipe_inode_info	*i_pipe;
513 	struct block_device	*i_bdev;
514 	struct cdev		*i_cdev;
515 	int			i_cindex;
516 
517 	__u32			i_generation;
518 
519 #ifdef CONFIG_DNOTIFY
520 	unsigned long		i_dnotify_mask; /* Directory notify events */
521 	struct dnotify_struct	*i_dnotify; /* for directory notifications */
522 #endif
523 
524 #ifdef CONFIG_INOTIFY
525 	struct list_head	inotify_watches; /* watches on this inode */
526 	struct mutex		inotify_mutex;	/* protects the watches list */
527 #endif
528 
529 	unsigned long		i_state;
530 	unsigned long		dirtied_when;	/* jiffies of first dirtying */
531 
532 	unsigned int		i_flags;
533 
534 	atomic_t		i_writecount;
535 	void			*i_security;
536 	union {
537 		void		*generic_ip;
538 	} u;
539 #ifdef __NEED_I_SIZE_ORDERED
540 	seqcount_t		i_size_seqcount;
541 #endif
542 };
543 
544 /*
545  * NOTE: in a 32bit arch with a preemptable kernel and
546  * an UP compile the i_size_read/write must be atomic
547  * with respect to the local cpu (unlike with preempt disabled),
548  * but they don't need to be atomic with respect to other cpus like in
549  * true SMP (so they need either to either locally disable irq around
550  * the read or for example on x86 they can be still implemented as a
551  * cmpxchg8b without the need of the lock prefix). For SMP compiles
552  * and 64bit archs it makes no difference if preempt is enabled or not.
553  */
554 static inline loff_t i_size_read(struct inode *inode)
555 {
556 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
557 	loff_t i_size;
558 	unsigned int seq;
559 
560 	do {
561 		seq = read_seqcount_begin(&inode->i_size_seqcount);
562 		i_size = inode->i_size;
563 	} while (read_seqcount_retry(&inode->i_size_seqcount, seq));
564 	return i_size;
565 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
566 	loff_t i_size;
567 
568 	preempt_disable();
569 	i_size = inode->i_size;
570 	preempt_enable();
571 	return i_size;
572 #else
573 	return inode->i_size;
574 #endif
575 }
576 
577 
578 static inline void i_size_write(struct inode *inode, loff_t i_size)
579 {
580 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
581 	write_seqcount_begin(&inode->i_size_seqcount);
582 	inode->i_size = i_size;
583 	write_seqcount_end(&inode->i_size_seqcount);
584 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
585 	preempt_disable();
586 	inode->i_size = i_size;
587 	preempt_enable();
588 #else
589 	inode->i_size = i_size;
590 #endif
591 }
592 
593 static inline unsigned iminor(struct inode *inode)
594 {
595 	return MINOR(inode->i_rdev);
596 }
597 
598 static inline unsigned imajor(struct inode *inode)
599 {
600 	return MAJOR(inode->i_rdev);
601 }
602 
603 extern struct block_device *I_BDEV(struct inode *inode);
604 
605 struct fown_struct {
606 	rwlock_t lock;          /* protects pid, uid, euid fields */
607 	int pid;		/* pid or -pgrp where SIGIO should be sent */
608 	uid_t uid, euid;	/* uid/euid of process setting the owner */
609 	void *security;
610 	int signum;		/* posix.1b rt signal to be delivered on IO */
611 };
612 
613 /*
614  * Track a single file's readahead state
615  */
616 struct file_ra_state {
617 	unsigned long start;		/* Current window */
618 	unsigned long size;
619 	unsigned long flags;		/* ra flags RA_FLAG_xxx*/
620 	unsigned long cache_hit;	/* cache hit count*/
621 	unsigned long prev_page;	/* Cache last read() position */
622 	unsigned long ahead_start;	/* Ahead window */
623 	unsigned long ahead_size;
624 	unsigned long ra_pages;		/* Maximum readahead window */
625 	unsigned long mmap_hit;		/* Cache hit stat for mmap accesses */
626 	unsigned long mmap_miss;	/* Cache miss stat for mmap accesses */
627 };
628 #define RA_FLAG_MISS 0x01	/* a cache miss occured against this file */
629 #define RA_FLAG_INCACHE 0x02	/* file is already in cache */
630 
631 struct file {
632 	/*
633 	 * fu_list becomes invalid after file_free is called and queued via
634 	 * fu_rcuhead for RCU freeing
635 	 */
636 	union {
637 		struct list_head	fu_list;
638 		struct rcu_head 	fu_rcuhead;
639 	} f_u;
640 	struct dentry		*f_dentry;
641 	struct vfsmount         *f_vfsmnt;
642 	const struct file_operations	*f_op;
643 	atomic_t		f_count;
644 	unsigned int 		f_flags;
645 	mode_t			f_mode;
646 	loff_t			f_pos;
647 	struct fown_struct	f_owner;
648 	unsigned int		f_uid, f_gid;
649 	struct file_ra_state	f_ra;
650 
651 	unsigned long		f_version;
652 	void			*f_security;
653 
654 	/* needed for tty driver, and maybe others */
655 	void			*private_data;
656 
657 #ifdef CONFIG_EPOLL
658 	/* Used by fs/eventpoll.c to link all the hooks to this file */
659 	struct list_head	f_ep_links;
660 	spinlock_t		f_ep_lock;
661 #endif /* #ifdef CONFIG_EPOLL */
662 	struct address_space	*f_mapping;
663 };
664 extern spinlock_t files_lock;
665 #define file_list_lock() spin_lock(&files_lock);
666 #define file_list_unlock() spin_unlock(&files_lock);
667 
668 #define get_file(x)	atomic_inc(&(x)->f_count)
669 #define file_count(x)	atomic_read(&(x)->f_count)
670 
671 #define	MAX_NON_LFS	((1UL<<31) - 1)
672 
673 /* Page cache limit. The filesystems should put that into their s_maxbytes
674    limits, otherwise bad things can happen in VM. */
675 #if BITS_PER_LONG==32
676 #define MAX_LFS_FILESIZE	(((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
677 #elif BITS_PER_LONG==64
678 #define MAX_LFS_FILESIZE 	0x7fffffffffffffffUL
679 #endif
680 
681 #define FL_POSIX	1
682 #define FL_FLOCK	2
683 #define FL_ACCESS	8	/* not trying to lock, just looking */
684 #define FL_LEASE	32	/* lease held on this file */
685 #define FL_SLEEP	128	/* A blocking lock */
686 
687 /*
688  * The POSIX file lock owner is determined by
689  * the "struct files_struct" in the thread group
690  * (or NULL for no owner - BSD locks).
691  *
692  * Lockd stuffs a "host" pointer into this.
