xref: /linux-6.15/include/linux/fs.h (revision eb2bce7f)
1 #ifndef _LINUX_FS_H
2 #define _LINUX_FS_H
3 
4 /*
5  * This file has definitions for some important file table
6  * structures etc.
7  */
8 
9 #include <linux/limits.h>
10 #include <linux/ioctl.h>
11 
12 /*
13  * It's silly to have NR_OPEN bigger than NR_FILE, but you can change
14  * the file limit at runtime and only root can increase the per-process
15  * nr_file rlimit, so it's safe to set up a ridiculously high absolute
16  * upper limit on files-per-process.
17  *
18  * Some programs (notably those using select()) may have to be
19  * recompiled to take full advantage of the new limits..
20  */
21 
22 /* Fixed constants first: */
23 #undef NR_OPEN
24 #define NR_OPEN (1024*1024)	/* Absolute upper limit on fd num */
25 #define INR_OPEN 1024		/* Initial setting for nfile rlimits */
26 
27 #define BLOCK_SIZE_BITS 10
28 #define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
29 
30 #define SEEK_SET	0	/* seek relative to beginning of file */
31 #define SEEK_CUR	1	/* seek relative to current file position */
32 #define SEEK_END	2	/* seek relative to end of file */
33 
34 /* And dynamically-tunable limits and defaults: */
35 struct files_stat_struct {
36 	int nr_files;		/* read only */
37 	int nr_free_files;	/* read only */
38 	int max_files;		/* tunable */
39 };
40 extern struct files_stat_struct files_stat;
41 extern int get_max_files(void);
42 
43 struct inodes_stat_t {
44 	int nr_inodes;
45 	int nr_unused;
46 	int dummy[5];
47 };
48 extern struct inodes_stat_t inodes_stat;
49 
50 extern int leases_enable, lease_break_time;
51 
52 #ifdef CONFIG_DNOTIFY
53 extern int dir_notify_enable;
54 #endif
55 
56 #define NR_FILE  8192	/* this can well be larger on a larger system */
57 
58 #define MAY_EXEC 1
59 #define MAY_WRITE 2
60 #define MAY_READ 4
61 #define MAY_APPEND 8
62 
63 #define FMODE_READ 1
64 #define FMODE_WRITE 2
65 
66 /* Internal kernel extensions */
67 #define FMODE_LSEEK	4
68 #define FMODE_PREAD	8
69 #define FMODE_PWRITE	FMODE_PREAD	/* These go hand in hand */
70 
71 /* File is being opened for execution. Primary users of this flag are
72    distributed filesystems that can use it to achieve correct ETXTBUSY
73    behavior for cross-node execution/opening_for_writing of files */
74 #define FMODE_EXEC	16
75 
76 #define RW_MASK		1
77 #define RWA_MASK	2
78 #define READ 0
79 #define WRITE 1
80 #define READA 2		/* read-ahead  - don't block if no resources */
81 #define SWRITE 3	/* for ll_rw_block() - wait for buffer lock */
82 #define READ_SYNC	(READ | (1 << BIO_RW_SYNC))
83 #define READ_META	(READ | (1 << BIO_RW_META))
84 #define WRITE_SYNC	(WRITE | (1 << BIO_RW_SYNC))
85 #define WRITE_BARRIER	((1 << BIO_RW) | (1 << BIO_RW_BARRIER))
86 
87 #define SEL_IN		1
88 #define SEL_OUT		2
89 #define SEL_EX		4
90 
91 /* public flags for file_system_type */
92 #define FS_REQUIRES_DEV 1
93 #define FS_BINARY_MOUNTDATA 2
94 #define FS_REVAL_DOT	16384	/* Check the paths ".", ".." for staleness */
95 #define FS_RENAME_DOES_D_MOVE	32768	/* FS will handle d_move()
96 					 * during rename() internally.
97 					 */
98 
99 /*
100  * These are the fs-independent mount-flags: up to 32 flags are supported
101  */
102 #define MS_RDONLY	 1	/* Mount read-only */
103 #define MS_NOSUID	 2	/* Ignore suid and sgid bits */
104 #define MS_NODEV	 4	/* Disallow access to device special files */
105 #define MS_NOEXEC	 8	/* Disallow program execution */
106 #define MS_SYNCHRONOUS	16	/* Writes are synced at once */
107 #define MS_REMOUNT	32	/* Alter flags of a mounted FS */
108 #define MS_MANDLOCK	64	/* Allow mandatory locks on an FS */
109 #define MS_DIRSYNC	128	/* Directory modifications are synchronous */
110 #define MS_NOATIME	1024	/* Do not update access times. */
111 #define MS_NODIRATIME	2048	/* Do not update directory access times */
112 #define MS_BIND		4096
113 #define MS_MOVE		8192
114 #define MS_REC		16384
115 #define MS_VERBOSE	32768	/* War is peace. Verbosity is silence.
116 				   MS_VERBOSE is deprecated. */
117 #define MS_SILENT	32768
118 #define MS_POSIXACL	(1<<16)	/* VFS does not apply the umask */
119 #define MS_UNBINDABLE	(1<<17)	/* change to unbindable */
120 #define MS_PRIVATE	(1<<18)	/* change to private */
121 #define MS_SLAVE	(1<<19)	/* change to slave */
122 #define MS_SHARED	(1<<20)	/* change to shared */
123 #define MS_RELATIME	(1<<21)	/* Update atime relative to mtime/ctime. */
124 #define MS_ACTIVE	(1<<30)
125 #define MS_NOUSER	(1<<31)
126 
127 /*
128  * Superblock flags that can be altered by MS_REMOUNT
129  */
130 #define MS_RMT_MASK	(MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK)
131 
132 /*
133  * Old magic mount flag and mask
134  */
135 #define MS_MGC_VAL 0xC0ED0000
136 #define MS_MGC_MSK 0xffff0000
137 
138 /* Inode flags - they have nothing to superblock flags now */
139 
140 #define S_SYNC		1	/* Writes are synced at once */
141 #define S_NOATIME	2	/* Do not update access times */
142 #define S_APPEND	4	/* Append-only file */
143 #define S_IMMUTABLE	8	/* Immutable file */
144 #define S_DEAD		16	/* removed, but still open directory */
145 #define S_NOQUOTA	32	/* Inode is not counted to quota */
146 #define S_DIRSYNC	64	/* Directory modifications are synchronous */
147 #define S_NOCMTIME	128	/* Do not update file c/mtime */
148 #define S_SWAPFILE	256	/* Do not truncate: swapon got its bmaps */
149 #define S_PRIVATE	512	/* Inode is fs-internal */
150 
151 /*
152  * Note that nosuid etc flags are inode-specific: setting some file-system
153  * flags just means all the inodes inherit those flags by default. It might be
154  * possible to override it selectively if you really wanted to with some
155  * ioctl() that is not currently implemented.
156  *
157  * Exception: MS_RDONLY is always applied to the entire file system.
158  *
159  * Unfortunately, it is possible to change a filesystems flags with it mounted
160  * with files in use.  This means that all of the inodes will not have their
161  * i_flags updated.  Hence, i_flags no longer inherit the superblock mount
162  * flags, so these have to be checked separately. -- [email protected]
163  */
164 #define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
165 
166 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
167 #define IS_SYNC(inode)		(__IS_FLG(inode, MS_SYNCHRONOUS) || \
168 					((inode)->i_flags & S_SYNC))
169 #define IS_DIRSYNC(inode)	(__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
170 					((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
171 #define IS_MANDLOCK(inode)	__IS_FLG(inode, MS_MANDLOCK)
172 #define IS_NOATIME(inode)   __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
173 
174 #define IS_NOQUOTA(inode)	((inode)->i_flags & S_NOQUOTA)
175 #define IS_APPEND(inode)	((inode)->i_flags & S_APPEND)
176 #define IS_IMMUTABLE(inode)	((inode)->i_flags & S_IMMUTABLE)
177 #define IS_POSIXACL(inode)	__IS_FLG(inode, MS_POSIXACL)
178 
179 #define IS_DEADDIR(inode)	((inode)->i_flags & S_DEAD)
180 #define IS_NOCMTIME(inode)	((inode)->i_flags & S_NOCMTIME)
181 #define IS_SWAPFILE(inode)	((inode)->i_flags & S_SWAPFILE)
182 #define IS_PRIVATE(inode)	((inode)->i_flags & S_PRIVATE)
183 
184 /* the read-only stuff doesn't really belong here, but any other place is
185    probably as bad and I don't want to create yet another include file. */
186 
187 #define BLKROSET   _IO(0x12,93)	/* set device read-only (0 = read-write) */
188 #define BLKROGET   _IO(0x12,94)	/* get read-only status (0 = read_write) */
189 #define BLKRRPART  _IO(0x12,95)	/* re-read partition table */
190 #define BLKGETSIZE _IO(0x12,96)	/* return device size /512 (long *arg) */
191 #define BLKFLSBUF  _IO(0x12,97)	/* flush buffer cache */
192 #define BLKRASET   _IO(0x12,98)	/* set read ahead for block device */
193 #define BLKRAGET   _IO(0x12,99)	/* get current read ahead setting */
194 #define BLKFRASET  _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
195 #define BLKFRAGET  _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
196 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
197 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
198 #define BLKSSZGET  _IO(0x12,104)/* get block device sector size */
199 #if 0
200 #define BLKPG      _IO(0x12,105)/* See blkpg.h */
201 
202 /* Some people are morons.  Do not use sizeof! */
203 
204 #define BLKELVGET  _IOR(0x12,106,size_t)/* elevator get */
205 #define BLKELVSET  _IOW(0x12,107,size_t)/* elevator set */
206 /* This was here just to show that the number is taken -
207    probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
208 #endif
209 /* A jump here: 108-111 have been used for various private purposes. */
210 #define BLKBSZGET  _IOR(0x12,112,size_t)
211 #define BLKBSZSET  _IOW(0x12,113,size_t)
212 #define BLKGETSIZE64 _IOR(0x12,114,size_t)	/* return device size in bytes (u64 *arg) */
213 #define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
214 #define BLKTRACESTART _IO(0x12,116)
215 #define BLKTRACESTOP _IO(0x12,117)
216 #define BLKTRACETEARDOWN _IO(0x12,118)
217 
218 #define BMAP_IOCTL 1		/* obsolete - kept for compatibility */
219 #define FIBMAP	   _IO(0x00,1)	/* bmap access */
220 #define FIGETBSZ   _IO(0x00,2)	/* get the block size used for bmap */
221 
222 #define	FS_IOC_GETFLAGS			_IOR('f', 1, long)
223 #define	FS_IOC_SETFLAGS			_IOW('f', 2, long)
224 #define	FS_IOC_GETVERSION		_IOR('v', 1, long)
225 #define	FS_IOC_SETVERSION		_IOW('v', 2, long)
226 #define FS_IOC32_GETFLAGS		_IOR('f', 1, int)
227 #define FS_IOC32_SETFLAGS		_IOW('f', 2, int)
228 #define FS_IOC32_GETVERSION		_IOR('v', 1, int)
229 #define FS_IOC32_SETVERSION		_IOW('v', 2, int)
230 
231 /*
232  * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS)
233  */
234 #define	FS_SECRM_FL			0x00000001 /* Secure deletion */
235 #define	FS_UNRM_FL			0x00000002 /* Undelete */
