xref: /linux-6.15/include/linux/fs.h (revision de2fe5e0)
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/config.h>
10 #include <linux/limits.h>
11 #include <linux/ioctl.h>
12 
13 /*
14  * It's silly to have NR_OPEN bigger than NR_FILE, but you can change
15  * the file limit at runtime and only root can increase the per-process
16  * nr_file rlimit, so it's safe to set up a ridiculously high absolute
17  * upper limit on files-per-process.
18  *
19  * Some programs (notably those using select()) may have to be
20  * recompiled to take full advantage of the new limits..
21  */
22 
23 /* Fixed constants first: */
24 #undef NR_OPEN
25 #define NR_OPEN (1024*1024)	/* Absolute upper limit on fd num */
26 #define INR_OPEN 1024		/* Initial setting for nfile rlimits */
27 
28 #define BLOCK_SIZE_BITS 10
29 #define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
30 
31 /* And dynamically-tunable limits and defaults: */
32 struct files_stat_struct {
33 	int nr_files;		/* read only */
34 	int nr_free_files;	/* read only */
35 	int max_files;		/* tunable */
36 };
37 extern struct files_stat_struct files_stat;
38 extern int get_max_files(void);
39 
40 struct inodes_stat_t {
41 	int nr_inodes;
42 	int nr_unused;
43 	int dummy[5];
44 };
45 extern struct inodes_stat_t inodes_stat;
46 
47 extern int leases_enable, lease_break_time;
48 
49 #ifdef CONFIG_DNOTIFY
50 extern int dir_notify_enable;
51 #endif
52 
53 #define NR_FILE  8192	/* this can well be larger on a larger system */
54 
55 #define MAY_EXEC 1
56 #define MAY_WRITE 2
57 #define MAY_READ 4
58 #define MAY_APPEND 8
59 
60 #define FMODE_READ 1
61 #define FMODE_WRITE 2
62 
63 /* Internal kernel extensions */
64 #define FMODE_LSEEK	4
65 #define FMODE_PREAD	8
66 #define FMODE_PWRITE	FMODE_PREAD	/* These go hand in hand */
67 
68 /* File is being opened for execution. Primary users of this flag are
69    distributed filesystems that can use it to achieve correct ETXTBUSY
70    behavior for cross-node execution/opening_for_writing of files */
71 #define FMODE_EXEC	16
72 
73 #define RW_MASK		1
74 #define RWA_MASK	2
75 #define READ 0
76 #define WRITE 1
77 #define READA 2		/* read-ahead  - don't block if no resources */
78 #define SWRITE 3	/* for ll_rw_block() - wait for buffer lock */
79 #define SPECIAL 4	/* For non-blockdevice requests in request queue */
80 #define READ_SYNC	(READ | (1 << BIO_RW_SYNC))
81 #define WRITE_SYNC	(WRITE | (1 << BIO_RW_SYNC))
82 #define WRITE_BARRIER	((1 << BIO_RW) | (1 << BIO_RW_BARRIER))
83 
84 #define SEL_IN		1
85 #define SEL_OUT		2
86 #define SEL_EX		4
87 
88 /* public flags for file_system_type */
89 #define FS_REQUIRES_DEV 1
90 #define FS_BINARY_MOUNTDATA 2
91 #define FS_REVAL_DOT	16384	/* Check the paths ".", ".." for staleness */
92 #define FS_ODD_RENAME	32768	/* Temporary stuff; will go away as soon
93 				  * as nfs_rename() will be cleaned up
94 				  */
95 /*
96  * These are the fs-independent mount-flags: up to 32 flags are supported
97  */
98 #define MS_RDONLY	 1	/* Mount read-only */
99 #define MS_NOSUID	 2	/* Ignore suid and sgid bits */
100 #define MS_NODEV	 4	/* Disallow access to device special files */
101 #define MS_NOEXEC	 8	/* Disallow program execution */
102 #define MS_SYNCHRONOUS	16	/* Writes are synced at once */
103 #define MS_REMOUNT	32	/* Alter flags of a mounted FS */
104 #define MS_MANDLOCK	64	/* Allow mandatory locks on an FS */
105 #define MS_DIRSYNC	128	/* Directory modifications are synchronous */
106 #define MS_NOATIME	1024	/* Do not update access times. */
107 #define MS_NODIRATIME	2048	/* Do not update directory access times */
108 #define MS_BIND		4096
109 #define MS_MOVE		8192
110 #define MS_REC		16384
111 #define MS_VERBOSE	32768	/* War is peace. Verbosity is silence.
112 				   MS_VERBOSE is deprecated. */
113 #define MS_SILENT	32768
114 #define MS_POSIXACL	(1<<16)	/* VFS does not apply the umask */
115 #define MS_UNBINDABLE	(1<<17)	/* change to unbindable */
116 #define MS_PRIVATE	(1<<18)	/* change to private */
117 #define MS_SLAVE	(1<<19)	/* change to slave */
118 #define MS_SHARED	(1<<20)	/* change to shared */
119 #define MS_ACTIVE	(1<<30)
120 #define MS_NOUSER	(1<<31)
121 
122 /*
123  * Superblock flags that can be altered by MS_REMOUNT
124  */
125 #define MS_RMT_MASK	(MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK)
126 
127 /*
128  * Old magic mount flag and mask
129  */
130 #define MS_MGC_VAL 0xC0ED0000
131 #define MS_MGC_MSK 0xffff0000
132 
133 /* Inode flags - they have nothing to superblock flags now */
134 
135 #define S_SYNC		1	/* Writes are synced at once */
136 #define S_NOATIME	2	/* Do not update access times */
137 #define S_APPEND	4	/* Append-only file */
138 #define S_IMMUTABLE	8	/* Immutable file */
139 #define S_DEAD		16	/* removed, but still open directory */
140 #define S_NOQUOTA	32	/* Inode is not counted to quota */
141 #define S_DIRSYNC	64	/* Directory modifications are synchronous */
142 #define S_NOCMTIME	128	/* Do not update file c/mtime */
143 #define S_SWAPFILE	256	/* Do not truncate: swapon got its bmaps */
144 #define S_PRIVATE	512	/* Inode is fs-internal */
145 
146 /*
147  * Note that nosuid etc flags are inode-specific: setting some file-system
148  * flags just means all the inodes inherit those flags by default. It might be
149  * possible to override it selectively if you really wanted to with some
150  * ioctl() that is not currently implemented.
151  *
152  * Exception: MS_RDONLY is always applied to the entire file system.
153  *
154  * Unfortunately, it is possible to change a filesystems flags with it mounted
155  * with files in use.  This means that all of the inodes will not have their
156  * i_flags updated.  Hence, i_flags no longer inherit the superblock mount
157  * flags, so these have to be checked separately. -- [email protected]
158  */
159 #define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
160 
161 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
162 #define IS_SYNC(inode)		(__IS_FLG(inode, MS_SYNCHRONOUS) || \
163 					((inode)->i_flags & S_SYNC))
164 #define IS_DIRSYNC(inode)	(__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
165 					((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
166 #define IS_MANDLOCK(inode)	__IS_FLG(inode, MS_MANDLOCK)
167 
168 #define IS_NOQUOTA(inode)	((inode)->i_flags & S_NOQUOTA)
169 #define IS_APPEND(inode)	((inode)->i_flags & S_APPEND)
170 #define IS_IMMUTABLE(inode)	((inode)->i_flags & S_IMMUTABLE)
171 #define IS_POSIXACL(inode)	__IS_FLG(inode, MS_POSIXACL)
172 
173 #define IS_DEADDIR(inode)	((inode)->i_flags & S_DEAD)
174 #define IS_NOCMTIME(inode)	((inode)->i_flags & S_NOCMTIME)
175 #define IS_SWAPFILE(inode)	((inode)->i_flags & S_SWAPFILE)
176 #define IS_PRIVATE(inode)	((inode)->i_flags & S_PRIVATE)
177 
178 /* the read-only stuff doesn't really belong here, but any other place is
179    probably as bad and I don't want to create yet another include file. */
180 
181 #define BLKROSET   _IO(0x12,93)	/* set device read-only (0 = read-write) */
182 #define BLKROGET   _IO(0x12,94)	/* get read-only status (0 = read_write) */
183 #define BLKRRPART  _IO(0x12,95)	/* re-read partition table */
184 #define BLKGETSIZE _IO(0x12,96)	/* return device size /512 (long *arg) */
185 #define BLKFLSBUF  _IO(0x12,97)	/* flush buffer cache */
186 #define BLKRASET   _IO(0x12,98)	/* set read ahead for block device */
187 #define BLKRAGET   _IO(0x12,99)	/* get current read ahead setting */
188 #define BLKFRASET  _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
189 #define BLKFRAGET  _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
190 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
191 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
192 #define BLKSSZGET  _IO(0x12,104)/* get block device sector size */
193 #if 0
194 #define BLKPG      _IO(0x12,105)/* See blkpg.h */
195 
196 /* Some people are morons.  Do not use sizeof! */
197 
198 #define BLKELVGET  _IOR(0x12,106,size_t)/* elevator get */
199 #define BLKELVSET  _IOW(0x12,107,size_t)/* elevator set */
200 /* This was here just to show that the number is taken -
201    probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
202 #endif
203 /* A jump here: 108-111 have been used for various private purposes. */
204 #define BLKBSZGET  _IOR(0x12,112,size_t)
205 #define BLKBSZSET  _IOW(0x12,113,size_t)
206 #define BLKGETSIZE64 _IOR(0x12,114,size_t)	/* return device size in bytes (u64 *arg) */
207 #define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
208 #define BLKTRACESTART _IO(0x12,116)
209 #define BLKTRACESTOP _IO(0x12,117)
210 #define BLKTRACETEARDOWN _IO(0x12,118)
211 
212 #define BMAP_IOCTL 1		/* obsolete - kept for compatibility */
213 #define FIBMAP	   _IO(0x00,1)	/* bmap access */
214 #define FIGETBSZ   _IO(0x00,2)	/* get the block size used for bmap */
215 
216 #ifdef __KERNEL__
217 
218 #include <linux/linkage.h>
219 #include <linux/wait.h>
220 #include <linux/types.h>
221 #include <linux/kdev_t.h>
222 #include <linux/dcache.h>
223 #include <linux/stat.h>
224 #include <linux/cache.h>
225 #include <linux/kobject.h>
226 #include <linux/list.h>
227 #include <linux/radix-tree.h>
228 #include <linux/prio_tree.h>
229 #include <linux/init.h>
230 #include <linux/sched.h>
231 #include <linux/mutex.h>
232 
233 #include <asm/atomic.h>
234 #include <asm/semaphore.h>
235 #include <asm/byteorder.h>
236 
237 struct hd_geometry;
238 struct iovec;
239 struct nameidata;
240 struct kiocb;
241 struct pipe_inode_info;
242 struct poll_table_struct;
243 struct kstatfs;
244 struct vm_area_struct;
245 struct vfsmount;
246 
247 extern void __init inode_init(unsigned long);
248 extern void __init inode_init_early(void);
249 extern void __init mnt_init(unsigned long);
250 extern void __init files_init(unsigned long);
251 
252 struct buffer_head;
253 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
254 			struct buffer_head *bh_result, int create);
255 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
256 			ssize_t bytes, void *private);
257 
258 /*
259  * Attribute flags.  These should be or-ed together to figure out what
260  * has been changed!
