xref: /openbmc/linux/include/linux/fs.h (revision 22fd411a)
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 #include <linux/blk_types.h>
12 #include <linux/types.h>
13 
14 /*
15  * It's silly to have NR_OPEN bigger than NR_FILE, but you can change
16  * the file limit at runtime and only root can increase the per-process
17  * nr_file rlimit, so it's safe to set up a ridiculously high absolute
18  * upper limit on files-per-process.
19  *
20  * Some programs (notably those using select()) may have to be
21  * recompiled to take full advantage of the new limits..
22  */
23 
24 /* Fixed constants first: */
25 #undef NR_OPEN
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 #define SEEK_SET	0	/* seek relative to beginning of file */
32 #define SEEK_CUR	1	/* seek relative to current file position */
33 #define SEEK_END	2	/* seek relative to end of file */
34 #define SEEK_MAX	SEEK_END
35 
36 struct fstrim_range {
37 	__u64 start;
38 	__u64 len;
39 	__u64 minlen;
40 };
41 
42 /* And dynamically-tunable limits and defaults: */
43 struct files_stat_struct {
44 	unsigned long nr_files;		/* read only */
45 	unsigned long nr_free_files;	/* read only */
46 	unsigned long max_files;		/* tunable */
47 };
48 
49 struct inodes_stat_t {
50 	int nr_inodes;
51 	int nr_unused;
52 	int dummy[5];		/* padding for sysctl ABI compatibility */
53 };
54 
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 #define MAY_ACCESS 16
63 #define MAY_OPEN 32
64 #define MAY_CHDIR 64
65 
66 /*
67  * flags in file.f_mode.  Note that FMODE_READ and FMODE_WRITE must correspond
68  * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
69  */
70 
71 /* file is open for reading */
72 #define FMODE_READ		((__force fmode_t)0x1)
73 /* file is open for writing */
74 #define FMODE_WRITE		((__force fmode_t)0x2)
75 /* file is seekable */
76 #define FMODE_LSEEK		((__force fmode_t)0x4)
77 /* file can be accessed using pread */
78 #define FMODE_PREAD		((__force fmode_t)0x8)
79 /* file can be accessed using pwrite */
80 #define FMODE_PWRITE		((__force fmode_t)0x10)
81 /* File is opened for execution with sys_execve / sys_uselib */
82 #define FMODE_EXEC		((__force fmode_t)0x20)
83 /* File is opened with O_NDELAY (only set for block devices) */
84 #define FMODE_NDELAY		((__force fmode_t)0x40)
85 /* File is opened with O_EXCL (only set for block devices) */
86 #define FMODE_EXCL		((__force fmode_t)0x80)
87 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
88    (specialy hack for floppy.c) */
89 #define FMODE_WRITE_IOCTL	((__force fmode_t)0x100)
90 
91 /*
92  * Don't update ctime and mtime.
93  *
94  * Currently a special hack for the XFS open_by_handle ioctl, but we'll
95  * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
96  */
97 #define FMODE_NOCMTIME		((__force fmode_t)0x800)
98 
99 /* Expect random access pattern */
100 #define FMODE_RANDOM		((__force fmode_t)0x1000)
101 
102 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
103 #define FMODE_UNSIGNED_OFFSET	((__force fmode_t)0x2000)
104 
105 /* File was opened by fanotify and shouldn't generate fanotify events */
106 #define FMODE_NONOTIFY		((__force fmode_t)0x1000000)
107 
108 /*
109  * The below are the various read and write types that we support. Some of
110  * them include behavioral modifiers that send information down to the
111  * block layer and IO scheduler. Terminology:
112  *
113  *	The block layer uses device plugging to defer IO a little bit, in
114  *	the hope that we will see more IO very shortly. This increases
115  *	coalescing of adjacent IO and thus reduces the number of IOs we
116  *	have to send to the device. It also allows for better queuing,
117  *	if the IO isn't mergeable. If the caller is going to be waiting
118  *	for the IO, then he must ensure that the device is unplugged so
119  *	that the IO is dispatched to the driver.
120  *
121  *	All IO is handled async in Linux. This is fine for background
122  *	writes, but for reads or writes that someone waits for completion
123  *	on, we want to notify the block layer and IO scheduler so that they
124  *	know about it. That allows them to make better scheduling
125  *	decisions. So when the below references 'sync' and 'async', it
126  *	is referencing this priority hint.
127  *
128  * With that in mind, the available types are:
129  *
130  * READ			A normal read operation. Device will be plugged.
131  * READ_SYNC		A synchronous read. Device is not plugged, caller can
132  *			immediately wait on this read without caring about
133  *			unplugging.
134  * READA		Used for read-ahead operations. Lower priority, and the
135  *			block layer could (in theory) choose to ignore this
136  *			request if it runs into resource problems.
137  * WRITE		A normal async write. Device will be plugged.
138  * WRITE_SYNC_PLUG	Synchronous write. Identical to WRITE, but passes down
139  *			the hint that someone will be waiting on this IO
140  *			shortly. The device must still be unplugged explicitly,
141  *			WRITE_SYNC_PLUG does not do this as we could be
142  *			submitting more writes before we actually wait on any
143  *			of them.
144  * WRITE_SYNC		Like WRITE_SYNC_PLUG, but also unplugs the device
145  *			immediately after submission. The write equivalent
146  *			of READ_SYNC.
147  * WRITE_ODIRECT_PLUG	Special case write for O_DIRECT only.
148  * WRITE_FLUSH		Like WRITE_SYNC but with preceding cache flush.
149  * WRITE_FUA		Like WRITE_SYNC but data is guaranteed to be on
150  *			non-volatile media on completion.
151  * WRITE_FLUSH_FUA	Combination of WRITE_FLUSH and FUA. The IO is preceded
152  *			by a cache flush and data is guaranteed to be on
153  *			non-volatile media on completion.
154  *
155  */
156 #define RW_MASK			REQ_WRITE
157 #define RWA_MASK		REQ_RAHEAD
158 
159 #define READ			0
160 #define WRITE			RW_MASK
161 #define READA			RWA_MASK
162 
163 #define READ_SYNC		(READ | REQ_SYNC | REQ_UNPLUG)
164 #define READ_META		(READ | REQ_META)
165 #define WRITE_SYNC_PLUG		(WRITE | REQ_SYNC | REQ_NOIDLE)
166 #define WRITE_SYNC		(WRITE | REQ_SYNC | REQ_NOIDLE | REQ_UNPLUG)
167 #define WRITE_ODIRECT_PLUG	(WRITE | REQ_SYNC)
168 #define WRITE_META		(WRITE | REQ_META)
169 #define WRITE_FLUSH		(WRITE | REQ_SYNC | REQ_NOIDLE | REQ_UNPLUG | \
170 				 REQ_FLUSH)
171 #define WRITE_FUA		(WRITE | REQ_SYNC | REQ_NOIDLE | REQ_UNPLUG | \
172 				 REQ_FUA)
173 #define WRITE_FLUSH_FUA		(WRITE | REQ_SYNC | REQ_NOIDLE | REQ_UNPLUG | \
174 				 REQ_FLUSH | REQ_FUA)
175 
176 #define SEL_IN		1
177 #define SEL_OUT		2
178 #define SEL_EX		4
179 
180 /* public flags for file_system_type */
181 #define FS_REQUIRES_DEV 1
182 #define FS_BINARY_MOUNTDATA 2
183 #define FS_HAS_SUBTYPE 4
184 #define FS_REVAL_DOT	16384	/* Check the paths ".", ".." for staleness */
185 #define FS_RENAME_DOES_D_MOVE	32768	/* FS will handle d_move()
186 					 * during rename() internally.
187 					 */
188 
189 /*
190  * These are the fs-independent mount-flags: up to 32 flags are supported
191  */
192 #define MS_RDONLY	 1	/* Mount read-only */
193 #define MS_NOSUID	 2	/* Ignore suid and sgid bits */
194 #define MS_NODEV	 4	/* Disallow access to device special files */
195 #define MS_NOEXEC	 8	/* Disallow program execution */
196 #define MS_SYNCHRONOUS	16	/* Writes are synced at once */
197 #define MS_REMOUNT	32	/* Alter flags of a mounted FS */
198 #define MS_MANDLOCK	64	/* Allow mandatory locks on an FS */
199 #define MS_DIRSYNC	128	/* Directory modifications are synchronous */
200 #define MS_NOATIME	1024	/* Do not update access times. */
201 #define MS_NODIRATIME	2048	/* Do not update directory access times */
202 #define MS_BIND		4096
203 #define MS_MOVE		8192
204 #define MS_REC		16384
205 #define MS_VERBOSE	32768	/* War is peace. Verbosity is silence.
206 				   MS_VERBOSE is deprecated. */
207 #define MS_SILENT	32768
208 #define MS_POSIXACL	(1<<16)	/* VFS does not apply the umask */
209 #define MS_UNBINDABLE	(1<<17)	/* change to unbindable */
210 #define MS_PRIVATE	(1<<18)	/* change to private */
211 #define MS_SLAVE	(1<<19)	/* change to slave */
212 #define MS_SHARED	(1<<20)	/* change to shared */
213 #define MS_RELATIME	(1<<21)	/* Update atime relative to mtime/ctime. */
214 #define MS_KERNMOUNT	(1<<22) /* this is a kern_mount call */
215 #define MS_I_VERSION	(1<<23) /* Update inode I_version field */
216 #define MS_STRICTATIME	(1<<24) /* Always perform atime updates */
217 #define MS_BORN		(1<<29)
218 #define MS_ACTIVE	(1<<30)
219 #define MS_NOUSER	(1<<31)
220 
221 /*
222  * Superblock flags that can be altered by MS_REMOUNT
223  */
224 #define MS_RMT_MASK	(MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_I_VERSION)
225 
226 /*
227  * Old magic mount flag and mask
228  */
229 #define MS_MGC_VAL 0xC0ED0000
230 #define MS_MGC_MSK 0xffff0000
231 
232 /* Inode flags - they have nothing to superblock flags now */
233 
234 #define S_SYNC		1	/* Writes are synced at once */
235 #define S_NOATIME	2	/* Do not update access times */
236 #define S_APPEND	4	/* Append-only file */
237 #define S_IMMUTABLE	8	/* Immutable file */
238 #define S_DEAD		16	/* removed, but still open directory */
239 #define S_NOQUOTA	32	/* Inode is not counted to quota */
240 #define S_DIRSYNC	64	/* Directory modifications are synchronous */
241 #define S_NOCMTIME	128	/* Do not update file c/mtime */
242 #define S_SWAPFILE	256	/* Do not truncate: swapon got its bmaps */
243 #define S_PRIVATE	512	/* Inode is fs-internal */
244 #define S_IMA		1024	/* Inode has an associated IMA struct */
245 #define S_AUTOMOUNT	2048	/* Automount/referral quasi-directory */
246 
247 /*
248  * Note that nosuid etc flags are inode-specific: setting some file-system
249  * flags just means all the inodes inherit those flags by default. It might be
250  * possible to override it selectively if you really wanted to with some
251  * ioctl() that is not currently implemented.
252  *
253  * Exception: MS_RDONLY is always applied to the entire file system.