236 #define	FS_COMPR_FL			0x00000004 /* Compress file */
237 #define FS_SYNC_FL			0x00000008 /* Synchronous updates */
238 #define FS_IMMUTABLE_FL			0x00000010 /* Immutable file */
239 #define FS_APPEND_FL			0x00000020 /* writes to file may only append */
240 #define FS_NODUMP_FL			0x00000040 /* do not dump file */
241 #define FS_NOATIME_FL			0x00000080 /* do not update atime */
242 /* Reserved for compression usage... */
243 #define FS_DIRTY_FL			0x00000100
244 #define FS_COMPRBLK_FL			0x00000200 /* One or more compressed clusters */
245 #define FS_NOCOMP_FL			0x00000400 /* Don't compress */
246 #define FS_ECOMPR_FL			0x00000800 /* Compression error */
247 /* End compression flags --- maybe not all used */
248 #define FS_BTREE_FL			0x00001000 /* btree format dir */
249 #define FS_INDEX_FL			0x00001000 /* hash-indexed directory */
250 #define FS_IMAGIC_FL			0x00002000 /* AFS directory */
251 #define FS_JOURNAL_DATA_FL		0x00004000 /* Reserved for ext3 */
252 #define FS_NOTAIL_FL			0x00008000 /* file tail should not be merged */
253 #define FS_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
254 #define FS_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
255 #define FS_EXTENT_FL			0x00080000 /* Extents */
256 #define FS_DIRECTIO_FL			0x00100000 /* Use direct i/o */
257 #define FS_RESERVED_FL			0x80000000 /* reserved for ext2 lib */
258 
259 #define FS_FL_USER_VISIBLE		0x0003DFFF /* User visible flags */
260 #define FS_FL_USER_MODIFIABLE		0x000380FF /* User modifiable flags */
261 
262 
263 #define SYNC_FILE_RANGE_WAIT_BEFORE	1
264 #define SYNC_FILE_RANGE_WRITE		2
265 #define SYNC_FILE_RANGE_WAIT_AFTER	4
266 
267 #ifdef __KERNEL__
268 
269 #include <linux/linkage.h>
270 #include <linux/wait.h>
271 #include <linux/types.h>
272 #include <linux/kdev_t.h>
273 #include <linux/dcache.h>
274 #include <linux/namei.h>
275 #include <linux/stat.h>
276 #include <linux/cache.h>
277 #include <linux/kobject.h>
278 #include <linux/list.h>
279 #include <linux/radix-tree.h>
280 #include <linux/prio_tree.h>
281 #include <linux/init.h>
282 #include <linux/pid.h>
283 #include <linux/mutex.h>
284 
285 #include <asm/atomic.h>
286 #include <asm/semaphore.h>
287 #include <asm/byteorder.h>
288 
289 struct hd_geometry;
290 struct iovec;
291 struct nameidata;
292 struct kiocb;
293 struct pipe_inode_info;
294 struct poll_table_struct;
295 struct kstatfs;
296 struct vm_area_struct;
297 struct vfsmount;
298 
299 extern void __init inode_init(unsigned long);
300 extern void __init inode_init_early(void);
301 extern void __init mnt_init(unsigned long);
302 extern void __init files_init(unsigned long);
303 
304 struct buffer_head;
305 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
306 			struct buffer_head *bh_result, int create);
307 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
308 			ssize_t bytes, void *private);
309 
310 /*
311  * Attribute flags.  These should be or-ed together to figure out what
312  * has been changed!
313  */
314 #define ATTR_MODE	1
315 #define ATTR_UID	2
316 #define ATTR_GID	4
317 #define ATTR_SIZE	8
318 #define ATTR_ATIME	16
319 #define ATTR_MTIME	32
320 #define ATTR_CTIME	64
321 #define ATTR_ATIME_SET	128
322 #define ATTR_MTIME_SET	256
323 #define ATTR_FORCE	512	/* Not a change, but a change it */
324 #define ATTR_ATTR_FLAG	1024
325 #define ATTR_KILL_SUID	2048
326 #define ATTR_KILL_SGID	4096
327 #define ATTR_FILE	8192
328 
329 /*
330  * This is the Inode Attributes structure, used for notify_change().  It
331  * uses the above definitions as flags, to know which values have changed.
332  * Also, in this manner, a Filesystem can look at only the values it cares
333  * about.  Basically, these are the attributes that the VFS layer can
334  * request to change from the FS layer.
335  *
336  * Derek Atkins <[email protected]> 94-10-20
337  */
338 struct iattr {
339 	unsigned int	ia_valid;
340 	umode_t		ia_mode;
341 	uid_t		ia_uid;
342 	gid_t		ia_gid;
343 	loff_t		ia_size;
344 	struct timespec	ia_atime;
345 	struct timespec	ia_mtime;
346 	struct timespec	ia_ctime;
347 
348 	/*
349 	 * Not an attribute, but an auxilary info for filesystems wanting to
350 	 * implement an ftruncate() like method.  NOTE: filesystem should
351 	 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
352 	 */
353 	struct file	*ia_file;
354 };
355 
356 /*
357  * Includes for diskquotas.
358  */
359 #include <linux/quota.h>
360 
361 /**
362  * enum positive_aop_returns - aop return codes with specific semantics
363  *
364  * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
365  * 			    completed, that the page is still locked, and
366  * 			    should be considered active.  The VM uses this hint
367  * 			    to return the page to the active list -- it won't
368  * 			    be a candidate for writeback again in the near
369  * 			    future.  Other callers must be careful to unlock
370  * 			    the page if they get this return.  Returned by
371  * 			    writepage();
372  *
373  * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
374  *  			unlocked it and the page might have been truncated.
375  *  			The caller should back up to acquiring a new page and
376  *  			trying again.  The aop will be taking reasonable
377  *  			precautions not to livelock.  If the caller held a page
378  *  			reference, it should drop it before retrying.  Returned
379  *  			by readpage(), prepare_write(), and commit_write().
380  *
381  * address_space_operation functions return these large constants to indicate
382  * special semantics to the caller.  These are much larger than the bytes in a
383  * page to allow for functions that return the number of bytes operated on in a
384  * given page.
385  */
386 
387 enum positive_aop_returns {
388 	AOP_WRITEPAGE_ACTIVATE	= 0x80000,
389 	AOP_TRUNCATED_PAGE	= 0x80001,
390 };
391 
392 /*
393  * oh the beauties of C type declarations.
394  */
395 struct page;
396 struct address_space;
397 struct writeback_control;
398 
399 struct address_space_operations {
400 	int (*writepage)(struct page *page, struct writeback_control *wbc);
401 	int (*readpage)(struct file *, struct page *);
402 	void (*sync_page)(struct page *);
403 
404 	/* Write back some dirty pages from this mapping. */
405 	int (*writepages)(struct address_space *, struct writeback_control *);
406 
407 	/* Set a page dirty.  Return true if this dirtied it */
408 	int (*set_page_dirty)(struct page *page);
409 
410 	int (*readpages)(struct file *filp, struct address_space *mapping,
411 			struct list_head *pages, unsigned nr_pages);
412 
413 	/*
414 	 * ext3 requires that a successful prepare_write() call be followed
415 	 * by a commit_write() call - they must be balanced
416 	 */
417 	int (*prepare_write)(struct file *, struct page *, unsigned, unsigned);
418 	int (*commit_write)(struct file *, struct page *, unsigned, unsigned);
419 	/* Unfortunately this kludge is needed for FIBMAP. Don't use it */
420 	sector_t (*bmap)(struct address_space *, sector_t);
421 	void (*invalidatepage) (struct page *, unsigned long);
422 	int (*releasepage) (struct page *, gfp_t);
423 	ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
424 			loff_t offset, unsigned long nr_segs);
425 	struct page* (*get_xip_page)(struct address_space *, sector_t,
426 			int);
427 	/* migrate the contents of a page to the specified target */
428 	int (*migratepage) (struct address_space *,
429 			struct page *, struct page *);
430 	int (*launder_page) (struct page *);
431 };
432 
433 struct backing_dev_info;
434 struct address_space {
435 	struct inode		*host;		/* owner: inode, block_device */
436 	struct radix_tree_root	page_tree;	/* radix tree of all pages */
437 	rwlock_t		tree_lock;	/* and rwlock protecting it */
438 	unsigned int		i_mmap_writable;/* count VM_SHARED mappings */
439 	struct prio_tree_root	i_mmap;		/* tree of private and shared mappings */
440 	struct list_head	i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
441 	spinlock_t		i_mmap_lock;	/* protect tree, count, list */
442 	unsigned int		truncate_count;	/* Cover race condition with truncate */
443 	unsigned long		nrpages;	/* number of total pages */
444 	pgoff_t			writeback_index;/* writeback starts here */
445 	const struct address_space_operations *a_ops;	/* methods */
446 	unsigned long		flags;		/* error bits/gfp mask */
447 	struct backing_dev_info *backing_dev_info; /* device readahead, etc */
448 	spinlock_t		private_lock;	/* for use by the address_space */
449 	struct list_head	private_list;	/* ditto */
450 	struct address_space	*assoc_mapping;	/* ditto */
451 } __attribute__((aligned(sizeof(long))));
452 	/*
453 	 * On most architectures that alignment is already the case; but
454 	 * must be enforced here for CRIS, to let the least signficant bit
455 	 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
456 	 */
457 
458 struct block_device {
459 	dev_t			bd_dev;  /* not a kdev_t - it's a search key */
460 	struct inode *		bd_inode;	/* will die */
461 	int			bd_openers;
462 	struct mutex		bd_mutex;	/* open/close mutex */
463 	struct semaphore	bd_mount_sem;
464 	struct list_head	bd_inodes;
465 	void *			bd_holder;
466 	int			bd_holders;
467 #ifdef CONFIG_SYSFS
468 	struct list_head	bd_holder_list;
469 #endif
470 	struct block_device *	bd_contains;
471 	unsigned		bd_block_size;
472 	struct hd_struct *	bd_part;
473 	/* number of times partitions within this device have been opened. */
474 	unsigned		bd_part_count;
475 	int			bd_invalidated;
476 	struct gendisk *	bd_disk;
477 	struct list_head	bd_list;
478 	struct backing_dev_info *bd_inode_backing_dev_info;
479 	/*
480 	 * Private data.  You must have bd_claim'ed the block_device
481 	 * to use this.  NOTE:  bd_claim allows an owner to claim
482 	 * the same device multiple times, the owner must take special
483 	 * care to not mess up bd_private for that case.