261  */
262 #define ATTR_MODE	1
263 #define ATTR_UID	2
264 #define ATTR_GID	4
265 #define ATTR_SIZE	8
266 #define ATTR_ATIME	16
267 #define ATTR_MTIME	32
268 #define ATTR_CTIME	64
269 #define ATTR_ATIME_SET	128
270 #define ATTR_MTIME_SET	256
271 #define ATTR_FORCE	512	/* Not a change, but a change it */
272 #define ATTR_ATTR_FLAG	1024
273 #define ATTR_KILL_SUID	2048
274 #define ATTR_KILL_SGID	4096
275 #define ATTR_FILE	8192
276 
277 /*
278  * This is the Inode Attributes structure, used for notify_change().  It
279  * uses the above definitions as flags, to know which values have changed.
280  * Also, in this manner, a Filesystem can look at only the values it cares
281  * about.  Basically, these are the attributes that the VFS layer can
282  * request to change from the FS layer.
283  *
284  * Derek Atkins <[email protected]> 94-10-20
285  */
286 struct iattr {
287 	unsigned int	ia_valid;
288 	umode_t		ia_mode;
289 	uid_t		ia_uid;
290 	gid_t		ia_gid;
291 	loff_t		ia_size;
292 	struct timespec	ia_atime;
293 	struct timespec	ia_mtime;
294 	struct timespec	ia_ctime;
295 
296 	/*
297 	 * Not an attribute, but an auxilary info for filesystems wanting to
298 	 * implement an ftruncate() like method.  NOTE: filesystem should
299 	 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
300 	 */
301 	struct file	*ia_file;
302 };
303 
304 /*
305  * Includes for diskquotas.
306  */
307 #include <linux/quota.h>
308 
309 /**
310  * enum positive_aop_returns - aop return codes with specific semantics
311  *
312  * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
313  * 			    completed, that the page is still locked, and
314  * 			    should be considered active.  The VM uses this hint
315  * 			    to return the page to the active list -- it won't
316  * 			    be a candidate for writeback again in the near
317  * 			    future.  Other callers must be careful to unlock
318  * 			    the page if they get this return.  Returned by
319  * 			    writepage();
320  *
321  * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
322  *  			unlocked it and the page might have been truncated.
323  *  			The caller should back up to acquiring a new page and
324  *  			trying again.  The aop will be taking reasonable
325  *  			precautions not to livelock.  If the caller held a page
326  *  			reference, it should drop it before retrying.  Returned
327  *  			by readpage(), prepare_write(), and commit_write().
328  *
329  * address_space_operation functions return these large constants to indicate
330  * special semantics to the caller.  These are much larger than the bytes in a
331  * page to allow for functions that return the number of bytes operated on in a
332  * given page.
333  */
334 
335 enum positive_aop_returns {
336 	AOP_WRITEPAGE_ACTIVATE	= 0x80000,
337 	AOP_TRUNCATED_PAGE	= 0x80001,
338 };
339 
340 /*
341  * oh the beauties of C type declarations.
342  */
343 struct page;
344 struct address_space;
345 struct writeback_control;
346 
347 struct address_space_operations {
348 	int (*writepage)(struct page *page, struct writeback_control *wbc);
349 	int (*readpage)(struct file *, struct page *);
350 	void (*sync_page)(struct page *);
351 
352 	/* Write back some dirty pages from this mapping. */
353 	int (*writepages)(struct address_space *, struct writeback_control *);
354 
355 	/* Set a page dirty.  Return true if this dirtied it */
356 	int (*set_page_dirty)(struct page *page);
357 
358 	int (*readpages)(struct file *filp, struct address_space *mapping,
359 			struct list_head *pages, unsigned nr_pages);
360 
361 	/*
362 	 * ext3 requires that a successful prepare_write() call be followed
363 	 * by a commit_write() call - they must be balanced
364 	 */
365 	int (*prepare_write)(struct file *, struct page *, unsigned, unsigned);
366 	int (*commit_write)(struct file *, struct page *, unsigned, unsigned);
367 	/* Unfortunately this kludge is needed for FIBMAP. Don't use it */
368 	sector_t (*bmap)(struct address_space *, sector_t);
369 	void (*invalidatepage) (struct page *, unsigned long);
370 	int (*releasepage) (struct page *, gfp_t);
371 	ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
372 			loff_t offset, unsigned long nr_segs);
373 	struct page* (*get_xip_page)(struct address_space *, sector_t,
374 			int);
375 	/* migrate the contents of a page to the specified target */
376 	int (*migratepage) (struct page *, struct page *);
377 };
378 
379 struct backing_dev_info;
380 struct address_space {
381 	struct inode		*host;		/* owner: inode, block_device */
382 	struct radix_tree_root	page_tree;	/* radix tree of all pages */
383 	rwlock_t		tree_lock;	/* and rwlock protecting it */
384 	unsigned int		i_mmap_writable;/* count VM_SHARED mappings */
385 	struct prio_tree_root	i_mmap;		/* tree of private and shared mappings */
386 	struct list_head	i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
387 	spinlock_t		i_mmap_lock;	/* protect tree, count, list */
388 	unsigned int		truncate_count;	/* Cover race condition with truncate */
389 	unsigned long		nrpages;	/* number of total pages */
390 	pgoff_t			writeback_index;/* writeback starts here */
391 	struct address_space_operations *a_ops;	/* methods */
392 	unsigned long		flags;		/* error bits/gfp mask */
393 	struct backing_dev_info *backing_dev_info; /* device readahead, etc */
394 	spinlock_t		private_lock;	/* for use by the address_space */
395 	struct list_head	private_list;	/* ditto */
396 	struct address_space	*assoc_mapping;	/* ditto */
397 } __attribute__((aligned(sizeof(long))));
398 	/*
399 	 * On most architectures that alignment is already the case; but
400 	 * must be enforced here for CRIS, to let the least signficant bit
401 	 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
402 	 */
403 
404 struct block_device {
405 	dev_t			bd_dev;  /* not a kdev_t - it's a search key */
406 	struct inode *		bd_inode;	/* will die */
407 	int			bd_openers;
408 	struct mutex		bd_mutex;	/* open/close mutex */
409 	struct mutex		bd_mount_mutex;	/* mount mutex */
410 	struct list_head	bd_inodes;
411 	void *			bd_holder;
412 	int			bd_holders;
413 #ifdef CONFIG_SYSFS
414 	struct list_head	bd_holder_list;
415 #endif
416 	struct block_device *	bd_contains;
417 	unsigned		bd_block_size;
418 	struct hd_struct *	bd_part;
419 	/* number of times partitions within this device have been opened. */
420 	unsigned		bd_part_count;
421 	int			bd_invalidated;
422 	struct gendisk *	bd_disk;
423 	struct list_head	bd_list;
424 	struct backing_dev_info *bd_inode_backing_dev_info;
425 	/*
426 	 * Private data.  You must have bd_claim'ed the block_device
427 	 * to use this.  NOTE:  bd_claim allows an owner to claim
428 	 * the same device multiple times, the owner must take special
429 	 * care to not mess up bd_private for that case.
430 	 */
431 	unsigned long		bd_private;
432 };
433 
434 /*
435  * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
436  * radix trees
437  */
438 #define PAGECACHE_TAG_DIRTY	0
439 #define PAGECACHE_TAG_WRITEBACK	1
440 
441 int mapping_tagged(struct address_space *mapping, int tag);
442 
443 /*
444  * Might pages of this file be mapped into userspace?
445  */
446 static inline int mapping_mapped(struct address_space *mapping)
447 {
448 	return	!prio_tree_empty(&mapping->i_mmap) ||
449 		!list_empty(&mapping->i_mmap_nonlinear);
450 }
451 
452 /*
453  * Might pages of this file have been modified in userspace?
454  * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
455  * marks vma as VM_SHARED if it is shared, and the file was opened for
456  * writing i.e. vma may be mprotected writable even if now readonly.
457  */
458 static inline int mapping_writably_mapped(struct address_space *mapping)
459 {
460 	return mapping->i_mmap_writable != 0;
461 }
462 
463 /*
464  * Use sequence counter to get consistent i_size on 32-bit processors.