254  *
255  * Unfortunately, it is possible to change a filesystems flags with it mounted
256  * with files in use.  This means that all of the inodes will not have their
257  * i_flags updated.  Hence, i_flags no longer inherit the superblock mount
258  * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
259  */
260 #define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
261 
262 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
263 #define IS_SYNC(inode)		(__IS_FLG(inode, MS_SYNCHRONOUS) || \
264 					((inode)->i_flags & S_SYNC))
265 #define IS_DIRSYNC(inode)	(__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
266 					((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
267 #define IS_MANDLOCK(inode)	__IS_FLG(inode, MS_MANDLOCK)
268 #define IS_NOATIME(inode)   __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
269 #define IS_I_VERSION(inode)   __IS_FLG(inode, MS_I_VERSION)
270 
271 #define IS_NOQUOTA(inode)	((inode)->i_flags & S_NOQUOTA)
272 #define IS_APPEND(inode)	((inode)->i_flags & S_APPEND)
273 #define IS_IMMUTABLE(inode)	((inode)->i_flags & S_IMMUTABLE)
274 #define IS_POSIXACL(inode)	__IS_FLG(inode, MS_POSIXACL)
275 
276 #define IS_DEADDIR(inode)	((inode)->i_flags & S_DEAD)
277 #define IS_NOCMTIME(inode)	((inode)->i_flags & S_NOCMTIME)
278 #define IS_SWAPFILE(inode)	((inode)->i_flags & S_SWAPFILE)
279 #define IS_PRIVATE(inode)	((inode)->i_flags & S_PRIVATE)
280 #define IS_IMA(inode)		((inode)->i_flags & S_IMA)
281 #define IS_AUTOMOUNT(inode)	((inode)->i_flags & S_AUTOMOUNT)
282 
283 /* the read-only stuff doesn't really belong here, but any other place is
284    probably as bad and I don't want to create yet another include file. */
285 
286 #define BLKROSET   _IO(0x12,93)	/* set device read-only (0 = read-write) */
287 #define BLKROGET   _IO(0x12,94)	/* get read-only status (0 = read_write) */
288 #define BLKRRPART  _IO(0x12,95)	/* re-read partition table */
289 #define BLKGETSIZE _IO(0x12,96)	/* return device size /512 (long *arg) */
290 #define BLKFLSBUF  _IO(0x12,97)	/* flush buffer cache */
291 #define BLKRASET   _IO(0x12,98)	/* set read ahead for block device */
292 #define BLKRAGET   _IO(0x12,99)	/* get current read ahead setting */
293 #define BLKFRASET  _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
294 #define BLKFRAGET  _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
295 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
296 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
297 #define BLKSSZGET  _IO(0x12,104)/* get block device sector size */
298 #if 0
299 #define BLKPG      _IO(0x12,105)/* See blkpg.h */
300 
301 /* Some people are morons.  Do not use sizeof! */
302 
303 #define BLKELVGET  _IOR(0x12,106,size_t)/* elevator get */
304 #define BLKELVSET  _IOW(0x12,107,size_t)/* elevator set */
305 /* This was here just to show that the number is taken -
306    probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
307 #endif
308 /* A jump here: 108-111 have been used for various private purposes. */
309 #define BLKBSZGET  _IOR(0x12,112,size_t)
310 #define BLKBSZSET  _IOW(0x12,113,size_t)
311 #define BLKGETSIZE64 _IOR(0x12,114,size_t)	/* return device size in bytes (u64 *arg) */
312 #define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
313 #define BLKTRACESTART _IO(0x12,116)
314 #define BLKTRACESTOP _IO(0x12,117)
315 #define BLKTRACETEARDOWN _IO(0x12,118)
316 #define BLKDISCARD _IO(0x12,119)
317 #define BLKIOMIN _IO(0x12,120)
318 #define BLKIOOPT _IO(0x12,121)
319 #define BLKALIGNOFF _IO(0x12,122)
320 #define BLKPBSZGET _IO(0x12,123)
321 #define BLKDISCARDZEROES _IO(0x12,124)
322 #define BLKSECDISCARD _IO(0x12,125)
323 
324 #define BMAP_IOCTL 1		/* obsolete - kept for compatibility */
325 #define FIBMAP	   _IO(0x00,1)	/* bmap access */
326 #define FIGETBSZ   _IO(0x00,2)	/* get the block size used for bmap */
327 #define FIFREEZE	_IOWR('X', 119, int)	/* Freeze */
328 #define FITHAW		_IOWR('X', 120, int)	/* Thaw */
329 #define FITRIM		_IOWR('X', 121, struct fstrim_range)	/* Trim */
330 
331 #define	FS_IOC_GETFLAGS			_IOR('f', 1, long)
332 #define	FS_IOC_SETFLAGS			_IOW('f', 2, long)
333 #define	FS_IOC_GETVERSION		_IOR('v', 1, long)
334 #define	FS_IOC_SETVERSION		_IOW('v', 2, long)
335 #define FS_IOC_FIEMAP			_IOWR('f', 11, struct fiemap)
336 #define FS_IOC32_GETFLAGS		_IOR('f', 1, int)
337 #define FS_IOC32_SETFLAGS		_IOW('f', 2, int)
338 #define FS_IOC32_GETVERSION		_IOR('v', 1, int)
339 #define FS_IOC32_SETVERSION		_IOW('v', 2, int)
340 
341 /*
342  * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS)
343  */
344 #define	FS_SECRM_FL			0x00000001 /* Secure deletion */
345 #define	FS_UNRM_FL			0x00000002 /* Undelete */
346 #define	FS_COMPR_FL			0x00000004 /* Compress file */
347 #define FS_SYNC_FL			0x00000008 /* Synchronous updates */
348 #define FS_IMMUTABLE_FL			0x00000010 /* Immutable file */
349 #define FS_APPEND_FL			0x00000020 /* writes to file may only append */
350 #define FS_NODUMP_FL			0x00000040 /* do not dump file */
351 #define FS_NOATIME_FL			0x00000080 /* do not update atime */
352 /* Reserved for compression usage... */
353 #define FS_DIRTY_FL			0x00000100
354 #define FS_COMPRBLK_FL			0x00000200 /* One or more compressed clusters */
355 #define FS_NOCOMP_FL			0x00000400 /* Don't compress */
356 #define FS_ECOMPR_FL			0x00000800 /* Compression error */
357 /* End compression flags --- maybe not all used */
358 #define FS_BTREE_FL			0x00001000 /* btree format dir */
359 #define FS_INDEX_FL			0x00001000 /* hash-indexed directory */
360 #define FS_IMAGIC_FL			0x00002000 /* AFS directory */
361 #define FS_JOURNAL_DATA_FL		0x00004000 /* Reserved for ext3 */
362 #define FS_NOTAIL_FL			0x00008000 /* file tail should not be merged */
363 #define FS_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
364 #define FS_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
365 #define FS_EXTENT_FL			0x00080000 /* Extents */
366 #define FS_DIRECTIO_FL			0x00100000 /* Use direct i/o */
367 #define FS_RESERVED_FL			0x80000000 /* reserved for ext2 lib */
368 
369 #define FS_FL_USER_VISIBLE		0x0003DFFF /* User visible flags */
370 #define FS_FL_USER_MODIFIABLE		0x000380FF /* User modifiable flags */
371 
372 
373 #define SYNC_FILE_RANGE_WAIT_BEFORE	1
374 #define SYNC_FILE_RANGE_WRITE		2
375 #define SYNC_FILE_RANGE_WAIT_AFTER	4
376 
377 #ifdef __KERNEL__
378 
379 #include <linux/linkage.h>
380 #include <linux/wait.h>
381 #include <linux/types.h>
382 #include <linux/kdev_t.h>
383 #include <linux/dcache.h>
384 #include <linux/path.h>
385 #include <linux/stat.h>
386 #include <linux/cache.h>
387 #include <linux/list.h>
388 #include <linux/radix-tree.h>
389 #include <linux/prio_tree.h>
390 #include <linux/init.h>
391 #include <linux/pid.h>
392 #include <linux/mutex.h>
393 #include <linux/capability.h>
394 #include <linux/semaphore.h>
395 #include <linux/fiemap.h>
396 #include <linux/rculist_bl.h>
397 
398 #include <asm/atomic.h>
399 #include <asm/byteorder.h>
400 
401 struct export_operations;
402 struct hd_geometry;
403 struct iovec;
404 struct nameidata;
405 struct kiocb;
406 struct kobject;
407 struct pipe_inode_info;
408 struct poll_table_struct;
409 struct kstatfs;
410 struct vm_area_struct;
411 struct vfsmount;
412 struct cred;
413 
414 extern void __init inode_init(void);
415 extern void __init inode_init_early(void);
416 extern void __init files_init(unsigned long);
417 
418 extern struct files_stat_struct files_stat;
419 extern unsigned long get_max_files(void);
420 extern int sysctl_nr_open;
421 extern struct inodes_stat_t inodes_stat;
422 extern int leases_enable, lease_break_time;
423 
424 struct buffer_head;
425 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
426 			struct buffer_head *bh_result, int create);
427 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
428 			ssize_t bytes, void *private, int ret,
429 			bool is_async);
430 
431 /*
432  * Attribute flags.  These should be or-ed together to figure out what
433  * has been changed!
434  */
435 #define ATTR_MODE	(1 << 0)
436 #define ATTR_UID	(1 << 1)
437 #define ATTR_GID	(1 << 2)
438 #define ATTR_SIZE	(1 << 3)
439 #define ATTR_ATIME	(1 << 4)
440 #define ATTR_MTIME	(1 << 5)
441 #define ATTR_CTIME	(1 << 6)
442 #define ATTR_ATIME_SET	(1 << 7)
443 #define ATTR_MTIME_SET	(1 << 8)
444 #define ATTR_FORCE	(1 << 9) /* Not a change, but a change it */
445 #define ATTR_ATTR_FLAG	(1 << 10)
446 #define ATTR_KILL_SUID	(1 << 11)
447 #define ATTR_KILL_SGID	(1 << 12)
448 #define ATTR_FILE	(1 << 13)
449 #define ATTR_KILL_PRIV	(1 << 14)
450 #define ATTR_OPEN	(1 << 15) /* Truncating from open(O_TRUNC) */
451 #define ATTR_TIMES_SET	(1 << 16)
452 
453 /*
454  * This is the Inode Attributes structure, used for notify_change().  It
455  * uses the above definitions as flags, to know which values have changed.
456  * Also, in this manner, a Filesystem can look at only the values it cares
457  * about.  Basically, these are the attributes that the VFS layer can
458  * request to change from the FS layer.
459  *
460  * Derek Atkins <warlord@MIT.EDU> 94-10-20
461  */
462 struct iattr {
463 	unsigned int	ia_valid;
464 	umode_t		ia_mode;
465 	uid_t		ia_uid;
466 	gid_t		ia_gid;
467 	loff_t		ia_size;
468 	struct timespec	ia_atime;
469 	struct timespec	ia_mtime;
470 	struct timespec	ia_ctime;
471 
472 	/*
473 	 * Not an attribute, but an auxilary info for filesystems wanting to
474 	 * implement an ftruncate() like method.  NOTE: filesystem should
475 	 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
476 	 */
477 	struct file	*ia_file;
478 };
479 
480 /*
481  * Includes for diskquotas.
482  */
483 #include <linux/quota.h>
484 
485 /**
486  * enum positive_aop_returns - aop return codes with specific semantics
487  *
488  * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
489  * 			    completed, that the page is still locked, and
490  * 			    should be considered active.  The VM uses this hint
491  * 			    to return the page to the active list -- it won't
492  * 			    be a candidate for writeback again in the near
493  * 			    future.  Other callers must be careful to unlock
494  * 			    the page if they get this return.  Returned by
495  * 			    writepage();
496  *
497  * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
498  *  			unlocked it and the page might have been truncated.
499  *  			The caller should back up to acquiring a new page and
500  *  			trying again.  The aop will be taking reasonable
501  *  			precautions not to livelock.  If the caller held a page
502  *  			reference, it should drop it before retrying.  Returned
503  *  			by readpage().
504  *
505  * address_space_operation functions return these large constants to indicate
506  * special semantics to the caller.  These are much larger than the bytes in a
507  * page to allow for functions that return the number of bytes operated on in a
508  * given page.
509  */
510 
511 enum positive_aop_returns {
512 	AOP_WRITEPAGE_ACTIVATE	= 0x80000,
513 	AOP_TRUNCATED_PAGE	= 0x80001,
514 };
515 
516 #define AOP_FLAG_UNINTERRUPTIBLE	0x0001 /* will not do a short write */
517 #define AOP_FLAG_CONT_EXPAND		0x0002 /* called from cont_expand */
518 #define AOP_FLAG_NOFS			0x0004 /* used by filesystem to direct
519 						* helper code (eg buffer layer)
520 						* to clear GFP_FS from alloc */
521 
522 /*
523  * oh the beauties of C type declarations.