484 	 */
485 	unsigned long		bd_private;
486 };
487 
488 /*
489  * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
490  * radix trees
491  */
492 #define PAGECACHE_TAG_DIRTY	0
493 #define PAGECACHE_TAG_WRITEBACK	1
494 
495 int mapping_tagged(struct address_space *mapping, int tag);
496 
497 /*
498  * Might pages of this file be mapped into userspace?
499  */
500 static inline int mapping_mapped(struct address_space *mapping)
501 {
502 	return	!prio_tree_empty(&mapping->i_mmap) ||
503 		!list_empty(&mapping->i_mmap_nonlinear);
504 }
505 
506 /*
507  * Might pages of this file have been modified in userspace?
508  * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
509  * marks vma as VM_SHARED if it is shared, and the file was opened for
510  * writing i.e. vma may be mprotected writable even if now readonly.
511  */
512 static inline int mapping_writably_mapped(struct address_space *mapping)
513 {
514 	return mapping->i_mmap_writable != 0;
515 }
516 
517 /*
518  * Use sequence counter to get consistent i_size on 32-bit processors.
519  */
520 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
521 #include <linux/seqlock.h>
522 #define __NEED_I_SIZE_ORDERED
523 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
524 #else
525 #define i_size_ordered_init(inode) do { } while (0)
526 #endif
527 
528 struct inode {
529 	struct hlist_node	i_hash;
530 	struct list_head	i_list;
531 	struct list_head	i_sb_list;
532 	struct list_head	i_dentry;
533 	unsigned long		i_ino;
534 	atomic_t		i_count;
535 	unsigned int		i_nlink;
536 	uid_t			i_uid;
537 	gid_t			i_gid;
538 	dev_t			i_rdev;
539 	unsigned long		i_version;
540 	loff_t			i_size;
541 #ifdef __NEED_I_SIZE_ORDERED
542 	seqcount_t		i_size_seqcount;
543 #endif
544 	struct timespec		i_atime;
545 	struct timespec		i_mtime;
546 	struct timespec		i_ctime;
547 	unsigned int		i_blkbits;
548 	blkcnt_t		i_blocks;
549 	unsigned short          i_bytes;
550 	umode_t			i_mode;
551 	spinlock_t		i_lock;	/* i_blocks, i_bytes, maybe i_size */
552 	struct mutex		i_mutex;
553 	struct rw_semaphore	i_alloc_sem;
554 	const struct inode_operations	*i_op;
555 	const struct file_operations	*i_fop;	/* former ->i_op->default_file_ops */
556 	struct super_block	*i_sb;
557 	struct file_lock	*i_flock;
558 	struct address_space	*i_mapping;
559 	struct address_space	i_data;
560 #ifdef CONFIG_QUOTA
561 	struct dquot		*i_dquot[MAXQUOTAS];
562 #endif
563 	struct list_head	i_devices;
564 	union {
565 		struct pipe_inode_info	*i_pipe;
566 		struct block_device	*i_bdev;
567 		struct cdev		*i_cdev;
568 	};
569 	int			i_cindex;
570 
571 	__u32			i_generation;
572 
573 #ifdef CONFIG_DNOTIFY
574 	unsigned long		i_dnotify_mask; /* Directory notify events */
575 	struct dnotify_struct	*i_dnotify; /* for directory notifications */
576 #endif
577 
578 #ifdef CONFIG_INOTIFY
579 	struct list_head	inotify_watches; /* watches on this inode */
580 	struct mutex		inotify_mutex;	/* protects the watches list */
581 #endif
582 
583 	unsigned long		i_state;
584 	unsigned long		dirtied_when;	/* jiffies of first dirtying */
585 
586 	unsigned int		i_flags;
587 
588 	atomic_t		i_writecount;
589 #ifdef CONFIG_SECURITY
590 	void			*i_security;
591 #endif
592 	void			*i_private; /* fs or device private pointer */
593 };
594 
595 /*
596  * inode->i_mutex nesting subclasses for the lock validator:
597  *
598  * 0: the object of the current VFS operation
599  * 1: parent
600  * 2: child/target
601  * 3: quota file
602  *
603  * The locking order between these classes is
604  * parent -> child -> normal -> xattr -> quota
605  */
606 enum inode_i_mutex_lock_class
607 {
608 	I_MUTEX_NORMAL,
609 	I_MUTEX_PARENT,
610 	I_MUTEX_CHILD,
611 	I_MUTEX_XATTR,
612 	I_MUTEX_QUOTA
613 };
614 
615 extern void inode_double_lock(struct inode *inode1, struct inode *inode2);
616 extern void inode_double_unlock(struct inode *inode1, struct inode *inode2);
617 
618 /*
619  * NOTE: in a 32bit arch with a preemptable kernel and
620  * an UP compile the i_size_read/write must be atomic
621  * with respect to the local cpu (unlike with preempt disabled),
622  * but they don't need to be atomic with respect to other cpus like in
623  * true SMP (so they need either to either locally disable irq around
624  * the read or for example on x86 they can be still implemented as a
625  * cmpxchg8b without the need of the lock prefix). For SMP compiles
626  * and 64bit archs it makes no difference if preempt is enabled or not.
627  */
628 static inline loff_t i_size_read(const struct inode *inode)
629 {
630 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
631 	loff_t i_size;
632 	unsigned int seq;
633 
634 	do {
635 		seq = read_seqcount_begin(&inode->i_size_seqcount);
636 		i_size = inode->i_size;
637 	} while (read_seqcount_retry(&inode->i_size_seqcount, seq));
638 	return i_size;
639 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
640 	loff_t i_size;
641 
642 	preempt_disable();
643 	i_size = inode->i_size;
644 	preempt_enable();
645 	return i_size;
646 #else
647 	return inode->i_size;
648 #endif
649 }
650 
651 /*
652  * NOTE: unlike i_size_read(), i_size_write() does need locking around it
653  * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
654  * can be lost, resulting in subsequent i_size_read() calls spinning forever.
655  */
656 static inline void i_size_write(struct inode *inode, loff_t i_size)
657 {
658 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
659 	write_seqcount_begin(&inode->i_size_seqcount);
660 	inode->i_size = i_size;
661 	write_seqcount_end(&inode->i_size_seqcount);
662 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
663 	preempt_disable();
664 	inode->i_size = i_size;
665 	preempt_enable();
666 #else
667 	inode->i_size = i_size;
668 #endif
669 }
670 
671 static inline unsigned iminor(const struct inode *inode)
672 {
673 	return MINOR(inode->i_rdev);
674 }
675 
676 static inline unsigned imajor(const struct inode *inode)
677 {
678 	return MAJOR(inode->i_rdev);
679 }
680 
681 extern struct block_device *I_BDEV(struct inode *inode);
682 
683 struct fown_struct {
684 	rwlock_t lock;          /* protects pid, uid, euid fields */
685 	struct pid *pid;	/* pid or -pgrp where SIGIO should be sent */
686 	enum pid_type pid_type;	/* Kind of process group SIGIO should be sent to */
687 	uid_t uid, euid;	/* uid/euid of process setting the owner */
688 	int signum;		/* posix.1b rt signal to be delivered on IO */
689 };
690 
691 /*
692  * Track a single file's readahead state
693  */
694 struct file_ra_state {
695 	unsigned long start;		/* Current window */
696 	unsigned long size;
697 	unsigned long flags;		/* ra flags RA_FLAG_xxx*/
698 	unsigned long cache_hit;	/* cache hit count*/
699 	unsigned long prev_index;	/* Cache last read() position */
700 	unsigned long ahead_start;	/* Ahead window */
701 	unsigned long ahead_size;
702 	unsigned long ra_pages;		/* Maximum readahead window */
703 	unsigned long mmap_hit;		/* Cache hit stat for mmap accesses */
704 	unsigned long mmap_miss;	/* Cache miss stat for mmap accesses */
705 	unsigned int prev_offset;	/* Offset where last read() ended in a page */
706 };
707 #define RA_FLAG_MISS 0x01	/* a cache miss occured against this file */
708 #define RA_FLAG_INCACHE 0x02	/* file is already in cache */
709 
710 struct file {
711 	/*
712 	 * fu_list becomes invalid after file_free is called and queued via
713 	 * fu_rcuhead for RCU freeing
714 	 */
715 	union {
716 		struct list_head	fu_list;
717 		struct rcu_head 	fu_rcuhead;
718 	} f_u;
719 	struct path		f_path;
720 #define f_dentry	f_path.dentry
721 #define f_vfsmnt	f_path.mnt
722 	const struct file_operations	*f_op;
723 	atomic_t		f_count;
724 	unsigned int 		f_flags;
725 	mode_t			f_mode;
726 	loff_t			f_pos;
727 	struct fown_struct	f_owner;
728 	unsigned int		f_uid, f_gid;
729 	struct file_ra_state	f_ra;
730 
731 	unsigned long		f_version;
732 #ifdef CONFIG_SECURITY
733 	void			*f_security;
734 #endif
735 	/* needed for tty driver, and maybe others */
736 	void			*private_data;
737 
738 #ifdef CONFIG_EPOLL
739 	/* Used by fs/eventpoll.c to link all the hooks to this file */
740 	struct list_head	f_ep_links;
741 	spinlock_t		f_ep_lock;
742 #endif /* #ifdef CONFIG_EPOLL */
743 	struct address_space	*f_mapping;
744 };
745 extern spinlock_t files_lock;
746 #define file_list_lock() spin_lock(&files_lock);
747 #define file_list_unlock() spin_unlock(&files_lock);
748 
749 #define get_file(x)	atomic_inc(&(x)->f_count)
750 #define file_count(x)	atomic_read(&(x)->f_count)
751 
752 #define	MAX_NON_LFS	((1UL<<31) - 1)
753 
754 /* Page cache limit. The filesystems should put that into their s_maxbytes
755    limits, otherwise bad things can happen in VM. */
756 #if BITS_PER_LONG==32
757 #define MAX_LFS_FILESIZE	(((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
758 #elif BITS_PER_LONG==64
759 #define MAX_LFS_FILESIZE 	0x7fffffffffffffffUL
760 #endif
761 
762 #define FL_POSIX	1
763 #define FL_FLOCK	2
764 #define FL_ACCESS	8	/* not trying to lock, just looking */
765 #define FL_EXISTS	16	/* when unlocking, test for existence */
766 #define FL_LEASE	32	/* lease held on this file */
767 #define FL_CLOSE	64	/* unlock on close */
768 #define FL_SLEEP	128	/* A blocking lock */
769 
770 /*
771  * The POSIX file lock owner is determined by
772  * the "struct files_struct" in the thread group
773  * (or NULL for no owner - BSD locks).