465  */
466 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
467 #include <linux/seqlock.h>
468 #define __NEED_I_SIZE_ORDERED
469 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
470 #else
471 #define i_size_ordered_init(inode) do { } while (0)
472 #endif
473 
474 struct inode {
475 	struct hlist_node	i_hash;
476 	struct list_head	i_list;
477 	struct list_head	i_sb_list;
478 	struct list_head	i_dentry;
479 	unsigned long		i_ino;
480 	atomic_t		i_count;
481 	umode_t			i_mode;
482 	unsigned int		i_nlink;
483 	uid_t			i_uid;
484 	gid_t			i_gid;
485 	dev_t			i_rdev;
486 	loff_t			i_size;
487 	struct timespec		i_atime;
488 	struct timespec		i_mtime;
489 	struct timespec		i_ctime;
490 	unsigned int		i_blkbits;
491 	unsigned long		i_blksize;
492 	unsigned long		i_version;
493 	blkcnt_t		i_blocks;
494 	unsigned short          i_bytes;
495 	spinlock_t		i_lock;	/* i_blocks, i_bytes, maybe i_size */
496 	struct mutex		i_mutex;
497 	struct rw_semaphore	i_alloc_sem;
498 	struct inode_operations	*i_op;
499 	const struct file_operations	*i_fop;	/* former ->i_op->default_file_ops */
500 	struct super_block	*i_sb;
501 	struct file_lock	*i_flock;
502 	struct address_space	*i_mapping;
503 	struct address_space	i_data;
504 #ifdef CONFIG_QUOTA
505 	struct dquot		*i_dquot[MAXQUOTAS];
506 #endif
507 	/* These three should probably be a union */
508 	struct list_head	i_devices;
509 	struct pipe_inode_info	*i_pipe;
510 	struct block_device	*i_bdev;
511 	struct cdev		*i_cdev;
512 	int			i_cindex;
513 
514 	__u32			i_generation;
515 
516 #ifdef CONFIG_DNOTIFY
517 	unsigned long		i_dnotify_mask; /* Directory notify events */
518 	struct dnotify_struct	*i_dnotify; /* for directory notifications */
519 #endif
520 
521 #ifdef CONFIG_INOTIFY
522 	struct list_head	inotify_watches; /* watches on this inode */
523 	struct mutex		inotify_mutex;	/* protects the watches list */
524 #endif
525 
526 	unsigned long		i_state;
527 	unsigned long		dirtied_when;	/* jiffies of first dirtying */
528 
529 	unsigned int		i_flags;
530 
531 	atomic_t		i_writecount;
532 	void			*i_security;
533 	union {
534 		void		*generic_ip;
535 	} u;
536 #ifdef __NEED_I_SIZE_ORDERED
537 	seqcount_t		i_size_seqcount;
538 #endif
539 };
540 
541 /*
542  * NOTE: in a 32bit arch with a preemptable kernel and
543  * an UP compile the i_size_read/write must be atomic
544  * with respect to the local cpu (unlike with preempt disabled),
545  * but they don't need to be atomic with respect to other cpus like in
546  * true SMP (so they need either to either locally disable irq around
547  * the read or for example on x86 they can be still implemented as a
548  * cmpxchg8b without the need of the lock prefix). For SMP compiles
549  * and 64bit archs it makes no difference if preempt is enabled or not.
550  */
551 static inline loff_t i_size_read(struct inode *inode)
552 {
553 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
554 	loff_t i_size;
555 	unsigned int seq;
556 
557 	do {
558 		seq = read_seqcount_begin(&inode->i_size_seqcount);
559 		i_size = inode->i_size;
560 	} while (read_seqcount_retry(&inode->i_size_seqcount, seq));
561 	return i_size;
562 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
563 	loff_t i_size;
564 
565 	preempt_disable();
566 	i_size = inode->i_size;
567 	preempt_enable();
568 	return i_size;
569 #else
570 	return inode->i_size;
571 #endif
572 }
573 
574 
575 static inline void i_size_write(struct inode *inode, loff_t i_size)
576 {
577 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
578 	write_seqcount_begin(&inode->i_size_seqcount);
579 	inode->i_size = i_size;
580 	write_seqcount_end(&inode->i_size_seqcount);
581 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
582 	preempt_disable();
583 	inode->i_size = i_size;
584 	preempt_enable();
585 #else
586 	inode->i_size = i_size;
587 #endif
588 }
589 
590 static inline unsigned iminor(struct inode *inode)
591 {
592 	return MINOR(inode->i_rdev);
593 }
594 
595 static inline unsigned imajor(struct inode *inode)
596 {
597 	return MAJOR(inode->i_rdev);
598 }
599 
600 extern struct block_device *I_BDEV(struct inode *inode);
601 
602 struct fown_struct {
603 	rwlock_t lock;          /* protects pid, uid, euid fields */
604 	int pid;		/* pid or -pgrp where SIGIO should be sent */
605 	uid_t uid, euid;	/* uid/euid of process setting the owner */
606 	void *security;
607 	int signum;		/* posix.1b rt signal to be delivered on IO */
608 };
609 
610 /*
611  * Track a single file's readahead state
612  */
613 struct file_ra_state {
614 	unsigned long start;		/* Current window */
615 	unsigned long size;
616 	unsigned long flags;		/* ra flags RA_FLAG_xxx*/
617 	unsigned long cache_hit;	/* cache hit count*/
618 	unsigned long prev_page;	/* Cache last read() position */
619 	unsigned long ahead_start;	/* Ahead window */
620 	unsigned long ahead_size;
621 	unsigned long ra_pages;		/* Maximum readahead window */
622 	unsigned long mmap_hit;		/* Cache hit stat for mmap accesses */
623 	unsigned long mmap_miss;	/* Cache miss stat for mmap accesses */
624 };
625 #define RA_FLAG_MISS 0x01	/* a cache miss occured against this file */
626 #define RA_FLAG_INCACHE 0x02	/* file is already in cache */
627 
628 struct file {
629 	/*
630 	 * fu_list becomes invalid after file_free is called and queued via
631 	 * fu_rcuhead for RCU freeing
632 	 */
633 	union {
634 		struct list_head	fu_list;
635 		struct rcu_head 	fu_rcuhead;
636 	} f_u;
637 	struct dentry		*f_dentry;
638 	struct vfsmount         *f_vfsmnt;
639 	const struct file_operations	*f_op;
640 	atomic_t		f_count;
641 	unsigned int 		f_flags;
642 	mode_t			f_mode;
643 	loff_t			f_pos;
644 	struct fown_struct	f_owner;
645 	unsigned int		f_uid, f_gid;
646 	struct file_ra_state	f_ra;
647 
648 	unsigned long		f_version;
649 	void			*f_security;
650 
651 	/* needed for tty driver, and maybe others */
652 	void			*private_data;
653 
654 #ifdef CONFIG_EPOLL
655 	/* Used by fs/eventpoll.c to link all the hooks to this file */
656 	struct list_head	f_ep_links;
657 	spinlock_t		f_ep_lock;
658 #endif /* #ifdef CONFIG_EPOLL */
659 	struct address_space	*f_mapping;
660 };
661 extern spinlock_t files_lock;
662 #define file_list_lock() spin_lock(&files_lock);
663 #define file_list_unlock() spin_unlock(&files_lock);
664 
665 #define get_file(x)	atomic_inc(&(x)->f_count)
666 #define file_count(x)	atomic_read(&(x)->f_count)
667 
668 #define	MAX_NON_LFS	((1UL<<31) - 1)
669 
670 /* Page cache limit. The filesystems should put that into their s_maxbytes
671    limits, otherwise bad things can happen in VM. */
672 #if BITS_PER_LONG==32
673 #define MAX_LFS_FILESIZE	(((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
674 #elif BITS_PER_LONG==64
675 #define MAX_LFS_FILESIZE 	0x7fffffffffffffffUL
676 #endif
677 
678 #define FL_POSIX	1
679 #define FL_FLOCK	2
680 #define FL_ACCESS	8	/* not trying to lock, just looking */
681 #define FL_LEASE	32	/* lease held on this file */
682 #define FL_SLEEP	128	/* A blocking lock */
683 
684 /*
685  * The POSIX file lock owner is determined by
686  * the "struct files_struct" in the thread group
687  * (or NULL for no owner - BSD locks).
688  *
689  * Lockd stuffs a "host" pointer into this.