693  */
694 typedef struct files_struct *fl_owner_t;
695 
696 struct file_lock_operations {
697 	void (*fl_insert)(struct file_lock *);	/* lock insertion callback */
698 	void (*fl_remove)(struct file_lock *);	/* lock removal callback */
699 	void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
700 	void (*fl_release_private)(struct file_lock *);
701 };
702 
703 struct lock_manager_operations {
704 	int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
705 	void (*fl_notify)(struct file_lock *);	/* unblock callback */
706 	void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
707 	void (*fl_release_private)(struct file_lock *);
708 	void (*fl_break)(struct file_lock *);
709 	int (*fl_mylease)(struct file_lock *, struct file_lock *);
710 	int (*fl_change)(struct file_lock **, int);
711 };
712 
713 /* that will die - we need it for nfs_lock_info */
714 #include <linux/nfs_fs_i.h>
715 
716 struct file_lock {
717 	struct file_lock *fl_next;	/* singly linked list for this inode  */
718 	struct list_head fl_link;	/* doubly linked list of all locks */
719 	struct list_head fl_block;	/* circular list of blocked processes */
720 	fl_owner_t fl_owner;
721 	unsigned int fl_pid;
722 	wait_queue_head_t fl_wait;
723 	struct file *fl_file;
724 	unsigned char fl_flags;
725 	unsigned char fl_type;
726 	loff_t fl_start;
727 	loff_t fl_end;
728 
729 	struct fasync_struct *	fl_fasync; /* for lease break notifications */
730 	unsigned long fl_break_time;	/* for nonblocking lease breaks */
731 
732 	struct file_lock_operations *fl_ops;	/* Callbacks for filesystems */
733 	struct lock_manager_operations *fl_lmops;	/* Callbacks for lockmanagers */
734 	union {
735 		struct nfs_lock_info	nfs_fl;
736 		struct nfs4_lock_info	nfs4_fl;
737 	} fl_u;
738 };
739 
740 /* The following constant reflects the upper bound of the file/locking space */
741 #ifndef OFFSET_MAX
742 #define INT_LIMIT(x)	(~((x)1 << (sizeof(x)*8 - 1)))
743 #define OFFSET_MAX	INT_LIMIT(loff_t)
744 #define OFFT_OFFSET_MAX	INT_LIMIT(off_t)
745 #endif
746 
747 #include <linux/fcntl.h>
748 
749 extern int fcntl_getlk(struct file *, struct flock __user *);
750 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
751 			struct flock __user *);
752 
753 #if BITS_PER_LONG == 32
754 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
755 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
756 			struct flock64 __user *);
757 #endif
758 
759 extern void send_sigio(struct fown_struct *fown, int fd, int band);
760 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
761 extern int fcntl_getlease(struct file *filp);
762 
763 /* fs/sync.c */
764 extern int do_sync_file_range(struct file *file, loff_t offset, loff_t endbyte,
765 			unsigned int flags);
766 
767 /* fs/locks.c */
768 extern void locks_init_lock(struct file_lock *);
769 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
770 extern void locks_remove_posix(struct file *, fl_owner_t);
771 extern void locks_remove_flock(struct file *);
772 extern int posix_test_lock(struct file *, struct file_lock *, struct file_lock *);
773 extern int posix_lock_file_conf(struct file *, struct file_lock *, struct file_lock *);
774 extern int posix_lock_file(struct file *, struct file_lock *);
775 extern int posix_lock_file_wait(struct file *, struct file_lock *);
776 extern int posix_unblock_lock(struct file *, struct file_lock *);
777 extern int posix_locks_deadlock(struct file_lock *, struct file_lock *);
778 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
779 extern int __break_lease(struct inode *inode, unsigned int flags);
780 extern void lease_get_mtime(struct inode *, struct timespec *time);
781 extern int setlease(struct file *, long, struct file_lock **);
782 extern int lease_modify(struct file_lock **, int);
783 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
784 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
785 
786 struct fasync_struct {
787 	int	magic;
788 	int	fa_fd;
789 	struct	fasync_struct	*fa_next; /* singly linked list */
790 	struct	file 		*fa_file;
791 };
792 
793 #define FASYNC_MAGIC 0x4601
794 
795 /* SMP safe fasync helpers: */
796 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
797 /* can be called from interrupts */
798 extern void kill_fasync(struct fasync_struct **, int, int);
799 /* only for net: no internal synchronization */
800 extern void __kill_fasync(struct fasync_struct *, int, int);
801 
802 extern int f_setown(struct file *filp, unsigned long arg, int force);
803 extern void f_delown(struct file *filp);
804 extern int send_sigurg(struct fown_struct *fown);
805 
806 /*
807  *	Umount options
808  */
809 
810 #define MNT_FORCE	0x00000001	/* Attempt to forcibily umount */
811 #define MNT_DETACH	0x00000002	/* Just detach from the tree */
812 #define MNT_EXPIRE	0x00000004	/* Mark for expiry */
813 
814 extern struct list_head super_blocks;
815 extern spinlock_t sb_lock;
816 
817 #define sb_entry(list)	list_entry((list), struct super_block, s_list)
818 #define S_BIAS (1<<30)
819 struct super_block {
820 	struct list_head	s_list;		/* Keep this first */
821 	dev_t			s_dev;		/* search index; _not_ kdev_t */
822 	unsigned long		s_blocksize;
823 	unsigned char		s_blocksize_bits;
824 	unsigned char		s_dirt;
825 	unsigned long long	s_maxbytes;	/* Max file size */
826 	struct file_system_type	*s_type;
827 	struct super_operations	*s_op;
828 	struct dquot_operations	*dq_op;
829  	struct quotactl_ops	*s_qcop;
830 	struct export_operations *s_export_op;
831 	unsigned long		s_flags;
832 	unsigned long		s_magic;
833 	struct dentry		*s_root;
834 	struct rw_semaphore	s_umount;
835 	struct mutex		s_lock;
836 	int			s_count;
837 	int			s_syncing;
838 	int			s_need_sync_fs;
839 	atomic_t		s_active;
840 	void                    *s_security;
841 	struct xattr_handler	**s_xattr;
842 
843 	struct list_head	s_inodes;	/* all inodes */
844 	struct list_head	s_dirty;	/* dirty inodes */
845 	struct list_head	s_io;		/* parked for writeback */
846 	struct hlist_head	s_anon;		/* anonymous dentries for (nfs) exporting */
847 	struct list_head	s_files;
848 
849 	struct block_device	*s_bdev;
850 	struct list_head	s_instances;
851 	struct quota_info	s_dquot;	/* Diskquota specific options */
852 
853 	int			s_frozen;
854 	wait_queue_head_t	s_wait_unfrozen;
855 
856 	char s_id[32];				/* Informational name */
857 
858 	void 			*s_fs_info;	/* Filesystem private info */
859 
860 	/*
861 	 * The next field is for VFS *only*. No filesystems have any business
862 	 * even looking at it. You had been warned.
863 	 */
864 	struct mutex s_vfs_rename_mutex;	/* Kludge */
865 
866 	/* Granularity of c/m/atime in ns.
867 	   Cannot be worse than a second */
868 	u32		   s_time_gran;
869 };
870 
871 extern struct timespec current_fs_time(struct super_block *sb);
872 
873 /*
874  * Snapshotting support.