524  */
525 struct page;
526 struct address_space;
527 struct writeback_control;
528 
529 struct iov_iter {
530 	const struct iovec *iov;
531 	unsigned long nr_segs;
532 	size_t iov_offset;
533 	size_t count;
534 };
535 
536 size_t iov_iter_copy_from_user_atomic(struct page *page,
537 		struct iov_iter *i, unsigned long offset, size_t bytes);
538 size_t iov_iter_copy_from_user(struct page *page,
539 		struct iov_iter *i, unsigned long offset, size_t bytes);
540 void iov_iter_advance(struct iov_iter *i, size_t bytes);
541 int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
542 size_t iov_iter_single_seg_count(struct iov_iter *i);
543 
544 static inline void iov_iter_init(struct iov_iter *i,
545 			const struct iovec *iov, unsigned long nr_segs,
546 			size_t count, size_t written)
547 {
548 	i->iov = iov;
549 	i->nr_segs = nr_segs;
550 	i->iov_offset = 0;
551 	i->count = count + written;
552 
553 	iov_iter_advance(i, written);
554 }
555 
556 static inline size_t iov_iter_count(struct iov_iter *i)
557 {
558 	return i->count;
559 }
560 
561 /*
562  * "descriptor" for what we're up to with a read.
563  * This allows us to use the same read code yet
564  * have multiple different users of the data that
565  * we read from a file.
566  *
567  * The simplest case just copies the data to user
568  * mode.
569  */
570 typedef struct {
571 	size_t written;
572 	size_t count;
573 	union {
574 		char __user *buf;
575 		void *data;
576 	} arg;
577 	int error;
578 } read_descriptor_t;
579 
580 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
581 		unsigned long, unsigned long);
582 
583 struct address_space_operations {
584 	int (*writepage)(struct page *page, struct writeback_control *wbc);
585 	int (*readpage)(struct file *, struct page *);
586 	void (*sync_page)(struct page *);
587 
588 	/* Write back some dirty pages from this mapping. */
589 	int (*writepages)(struct address_space *, struct writeback_control *);
590 
591 	/* Set a page dirty.  Return true if this dirtied it */
592 	int (*set_page_dirty)(struct page *page);
593 
594 	int (*readpages)(struct file *filp, struct address_space *mapping,
595 			struct list_head *pages, unsigned nr_pages);
596 
597 	int (*write_begin)(struct file *, struct address_space *mapping,
598 				loff_t pos, unsigned len, unsigned flags,
599 				struct page **pagep, void **fsdata);
600 	int (*write_end)(struct file *, struct address_space *mapping,
601 				loff_t pos, unsigned len, unsigned copied,
602 				struct page *page, void *fsdata);
603 
604 	/* Unfortunately this kludge is needed for FIBMAP. Don't use it */
605 	sector_t (*bmap)(struct address_space *, sector_t);
606 	void (*invalidatepage) (struct page *, unsigned long);
607 	int (*releasepage) (struct page *, gfp_t);
608 	void (*freepage)(struct page *);
609 	ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
610 			loff_t offset, unsigned long nr_segs);
611 	int (*get_xip_mem)(struct address_space *, pgoff_t, int,
612 						void **, unsigned long *);
613 	/* migrate the contents of a page to the specified target */
614 	int (*migratepage) (struct address_space *,
615 			struct page *, struct page *);
616 	int (*launder_page) (struct page *);
617 	int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
618 					unsigned long);
619 	int (*error_remove_page)(struct address_space *, struct page *);
620 };
621 
622 /*
623  * pagecache_write_begin/pagecache_write_end must be used by general code
624  * to write into the pagecache.
625  */
626 int pagecache_write_begin(struct file *, struct address_space *mapping,
627 				loff_t pos, unsigned len, unsigned flags,
628 				struct page **pagep, void **fsdata);
629 
630 int pagecache_write_end(struct file *, struct address_space *mapping,
631 				loff_t pos, unsigned len, unsigned copied,
632 				struct page *page, void *fsdata);
633 
634 struct backing_dev_info;
635 struct address_space {
636 	struct inode		*host;		/* owner: inode, block_device */
637 	struct radix_tree_root	page_tree;	/* radix tree of all pages */
638 	spinlock_t		tree_lock;	/* and lock protecting it */
639 	unsigned int		i_mmap_writable;/* count VM_SHARED mappings */
640 	struct prio_tree_root	i_mmap;		/* tree of private and shared mappings */
641 	struct list_head	i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
642 	spinlock_t		i_mmap_lock;	/* protect tree, count, list */
643 	unsigned int		truncate_count;	/* Cover race condition with truncate */
644 	unsigned long		nrpages;	/* number of total pages */
645 	pgoff_t			writeback_index;/* writeback starts here */
646 	const struct address_space_operations *a_ops;	/* methods */
647 	unsigned long		flags;		/* error bits/gfp mask */
648 	struct backing_dev_info *backing_dev_info; /* device readahead, etc */
649 	spinlock_t		private_lock;	/* for use by the address_space */
650 	struct list_head	private_list;	/* ditto */
651 	struct address_space	*assoc_mapping;	/* ditto */
652 } __attribute__((aligned(sizeof(long))));
653 	/*
654 	 * On most architectures that alignment is already the case; but
655 	 * must be enforced here for CRIS, to let the least signficant bit
656 	 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
657 	 */
658 
659 struct block_device {
660 	dev_t			bd_dev;  /* not a kdev_t - it's a search key */
661 	struct inode *		bd_inode;	/* will die */
662 	struct super_block *	bd_super;
663 	int			bd_openers;
664 	struct mutex		bd_mutex;	/* open/close mutex */
665 	struct list_head	bd_inodes;
666 	void *			bd_claiming;
667 	void *			bd_holder;
668 	int			bd_holders;
669 	bool			bd_write_holder;
670 #ifdef CONFIG_SYSFS
671 	struct list_head	bd_holder_disks;
672 #endif
673 	struct block_device *	bd_contains;
674 	unsigned		bd_block_size;
675 	struct hd_struct *	bd_part;
676 	/* number of times partitions within this device have been opened. */
677 	unsigned		bd_part_count;
678 	int			bd_invalidated;
679 	struct gendisk *	bd_disk;
680 	struct list_head	bd_list;
681 	/*
682 	 * Private data.  You must have bd_claim'ed the block_device
683 	 * to use this.  NOTE:  bd_claim allows an owner to claim
684 	 * the same device multiple times, the owner must take special
685 	 * care to not mess up bd_private for that case.
686 	 */
687 	unsigned long		bd_private;
688 
689 	/* The counter of freeze processes */
690 	int			bd_fsfreeze_count;
691 	/* Mutex for freeze */
692 	struct mutex		bd_fsfreeze_mutex;
693 };
694 
695 /*
696  * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
697  * radix trees
698  */
699 #define PAGECACHE_TAG_DIRTY	0
700 #define PAGECACHE_TAG_WRITEBACK	1
701 #define PAGECACHE_TAG_TOWRITE	2
702 
703 int mapping_tagged(struct address_space *mapping, int tag);
704 
705 /*
706  * Might pages of this file be mapped into userspace?
707  */
708 static inline int mapping_mapped(struct address_space *mapping)
709 {
710 	return	!prio_tree_empty(&mapping->i_mmap) ||
711 		!list_empty(&mapping->i_mmap_nonlinear);
712 }
713 
714 /*
715  * Might pages of this file have been modified in userspace?
716  * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
717  * marks vma as VM_SHARED if it is shared, and the file was opened for
718  * writing i.e. vma may be mprotected writable even if now readonly.
719  */
720 static inline int mapping_writably_mapped(struct address_space *mapping)
721 {
722 	return mapping->i_mmap_writable != 0;
723 }
724 
725 /*
726  * Use sequence counter to get consistent i_size on 32-bit processors.
727  */
728 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
729 #include <linux/seqlock.h>
730 #define __NEED_I_SIZE_ORDERED
731 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
732 #else
733 #define i_size_ordered_init(inode) do { } while (0)
734 #endif
735 
736 struct posix_acl;
737 #define ACL_NOT_CACHED ((void *)(-1))
738 
739 struct inode {
740 	/* RCU path lookup touches following: */
741 	umode_t			i_mode;
742 	uid_t			i_uid;
743 	gid_t			i_gid;
744 	const struct inode_operations	*i_op;
745 	struct super_block	*i_sb;
746 
747 	spinlock_t		i_lock;	/* i_blocks, i_bytes, maybe i_size */
748 	unsigned int		i_flags;
749 	struct mutex		i_mutex;
750 
751 	unsigned long		i_state;
752 	unsigned long		dirtied_when;	/* jiffies of first dirtying */
753 
754 	struct hlist_node	i_hash;
755 	struct list_head	i_wb_list;	/* backing dev IO list */
756 	struct list_head	i_lru;		/* inode LRU list */
757 	struct list_head	i_sb_list;
758 	union {
759 		struct list_head	i_dentry;
760 		struct rcu_head		i_rcu;
761 	};
762 	unsigned long		i_ino;
763 	atomic_t		i_count;
764 	unsigned int		i_nlink;
765 	dev_t			i_rdev;
766 	unsigned int		i_blkbits;
767 	u64			i_version;
768 	loff_t			i_size;
769 #ifdef __NEED_I_SIZE_ORDERED
770 	seqcount_t		i_size_seqcount;
771 #endif
772 	struct timespec		i_atime;
773 	struct timespec		i_mtime;
774 	struct timespec		i_ctime;
775 	blkcnt_t		i_blocks;
776 	unsigned short          i_bytes;
777 	struct rw_semaphore	i_alloc_sem;
778 	const struct file_operations	*i_fop;	/* former ->i_op->default_file_ops */
779 	struct file_lock	*i_flock;
780 	struct address_space	*i_mapping;
781 	struct address_space	i_data;
782 #ifdef CONFIG_QUOTA
783 	struct dquot		*i_dquot[MAXQUOTAS];
784 #endif
785 	struct list_head	i_devices;
786 	union {
787 		struct pipe_inode_info	*i_pipe;
788 		struct block_device	*i_bdev;
789 		struct cdev		*i_cdev;
790 	};
791 
792 	__u32			i_generation;
793 
794 #ifdef CONFIG_FSNOTIFY
795 	__u32			i_fsnotify_mask; /* all events this inode cares about */
796 	struct hlist_head	i_fsnotify_marks;
797 #endif
798 
799 #ifdef CONFIG_IMA
800 	/* protected by i_lock */
801 	unsigned int		i_readcount; /* struct files open RO */
802 #endif
803 	atomic_t		i_writecount;
804 #ifdef CONFIG_SECURITY
805 	void			*i_security;
806 #endif
807 #ifdef CONFIG_FS_POSIX_ACL
808 	struct posix_acl	*i_acl;
809 	struct posix_acl	*i_default_acl;
810 #endif
811 	void			*i_private; /* fs or device private pointer */
812 };
813 
814 static inline int inode_unhashed(struct inode *inode)
815 {
816 	return hlist_unhashed(&inode->i_hash);
817 }
818 
819 /*
820  * inode->i_mutex nesting subclasses for the lock validator:
821  *
822  * 0: the object of the current VFS operation
823  * 1: parent
824  * 2: child/target
825  * 3: quota file
826  *
827  * The locking order between these classes is
828  * parent -> child -> normal -> xattr -> quota
829  */
830 enum inode_i_mutex_lock_class
831 {
832 	I_MUTEX_NORMAL,
833 	I_MUTEX_PARENT,
834 	I_MUTEX_CHILD,
835 	I_MUTEX_XATTR,
836 	I_MUTEX_QUOTA
837 };
838 
839 /*
840  * NOTE: in a 32bit arch with a preemptable kernel and
841  * an UP compile the i_size_read/write must be atomic
842  * with respect to the local cpu (unlike with preempt disabled),
843  * but they don't need to be atomic with respect to other cpus like in
844  * true SMP (so they need either to either locally disable irq around
845  * the read or for example on x86 they can be still implemented as a
846  * cmpxchg8b without the need of the lock prefix). For SMP compiles
847  * and 64bit archs it makes no difference if preempt is enabled or not.
848  */
849 static inline loff_t i_size_read(const struct inode *inode)
850 {
851 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
852 	loff_t i_size;
853 	unsigned int seq;
854 
855 	do {
856 		seq = read_seqcount_begin(&inode->i_size_seqcount);
857 		i_size = inode->i_size;
858 	} while (read_seqcount_retry(&inode->i_size_seqcount, seq));
859 	return i_size;
860 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
861 	loff_t i_size;
862 
863 	preempt_disable();
864 	i_size = inode->i_size;
865 	preempt_enable();
866 	return i_size;
867 #else
868 	return inode->i_size;
869 #endif
870 }
871 
872 /*
873  * NOTE: unlike i_size_read(), i_size_write() does need locking around it
874  * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
875  * can be lost, resulting in subsequent i_size_read() calls spinning forever.
