xref: /linux-6.15/include/linux/fs.h (revision f7be3455)
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_CUR 1024	/* Initial setting for nfile rlimits */
27 #define INR_OPEN_MAX 4096	/* Hard limit for nfile rlimits */
28 
29 #define BLOCK_SIZE_BITS 10
30 #define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
31 
32 #define SEEK_SET	0	/* seek relative to beginning of file */
33 #define SEEK_CUR	1	/* seek relative to current file position */
34 #define SEEK_END	2	/* seek relative to end of file */
35 #define SEEK_DATA	3	/* seek to the next data */
36 #define SEEK_HOLE	4	/* seek to the next hole */
37 #define SEEK_MAX	SEEK_HOLE
38 
39 struct fstrim_range {
40 	__u64 start;
41 	__u64 len;
42 	__u64 minlen;
43 };
44 
45 /* And dynamically-tunable limits and defaults: */
46 struct files_stat_struct {
47 	unsigned long nr_files;		/* read only */
48 	unsigned long nr_free_files;	/* read only */
49 	unsigned long max_files;		/* tunable */
50 };
51 
52 struct inodes_stat_t {
53 	int nr_inodes;
54 	int nr_unused;
55 	int dummy[5];		/* padding for sysctl ABI compatibility */
56 };
57 
58 
59 #define NR_FILE  8192	/* this can well be larger on a larger system */
60 
61 #define MAY_EXEC		0x00000001
62 #define MAY_WRITE		0x00000002
63 #define MAY_READ		0x00000004
64 #define MAY_APPEND		0x00000008
65 #define MAY_ACCESS		0x00000010
66 #define MAY_OPEN		0x00000020
67 #define MAY_CHDIR		0x00000040
68 /* called from RCU mode, don't block */
69 #define MAY_NOT_BLOCK		0x00000080
70 
71 /*
72  * flags in file.f_mode.  Note that FMODE_READ and FMODE_WRITE must correspond
73  * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
74  */
75 
76 /* file is open for reading */
77 #define FMODE_READ		((__force fmode_t)0x1)
78 /* file is open for writing */
79 #define FMODE_WRITE		((__force fmode_t)0x2)
80 /* file is seekable */
81 #define FMODE_LSEEK		((__force fmode_t)0x4)
82 /* file can be accessed using pread */
83 #define FMODE_PREAD		((__force fmode_t)0x8)
84 /* file can be accessed using pwrite */
85 #define FMODE_PWRITE		((__force fmode_t)0x10)
86 /* File is opened for execution with sys_execve / sys_uselib */
87 #define FMODE_EXEC		((__force fmode_t)0x20)
88 /* File is opened with O_NDELAY (only set for block devices) */
89 #define FMODE_NDELAY		((__force fmode_t)0x40)
90 /* File is opened with O_EXCL (only set for block devices) */
91 #define FMODE_EXCL		((__force fmode_t)0x80)
92 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
93    (specialy hack for floppy.c) */
94 #define FMODE_WRITE_IOCTL	((__force fmode_t)0x100)
95 
96 /*
97  * Don't update ctime and mtime.
98  *
99  * Currently a special hack for the XFS open_by_handle ioctl, but we'll
100  * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
101  */
102 #define FMODE_NOCMTIME		((__force fmode_t)0x800)
103 
104 /* Expect random access pattern */
105 #define FMODE_RANDOM		((__force fmode_t)0x1000)
106 
107 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
108 #define FMODE_UNSIGNED_OFFSET	((__force fmode_t)0x2000)
109 
110 /* File is opened with O_PATH; almost nothing can be done with it */
111 #define FMODE_PATH		((__force fmode_t)0x4000)
112 
113 /* File was opened by fanotify and shouldn't generate fanotify events */
114 #define FMODE_NONOTIFY		((__force fmode_t)0x1000000)
115 
116 /*
117  * The below are the various read and write types that we support. Some of
118  * them include behavioral modifiers that send information down to the
119  * block layer and IO scheduler. Terminology:
120  *
121  *	The block layer uses device plugging to defer IO a little bit, in
122  *	the hope that we will see more IO very shortly. This increases
123  *	coalescing of adjacent IO and thus reduces the number of IOs we
124  *	have to send to the device. It also allows for better queuing,
125  *	if the IO isn't mergeable. If the caller is going to be waiting
126  *	for the IO, then he must ensure that the device is unplugged so
127  *	that the IO is dispatched to the driver.
128  *
129  *	All IO is handled async in Linux. This is fine for background
130  *	writes, but for reads or writes that someone waits for completion
131  *	on, we want to notify the block layer and IO scheduler so that they
132  *	know about it. That allows them to make better scheduling
133  *	decisions. So when the below references 'sync' and 'async', it
134  *	is referencing this priority hint.
135  *
136  * With that in mind, the available types are:
137  *
138  * READ			A normal read operation. Device will be plugged.
139  * READ_SYNC		A synchronous read. Device is not plugged, caller can
140  *			immediately wait on this read without caring about
141  *			unplugging.
142  * READA		Used for read-ahead operations. Lower priority, and the
143  *			block layer could (in theory) choose to ignore this
144  *			request if it runs into resource problems.
145  * WRITE		A normal async write. Device will be plugged.
146  * WRITE_SYNC		Synchronous write. Identical to WRITE, but passes down
147  *			the hint that someone will be waiting on this IO
148  *			shortly. The write equivalent of READ_SYNC.
149  * WRITE_ODIRECT	Special case write for O_DIRECT only.
150  * WRITE_FLUSH		Like WRITE_SYNC but with preceding cache flush.
151  * WRITE_FUA		Like WRITE_SYNC but data is guaranteed to be on
152  *			non-volatile media on completion.
153  * WRITE_FLUSH_FUA	Combination of WRITE_FLUSH and FUA. The IO is preceded
154  *			by a cache flush and data is guaranteed to be on
155  *			non-volatile media on completion.
156  *
157  */
158 #define RW_MASK			REQ_WRITE
159 #define RWA_MASK		REQ_RAHEAD
160 
161 #define READ			0
162 #define WRITE			RW_MASK
163 #define READA			RWA_MASK
164 
165 #define READ_SYNC		(READ | REQ_SYNC)
166 #define WRITE_SYNC		(WRITE | REQ_SYNC | REQ_NOIDLE)
167 #define WRITE_ODIRECT		(WRITE | REQ_SYNC)
168 #define WRITE_FLUSH		(WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
169 #define WRITE_FUA		(WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
170 #define WRITE_FLUSH_FUA		(WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
171 
172 #define SEL_IN		1
173 #define SEL_OUT		2
174 #define SEL_EX		4
175 
176 /* public flags for file_system_type */
177 #define FS_REQUIRES_DEV 1
178 #define FS_BINARY_MOUNTDATA 2
179 #define FS_HAS_SUBTYPE 4
180 #define FS_REVAL_DOT	16384	/* Check the paths ".", ".." for staleness */
181 #define FS_RENAME_DOES_D_MOVE	32768	/* FS will handle d_move()
182 					 * during rename() internally.
183 					 */
184 
185 /*
186  * These are the fs-independent mount-flags: up to 32 flags are supported
187  */
188 #define MS_RDONLY	 1	/* Mount read-only */
189 #define MS_NOSUID	 2	/* Ignore suid and sgid bits */
190 #define MS_NODEV	 4	/* Disallow access to device special files */
191 #define MS_NOEXEC	 8	/* Disallow program execution */
192 #define MS_SYNCHRONOUS	16	/* Writes are synced at once */
193 #define MS_REMOUNT	32	/* Alter flags of a mounted FS */
194 #define MS_MANDLOCK	64	/* Allow mandatory locks on an FS */
195 #define MS_DIRSYNC	128	/* Directory modifications are synchronous */
196 #define MS_NOATIME	1024	/* Do not update access times. */
197 #define MS_NODIRATIME	2048	/* Do not update directory access times */
198 #define MS_BIND		4096
199 #define MS_MOVE		8192
200 #define MS_REC		16384
201 #define MS_VERBOSE	32768	/* War is peace. Verbosity is silence.
202 				   MS_VERBOSE is deprecated. */
203 #define MS_SILENT	32768
204 #define MS_POSIXACL	(1<<16)	/* VFS does not apply the umask */
205 #define MS_UNBINDABLE	(1<<17)	/* change to unbindable */
206 #define MS_PRIVATE	(1<<18)	/* change to private */
207 #define MS_SLAVE	(1<<19)	/* change to slave */
208 #define MS_SHARED	(1<<20)	/* change to shared */
209 #define MS_RELATIME	(1<<21)	/* Update atime relative to mtime/ctime. */
210 #define MS_KERNMOUNT	(1<<22) /* this is a kern_mount call */
211 #define MS_I_VERSION	(1<<23) /* Update inode I_version field */
212 #define MS_STRICTATIME	(1<<24) /* Always perform atime updates */
213 #define MS_NOSEC	(1<<28)
214 #define MS_BORN		(1<<29)
215 #define MS_ACTIVE	(1<<30)
216 #define MS_NOUSER	(1<<31)
217 
218 /*
219  * Superblock flags that can be altered by MS_REMOUNT
220  */
221 #define MS_RMT_MASK	(MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_I_VERSION)
222 
223 /*
224  * Old magic mount flag and mask
225  */
226 #define MS_MGC_VAL 0xC0ED0000
227 #define MS_MGC_MSK 0xffff0000
228 
229 /* Inode flags - they have nothing to superblock flags now */
230 
231 #define S_SYNC		1	/* Writes are synced at once */
232 #define S_NOATIME	2	/* Do not update access times */
233 #define S_APPEND	4	/* Append-only file */
234 #define S_IMMUTABLE	8	/* Immutable file */
235 #define S_DEAD		16	/* removed, but still open directory */
236 #define S_NOQUOTA	32	/* Inode is not counted to quota */
237 #define S_DIRSYNC	64	/* Directory modifications are synchronous */
238 #define S_NOCMTIME	128	/* Do not update file c/mtime */
239 #define S_SWAPFILE	256	/* Do not truncate: swapon got its bmaps */
240 #define S_PRIVATE	512	/* Inode is fs-internal */
241 #define S_IMA		1024	/* Inode has an associated IMA struct */
242 #define S_AUTOMOUNT	2048	/* Automount/referral quasi-directory */
243 #define S_NOSEC		4096	/* no suid or xattr security attributes */
244 
245 /*
246  * Note that nosuid etc flags are inode-specific: setting some file-system
247  * flags just means all the inodes inherit those flags by default. It might be
248  * possible to override it selectively if you really wanted to with some
249  * ioctl() that is not currently implemented.