875  */
876 enum {
877 	SB_UNFROZEN = 0,
878 	SB_FREEZE_WRITE	= 1,
879 	SB_FREEZE_TRANS = 2,
880 };
881 
882 #define vfs_check_frozen(sb, level) \
883 	wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
884 
885 static inline void get_fs_excl(void)
886 {
887 	atomic_inc(&current->fs_excl);
888 }
889 
890 static inline void put_fs_excl(void)
891 {
892 	atomic_dec(&current->fs_excl);
893 }
894 
895 static inline int has_fs_excl(void)
896 {
897 	return atomic_read(&current->fs_excl);
898 }
899 
900 
901 /*
902  * Superblock locking.
903  */
904 static inline void lock_super(struct super_block * sb)
905 {
906 	get_fs_excl();
907 	mutex_lock(&sb->s_lock);
908 }
909 
910 static inline void unlock_super(struct super_block * sb)
911 {
912 	put_fs_excl();
913 	mutex_unlock(&sb->s_lock);
914 }
915 
916 /*
917  * VFS helper functions..
918  */
919 extern int vfs_permission(struct nameidata *, int);
920 extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
921 extern int vfs_mkdir(struct inode *, struct dentry *, int);
922 extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
923 extern int vfs_symlink(struct inode *, struct dentry *, const char *, int);
924 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
925 extern int vfs_rmdir(struct inode *, struct dentry *);
926 extern int vfs_unlink(struct inode *, struct dentry *);
927 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
928 
929 /*
930  * VFS dentry helper functions.
931  */
932 extern void dentry_unhash(struct dentry *dentry);
933 
934 /*
935  * VFS file helper functions.
936  */
937 extern int file_permission(struct file *, int);
938 
939 /*
940  * File types
941  *
942  * NOTE! These match bits 12..15 of stat.st_mode
943  * (ie "(i_mode >> 12) & 15").
944  */
945 #define DT_UNKNOWN	0
946 #define DT_FIFO		1
947 #define DT_CHR		2
948 #define DT_DIR		4
949 #define DT_BLK		6
950 #define DT_REG		8
951 #define DT_LNK		10
952 #define DT_SOCK		12
953 #define DT_WHT		14
954 
955 #define OSYNC_METADATA	(1<<0)
956 #define OSYNC_DATA	(1<<1)
957 #define OSYNC_INODE	(1<<2)
958 int generic_osync_inode(struct inode *, struct address_space *, int);
959 
960 /*
961  * This is the "filldir" function type, used by readdir() to let
962  * the kernel specify what kind of dirent layout it wants to have.
963  * This allows the kernel to read directories into kernel space or
964  * to have different dirent layouts depending on the binary type.
965  */
966 typedef int (*filldir_t)(void *, const char *, int, loff_t, ino_t, unsigned);
967 
968 struct block_device_operations {
969 	int (*open) (struct inode *, struct file *);
970 	int (*release) (struct inode *, struct file *);
971 	int (*ioctl) (struct inode *, struct file *, unsigned, unsigned long);
972 	long (*unlocked_ioctl) (struct file *, unsigned, unsigned long);
973 	long (*compat_ioctl) (struct file *, unsigned, unsigned long);
974 	int (*direct_access) (struct block_device *, sector_t, unsigned long *);
975 	int (*media_changed) (struct gendisk *);
976 	int (*revalidate_disk) (struct gendisk *);
977 	int (*getgeo)(struct block_device *, struct hd_geometry *);
978 	struct module *owner;
979 };
980 
981 /*
982  * "descriptor" for what we're up to with a read for sendfile().
983  * This allows us to use the same read code yet
984  * have multiple different users of the data that
985  * we read from a file.
986  *
987  * The simplest case just copies the data to user
988  * mode.
989  */
990 typedef struct {
991 	size_t written;
992 	size_t count;
993 	union {
994 		char __user * buf;
995 		void *data;
996 	} arg;
997 	int error;
998 } read_descriptor_t;
999 
1000 typedef int (*read_actor_t)(read_descriptor_t *, struct page *, unsigned long, unsigned long);
1001 
1002 /* These macros are for out of kernel modules to test that
1003  * the kernel supports the unlocked_ioctl and compat_ioctl
1004  * fields in struct file_operations. */
1005 #define HAVE_COMPAT_IOCTL 1
1006 #define HAVE_UNLOCKED_IOCTL 1
1007 
1008 /*
1009  * NOTE:
1010  * read, write, poll, fsync, readv, writev, unlocked_ioctl and compat_ioctl
1011  * can be called without the big kernel lock held in all filesystems.
1012  */
1013 struct file_operations {
1014 	struct module *owner;
1015 	loff_t (*llseek) (struct file *, loff_t, int);
1016 	ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1017 	ssize_t (*aio_read) (struct kiocb *, char __user *, size_t, loff_t);
1018 	ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1019 	ssize_t (*aio_write) (struct kiocb *, const char __user *, size_t, loff_t);
1020 	int (*readdir) (struct file *, void *, filldir_t);
1021 	unsigned int (*poll) (struct file *, struct poll_table_struct *);
1022 	int (*ioctl) (struct inode *, struct file *, unsigned int, unsigned long);
1023 	long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1024 	long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1025 	int (*mmap) (struct file *, struct vm_area_struct *);
1026 	int (*open) (struct inode *, struct file *);
1027 	int (*flush) (struct file *);
1028 	int (*release) (struct inode *, struct file *);
1029 	int (*fsync) (struct file *, struct dentry *, int datasync);
1030 	int (*aio_fsync) (struct kiocb *, int datasync);
1031 	int (*fasync) (int, struct file *, int);
1032 	int (*lock) (struct file *, int, struct file_lock *);
1033 	ssize_t (*readv) (struct file *, const struct iovec *, unsigned long, loff_t *);
1034 	ssize_t (*writev) (struct file *, const struct iovec *, unsigned long, loff_t *);
1035 	ssize_t (*sendfile) (struct file *, loff_t *, size_t, read_actor_t, void *);
1036 	ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1037 	unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1038 	int (*check_flags)(int);
1039 	int (*dir_notify)(struct file *filp, unsigned long arg);
1040 	int (*flock) (struct file *, int, struct file_lock *);
1041 	ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1042 	ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1043 };
1044 
1045 struct inode_operations {
1046 	int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1047 	struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1048 	int (*link) (struct dentry *,struct inode *,struct dentry *);
1049 	int (*unlink) (struct inode *,struct dentry *);
1050 	int (*symlink) (struct inode *,struct dentry *,const char *);
1051 	int (*mkdir) (struct inode *,struct dentry *,int);
1052 	int (*rmdir) (struct inode *,struct dentry *);
1053 	int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1054 	int (*rename) (struct inode *, struct dentry *,
1055 			struct inode *, struct dentry *);
1056 	int (*readlink) (struct dentry *, char __user *,int);
1057 	void * (*follow_link) (struct dentry *, struct nameidata *);
1058 	void (*put_link) (struct dentry *, struct nameidata *, void *);
1059 	void (*truncate) (struct inode *);
1060 	int (*permission) (struct inode *, int, struct nameidata *);
1061 	int (*setattr) (struct dentry *, struct iattr *);
1062 	int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1063 	int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1064 	ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1065 	ssize_t (*listxattr) (struct dentry *, char *, size_t);
1066 	int (*removexattr) (struct dentry *, const char *);
1067 	void (*truncate_range)(struct inode *, loff_t, loff_t);
1068 };
1069 
1070 struct seq_file;
1071 
1072 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1073 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1074 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1075 		unsigned long, loff_t *);
1076 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1077 		unsigned long, loff_t *);
1078 
1079 /*
1080  * NOTE: write_inode, delete_inode, clear_inode, put_inode can be called
1081  * without the big kernel lock held in all filesystems.