876  */
877 static inline void i_size_write(struct inode *inode, loff_t i_size)
878 {
879 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
880 	write_seqcount_begin(&inode->i_size_seqcount);
881 	inode->i_size = i_size;
882 	write_seqcount_end(&inode->i_size_seqcount);
883 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
884 	preempt_disable();
885 	inode->i_size = i_size;
886 	preempt_enable();
887 #else
888 	inode->i_size = i_size;
889 #endif
890 }
891 
892 static inline unsigned iminor(const struct inode *inode)
893 {
894 	return MINOR(inode->i_rdev);
895 }
896 
897 static inline unsigned imajor(const struct inode *inode)
898 {
899 	return MAJOR(inode->i_rdev);
900 }
901 
902 extern struct block_device *I_BDEV(struct inode *inode);
903 
904 struct fown_struct {
905 	rwlock_t lock;          /* protects pid, uid, euid fields */
906 	struct pid *pid;	/* pid or -pgrp where SIGIO should be sent */
907 	enum pid_type pid_type;	/* Kind of process group SIGIO should be sent to */
908 	uid_t uid, euid;	/* uid/euid of process setting the owner */
909 	int signum;		/* posix.1b rt signal to be delivered on IO */
910 };
911 
912 /*
913  * Track a single file's readahead state
914  */
915 struct file_ra_state {
916 	pgoff_t start;			/* where readahead started */
917 	unsigned int size;		/* # of readahead pages */
918 	unsigned int async_size;	/* do asynchronous readahead when
919 					   there are only # of pages ahead */
920 
921 	unsigned int ra_pages;		/* Maximum readahead window */
922 	unsigned int mmap_miss;		/* Cache miss stat for mmap accesses */
923 	loff_t prev_pos;		/* Cache last read() position */
924 };
925 
926 /*
927  * Check if @index falls in the readahead windows.
928  */
929 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
930 {
931 	return (index >= ra->start &&
932 		index <  ra->start + ra->size);
933 }
934 
935 #define FILE_MNT_WRITE_TAKEN	1
936 #define FILE_MNT_WRITE_RELEASED	2
937 
938 struct file {
939 	/*
940 	 * fu_list becomes invalid after file_free is called and queued via
941 	 * fu_rcuhead for RCU freeing
942 	 */
943 	union {
944 		struct list_head	fu_list;
945 		struct rcu_head 	fu_rcuhead;
946 	} f_u;
947 	struct path		f_path;
948 #define f_dentry	f_path.dentry
949 #define f_vfsmnt	f_path.mnt
950 	const struct file_operations	*f_op;
951 	spinlock_t		f_lock;  /* f_ep_links, f_flags, no IRQ */
952 #ifdef CONFIG_SMP
953 	int			f_sb_list_cpu;
954 #endif
955 	atomic_long_t		f_count;
956 	unsigned int 		f_flags;
957 	fmode_t			f_mode;
958 	loff_t			f_pos;
959 	struct fown_struct	f_owner;
960 	const struct cred	*f_cred;
961 	struct file_ra_state	f_ra;
962 
963 	u64			f_version;
964 #ifdef CONFIG_SECURITY
965 	void			*f_security;
966 #endif
967 	/* needed for tty driver, and maybe others */
968 	void			*private_data;
969 
970 #ifdef CONFIG_EPOLL
971 	/* Used by fs/eventpoll.c to link all the hooks to this file */
972 	struct list_head	f_ep_links;
973 #endif /* #ifdef CONFIG_EPOLL */
974 	struct address_space	*f_mapping;
975 #ifdef CONFIG_DEBUG_WRITECOUNT
976 	unsigned long f_mnt_write_state;
977 #endif
978 };
979 
980 #define get_file(x)	atomic_long_inc(&(x)->f_count)
981 #define fput_atomic(x)	atomic_long_add_unless(&(x)->f_count, -1, 1)
982 #define file_count(x)	atomic_long_read(&(x)->f_count)
983 
984 #ifdef CONFIG_DEBUG_WRITECOUNT
985 static inline void file_take_write(struct file *f)
986 {
987 	WARN_ON(f->f_mnt_write_state != 0);
988 	f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
989 }
990 static inline void file_release_write(struct file *f)
991 {
992 	f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
993 }
994 static inline void file_reset_write(struct file *f)
995 {
996 	f->f_mnt_write_state = 0;
997 }
998 static inline void file_check_state(struct file *f)
999 {
1000 	/*
1001 	 * At this point, either both or neither of these bits
1002 	 * should be set.
1003 	 */
1004 	WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
1005 	WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
1006 }
1007 static inline int file_check_writeable(struct file *f)
1008 {
1009 	if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
1010 		return 0;
1011 	printk(KERN_WARNING "writeable file with no "
1012 			    "mnt_want_write()\n");
1013 	WARN_ON(1);
1014 	return -EINVAL;
1015 }
1016 #else /* !CONFIG_DEBUG_WRITECOUNT */
1017 static inline void file_take_write(struct file *filp) {}
1018 static inline void file_release_write(struct file *filp) {}
1019 static inline void file_reset_write(struct file *filp) {}
1020 static inline void file_check_state(struct file *filp) {}
1021 static inline int file_check_writeable(struct file *filp)
1022 {
1023 	return 0;
1024 }
1025 #endif /* CONFIG_DEBUG_WRITECOUNT */
1026 
1027 #define	MAX_NON_LFS	((1UL<<31) - 1)
1028 
1029 /* Page cache limit. The filesystems should put that into their s_maxbytes
1030    limits, otherwise bad things can happen in VM. */
1031 #if BITS_PER_LONG==32
1032 #define MAX_LFS_FILESIZE	(((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
1033 #elif BITS_PER_LONG==64
1034 #define MAX_LFS_FILESIZE 	0x7fffffffffffffffUL
1035 #endif
1036 
1037 #define FL_POSIX	1
1038 #define FL_FLOCK	2
1039 #define FL_ACCESS	8	/* not trying to lock, just looking */
1040 #define FL_EXISTS	16	/* when unlocking, test for existence */
1041 #define FL_LEASE	32	/* lease held on this file */
1042 #define FL_CLOSE	64	/* unlock on close */
1043 #define FL_SLEEP	128	/* A blocking lock */
1044 
1045 /*
1046  * Special return value from posix_lock_file() and vfs_lock_file() for
1047  * asynchronous locking.
1048  */
1049 #define FILE_LOCK_DEFERRED 1
1050 
1051 /*
1052  * The POSIX file lock owner is determined by
1053  * the "struct files_struct" in the thread group
1054  * (or NULL for no owner - BSD locks).
1055  *
1056  * Lockd stuffs a "host" pointer into this.
1057  */
1058 typedef struct files_struct *fl_owner_t;
1059 
1060 struct file_lock_operations {
1061 	void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1062 	void (*fl_release_private)(struct file_lock *);
1063 };
1064 
1065 struct lock_manager_operations {
1066 	int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
1067 	void (*fl_notify)(struct file_lock *);	/* unblock callback */
1068 	int (*fl_grant)(struct file_lock *, struct file_lock *, int);
1069 	void (*fl_release_private)(struct file_lock *);
1070 	void (*fl_break)(struct file_lock *);
1071 	int (*fl_change)(struct file_lock **, int);
1072 };
1073 
1074 struct lock_manager {
1075 	struct list_head list;
1076 };
1077 
1078 void locks_start_grace(struct lock_manager *);
1079 void locks_end_grace(struct lock_manager *);
1080 int locks_in_grace(void);
1081 
1082 /* that will die - we need it for nfs_lock_info */
1083 #include <linux/nfs_fs_i.h>
1084 
1085 struct file_lock {
1086 	struct file_lock *fl_next;	/* singly linked list for this inode  */
1087 	struct list_head fl_link;	/* doubly linked list of all locks */
1088 	struct list_head fl_block;	/* circular list of blocked processes */
1089 	fl_owner_t fl_owner;
1090 	unsigned char fl_flags;
1091 	unsigned char fl_type;
1092 	unsigned int fl_pid;
1093 	struct pid *fl_nspid;
1094 	wait_queue_head_t fl_wait;
1095 	struct file *fl_file;
1096 	loff_t fl_start;
1097 	loff_t fl_end;
1098 
1099 	struct fasync_struct *	fl_fasync; /* for lease break notifications */
1100 	unsigned long fl_break_time;	/* for nonblocking lease breaks */
1101 
1102 	const struct file_lock_operations *fl_ops;	/* Callbacks for filesystems */
1103 	const struct lock_manager_operations *fl_lmops;	/* Callbacks for lockmanagers */
1104 	union {
1105 		struct nfs_lock_info	nfs_fl;
1106 		struct nfs4_lock_info	nfs4_fl;
1107 		struct {
1108 			struct list_head link;	/* link in AFS vnode's pending_locks list */
1109 			int state;		/* state of grant or error if -ve */
1110 		} afs;
1111 	} fl_u;
1112 };
1113 
1114 /* The following constant reflects the upper bound of the file/locking space */
1115 #ifndef OFFSET_MAX
1116 #define INT_LIMIT(x)	(~((x)1 << (sizeof(x)*8 - 1)))
1117 #define OFFSET_MAX	INT_LIMIT(loff_t)
1118 #define OFFT_OFFSET_MAX	INT_LIMIT(off_t)
1119 #endif
1120 
1121 #include <linux/fcntl.h>
1122 
1123 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1124 
1125 #ifdef CONFIG_FILE_LOCKING
1126 extern int fcntl_getlk(struct file *, struct flock __user *);
1127 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1128 			struct flock __user *);
1129 
1130 #if BITS_PER_LONG == 32
1131 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
1132 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1133 			struct flock64 __user *);
1134 #endif
1135 
1136 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1137 extern int fcntl_getlease(struct file *filp);
1138 
1139 /* fs/locks.c */
1140 void locks_free_lock(struct file_lock *fl);
1141 extern void locks_init_lock(struct file_lock *);
1142 extern struct file_lock * locks_alloc_lock(void);
1143 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1144 extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
1145 extern void locks_remove_posix(struct file *, fl_owner_t);
1146 extern void locks_remove_flock(struct file *);
1147 extern void locks_release_private(struct file_lock *);
1148 extern void posix_test_lock(struct file *, struct file_lock *);
1149 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1150 extern int posix_lock_file_wait(struct file *, struct file_lock *);
1151 extern int posix_unblock_lock(struct file *, struct file_lock *);
1152 extern int vfs_test_lock(struct file *, struct file_lock *);
1153 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1154 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1155 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1156 extern int __break_lease(struct inode *inode, unsigned int flags);
1157 extern void lease_get_mtime(struct inode *, struct timespec *time);
1158 extern int generic_setlease(struct file *, long, struct file_lock **);
1159 extern int vfs_setlease(struct file *, long, struct file_lock **);
1160 extern int lease_modify(struct file_lock **, int);
1161 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1162 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1163 extern void lock_flocks(void);
1164 extern void unlock_flocks(void);
1165 #else /* !CONFIG_FILE_LOCKING */
1166 static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1167 {
1168 	return -EINVAL;
1169 }
1170 
1171 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1172 			      unsigned int cmd, struct flock __user *user)
1173 {
1174 	return -EACCES;
1175 }
1176 
1177 #if BITS_PER_LONG == 32
1178 static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1179 {
1180 	return -EINVAL;
1181 }
1182 
1183 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1184 				unsigned int cmd, struct flock64 __user *user)
1185 {
1186 	return -EACCES;
1187 }
1188 #endif
1189 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1190 {
1191 	return 0;
1192 }
1193 
1194 static inline int fcntl_getlease(struct file *filp)
1195 {
1196 	return 0;
1197 }
1198 
1199 static inline void locks_init_lock(struct file_lock *fl)
1200 {
1201 	return;
1202 }
1203 
1204 static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1205 {
1206 	return;
1207 }
1208 
1209 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1210 {
1211 	return;
1212 }
1213 
1214 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1215 {
1216 	return;
1217 }
1218 
1219 static inline void locks_remove_flock(struct file *filp)
1220 {
1221 	return;
1222 }
1223 
1224 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1225 {
1226 	return;
1227 }
1228 
1229 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1230 				  struct file_lock *conflock)
1231 {
1232 	return -ENOLCK;
1233 }
1234 
1235 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1236 {
1237 	return -ENOLCK;
1238 }
1239 
1240 static inline int posix_unblock_lock(struct file *filp,
1241 				     struct file_lock *waiter)
1242 {
1243 	return -ENOENT;
1244 }
1245 
1246 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1247 {
1248 	return 0;
1249 }
1250 
1251 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1252 				struct file_lock *fl, struct file_lock *conf)
1253 {
1254 	return -ENOLCK;
1255 }
1256 
1257 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1258 {
1259 	return 0;
1260 }
1261 
1262 static inline int flock_lock_file_wait(struct file *filp,
1263 				       struct file_lock *request)
1264 {
1265 	return -ENOLCK;
1266 }
1267 
1268 static inline int __break_lease(struct inode *inode, unsigned int mode)
1269 {
1270 	return 0;
1271 }
1272 
1273 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1274 {
1275 	return;
1276 }
1277 
1278 static inline int generic_setlease(struct file *filp, long arg,
1279 				    struct file_lock **flp)