690  */
691 typedef struct files_struct *fl_owner_t;
692 
693 struct file_lock_operations {
694 	void (*fl_insert)(struct file_lock *);	/* lock insertion callback */
695 	void (*fl_remove)(struct file_lock *);	/* lock removal callback */
696 	void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
697 	void (*fl_release_private)(struct file_lock *);
698 };
699 
700 struct lock_manager_operations {
701 	int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
702 	void (*fl_notify)(struct file_lock *);	/* unblock callback */
703 	void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
704 	void (*fl_release_private)(struct file_lock *);
705 	void (*fl_break)(struct file_lock *);
706 	int (*fl_mylease)(struct file_lock *, struct file_lock *);
707 	int (*fl_change)(struct file_lock **, int);
708 };
709 
710 /* that will die - we need it for nfs_lock_info */
711 #include <linux/nfs_fs_i.h>
712 
713 struct file_lock {
714 	struct file_lock *fl_next;	/* singly linked list for this inode  */
715 	struct list_head fl_link;	/* doubly linked list of all locks */
716 	struct list_head fl_block;	/* circular list of blocked processes */
717 	fl_owner_t fl_owner;
718 	unsigned int fl_pid;
719 	wait_queue_head_t fl_wait;
720 	struct file *fl_file;
721 	unsigned char fl_flags;
722 	unsigned char fl_type;
723 	loff_t fl_start;
724 	loff_t fl_end;
725 
726 	struct fasync_struct *	fl_fasync; /* for lease break notifications */
727 	unsigned long fl_break_time;	/* for nonblocking lease breaks */
728 
729 	struct file_lock_operations *fl_ops;	/* Callbacks for filesystems */
730 	struct lock_manager_operations *fl_lmops;	/* Callbacks for lockmanagers */
731 	union {
732 		struct nfs_lock_info	nfs_fl;
733 		struct nfs4_lock_info	nfs4_fl;
734 	} fl_u;
735 };
736 
737 /* The following constant reflects the upper bound of the file/locking space */
738 #ifndef OFFSET_MAX
739 #define INT_LIMIT(x)	(~((x)1 << (sizeof(x)*8 - 1)))
740 #define OFFSET_MAX	INT_LIMIT(loff_t)
741 #define OFFT_OFFSET_MAX	INT_LIMIT(off_t)
742 #endif
743 
744 #include <linux/fcntl.h>
745 
746 extern int fcntl_getlk(struct file *, struct flock __user *);
747 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
748 			struct flock __user *);
749 
750 #if BITS_PER_LONG == 32
751 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
752 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
753 			struct flock64 __user *);
754 #endif
755 
756 extern void send_sigio(struct fown_struct *fown, int fd, int band);
757 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
758 extern int fcntl_getlease(struct file *filp);
759 
760 /* fs/locks.c */
761 extern void locks_init_lock(struct file_lock *);
762 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
763 extern void locks_remove_posix(struct file *, fl_owner_t);
764 extern void locks_remove_flock(struct file *);
765 extern int posix_test_lock(struct file *, struct file_lock *, struct file_lock *);
766 extern int posix_lock_file_conf(struct file *, struct file_lock *, struct file_lock *);
767 extern int posix_lock_file(struct file *, struct file_lock *);
768 extern int posix_lock_file_wait(struct file *, struct file_lock *);
769 extern int posix_unblock_lock(struct file *, struct file_lock *);
770 extern int posix_locks_deadlock(struct file_lock *, struct file_lock *);
771 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
772 extern int __break_lease(struct inode *inode, unsigned int flags);
773 extern void lease_get_mtime(struct inode *, struct timespec *time);
774 extern int setlease(struct file *, long, struct file_lock **);
775 extern int lease_modify(struct file_lock **, int);
776 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
777 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
778 extern void steal_locks(fl_owner_t from);
779 
780 struct fasync_struct {
781 	int	magic;
782 	int	fa_fd;
783 	struct	fasync_struct	*fa_next; /* singly linked list */
784 	struct	file 		*fa_file;
785 };
786 
787 #define FASYNC_MAGIC 0x4601
788 
789 /* SMP safe fasync helpers: */
790 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
791 /* can be called from interrupts */
792 extern void kill_fasync(struct fasync_struct **, int, int);
793 /* only for net: no internal synchronization */
794 extern void __kill_fasync(struct fasync_struct *, int, int);
795 
796 extern int f_setown(struct file *filp, unsigned long arg, int force);
797 extern void f_delown(struct file *filp);
798 extern int send_sigurg(struct fown_struct *fown);
799 
800 /*
801  *	Umount options
802  */
803 
804 #define MNT_FORCE	0x00000001	/* Attempt to forcibily umount */
805 #define MNT_DETACH	0x00000002	/* Just detach from the tree */
806 #define MNT_EXPIRE	0x00000004	/* Mark for expiry */
807 
808 extern struct list_head super_blocks;
809 extern spinlock_t sb_lock;
810 
811 #define sb_entry(list)	list_entry((list), struct super_block, s_list)
812 #define S_BIAS (1<<30)
813 struct super_block {
814 	struct list_head	s_list;		/* Keep this first */
815 	dev_t			s_dev;		/* search index; _not_ kdev_t */
816 	unsigned long		s_blocksize;
817 	unsigned char		s_blocksize_bits;
818 	unsigned char		s_dirt;
819 	unsigned long long	s_maxbytes;	/* Max file size */
820 	struct file_system_type	*s_type;
821 	struct super_operations	*s_op;
822 	struct dquot_operations	*dq_op;
823  	struct quotactl_ops	*s_qcop;
824 	struct export_operations *s_export_op;
825 	unsigned long		s_flags;
826 	unsigned long		s_magic;
827 	struct dentry		*s_root;
828 	struct rw_semaphore	s_umount;
829 	struct mutex		s_lock;
830 	int			s_count;
831 	int			s_syncing;
832 	int			s_need_sync_fs;
833 	atomic_t		s_active;
834 	void                    *s_security;
835 	struct xattr_handler	**s_xattr;
836 
837 	struct list_head	s_inodes;	/* all inodes */
838 	struct list_head	s_dirty;	/* dirty inodes */
839 	struct list_head	s_io;		/* parked for writeback */
840 	struct hlist_head	s_anon;		/* anonymous dentries for (nfs) exporting */
841 	struct list_head	s_files;
842 
843 	struct block_device	*s_bdev;
844 	struct list_head	s_instances;
845 	struct quota_info	s_dquot;	/* Diskquota specific options */
846 
847 	int			s_frozen;
848 	wait_queue_head_t	s_wait_unfrozen;
849 
850 	char s_id[32];				/* Informational name */
851 
852 	void 			*s_fs_info;	/* Filesystem private info */
853 
854 	/*
855 	 * The next field is for VFS *only*. No filesystems have any business
856 	 * even looking at it. You had been warned.
857 	 */
858 	struct mutex s_vfs_rename_mutex;	/* Kludge */
859 
860 	/* Granuality of c/m/atime in ns.
861 	   Cannot be worse than a second */
862 	u32		   s_time_gran;
863 };
864 
865 extern struct timespec current_fs_time(struct super_block *sb);
866 
867 /*
868  * Snapshotting support.
869  */
870 enum {
871 	SB_UNFROZEN = 0,
872 	SB_FREEZE_WRITE	= 1,
873 	SB_FREEZE_TRANS = 2,
874 };
875 
876 #define vfs_check_frozen(sb, level) \
877 	wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
878 
879 static inline void get_fs_excl(void)
880 {
881 	atomic_inc(&current->fs_excl);
882 }
883 
884 static inline void put_fs_excl(void)
885 {
886 	atomic_dec(&current->fs_excl);
887 }
888 
889 static inline int has_fs_excl(void)
890 {
891 	return atomic_read(&current->fs_excl);
892 }
893 
894 
895 /*
896  * Superblock locking.
897  */
898 static inline void lock_super(struct super_block * sb)
899 {
900 	get_fs_excl();
901 	mutex_lock(&sb->s_lock);
902 }
903 
904 static inline void unlock_super(struct super_block * sb)
905 {
906 	put_fs_excl();
907 	mutex_unlock(&sb->s_lock);
908 }
909 
910 /*
911  * VFS helper functions..
912  */
913 extern int vfs_permission(struct nameidata *, int);
914 extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
915 extern int vfs_mkdir(struct inode *, struct dentry *, int);
916 extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
917 extern int vfs_symlink(struct inode *, struct dentry *, const char *, int);
918 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
919 extern int vfs_rmdir(struct inode *, struct dentry *);
920 extern int vfs_unlink(struct inode *, struct dentry *);
921 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
922 
923 /*
924  * VFS dentry helper functions.
925  */
926 extern void dentry_unhash(struct dentry *dentry);
927 
928 /*
929  * VFS file helper functions.
930  */
931 extern int file_permission(struct file *, int);
932 
933 /*
934  * File types
935  *
936  * NOTE! These match bits 12..15 of stat.st_mode
937  * (ie "(i_mode >> 12) & 15").
938  */
939 #define DT_UNKNOWN	0
940 #define DT_FIFO		1
941 #define DT_CHR		2
942 #define DT_DIR		4
943 #define DT_BLK		6
944 #define DT_REG		8
945 #define DT_LNK		10
946 #define DT_SOCK		12
947 #define DT_WHT		14
948 
949 #define OSYNC_METADATA	(1<<0)
950 #define OSYNC_DATA	(1<<1)
951 #define OSYNC_INODE	(1<<2)
952 int generic_osync_inode(struct inode *, struct address_space *, int);
953 
954 /*
955  * This is the "filldir" function type, used by readdir() to let
956  * the kernel specify what kind of dirent layout it wants to have.
957  * This allows the kernel to read directories into kernel space or
958  * to have different dirent layouts depending on the binary type.
959  */
960 typedef int (*filldir_t)(void *, const char *, int, loff_t, ino_t, unsigned);
961 
962 struct block_device_operations {
963 	int (*open) (struct inode *, struct file *);
964 	int (*release) (struct inode *, struct file *);
965 	int (*ioctl) (struct inode *, struct file *, unsigned, unsigned long);
966 	long (*unlocked_ioctl) (struct file *, unsigned, unsigned long);
967 	long (*compat_ioctl) (struct file *, unsigned, unsigned long);
968 	int (*direct_access) (struct block_device *, sector_t, unsigned long *);
969 	int (*media_changed) (struct gendisk *);
970 	int (*revalidate_disk) (struct gendisk *);
971 	int (*getgeo)(struct block_device *, struct hd_geometry *);
972 	struct module *owner;
973 };
974 
975 /*
976  * "descriptor" for what we're up to with a read for sendfile().
977  * This allows us to use the same read code yet
978  * have multiple different users of the data that
979  * we read from a file.
980  *
981  * The simplest case just copies the data to user
982  * mode.
983  */
984 typedef struct {
985 	size_t written;
986 	size_t count;
987 	union {
988 		char __user * buf;
989 		void *data;
990 	} arg;
991 	int error;
992 } read_descriptor_t;
993 
994 typedef int (*read_actor_t)(read_descriptor_t *, struct page *, unsigned long, unsigned long);
995 
996 /* These macros are for out of kernel modules to test that
997  * the kernel supports the unlocked_ioctl and compat_ioctl
998  * fields in struct file_operations. */
999 #define HAVE_COMPAT_IOCTL 1
1000 #define HAVE_UNLOCKED_IOCTL 1
1001 
1002 /*
1003  * NOTE:
1004  * read, write, poll, fsync, readv, writev, unlocked_ioctl and compat_ioctl
1005  * can be called without the big kernel lock held in all filesystems.