250  *
251  * Exception: MS_RDONLY is always applied to the entire file system.
252  *
253  * Unfortunately, it is possible to change a filesystems flags with it mounted
254  * with files in use.  This means that all of the inodes will not have their
255  * i_flags updated.  Hence, i_flags no longer inherit the superblock mount
256  * flags, so these have to be checked separately. -- [email protected]
257  */
258 #define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
259 
260 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
261 #define IS_SYNC(inode)		(__IS_FLG(inode, MS_SYNCHRONOUS) || \
262 					((inode)->i_flags & S_SYNC))
263 #define IS_DIRSYNC(inode)	(__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
264 					((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
265 #define IS_MANDLOCK(inode)	__IS_FLG(inode, MS_MANDLOCK)
266 #define IS_NOATIME(inode)   __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
267 #define IS_I_VERSION(inode)   __IS_FLG(inode, MS_I_VERSION)
268 
269 #define IS_NOQUOTA(inode)	((inode)->i_flags & S_NOQUOTA)
270 #define IS_APPEND(inode)	((inode)->i_flags & S_APPEND)
271 #define IS_IMMUTABLE(inode)	((inode)->i_flags & S_IMMUTABLE)
272 #define IS_POSIXACL(inode)	__IS_FLG(inode, MS_POSIXACL)
273 
274 #define IS_DEADDIR(inode)	((inode)->i_flags & S_DEAD)
275 #define IS_NOCMTIME(inode)	((inode)->i_flags & S_NOCMTIME)
276 #define IS_SWAPFILE(inode)	((inode)->i_flags & S_SWAPFILE)
277 #define IS_PRIVATE(inode)	((inode)->i_flags & S_PRIVATE)
278 #define IS_IMA(inode)		((inode)->i_flags & S_IMA)
279 #define IS_AUTOMOUNT(inode)	((inode)->i_flags & S_AUTOMOUNT)
280 #define IS_NOSEC(inode)		((inode)->i_flags & S_NOSEC)
281 
282 /* the read-only stuff doesn't really belong here, but any other place is
283    probably as bad and I don't want to create yet another include file. */
284 
285 #define BLKROSET   _IO(0x12,93)	/* set device read-only (0 = read-write) */
286 #define BLKROGET   _IO(0x12,94)	/* get read-only status (0 = read_write) */
287 #define BLKRRPART  _IO(0x12,95)	/* re-read partition table */
288 #define BLKGETSIZE _IO(0x12,96)	/* return device size /512 (long *arg) */
289 #define BLKFLSBUF  _IO(0x12,97)	/* flush buffer cache */
290 #define BLKRASET   _IO(0x12,98)	/* set read ahead for block device */
291 #define BLKRAGET   _IO(0x12,99)	/* get current read ahead setting */
292 #define BLKFRASET  _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
293 #define BLKFRAGET  _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
294 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
295 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
296 #define BLKSSZGET  _IO(0x12,104)/* get block device sector size */
297 #if 0
298 #define BLKPG      _IO(0x12,105)/* See blkpg.h */
299 
300 /* Some people are morons.  Do not use sizeof! */
301 
302 #define BLKELVGET  _IOR(0x12,106,size_t)/* elevator get */
303 #define BLKELVSET  _IOW(0x12,107,size_t)/* elevator set */
304 /* This was here just to show that the number is taken -
305    probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
306 #endif
307 /* A jump here: 108-111 have been used for various private purposes. */
308 #define BLKBSZGET  _IOR(0x12,112,size_t)
309 #define BLKBSZSET  _IOW(0x12,113,size_t)
310 #define BLKGETSIZE64 _IOR(0x12,114,size_t)	/* return device size in bytes (u64 *arg) */
311 #define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
312 #define BLKTRACESTART _IO(0x12,116)
313 #define BLKTRACESTOP _IO(0x12,117)
314 #define BLKTRACETEARDOWN _IO(0x12,118)
315 #define BLKDISCARD _IO(0x12,119)
316 #define BLKIOMIN _IO(0x12,120)
317 #define BLKIOOPT _IO(0x12,121)
318 #define BLKALIGNOFF _IO(0x12,122)
319 #define BLKPBSZGET _IO(0x12,123)
320 #define BLKDISCARDZEROES _IO(0x12,124)
321 #define BLKSECDISCARD _IO(0x12,125)
322 
323 #define BMAP_IOCTL 1		/* obsolete - kept for compatibility */
324 #define FIBMAP	   _IO(0x00,1)	/* bmap access */
325 #define FIGETBSZ   _IO(0x00,2)	/* get the block size used for bmap */
326 #define FIFREEZE	_IOWR('X', 119, int)	/* Freeze */
327 #define FITHAW		_IOWR('X', 120, int)	/* Thaw */
328 #define FITRIM		_IOWR('X', 121, struct fstrim_range)	/* Trim */
329 
330 #define	FS_IOC_GETFLAGS			_IOR('f', 1, long)
331 #define	FS_IOC_SETFLAGS			_IOW('f', 2, long)
332 #define	FS_IOC_GETVERSION		_IOR('v', 1, long)
333 #define	FS_IOC_SETVERSION		_IOW('v', 2, long)
334 #define FS_IOC_FIEMAP			_IOWR('f', 11, struct fiemap)
335 #define FS_IOC32_GETFLAGS		_IOR('f', 1, int)
336 #define FS_IOC32_SETFLAGS		_IOW('f', 2, int)
337 #define FS_IOC32_GETVERSION		_IOR('v', 1, int)
338 #define FS_IOC32_SETVERSION		_IOW('v', 2, int)
339 
340 /*
341  * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS)
342  */
343 #define	FS_SECRM_FL			0x00000001 /* Secure deletion */
344 #define	FS_UNRM_FL			0x00000002 /* Undelete */
345 #define	FS_COMPR_FL			0x00000004 /* Compress file */
346 #define FS_SYNC_FL			0x00000008 /* Synchronous updates */
347 #define FS_IMMUTABLE_FL			0x00000010 /* Immutable file */
348 #define FS_APPEND_FL			0x00000020 /* writes to file may only append */
349 #define FS_NODUMP_FL			0x00000040 /* do not dump file */
350 #define FS_NOATIME_FL			0x00000080 /* do not update atime */
351 /* Reserved for compression usage... */
352 #define FS_DIRTY_FL			0x00000100
353 #define FS_COMPRBLK_FL			0x00000200 /* One or more compressed clusters */
354 #define FS_NOCOMP_FL			0x00000400 /* Don't compress */
355 #define FS_ECOMPR_FL			0x00000800 /* Compression error */
356 /* End compression flags --- maybe not all used */
357 #define FS_BTREE_FL			0x00001000 /* btree format dir */
358 #define FS_INDEX_FL			0x00001000 /* hash-indexed directory */
359 #define FS_IMAGIC_FL			0x00002000 /* AFS directory */
360 #define FS_JOURNAL_DATA_FL		0x00004000 /* Reserved for ext3 */
361 #define FS_NOTAIL_FL			0x00008000 /* file tail should not be merged */
362 #define FS_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
363 #define FS_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
364 #define FS_EXTENT_FL			0x00080000 /* Extents */
365 #define FS_DIRECTIO_FL			0x00100000 /* Use direct i/o */
366 #define FS_NOCOW_FL			0x00800000 /* Do not cow file */
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/kdev_t.h>
382 #include <linux/dcache.h>
383 #include <linux/path.h>
384 #include <linux/stat.h>
385 #include <linux/cache.h>
386 #include <linux/list.h>
387 #include <linux/radix-tree.h>
388 #include <linux/prio_tree.h>
389 #include <linux/init.h>
390 #include <linux/pid.h>
391 #include <linux/mutex.h>
392 #include <linux/capability.h>
393 #include <linux/semaphore.h>
394 #include <linux/fiemap.h>
395 #include <linux/rculist_bl.h>
396 #include <linux/shrinker.h>
397 #include <linux/atomic.h>
398 
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 <[email protected]> 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 auxiliary 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 
587 	/* Write back some dirty pages from this mapping. */
588 	int (*writepages)(struct address_space *, struct writeback_control *);
589 
590 	/* Set a page dirty.  Return true if this dirtied it */
591 	int (*set_page_dirty)(struct page *page);
592 
593 	int (*readpages)(struct file *filp, struct address_space *mapping,
594 			struct list_head *pages, unsigned nr_pages);
595 
596 	int (*write_begin)(struct file *, struct address_space *mapping,
597 				loff_t pos, unsigned len, unsigned flags,
598 				struct page **pagep, void **fsdata);
599 	int (*write_end)(struct file *, struct address_space *mapping,
600 				loff_t pos, unsigned len, unsigned copied,
601 				struct page *page, void *fsdata);
602 
603 	/* Unfortunately this kludge is needed for FIBMAP. Don't use it */
604 	sector_t (*bmap)(struct address_space *, sector_t);
605 	void (*invalidatepage) (struct page *, unsigned long);
606 	int (*releasepage) (struct page *, gfp_t);
607 	void (*freepage)(struct page *);
608 	ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
609 			loff_t offset, unsigned long nr_segs);
610 	int (*get_xip_mem)(struct address_space *, pgoff_t, int,
611 						void **, unsigned long *);
612 	/* migrate the contents of a page to the specified target */
613 	int (*migratepage) (struct address_space *,
614 			struct page *, struct page *);
615 	int (*launder_page) (struct page *);
616 	int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
617 					unsigned long);
618 	int (*error_remove_page)(struct address_space *, struct page *);
619 };
620 
621 extern const struct address_space_operations empty_aops;
622 
623 /*
624  * pagecache_write_begin/pagecache_write_end must be used by general code
625  * to write into the pagecache.