1280 {
1281 	return -EINVAL;
1282 }
1283 
1284 static inline int vfs_setlease(struct file *filp, long arg,
1285 			       struct file_lock **lease)
1286 {
1287 	return -EINVAL;
1288 }
1289 
1290 static inline int lease_modify(struct file_lock **before, int arg)
1291 {
1292 	return -EINVAL;
1293 }
1294 
1295 static inline int lock_may_read(struct inode *inode, loff_t start,
1296 				unsigned long len)
1297 {
1298 	return 1;
1299 }
1300 
1301 static inline int lock_may_write(struct inode *inode, loff_t start,
1302 				 unsigned long len)
1303 {
1304 	return 1;
1305 }
1306 
1307 static inline void lock_flocks(void)
1308 {
1309 }
1310 
1311 static inline void unlock_flocks(void)
1312 {
1313 }
1314 
1315 #endif /* !CONFIG_FILE_LOCKING */
1316 
1317 
1318 struct fasync_struct {
1319 	spinlock_t		fa_lock;
1320 	int			magic;
1321 	int			fa_fd;
1322 	struct fasync_struct	*fa_next; /* singly linked list */
1323 	struct file		*fa_file;
1324 	struct rcu_head		fa_rcu;
1325 };
1326 
1327 #define FASYNC_MAGIC 0x4601
1328 
1329 /* SMP safe fasync helpers: */
1330 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1331 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1332 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1333 extern struct fasync_struct *fasync_alloc(void);
1334 extern void fasync_free(struct fasync_struct *);
1335 
1336 /* can be called from interrupts */
1337 extern void kill_fasync(struct fasync_struct **, int, int);
1338 
1339 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1340 extern int f_setown(struct file *filp, unsigned long arg, int force);
1341 extern void f_delown(struct file *filp);
1342 extern pid_t f_getown(struct file *filp);
1343 extern int send_sigurg(struct fown_struct *fown);
1344 
1345 /*
1346  *	Umount options
1347  */
1348 
1349 #define MNT_FORCE	0x00000001	/* Attempt to forcibily umount */
1350 #define MNT_DETACH	0x00000002	/* Just detach from the tree */
1351 #define MNT_EXPIRE	0x00000004	/* Mark for expiry */
1352 #define UMOUNT_NOFOLLOW	0x00000008	/* Don't follow symlink on umount */
1353 #define UMOUNT_UNUSED	0x80000000	/* Flag guaranteed to be unused */
1354 
1355 extern struct list_head super_blocks;
1356 extern spinlock_t sb_lock;
1357 
1358 struct super_block {
1359 	struct list_head	s_list;		/* Keep this first */
1360 	dev_t			s_dev;		/* search index; _not_ kdev_t */
1361 	unsigned char		s_dirt;
1362 	unsigned char		s_blocksize_bits;
1363 	unsigned long		s_blocksize;
1364 	loff_t			s_maxbytes;	/* Max file size */
1365 	struct file_system_type	*s_type;
1366 	const struct super_operations	*s_op;
1367 	const struct dquot_operations	*dq_op;
1368 	const struct quotactl_ops	*s_qcop;
1369 	const struct export_operations *s_export_op;
1370 	unsigned long		s_flags;
1371 	unsigned long		s_magic;
1372 	struct dentry		*s_root;
1373 	struct rw_semaphore	s_umount;
1374 	struct mutex		s_lock;
1375 	int			s_count;
1376 	atomic_t		s_active;
1377 #ifdef CONFIG_SECURITY
1378 	void                    *s_security;
1379 #endif
1380 	const struct xattr_handler **s_xattr;
1381 
1382 	struct list_head	s_inodes;	/* all inodes */
1383 	struct hlist_bl_head	s_anon;		/* anonymous dentries for (nfs) exporting */
1384 #ifdef CONFIG_SMP
1385 	struct list_head __percpu *s_files;
1386 #else
1387 	struct list_head	s_files;
1388 #endif
1389 	/* s_dentry_lru, s_nr_dentry_unused protected by dcache.c lru locks */
1390 	struct list_head	s_dentry_lru;	/* unused dentry lru */
1391 	int			s_nr_dentry_unused;	/* # of dentry on lru */
1392 
1393 	struct block_device	*s_bdev;
1394 	struct backing_dev_info *s_bdi;
1395 	struct mtd_info		*s_mtd;
1396 	struct list_head	s_instances;
1397 	struct quota_info	s_dquot;	/* Diskquota specific options */
1398 
1399 	int			s_frozen;
1400 	wait_queue_head_t	s_wait_unfrozen;
1401 
1402 	char s_id[32];				/* Informational name */
1403 
1404 	void 			*s_fs_info;	/* Filesystem private info */
1405 	fmode_t			s_mode;
1406 
1407 	/* Granularity of c/m/atime in ns.
1408 	   Cannot be worse than a second */
1409 	u32		   s_time_gran;
1410 
1411 	/*
1412 	 * The next field is for VFS *only*. No filesystems have any business
1413 	 * even looking at it. You had been warned.
1414 	 */
1415 	struct mutex s_vfs_rename_mutex;	/* Kludge */
1416 
1417 	/*
1418 	 * Filesystem subtype.  If non-empty the filesystem type field
1419 	 * in /proc/mounts will be "type.subtype"
1420 	 */
1421 	char *s_subtype;
1422 
1423 	/*
1424 	 * Saved mount options for lazy filesystems using
1425 	 * generic_show_options()
1426 	 */
1427 	char __rcu *s_options;
1428 	const struct dentry_operations *s_d_op; /* default d_op for dentries */
1429 };
1430 
1431 extern struct timespec current_fs_time(struct super_block *sb);
1432 
1433 /*
1434  * Snapshotting support.
1435  */
1436 enum {
1437 	SB_UNFROZEN = 0,
1438 	SB_FREEZE_WRITE	= 1,
1439 	SB_FREEZE_TRANS = 2,
1440 };
1441 
1442 #define vfs_check_frozen(sb, level) \
1443 	wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
1444 
1445 #define get_fs_excl() atomic_inc(&current->fs_excl)
1446 #define put_fs_excl() atomic_dec(&current->fs_excl)
1447 #define has_fs_excl() atomic_read(&current->fs_excl)
1448 
1449 #define is_owner_or_cap(inode)	\
1450 	((current_fsuid() == (inode)->i_uid) || capable(CAP_FOWNER))
1451 
1452 /* not quite ready to be deprecated, but... */
1453 extern void lock_super(struct super_block *);
1454 extern void unlock_super(struct super_block *);
1455 
1456 /*
1457  * VFS helper functions..
1458  */
1459 extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
1460 extern int vfs_mkdir(struct inode *, struct dentry *, int);
1461 extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
1462 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1463 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1464 extern int vfs_rmdir(struct inode *, struct dentry *);
1465 extern int vfs_unlink(struct inode *, struct dentry *);
1466 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1467 
1468 /*
1469  * VFS dentry helper functions.
1470  */
1471 extern void dentry_unhash(struct dentry *dentry);
1472 
1473 /*
1474  * VFS file helper functions.
1475  */
1476 extern int file_permission(struct file *, int);
1477 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1478 			mode_t mode);
1479 /*
1480  * VFS FS_IOC_FIEMAP helper definitions.
1481  */
1482 struct fiemap_extent_info {
1483 	unsigned int fi_flags;		/* Flags as passed from user */
1484 	unsigned int fi_extents_mapped;	/* Number of mapped extents */
1485 	unsigned int fi_extents_max;	/* Size of fiemap_extent array */
1486 	struct fiemap_extent __user *fi_extents_start; /* Start of
1487 							fiemap_extent array */
1488 };
1489 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1490 			    u64 phys, u64 len, u32 flags);
1491 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1492 
1493 /*
1494  * File types
1495  *
1496  * NOTE! These match bits 12..15 of stat.st_mode
1497  * (ie "(i_mode >> 12) & 15").
1498  */
1499 #define DT_UNKNOWN	0
1500 #define DT_FIFO		1
1501 #define DT_CHR		2
1502 #define DT_DIR		4
1503 #define DT_BLK		6
1504 #define DT_REG		8
1505 #define DT_LNK		10
1506 #define DT_SOCK		12
1507 #define DT_WHT		14
1508 
1509 /*
1510  * This is the "filldir" function type, used by readdir() to let
1511  * the kernel specify what kind of dirent layout it wants to have.
1512  * This allows the kernel to read directories into kernel space or
1513  * to have different dirent layouts depending on the binary type.
1514  */
1515 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1516 struct block_device_operations;
1517 
1518 /* These macros are for out of kernel modules to test that
1519  * the kernel supports the unlocked_ioctl and compat_ioctl
1520  * fields in struct file_operations. */
1521 #define HAVE_COMPAT_IOCTL 1
1522 #define HAVE_UNLOCKED_IOCTL 1
1523 
1524 /*
1525  * NOTE:
1526  * all file operations except setlease can be called without
1527  * the big kernel lock held in all filesystems.
1528  */
1529 struct file_operations {
1530 	struct module *owner;
1531 	loff_t (*llseek) (struct file *, loff_t, int);
1532 	ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1533 	ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1534 	ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1535 	ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1536 	int (*readdir) (struct file *, void *, filldir_t);
1537 	unsigned int (*poll) (struct file *, struct poll_table_struct *);
1538 	long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1539 	long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1540 	int (*mmap) (struct file *, struct vm_area_struct *);
1541 	int (*open) (struct inode *, struct file *);
1542 	int (*flush) (struct file *, fl_owner_t id);
1543 	int (*release) (struct inode *, struct file *);
1544 	int (*fsync) (struct file *, int datasync);
1545 	int (*aio_fsync) (struct kiocb *, int datasync);
1546 	int (*fasync) (int, struct file *, int);
1547 	int (*lock) (struct file *, int, struct file_lock *);
1548 	ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1549 	unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1550 	int (*check_flags)(int);
1551 	int (*flock) (struct file *, int, struct file_lock *);
1552 	ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1553 	ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1554 	int (*setlease)(struct file *, long, struct file_lock **);
1555 	long (*fallocate)(struct file *file, int mode, loff_t offset,
1556 			  loff_t len);
1557 };
1558 
1559 #define IPERM_FLAG_RCU	0x0001
1560 
1561 struct inode_operations {
1562 	struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1563 	void * (*follow_link) (struct dentry *, struct nameidata *);
1564 	int (*permission) (struct inode *, int, unsigned int);
1565 	int (*check_acl)(struct inode *, int, unsigned int);
1566 
1567 	int (*readlink) (struct dentry *, char __user *,int);
1568 	void (*put_link) (struct dentry *, struct nameidata *, void *);
1569 
1570 	int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1571 	int (*link) (struct dentry *,struct inode *,struct dentry *);
1572 	int (*unlink) (struct inode *,struct dentry *);
1573 	int (*symlink) (struct inode *,struct dentry *,const char *);
1574 	int (*mkdir) (struct inode *,struct dentry *,int);
1575 	int (*rmdir) (struct inode *,struct dentry *);
1576 	int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1577 	int (*rename) (struct inode *, struct dentry *,
1578 			struct inode *, struct dentry *);
1579 	void (*truncate) (struct inode *);
1580 	int (*setattr) (struct dentry *, struct iattr *);
1581 	int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1582 	int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1583 	ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1584 	ssize_t (*listxattr) (struct dentry *, char *, size_t);
1585 	int (*removexattr) (struct dentry *, const char *);
1586 	void (*truncate_range)(struct inode *, loff_t, loff_t);
1587 	int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1588 		      u64 len);
1589 } ____cacheline_aligned;
1590 
1591 struct seq_file;
1592 
1593 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1594 				unsigned long nr_segs, unsigned long fast_segs,
1595 				struct iovec *fast_pointer,
1596 				struct iovec **ret_pointer);
1597 
1598 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1599 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1600 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1601 		unsigned long, loff_t *);
1602 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1603 		unsigned long, loff_t *);
1604 
1605 struct super_operations {
1606    	struct inode *(*alloc_inode)(struct super_block *sb);
1607 	void (*destroy_inode)(struct inode *);
1608 
1609    	void (*dirty_inode) (struct inode *);
1610 	int (*write_inode) (struct inode *, struct writeback_control *wbc);
1611 	int (*drop_inode) (struct inode *);
1612 	void (*evict_inode) (struct inode *);
1613 	void (*put_super) (struct super_block *);
1614 	void (*write_super) (struct super_block *);
1615 	int (*sync_fs)(struct super_block *sb, int wait);
1616 	int (*freeze_fs) (struct super_block *);
1617 	int (*unfreeze_fs) (struct super_block *);
1618 	int (*statfs) (struct dentry *, struct kstatfs *);
1619 	int (*remount_fs) (struct super_block *, int *, char *);
1620 	void (*umount_begin) (struct super_block *);
1621 
1622 	int (*show_options)(struct seq_file *, struct vfsmount *);
1623 	int (*show_stats)(struct seq_file *, struct vfsmount *);
1624 #ifdef CONFIG_QUOTA
1625 	ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1626 	ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1627 #endif
1628 	int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1629 };
1630 
1631 /*
1632  * Inode state bits.  Protected by inode_lock.