774  *
775  * Lockd stuffs a "host" pointer into this.
776  */
777 typedef struct files_struct *fl_owner_t;
778 
779 struct file_lock_operations {
780 	void (*fl_insert)(struct file_lock *);	/* lock insertion callback */
781 	void (*fl_remove)(struct file_lock *);	/* lock removal callback */
782 	void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
783 	void (*fl_release_private)(struct file_lock *);
784 };
785 
786 struct lock_manager_operations {
787 	int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
788 	void (*fl_notify)(struct file_lock *);	/* unblock callback */
789 	void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
790 	void (*fl_release_private)(struct file_lock *);
791 	void (*fl_break)(struct file_lock *);
792 	int (*fl_mylease)(struct file_lock *, struct file_lock *);
793 	int (*fl_change)(struct file_lock **, int);
794 };
795 
796 /* that will die - we need it for nfs_lock_info */
797 #include <linux/nfs_fs_i.h>
798 
799 struct file_lock {
800 	struct file_lock *fl_next;	/* singly linked list for this inode  */
801 	struct list_head fl_link;	/* doubly linked list of all locks */
802 	struct list_head fl_block;	/* circular list of blocked processes */
803 	fl_owner_t fl_owner;
804 	unsigned int fl_pid;
805 	wait_queue_head_t fl_wait;
806 	struct file *fl_file;
807 	unsigned char fl_flags;
808 	unsigned char fl_type;
809 	loff_t fl_start;
810 	loff_t fl_end;
811 
812 	struct fasync_struct *	fl_fasync; /* for lease break notifications */
813 	unsigned long fl_break_time;	/* for nonblocking lease breaks */
814 
815 	struct file_lock_operations *fl_ops;	/* Callbacks for filesystems */
816 	struct lock_manager_operations *fl_lmops;	/* Callbacks for lockmanagers */
817 	union {
818 		struct nfs_lock_info	nfs_fl;
819 		struct nfs4_lock_info	nfs4_fl;
820 	} fl_u;
821 };
822 
823 /* The following constant reflects the upper bound of the file/locking space */
824 #ifndef OFFSET_MAX
825 #define INT_LIMIT(x)	(~((x)1 << (sizeof(x)*8 - 1)))
826 #define OFFSET_MAX	INT_LIMIT(loff_t)
827 #define OFFT_OFFSET_MAX	INT_LIMIT(off_t)
828 #endif
829 
830 #include <linux/fcntl.h>
831 
832 extern int fcntl_getlk(struct file *, struct flock __user *);
833 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
834 			struct flock __user *);
835 
836 #if BITS_PER_LONG == 32
837 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
838 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
839 			struct flock64 __user *);
840 #endif
841 
842 extern void send_sigio(struct fown_struct *fown, int fd, int band);
843 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
844 extern int fcntl_getlease(struct file *filp);
845 
846 /* fs/sync.c */
847 extern int do_sync_mapping_range(struct address_space *mapping, loff_t offset,
848 			loff_t endbyte, unsigned int flags);
849 static inline int do_sync_file_range(struct file *file, loff_t offset,
850 			loff_t endbyte, unsigned int flags)
851 {
852 	return do_sync_mapping_range(file->f_mapping, offset, endbyte, flags);
853 }
854 
855 /* fs/locks.c */
856 extern void locks_init_lock(struct file_lock *);
857 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
858 extern void locks_remove_posix(struct file *, fl_owner_t);
859 extern void locks_remove_flock(struct file *);
860 extern int posix_test_lock(struct file *, struct file_lock *, struct file_lock *);
861 extern int posix_lock_file_conf(struct file *, struct file_lock *, struct file_lock *);
862 extern int posix_lock_file(struct file *, struct file_lock *);
863 extern int posix_lock_file_wait(struct file *, struct file_lock *);
864 extern int posix_unblock_lock(struct file *, struct file_lock *);
865 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
866 extern int __break_lease(struct inode *inode, unsigned int flags);
867 extern void lease_get_mtime(struct inode *, struct timespec *time);
868 extern int setlease(struct file *, long, struct file_lock **);
869 extern int lease_modify(struct file_lock **, int);
870 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
871 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
872 
873 struct fasync_struct {
874 	int	magic;
875 	int	fa_fd;
876 	struct	fasync_struct	*fa_next; /* singly linked list */
877 	struct	file 		*fa_file;
878 };
879 
880 #define FASYNC_MAGIC 0x4601
881 
882 /* SMP safe fasync helpers: */
883 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
884 /* can be called from interrupts */
885 extern void kill_fasync(struct fasync_struct **, int, int);
886 /* only for net: no internal synchronization */
887 extern void __kill_fasync(struct fasync_struct *, int, int);
888 
889 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
890 extern int f_setown(struct file *filp, unsigned long arg, int force);
891 extern void f_delown(struct file *filp);
892 extern pid_t f_getown(struct file *filp);
893 extern int send_sigurg(struct fown_struct *fown);
894 
895 /*
896  *	Umount options
897  */
898 
899 #define MNT_FORCE	0x00000001	/* Attempt to forcibily umount */
900 #define MNT_DETACH	0x00000002	/* Just detach from the tree */
901 #define MNT_EXPIRE	0x00000004	/* Mark for expiry */
902 
903 extern struct list_head super_blocks;
904 extern spinlock_t sb_lock;
905 
906 #define sb_entry(list)	list_entry((list), struct super_block, s_list)
907 #define S_BIAS (1<<30)
908 struct super_block {
909 	struct list_head	s_list;		/* Keep this first */
910 	dev_t			s_dev;		/* search index; _not_ kdev_t */
911 	unsigned long		s_blocksize;
912 	unsigned char		s_blocksize_bits;
913 	unsigned char		s_dirt;
914 	unsigned long long	s_maxbytes;	/* Max file size */
915 	struct file_system_type	*s_type;
916 	const struct super_operations	*s_op;
917 	struct dquot_operations	*dq_op;
918  	struct quotactl_ops	*s_qcop;
919 	struct export_operations *s_export_op;
920 	unsigned long		s_flags;
921 	unsigned long		s_magic;
922 	struct dentry		*s_root;
923 	struct rw_semaphore	s_umount;
924 	struct mutex		s_lock;
925 	int			s_count;
926 	int			s_syncing;
927 	int			s_need_sync_fs;
928 	atomic_t		s_active;
929 #ifdef CONFIG_SECURITY
930 	void                    *s_security;
931 #endif
932 	struct xattr_handler	**s_xattr;
933 
934 	struct list_head	s_inodes;	/* all inodes */
935 	struct list_head	s_dirty;	/* dirty inodes */
936 	struct list_head	s_io;		/* parked for writeback */
937 	struct hlist_head	s_anon;		/* anonymous dentries for (nfs) exporting */
938 	struct list_head	s_files;
939 
940 	struct block_device	*s_bdev;
941 	struct list_head	s_instances;
942 	struct quota_info	s_dquot;	/* Diskquota specific options */
943 
944 	int			s_frozen;
945 	wait_queue_head_t	s_wait_unfrozen;
946 
947 	char s_id[32];				/* Informational name */
948 
949 	void 			*s_fs_info;	/* Filesystem private info */
950 
951 	/*
952 	 * The next field is for VFS *only*. No filesystems have any business
953 	 * even looking at it. You had been warned.
954 	 */
955 	struct mutex s_vfs_rename_mutex;	/* Kludge */
956 
957 	/* Granularity of c/m/atime in ns.
958 	   Cannot be worse than a second */
959 	u32		   s_time_gran;
960 };
961 
962 extern struct timespec current_fs_time(struct super_block *sb);
963 
964 /*
965  * Snapshotting support.
966  */
967 enum {
968 	SB_UNFROZEN = 0,
969 	SB_FREEZE_WRITE	= 1,
970 	SB_FREEZE_TRANS = 2,
971 };
972 
973 #define vfs_check_frozen(sb, level) \
974 	wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
975 
976 #define get_fs_excl() atomic_inc(&current->fs_excl)
977 #define put_fs_excl() atomic_dec(&current->fs_excl)
978 #define has_fs_excl() atomic_read(&current->fs_excl)
979 
980 /* not quite ready to be deprecated, but... */
981 extern void lock_super(struct super_block *);
982 extern void unlock_super(struct super_block *);
983 
984 /*
985  * VFS helper functions..
986  */
987 extern int vfs_permission(struct nameidata *, int);
988 extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
989 extern int vfs_mkdir(struct inode *, struct dentry *, int);
990 extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
991 extern int vfs_symlink(struct inode *, struct dentry *, const char *, int);
992 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
993 extern int vfs_rmdir(struct inode *, struct dentry *);
994 extern int vfs_unlink(struct inode *, struct dentry *);
995 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
996 
997 /*
998  * VFS dentry helper functions.
999  */
1000 extern void dentry_unhash(struct dentry *dentry);
1001 
1002 /*
1003  * VFS file helper functions.
1004  */
1005 extern int file_permission(struct file *, int);
1006 
1007 /*
1008  * File types
1009  *
1010  * NOTE! These match bits 12..15 of stat.st_mode
1011  * (ie "(i_mode >> 12) & 15").