1082  */
1083 struct super_operations {
1084    	struct inode *(*alloc_inode)(struct super_block *sb);
1085 	void (*destroy_inode)(struct inode *);
1086 
1087 	void (*read_inode) (struct inode *);
1088 
1089    	void (*dirty_inode) (struct inode *);
1090 	int (*write_inode) (struct inode *, int);
1091 	void (*put_inode) (struct inode *);
1092 	void (*drop_inode) (struct inode *);
1093 	void (*delete_inode) (struct inode *);
1094 	void (*put_super) (struct super_block *);
1095 	void (*write_super) (struct super_block *);
1096 	int (*sync_fs)(struct super_block *sb, int wait);
1097 	void (*write_super_lockfs) (struct super_block *);
1098 	void (*unlockfs) (struct super_block *);
1099 	int (*statfs) (struct super_block *, struct kstatfs *);
1100 	int (*remount_fs) (struct super_block *, int *, char *);
1101 	void (*clear_inode) (struct inode *);
1102 	void (*umount_begin) (struct super_block *);
1103 
1104 	int (*show_options)(struct seq_file *, struct vfsmount *);
1105 	int (*show_stats)(struct seq_file *, struct vfsmount *);
1106 
1107 	ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1108 	ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1109 };
1110 
1111 /* Inode state bits.  Protected by inode_lock. */
1112 #define I_DIRTY_SYNC		1 /* Not dirty enough for O_DATASYNC */
1113 #define I_DIRTY_DATASYNC	2 /* Data-related inode changes pending */
1114 #define I_DIRTY_PAGES		4 /* Data-related inode changes pending */
1115 #define __I_LOCK		3
1116 #define I_LOCK			(1 << __I_LOCK)
1117 #define I_FREEING		16
1118 #define I_CLEAR			32
1119 #define I_NEW			64
1120 #define I_WILL_FREE		128
1121 
1122 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1123 
1124 extern void __mark_inode_dirty(struct inode *, int);
1125 static inline void mark_inode_dirty(struct inode *inode)
1126 {
1127 	__mark_inode_dirty(inode, I_DIRTY);
1128 }
1129 
1130 static inline void mark_inode_dirty_sync(struct inode *inode)
1131 {
1132 	__mark_inode_dirty(inode, I_DIRTY_SYNC);
1133 }
1134 
1135 static inline void inode_inc_link_count(struct inode *inode)
1136 {
1137 	inode->i_nlink++;
1138 	mark_inode_dirty(inode);
1139 }
1140 
1141 static inline void inode_dec_link_count(struct inode *inode)
1142 {
1143 	inode->i_nlink--;
1144 	mark_inode_dirty(inode);
1145 }
1146 
1147 extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
1148 static inline void file_accessed(struct file *file)
1149 {
1150 	if (!(file->f_flags & O_NOATIME))
1151 		touch_atime(file->f_vfsmnt, file->f_dentry);
1152 }
1153 
1154 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1155 
1156 /**
1157  * struct export_operations - for nfsd to communicate with file systems
1158  * @decode_fh:      decode a file handle fragment and return a &struct dentry
1159  * @encode_fh:      encode a file handle fragment from a dentry
1160  * @get_name:       find the name for a given inode in a given directory
1161  * @get_parent:     find the parent of a given directory
1162  * @get_dentry:     find a dentry for the inode given a file handle sub-fragment
1163  * @find_exported_dentry:
1164  *	set by the exporting module to a standard helper function.
1165  *
1166  * Description:
1167  *    The export_operations structure provides a means for nfsd to communicate
1168  *    with a particular exported file system  - particularly enabling nfsd and
1169  *    the filesystem to co-operate when dealing with file handles.
1170  *
1171  *    export_operations contains two basic operation for dealing with file
1172  *    handles, decode_fh() and encode_fh(), and allows for some other
1173  *    operations to be defined which standard helper routines use to get
1174  *    specific information from the filesystem.
1175  *
1176  *    nfsd encodes information use to determine which filesystem a filehandle
1177  *    applies to in the initial part of the file handle.  The remainder, termed
1178  *    a file handle fragment, is controlled completely by the filesystem.  The
1179  *    standard helper routines assume that this fragment will contain one or
1180  *    two sub-fragments, one which identifies the file, and one which may be
1181  *    used to identify the (a) directory containing the file.
1182  *
1183  *    In some situations, nfsd needs to get a dentry which is connected into a
1184  *    specific part of the file tree.  To allow for this, it passes the
1185  *    function acceptable() together with a @context which can be used to see
1186  *    if the dentry is acceptable.  As there can be multiple dentrys for a
1187  *    given file, the filesystem should check each one for acceptability before
1188  *    looking for the next.  As soon as an acceptable one is found, it should
1189  *    be returned.
1190  *
1191  * decode_fh:
1192  *    @decode_fh is given a &struct super_block (@sb), a file handle fragment
1193  *    (@fh, @fh_len) and an acceptability testing function (@acceptable,
1194  *    @context).  It should return a &struct dentry which refers to the same
1195  *    file that the file handle fragment refers to,  and which passes the
1196  *    acceptability test.  If it cannot, it should return a %NULL pointer if
1197  *    the file was found but no acceptable &dentries were available, or a
1198  *    %ERR_PTR error code indicating why it couldn't be found (e.g. %ENOENT or
1199  *    %ENOMEM).
1200  *
1201  * encode_fh:
1202  *    @encode_fh should store in the file handle fragment @fh (using at most
1203  *    @max_len bytes) information that can be used by @decode_fh to recover the
1204  *    file refered to by the &struct dentry @de.  If the @connectable flag is
1205  *    set, the encode_fh() should store sufficient information so that a good
1206  *    attempt can be made to find not only the file but also it's place in the
1207  *    filesystem.   This typically means storing a reference to de->d_parent in
1208  *    the filehandle fragment.  encode_fh() should return the number of bytes
1209  *    stored or a negative error code such as %-ENOSPC
1210  *
1211  * get_name:
1212  *    @get_name should find a name for the given @child in the given @parent
1213  *    directory.  The name should be stored in the @name (with the
1214  *    understanding that it is already pointing to a a %NAME_MAX+1 sized
1215  *    buffer.   get_name() should return %0 on success, a negative error code
1216  *    or error.  @get_name will be called without @parent->i_mutex held.
1217  *
1218  * get_parent:
1219  *    @get_parent should find the parent directory for the given @child which
1220  *    is also a directory.  In the event that it cannot be found, or storage
1221  *    space cannot be allocated, a %ERR_PTR should be returned.