1633  *
1634  * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1635  * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1636  *
1637  * Four bits define the lifetime of an inode.  Initially, inodes are I_NEW,
1638  * until that flag is cleared.  I_WILL_FREE, I_FREEING and I_CLEAR are set at
1639  * various stages of removing an inode.
1640  *
1641  * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1642  *
1643  * I_DIRTY_SYNC		Inode is dirty, but doesn't have to be written on
1644  *			fdatasync().  i_atime is the usual cause.
1645  * I_DIRTY_DATASYNC	Data-related inode changes pending. We keep track of
1646  *			these changes separately from I_DIRTY_SYNC so that we
1647  *			don't have to write inode on fdatasync() when only
1648  *			mtime has changed in it.
1649  * I_DIRTY_PAGES	Inode has dirty pages.  Inode itself may be clean.
1650  * I_NEW		Serves as both a mutex and completion notification.
1651  *			New inodes set I_NEW.  If two processes both create
1652  *			the same inode, one of them will release its inode and
1653  *			wait for I_NEW to be released before returning.
1654  *			Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1655  *			also cause waiting on I_NEW, without I_NEW actually
1656  *			being set.  find_inode() uses this to prevent returning
1657  *			nearly-dead inodes.
1658  * I_WILL_FREE		Must be set when calling write_inode_now() if i_count
1659  *			is zero.  I_FREEING must be set when I_WILL_FREE is
1660  *			cleared.
1661  * I_FREEING		Set when inode is about to be freed but still has dirty
1662  *			pages or buffers attached or the inode itself is still
1663  *			dirty.
1664  * I_CLEAR		Added by end_writeback().  In this state the inode is clean
1665  *			and can be destroyed.  Inode keeps I_FREEING.
1666  *
1667  *			Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1668  *			prohibited for many purposes.  iget() must wait for
1669  *			the inode to be completely released, then create it
1670  *			anew.  Other functions will just ignore such inodes,
1671  *			if appropriate.  I_NEW is used for waiting.
1672  *
1673  * I_SYNC		Synchonized write of dirty inode data.  The bits is
1674  *			set during data writeback, and cleared with a wakeup
1675  *			on the bit address once it is done.
1676  *
1677  * Q: What is the difference between I_WILL_FREE and I_FREEING?
1678  */
1679 #define I_DIRTY_SYNC		(1 << 0)
1680 #define I_DIRTY_DATASYNC	(1 << 1)
1681 #define I_DIRTY_PAGES		(1 << 2)
1682 #define __I_NEW			3
1683 #define I_NEW			(1 << __I_NEW)
1684 #define I_WILL_FREE		(1 << 4)
1685 #define I_FREEING		(1 << 5)
1686 #define I_CLEAR			(1 << 6)
1687 #define __I_SYNC		7
1688 #define I_SYNC			(1 << __I_SYNC)
1689 #define I_REFERENCED		(1 << 8)
1690 
1691 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1692 
1693 extern void __mark_inode_dirty(struct inode *, int);
1694 static inline void mark_inode_dirty(struct inode *inode)
1695 {
1696 	__mark_inode_dirty(inode, I_DIRTY);
1697 }
1698 
1699 static inline void mark_inode_dirty_sync(struct inode *inode)
1700 {
1701 	__mark_inode_dirty(inode, I_DIRTY_SYNC);
1702 }
1703 
1704 /**
1705  * inc_nlink - directly increment an inode's link count
1706  * @inode: inode
1707  *
1708  * This is a low-level filesystem helper to replace any
1709  * direct filesystem manipulation of i_nlink.  Currently,
1710  * it is only here for parity with dec_nlink().
1711  */
1712 static inline void inc_nlink(struct inode *inode)
1713 {
1714 	inode->i_nlink++;
1715 }
1716 
1717 static inline void inode_inc_link_count(struct inode *inode)
1718 {
1719 	inc_nlink(inode);
1720 	mark_inode_dirty(inode);
1721 }
1722 
1723 /**
1724  * drop_nlink - directly drop an inode's link count
1725  * @inode: inode
1726  *
1727  * This is a low-level filesystem helper to replace any
1728  * direct filesystem manipulation of i_nlink.  In cases
1729  * where we are attempting to track writes to the
1730  * filesystem, a decrement to zero means an imminent
1731  * write when the file is truncated and actually unlinked
1732  * on the filesystem.
1733  */
1734 static inline void drop_nlink(struct inode *inode)
1735 {
1736 	inode->i_nlink--;
1737 }
1738 
1739 /**
1740  * clear_nlink - directly zero an inode's link count
1741  * @inode: inode
1742  *
1743  * This is a low-level filesystem helper to replace any
1744  * direct filesystem manipulation of i_nlink.  See
1745  * drop_nlink() for why we care about i_nlink hitting zero.
1746  */
1747 static inline void clear_nlink(struct inode *inode)
1748 {
1749 	inode->i_nlink = 0;
1750 }
1751 
1752 static inline void inode_dec_link_count(struct inode *inode)
1753 {
1754 	drop_nlink(inode);
1755 	mark_inode_dirty(inode);
1756 }
1757 
1758 /**
1759  * inode_inc_iversion - increments i_version
1760  * @inode: inode that need to be updated
1761  *
1762  * Every time the inode is modified, the i_version field will be incremented.
1763  * The filesystem has to be mounted with i_version flag
1764  */
1765 
1766 static inline void inode_inc_iversion(struct inode *inode)
1767 {
1768        spin_lock(&inode->i_lock);
1769        inode->i_version++;
1770        spin_unlock(&inode->i_lock);
1771 }
1772 
1773 extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
1774 static inline void file_accessed(struct file *file)
1775 {
1776 	if (!(file->f_flags & O_NOATIME))
1777 		touch_atime(file->f_path.mnt, file->f_path.dentry);
1778 }
1779 
1780 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1781 int sync_inode_metadata(struct inode *inode, int wait);
1782 
1783 struct file_system_type {
1784 	const char *name;
1785 	int fs_flags;
1786 	int (*get_sb) (struct file_system_type *, int,
1787 		       const char *, void *, struct vfsmount *);
1788 	struct dentry *(*mount) (struct file_system_type *, int,
1789 		       const char *, void *);
1790 	void (*kill_sb) (struct super_block *);
1791 	struct module *owner;
1792 	struct file_system_type * next;
1793 	struct list_head fs_supers;
1794 
1795 	struct lock_class_key s_lock_key;
1796 	struct lock_class_key s_umount_key;
1797 	struct lock_class_key s_vfs_rename_key;
1798 
1799 	struct lock_class_key i_lock_key;
1800 	struct lock_class_key i_mutex_key;
1801 	struct lock_class_key i_mutex_dir_key;
1802 	struct lock_class_key i_alloc_sem_key;
1803 };
1804 
1805 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1806 	void *data, int (*fill_super)(struct super_block *, void *, int));
1807 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1808 	int flags, const char *dev_name, void *data,
1809 	int (*fill_super)(struct super_block *, void *, int));
1810 extern int get_sb_bdev(struct file_system_type *fs_type,
1811 	int flags, const char *dev_name, void *data,
1812 	int (*fill_super)(struct super_block *, void *, int),
1813 	struct vfsmount *mnt);
1814 extern struct dentry *mount_single(struct file_system_type *fs_type,
1815 	int flags, void *data,
1816 	int (*fill_super)(struct super_block *, void *, int));
1817 extern int get_sb_single(struct file_system_type *fs_type,
1818 	int flags, void *data,
1819 	int (*fill_super)(struct super_block *, void *, int),
1820 	struct vfsmount *mnt);
1821 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1822 	int flags, void *data,
1823 	int (*fill_super)(struct super_block *, void *, int));
1824 extern int get_sb_nodev(struct file_system_type *fs_type,
1825 	int flags, void *data,
1826 	int (*fill_super)(struct super_block *, void *, int),
1827 	struct vfsmount *mnt);
1828 void generic_shutdown_super(struct super_block *sb);
1829 void kill_block_super(struct super_block *sb);
1830 void kill_anon_super(struct super_block *sb);
1831 void kill_litter_super(struct super_block *sb);
1832 void deactivate_super(struct super_block *sb);
1833 void deactivate_locked_super(struct super_block *sb);
1834 int set_anon_super(struct super_block *s, void *data);
1835 struct super_block *sget(struct file_system_type *type,
1836 			int (*test)(struct super_block *,void *),
1837 			int (*set)(struct super_block *,void *),
1838 			void *data);
1839 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1840 	const struct super_operations *ops,
1841 	const struct dentry_operations *dops,
1842 	unsigned long);
1843 extern void simple_set_mnt(struct vfsmount *mnt, struct super_block *sb);
1844 
1845 static inline void sb_mark_dirty(struct super_block *sb)
1846 {
1847 	sb->s_dirt = 1;
1848 }
1849 static inline void sb_mark_clean(struct super_block *sb)
1850 {
1851 	sb->s_dirt = 0;
1852 }
1853 static inline int sb_is_dirty(struct super_block *sb)
1854 {
1855 	return sb->s_dirt;
1856 }
1857 
1858 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1859 #define fops_get(fops) \
1860 	(((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1861 #define fops_put(fops) \
1862 	do { if (fops) module_put((fops)->owner); } while(0)
1863 
1864 extern int register_filesystem(struct file_system_type *);
1865 extern int unregister_filesystem(struct file_system_type *);
1866 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1867 #define kern_mount(type) kern_mount_data(type, NULL)
1868 extern int may_umount_tree(struct vfsmount *);
1869 extern int may_umount(struct vfsmount *);
1870 extern long do_mount(char *, char *, char *, unsigned long, void *);
1871 extern struct vfsmount *collect_mounts(struct path *);
1872 extern void drop_collected_mounts(struct vfsmount *);
1873 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1874 			  struct vfsmount *);
1875 extern int vfs_statfs(struct path *, struct kstatfs *);
1876 extern int statfs_by_dentry(struct dentry *, struct kstatfs *);
1877 extern int freeze_super(struct super_block *super);
1878 extern int thaw_super(struct super_block *super);
1879 
1880 extern int current_umask(void);
1881 
1882 /* /sys/fs */
1883 extern struct kobject *fs_kobj;
1884 
1885 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1886 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1887 
1888 #define FLOCK_VERIFY_READ  1
1889 #define FLOCK_VERIFY_WRITE 2
1890 
1891 #ifdef CONFIG_FILE_LOCKING
1892 extern int locks_mandatory_locked(struct inode *);
1893 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1894 
1895 /*
1896  * Candidates for mandatory locking have the setgid bit set
1897  * but no group execute bit -  an otherwise meaningless combination.