626  */
627 int pagecache_write_begin(struct file *, struct address_space *mapping,
628 				loff_t pos, unsigned len, unsigned flags,
629 				struct page **pagep, void **fsdata);
630 
631 int pagecache_write_end(struct file *, struct address_space *mapping,
632 				loff_t pos, unsigned len, unsigned copied,
633 				struct page *page, void *fsdata);
634 
635 struct backing_dev_info;
636 struct address_space {
637 	struct inode		*host;		/* owner: inode, block_device */
638 	struct radix_tree_root	page_tree;	/* radix tree of all pages */
639 	spinlock_t		tree_lock;	/* and lock protecting it */
640 	unsigned int		i_mmap_writable;/* count VM_SHARED mappings */
641 	struct prio_tree_root	i_mmap;		/* tree of private and shared mappings */
642 	struct list_head	i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
643 	struct mutex		i_mmap_mutex;	/* protect tree, count, list */
644 	/* Protected by tree_lock together with the radix tree */
645 	unsigned long		nrpages;	/* number of total pages */
646 	pgoff_t			writeback_index;/* writeback starts here */
647 	const struct address_space_operations *a_ops;	/* methods */
648 	unsigned long		flags;		/* error bits/gfp mask */
649 	struct backing_dev_info *backing_dev_info; /* device readahead, etc */
650 	spinlock_t		private_lock;	/* for use by the address_space */
651 	struct list_head	private_list;	/* ditto */
652 	struct address_space	*assoc_mapping;	/* ditto */
653 } __attribute__((aligned(sizeof(long))));
654 	/*
655 	 * On most architectures that alignment is already the case; but
656 	 * must be enforced here for CRIS, to let the least significant bit
657 	 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
658 	 */
659 
660 struct block_device {
661 	dev_t			bd_dev;  /* not a kdev_t - it's a search key */
662 	int			bd_openers;
663 	struct inode *		bd_inode;	/* will die */
664 	struct super_block *	bd_super;
665 	struct mutex		bd_mutex;	/* open/close mutex */
666 	struct list_head	bd_inodes;
667 	void *			bd_claiming;
668 	void *			bd_holder;
669 	int			bd_holders;
670 	bool			bd_write_holder;
671 #ifdef CONFIG_SYSFS
672 	struct list_head	bd_holder_disks;
673 #endif
674 	struct block_device *	bd_contains;
675 	unsigned		bd_block_size;
676 	struct hd_struct *	bd_part;
677 	/* number of times partitions within this device have been opened. */
678 	unsigned		bd_part_count;
679 	int			bd_invalidated;
680 	struct gendisk *	bd_disk;
681 	struct list_head	bd_list;
682 	/*
683 	 * Private data.  You must have bd_claim'ed the block_device
684 	 * to use this.  NOTE:  bd_claim allows an owner to claim
685 	 * the same device multiple times, the owner must take special
686 	 * care to not mess up bd_private for that case.
687 	 */
688 	unsigned long		bd_private;
689 
690 	/* The counter of freeze processes */
691 	int			bd_fsfreeze_count;
692 	/* Mutex for freeze */
693 	struct mutex		bd_fsfreeze_mutex;
694 };
695 
696 /*
697  * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
698  * radix trees
699  */
700 #define PAGECACHE_TAG_DIRTY	0
701 #define PAGECACHE_TAG_WRITEBACK	1
702 #define PAGECACHE_TAG_TOWRITE	2
703 
704 int mapping_tagged(struct address_space *mapping, int tag);
705 
706 /*
707  * Might pages of this file be mapped into userspace?
708  */
709 static inline int mapping_mapped(struct address_space *mapping)
710 {
711 	return	!prio_tree_empty(&mapping->i_mmap) ||
712 		!list_empty(&mapping->i_mmap_nonlinear);
713 }
714 
715 /*
716  * Might pages of this file have been modified in userspace?
717  * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
718  * marks vma as VM_SHARED if it is shared, and the file was opened for
719  * writing i.e. vma may be mprotected writable even if now readonly.
720  */
721 static inline int mapping_writably_mapped(struct address_space *mapping)
722 {
723 	return mapping->i_mmap_writable != 0;
724 }
725 
726 /*
727  * Use sequence counter to get consistent i_size on 32-bit processors.
728  */
729 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
730 #include <linux/seqlock.h>
731 #define __NEED_I_SIZE_ORDERED
732 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
733 #else
734 #define i_size_ordered_init(inode) do { } while (0)
735 #endif
736 
737 struct posix_acl;
738 #define ACL_NOT_CACHED ((void *)(-1))
739 
740 #define IOP_FASTPERM	0x0001
741 #define IOP_LOOKUP	0x0002
742 #define IOP_NOFOLLOW	0x0004
743 
744 /*
745  * Keep mostly read-only and often accessed (especially for
746  * the RCU path lookup and 'stat' data) fields at the beginning
747  * of the 'struct inode'
748  */
749 struct inode {
750 	umode_t			i_mode;
751 	unsigned short		i_opflags;
752 	uid_t			i_uid;
753 	gid_t			i_gid;
754 	unsigned int		i_flags;
755 
756 #ifdef CONFIG_FS_POSIX_ACL
757 	struct posix_acl	*i_acl;
758 	struct posix_acl	*i_default_acl;
759 #endif
760 
761 	const struct inode_operations	*i_op;
762 	struct super_block	*i_sb;
763 	struct address_space	*i_mapping;
764 
765 #ifdef CONFIG_SECURITY
766 	void			*i_security;
767 #endif
768 
769 	/* Stat data, not accessed from path walking */
770 	unsigned long		i_ino;
771 	unsigned int		i_nlink;
772 	dev_t			i_rdev;
773 	loff_t			i_size;
774 	struct timespec		i_atime;
775 	struct timespec		i_mtime;
776 	struct timespec		i_ctime;
777 	unsigned int		i_blkbits;
778 	blkcnt_t		i_blocks;
779 
780 #ifdef __NEED_I_SIZE_ORDERED
781 	seqcount_t		i_size_seqcount;
782 #endif
783 
784 	/* Misc */
785 	unsigned long		i_state;
786 	spinlock_t		i_lock;	/* i_blocks, i_bytes, maybe i_size */
787 	struct mutex		i_mutex;
788 
789 	unsigned long		dirtied_when;	/* jiffies of first dirtying */
790 
791 	struct hlist_node	i_hash;
792 	struct list_head	i_wb_list;	/* backing dev IO list */
793 	struct list_head	i_lru;		/* inode LRU list */
794 	struct list_head	i_sb_list;
795 	union {
796 		struct list_head	i_dentry;
797 		struct rcu_head		i_rcu;
798 	};
799 	atomic_t		i_count;
800 	u64			i_version;
801 	unsigned short          i_bytes;
802 	atomic_t		i_dio_count;
803 	const struct file_operations	*i_fop;	/* former ->i_op->default_file_ops */
804 	struct file_lock	*i_flock;
805 	struct address_space	i_data;
806 #ifdef CONFIG_QUOTA
807 	struct dquot		*i_dquot[MAXQUOTAS];
808 #endif
809 	struct list_head	i_devices;
810 	union {
811 		struct pipe_inode_info	*i_pipe;
812 		struct block_device	*i_bdev;
813 		struct cdev		*i_cdev;
814 	};
815 
816 	__u32			i_generation;
817 
818 #ifdef CONFIG_FSNOTIFY
819 	__u32			i_fsnotify_mask; /* all events this inode cares about */
820 	struct hlist_head	i_fsnotify_marks;
821 #endif
822 
823 #ifdef CONFIG_IMA
824 	atomic_t		i_readcount; /* struct files open RO */
825 #endif
826 	atomic_t		i_writecount;
827 	void			*i_private; /* fs or device private pointer */
828 };
829 
830 static inline int inode_unhashed(struct inode *inode)
831 {
832 	return hlist_unhashed(&inode->i_hash);
833 }
834 
835 /*
836  * inode->i_mutex nesting subclasses for the lock validator:
837  *
838  * 0: the object of the current VFS operation
839  * 1: parent
840  * 2: child/target
841  * 3: quota file
842  *
843  * The locking order between these classes is
844  * parent -> child -> normal -> xattr -> quota
845  */
846 enum inode_i_mutex_lock_class
847 {
848 	I_MUTEX_NORMAL,
849 	I_MUTEX_PARENT,
850 	I_MUTEX_CHILD,
851 	I_MUTEX_XATTR,
852 	I_MUTEX_QUOTA
853 };
854 
855 /*
856  * NOTE: in a 32bit arch with a preemptable kernel and
857  * an UP compile the i_size_read/write must be atomic
858  * with respect to the local cpu (unlike with preempt disabled),
859  * but they don't need to be atomic with respect to other cpus like in
860  * true SMP (so they need either to either locally disable irq around
861  * the read or for example on x86 they can be still implemented as a
862  * cmpxchg8b without the need of the lock prefix). For SMP compiles
863  * and 64bit archs it makes no difference if preempt is enabled or not.
864  */
865 static inline loff_t i_size_read(const struct inode *inode)
866 {
867 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
868 	loff_t i_size;
869 	unsigned int seq;
870 
871 	do {
872 		seq = read_seqcount_begin(&inode->i_size_seqcount);
873 		i_size = inode->i_size;
874 	} while (read_seqcount_retry(&inode->i_size_seqcount, seq));
875 	return i_size;
876 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
877 	loff_t i_size;
878 
879 	preempt_disable();
880 	i_size = inode->i_size;
881 	preempt_enable();
882 	return i_size;
883 #else
884 	return inode->i_size;
885 #endif
886 }
887 
888 /*
889  * NOTE: unlike i_size_read(), i_size_write() does need locking around it
890  * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
891  * can be lost, resulting in subsequent i_size_read() calls spinning forever.