1222  *
1223  * get_dentry:
1224  *    Given a &super_block (@sb) and a pointer to a file-system specific inode
1225  *    identifier, possibly an inode number, (@inump) get_dentry() should find
1226  *    the identified inode and return a dentry for that inode.  Any suitable
1227  *    dentry can be returned including, if necessary, a new dentry created with
1228  *    d_alloc_root.  The caller can then find any other extant dentrys by
1229  *    following the d_alias links.  If a new dentry was created using
1230  *    d_alloc_root, DCACHE_NFSD_DISCONNECTED should be set, and the dentry
1231  *    should be d_rehash()ed.
1232  *
1233  *    If the inode cannot be found, either a %NULL pointer or an %ERR_PTR code
1234  *    can be returned.  The @inump will be whatever was passed to
1235  *    nfsd_find_fh_dentry() in either the @obj or @parent parameters.
1236  *
1237  * Locking rules:
1238  *    get_parent is called with child->d_inode->i_mutex down
1239  *    get_name is not (which is possibly inconsistent)
1240  */
1241 
1242 struct export_operations {
1243 	struct dentry *(*decode_fh)(struct super_block *sb, __u32 *fh, int fh_len, int fh_type,
1244 			 int (*acceptable)(void *context, struct dentry *de),
1245 			 void *context);
1246 	int (*encode_fh)(struct dentry *de, __u32 *fh, int *max_len,
1247 			 int connectable);
1248 
1249 	/* the following are only called from the filesystem itself */
1250 	int (*get_name)(struct dentry *parent, char *name,
1251 			struct dentry *child);
1252 	struct dentry * (*get_parent)(struct dentry *child);
1253 	struct dentry * (*get_dentry)(struct super_block *sb, void *inump);
1254 
1255 	/* This is set by the exporting module to a standard helper */
1256 	struct dentry * (*find_exported_dentry)(
1257 		struct super_block *sb, void *obj, void *parent,
1258 		int (*acceptable)(void *context, struct dentry *de),
1259 		void *context);
1260 
1261 
1262 };
1263 
1264 extern struct dentry *
1265 find_exported_dentry(struct super_block *sb, void *obj, void *parent,
1266 		     int (*acceptable)(void *context, struct dentry *de),
1267 		     void *context);
1268 
1269 struct file_system_type {
1270 	const char *name;
1271 	int fs_flags;
1272 	struct super_block *(*get_sb) (struct file_system_type *, int,
1273 				       const char *, void *);
1274 	void (*kill_sb) (struct super_block *);
1275 	struct module *owner;
1276 	struct file_system_type * next;
1277 	struct list_head fs_supers;
1278 };
1279 
1280 struct super_block *get_sb_bdev(struct file_system_type *fs_type,
1281 	int flags, const char *dev_name, void *data,
1282 	int (*fill_super)(struct super_block *, void *, int));
1283 struct super_block *get_sb_single(struct file_system_type *fs_type,
1284 	int flags, void *data,
1285 	int (*fill_super)(struct super_block *, void *, int));
1286 struct super_block *get_sb_nodev(struct file_system_type *fs_type,
1287 	int flags, void *data,
1288 	int (*fill_super)(struct super_block *, void *, int));
1289 void generic_shutdown_super(struct super_block *sb);
1290 void kill_block_super(struct super_block *sb);
1291 void kill_anon_super(struct super_block *sb);
1292 void kill_litter_super(struct super_block *sb);
1293 void deactivate_super(struct super_block *sb);
1294 int set_anon_super(struct super_block *s, void *data);
1295 struct super_block *sget(struct file_system_type *type,
1296 			int (*test)(struct super_block *,void *),
1297 			int (*set)(struct super_block *,void *),
1298 			void *data);
1299 struct super_block *get_sb_pseudo(struct file_system_type *, char *,
1300 			struct super_operations *ops, unsigned long);
1301 int __put_super(struct super_block *sb);
1302 int __put_super_and_need_restart(struct super_block *sb);
1303 void unnamed_dev_init(void);
1304 
1305 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1306 #define fops_get(fops) \
1307 	(((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1308 #define fops_put(fops) \
1309 	do { if (fops) module_put((fops)->owner); } while(0)
1310 
1311 extern int register_filesystem(struct file_system_type *);
1312 extern int unregister_filesystem(struct file_system_type *);
1313 extern struct vfsmount *kern_mount(struct file_system_type *);
1314 extern int may_umount_tree(struct vfsmount *);
1315 extern int may_umount(struct vfsmount *);
1316 extern void umount_tree(struct vfsmount *, int, struct list_head *);
1317 extern void release_mounts(struct list_head *);
1318 extern long do_mount(char *, char *, char *, unsigned long, void *);
1319 extern struct vfsmount *copy_tree(struct vfsmount *, struct dentry *, int);
1320 extern void mnt_set_mountpoint(struct vfsmount *, struct dentry *,
1321 				  struct vfsmount *);
1322 
1323 extern int vfs_statfs(struct super_block *, struct kstatfs *);
1324 
1325 /* /sys/fs */
1326 extern struct subsystem fs_subsys;
1327 
1328 #define FLOCK_VERIFY_READ  1
1329 #define FLOCK_VERIFY_WRITE 2
1330 
1331 extern int locks_mandatory_locked(struct inode *);
1332 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1333 
1334 /*
1335  * Candidates for mandatory locking have the setgid bit set
1336  * but no group execute bit -  an otherwise meaningless combination.