1012  */
1013 #define DT_UNKNOWN	0
1014 #define DT_FIFO		1
1015 #define DT_CHR		2
1016 #define DT_DIR		4
1017 #define DT_BLK		6
1018 #define DT_REG		8
1019 #define DT_LNK		10
1020 #define DT_SOCK		12
1021 #define DT_WHT		14
1022 
1023 #define OSYNC_METADATA	(1<<0)
1024 #define OSYNC_DATA	(1<<1)
1025 #define OSYNC_INODE	(1<<2)
1026 int generic_osync_inode(struct inode *, struct address_space *, int);
1027 
1028 /*
1029  * This is the "filldir" function type, used by readdir() to let
1030  * the kernel specify what kind of dirent layout it wants to have.
1031  * This allows the kernel to read directories into kernel space or
1032  * to have different dirent layouts depending on the binary type.
1033  */
1034 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1035 
1036 struct block_device_operations {
1037 	int (*open) (struct inode *, struct file *);
1038 	int (*release) (struct inode *, struct file *);
1039 	int (*ioctl) (struct inode *, struct file *, unsigned, unsigned long);
1040 	long (*unlocked_ioctl) (struct file *, unsigned, unsigned long);
1041 	long (*compat_ioctl) (struct file *, unsigned, unsigned long);
1042 	int (*direct_access) (struct block_device *, sector_t, unsigned long *);
1043 	int (*media_changed) (struct gendisk *);
1044 	int (*revalidate_disk) (struct gendisk *);
1045 	int (*getgeo)(struct block_device *, struct hd_geometry *);
1046 	struct module *owner;
1047 };
1048 
1049 /*
1050  * "descriptor" for what we're up to with a read for sendfile().
1051  * This allows us to use the same read code yet
1052  * have multiple different users of the data that
1053  * we read from a file.
1054  *
1055  * The simplest case just copies the data to user
1056  * mode.
1057  */
1058 typedef struct {
1059 	size_t written;
1060 	size_t count;
1061 	union {
1062 		char __user * buf;
1063 		void *data;
1064 	} arg;
1065 	int error;
1066 } read_descriptor_t;
1067 
1068 typedef int (*read_actor_t)(read_descriptor_t *, struct page *, unsigned long, unsigned long);
1069 
1070 /* These macros are for out of kernel modules to test that
1071  * the kernel supports the unlocked_ioctl and compat_ioctl
1072  * fields in struct file_operations. */
1073 #define HAVE_COMPAT_IOCTL 1
1074 #define HAVE_UNLOCKED_IOCTL 1
1075 
1076 /*
1077  * NOTE:
1078  * read, write, poll, fsync, readv, writev, unlocked_ioctl and compat_ioctl
1079  * can be called without the big kernel lock held in all filesystems.
1080  */
1081 struct file_operations {
1082 	struct module *owner;
1083 	loff_t (*llseek) (struct file *, loff_t, int);
1084 	ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1085 	ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1086 	ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1087 	ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1088 	int (*readdir) (struct file *, void *, filldir_t);
1089 	unsigned int (*poll) (struct file *, struct poll_table_struct *);
1090 	int (*ioctl) (struct inode *, struct file *, unsigned int, unsigned long);
1091 	long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1092 	long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1093 	int (*mmap) (struct file *, struct vm_area_struct *);
1094 	int (*open) (struct inode *, struct file *);
1095 	int (*flush) (struct file *, fl_owner_t id);
1096 	int (*release) (struct inode *, struct file *);
1097 	int (*fsync) (struct file *, struct dentry *, int datasync);
1098 	int (*aio_fsync) (struct kiocb *, int datasync);
1099 	int (*fasync) (int, struct file *, int);
1100 	int (*lock) (struct file *, int, struct file_lock *);
1101 	ssize_t (*sendfile) (struct file *, loff_t *, size_t, read_actor_t, void *);
1102 	ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1103 	unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1104 	int (*check_flags)(int);
1105 	int (*dir_notify)(struct file *filp, unsigned long arg);
1106 	int (*flock) (struct file *, int, struct file_lock *);
1107 	ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1108 	ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1109 };
1110 
1111 struct inode_operations {
1112 	int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1113 	struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1114 	int (*link) (struct dentry *,struct inode *,struct dentry *);
1115 	int (*unlink) (struct inode *,struct dentry *);
1116 	int (*symlink) (struct inode *,struct dentry *,const char *);
1117 	int (*mkdir) (struct inode *,struct dentry *,int);
1118 	int (*rmdir) (struct inode *,struct dentry *);
1119 	int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1120 	int (*rename) (struct inode *, struct dentry *,
1121 			struct inode *, struct dentry *);
1122 	int (*readlink) (struct dentry *, char __user *,int);
1123 	void * (*follow_link) (struct dentry *, struct nameidata *);
1124 	void (*put_link) (struct dentry *, struct nameidata *, void *);
1125 	void (*truncate) (struct inode *);
1126 	int (*permission) (struct inode *, int, struct nameidata *);
1127 	int (*setattr) (struct dentry *, struct iattr *);
1128 	int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1129 	int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1130 	ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1131 	ssize_t (*listxattr) (struct dentry *, char *, size_t);
1132 	int (*removexattr) (struct dentry *, const char *);
1133 	void (*truncate_range)(struct inode *, loff_t, loff_t);
1134 };
1135 
1136 struct seq_file;
1137 
1138 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1139 				unsigned long nr_segs, unsigned long fast_segs,
1140 				struct iovec *fast_pointer,
1141 				struct iovec **ret_pointer);
1142 
1143 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1144 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1145 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1146 		unsigned long, loff_t *);
1147 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1148 		unsigned long, loff_t *);
1149 
1150 /*
1151  * NOTE: write_inode, delete_inode, clear_inode, put_inode can be called
1152  * without the big kernel lock held in all filesystems.
1153  */
1154 struct super_operations {
1155    	struct inode *(*alloc_inode)(struct super_block *sb);
1156 	void (*destroy_inode)(struct inode *);
1157 
1158 	void (*read_inode) (struct inode *);
1159 
1160    	void (*dirty_inode) (struct inode *);
1161 	int (*write_inode) (struct inode *, int);
1162 	void (*put_inode) (struct inode *);
1163 	void (*drop_inode) (struct inode *);
1164 	void (*delete_inode) (struct inode *);
1165 	void (*put_super) (struct super_block *);
1166 	void (*write_super) (struct super_block *);
1167 	int (*sync_fs)(struct super_block *sb, int wait);
1168 	void (*write_super_lockfs) (struct super_block *);
1169 	void (*unlockfs) (struct super_block *);
1170 	int (*statfs) (struct dentry *, struct kstatfs *);
1171 	int (*remount_fs) (struct super_block *, int *, char *);
1172 	void (*clear_inode) (struct inode *);
1173 	void (*umount_begin) (struct vfsmount *, int);
1174 
1175 	int (*show_options)(struct seq_file *, struct vfsmount *);
1176 	int (*show_stats)(struct seq_file *, struct vfsmount *);
1177 #ifdef CONFIG_QUOTA
1178 	ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1179 	ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1180 #endif
1181 };
1182 
1183 /* Inode state bits.  Protected by inode_lock. */
1184 #define I_DIRTY_SYNC		1 /* Not dirty enough for O_DATASYNC */
1185 #define I_DIRTY_DATASYNC	2 /* Data-related inode changes pending */
1186 #define I_DIRTY_PAGES		4 /* Data-related inode changes pending */
1187 #define __I_LOCK		3
1188 #define I_LOCK			(1 << __I_LOCK)
1189 #define I_FREEING		16
1190 #define I_CLEAR			32
1191 #define I_NEW			64
1192 #define I_WILL_FREE		128
1193 
1194 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1195 
1196 extern void __mark_inode_dirty(struct inode *, int);
1197 static inline void mark_inode_dirty(struct inode *inode)
1198 {
1199 	__mark_inode_dirty(inode, I_DIRTY);
1200 }
1201 
1202 static inline void mark_inode_dirty_sync(struct inode *inode)
1203 {
1204 	__mark_inode_dirty(inode, I_DIRTY_SYNC);
1205 }
1206 
1207 static inline void inc_nlink(struct inode *inode)
1208 {
1209 	inode->i_nlink++;
1210 }
1211 
1212 static inline void inode_inc_link_count(struct inode *inode)
1213 {
1214 	inc_nlink(inode);
1215 	mark_inode_dirty(inode);
1216 }
1217 
1218 static inline void drop_nlink(struct inode *inode)
1219 {
1220 	inode->i_nlink--;
1221 }
1222 
1223 static inline void clear_nlink(struct inode *inode)
1224 {
1225 	inode->i_nlink = 0;
1226 }
1227 
1228 static inline void inode_dec_link_count(struct inode *inode)
1229 {
1230 	drop_nlink(inode);
1231 	mark_inode_dirty(inode);
1232 }
1233 
1234 extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
1235 static inline void file_accessed(struct file *file)
1236 {
1237 	if (!(file->f_flags & O_NOATIME))
1238 		touch_atime(file->f_path.mnt, file->f_path.dentry);
1239 }
1240 
1241 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1242 
1243 /**
1244  * struct export_operations - for nfsd to communicate with file systems
1245  * @decode_fh:      decode a file handle fragment and return a &struct dentry
1246  * @encode_fh:      encode a file handle fragment from a dentry
1247  * @get_name:       find the name for a given inode in a given directory
1248  * @get_parent:     find the parent of a given directory
1249  * @get_dentry:     find a dentry for the inode given a file handle sub-fragment
1250  * @find_exported_dentry:
1251  *	set by the exporting module to a standard helper function.
1252  *
1253  * Description:
1254  *    The export_operations structure provides a means for nfsd to communicate
1255  *    with a particular exported file system  - particularly enabling nfsd and
1256  *    the filesystem to co-operate when dealing with file handles.
1257  *
1258  *    export_operations contains two basic operation for dealing with file
1259  *    handles, decode_fh() and encode_fh(), and allows for some other
1260  *    operations to be defined which standard helper routines use to get
1261  *    specific information from the filesystem.
1262  *
1263  *    nfsd encodes information use to determine which filesystem a filehandle
1264  *    applies to in the initial part of the file handle.  The remainder, termed
1265  *    a file handle fragment, is controlled completely by the filesystem.  The
1266  *    standard helper routines assume that this fragment will contain one or
1267  *    two sub-fragments, one which identifies the file, and one which may be
1268  *    used to identify the (a) directory containing the file.
1269  *
1270  *    In some situations, nfsd needs to get a dentry which is connected into a
1271  *    specific part of the file tree.  To allow for this, it passes the
1272  *    function acceptable() together with a @context which can be used to see
1273  *    if the dentry is acceptable.  As there can be multiple dentrys for a
1274  *    given file, the filesystem should check each one for acceptability before
1275  *    looking for the next.  As soon as an acceptable one is found, it should
1276  *    be returned.