892  */
893 static inline void i_size_write(struct inode *inode, loff_t i_size)
894 {
895 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
896 	write_seqcount_begin(&inode->i_size_seqcount);
897 	inode->i_size = i_size;
898 	write_seqcount_end(&inode->i_size_seqcount);
899 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
900 	preempt_disable();
901 	inode->i_size = i_size;
902 	preempt_enable();
903 #else
904 	inode->i_size = i_size;
905 #endif
906 }
907 
908 static inline unsigned iminor(const struct inode *inode)
909 {
910 	return MINOR(inode->i_rdev);
911 }
912 
913 static inline unsigned imajor(const struct inode *inode)
914 {
915 	return MAJOR(inode->i_rdev);
916 }
917 
918 extern struct block_device *I_BDEV(struct inode *inode);
919 
920 struct fown_struct {
921 	rwlock_t lock;          /* protects pid, uid, euid fields */
922 	struct pid *pid;	/* pid or -pgrp where SIGIO should be sent */
923 	enum pid_type pid_type;	/* Kind of process group SIGIO should be sent to */
924 	uid_t uid, euid;	/* uid/euid of process setting the owner */
925 	int signum;		/* posix.1b rt signal to be delivered on IO */
926 };
927 
928 /*
929  * Track a single file's readahead state
930  */
931 struct file_ra_state {
932 	pgoff_t start;			/* where readahead started */
933 	unsigned int size;		/* # of readahead pages */
934 	unsigned int async_size;	/* do asynchronous readahead when
935 					   there are only # of pages ahead */
936 
937 	unsigned int ra_pages;		/* Maximum readahead window */
938 	unsigned int mmap_miss;		/* Cache miss stat for mmap accesses */
939 	loff_t prev_pos;		/* Cache last read() position */
940 };
941 
942 /*
943  * Check if @index falls in the readahead windows.
944  */
945 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
946 {
947 	return (index >= ra->start &&
948 		index <  ra->start + ra->size);
949 }
950 
951 #define FILE_MNT_WRITE_TAKEN	1
952 #define FILE_MNT_WRITE_RELEASED	2
953 
954 struct file {
955 	/*
956 	 * fu_list becomes invalid after file_free is called and queued via
957 	 * fu_rcuhead for RCU freeing
958 	 */
959 	union {
960 		struct list_head	fu_list;
961 		struct rcu_head 	fu_rcuhead;
962 	} f_u;
963 	struct path		f_path;
964 #define f_dentry	f_path.dentry
965 #define f_vfsmnt	f_path.mnt
966 	const struct file_operations	*f_op;
967 
968 	/*
969 	 * Protects f_ep_links, f_flags, f_pos vs i_size in lseek SEEK_CUR.
970 	 * Must not be taken from IRQ context.
971 	 */
972 	spinlock_t		f_lock;
973 #ifdef CONFIG_SMP
974 	int			f_sb_list_cpu;
975 #endif
976 	atomic_long_t		f_count;
977 	unsigned int 		f_flags;
978 	fmode_t			f_mode;
979 	loff_t			f_pos;
980 	struct fown_struct	f_owner;
981 	const struct cred	*f_cred;
982 	struct file_ra_state	f_ra;
983 
984 	u64			f_version;
985 #ifdef CONFIG_SECURITY
986 	void			*f_security;
987 #endif
988 	/* needed for tty driver, and maybe others */
989 	void			*private_data;
990 
991 #ifdef CONFIG_EPOLL
992 	/* Used by fs/eventpoll.c to link all the hooks to this file */
993 	struct list_head	f_ep_links;
994 #endif /* #ifdef CONFIG_EPOLL */
995 	struct address_space	*f_mapping;
996 #ifdef CONFIG_DEBUG_WRITECOUNT
997 	unsigned long f_mnt_write_state;
998 #endif
999 };
1000 
1001 struct file_handle {
1002 	__u32 handle_bytes;
1003 	int handle_type;
1004 	/* file identifier */
1005 	unsigned char f_handle[0];
1006 };
1007 
1008 #define get_file(x)	atomic_long_inc(&(x)->f_count)
1009 #define fput_atomic(x)	atomic_long_add_unless(&(x)->f_count, -1, 1)
1010 #define file_count(x)	atomic_long_read(&(x)->f_count)
1011 
1012 #ifdef CONFIG_DEBUG_WRITECOUNT
1013 static inline void file_take_write(struct file *f)
1014 {
1015 	WARN_ON(f->f_mnt_write_state != 0);
1016 	f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
1017 }
1018 static inline void file_release_write(struct file *f)
1019 {
1020 	f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
1021 }
1022 static inline void file_reset_write(struct file *f)
1023 {
1024 	f->f_mnt_write_state = 0;
1025 }
1026 static inline void file_check_state(struct file *f)
1027 {
1028 	/*
1029 	 * At this point, either both or neither of these bits
1030 	 * should be set.
1031 	 */
1032 	WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
1033 	WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
1034 }
1035 static inline int file_check_writeable(struct file *f)
1036 {
1037 	if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
1038 		return 0;
1039 	printk(KERN_WARNING "writeable file with no "
1040 			    "mnt_want_write()\n");
1041 	WARN_ON(1);
1042 	return -EINVAL;
1043 }
1044 #else /* !CONFIG_DEBUG_WRITECOUNT */
1045 static inline void file_take_write(struct file *filp) {}
1046 static inline void file_release_write(struct file *filp) {}
1047 static inline void file_reset_write(struct file *filp) {}
1048 static inline void file_check_state(struct file *filp) {}
1049 static inline int file_check_writeable(struct file *filp)
1050 {
1051 	return 0;
1052 }
1053 #endif /* CONFIG_DEBUG_WRITECOUNT */
1054 
1055 #define	MAX_NON_LFS	((1UL<<31) - 1)
1056 
1057 /* Page cache limit. The filesystems should put that into their s_maxbytes
1058    limits, otherwise bad things can happen in VM. */
1059 #if BITS_PER_LONG==32
1060 #define MAX_LFS_FILESIZE	(((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
1061 #elif BITS_PER_LONG==64
1062 #define MAX_LFS_FILESIZE 	0x7fffffffffffffffUL
1063 #endif
1064 
1065 #define FL_POSIX	1
1066 #define FL_FLOCK	2
1067 #define FL_ACCESS	8	/* not trying to lock, just looking */
1068 #define FL_EXISTS	16	/* when unlocking, test for existence */
1069 #define FL_LEASE	32	/* lease held on this file */
1070 #define FL_CLOSE	64	/* unlock on close */
1071 #define FL_SLEEP	128	/* A blocking lock */
1072 #define FL_DOWNGRADE_PENDING	256 /* Lease is being downgraded */
1073 #define FL_UNLOCK_PENDING	512 /* Lease is being broken */
1074 
1075 /*
1076  * Special return value from posix_lock_file() and vfs_lock_file() for
1077  * asynchronous locking.
1078  */
1079 #define FILE_LOCK_DEFERRED 1
1080 
1081 /*
1082  * The POSIX file lock owner is determined by
1083  * the "struct files_struct" in the thread group
1084  * (or NULL for no owner - BSD locks).
1085  *
1086  * Lockd stuffs a "host" pointer into this.
1087  */
1088 typedef struct files_struct *fl_owner_t;
1089 
1090 struct file_lock_operations {
1091 	void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1092 	void (*fl_release_private)(struct file_lock *);
1093 };
1094 
1095 struct lock_manager_operations {
1096 	int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
1097 	void (*lm_notify)(struct file_lock *);	/* unblock callback */
1098 	int (*lm_grant)(struct file_lock *, struct file_lock *, int);
1099 	void (*lm_release_private)(struct file_lock *);
1100 	void (*lm_break)(struct file_lock *);
1101 	int (*lm_change)(struct file_lock **, int);
1102 };
1103 
1104 struct lock_manager {
1105 	struct list_head list;
1106 };
1107 
1108 void locks_start_grace(struct lock_manager *);
1109 void locks_end_grace(struct lock_manager *);
1110 int locks_in_grace(void);
1111 
1112 /* that will die - we need it for nfs_lock_info */
1113 #include <linux/nfs_fs_i.h>
1114 
1115 struct file_lock {
1116 	struct file_lock *fl_next;	/* singly linked list for this inode  */
1117 	struct list_head fl_link;	/* doubly linked list of all locks */
1118 	struct list_head fl_block;	/* circular list of blocked processes */
1119 	fl_owner_t fl_owner;
1120 	unsigned int fl_flags;
1121 	unsigned char fl_type;
1122 	unsigned int fl_pid;
1123 	struct pid *fl_nspid;
1124 	wait_queue_head_t fl_wait;
1125 	struct file *fl_file;
1126 	loff_t fl_start;
1127 	loff_t fl_end;
1128 
1129 	struct fasync_struct *	fl_fasync; /* for lease break notifications */
1130 	/* for lease breaks: */
1131 	unsigned long fl_break_time;
1132 	unsigned long fl_downgrade_time;
1133 
1134 	const struct file_lock_operations *fl_ops;	/* Callbacks for filesystems */
1135 	const struct lock_manager_operations *fl_lmops;	/* Callbacks for lockmanagers */
1136 	union {
1137 		struct nfs_lock_info	nfs_fl;
1138 		struct nfs4_lock_info	nfs4_fl;
1139 		struct {
1140 			struct list_head link;	/* link in AFS vnode's pending_locks list */
1141 			int state;		/* state of grant or error if -ve */
1142 		} afs;
1143 	} fl_u;
1144 };
1145 
1146 /* The following constant reflects the upper bound of the file/locking space */
1147 #ifndef OFFSET_MAX
1148 #define INT_LIMIT(x)	(~((x)1 << (sizeof(x)*8 - 1)))
1149 #define OFFSET_MAX	INT_LIMIT(loff_t)
1150 #define OFFT_OFFSET_MAX	INT_LIMIT(off_t)
1151 #endif
1152 
1153 #include <linux/fcntl.h>
1154 
1155 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1156 
1157 #ifdef CONFIG_FILE_LOCKING
1158 extern int fcntl_getlk(struct file *, struct flock __user *);
1159 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1160 			struct flock __user *);
1161 
1162 #if BITS_PER_LONG == 32
1163 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
1164 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1165 			struct flock64 __user *);
1166 #endif
1167 
1168 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1169 extern int fcntl_getlease(struct file *filp);
1170 
1171 /* fs/locks.c */
1172 void locks_free_lock(struct file_lock *fl);
1173 extern void locks_init_lock(struct file_lock *);
1174 extern struct file_lock * locks_alloc_lock(void);
1175 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1176 extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
1177 extern void locks_remove_posix(struct file *, fl_owner_t);
1178 extern void locks_remove_flock(struct file *);
1179 extern void locks_release_private(struct file_lock *);
1180 extern void posix_test_lock(struct file *, struct file_lock *);