1006  */
1007 struct file_operations {
1008 	struct module *owner;
1009 	loff_t (*llseek) (struct file *, loff_t, int);
1010 	ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1011 	ssize_t (*aio_read) (struct kiocb *, char __user *, size_t, loff_t);
1012 	ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1013 	ssize_t (*aio_write) (struct kiocb *, const char __user *, size_t, loff_t);
1014 	int (*readdir) (struct file *, void *, filldir_t);
1015 	unsigned int (*poll) (struct file *, struct poll_table_struct *);
1016 	int (*ioctl) (struct inode *, struct file *, unsigned int, unsigned long);
1017 	long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1018 	long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1019 	int (*mmap) (struct file *, struct vm_area_struct *);
1020 	int (*open) (struct inode *, struct file *);
1021 	int (*flush) (struct file *);
1022 	int (*release) (struct inode *, struct file *);
1023 	int (*fsync) (struct file *, struct dentry *, int datasync);
1024 	int (*aio_fsync) (struct kiocb *, int datasync);
1025 	int (*fasync) (int, struct file *, int);
1026 	int (*lock) (struct file *, int, struct file_lock *);
1027 	ssize_t (*readv) (struct file *, const struct iovec *, unsigned long, loff_t *);
1028 	ssize_t (*writev) (struct file *, const struct iovec *, unsigned long, loff_t *);
1029 	ssize_t (*sendfile) (struct file *, loff_t *, size_t, read_actor_t, void *);
1030 	ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1031 	unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1032 	int (*check_flags)(int);
1033 	int (*dir_notify)(struct file *filp, unsigned long arg);
1034 	int (*flock) (struct file *, int, struct file_lock *);
1035 	ssize_t (*splice_write)(struct inode *, struct file *, size_t, unsigned int);
1036 	ssize_t (*splice_read)(struct file *, struct inode *, size_t, unsigned int);
1037 };
1038 
1039 struct inode_operations {
1040 	int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1041 	struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1042 	int (*link) (struct dentry *,struct inode *,struct dentry *);
1043 	int (*unlink) (struct inode *,struct dentry *);
1044 	int (*symlink) (struct inode *,struct dentry *,const char *);
1045 	int (*mkdir) (struct inode *,struct dentry *,int);
1046 	int (*rmdir) (struct inode *,struct dentry *);
1047 	int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1048 	int (*rename) (struct inode *, struct dentry *,
1049 			struct inode *, struct dentry *);
1050 	int (*readlink) (struct dentry *, char __user *,int);
1051 	void * (*follow_link) (struct dentry *, struct nameidata *);
1052 	void (*put_link) (struct dentry *, struct nameidata *, void *);
1053 	void (*truncate) (struct inode *);
1054 	int (*permission) (struct inode *, int, struct nameidata *);
1055 	int (*setattr) (struct dentry *, struct iattr *);
1056 	int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1057 	int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1058 	ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1059 	ssize_t (*listxattr) (struct dentry *, char *, size_t);
1060 	int (*removexattr) (struct dentry *, const char *);
1061 	void (*truncate_range)(struct inode *, loff_t, loff_t);
1062 };
1063 
1064 struct seq_file;
1065 
1066 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1067 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1068 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1069 		unsigned long, loff_t *);
1070 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1071 		unsigned long, loff_t *);
1072 
1073 /*
1074  * NOTE: write_inode, delete_inode, clear_inode, put_inode can be called
1075  * without the big kernel lock held in all filesystems.
1076  */
1077 struct super_operations {
1078    	struct inode *(*alloc_inode)(struct super_block *sb);
1079 	void (*destroy_inode)(struct inode *);
1080 
1081 	void (*read_inode) (struct inode *);
1082 
1083    	void (*dirty_inode) (struct inode *);
1084 	int (*write_inode) (struct inode *, int);
1085 	void (*put_inode) (struct inode *);
1086 	void (*drop_inode) (struct inode *);
1087 	void (*delete_inode) (struct inode *);
1088 	void (*put_super) (struct super_block *);
1089 	void (*write_super) (struct super_block *);
1090 	int (*sync_fs)(struct super_block *sb, int wait);
1091 	void (*write_super_lockfs) (struct super_block *);
1092 	void (*unlockfs) (struct super_block *);
1093 	int (*statfs) (struct super_block *, struct kstatfs *);
1094 	int (*remount_fs) (struct super_block *, int *, char *);
1095 	void (*clear_inode) (struct inode *);
1096 	void (*umount_begin) (struct super_block *);
1097 
1098 	int (*show_options)(struct seq_file *, struct vfsmount *);
1099 	int (*show_stats)(struct seq_file *, struct vfsmount *);
1100 
1101 	ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1102 	ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1103 };
1104 
1105 /* Inode state bits.  Protected by inode_lock. */
1106 #define I_DIRTY_SYNC		1 /* Not dirty enough for O_DATASYNC */
1107 #define I_DIRTY_DATASYNC	2 /* Data-related inode changes pending */
1108 #define I_DIRTY_PAGES		4 /* Data-related inode changes pending */
1109 #define __I_LOCK		3
1110 #define I_LOCK			(1 << __I_LOCK)
1111 #define I_FREEING		16
1112 #define I_CLEAR			32
1113 #define I_NEW			64
1114 #define I_WILL_FREE		128
1115 
1116 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1117 
1118 extern void __mark_inode_dirty(struct inode *, int);
1119 static inline void mark_inode_dirty(struct inode *inode)
1120 {
1121 	__mark_inode_dirty(inode, I_DIRTY);
1122 }
1123 
1124 static inline void mark_inode_dirty_sync(struct inode *inode)
1125 {
1126 	__mark_inode_dirty(inode, I_DIRTY_SYNC);
1127 }
1128 
1129 static inline void inode_inc_link_count(struct inode *inode)
1130 {
1131 	inode->i_nlink++;
1132 	mark_inode_dirty(inode);
1133 }
1134 
1135 static inline void inode_dec_link_count(struct inode *inode)
1136 {
1137 	inode->i_nlink--;
1138 	mark_inode_dirty(inode);
1139 }
1140 
1141 extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
1142 static inline void file_accessed(struct file *file)
1143 {
1144 	if (!(file->f_flags & O_NOATIME))
1145 		touch_atime(file->f_vfsmnt, file->f_dentry);
1146 }
1147 
1148 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1149 
1150 /**
1151  * struct export_operations - for nfsd to communicate with file systems
1152  * @decode_fh:      decode a file handle fragment and return a &struct dentry
1153  * @encode_fh:      encode a file handle fragment from a dentry
1154  * @get_name:       find the name for a given inode in a given directory
1155  * @get_parent:     find the parent of a given directory
1156  * @get_dentry:     find a dentry for the inode given a file handle sub-fragment
1157  * @find_exported_dentry:
1158  *	set by the exporting module to a standard helper function.
1159  *
1160  * Description:
1161  *    The export_operations structure provides a means for nfsd to communicate
1162  *    with a particular exported file system  - particularly enabling nfsd and
1163  *    the filesystem to co-operate when dealing with file handles.
1164  *
1165  *    export_operations contains two basic operation for dealing with file
1166  *    handles, decode_fh() and encode_fh(), and allows for some other
1167  *    operations to be defined which standard helper routines use to get
1168  *    specific information from the filesystem.
1169  *
1170  *    nfsd encodes information use to determine which filesystem a filehandle
1171  *    applies to in the initial part of the file handle.  The remainder, termed
1172  *    a file handle fragment, is controlled completely by the filesystem.  The
1173  *    standard helper routines assume that this fragment will contain one or
1174  *    two sub-fragments, one which identifies the file, and one which may be
1175  *    used to identify the (a) directory containing the file.
1176  *
1177  *    In some situations, nfsd needs to get a dentry which is connected into a
1178  *    specific part of the file tree.  To allow for this, it passes the
1179  *    function acceptable() together with a @context which can be used to see
1180  *    if the dentry is acceptable.  As there can be multiple dentrys for a
1181  *    given file, the filesystem should check each one for acceptability before
1182  *    looking for the next.  As soon as an acceptable one is found, it should
1183  *    be returned.
1184  *
1185  * decode_fh:
1186  *    @decode_fh is given a &struct super_block (@sb), a file handle fragment
1187  *    (@fh, @fh_len) and an acceptability testing function (@acceptable,
1188  *    @context).  It should return a &struct dentry which refers to the same
1189  *    file that the file handle fragment refers to,  and which passes the
1190  *    acceptability test.  If it cannot, it should return a %NULL pointer if
1191  *    the file was found but no acceptable &dentries were available, or a
1192  *    %ERR_PTR error code indicating why it couldn't be found (e.g. %ENOENT or
1193  *    %ENOMEM).
1194  *
1195  * encode_fh:
1196  *    @encode_fh should store in the file handle fragment @fh (using at most
1197  *    @max_len bytes) information that can be used by @decode_fh to recover the
1198  *    file refered to by the &struct dentry @de.  If the @connectable flag is
1199  *    set, the encode_fh() should store sufficient information so that a good
1200  *    attempt can be made to find not only the file but also it's place in the
1201  *    filesystem.   This typically means storing a reference to de->d_parent in
1202  *    the filehandle fragment.  encode_fh() should return the number of bytes
1203  *    stored or a negative error code such as %-ENOSPC
1204  *
1205  * get_name:
1206  *    @get_name should find a name for the given @child in the given @parent
1207  *    directory.  The name should be stored in the @name (with the
1208  *    understanding that it is already pointing to a a %NAME_MAX+1 sized
1209  *    buffer.   get_name() should return %0 on success, a negative error code
1210  *    or error.  @get_name will be called without @parent->i_mutex held.