1898  */
1899 
1900 static inline int __mandatory_lock(struct inode *ino)
1901 {
1902 	return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1903 }
1904 
1905 /*
1906  * ... and these candidates should be on MS_MANDLOCK mounted fs,
1907  * otherwise these will be advisory locks
1908  */
1909 
1910 static inline int mandatory_lock(struct inode *ino)
1911 {
1912 	return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1913 }
1914 
1915 static inline int locks_verify_locked(struct inode *inode)
1916 {
1917 	if (mandatory_lock(inode))
1918 		return locks_mandatory_locked(inode);
1919 	return 0;
1920 }
1921 
1922 static inline int locks_verify_truncate(struct inode *inode,
1923 				    struct file *filp,
1924 				    loff_t size)
1925 {
1926 	if (inode->i_flock && mandatory_lock(inode))
1927 		return locks_mandatory_area(
1928 			FLOCK_VERIFY_WRITE, inode, filp,
1929 			size < inode->i_size ? size : inode->i_size,
1930 			(size < inode->i_size ? inode->i_size - size
1931 			 : size - inode->i_size)
1932 		);
1933 	return 0;
1934 }
1935 
1936 static inline int break_lease(struct inode *inode, unsigned int mode)
1937 {
1938 	if (inode->i_flock)
1939 		return __break_lease(inode, mode);
1940 	return 0;
1941 }
1942 #else /* !CONFIG_FILE_LOCKING */
1943 static inline int locks_mandatory_locked(struct inode *inode)
1944 {
1945 	return 0;
1946 }
1947 
1948 static inline int locks_mandatory_area(int rw, struct inode *inode,
1949 				       struct file *filp, loff_t offset,
1950 				       size_t count)
1951 {
1952 	return 0;
1953 }
1954 
1955 static inline int __mandatory_lock(struct inode *inode)
1956 {
1957 	return 0;
1958 }
1959 
1960 static inline int mandatory_lock(struct inode *inode)
1961 {
1962 	return 0;
1963 }
1964 
1965 static inline int locks_verify_locked(struct inode *inode)
1966 {
1967 	return 0;
1968 }
1969 
1970 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
1971 					size_t size)
1972 {
1973 	return 0;
1974 }
1975 
1976 static inline int break_lease(struct inode *inode, unsigned int mode)
1977 {
1978 	return 0;
1979 }
1980 
1981 #endif /* CONFIG_FILE_LOCKING */
1982 
1983 /* fs/open.c */
1984 
1985 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
1986 		       struct file *filp);
1987 extern int do_fallocate(struct file *file, int mode, loff_t offset,
1988 			loff_t len);
1989 extern long do_sys_open(int dfd, const char __user *filename, int flags,
1990 			int mode);
1991 extern struct file *filp_open(const char *, int, int);
1992 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int,
1993 				 const struct cred *);
1994 extern int filp_close(struct file *, fl_owner_t id);
1995 extern char * getname(const char __user *);
1996 
1997 /* fs/ioctl.c */
1998 
1999 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2000 
2001 /* fs/dcache.c */
2002 extern void __init vfs_caches_init_early(void);
2003 extern void __init vfs_caches_init(unsigned long);
2004 
2005 extern struct kmem_cache *names_cachep;
2006 
2007 #define __getname_gfp(gfp)	kmem_cache_alloc(names_cachep, (gfp))
2008 #define __getname()		__getname_gfp(GFP_KERNEL)
2009 #define __putname(name)		kmem_cache_free(names_cachep, (void *)(name))
2010 #ifndef CONFIG_AUDITSYSCALL
2011 #define putname(name)   __putname(name)
2012 #else
2013 extern void putname(const char *name);
2014 #endif
2015 
2016 #ifdef CONFIG_BLOCK
2017 extern int register_blkdev(unsigned int, const char *);
2018 extern void unregister_blkdev(unsigned int, const char *);
2019 extern struct block_device *bdget(dev_t);
2020 extern struct block_device *bdgrab(struct block_device *bdev);
2021 extern void bd_set_size(struct block_device *, loff_t size);
2022 extern void bd_forget(struct inode *inode);
2023 extern void bdput(struct block_device *);
2024 extern void invalidate_bdev(struct block_device *);
2025 extern int sync_blockdev(struct block_device *bdev);
2026 extern struct super_block *freeze_bdev(struct block_device *);
2027 extern void emergency_thaw_all(void);
2028 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2029 extern int fsync_bdev(struct block_device *);
2030 #else
2031 static inline void bd_forget(struct inode *inode) {}
2032 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2033 static inline void invalidate_bdev(struct block_device *bdev) {}
2034 
2035 static inline struct super_block *freeze_bdev(struct block_device *sb)
2036 {
2037 	return NULL;
2038 }
2039 
2040 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2041 {
2042 	return 0;
2043 }
2044 #endif
2045 extern int sync_filesystem(struct super_block *);
2046 extern const struct file_operations def_blk_fops;
2047 extern const struct file_operations def_chr_fops;
2048 extern const struct file_operations bad_sock_fops;
2049 extern const struct file_operations def_fifo_fops;
2050 #ifdef CONFIG_BLOCK
2051 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2052 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2053 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2054 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2055 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2056 					       void *holder);
2057 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2058 					      void *holder);
2059 extern int blkdev_put(struct block_device *bdev, fmode_t mode);
2060 #ifdef CONFIG_SYSFS
2061 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2062 extern void bd_unlink_disk_holder(struct block_device *bdev,
2063 				  struct gendisk *disk);
2064 #else
2065 static inline int bd_link_disk_holder(struct block_device *bdev,
2066 				      struct gendisk *disk)
2067 {
2068 	return 0;
2069 }
2070 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2071 					 struct gendisk *disk)
2072 {
2073 }
2074 #endif
2075 #endif
2076 
2077 /* fs/char_dev.c */
2078 #define CHRDEV_MAJOR_HASH_SIZE	255
2079 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2080 extern int register_chrdev_region(dev_t, unsigned, const char *);
2081 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2082 			     unsigned int count, const char *name,
2083 			     const struct file_operations *fops);
2084 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2085 				unsigned int count, const char *name);
2086 extern void unregister_chrdev_region(dev_t, unsigned);
2087 extern void chrdev_show(struct seq_file *,off_t);
2088 
2089 static inline int register_chrdev(unsigned int major, const char *name,
2090 				  const struct file_operations *fops)
2091 {
2092 	return __register_chrdev(major, 0, 256, name, fops);
2093 }
2094 
2095 static inline void unregister_chrdev(unsigned int major, const char *name)
2096 {
2097 	__unregister_chrdev(major, 0, 256, name);
2098 }
2099 
2100 /* fs/block_dev.c */
2101 #define BDEVNAME_SIZE	32	/* Largest string for a blockdev identifier */
2102 #define BDEVT_SIZE	10	/* Largest string for MAJ:MIN for blkdev */
2103 
2104 #ifdef CONFIG_BLOCK
2105 #define BLKDEV_MAJOR_HASH_SIZE	255
2106 extern const char *__bdevname(dev_t, char *buffer);
2107 extern const char *bdevname(struct block_device *bdev, char *buffer);
2108 extern struct block_device *lookup_bdev(const char *);
2109 extern void blkdev_show(struct seq_file *,off_t);
2110 
2111 #else
2112 #define BLKDEV_MAJOR_HASH_SIZE	0
2113 #endif
2114 
2115 extern void init_special_inode(struct inode *, umode_t, dev_t);
2116 
2117 /* Invalid inode operations -- fs/bad_inode.c */
2118 extern void make_bad_inode(struct inode *);
2119 extern int is_bad_inode(struct inode *);
2120 
2121 extern const struct file_operations read_pipefifo_fops;
2122 extern const struct file_operations write_pipefifo_fops;
2123 extern const struct file_operations rdwr_pipefifo_fops;
2124 
2125 extern int fs_may_remount_ro(struct super_block *);
2126 
2127 #ifdef CONFIG_BLOCK
2128 /*
2129  * return READ, READA, or WRITE
2130  */
2131 #define bio_rw(bio)		((bio)->bi_rw & (RW_MASK | RWA_MASK))
2132 
2133 /*
2134  * return data direction, READ or WRITE
2135  */
2136 #define bio_data_dir(bio)	((bio)->bi_rw & 1)
2137 
2138 extern void check_disk_size_change(struct gendisk *disk,
2139 				   struct block_device *bdev);
2140 extern int revalidate_disk(struct gendisk *);
2141 extern int check_disk_change(struct block_device *);
2142 extern int __invalidate_device(struct block_device *);
2143 extern int invalidate_partition(struct gendisk *, int);
2144 #endif
2145 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2146 					pgoff_t start, pgoff_t end);
2147 
2148 static inline void invalidate_remote_inode(struct inode *inode)
2149 {
2150 	if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2151 	    S_ISLNK(inode->i_mode))
2152 		invalidate_mapping_pages(inode->i_mapping, 0, -1);
2153 }
2154 extern int invalidate_inode_pages2(struct address_space *mapping);
2155 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2156 					 pgoff_t start, pgoff_t end);
2157 extern int write_inode_now(struct inode *, int);
2158 extern int filemap_fdatawrite(struct address_space *);
2159 extern int filemap_flush(struct address_space *);
2160 extern int filemap_fdatawait(struct address_space *);
2161 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2162 				   loff_t lend);
2163 extern int filemap_write_and_wait(struct address_space *mapping);
2164 extern int filemap_write_and_wait_range(struct address_space *mapping,
2165 				        loff_t lstart, loff_t lend);
2166 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2167 				loff_t start, loff_t end, int sync_mode);
2168 extern int filemap_fdatawrite_range(struct address_space *mapping,
2169 				loff_t start, loff_t end);
2170 
2171 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2172 			   int datasync);
2173 extern int vfs_fsync(struct file *file, int datasync);
2174 extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
2175 extern void sync_supers(void);
2176 extern void emergency_sync(void);
2177 extern void emergency_remount(void);
2178 #ifdef CONFIG_BLOCK
2179 extern sector_t bmap(struct inode *, sector_t);
2180 #endif
2181 extern int notify_change(struct dentry *, struct iattr *);
2182 extern int inode_permission(struct inode *, int);
2183 extern int generic_permission(struct inode *, int, unsigned int,
2184 		int (*check_acl)(struct inode *, int, unsigned int));
2185 
2186 static inline bool execute_ok(struct inode *inode)
2187 {
2188 	return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2189 }
2190 
2191 extern int get_write_access(struct inode *);
2192 extern int deny_write_access(struct file *);
2193 static inline void put_write_access(struct inode * inode)
2194 {
2195 	atomic_dec(&inode->i_writecount);
2196 }
2197 static inline void allow_write_access(struct file *file)
2198 {
2199 	if (file)
2200 		atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
2201 }
2202 extern int do_pipe_flags(int *, int);
2203 extern struct file *create_read_pipe(struct file *f, int flags);
2204 extern struct file *create_write_pipe(int flags);
2205 extern void free_write_pipe(struct file *);
2206 
2207 extern struct file *do_filp_open(int dfd, const char *pathname,
2208 		int open_flag, int mode, int acc_mode);
2209 extern int may_open(struct path *, int, int);
2210 
2211 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2212 extern struct file * open_exec(const char *);
2213 
2214 /* fs/dcache.c -- generic fs support functions */
2215 extern int is_subdir(struct dentry *, struct dentry *);
2216 extern int path_is_under(struct path *, struct path *);
2217 extern ino_t find_inode_number(struct dentry *, struct qstr *);
2218 
2219 #include <linux/err.h>
2220 
2221 /* needed for stackable file system support */
2222 extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
2223 
2224 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
2225 
2226 extern int inode_init_always(struct super_block *, struct inode *);
2227 extern void inode_init_once(struct inode *);
2228 extern void ihold(struct inode * inode);
2229 extern void iput(struct inode *);
2230 extern struct inode * igrab(struct inode *);
2231 extern ino_t iunique(struct super_block *, ino_t);
2232 extern int inode_needs_sync(struct inode *inode);
2233 extern int generic_delete_inode(struct inode *inode);
2234 extern int generic_drop_inode(struct inode *inode);
2235 
2236 extern struct inode *ilookup5_nowait(struct super_block *sb,
2237 		unsigned long hashval, int (*test)(struct inode *, void *),
2238 		void *data);
2239 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2240 		int (*test)(struct inode *, void *), void *data);
2241 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2242 
2243 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2244 extern struct inode * iget_locked(struct super_block *, unsigned long);
2245 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2246 extern int insert_inode_locked(struct inode *);
2247 extern void unlock_new_inode(struct inode *);
2248 extern unsigned int get_next_ino(void);
2249 
2250 extern void __iget(struct inode * inode);
2251 extern void iget_failed(struct inode *);
2252 extern void end_writeback(struct inode *);
2253 extern void __destroy_inode(struct inode *);
2254 extern struct inode *new_inode(struct super_block *);
2255 extern void free_inode_nonrcu(struct inode *inode);
2256 extern int should_remove_suid(struct dentry *);