1337  */
1338 #define MANDATORY_LOCK(inode) \
1339 	(IS_MANDLOCK(inode) && ((inode)->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID)
1340 
1341 static inline int locks_verify_locked(struct inode *inode)
1342 {
1343 	if (MANDATORY_LOCK(inode))
1344 		return locks_mandatory_locked(inode);
1345 	return 0;
1346 }
1347 
1348 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1349 
1350 static inline int locks_verify_truncate(struct inode *inode,
1351 				    struct file *filp,
1352 				    loff_t size)
1353 {
1354 	if (inode->i_flock && MANDATORY_LOCK(inode))
1355 		return locks_mandatory_area(
1356 			FLOCK_VERIFY_WRITE, inode, filp,
1357 			size < inode->i_size ? size : inode->i_size,
1358 			(size < inode->i_size ? inode->i_size - size
1359 			 : size - inode->i_size)
1360 		);
1361 	return 0;
1362 }
1363 
1364 static inline int break_lease(struct inode *inode, unsigned int mode)
1365 {
1366 	if (inode->i_flock)
1367 		return __break_lease(inode, mode);
1368 	return 0;
1369 }
1370 
1371 /* fs/open.c */
1372 
1373 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
1374 		       struct file *filp);
1375 extern long do_sys_open(int fdf, const char __user *filename, int flags,
1376 			int mode);
1377 extern struct file *filp_open(const char *, int, int);
1378 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int);
1379 extern int filp_close(struct file *, fl_owner_t id);
1380 extern char * getname(const char __user *);
1381 
1382 /* fs/dcache.c */
1383 extern void __init vfs_caches_init_early(void);
1384 extern void __init vfs_caches_init(unsigned long);
1385 
1386 #define __getname()	kmem_cache_alloc(names_cachep, SLAB_KERNEL)
1387 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
1388 #ifndef CONFIG_AUDITSYSCALL
1389 #define putname(name)   __putname(name)
1390 #else
1391 extern void putname(const char *name);
1392 #endif
1393 
1394 extern int register_blkdev(unsigned int, const char *);
1395 extern int unregister_blkdev(unsigned int, const char *);
1396 extern struct block_device *bdget(dev_t);
1397 extern void bd_set_size(struct block_device *, loff_t size);
1398 extern void bd_forget(struct inode *inode);
1399 extern void bdput(struct block_device *);
1400 extern struct block_device *open_by_devnum(dev_t, unsigned);
1401 extern const struct file_operations def_blk_fops;
1402 extern struct address_space_operations def_blk_aops;
1403 extern const struct file_operations def_chr_fops;
1404 extern const struct file_operations bad_sock_fops;
1405 extern const struct file_operations def_fifo_fops;
1406 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
1407 extern int blkdev_ioctl(struct inode *, struct file *, unsigned, unsigned long);
1408 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
1409 extern int blkdev_get(struct block_device *, mode_t, unsigned);
1410 extern int blkdev_put(struct block_device *);
1411 extern int bd_claim(struct block_device *, void *);
1412 extern void bd_release(struct block_device *);
1413 #ifdef CONFIG_SYSFS
1414 extern int bd_claim_by_disk(struct block_device *, void *, struct gendisk *);
1415 extern void bd_release_from_disk(struct block_device *, struct gendisk *);
1416 #else
1417 #define bd_claim_by_disk(bdev, holder, disk)	bd_claim(bdev, holder)
1418 #define bd_release_from_disk(bdev, disk)	bd_release(bdev)
1419 #endif
1420 
1421 /* fs/char_dev.c */
1422 #define CHRDEV_MAJOR_HASH_SIZE	255
1423 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
1424 extern int register_chrdev_region(dev_t, unsigned, const char *);
1425 extern int register_chrdev(unsigned int, const char *,
1426 			   const struct file_operations *);
1427 extern int unregister_chrdev(unsigned int, const char *);
1428 extern void unregister_chrdev_region(dev_t, unsigned);
1429 extern int chrdev_open(struct inode *, struct file *);
1430 extern void chrdev_show(struct seq_file *,off_t);
1431 
1432 /* fs/block_dev.c */
1433 #define BLKDEV_MAJOR_HASH_SIZE	255
1434 #define BDEVNAME_SIZE	32	/* Largest string for a blockdev identifier */
1435 extern const char *__bdevname(dev_t, char *buffer);
1436 extern const char *bdevname(struct block_device *bdev, char *buffer);
1437 extern struct block_device *lookup_bdev(const char *);
1438 extern struct block_device *open_bdev_excl(const char *, int, void *);
1439 extern void close_bdev_excl(struct block_device *);
1440 extern void blkdev_show(struct seq_file *,off_t);
1441 
1442 extern void init_special_inode(struct inode *, umode_t, dev_t);
1443 
1444 /* Invalid inode operations -- fs/bad_inode.c */
1445 extern void make_bad_inode(struct inode *);
1446 extern int is_bad_inode(struct inode *);
1447 
1448 extern const struct file_operations read_fifo_fops;
1449 extern const struct file_operations write_fifo_fops;
1450 extern const struct file_operations rdwr_fifo_fops;
1451 
1452 extern int fs_may_remount_ro(struct super_block *);
1453 
1454 /*
1455  * return READ, READA, or WRITE
1456  */
1457 #define bio_rw(bio)		((bio)->bi_rw & (RW_MASK | RWA_MASK))
1458 
1459 /*
1460  * return data direction, READ or WRITE
1461  */
1462 #define bio_data_dir(bio)	((bio)->bi_rw & 1)
1463 
1464 extern int check_disk_change(struct block_device *);
1465 extern int invalidate_inodes(struct super_block *);
1466 extern int __invalidate_device(struct block_device *);
1467 extern int invalidate_partition(struct gendisk *, int);
1468 unsigned long invalidate_mapping_pages(struct address_space *mapping,
1469 					pgoff_t start, pgoff_t end);
1470 unsigned long invalidate_inode_pages(struct address_space *mapping);
1471 static inline void invalidate_remote_inode(struct inode *inode)
1472 {
1473 	if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
1474 	    S_ISLNK(inode->i_mode))
1475 		invalidate_inode_pages(inode->i_mapping);
1476 }
1477 extern int invalidate_inode_pages2(struct address_space *mapping);
1478 extern int invalidate_inode_pages2_range(struct address_space *mapping,
1479 					 pgoff_t start, pgoff_t end);
1480 extern int write_inode_now(struct inode *, int);
1481 extern int filemap_fdatawrite(struct address_space *);
1482 extern int filemap_flush(struct address_space *);
1483 extern int filemap_fdatawait(struct address_space *);
1484 extern int filemap_write_and_wait(struct address_space *mapping);
1485 extern int filemap_write_and_wait_range(struct address_space *mapping,
1486 				        loff_t lstart, loff_t lend);
1487 extern int wait_on_page_writeback_range(struct address_space *mapping,
1488 				pgoff_t start, pgoff_t end);
1489 extern int __filemap_fdatawrite_range(struct address_space *mapping,
1490 				loff_t start, loff_t end, int sync_mode);
1491 
1492 extern long do_fsync(struct file *file, int datasync);
1493 extern void sync_supers(void);
1494 extern void sync_filesystems(int wait);
1495 extern void emergency_sync(void);
1496 extern void emergency_remount(void);
1497 extern int do_remount_sb(struct super_block *sb, int flags,
1498 			 void *data, int force);
1499 extern sector_t bmap(struct inode *, sector_t);
1500 extern int notify_change(struct dentry *, struct iattr *);
1501 extern int permission(struct inode *, int, struct nameidata *);
1502 extern int generic_permission(struct inode *, int,
1503 		int (*check_acl)(struct inode *, int));
1504 
1505 extern int get_write_access(struct inode *);
1506 extern int deny_write_access(struct file *);
1507 static inline void put_write_access(struct inode * inode)
1508 {
1509 	atomic_dec(&inode->i_writecount);
1510 }
1511 static inline void allow_write_access(struct file *file)
1512 {
1513 	if (file)
1514 		atomic_inc(&file->f_dentry->d_inode->i_writecount);
1515 }
1516 extern int do_pipe(int *);
1517 
1518 extern int open_namei(int dfd, const char *, int, int, struct nameidata *);
1519 extern int may_open(struct nameidata *, int, int);
1520 
1521 extern int kernel_read(struct file *, unsigned long, char *, unsigned long);