1211  *
1212  * get_parent:
1213  *    @get_parent should find the parent directory for the given @child which
1214  *    is also a directory.  In the event that it cannot be found, or storage
1215  *    space cannot be allocated, a %ERR_PTR should be returned.
1216  *
1217  * get_dentry:
1218  *    Given a &super_block (@sb) and a pointer to a file-system specific inode
1219  *    identifier, possibly an inode number, (@inump) get_dentry() should find
1220  *    the identified inode and return a dentry for that inode.  Any suitable
1221  *    dentry can be returned including, if necessary, a new dentry created with
1222  *    d_alloc_root.  The caller can then find any other extant dentrys by
1223  *    following the d_alias links.  If a new dentry was created using
1224  *    d_alloc_root, DCACHE_NFSD_DISCONNECTED should be set, and the dentry
1225  *    should be d_rehash()ed.
1226  *
1227  *    If the inode cannot be found, either a %NULL pointer or an %ERR_PTR code
1228  *    can be returned.  The @inump will be whatever was passed to
1229  *    nfsd_find_fh_dentry() in either the @obj or @parent parameters.
1230  *
1231  * Locking rules:
1232  *    get_parent is called with child->d_inode->i_mutex down
1233  *    get_name is not (which is possibly inconsistent)
1234  */
1235 
1236 struct export_operations {
1237 	struct dentry *(*decode_fh)(struct super_block *sb, __u32 *fh, int fh_len, int fh_type,
1238 			 int (*acceptable)(void *context, struct dentry *de),
1239 			 void *context);
1240 	int (*encode_fh)(struct dentry *de, __u32 *fh, int *max_len,
1241 			 int connectable);
1242 
1243 	/* the following are only called from the filesystem itself */
1244 	int (*get_name)(struct dentry *parent, char *name,
1245 			struct dentry *child);
1246 	struct dentry * (*get_parent)(struct dentry *child);
1247 	struct dentry * (*get_dentry)(struct super_block *sb, void *inump);
1248 
1249 	/* This is set by the exporting module to a standard helper */
1250 	struct dentry * (*find_exported_dentry)(
1251 		struct super_block *sb, void *obj, void *parent,
1252 		int (*acceptable)(void *context, struct dentry *de),
1253 		void *context);
1254 
1255 
1256 };
1257 
1258 extern struct dentry *
1259 find_exported_dentry(struct super_block *sb, void *obj, void *parent,
1260 		     int (*acceptable)(void *context, struct dentry *de),
1261 		     void *context);
1262 
1263 struct file_system_type {
1264 	const char *name;
1265 	int fs_flags;
1266 	struct super_block *(*get_sb) (struct file_system_type *, int,
1267 				       const char *, void *);
1268 	void (*kill_sb) (struct super_block *);
1269 	struct module *owner;
1270 	struct file_system_type * next;
1271 	struct list_head fs_supers;
1272 };
1273 
1274 struct super_block *get_sb_bdev(struct file_system_type *fs_type,
1275 	int flags, const char *dev_name, void *data,
1276 	int (*fill_super)(struct super_block *, void *, int));
1277 struct super_block *get_sb_single(struct file_system_type *fs_type,
1278 	int flags, void *data,
1279 	int (*fill_super)(struct super_block *, void *, int));
1280 struct super_block *get_sb_nodev(struct file_system_type *fs_type,
1281 	int flags, void *data,
1282 	int (*fill_super)(struct super_block *, void *, int));
1283 void generic_shutdown_super(struct super_block *sb);
1284 void kill_block_super(struct super_block *sb);
1285 void kill_anon_super(struct super_block *sb);
1286 void kill_litter_super(struct super_block *sb);
1287 void deactivate_super(struct super_block *sb);
1288 int set_anon_super(struct super_block *s, void *data);
1289 struct super_block *sget(struct file_system_type *type,
1290 			int (*test)(struct super_block *,void *),
1291 			int (*set)(struct super_block *,void *),
1292 			void *data);
1293 struct super_block *get_sb_pseudo(struct file_system_type *, char *,
1294 			struct super_operations *ops, unsigned long);
1295 int __put_super(struct super_block *sb);
1296 int __put_super_and_need_restart(struct super_block *sb);
1297 void unnamed_dev_init(void);
1298 
1299 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1300 #define fops_get(fops) \
1301 	(((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1302 #define fops_put(fops) \
1303 	do { if (fops) module_put((fops)->owner); } while(0)
1304 
1305 extern int register_filesystem(struct file_system_type *);
1306 extern int unregister_filesystem(struct file_system_type *);
1307 extern struct vfsmount *kern_mount(struct file_system_type *);
1308 extern int may_umount_tree(struct vfsmount *);
1309 extern int may_umount(struct vfsmount *);
1310 extern void umount_tree(struct vfsmount *, int, struct list_head *);
1311 extern void release_mounts(struct list_head *);
1312 extern long do_mount(char *, char *, char *, unsigned long, void *);
1313 extern struct vfsmount *copy_tree(struct vfsmount *, struct dentry *, int);
1314 extern void mnt_set_mountpoint(struct vfsmount *, struct dentry *,
1315 				  struct vfsmount *);
1316 
1317 extern int vfs_statfs(struct super_block *, struct kstatfs *);
1318 
1319 /* /sys/fs */
1320 extern struct subsystem fs_subsys;
1321 
1322 #define FLOCK_VERIFY_READ  1
1323 #define FLOCK_VERIFY_WRITE 2
1324 
1325 extern int locks_mandatory_locked(struct inode *);
1326 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1327 
1328 /*
1329  * Candidates for mandatory locking have the setgid bit set
1330  * but no group execute bit -  an otherwise meaningless combination.
1331  */
1332 #define MANDATORY_LOCK(inode) \
1333 	(IS_MANDLOCK(inode) && ((inode)->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID)
1334 
1335 static inline int locks_verify_locked(struct inode *inode)
1336 {
1337 	if (MANDATORY_LOCK(inode))
1338 		return locks_mandatory_locked(inode);
1339 	return 0;
1340 }
1341 
1342 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1343 
1344 static inline int locks_verify_truncate(struct inode *inode,
1345 				    struct file *filp,
1346 				    loff_t size)
1347 {
1348 	if (inode->i_flock && MANDATORY_LOCK(inode))
1349 		return locks_mandatory_area(
1350 			FLOCK_VERIFY_WRITE, inode, filp,
1351 			size < inode->i_size ? size : inode->i_size,
1352 			(size < inode->i_size ? inode->i_size - size
1353 			 : size - inode->i_size)
1354 		);
1355 	return 0;
1356 }
1357 
1358 static inline int break_lease(struct inode *inode, unsigned int mode)
1359 {
1360 	if (inode->i_flock)
1361 		return __break_lease(inode, mode);
1362 	return 0;
1363 }
1364 
1365 /* fs/open.c */
1366 
1367 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
1368 		       struct file *filp);
1369 extern long do_sys_open(int fdf, const char __user *filename, int flags,
1370 			int mode);
1371 extern struct file *filp_open(const char *, int, int);
1372 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int);
1373 extern int filp_close(struct file *, fl_owner_t id);
1374 extern char * getname(const char __user *);
1375 
1376 /* fs/dcache.c */
1377 extern void __init vfs_caches_init_early(void);
1378 extern void __init vfs_caches_init(unsigned long);
1379 
1380 #define __getname()	kmem_cache_alloc(names_cachep, SLAB_KERNEL)
1381 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
1382 #ifndef CONFIG_AUDITSYSCALL
1383 #define putname(name)   __putname(name)
1384 #else
1385 extern void putname(const char *name);
1386 #endif
1387 
1388 extern int register_blkdev(unsigned int, const char *);
1389 extern int unregister_blkdev(unsigned int, const char *);
1390 extern struct block_device *bdget(dev_t);
1391 extern void bd_set_size(struct block_device *, loff_t size);
1392 extern void bd_forget(struct inode *inode);
1393 extern void bdput(struct block_device *);
1394 extern struct block_device *open_by_devnum(dev_t, unsigned);
1395 extern const struct file_operations def_blk_fops;
1396 extern struct address_space_operations def_blk_aops;
1397 extern const struct file_operations def_chr_fops;
1398 extern const struct file_operations bad_sock_fops;
1399 extern const struct file_operations def_fifo_fops;
1400 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
1401 extern int blkdev_ioctl(struct inode *, struct file *, unsigned, unsigned long);
1402 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
1403 extern int blkdev_get(struct block_device *, mode_t, unsigned);
1404 extern int blkdev_put(struct block_device *);
1405 extern int bd_claim(struct block_device *, void *);
1406 extern void bd_release(struct block_device *);
1407 #ifdef CONFIG_SYSFS
1408 extern int bd_claim_by_disk(struct block_device *, void *, struct gendisk *);
1409 extern void bd_release_from_disk(struct block_device *, struct gendisk *);
1410 #else
1411 #define bd_claim_by_disk(bdev, holder, disk)	bd_claim(bdev, holder)
1412 #define bd_release_from_disk(bdev, disk)	bd_release(bdev)
1413 #endif
1414 
1415 /* fs/char_dev.c */
1416 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
1417 extern int register_chrdev_region(dev_t, unsigned, const char *);
1418 extern int register_chrdev(unsigned int, const char *,
1419 			   const struct file_operations *);
1420 extern int unregister_chrdev(unsigned int, const char *);
1421 extern void unregister_chrdev_region(dev_t, unsigned);
1422 extern int chrdev_open(struct inode *, struct file *);
1423 extern int get_chrdev_list(char *);
1424 extern void *acquire_chrdev_list(void);
1425 extern int count_chrdev_list(void);
1426 extern void *get_next_chrdev(void *);
1427 extern int get_chrdev_info(void *, int *, char **);
1428 extern void release_chrdev_list(void *);
1429 
1430 /* fs/block_dev.c */
1431 #define BDEVNAME_SIZE	32	/* Largest string for a blockdev identifier */
1432 extern const char *__bdevname(dev_t, char *buffer);