2257 extern int file_remove_suid(struct file *);
2258 
2259 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2260 extern void remove_inode_hash(struct inode *);
2261 static inline void insert_inode_hash(struct inode *inode)
2262 {
2263 	__insert_inode_hash(inode, inode->i_ino);
2264 }
2265 extern void inode_sb_list_add(struct inode *inode);
2266 
2267 #ifdef CONFIG_BLOCK
2268 extern void submit_bio(int, struct bio *);
2269 extern int bdev_read_only(struct block_device *);
2270 #endif
2271 extern int set_blocksize(struct block_device *, int);
2272 extern int sb_set_blocksize(struct super_block *, int);
2273 extern int sb_min_blocksize(struct super_block *, int);
2274 
2275 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2276 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2277 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
2278 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2279 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2280 extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2281 		loff_t *);
2282 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2283 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2284 		unsigned long *, loff_t, loff_t *, size_t, size_t);
2285 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2286 		unsigned long, loff_t, loff_t *, size_t, ssize_t);
2287 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2288 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2289 extern int generic_segment_checks(const struct iovec *iov,
2290 		unsigned long *nr_segs, size_t *count, int access_flags);
2291 
2292 /* fs/block_dev.c */
2293 extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2294 				unsigned long nr_segs, loff_t pos);
2295 extern int blkdev_fsync(struct file *filp, int datasync);
2296 
2297 /* fs/splice.c */
2298 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2299 		struct pipe_inode_info *, size_t, unsigned int);
2300 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2301 		struct pipe_inode_info *, size_t, unsigned int);
2302 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
2303 		struct file *, loff_t *, size_t, unsigned int);
2304 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2305 		struct file *out, loff_t *, size_t len, unsigned int flags);
2306 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2307 		size_t len, unsigned int flags);
2308 
2309 extern void
2310 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2311 extern loff_t noop_llseek(struct file *file, loff_t offset, int origin);
2312 extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
2313 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
2314 extern loff_t generic_file_llseek_unlocked(struct file *file, loff_t offset,
2315 			int origin);
2316 extern int generic_file_open(struct inode * inode, struct file * filp);
2317 extern int nonseekable_open(struct inode * inode, struct file * filp);
2318 
2319 #ifdef CONFIG_FS_XIP
2320 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2321 			     loff_t *ppos);
2322 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2323 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2324 			      size_t len, loff_t *ppos);
2325 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2326 #else
2327 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2328 {
2329 	return 0;
2330 }
2331 #endif
2332 
2333 #ifdef CONFIG_BLOCK
2334 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2335 			    loff_t file_offset);
2336 
2337 enum {
2338 	/* need locking between buffered and direct access */
2339 	DIO_LOCKING	= 0x01,
2340 
2341 	/* filesystem does not support filling holes */
2342 	DIO_SKIP_HOLES	= 0x02,
2343 };
2344 
2345 void dio_end_io(struct bio *bio, int error);
2346 
2347 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2348 	struct block_device *bdev, const struct iovec *iov, loff_t offset,
2349 	unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2350 	dio_submit_t submit_io,	int flags);
2351 
2352 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2353 	struct inode *inode, struct block_device *bdev, const struct iovec *iov,
2354 	loff_t offset, unsigned long nr_segs, get_block_t get_block,
2355 	dio_iodone_t end_io)
2356 {
2357 	return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
2358 				    nr_segs, get_block, end_io, NULL,
2359 				    DIO_LOCKING | DIO_SKIP_HOLES);
2360 }
2361 #endif
2362 
2363 extern const struct file_operations generic_ro_fops;
2364 
2365 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2366 
2367 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2368 extern int vfs_follow_link(struct nameidata *, const char *);
2369 extern int page_readlink(struct dentry *, char __user *, int);
2370 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2371 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2372 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2373 		int nofs);
2374 extern int page_symlink(struct inode *inode, const char *symname, int len);
2375 extern const struct inode_operations page_symlink_inode_operations;
2376 extern int generic_readlink(struct dentry *, char __user *, int);
2377 extern void generic_fillattr(struct inode *, struct kstat *);
2378 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2379 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2380 void inode_add_bytes(struct inode *inode, loff_t bytes);
2381 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2382 loff_t inode_get_bytes(struct inode *inode);
2383 void inode_set_bytes(struct inode *inode, loff_t bytes);
2384 
2385 extern int vfs_readdir(struct file *, filldir_t, void *);
2386 
2387 extern int vfs_stat(const char __user *, struct kstat *);
2388 extern int vfs_lstat(const char __user *, struct kstat *);
2389 extern int vfs_fstat(unsigned int, struct kstat *);
2390 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2391 
2392 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2393 		    unsigned long arg);
2394 extern int __generic_block_fiemap(struct inode *inode,
2395 				  struct fiemap_extent_info *fieinfo,
2396 				  loff_t start, loff_t len,
2397 				  get_block_t *get_block);
2398 extern int generic_block_fiemap(struct inode *inode,
2399 				struct fiemap_extent_info *fieinfo, u64 start,
2400 				u64 len, get_block_t *get_block);
2401 
2402 extern void get_filesystem(struct file_system_type *fs);
2403 extern void put_filesystem(struct file_system_type *fs);
2404 extern struct file_system_type *get_fs_type(const char *name);
2405 extern struct super_block *get_super(struct block_device *);
2406 extern struct super_block *get_active_super(struct block_device *bdev);
2407 extern struct super_block *user_get_super(dev_t);
2408 extern void drop_super(struct super_block *sb);
2409 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2410 
2411 extern int dcache_dir_open(struct inode *, struct file *);
2412 extern int dcache_dir_close(struct inode *, struct file *);
2413 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2414 extern int dcache_readdir(struct file *, void *, filldir_t);
2415 extern int simple_setattr(struct dentry *, struct iattr *);
2416 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2417 extern int simple_statfs(struct dentry *, struct kstatfs *);
2418 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2419 extern int simple_unlink(struct inode *, struct dentry *);
2420 extern int simple_rmdir(struct inode *, struct dentry *);
2421 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2422 extern int noop_fsync(struct file *, int);
2423 extern int simple_empty(struct dentry *);
2424 extern int simple_readpage(struct file *file, struct page *page);
2425 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2426 			loff_t pos, unsigned len, unsigned flags,
2427 			struct page **pagep, void **fsdata);
2428 extern int simple_write_end(struct file *file, struct address_space *mapping,
2429 			loff_t pos, unsigned len, unsigned copied,
2430 			struct page *page, void *fsdata);
2431 
2432 extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
2433 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2434 extern const struct file_operations simple_dir_operations;
2435 extern const struct inode_operations simple_dir_inode_operations;
2436 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2437 struct dentry *d_alloc_name(struct dentry *, const char *);
2438 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2439 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2440 extern void simple_release_fs(struct vfsmount **mount, int *count);
2441 
2442 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2443 			loff_t *ppos, const void *from, size_t available);
2444 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2445 		const void __user *from, size_t count);
2446 
2447 extern int generic_file_fsync(struct file *, int);
2448 
2449 extern int generic_check_addressable(unsigned, u64);
2450 
2451 #ifdef CONFIG_MIGRATION
2452 extern int buffer_migrate_page(struct address_space *,
2453 				struct page *, struct page *);
2454 #else
2455 #define buffer_migrate_page NULL
2456 #endif
2457 
2458 extern int inode_change_ok(const struct inode *, struct iattr *);
2459 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2460 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2461 
2462 extern void file_update_time(struct file *file);
2463 
2464 extern int generic_show_options(struct seq_file *m, struct vfsmount *mnt);
2465 extern void save_mount_options(struct super_block *sb, char *options);
2466 extern void replace_mount_options(struct super_block *sb, char *options);
2467 
2468 static inline ino_t parent_ino(struct dentry *dentry)
2469 {
2470 	ino_t res;
2471 
2472 	/*
2473 	 * Don't strictly need d_lock here? If the parent ino could change
2474 	 * then surely we'd have a deeper race in the caller?
2475 	 */
2476 	spin_lock(&dentry->d_lock);
2477 	res = dentry->d_parent->d_inode->i_ino;
2478 	spin_unlock(&dentry->d_lock);
2479 	return res;
2480 }
2481 
2482 /* Transaction based IO helpers */
2483 
2484 /*
2485  * An argresp is stored in an allocated page and holds the
2486  * size of the argument or response, along with its content
2487  */
2488 struct simple_transaction_argresp {
2489 	ssize_t size;
2490 	char data[0];
2491 };
2492 
2493 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2494 
2495 char *simple_transaction_get(struct file *file, const char __user *buf,
2496 				size_t size);
2497 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2498 				size_t size, loff_t *pos);
2499 int simple_transaction_release(struct inode *inode, struct file *file);
2500 
2501 void simple_transaction_set(struct file *file, size_t n);
2502 
2503 /*
2504  * simple attribute files
2505  *
2506  * These attributes behave similar to those in sysfs:
2507  *
2508  * Writing to an attribute immediately sets a value, an open file can be
2509  * written to multiple times.
2510  *
2511  * Reading from an attribute creates a buffer from the value that might get
2512  * read with multiple read calls. When the attribute has been read
2513  * completely, no further read calls are possible until the file is opened
2514  * again.
2515  *
2516  * All attributes contain a text representation of a numeric value
2517  * that are accessed with the get() and set() functions.
2518  */
2519 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt)		\
2520 static int __fops ## _open(struct inode *inode, struct file *file)	\
2521 {									\
2522 	__simple_attr_check_format(__fmt, 0ull);			\
2523 	return simple_attr_open(inode, file, __get, __set, __fmt);	\
2524 }									\
2525 static const struct file_operations __fops = {				\
2526 	.owner	 = THIS_MODULE,						\
2527 	.open	 = __fops ## _open,					\
2528 	.release = simple_attr_release,					\
2529 	.read	 = simple_attr_read,					\
2530 	.write	 = simple_attr_write,					\
2531 	.llseek	 = generic_file_llseek,					\
2532 };
2533 
2534 static inline void __attribute__((format(printf, 1, 2)))
2535 __simple_attr_check_format(const char *fmt, ...)
2536 {
2537 	/* don't do anything, just let the compiler check the arguments; */
2538 }
2539 
2540 int simple_attr_open(struct inode *inode, struct file *file,
2541 		     int (*get)(void *, u64 *), int (*set)(void *, u64),
2542 		     const char *fmt);
2543 int simple_attr_release(struct inode *inode, struct file *file);
2544 ssize_t simple_attr_read(struct file *file, char __user *buf,
2545 			 size_t len, loff_t *ppos);
2546 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2547 			  size_t len, loff_t *ppos);
2548 
2549 struct ctl_table;
2550 int proc_nr_files(struct ctl_table *table, int write,
2551 		  void __user *buffer, size_t *lenp, loff_t *ppos);
2552 int proc_nr_dentry(struct ctl_table *table, int write,
2553 		  void __user *buffer, size_t *lenp, loff_t *ppos);
2554 int proc_nr_inodes(struct ctl_table *table, int write,
2555 		   void __user *buffer, size_t *lenp, loff_t *ppos);
2556 int __init get_filesystem_list(char *buf);
2557 
2558 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2559 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2560 					    (flag & FMODE_NONOTIFY)))
2561 
2562 #endif /* __KERNEL__ */
2563 #endif /* _LINUX_FS_H */
2564