1181 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1182 extern int posix_lock_file_wait(struct file *, struct file_lock *);
1183 extern int posix_unblock_lock(struct file *, struct file_lock *);
1184 extern int vfs_test_lock(struct file *, struct file_lock *);
1185 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1186 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1187 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1188 extern int __break_lease(struct inode *inode, unsigned int flags);
1189 extern void lease_get_mtime(struct inode *, struct timespec *time);
1190 extern int generic_setlease(struct file *, long, struct file_lock **);
1191 extern int vfs_setlease(struct file *, long, struct file_lock **);
1192 extern int lease_modify(struct file_lock **, int);
1193 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1194 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1195 extern void lock_flocks(void);
1196 extern void unlock_flocks(void);
1197 #else /* !CONFIG_FILE_LOCKING */
1198 static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1199 {
1200 	return -EINVAL;
1201 }
1202 
1203 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1204 			      unsigned int cmd, struct flock __user *user)
1205 {
1206 	return -EACCES;
1207 }
1208 
1209 #if BITS_PER_LONG == 32
1210 static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1211 {
1212 	return -EINVAL;
1213 }
1214 
1215 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1216 				unsigned int cmd, struct flock64 __user *user)
1217 {
1218 	return -EACCES;
1219 }
1220 #endif
1221 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1222 {
1223 	return 0;
1224 }
1225 
1226 static inline int fcntl_getlease(struct file *filp)
1227 {
1228 	return 0;
1229 }
1230 
1231 static inline void locks_init_lock(struct file_lock *fl)
1232 {
1233 	return;
1234 }
1235 
1236 static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1237 {
1238 	return;
1239 }
1240 
1241 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1242 {
1243 	return;
1244 }
1245 
1246 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1247 {
1248 	return;
1249 }
1250 
1251 static inline void locks_remove_flock(struct file *filp)
1252 {
1253 	return;
1254 }
1255 
1256 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1257 {
1258 	return;
1259 }
1260 
1261 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1262 				  struct file_lock *conflock)
1263 {
1264 	return -ENOLCK;
1265 }
1266 
1267 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1268 {
1269 	return -ENOLCK;
1270 }
1271 
1272 static inline int posix_unblock_lock(struct file *filp,
1273 				     struct file_lock *waiter)
1274 {
1275 	return -ENOENT;
1276 }
1277 
1278 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1279 {
1280 	return 0;
1281 }
1282 
1283 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1284 				struct file_lock *fl, struct file_lock *conf)
1285 {
1286 	return -ENOLCK;
1287 }
1288 
1289 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1290 {
1291 	return 0;
1292 }
1293 
1294 static inline int flock_lock_file_wait(struct file *filp,
1295 				       struct file_lock *request)
1296 {
1297 	return -ENOLCK;
1298 }
1299 
1300 static inline int __break_lease(struct inode *inode, unsigned int mode)
1301 {
1302 	return 0;
1303 }
1304 
1305 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1306 {
1307 	return;
1308 }
1309 
1310 static inline int generic_setlease(struct file *filp, long arg,
1311 				    struct file_lock **flp)
1312 {
1313 	return -EINVAL;
1314 }
1315 
1316 static inline int vfs_setlease(struct file *filp, long arg,
1317 			       struct file_lock **lease)
1318 {
1319 	return -EINVAL;
1320 }
1321 
1322 static inline int lease_modify(struct file_lock **before, int arg)
1323 {
1324 	return -EINVAL;
1325 }
1326 
1327 static inline int lock_may_read(struct inode *inode, loff_t start,
1328 				unsigned long len)
1329 {
1330 	return 1;
1331 }
1332 
1333 static inline int lock_may_write(struct inode *inode, loff_t start,
1334 				 unsigned long len)
1335 {
1336 	return 1;
1337 }
1338 
1339 static inline void lock_flocks(void)
1340 {
1341 }
1342 
1343 static inline void unlock_flocks(void)
1344 {
1345 }
1346 
1347 #endif /* !CONFIG_FILE_LOCKING */
1348 
1349 
1350 struct fasync_struct {
1351 	spinlock_t		fa_lock;
1352 	int			magic;
1353 	int			fa_fd;
1354 	struct fasync_struct	*fa_next; /* singly linked list */
1355 	struct file		*fa_file;
1356 	struct rcu_head		fa_rcu;
1357 };
1358 
1359 #define FASYNC_MAGIC 0x4601
1360 
1361 /* SMP safe fasync helpers: */
1362 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1363 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1364 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1365 extern struct fasync_struct *fasync_alloc(void);
1366 extern void fasync_free(struct fasync_struct *);
1367 
1368 /* can be called from interrupts */
1369 extern void kill_fasync(struct fasync_struct **, int, int);
1370 
1371 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1372 extern int f_setown(struct file *filp, unsigned long arg, int force);
1373 extern void f_delown(struct file *filp);
1374 extern pid_t f_getown(struct file *filp);
1375 extern int send_sigurg(struct fown_struct *fown);
1376 
1377 /*
1378  *	Umount options
1379  */
1380 
1381 #define MNT_FORCE	0x00000001	/* Attempt to forcibily umount */
1382 #define MNT_DETACH	0x00000002	/* Just detach from the tree */
1383 #define MNT_EXPIRE	0x00000004	/* Mark for expiry */
1384 #define UMOUNT_NOFOLLOW	0x00000008	/* Don't follow symlink on umount */
1385 #define UMOUNT_UNUSED	0x80000000	/* Flag guaranteed to be unused */
1386 
1387 extern struct list_head super_blocks;
1388 extern spinlock_t sb_lock;
1389 
1390 struct super_block {
1391 	struct list_head	s_list;		/* Keep this first */
1392 	dev_t			s_dev;		/* search index; _not_ kdev_t */
1393 	unsigned char		s_dirt;
1394 	unsigned char		s_blocksize_bits;
1395 	unsigned long		s_blocksize;
1396 	loff_t			s_maxbytes;	/* Max file size */
1397 	struct file_system_type	*s_type;
1398 	const struct super_operations	*s_op;
1399 	const struct dquot_operations	*dq_op;
1400 	const struct quotactl_ops	*s_qcop;
1401 	const struct export_operations *s_export_op;
1402 	unsigned long		s_flags;
1403 	unsigned long		s_magic;
1404 	struct dentry		*s_root;
1405 	struct rw_semaphore	s_umount;
1406 	struct mutex		s_lock;
1407 	int			s_count;
1408 	atomic_t		s_active;
1409 #ifdef CONFIG_SECURITY
1410 	void                    *s_security;
1411 #endif
1412 	const struct xattr_handler **s_xattr;
1413 
1414 	struct list_head	s_inodes;	/* all inodes */
1415 	struct hlist_bl_head	s_anon;		/* anonymous dentries for (nfs) exporting */
1416 #ifdef CONFIG_SMP
1417 	struct list_head __percpu *s_files;
1418 #else
1419 	struct list_head	s_files;
1420 #endif
1421 	/* s_dentry_lru, s_nr_dentry_unused protected by dcache.c lru locks */
1422 	struct list_head	s_dentry_lru;	/* unused dentry lru */
1423 	int			s_nr_dentry_unused;	/* # of dentry on lru */
1424 
1425 	/* s_inode_lru_lock protects s_inode_lru and s_nr_inodes_unused */
1426 	spinlock_t		s_inode_lru_lock ____cacheline_aligned_in_smp;
1427 	struct list_head	s_inode_lru;		/* unused inode lru */
1428 	int			s_nr_inodes_unused;	/* # of inodes on lru */
1429 
1430 	struct block_device	*s_bdev;
1431 	struct backing_dev_info *s_bdi;
1432 	struct mtd_info		*s_mtd;
1433 	struct list_head	s_instances;
1434 	struct quota_info	s_dquot;	/* Diskquota specific options */
1435 
1436 	int			s_frozen;
1437 	wait_queue_head_t	s_wait_unfrozen;
1438 
1439 	char s_id[32];				/* Informational name */
1440 	u8 s_uuid[16];				/* UUID */
1441 
1442 	void 			*s_fs_info;	/* Filesystem private info */
1443 	fmode_t			s_mode;
1444 
1445 	/* Granularity of c/m/atime in ns.
1446 	   Cannot be worse than a second */
1447 	u32		   s_time_gran;
1448 
1449 	/*
1450 	 * The next field is for VFS *only*. No filesystems have any business
1451 	 * even looking at it. You had been warned.
1452 	 */
1453 	struct mutex s_vfs_rename_mutex;	/* Kludge */
1454 
1455 	/*
1456 	 * Filesystem subtype.  If non-empty the filesystem type field
1457 	 * in /proc/mounts will be "type.subtype"
1458 	 */
1459 	char *s_subtype;
1460 
1461 	/*
1462 	 * Saved mount options for lazy filesystems using
1463 	 * generic_show_options()
1464 	 */
1465 	char __rcu *s_options;
1466 	const struct dentry_operations *s_d_op; /* default d_op for dentries */
1467 
1468 	/*
1469 	 * Saved pool identifier for cleancache (-1 means none)
1470 	 */
1471 	int cleancache_poolid;
1472 
1473 	struct shrinker s_shrink;	/* per-sb shrinker handle */
1474 };
1475 
1476 /* superblock cache pruning functions */
1477 extern void prune_icache_sb(struct super_block *sb, int nr_to_scan);
1478 extern void prune_dcache_sb(struct super_block *sb, int nr_to_scan);
1479 
1480 extern struct timespec current_fs_time(struct super_block *sb);
1481 
1482 /*
1483  * Snapshotting support.
1484  */
1485 enum {
1486 	SB_UNFROZEN = 0,
1487 	SB_FREEZE_WRITE	= 1,
1488 	SB_FREEZE_TRANS = 2,
1489 };
1490 
1491 #define vfs_check_frozen(sb, level) \
1492 	wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
1493 
1494 /*
1495  * until VFS tracks user namespaces for inodes, just make all files
1496  * belong to init_user_ns
1497  */
1498 extern struct user_namespace init_user_ns;
1499 #define inode_userns(inode) (&init_user_ns)
1500 extern bool inode_owner_or_capable(const struct inode *inode);
1501 
1502 /* not quite ready to be deprecated, but... */
1503 extern void lock_super(struct super_block *);
1504 extern void unlock_super(struct super_block *);
1505 
1506 /*
1507  * VFS helper functions..