1277  *
1278  * decode_fh:
1279  *    @decode_fh is given a &struct super_block (@sb), a file handle fragment
1280  *    (@fh, @fh_len) and an acceptability testing function (@acceptable,
1281  *    @context).  It should return a &struct dentry which refers to the same
1282  *    file that the file handle fragment refers to,  and which passes the
1283  *    acceptability test.  If it cannot, it should return a %NULL pointer if
1284  *    the file was found but no acceptable &dentries were available, or a
1285  *    %ERR_PTR error code indicating why it couldn't be found (e.g. %ENOENT or
1286  *    %ENOMEM).
1287  *
1288  * encode_fh:
1289  *    @encode_fh should store in the file handle fragment @fh (using at most
1290  *    @max_len bytes) information that can be used by @decode_fh to recover the
1291  *    file refered to by the &struct dentry @de.  If the @connectable flag is
1292  *    set, the encode_fh() should store sufficient information so that a good
1293  *    attempt can be made to find not only the file but also it's place in the
1294  *    filesystem.   This typically means storing a reference to de->d_parent in
1295  *    the filehandle fragment.  encode_fh() should return the number of bytes
1296  *    stored or a negative error code such as %-ENOSPC
1297  *
1298  * get_name:
1299  *    @get_name should find a name for the given @child in the given @parent
1300  *    directory.  The name should be stored in the @name (with the
1301  *    understanding that it is already pointing to a a %NAME_MAX+1 sized
1302  *    buffer.   get_name() should return %0 on success, a negative error code
1303  *    or error.  @get_name will be called without @parent->i_mutex held.
1304  *
1305  * get_parent:
1306  *    @get_parent should find the parent directory for the given @child which
1307  *    is also a directory.  In the event that it cannot be found, or storage
1308  *    space cannot be allocated, a %ERR_PTR should be returned.
1309  *
1310  * get_dentry:
1311  *    Given a &super_block (@sb) and a pointer to a file-system specific inode
1312  *    identifier, possibly an inode number, (@inump) get_dentry() should find
1313  *    the identified inode and return a dentry for that inode.  Any suitable
1314  *    dentry can be returned including, if necessary, a new dentry created with
1315  *    d_alloc_root.  The caller can then find any other extant dentrys by
1316  *    following the d_alias links.  If a new dentry was created using
1317  *    d_alloc_root, DCACHE_NFSD_DISCONNECTED should be set, and the dentry
1318  *    should be d_rehash()ed.
1319  *
1320  *    If the inode cannot be found, either a %NULL pointer or an %ERR_PTR code
1321  *    can be returned.  The @inump will be whatever was passed to
1322  *    nfsd_find_fh_dentry() in either the @obj or @parent parameters.
1323  *
1324  * Locking rules:
1325  *    get_parent is called with child->d_inode->i_mutex down
1326  *    get_name is not (which is possibly inconsistent)
1327  */
1328 
1329 struct export_operations {
1330 	struct dentry *(*decode_fh)(struct super_block *sb, __u32 *fh, int fh_len, int fh_type,
1331 			 int (*acceptable)(void *context, struct dentry *de),
1332 			 void *context);
1333 	int (*encode_fh)(struct dentry *de, __u32 *fh, int *max_len,
1334 			 int connectable);
1335 
1336 	/* the following are only called from the filesystem itself */
1337 	int (*get_name)(struct dentry *parent, char *name,
1338 			struct dentry *child);
1339 	struct dentry * (*get_parent)(struct dentry *child);
1340 	struct dentry * (*get_dentry)(struct super_block *sb, void *inump);
1341 
1342 	/* This is set by the exporting module to a standard helper */
1343 	struct dentry * (*find_exported_dentry)(
1344 		struct super_block *sb, void *obj, void *parent,
1345 		int (*acceptable)(void *context, struct dentry *de),
1346 		void *context);
1347 
1348 
1349 };
1350 
1351 extern struct dentry *
1352 find_exported_dentry(struct super_block *sb, void *obj, void *parent,
1353 		     int (*acceptable)(void *context, struct dentry *de),
1354 		     void *context);
1355 
1356 struct file_system_type {
1357 	const char *name;
1358 	int fs_flags;
1359 	int (*get_sb) (struct file_system_type *, int,
1360 		       const char *, void *, struct vfsmount *);
1361 	void (*kill_sb) (struct super_block *);
1362 	struct module *owner;
1363 	struct file_system_type * next;
1364 	struct list_head fs_supers;
1365 	struct lock_class_key s_lock_key;
1366 	struct lock_class_key s_umount_key;
1367 };
1368 
1369 extern int get_sb_bdev(struct file_system_type *fs_type,
1370 	int flags, const char *dev_name, void *data,
1371 	int (*fill_super)(struct super_block *, void *, int),
1372 	struct vfsmount *mnt);
1373 extern int get_sb_single(struct file_system_type *fs_type,
1374 	int flags, void *data,
1375 	int (*fill_super)(struct super_block *, void *, int),
1376 	struct vfsmount *mnt);
1377 extern int get_sb_nodev(struct file_system_type *fs_type,
1378 	int flags, void *data,
1379 	int (*fill_super)(struct super_block *, void *, int),
1380 	struct vfsmount *mnt);
1381 void generic_shutdown_super(struct super_block *sb);
1382 void kill_block_super(struct super_block *sb);
1383 void kill_anon_super(struct super_block *sb);
1384 void kill_litter_super(struct super_block *sb);
1385 void deactivate_super(struct super_block *sb);
1386 int set_anon_super(struct super_block *s, void *data);
1387 struct super_block *sget(struct file_system_type *type,
1388 			int (*test)(struct super_block *,void *),
1389 			int (*set)(struct super_block *,void *),
1390 			void *data);
1391 extern int get_sb_pseudo(struct file_system_type *, char *,
1392 	const struct super_operations *ops, unsigned long,
1393 	struct vfsmount *mnt);
1394 extern int simple_set_mnt(struct vfsmount *mnt, struct super_block *sb);
1395 int __put_super(struct super_block *sb);
1396 int __put_super_and_need_restart(struct super_block *sb);
1397 void unnamed_dev_init(void);
1398 
1399 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1400 #define fops_get(fops) \
1401 	(((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1402 #define fops_put(fops) \
1403 	do { if (fops) module_put((fops)->owner); } while(0)
1404 
1405 extern int register_filesystem(struct file_system_type *);
1406 extern int unregister_filesystem(struct file_system_type *);
1407 extern struct vfsmount *kern_mount(struct file_system_type *);
1408 extern int may_umount_tree(struct vfsmount *);
1409 extern int may_umount(struct vfsmount *);
1410 extern void umount_tree(struct vfsmount *, int, struct list_head *);
1411 extern void release_mounts(struct list_head *);
1412 extern long do_mount(char *, char *, char *, unsigned long, void *);
1413 extern struct vfsmount *copy_tree(struct vfsmount *, struct dentry *, int);
1414 extern void mnt_set_mountpoint(struct vfsmount *, struct dentry *,
1415 				  struct vfsmount *);
1416 
1417 extern int vfs_statfs(struct dentry *, struct kstatfs *);
1418 
1419 /* /sys/fs */
1420 extern struct kset fs_subsys;
1421 
1422 #define FLOCK_VERIFY_READ  1
1423 #define FLOCK_VERIFY_WRITE 2
1424 
1425 extern int locks_mandatory_locked(struct inode *);
1426 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1427 
1428 /*
1429  * Candidates for mandatory locking have the setgid bit set
1430  * but no group execute bit -  an otherwise meaningless combination.