1522 extern struct file * open_exec(const char *);
1523 
1524 /* fs/dcache.c -- generic fs support functions */
1525 extern int is_subdir(struct dentry *, struct dentry *);
1526 extern ino_t find_inode_number(struct dentry *, struct qstr *);
1527 
1528 #include <linux/err.h>
1529 
1530 /* needed for stackable file system support */
1531 extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
1532 
1533 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
1534 
1535 extern void inode_init_once(struct inode *);
1536 extern void iput(struct inode *);
1537 extern struct inode * igrab(struct inode *);
1538 extern ino_t iunique(struct super_block *, ino_t);
1539 extern int inode_needs_sync(struct inode *inode);
1540 extern void generic_delete_inode(struct inode *inode);
1541 extern void generic_drop_inode(struct inode *inode);
1542 
1543 extern struct inode *ilookup5_nowait(struct super_block *sb,
1544 		unsigned long hashval, int (*test)(struct inode *, void *),
1545 		void *data);
1546 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
1547 		int (*test)(struct inode *, void *), void *data);
1548 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
1549 
1550 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
1551 extern struct inode * iget_locked(struct super_block *, unsigned long);
1552 extern void unlock_new_inode(struct inode *);
1553 
1554 static inline struct inode *iget(struct super_block *sb, unsigned long ino)
1555 {
1556 	struct inode *inode = iget_locked(sb, ino);
1557 
1558 	if (inode && (inode->i_state & I_NEW)) {
1559 		sb->s_op->read_inode(inode);
1560 		unlock_new_inode(inode);
1561 	}
1562 
1563 	return inode;
1564 }
1565 
1566 extern void __iget(struct inode * inode);
1567 extern void clear_inode(struct inode *);
1568 extern void destroy_inode(struct inode *);
1569 extern struct inode *new_inode(struct super_block *);
1570 extern int remove_suid(struct dentry *);
1571 extern void remove_dquot_ref(struct super_block *, int, struct list_head *);
1572 
1573 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
1574 extern void remove_inode_hash(struct inode *);
1575 static inline void insert_inode_hash(struct inode *inode) {
1576 	__insert_inode_hash(inode, inode->i_ino);
1577 }
1578 
1579 extern struct file * get_empty_filp(void);
1580 extern void file_move(struct file *f, struct list_head *list);
1581 extern void file_kill(struct file *f);
1582 struct bio;
1583 extern void submit_bio(int, struct bio *);
1584 extern int bdev_read_only(struct block_device *);
1585 extern int set_blocksize(struct block_device *, int);
1586 extern int sb_set_blocksize(struct super_block *, int);
1587 extern int sb_min_blocksize(struct super_block *, int);
1588 
1589 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
1590 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
1591 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1592 extern int file_send_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1593 extern ssize_t generic_file_read(struct file *, char __user *, size_t, loff_t *);
1594 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
1595 extern ssize_t generic_file_write(struct file *, const char __user *, size_t, loff_t *);
1596 extern ssize_t generic_file_aio_read(struct kiocb *, char __user *, size_t, loff_t);
1597 extern ssize_t __generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t *);
1598 extern ssize_t generic_file_aio_write(struct kiocb *, const char __user *, size_t, loff_t);
1599 extern ssize_t generic_file_aio_write_nolock(struct kiocb *, const struct iovec *,
1600 		unsigned long, loff_t *);
1601 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
1602 		unsigned long *, loff_t, loff_t *, size_t, size_t);
1603 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
1604 		unsigned long, loff_t, loff_t *, size_t, ssize_t);
1605 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
1606 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
1607 ssize_t generic_file_write_nolock(struct file *file, const struct iovec *iov,
1608 				unsigned long nr_segs, loff_t *ppos);
1609 extern ssize_t generic_file_sendfile(struct file *, loff_t *, size_t, read_actor_t, void *);
1610 extern void do_generic_mapping_read(struct address_space *mapping,
1611 				    struct file_ra_state *, struct file *,
1612 				    loff_t *, read_descriptor_t *, read_actor_t);
1613 
1614 /* fs/splice.c */
1615 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
1616 		struct pipe_inode_info *, size_t, unsigned int);
1617 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
1618 		struct file *, loff_t *, size_t, unsigned int);
1619 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
1620 		struct file *out, loff_t *, size_t len, unsigned int flags);
1621 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
1622 		size_t len, unsigned int flags);
1623 
1624 extern void
1625 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
1626 extern ssize_t generic_file_readv(struct file *filp, const struct iovec *iov,
1627 	unsigned long nr_segs, loff_t *ppos);
1628 ssize_t generic_file_writev(struct file *filp, const struct iovec *iov,
1629 			unsigned long nr_segs, loff_t *ppos);
1630 extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
1631 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
1632 extern loff_t remote_llseek(struct file *file, loff_t offset, int origin);
1633 extern int generic_file_open(struct inode * inode, struct file * filp);
1634 extern int nonseekable_open(struct inode * inode, struct file * filp);
1635 
1636 #ifdef CONFIG_FS_XIP
1637 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
1638 			     loff_t *ppos);
1639 extern ssize_t xip_file_sendfile(struct file *in_file, loff_t *ppos,
1640 				 size_t count, read_actor_t actor,
1641 				 void *target);
1642 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
1643 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
1644 			      size_t len, loff_t *ppos);
1645 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
1646 #else
1647 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
1648 {
1649 	return 0;
1650 }
1651 #endif
1652 
1653 static inline void do_generic_file_read(struct file * filp, loff_t *ppos,
1654 					read_descriptor_t * desc,
1655 					read_actor_t actor)
1656 {
1657 	do_generic_mapping_read(filp->f_mapping,
1658 				&filp->f_ra,
1659 				filp,
1660 				ppos,
1661 				desc,
1662 				actor);
1663 }
1664 
1665 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
1666 	struct block_device *bdev, const struct iovec *iov, loff_t offset,
1667 	unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
1668 	int lock_type);
1669 
1670 enum {
1671 	DIO_LOCKING = 1, /* need locking between buffered and direct access */
1672 	DIO_NO_LOCKING,  /* bdev; no locking at all between buffered/direct */
1673 	DIO_OWN_LOCKING, /* filesystem locks buffered and direct internally */
1674 };
1675 
1676 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
1677 	struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1678 	loff_t offset, unsigned long nr_segs, get_block_t get_block,
1679 	dio_iodone_t end_io)
1680 {
1681 	return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1682 				nr_segs, get_block, end_io, DIO_LOCKING);
1683 }
1684 
1685 static inline ssize_t blockdev_direct_IO_no_locking(int rw, struct kiocb *iocb,
1686 	struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1687 	loff_t offset, unsigned long nr_segs, get_block_t get_block,
1688 	dio_iodone_t end_io)
1689 {
1690 	return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1691 				nr_segs, get_block, end_io, DIO_NO_LOCKING);
1692 }
1693 
1694 static inline ssize_t blockdev_direct_IO_own_locking(int rw, struct kiocb *iocb,
1695 	struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1696 	loff_t offset, unsigned long nr_segs, get_block_t get_block,