1433 extern const char *bdevname(struct block_device *bdev, char *buffer);
1434 extern struct block_device *lookup_bdev(const char *);
1435 extern struct block_device *open_bdev_excl(const char *, int, void *);
1436 extern void close_bdev_excl(struct block_device *);
1437 extern void *acquire_blkdev_list(void);
1438 extern int count_blkdev_list(void);
1439 extern void *get_next_blkdev(void *);
1440 extern int get_blkdev_info(void *, int *, char **);
1441 extern void release_blkdev_list(void *);
1442 
1443 extern void init_special_inode(struct inode *, umode_t, dev_t);
1444 
1445 /* Invalid inode operations -- fs/bad_inode.c */
1446 extern void make_bad_inode(struct inode *);
1447 extern int is_bad_inode(struct inode *);
1448 
1449 extern const struct file_operations read_fifo_fops;
1450 extern const struct file_operations write_fifo_fops;
1451 extern const struct file_operations rdwr_fifo_fops;
1452 
1453 extern int fs_may_remount_ro(struct super_block *);
1454 
1455 /*
1456  * return READ, READA, or WRITE
1457  */
1458 #define bio_rw(bio)		((bio)->bi_rw & (RW_MASK | RWA_MASK))
1459 
1460 /*
1461  * return data direction, READ or WRITE
1462  */
1463 #define bio_data_dir(bio)	((bio)->bi_rw & 1)
1464 
1465 extern int check_disk_change(struct block_device *);
1466 extern int invalidate_inodes(struct super_block *);
1467 extern int __invalidate_device(struct block_device *);
1468 extern int invalidate_partition(struct gendisk *, int);
1469 unsigned long invalidate_mapping_pages(struct address_space *mapping,
1470 					pgoff_t start, pgoff_t end);
1471 unsigned long invalidate_inode_pages(struct address_space *mapping);
1472 static inline void invalidate_remote_inode(struct inode *inode)
1473 {
1474 	if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
1475 	    S_ISLNK(inode->i_mode))
1476 		invalidate_inode_pages(inode->i_mapping);
1477 }
1478 extern int invalidate_inode_pages2(struct address_space *mapping);
1479 extern int invalidate_inode_pages2_range(struct address_space *mapping,
1480 					 pgoff_t start, pgoff_t end);
1481 extern int write_inode_now(struct inode *, int);
1482 extern int filemap_fdatawrite(struct address_space *);
1483 extern int filemap_flush(struct address_space *);
1484 extern int filemap_fdatawait(struct address_space *);
1485 extern int filemap_write_and_wait(struct address_space *mapping);
1486 extern int filemap_write_and_wait_range(struct address_space *mapping,
1487 				        loff_t lstart, loff_t lend);
1488 extern int wait_on_page_writeback_range(struct address_space *mapping,
1489 				pgoff_t start, pgoff_t end);
1490 extern int __filemap_fdatawrite_range(struct address_space *mapping,
1491 				loff_t start, loff_t end, int sync_mode);
1492 
1493 extern long do_fsync(struct file *file, int datasync);
1494 extern void sync_supers(void);
1495 extern void sync_filesystems(int wait);
1496 extern void emergency_sync(void);
1497 extern void emergency_remount(void);
1498 extern int do_remount_sb(struct super_block *sb, int flags,
1499 			 void *data, int force);
1500 extern sector_t bmap(struct inode *, sector_t);
1501 extern int notify_change(struct dentry *, struct iattr *);
1502 extern int permission(struct inode *, int, struct nameidata *);
1503 extern int generic_permission(struct inode *, int,
1504 		int (*check_acl)(struct inode *, int));
1505 
1506 extern int get_write_access(struct inode *);
1507 extern int deny_write_access(struct file *);
1508 static inline void put_write_access(struct inode * inode)
1509 {
1510 	atomic_dec(&inode->i_writecount);
1511 }
1512 static inline void allow_write_access(struct file *file)
1513 {
1514 	if (file)
1515 		atomic_inc(&file->f_dentry->d_inode->i_writecount);
1516 }
1517 extern int do_pipe(int *);
1518 
1519 extern int open_namei(int dfd, const char *, int, int, struct nameidata *);
1520 extern int may_open(struct nameidata *, int, int);
1521 
1522 extern int kernel_read(struct file *, unsigned long, char *, unsigned long);
1523 extern struct file * open_exec(const char *);
1524 
1525 /* fs/dcache.c -- generic fs support functions */
1526 extern int is_subdir(struct dentry *, struct dentry *);
1527 extern ino_t find_inode_number(struct dentry *, struct qstr *);
1528 
1529 #include <linux/err.h>
1530 
1531 /* needed for stackable file system support */
1532 extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
1533 
1534 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
1535 
1536 extern void inode_init_once(struct inode *);
1537 extern void iput(struct inode *);
1538 extern struct inode * igrab(struct inode *);
1539 extern ino_t iunique(struct super_block *, ino_t);
1540 extern int inode_needs_sync(struct inode *inode);
1541 extern void generic_delete_inode(struct inode *inode);
1542 extern void generic_drop_inode(struct inode *inode);
1543 
1544 extern struct inode *ilookup5_nowait(struct super_block *sb,
1545 		unsigned long hashval, int (*test)(struct inode *, void *),
1546 		void *data);
1547 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
1548 		int (*test)(struct inode *, void *), void *data);
1549 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
1550 
1551 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
1552 extern struct inode * iget_locked(struct super_block *, unsigned long);
1553 extern void unlock_new_inode(struct inode *);
1554 
1555 static inline struct inode *iget(struct super_block *sb, unsigned long ino)
1556 {
1557 	struct inode *inode = iget_locked(sb, ino);
1558 
1559 	if (inode && (inode->i_state & I_NEW)) {
1560 		sb->s_op->read_inode(inode);
1561 		unlock_new_inode(inode);
1562 	}
1563 
1564 	return inode;
1565 }
1566 
1567 extern void __iget(struct inode * inode);
1568 extern void clear_inode(struct inode *);
1569 extern void destroy_inode(struct inode *);
1570 extern struct inode *new_inode(struct super_block *);
1571 extern int remove_suid(struct dentry *);
1572 extern void remove_dquot_ref(struct super_block *, int, struct list_head *);
1573 
1574 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
1575 extern void remove_inode_hash(struct inode *);
1576 static inline void insert_inode_hash(struct inode *inode) {
1577 	__insert_inode_hash(inode, inode->i_ino);
1578 }
1579 
1580 extern struct file * get_empty_filp(void);
1581 extern void file_move(struct file *f, struct list_head *list);
1582 extern void file_kill(struct file *f);
1583 struct bio;
1584 extern void submit_bio(int, struct bio *);
1585 extern int bdev_read_only(struct block_device *);
1586 extern int set_blocksize(struct block_device *, int);
1587 extern int sb_set_blocksize(struct super_block *, int);
1588 extern int sb_min_blocksize(struct super_block *, int);
1589 
1590 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
1591 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
1592 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1593 extern int file_send_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1594 extern ssize_t generic_file_read(struct file *, char __user *, size_t, loff_t *);
1595 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
1596 extern ssize_t generic_file_write(struct file *, const char __user *, size_t, loff_t *);
1597 extern ssize_t generic_file_aio_read(struct kiocb *, char __user *, size_t, loff_t);
1598 extern ssize_t __generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t *);
1599 extern ssize_t generic_file_aio_write(struct kiocb *, const char __user *, size_t, loff_t);
1600 extern ssize_t generic_file_aio_write_nolock(struct kiocb *, const struct iovec *,
1601 		unsigned long, loff_t *);
1602 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
1603 		unsigned long *, loff_t, loff_t *, size_t, size_t);
1604 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
1605 		unsigned long, loff_t, loff_t *, size_t, ssize_t);
1606 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
1607 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
1608 ssize_t generic_file_write_nolock(struct file *file, const struct iovec *iov,
1609 				unsigned long nr_segs, loff_t *ppos);
1610 extern ssize_t generic_file_sendfile(struct file *, loff_t *, size_t, read_actor_t, void *);
1611 extern void do_generic_mapping_read(struct address_space *mapping,
1612 				    struct file_ra_state *, struct file *,
1613 				    loff_t *, read_descriptor_t *, read_actor_t);
1614 extern ssize_t generic_file_splice_read(struct file *, struct inode *, size_t, unsigned int);
1615 extern ssize_t generic_file_splice_write(struct inode *, struct file *, size_t, unsigned int);
1616 extern void
1617 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
1618 extern ssize_t generic_file_readv(struct file *filp, const struct iovec *iov,
1619 	unsigned long nr_segs, loff_t *ppos);
1620 ssize_t generic_file_writev(struct file *filp, const struct iovec *iov,
1621 			unsigned long nr_segs, loff_t *ppos);
1622 extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
1623 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
1624 extern loff_t remote_llseek(struct file *file, loff_t offset, int origin);
1625 extern int generic_file_open(struct inode * inode, struct file * filp);
1626 extern int nonseekable_open(struct inode * inode, struct file * filp);
1627 
1628 #ifdef CONFIG_FS_XIP
1629 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
1630 			     loff_t *ppos);
1631 extern ssize_t xip_file_sendfile(struct file *in_file, loff_t *ppos,
1632 				 size_t count, read_actor_t actor,
1633 				 void *target);
1634 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
1635 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
1636 			      size_t len, loff_t *ppos);
1637 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
1638 #else