1431  */
1432 #define MANDATORY_LOCK(inode) \
1433 	(IS_MANDLOCK(inode) && ((inode)->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID)
1434 
1435 static inline int locks_verify_locked(struct inode *inode)
1436 {
1437 	if (MANDATORY_LOCK(inode))
1438 		return locks_mandatory_locked(inode);
1439 	return 0;
1440 }
1441 
1442 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1443 
1444 static inline int locks_verify_truncate(struct inode *inode,
1445 				    struct file *filp,
1446 				    loff_t size)
1447 {
1448 	if (inode->i_flock && MANDATORY_LOCK(inode))
1449 		return locks_mandatory_area(
1450 			FLOCK_VERIFY_WRITE, inode, filp,
1451 			size < inode->i_size ? size : inode->i_size,
1452 			(size < inode->i_size ? inode->i_size - size
1453 			 : size - inode->i_size)
1454 		);
1455 	return 0;
1456 }
1457 
1458 static inline int break_lease(struct inode *inode, unsigned int mode)
1459 {
1460 	if (inode->i_flock)
1461 		return __break_lease(inode, mode);
1462 	return 0;
1463 }
1464 
1465 /* fs/open.c */
1466 
1467 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
1468 		       struct file *filp);
1469 extern long do_sys_open(int fdf, const char __user *filename, int flags,
1470 			int mode);
1471 extern struct file *filp_open(const char *, int, int);
1472 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int);
1473 extern int filp_close(struct file *, fl_owner_t id);
1474 extern char * getname(const char __user *);
1475 
1476 /* fs/dcache.c */
1477 extern void __init vfs_caches_init_early(void);
1478 extern void __init vfs_caches_init(unsigned long);
1479 
1480 extern struct kmem_cache *names_cachep;
1481 
1482 #define __getname()	kmem_cache_alloc(names_cachep, GFP_KERNEL)
1483 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
1484 #ifndef CONFIG_AUDITSYSCALL
1485 #define putname(name)   __putname(name)
1486 #else
1487 extern void putname(const char *name);
1488 #endif
1489 
1490 #ifdef CONFIG_BLOCK
1491 extern int register_blkdev(unsigned int, const char *);
1492 extern int unregister_blkdev(unsigned int, const char *);
1493 extern struct block_device *bdget(dev_t);
1494 extern void bd_set_size(struct block_device *, loff_t size);
1495 extern void bd_forget(struct inode *inode);
1496 extern void bdput(struct block_device *);
1497 extern struct block_device *open_by_devnum(dev_t, unsigned);
1498 extern const struct address_space_operations def_blk_aops;
1499 #else
1500 static inline void bd_forget(struct inode *inode) {}
1501 #endif
1502 extern const struct file_operations def_blk_fops;
1503 extern const struct file_operations def_chr_fops;
1504 extern const struct file_operations bad_sock_fops;
1505 extern const struct file_operations def_fifo_fops;
1506 #ifdef CONFIG_BLOCK
1507 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
1508 extern int blkdev_ioctl(struct inode *, struct file *, unsigned, unsigned long);
1509 extern int blkdev_driver_ioctl(struct inode *inode, struct file *file,
1510 			       struct gendisk *disk, unsigned cmd,
1511 			       unsigned long arg);
1512 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
1513 extern int blkdev_get(struct block_device *, mode_t, unsigned);
1514 extern int blkdev_put(struct block_device *);
1515 extern int bd_claim(struct block_device *, void *);
1516 extern void bd_release(struct block_device *);
1517 #ifdef CONFIG_SYSFS
1518 extern int bd_claim_by_disk(struct block_device *, void *, struct gendisk *);
1519 extern void bd_release_from_disk(struct block_device *, struct gendisk *);
1520 #else
1521 #define bd_claim_by_disk(bdev, holder, disk)	bd_claim(bdev, holder)
1522 #define bd_release_from_disk(bdev, disk)	bd_release(bdev)
1523 #endif
1524 #endif
1525 
1526 /* fs/char_dev.c */
1527 #define CHRDEV_MAJOR_HASH_SIZE	255
1528 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
1529 extern int register_chrdev_region(dev_t, unsigned, const char *);
1530 extern int register_chrdev(unsigned int, const char *,
1531 			   const struct file_operations *);
1532 extern int unregister_chrdev(unsigned int, const char *);
1533 extern void unregister_chrdev_region(dev_t, unsigned);
1534 extern int chrdev_open(struct inode *, struct file *);
1535 extern void chrdev_show(struct seq_file *,off_t);
1536 
1537 /* fs/block_dev.c */
1538 #define BDEVNAME_SIZE	32	/* Largest string for a blockdev identifier */
1539 
1540 #ifdef CONFIG_BLOCK
1541 #define BLKDEV_MAJOR_HASH_SIZE	255
1542 extern const char *__bdevname(dev_t, char *buffer);
1543 extern const char *bdevname(struct block_device *bdev, char *buffer);
1544 extern struct block_device *lookup_bdev(const char *);
1545 extern struct block_device *open_bdev_excl(const char *, int, void *);
1546 extern void close_bdev_excl(struct block_device *);
1547 extern void blkdev_show(struct seq_file *,off_t);
1548 #else
1549 #define BLKDEV_MAJOR_HASH_SIZE	0
1550 #endif
1551 
1552 extern void init_special_inode(struct inode *, umode_t, dev_t);
1553 
1554 /* Invalid inode operations -- fs/bad_inode.c */
1555 extern void make_bad_inode(struct inode *);
1556 extern int is_bad_inode(struct inode *);
1557 
1558 extern const struct file_operations read_fifo_fops;
1559 extern const struct file_operations write_fifo_fops;
1560 extern const struct file_operations rdwr_fifo_fops;
1561 
1562 extern int fs_may_remount_ro(struct super_block *);
1563 
1564 #ifdef CONFIG_BLOCK
1565 /*
1566  * return READ, READA, or WRITE
1567  */
1568 #define bio_rw(bio)		((bio)->bi_rw & (RW_MASK | RWA_MASK))
1569 
1570 /*
1571  * return data direction, READ or WRITE
1572  */
1573 #define bio_data_dir(bio)	((bio)->bi_rw & 1)
1574 
1575 extern int check_disk_change(struct block_device *);
1576 extern int __invalidate_device(struct block_device *);
1577 extern int invalidate_partition(struct gendisk *, int);
1578 #endif
1579 extern int invalidate_inodes(struct super_block *);
1580 unsigned long invalidate_mapping_pages(struct address_space *mapping,
1581 					pgoff_t start, pgoff_t end);
1582 
1583 static inline unsigned long __deprecated
1584 invalidate_inode_pages(struct address_space *mapping)
1585 {
1586 	return invalidate_mapping_pages(mapping, 0, ~0UL);
1587 }
1588 
1589 static inline void invalidate_remote_inode(struct inode *inode)
1590 {
1591 	if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
1592 	    S_ISLNK(inode->i_mode))
1593 		invalidate_mapping_pages(inode->i_mapping, 0, -1);
1594 }
1595 extern int invalidate_inode_pages2(struct address_space *mapping);
1596 extern int invalidate_inode_pages2_range(struct address_space *mapping,
1597 					 pgoff_t start, pgoff_t end);
1598 extern int write_inode_now(struct inode *, int);
1599 extern int filemap_fdatawrite(struct address_space *);
1600 extern int filemap_flush(struct address_space *);
1601 extern int filemap_fdatawait(struct address_space *);
1602 extern int filemap_write_and_wait(struct address_space *mapping);
1603 extern int filemap_write_and_wait_range(struct address_space *mapping,
1604 				        loff_t lstart, loff_t lend);
1605 extern int wait_on_page_writeback_range(struct address_space *mapping,
1606 				pgoff_t start, pgoff_t end);
1607 extern int __filemap_fdatawrite_range(struct address_space *mapping,
1608 				loff_t start, loff_t end, int sync_mode);
1609 
1610 extern long do_fsync(struct file *file, int datasync);
1611 extern void sync_supers(void);
1612 extern void sync_filesystems(int wait);
1613 extern void __fsync_super(struct super_block *sb);
1614 extern void emergency_sync(void);
1615 extern void emergency_remount(void);
1616 extern int do_remount_sb(struct super_block *sb, int flags,
1617 			 void *data, int force);
1618 #ifdef CONFIG_BLOCK
1619 extern sector_t bmap(struct inode *, sector_t);
1620 #endif
1621 extern int notify_change(struct dentry *, struct iattr *);
1622 extern int permission(struct inode *, int, struct nameidata *);
1623 extern int generic_permission(struct inode *, int,
1624 		int (*check_acl)(struct inode *, int));
1625 
1626 extern int get_write_access(struct inode *);
1627 extern int deny_write_access(struct file *);
1628 static inline void put_write_access(struct inode * inode)
1629 {
1630 	atomic_dec(&inode->i_writecount);
1631 }
1632 static inline void allow_write_access(struct file *file)
1633 {
1634 	if (file)
1635 		atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
1636 }
1637 extern int do_pipe(int *);
1638 extern struct file *create_read_pipe(struct file *f);
1639 extern struct file *create_write_pipe(void);
1640 extern void free_write_pipe(struct file *);
1641 
1642 extern int open_namei(int dfd, const char *, int, int, struct nameidata *);
1643 extern int may_open(struct nameidata *, int, int);
1644 
1645 extern int kernel_read(struct file *, unsigned long, char *, unsigned long);
1646 extern struct file * open_exec(const char *);
1647 
1648 /* fs/dcache.c -- generic fs support functions */
1649 extern int is_subdir(struct dentry *, struct dentry *);
1650 extern ino_t find_inode_number(struct dentry *, struct qstr *);
1651 
1652 #include <linux/err.h>
1653 
1654 /* needed for stackable file system support */
1655 extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
1656 
1657 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
1658 
1659 extern void inode_init_once(struct inode *);
1660 extern void iput(struct inode *);
1661 extern struct inode * igrab(struct inode *);
1662 extern ino_t iunique(struct super_block *, ino_t);
1663 extern int inode_needs_sync(struct inode *inode);
1664 extern void generic_delete_inode(struct inode *inode);
1665 extern void generic_drop_inode(struct inode *inode);
1666 
1667 extern struct inode *ilookup5_nowait(struct super_block *sb,
1668 		unsigned long hashval, int (*test)(struct inode *, void *),
1669 		void *data);
1670 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
1671 		int (*test)(struct inode *, void *), void *data);
1672 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
1673 
1674 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
1675 extern struct inode * iget_locked(struct super_block *, unsigned long);
1676 extern void unlock_new_inode(struct inode *);
1677 
1678 static inline struct inode *iget(struct super_block *sb, unsigned long ino)
1679 {
1680 	struct inode *inode = iget_locked(sb, ino);
1681 
1682 	if (inode && (inode->i_state & I_NEW)) {
1683 		sb->s_op->read_inode(inode);
1684 		unlock_new_inode(inode);
1685 	}
1686 
1687 	return inode;
1688 }
1689 
1690 extern void __iget(struct inode * inode);
1691 extern void clear_inode(struct inode *);
1692 extern void destroy_inode(struct inode *);
1693 extern struct inode *new_inode(struct super_block *);
1694 extern int __remove_suid(struct dentry *, int);
1695 extern int should_remove_suid(struct dentry *);
1696 extern int remove_suid(struct dentry *);
1697 
1698 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
1699 extern void remove_inode_hash(struct inode *);
1700 static inline void insert_inode_hash(struct inode *inode) {
1701 	__insert_inode_hash(inode, inode->i_ino);
1702 }
1703 
1704 extern struct file * get_empty_filp(void);
1705 extern void file_move(struct file *f, struct list_head *list);
1706 extern void file_kill(struct file *f);
1707 #ifdef CONFIG_BLOCK
1708 struct bio;
1709 extern void submit_bio(int, struct bio *);
1710 extern int bdev_read_only(struct block_device *);
1711 #endif
1712 extern int set_blocksize(struct block_device *, int);
1713 extern int sb_set_blocksize(struct super_block *, int);
1714 extern int sb_min_blocksize(struct super_block *, int);
1715 
1716 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
1717 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
1718 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1719 extern int file_send_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
1720 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
1721 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
1722 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
1723 extern ssize_t generic_file_aio_write_nolock(struct kiocb *, const struct iovec *,