1697 	dio_iodone_t end_io)
1698 {
1699 	return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1700 				nr_segs, get_block, end_io, DIO_OWN_LOCKING);
1701 }
1702 
1703 extern const struct file_operations generic_ro_fops;
1704 
1705 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
1706 
1707 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
1708 extern int vfs_follow_link(struct nameidata *, const char *);
1709 extern int page_readlink(struct dentry *, char __user *, int);
1710 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
1711 extern void page_put_link(struct dentry *, struct nameidata *, void *);
1712 extern int __page_symlink(struct inode *inode, const char *symname, int len,
1713 		gfp_t gfp_mask);
1714 extern int page_symlink(struct inode *inode, const char *symname, int len);
1715 extern struct inode_operations page_symlink_inode_operations;
1716 extern int generic_readlink(struct dentry *, char __user *, int);
1717 extern void generic_fillattr(struct inode *, struct kstat *);
1718 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1719 void inode_add_bytes(struct inode *inode, loff_t bytes);
1720 void inode_sub_bytes(struct inode *inode, loff_t bytes);
1721 loff_t inode_get_bytes(struct inode *inode);
1722 void inode_set_bytes(struct inode *inode, loff_t bytes);
1723 
1724 extern int vfs_readdir(struct file *, filldir_t, void *);
1725 
1726 extern int vfs_stat(char __user *, struct kstat *);
1727 extern int vfs_lstat(char __user *, struct kstat *);
1728 extern int vfs_stat_fd(int dfd, char __user *, struct kstat *);
1729 extern int vfs_lstat_fd(int dfd, char __user *, struct kstat *);
1730 extern int vfs_fstat(unsigned int, struct kstat *);
1731 
1732 extern int vfs_ioctl(struct file *, unsigned int, unsigned int, unsigned long);
1733 
1734 extern struct file_system_type *get_fs_type(const char *name);
1735 extern struct super_block *get_super(struct block_device *);
1736 extern struct super_block *user_get_super(dev_t);
1737 extern void drop_super(struct super_block *sb);
1738 
1739 extern int dcache_dir_open(struct inode *, struct file *);
1740 extern int dcache_dir_close(struct inode *, struct file *);
1741 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
1742 extern int dcache_readdir(struct file *, void *, filldir_t);
1743 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1744 extern int simple_statfs(struct super_block *, struct kstatfs *);
1745 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
1746 extern int simple_unlink(struct inode *, struct dentry *);
1747 extern int simple_rmdir(struct inode *, struct dentry *);
1748 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1749 extern int simple_sync_file(struct file *, struct dentry *, int);
1750 extern int simple_empty(struct dentry *);
1751 extern int simple_readpage(struct file *file, struct page *page);
1752 extern int simple_prepare_write(struct file *file, struct page *page,
1753 			unsigned offset, unsigned to);
1754 extern int simple_commit_write(struct file *file, struct page *page,
1755 				unsigned offset, unsigned to);
1756 
1757 extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
1758 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
1759 extern const struct file_operations simple_dir_operations;
1760 extern struct inode_operations simple_dir_inode_operations;
1761 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
1762 struct dentry *d_alloc_name(struct dentry *, const char *);
1763 extern int simple_fill_super(struct super_block *, int, struct tree_descr *);
1764 extern int simple_pin_fs(char *name, struct vfsmount **mount, int *count);
1765 extern void simple_release_fs(struct vfsmount **mount, int *count);
1766 
1767 extern ssize_t simple_read_from_buffer(void __user *, size_t, loff_t *, const void *, size_t);
1768 
1769 #ifdef CONFIG_MIGRATION
1770 extern int buffer_migrate_page(struct page *, struct page *);
1771 #else
1772 #define buffer_migrate_page NULL
1773 #endif
1774 
1775 extern int inode_change_ok(struct inode *, struct iattr *);
1776 extern int __must_check inode_setattr(struct inode *, struct iattr *);
1777 
1778 extern void file_update_time(struct file *file);
1779 
1780 static inline ino_t parent_ino(struct dentry *dentry)
1781 {
1782 	ino_t res;
1783 
1784 	spin_lock(&dentry->d_lock);
1785 	res = dentry->d_parent->d_inode->i_ino;
1786 	spin_unlock(&dentry->d_lock);
1787 	return res;
1788 }
1789 
1790 /* kernel/fork.c */
1791 extern int unshare_files(void);
1792 
1793 /* Transaction based IO helpers */
1794 
1795 /*
1796  * An argresp is stored in an allocated page and holds the
1797  * size of the argument or response, along with its content
1798  */
1799 struct simple_transaction_argresp {
1800 	ssize_t size;
1801 	char data[0];
1802 };
1803 
1804 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
1805 
1806 char *simple_transaction_get(struct file *file, const char __user *buf,
1807 				size_t size);
1808 ssize_t simple_transaction_read(struct file *file, char __user *buf,
1809 				size_t size, loff_t *pos);
1810 int simple_transaction_release(struct inode *inode, struct file *file);
1811 
1812 static inline void simple_transaction_set(struct file *file, size_t n)
1813 {
1814 	struct simple_transaction_argresp *ar = file->private_data;
1815 
1816 	BUG_ON(n > SIMPLE_TRANSACTION_LIMIT);
1817 
1818 	/*
1819 	 * The barrier ensures that ar->size will really remain zero until
1820 	 * ar->data is ready for reading.
1821 	 */
1822 	smp_mb();
1823 	ar->size = n;
1824 }
1825 
1826 /*
1827  * simple attribute files
1828  *
1829  * These attributes behave similar to those in sysfs:
1830  *
1831  * Writing to an attribute immediately sets a value, an open file can be
1832  * written to multiple times.
1833  *
1834  * Reading from an attribute creates a buffer from the value that might get
1835  * read with multiple read calls. When the attribute has been read
1836  * completely, no further read calls are possible until the file is opened
1837  * again.
1838  *
1839  * All attributes contain a text representation of a numeric value
1840  * that are accessed with the get() and set() functions.
1841  */
1842 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt)		\
1843 static int __fops ## _open(struct inode *inode, struct file *file)	\
1844 {									\
1845 	__simple_attr_check_format(__fmt, 0ull);			\
1846 	return simple_attr_open(inode, file, __get, __set, __fmt);	\
1847 }									\
1848 static struct file_operations __fops = {				\
1849 	.owner	 = THIS_MODULE,						\
1850 	.open	 = __fops ## _open,					\
1851 	.release = simple_attr_close,					\
1852 	.read	 = simple_attr_read,					\
1853 	.write	 = simple_attr_write,					\
1854 };
1855 
1856 static inline void __attribute__((format(printf, 1, 2)))
1857 __simple_attr_check_format(const char *fmt, ...)
1858 {
1859 	/* don't do anything, just let the compiler check the arguments; */
1860 }
1861 
1862 int simple_attr_open(struct inode *inode, struct file *file,
1863 		     u64 (*get)(void *), void (*set)(void *, u64),
1864 		     const char *fmt);
1865 int simple_attr_close(struct inode *inode, struct file *file);
1866 ssize_t simple_attr_read(struct file *file, char __user *buf,
1867 			 size_t len, loff_t *ppos);
1868 ssize_t simple_attr_write(struct file *file, const char __user *buf,
1869 			  size_t len, loff_t *ppos);
1870 
1871 
1872 #ifdef CONFIG_SECURITY
1873 static inline char *alloc_secdata(void)
1874 {
1875 	return (char *)get_zeroed_page(GFP_KERNEL);
1876 }
1877 
1878 static inline void free_secdata(void *secdata)
1879 {
1880 	free_page((unsigned long)secdata);
1881 }
1882 #else
1883 static inline char *alloc_secdata(void)
1884 {
1885 	return (char *)1;
1886 }
1887 
1888 static inline void free_secdata(void *secdata)
1889 { }
1890 #endif	/* CONFIG_SECURITY */
1891 
1892 #endif /* __KERNEL__ */
1893 #endif /* _LINUX_FS_H */
1894