1639 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
1640 {
1641 	return 0;
1642 }
1643 #endif
1644 
1645 static inline void do_generic_file_read(struct file * filp, loff_t *ppos,
1646 					read_descriptor_t * desc,
1647 					read_actor_t actor)
1648 {
1649 	do_generic_mapping_read(filp->f_mapping,
1650 				&filp->f_ra,
1651 				filp,
1652 				ppos,
1653 				desc,
1654 				actor);
1655 }
1656 
1657 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
1658 	struct block_device *bdev, const struct iovec *iov, loff_t offset,
1659 	unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
1660 	int lock_type);
1661 
1662 enum {
1663 	DIO_LOCKING = 1, /* need locking between buffered and direct access */
1664 	DIO_NO_LOCKING,  /* bdev; no locking at all between buffered/direct */
1665 	DIO_OWN_LOCKING, /* filesystem locks buffered and direct internally */
1666 };
1667 
1668 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
1669 	struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1670 	loff_t offset, unsigned long nr_segs, get_block_t get_block,
1671 	dio_iodone_t end_io)
1672 {
1673 	return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1674 				nr_segs, get_block, end_io, DIO_LOCKING);
1675 }
1676 
1677 static inline ssize_t blockdev_direct_IO_no_locking(int rw, struct kiocb *iocb,
1678 	struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1679 	loff_t offset, unsigned long nr_segs, get_block_t get_block,
1680 	dio_iodone_t end_io)
1681 {
1682 	return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1683 				nr_segs, get_block, end_io, DIO_NO_LOCKING);
1684 }
1685 
1686 static inline ssize_t blockdev_direct_IO_own_locking(int rw, struct kiocb *iocb,
1687 	struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1688 	loff_t offset, unsigned long nr_segs, get_block_t get_block,
1689 	dio_iodone_t end_io)
1690 {
1691 	return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1692 				nr_segs, get_block, end_io, DIO_OWN_LOCKING);
1693 }
1694 
1695 extern const struct file_operations generic_ro_fops;
1696 
1697 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
1698 
1699 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
1700 extern int vfs_follow_link(struct nameidata *, const char *);
1701 extern int page_readlink(struct dentry *, char __user *, int);
1702 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
1703 extern void page_put_link(struct dentry *, struct nameidata *, void *);
1704 extern int __page_symlink(struct inode *inode, const char *symname, int len,
1705 		gfp_t gfp_mask);
1706 extern int page_symlink(struct inode *inode, const char *symname, int len);
1707 extern struct inode_operations page_symlink_inode_operations;
1708 extern int generic_readlink(struct dentry *, char __user *, int);
1709 extern void generic_fillattr(struct inode *, struct kstat *);
1710 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1711 void inode_add_bytes(struct inode *inode, loff_t bytes);
1712 void inode_sub_bytes(struct inode *inode, loff_t bytes);
1713 loff_t inode_get_bytes(struct inode *inode);
1714 void inode_set_bytes(struct inode *inode, loff_t bytes);
1715 
1716 extern int vfs_readdir(struct file *, filldir_t, void *);
1717 
1718 extern int vfs_stat(char __user *, struct kstat *);
1719 extern int vfs_lstat(char __user *, struct kstat *);
1720 extern int vfs_stat_fd(int dfd, char __user *, struct kstat *);
1721 extern int vfs_lstat_fd(int dfd, char __user *, struct kstat *);
1722 extern int vfs_fstat(unsigned int, struct kstat *);
1723 
1724 extern int vfs_ioctl(struct file *, unsigned int, unsigned int, unsigned long);
1725 
1726 extern struct file_system_type *get_fs_type(const char *name);
1727 extern struct super_block *get_super(struct block_device *);
1728 extern struct super_block *user_get_super(dev_t);
1729 extern void drop_super(struct super_block *sb);
1730 
1731 extern int dcache_dir_open(struct inode *, struct file *);
1732 extern int dcache_dir_close(struct inode *, struct file *);
1733 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
1734 extern int dcache_readdir(struct file *, void *, filldir_t);
1735 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1736 extern int simple_statfs(struct super_block *, struct kstatfs *);
1737 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
1738 extern int simple_unlink(struct inode *, struct dentry *);
1739 extern int simple_rmdir(struct inode *, struct dentry *);
1740 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1741 extern int simple_sync_file(struct file *, struct dentry *, int);
1742 extern int simple_empty(struct dentry *);
1743 extern int simple_readpage(struct file *file, struct page *page);
1744 extern int simple_prepare_write(struct file *file, struct page *page,
1745 			unsigned offset, unsigned to);
1746 extern int simple_commit_write(struct file *file, struct page *page,
1747 				unsigned offset, unsigned to);
1748 
1749 extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
1750 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
1751 extern const struct file_operations simple_dir_operations;
1752 extern struct inode_operations simple_dir_inode_operations;
1753 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
1754 struct dentry *d_alloc_name(struct dentry *, const char *);
1755 extern int simple_fill_super(struct super_block *, int, struct tree_descr *);
1756 extern int simple_pin_fs(char *name, struct vfsmount **mount, int *count);
1757 extern void simple_release_fs(struct vfsmount **mount, int *count);
1758 
1759 extern ssize_t simple_read_from_buffer(void __user *, size_t, loff_t *, const void *, size_t);
1760 
1761 #ifdef CONFIG_MIGRATION
1762 extern int buffer_migrate_page(struct page *, struct page *);
1763 #else
1764 #define buffer_migrate_page NULL
1765 #endif
1766 
1767 extern int inode_change_ok(struct inode *, struct iattr *);
1768 extern int __must_check inode_setattr(struct inode *, struct iattr *);
1769 
1770 extern void file_update_time(struct file *file);
1771 
1772 static inline ino_t parent_ino(struct dentry *dentry)
1773 {
1774 	ino_t res;
1775 
1776 	spin_lock(&dentry->d_lock);
1777 	res = dentry->d_parent->d_inode->i_ino;
1778 	spin_unlock(&dentry->d_lock);
1779 	return res;
1780 }
1781 
1782 /* kernel/fork.c */
1783 extern int unshare_files(void);
1784 
1785 /* Transaction based IO helpers */
1786 
1787 /*
1788  * An argresp is stored in an allocated page and holds the
1789  * size of the argument or response, along with its content
1790  */
1791 struct simple_transaction_argresp {
1792 	ssize_t size;
1793 	char data[0];
1794 };
1795 
1796 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
1797 
1798 char *simple_transaction_get(struct file *file, const char __user *buf,
1799 				size_t size);
1800 ssize_t simple_transaction_read(struct file *file, char __user *buf,
1801 				size_t size, loff_t *pos);
1802 int simple_transaction_release(struct inode *inode, struct file *file);
1803 
1804 static inline void simple_transaction_set(struct file *file, size_t n)
1805 {
1806 	struct simple_transaction_argresp *ar = file->private_data;
1807 
1808 	BUG_ON(n > SIMPLE_TRANSACTION_LIMIT);
1809 
1810 	/*
1811 	 * The barrier ensures that ar->size will really remain zero until
1812 	 * ar->data is ready for reading.
1813 	 */
1814 	smp_mb();
1815 	ar->size = n;
1816 }
1817 
1818 /*
1819  * simple attribute files
1820  *
1821  * These attributes behave similar to those in sysfs:
1822  *
1823  * Writing to an attribute immediately sets a value, an open file can be
1824  * written to multiple times.
1825  *
1826  * Reading from an attribute creates a buffer from the value that might get
1827  * read with multiple read calls. When the attribute has been read
1828  * completely, no further read calls are possible until the file is opened
1829  * again.
1830  *
1831  * All attributes contain a text representation of a numeric value
1832  * that are accessed with the get() and set() functions.
1833  */
1834 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt)		\
1835 static int __fops ## _open(struct inode *inode, struct file *file)	\
1836 {									\
1837 	__simple_attr_check_format(__fmt, 0ull);			\
1838 	return simple_attr_open(inode, file, __get, __set, __fmt);	\
1839 }									\
1840 static struct file_operations __fops = {				\
1841 	.owner	 = THIS_MODULE,						\
1842 	.open	 = __fops ## _open,					\
1843 	.release = simple_attr_close,					\
1844 	.read	 = simple_attr_read,					\
1845 	.write	 = simple_attr_write,					\
1846 };
1847 
1848 static inline void __attribute__((format(printf, 1, 2)))
1849 __simple_attr_check_format(const char *fmt, ...)
1850 {
1851 	/* don't do anything, just let the compiler check the arguments; */
1852 }
1853 
1854 int simple_attr_open(struct inode *inode, struct file *file,
1855 		     u64 (*get)(void *), void (*set)(void *, u64),
1856 		     const char *fmt);
1857 int simple_attr_close(struct inode *inode, struct file *file);
1858 ssize_t simple_attr_read(struct file *file, char __user *buf,
1859 			 size_t len, loff_t *ppos);
1860 ssize_t simple_attr_write(struct file *file, const char __user *buf,
1861 			  size_t len, loff_t *ppos);
1862 
1863 
1864 #ifdef CONFIG_SECURITY
1865 static inline char *alloc_secdata(void)
1866 {
1867 	return (char *)get_zeroed_page(GFP_KERNEL);
1868 }
1869 
1870 static inline void free_secdata(void *secdata)
1871 {
1872 	free_page((unsigned long)secdata);
1873 }
1874 #else
1875 static inline char *alloc_secdata(void)
1876 {
1877 	return (char *)1;
1878 }
1879 
1880 static inline void free_secdata(void *secdata)
1881 { }
1882 #endif	/* CONFIG_SECURITY */
1883 
1884 #endif /* __KERNEL__ */
1885 #endif /* _LINUX_FS_H */
1886