1724 		unsigned long, loff_t);
1725 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
1726 		unsigned long *, loff_t, loff_t *, size_t, size_t);
1727 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
1728 		unsigned long, loff_t, loff_t *, size_t, ssize_t);
1729 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
1730 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
1731 extern ssize_t generic_file_sendfile(struct file *, loff_t *, size_t, read_actor_t, void *);
1732 extern void do_generic_mapping_read(struct address_space *mapping,
1733 				    struct file_ra_state *, struct file *,
1734 				    loff_t *, read_descriptor_t *, read_actor_t);
1735 
1736 /* fs/splice.c */
1737 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
1738 		struct pipe_inode_info *, size_t, unsigned int);
1739 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
1740 		struct file *, loff_t *, size_t, unsigned int);
1741 extern ssize_t generic_file_splice_write_nolock(struct pipe_inode_info *,
1742 		struct file *, loff_t *, size_t, unsigned int);
1743 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
1744 		struct file *out, loff_t *, size_t len, unsigned int flags);
1745 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
1746 		size_t len, unsigned int flags);
1747 
1748 extern void
1749 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
1750 extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
1751 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
1752 extern loff_t remote_llseek(struct file *file, loff_t offset, int origin);
1753 extern int generic_file_open(struct inode * inode, struct file * filp);
1754 extern int nonseekable_open(struct inode * inode, struct file * filp);
1755 
1756 #ifdef CONFIG_FS_XIP
1757 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
1758 			     loff_t *ppos);
1759 extern ssize_t xip_file_sendfile(struct file *in_file, loff_t *ppos,
1760 				 size_t count, read_actor_t actor,
1761 				 void *target);
1762 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
1763 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
1764 			      size_t len, loff_t *ppos);
1765 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
1766 #else
1767 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
1768 {
1769 	return 0;
1770 }
1771 #endif
1772 
1773 static inline void do_generic_file_read(struct file * filp, loff_t *ppos,
1774 					read_descriptor_t * desc,
1775 					read_actor_t actor)
1776 {
1777 	do_generic_mapping_read(filp->f_mapping,
1778 				&filp->f_ra,
1779 				filp,
1780 				ppos,
1781 				desc,
1782 				actor);
1783 }
1784 
1785 #ifdef CONFIG_BLOCK
1786 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
1787 	struct block_device *bdev, const struct iovec *iov, loff_t offset,
1788 	unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
1789 	int lock_type);
1790 
1791 enum {
1792 	DIO_LOCKING = 1, /* need locking between buffered and direct access */
1793 	DIO_NO_LOCKING,  /* bdev; no locking at all between buffered/direct */
1794 	DIO_OWN_LOCKING, /* filesystem locks buffered and direct internally */
1795 };
1796 
1797 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
1798 	struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1799 	loff_t offset, unsigned long nr_segs, get_block_t get_block,
1800 	dio_iodone_t end_io)
1801 {
1802 	return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1803 				nr_segs, get_block, end_io, DIO_LOCKING);
1804 }
1805 
1806 static inline ssize_t blockdev_direct_IO_no_locking(int rw, struct kiocb *iocb,
1807 	struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1808 	loff_t offset, unsigned long nr_segs, get_block_t get_block,
1809 	dio_iodone_t end_io)
1810 {
1811 	return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1812 				nr_segs, get_block, end_io, DIO_NO_LOCKING);
1813 }
1814 
1815 static inline ssize_t blockdev_direct_IO_own_locking(int rw, struct kiocb *iocb,
1816 	struct inode *inode, struct block_device *bdev, const struct iovec *iov,
1817 	loff_t offset, unsigned long nr_segs, get_block_t get_block,
1818 	dio_iodone_t end_io)
1819 {
1820 	return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
1821 				nr_segs, get_block, end_io, DIO_OWN_LOCKING);
1822 }
1823 #endif
1824 
1825 extern const struct file_operations generic_ro_fops;
1826 
1827 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
1828 
1829 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
1830 extern int vfs_follow_link(struct nameidata *, const char *);
1831 extern int page_readlink(struct dentry *, char __user *, int);
1832 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
1833 extern void page_put_link(struct dentry *, struct nameidata *, void *);
1834 extern int __page_symlink(struct inode *inode, const char *symname, int len,
1835 		gfp_t gfp_mask);
1836 extern int page_symlink(struct inode *inode, const char *symname, int len);
1837 extern const struct inode_operations page_symlink_inode_operations;
1838 extern int generic_readlink(struct dentry *, char __user *, int);
1839 extern void generic_fillattr(struct inode *, struct kstat *);
1840 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1841 void inode_add_bytes(struct inode *inode, loff_t bytes);
1842 void inode_sub_bytes(struct inode *inode, loff_t bytes);
1843 loff_t inode_get_bytes(struct inode *inode);
1844 void inode_set_bytes(struct inode *inode, loff_t bytes);
1845 
1846 extern int vfs_readdir(struct file *, filldir_t, void *);
1847 
1848 extern int vfs_stat(char __user *, struct kstat *);
1849 extern int vfs_lstat(char __user *, struct kstat *);
1850 extern int vfs_stat_fd(int dfd, char __user *, struct kstat *);
1851 extern int vfs_lstat_fd(int dfd, char __user *, struct kstat *);
1852 extern int vfs_fstat(unsigned int, struct kstat *);
1853 
1854 extern int vfs_ioctl(struct file *, unsigned int, unsigned int, unsigned long);
1855 
1856 extern struct file_system_type *get_fs_type(const char *name);
1857 extern struct super_block *get_super(struct block_device *);
1858 extern struct super_block *user_get_super(dev_t);
1859 extern void drop_super(struct super_block *sb);
1860 
1861 extern int dcache_dir_open(struct inode *, struct file *);
1862 extern int dcache_dir_close(struct inode *, struct file *);
1863 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
1864 extern int dcache_readdir(struct file *, void *, filldir_t);
1865 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
1866 extern int simple_statfs(struct dentry *, struct kstatfs *);
1867 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
1868 extern int simple_unlink(struct inode *, struct dentry *);
1869 extern int simple_rmdir(struct inode *, struct dentry *);
1870 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1871 extern int simple_sync_file(struct file *, struct dentry *, int);
1872 extern int simple_empty(struct dentry *);
1873 extern int simple_readpage(struct file *file, struct page *page);
1874 extern int simple_prepare_write(struct file *file, struct page *page,
1875 			unsigned offset, unsigned to);
1876 extern int simple_commit_write(struct file *file, struct page *page,
1877 				unsigned offset, unsigned to);
1878 
1879 extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
1880 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
1881 extern const struct file_operations simple_dir_operations;
1882 extern const struct inode_operations simple_dir_inode_operations;
1883 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
1884 struct dentry *d_alloc_name(struct dentry *, const char *);
1885 extern int simple_fill_super(struct super_block *, int, struct tree_descr *);
1886 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
1887 extern void simple_release_fs(struct vfsmount **mount, int *count);
1888 
1889 extern ssize_t simple_read_from_buffer(void __user *, size_t, loff_t *, const void *, size_t);
1890 
1891 #ifdef CONFIG_MIGRATION
1892 extern int buffer_migrate_page(struct address_space *,
1893 				struct page *, struct page *);
1894 #else
1895 #define buffer_migrate_page NULL
1896 #endif
1897 
1898 extern int inode_change_ok(struct inode *, struct iattr *);
1899 extern int __must_check inode_setattr(struct inode *, struct iattr *);
1900 
1901 extern void file_update_time(struct file *file);
1902 
1903 static inline ino_t parent_ino(struct dentry *dentry)
1904 {
1905 	ino_t res;
1906 
1907 	spin_lock(&dentry->d_lock);
1908 	res = dentry->d_parent->d_inode->i_ino;
1909 	spin_unlock(&dentry->d_lock);
1910 	return res;
1911 }
1912 
1913 /* kernel/fork.c */
1914 extern int unshare_files(void);
1915 
1916 /* Transaction based IO helpers */
1917 
1918 /*
1919  * An argresp is stored in an allocated page and holds the
1920  * size of the argument or response, along with its content
1921  */
1922 struct simple_transaction_argresp {
1923 	ssize_t size;
1924 	char data[0];
1925 };
1926 
1927 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
1928 
1929 char *simple_transaction_get(struct file *file, const char __user *buf,
1930 				size_t size);
1931 ssize_t simple_transaction_read(struct file *file, char __user *buf,
1932 				size_t size, loff_t *pos);
1933 int simple_transaction_release(struct inode *inode, struct file *file);
1934 
1935 static inline void simple_transaction_set(struct file *file, size_t n)
1936 {
1937 	struct simple_transaction_argresp *ar = file->private_data;
1938 
1939 	BUG_ON(n > SIMPLE_TRANSACTION_LIMIT);
1940 
1941 	/*
1942 	 * The barrier ensures that ar->size will really remain zero until
1943 	 * ar->data is ready for reading.
1944 	 */
1945 	smp_mb();
1946 	ar->size = n;
1947 }
1948 
1949 /*
1950  * simple attribute files
1951  *
1952  * These attributes behave similar to those in sysfs:
1953  *
1954  * Writing to an attribute immediately sets a value, an open file can be
1955  * written to multiple times.
1956  *
1957  * Reading from an attribute creates a buffer from the value that might get
1958  * read with multiple read calls. When the attribute has been read
1959  * completely, no further read calls are possible until the file is opened
1960  * again.
1961  *
1962  * All attributes contain a text representation of a numeric value
1963  * that are accessed with the get() and set() functions.
1964  */
1965 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt)		\
1966 static int __fops ## _open(struct inode *inode, struct file *file)	\
1967 {									\
1968 	__simple_attr_check_format(__fmt, 0ull);			\
1969 	return simple_attr_open(inode, file, __get, __set, __fmt);	\
1970 }									\
1971 static struct file_operations __fops = {				\
1972 	.owner	 = THIS_MODULE,						\
1973 	.open	 = __fops ## _open,					\
1974 	.release = simple_attr_close,					\
1975 	.read	 = simple_attr_read,					\
1976 	.write	 = simple_attr_write,					\
1977 };
1978 
1979 static inline void __attribute__((format(printf, 1, 2)))
1980 __simple_attr_check_format(const char *fmt, ...)
1981 {
1982 	/* don't do anything, just let the compiler check the arguments; */
1983 }
1984 
1985 int simple_attr_open(struct inode *inode, struct file *file,
1986 		     u64 (*get)(void *), void (*set)(void *, u64),
1987 		     const char *fmt);
1988 int simple_attr_close(struct inode *inode, struct file *file);
1989 ssize_t simple_attr_read(struct file *file, char __user *buf,
1990 			 size_t len, loff_t *ppos);
1991 ssize_t simple_attr_write(struct file *file, const char __user *buf,
1992 			  size_t len, loff_t *ppos);
1993 
1994 
1995 #ifdef CONFIG_SECURITY
1996 static inline char *alloc_secdata(void)
1997 {
1998 	return (char *)get_zeroed_page(GFP_KERNEL);
1999 }
2000 
2001 static inline void free_secdata(void *secdata)
2002 {
2003 	free_page((unsigned long)secdata);
2004 }
2005 #else
2006 static inline char *alloc_secdata(void)
2007 {
2008 	return (char *)1;
2009 }
2010 
2011 static inline void free_secdata(void *secdata)
2012 { }
2013 #endif	/* CONFIG_SECURITY */
2014 
2015 #endif /* __KERNEL__ */
2016 #endif /* _LINUX_FS_H */
2017