1508  */
1509 extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
1510 extern int vfs_mkdir(struct inode *, struct dentry *, int);
1511 extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
1512 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1513 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1514 extern int vfs_rmdir(struct inode *, struct dentry *);
1515 extern int vfs_unlink(struct inode *, struct dentry *);
1516 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1517 
1518 /*
1519  * VFS dentry helper functions.
1520  */
1521 extern void dentry_unhash(struct dentry *dentry);
1522 
1523 /*
1524  * VFS file helper functions.
1525  */
1526 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1527 			mode_t mode);
1528 /*
1529  * VFS FS_IOC_FIEMAP helper definitions.
1530  */
1531 struct fiemap_extent_info {
1532 	unsigned int fi_flags;		/* Flags as passed from user */
1533 	unsigned int fi_extents_mapped;	/* Number of mapped extents */
1534 	unsigned int fi_extents_max;	/* Size of fiemap_extent array */
1535 	struct fiemap_extent __user *fi_extents_start; /* Start of
1536 							fiemap_extent array */
1537 };
1538 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1539 			    u64 phys, u64 len, u32 flags);
1540 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1541 
1542 /*
1543  * File types
1544  *
1545  * NOTE! These match bits 12..15 of stat.st_mode
1546  * (ie "(i_mode >> 12) & 15").
1547  */
1548 #define DT_UNKNOWN	0
1549 #define DT_FIFO		1
1550 #define DT_CHR		2
1551 #define DT_DIR		4
1552 #define DT_BLK		6
1553 #define DT_REG		8
1554 #define DT_LNK		10
1555 #define DT_SOCK		12
1556 #define DT_WHT		14
1557 
1558 /*
1559  * This is the "filldir" function type, used by readdir() to let
1560  * the kernel specify what kind of dirent layout it wants to have.
1561  * This allows the kernel to read directories into kernel space or
1562  * to have different dirent layouts depending on the binary type.
1563  */
1564 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1565 struct block_device_operations;
1566 
1567 /* These macros are for out of kernel modules to test that
1568  * the kernel supports the unlocked_ioctl and compat_ioctl
1569  * fields in struct file_operations. */
1570 #define HAVE_COMPAT_IOCTL 1
1571 #define HAVE_UNLOCKED_IOCTL 1
1572 
1573 struct file_operations {
1574 	struct module *owner;
1575 	loff_t (*llseek) (struct file *, loff_t, int);
1576 	ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1577 	ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1578 	ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1579 	ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1580 	int (*readdir) (struct file *, void *, filldir_t);
1581 	unsigned int (*poll) (struct file *, struct poll_table_struct *);
1582 	long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1583 	long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1584 	int (*mmap) (struct file *, struct vm_area_struct *);
1585 	int (*open) (struct inode *, struct file *);
1586 	int (*flush) (struct file *, fl_owner_t id);
1587 	int (*release) (struct inode *, struct file *);
1588 	int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1589 	int (*aio_fsync) (struct kiocb *, int datasync);
1590 	int (*fasync) (int, struct file *, int);
1591 	int (*lock) (struct file *, int, struct file_lock *);
1592 	ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1593 	unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1594 	int (*check_flags)(int);
1595 	int (*flock) (struct file *, int, struct file_lock *);
1596 	ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1597 	ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1598 	int (*setlease)(struct file *, long, struct file_lock **);
1599 	long (*fallocate)(struct file *file, int mode, loff_t offset,
1600 			  loff_t len);
1601 };
1602 
1603 struct inode_operations {
1604 	struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1605 	void * (*follow_link) (struct dentry *, struct nameidata *);
1606 	int (*permission) (struct inode *, int);
1607 	struct posix_acl * (*get_acl)(struct inode *, int);
1608 
1609 	int (*readlink) (struct dentry *, char __user *,int);
1610 	void (*put_link) (struct dentry *, struct nameidata *, void *);
1611 
1612 	int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1613 	int (*link) (struct dentry *,struct inode *,struct dentry *);
1614 	int (*unlink) (struct inode *,struct dentry *);
1615 	int (*symlink) (struct inode *,struct dentry *,const char *);
1616 	int (*mkdir) (struct inode *,struct dentry *,int);
1617 	int (*rmdir) (struct inode *,struct dentry *);
1618 	int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1619 	int (*rename) (struct inode *, struct dentry *,
1620 			struct inode *, struct dentry *);
1621 	void (*truncate) (struct inode *);
1622 	int (*setattr) (struct dentry *, struct iattr *);
1623 	int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1624 	int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1625 	ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1626 	ssize_t (*listxattr) (struct dentry *, char *, size_t);
1627 	int (*removexattr) (struct dentry *, const char *);
1628 	void (*truncate_range)(struct inode *, loff_t, loff_t);
1629 	int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1630 		      u64 len);
1631 } ____cacheline_aligned;
1632 
1633 struct seq_file;
1634 
1635 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1636 				unsigned long nr_segs, unsigned long fast_segs,
1637 				struct iovec *fast_pointer,
1638 				struct iovec **ret_pointer);
1639 
1640 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1641 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1642 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1643 		unsigned long, loff_t *);
1644 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1645 		unsigned long, loff_t *);
1646 
1647 struct super_operations {
1648    	struct inode *(*alloc_inode)(struct super_block *sb);
1649 	void (*destroy_inode)(struct inode *);
1650 
1651    	void (*dirty_inode) (struct inode *, int flags);
1652 	int (*write_inode) (struct inode *, struct writeback_control *wbc);
1653 	int (*drop_inode) (struct inode *);
1654 	void (*evict_inode) (struct inode *);
1655 	void (*put_super) (struct super_block *);
1656 	void (*write_super) (struct super_block *);
1657 	int (*sync_fs)(struct super_block *sb, int wait);
1658 	int (*freeze_fs) (struct super_block *);
1659 	int (*unfreeze_fs) (struct super_block *);
1660 	int (*statfs) (struct dentry *, struct kstatfs *);
1661 	int (*remount_fs) (struct super_block *, int *, char *);
1662 	void (*umount_begin) (struct super_block *);
1663 
1664 	int (*show_options)(struct seq_file *, struct vfsmount *);
1665 	int (*show_devname)(struct seq_file *, struct vfsmount *);
1666 	int (*show_path)(struct seq_file *, struct vfsmount *);
1667 	int (*show_stats)(struct seq_file *, struct vfsmount *);
1668 #ifdef CONFIG_QUOTA
1669 	ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1670 	ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1671 #endif
1672 	int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1673 	int (*nr_cached_objects)(struct super_block *);
1674 	void (*free_cached_objects)(struct super_block *, int);
1675 };
1676 
1677 /*
1678  * Inode state bits.  Protected by inode->i_lock
1679  *
1680  * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1681  * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1682  *
1683  * Four bits define the lifetime of an inode.  Initially, inodes are I_NEW,
1684  * until that flag is cleared.  I_WILL_FREE, I_FREEING and I_CLEAR are set at
1685  * various stages of removing an inode.
1686  *
1687  * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1688  *
1689  * I_DIRTY_SYNC		Inode is dirty, but doesn't have to be written on
1690  *			fdatasync().  i_atime is the usual cause.
1691  * I_DIRTY_DATASYNC	Data-related inode changes pending. We keep track of
1692  *			these changes separately from I_DIRTY_SYNC so that we
1693  *			don't have to write inode on fdatasync() when only
1694  *			mtime has changed in it.
1695  * I_DIRTY_PAGES	Inode has dirty pages.  Inode itself may be clean.
1696  * I_NEW		Serves as both a mutex and completion notification.
1697  *			New inodes set I_NEW.  If two processes both create
1698  *			the same inode, one of them will release its inode and
1699  *			wait for I_NEW to be released before returning.
1700  *			Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1701  *			also cause waiting on I_NEW, without I_NEW actually
1702  *			being set.  find_inode() uses this to prevent returning
1703  *			nearly-dead inodes.
1704  * I_WILL_FREE		Must be set when calling write_inode_now() if i_count
1705  *			is zero.  I_FREEING must be set when I_WILL_FREE is
1706  *			cleared.
1707  * I_FREEING		Set when inode is about to be freed but still has dirty
1708  *			pages or buffers attached or the inode itself is still
1709  *			dirty.
1710  * I_CLEAR		Added by end_writeback().  In this state the inode is clean
1711  *			and can be destroyed.  Inode keeps I_FREEING.
1712  *
1713  *			Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1714  *			prohibited for many purposes.  iget() must wait for
1715  *			the inode to be completely released, then create it
1716  *			anew.  Other functions will just ignore such inodes,
1717  *			if appropriate.  I_NEW is used for waiting.
1718  *
1719  * I_SYNC		Synchonized write of dirty inode data.  The bits is
1720  *			set during data writeback, and cleared with a wakeup
1721  *			on the bit address once it is done.
1722  *
1723  * I_REFERENCED		Marks the inode as recently references on the LRU list.
1724  *
1725  * I_DIO_WAKEUP		Never set.  Only used as a key for wait_on_bit().
1726  *
1727  * Q: What is the difference between I_WILL_FREE and I_FREEING?
1728  */
1729 #define I_DIRTY_SYNC		(1 << 0)
1730 #define I_DIRTY_DATASYNC	(1 << 1)
1731 #define I_DIRTY_PAGES		(1 << 2)
1732 #define __I_NEW			3
1733 #define I_NEW			(1 << __I_NEW)
1734 #define I_WILL_FREE		(1 << 4)
1735 #define I_FREEING		(1 << 5)
1736 #define I_CLEAR			(1 << 6)
1737 #define __I_SYNC		7
1738 #define I_SYNC			(1 << __I_SYNC)
1739 #define I_REFERENCED		(1 << 8)
1740 #define __I_DIO_WAKEUP		9
1741 #define I_DIO_WAKEUP		(1 << I_DIO_WAKEUP)
1742 
1743 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1744 
1745 extern void __mark_inode_dirty(struct inode *, int);
1746 static inline void mark_inode_dirty(struct inode *inode)
1747 {
1748 	__mark_inode_dirty(inode, I_DIRTY);
1749 }
1750 
1751 static inline void mark_inode_dirty_sync(struct inode *inode)
1752 {
1753 	__mark_inode_dirty(inode, I_DIRTY_SYNC);
1754 }
1755 
1756 /**
1757  * inc_nlink - directly increment an inode's link count
1758  * @inode: inode
1759  *
1760  * This is a low-level filesystem helper to replace any
1761  * direct filesystem manipulation of i_nlink.  Currently,
1762  * it is only here for parity with dec_nlink().
1763  */
1764 static inline void inc_nlink(struct inode *inode)
1765 {
1766 	inode->i_nlink++;
1767 }
1768 
1769 static inline void inode_inc_link_count(struct inode *inode)
1770 {
1771 	inc_nlink(inode);
1772 	mark_inode_dirty(inode);
1773 }
1774 
1775 /**
1776  * drop_nlink - directly drop an inode's link count
1777  * @inode: inode
1778  *
1779  * This is a low-level filesystem helper to replace any
1780  * direct filesystem manipulation of i_nlink.  In cases
1781  * where we are attempting to track writes to the
1782  * filesystem, a decrement to zero means an imminent
1783  * write when the file is truncated and actually unlinked
1784  * on the filesystem.
1785  */
1786 static inline void drop_nlink(struct inode *inode)
1787 {
1788 	inode->i_nlink--;
1789 }
1790 
1791 /**
1792  * clear_nlink - directly zero an inode's link count
1793  * @inode: inode
1794  *
1795  * This is a low-level filesystem helper to replace any
1796  * direct filesystem manipulation of i_nlink.  See
1797  * drop_nlink() for why we care about i_nlink hitting zero.
1798  */
1799 static inline void clear_nlink(struct inode *inode)
1800 {
1801 	inode->i_nlink = 0;
1802 }
1803 
1804 static inline void inode_dec_link_count(struct inode *inode)
1805 {
1806 	drop_nlink(inode);
1807 	mark_inode_dirty(inode);
1808 }
1809 
1810 /**
1811  * inode_inc_iversion - increments i_version
1812  * @inode: inode that need to be updated
1813  *
1814  * Every time the inode is modified, the i_version field will be incremented.
1815  * The filesystem has to be mounted with i_version flag
1816  */
1817 
1818 static inline void inode_inc_iversion(struct inode *inode)
1819 {
1820        spin_lock(&inode->i_lock);
1821        inode->i_version++;
1822        spin_unlock(&inode->i_lock);
1823 }
1824 
1825 extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
1826 static inline void file_accessed(struct file *file)
1827 {
1828 	if (!(file->f_flags & O_NOATIME))
1829 		touch_atime(file->f_path.mnt, file->f_path.dentry);
1830 }
1831 
1832 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1833 int sync_inode_metadata(struct inode *inode, int wait);
1834 
1835 struct file_system_type {
1836 	const char *name;
1837 	int fs_flags;
1838 	struct dentry *(*mount) (struct file_system_type *, int,
1839 		       const char *, void *);
1840 	void (*kill_sb) (struct super_block *);
1841 	struct module *owner;
1842 	struct file_system_type * next;
1843 	struct list_head fs_supers;
1844 
1845 	struct lock_class_key s_lock_key;
1846 	struct lock_class_key s_umount_key;
1847 	struct lock_class_key s_vfs_rename_key;
1848 
1849 	struct lock_class_key i_lock_key;
1850 	struct lock_class_key i_mutex_key;
1851 	struct lock_class_key i_mutex_dir_key;
1852 };
1853 
1854 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1855 	void *data, int (*fill_super)(struct super_block *, void *, int));
1856 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1857 	int flags, const char *dev_name, void *data,
1858 	int (*fill_super)(struct super_block *, void *, int));
1859 extern struct dentry *mount_single(struct file_system_type *fs_type,
1860 	int flags, void *data,
1861 	int (*fill_super)(struct super_block *, void *, int));
1862 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1863 	int flags, void *data,
1864 	int (*fill_super)(struct super_block *, void *, int));
1865 void generic_shutdown_super(struct super_block *sb);
1866 void kill_block_super(struct super_block *sb);
1867 void kill_anon_super(struct super_block *sb);
1868 void kill_litter_super(struct super_block *sb);
1869 void deactivate_super(struct super_block *sb);
1870 void deactivate_locked_super(struct super_block *sb);
1871 int set_anon_super(struct super_block *s, void *data);
1872 int get_anon_bdev(dev_t *);
1873 void free_anon_bdev(dev_t);
1874 struct super_block *sget(struct file_system_type *type,
1875 			int (*test)(struct super_block *,void *),
1876 			int (*set)(struct super_block *,void *),
1877 			void *data);
1878 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1879 	const struct super_operations *ops,
1880 	const struct dentry_operations *dops,
1881 	unsigned long);
1882 
1883 static inline void sb_mark_dirty(struct super_block *sb)
1884 {
1885 	sb->s_dirt = 1;
1886 }
1887 static inline void sb_mark_clean(struct super_block *sb)
1888 {
1889 	sb->s_dirt = 0;
1890 }
1891 static inline int sb_is_dirty(struct super_block *sb)
1892 {
1893 	return sb->s_dirt;
1894 }
1895 
1896 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1897 #define fops_get(fops) \
1898 	(((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1899 #define fops_put(fops) \
1900 	do { if (fops) module_put((fops)->owner); } while(0)
1901 
1902 extern int register_filesystem(struct file_system_type *);
1903 extern int unregister_filesystem(struct file_system_type *);
1904 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1905 #define kern_mount(type) kern_mount_data(type, NULL)
1906 extern void kern_unmount(struct vfsmount *mnt);
1907 extern int may_umount_tree(struct vfsmount *);
1908 extern int may_umount(struct vfsmount *);
1909 extern long do_mount(char *, char *, char *, unsigned long, void *);
1910 extern struct vfsmount *collect_mounts(struct path *);
1911 extern void drop_collected_mounts(struct vfsmount *);
1912 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1913 			  struct vfsmount *);
1914 extern int vfs_statfs(struct path *, struct kstatfs *);
1915 extern int user_statfs(const char __user *, struct kstatfs *);
1916 extern int fd_statfs(int, struct kstatfs *);
1917 extern int statfs_by_dentry(struct dentry *, struct kstatfs *);
1918 extern int freeze_super(struct super_block *super);
1919 extern int thaw_super(struct super_block *super);
1920 
1921 extern int current_umask(void);
1922 
1923 /* /sys/fs */
1924 extern struct kobject *fs_kobj;
1925 
1926 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1927 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1928 
1929 #define FLOCK_VERIFY_READ  1
1930 #define FLOCK_VERIFY_WRITE 2
1931 
1932 #ifdef CONFIG_FILE_LOCKING
1933 extern int locks_mandatory_locked(struct inode *);
1934 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1935 
1936 /*
1937  * Candidates for mandatory locking have the setgid bit set
1938  * but no group execute bit -  an otherwise meaningless combination.
1939  */
1940 
1941 static inline int __mandatory_lock(struct inode *ino)
1942 {
1943 	return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1944 }
1945 
1946 /*
1947  * ... and these candidates should be on MS_MANDLOCK mounted fs,
1948  * otherwise these will be advisory locks
1949  */
1950 
1951 static inline int mandatory_lock(struct inode *ino)
1952 {
1953 	return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1954 }
1955 
1956 static inline int locks_verify_locked(struct inode *inode)
1957 {
1958 	if (mandatory_lock(inode))
1959 		return locks_mandatory_locked(inode);
1960 	return 0;
1961 }
1962 
1963 static inline int locks_verify_truncate(struct inode *inode,
1964 				    struct file *filp,
1965 				    loff_t size)
1966 {
1967 	if (inode->i_flock && mandatory_lock(inode))
1968 		return locks_mandatory_area(
1969 			FLOCK_VERIFY_WRITE, inode, filp,
1970 			size < inode->i_size ? size : inode->i_size,
1971 			(size < inode->i_size ? inode->i_size - size
1972 			 : size - inode->i_size)
1973 		);
1974 	return 0;
1975 }
1976 
1977 static inline int break_lease(struct inode *inode, unsigned int mode)
1978 {
1979 	if (inode->i_flock)
1980 		return __break_lease(inode, mode);
1981 	return 0;
1982 }
1983 #else /* !CONFIG_FILE_LOCKING */
1984 static inline int locks_mandatory_locked(struct inode *inode)
1985 {
1986 	return 0;
1987 }
1988 
1989 static inline int locks_mandatory_area(int rw, struct inode *inode,
1990 				       struct file *filp, loff_t offset,
1991 				       size_t count)
1992 {
1993 	return 0;
1994 }
1995 
1996 static inline int __mandatory_lock(struct inode *inode)
1997 {
1998 	return 0;
1999 }
2000 
2001 static inline int mandatory_lock(struct inode *inode)
2002 {
2003 	return 0;
2004 }
2005 
2006 static inline int locks_verify_locked(struct inode *inode)
2007 {
2008 	return 0;
2009 }
2010 
2011 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2012 					size_t size)
2013 {
2014 	return 0;
2015 }
2016 
2017 static inline int break_lease(struct inode *inode, unsigned int mode)
2018 {
2019 	return 0;
2020 }
2021 
2022 #endif /* CONFIG_FILE_LOCKING */
2023 
2024 /* fs/open.c */
2025 
2026 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2027 		       struct file *filp);
2028 extern int do_fallocate(struct file *file, int mode, loff_t offset,
2029 			loff_t len);
2030 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2031 			int mode);
2032 extern struct file *filp_open(const char *, int, int);
2033 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2034 				   const char *, int);
2035 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int,
2036 				 const struct cred *);
2037 extern int filp_close(struct file *, fl_owner_t id);
2038 extern char * getname(const char __user *);
2039 
2040 /* fs/ioctl.c */
2041 
2042 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2043 
2044 /* fs/dcache.c */
2045 extern void __init vfs_caches_init_early(void);
2046 extern void __init vfs_caches_init(unsigned long);
2047 
2048 extern struct kmem_cache *names_cachep;
2049 
2050 #define __getname_gfp(gfp)	kmem_cache_alloc(names_cachep, (gfp))
2051 #define __getname()		__getname_gfp(GFP_KERNEL)
2052 #define __putname(name)		kmem_cache_free(names_cachep, (void *)(name))
2053 #ifndef CONFIG_AUDITSYSCALL
2054 #define putname(name)   __putname(name)
2055 #else
2056 extern void putname(const char *name);
2057 #endif
2058 
2059 #ifdef CONFIG_BLOCK
2060 extern int register_blkdev(unsigned int, const char *);
2061 extern void unregister_blkdev(unsigned int, const char *);
2062 extern struct block_device *bdget(dev_t);
2063 extern struct block_device *bdgrab(struct block_device *bdev);
2064 extern void bd_set_size(struct block_device *, loff_t size);
2065 extern void bd_forget(struct inode *inode);
2066 extern void bdput(struct block_device *);
2067 extern void invalidate_bdev(struct block_device *);
2068 extern int sync_blockdev(struct block_device *bdev);
2069 extern struct super_block *freeze_bdev(struct block_device *);
2070 extern void emergency_thaw_all(void);
2071 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2072 extern int fsync_bdev(struct block_device *);
2073 #else
2074 static inline void bd_forget(struct inode *inode) {}
2075 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2076 static inline void invalidate_bdev(struct block_device *bdev) {}
2077 
2078 static inline struct super_block *freeze_bdev(struct block_device *sb)
2079 {
2080 	return NULL;
2081 }
2082 
2083 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2084 {
2085 	return 0;
2086 }
2087 #endif
2088 extern int sync_filesystem(struct super_block *);
2089 extern const struct file_operations def_blk_fops;
2090 extern const struct file_operations def_chr_fops;
2091 extern const struct file_operations bad_sock_fops;
2092 extern const struct file_operations def_fifo_fops;
2093 #ifdef CONFIG_BLOCK
2094 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2095 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2096 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2097 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2098 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2099 					       void *holder);
2100 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2101 					      void *holder);
2102 extern int blkdev_put(struct block_device *bdev, fmode_t mode);
2103 #ifdef CONFIG_SYSFS
2104 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2105 extern void bd_unlink_disk_holder(struct block_device *bdev,
2106 				  struct gendisk *disk);
2107 #else
2108 static inline int bd_link_disk_holder(struct block_device *bdev,
2109 				      struct gendisk *disk)
2110 {
2111 	return 0;
2112 }
2113 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2114 					 struct gendisk *disk)
2115 {
2116 }
2117 #endif
2118 #endif
2119 
2120 /* fs/char_dev.c */
2121 #define CHRDEV_MAJOR_HASH_SIZE	255
2122 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2123 extern int register_chrdev_region(dev_t, unsigned, const char *);
2124 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2125 			     unsigned int count, const char *name,
2126 			     const struct file_operations *fops);
2127 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2128 				unsigned int count, const char *name);
2129 extern void unregister_chrdev_region(dev_t, unsigned);
2130 extern void chrdev_show(struct seq_file *,off_t);
2131 
2132 static inline int register_chrdev(unsigned int major, const char *name,
2133 				  const struct file_operations *fops)
2134 {
2135 	return __register_chrdev(major, 0, 256, name, fops);
2136 }
2137 
2138 static inline void unregister_chrdev(unsigned int major, const char *name)
2139 {
2140 	__unregister_chrdev(major, 0, 256, name);
2141 }
2142 
2143 /* fs/block_dev.c */
2144 #define BDEVNAME_SIZE	32	/* Largest string for a blockdev identifier */
2145 #define BDEVT_SIZE	10	/* Largest string for MAJ:MIN for blkdev */
2146 
2147 #ifdef CONFIG_BLOCK
2148 #define BLKDEV_MAJOR_HASH_SIZE	255
2149 extern const char *__bdevname(dev_t, char *buffer);
2150 extern const char *bdevname(struct block_device *bdev, char *buffer);
2151 extern struct block_device *lookup_bdev(const char *);
2152 extern void blkdev_show(struct seq_file *,off_t);
2153 
2154 #else
2155 #define BLKDEV_MAJOR_HASH_SIZE	0
2156 #endif
2157 
2158 extern void init_special_inode(struct inode *, umode_t, dev_t);
2159 
2160 /* Invalid inode operations -- fs/bad_inode.c */
2161 extern void make_bad_inode(struct inode *);
2162 extern int is_bad_inode(struct inode *);
2163 
2164 extern const struct file_operations read_pipefifo_fops;
2165 extern const struct file_operations write_pipefifo_fops;
2166 extern const struct file_operations rdwr_pipefifo_fops;
2167 
2168 extern int fs_may_remount_ro(struct super_block *);
2169 
2170 #ifdef CONFIG_BLOCK
2171 /*
2172  * return READ, READA, or WRITE
2173  */
2174 #define bio_rw(bio)		((bio)->bi_rw & (RW_MASK | RWA_MASK))
2175 
2176 /*
2177  * return data direction, READ or WRITE
2178  */
2179 #define bio_data_dir(bio)	((bio)->bi_rw & 1)
2180 
2181 extern void check_disk_size_change(struct gendisk *disk,
2182 				   struct block_device *bdev);
2183 extern int revalidate_disk(struct gendisk *);
2184 extern int check_disk_change(struct block_device *);
2185 extern int __invalidate_device(struct block_device *, bool);
2186 extern int invalidate_partition(struct gendisk *, int);
2187 #endif
2188 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2189 					pgoff_t start, pgoff_t end);
2190 
2191 static inline void invalidate_remote_inode(struct inode *inode)
2192 {
2193 	if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2194 	    S_ISLNK(inode->i_mode))
2195 		invalidate_mapping_pages(inode->i_mapping, 0, -1);
2196 }
2197 extern int invalidate_inode_pages2(struct address_space *mapping);
2198 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2199 					 pgoff_t start, pgoff_t end);
2200 extern int write_inode_now(struct inode *, int);
2201 extern int filemap_fdatawrite(struct address_space *);
2202 extern int filemap_flush(struct address_space *);
2203 extern int filemap_fdatawait(struct address_space *);
2204 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2205 				   loff_t lend);
2206 extern int filemap_write_and_wait(struct address_space *mapping);
2207 extern int filemap_write_and_wait_range(struct address_space *mapping,
2208 				        loff_t lstart, loff_t lend);
2209 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2210 				loff_t start, loff_t end, int sync_mode);
2211 extern int filemap_fdatawrite_range(struct address_space *mapping,
2212 				loff_t start, loff_t end);
2213 
2214 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2215 			   int datasync);
2216 extern int vfs_fsync(struct file *file, int datasync);
2217 extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
2218 extern void sync_supers(void);
2219 extern void emergency_sync(void);
2220 extern void emergency_remount(void);
2221 #ifdef CONFIG_BLOCK
2222 extern sector_t bmap(struct inode *, sector_t);
2223 #endif
2224 extern int notify_change(struct dentry *, struct iattr *);
2225 extern int inode_permission(struct inode *, int);
2226 extern int generic_permission(struct inode *, int);
2227 
2228 static inline bool execute_ok(struct inode *inode)
2229 {
2230 	return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2231 }
2232 
2233 /*
2234  * get_write_access() gets write permission for a file.
2235  * put_write_access() releases this write permission.
2236  * This is used for regular files.
2237  * We cannot support write (and maybe mmap read-write shared) accesses and
2238  * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2239  * can have the following values:
2240  * 0: no writers, no VM_DENYWRITE mappings
2241  * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2242  * > 0: (i_writecount) users are writing to the file.
2243  *
2244  * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2245  * except for the cases where we don't hold i_writecount yet. Then we need to
2246  * use {get,deny}_write_access() - these functions check the sign and refuse
2247  * to do the change if sign is wrong.
2248  */
2249 static inline int get_write_access(struct inode *inode)
2250 {
2251 	return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2252 }
2253 static inline int deny_write_access(struct file *file)
2254 {
2255 	struct inode *inode = file->f_path.dentry->d_inode;
2256 	return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2257 }
2258 static inline void put_write_access(struct inode * inode)
2259 {
2260 	atomic_dec(&inode->i_writecount);
2261 }
2262 static inline void allow_write_access(struct file *file)
2263 {
2264 	if (file)
2265 		atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
2266 }
2267 #ifdef CONFIG_IMA
2268 static inline void i_readcount_dec(struct inode *inode)
2269 {
2270 	BUG_ON(!atomic_read(&inode->i_readcount));
2271 	atomic_dec(&inode->i_readcount);
2272 }
2273 static inline void i_readcount_inc(struct inode *inode)
2274 {
2275 	atomic_inc(&inode->i_readcount);
2276 }
2277 #else
2278 static inline void i_readcount_dec(struct inode *inode)
2279 {
2280 	return;
2281 }
2282 static inline void i_readcount_inc(struct inode *inode)
2283 {
2284 	return;
2285 }
2286 #endif
2287 extern int do_pipe_flags(int *, int);
2288 extern struct file *create_read_pipe(struct file *f, int flags);
2289 extern struct file *create_write_pipe(int flags);
2290 extern void free_write_pipe(struct file *);
2291 
2292 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2293 extern struct file * open_exec(const char *);
2294 
2295 /* fs/dcache.c -- generic fs support functions */
2296 extern int is_subdir(struct dentry *, struct dentry *);
2297 extern int path_is_under(struct path *, struct path *);
2298 extern ino_t find_inode_number(struct dentry *, struct qstr *);
2299 
2300 #include <linux/err.h>
2301 
2302 /* needed for stackable file system support */
2303 extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
2304 
2305 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
2306 
2307 extern int inode_init_always(struct super_block *, struct inode *);
2308 extern void inode_init_once(struct inode *);
2309 extern void address_space_init_once(struct address_space *mapping);
2310 extern void ihold(struct inode * inode);
2311 extern void iput(struct inode *);
2312 extern struct inode * igrab(struct inode *);
2313 extern ino_t iunique(struct super_block *, ino_t);
2314 extern int inode_needs_sync(struct inode *inode);
2315 extern int generic_delete_inode(struct inode *inode);
2316 extern int generic_drop_inode(struct inode *inode);
2317 
2318 extern struct inode *ilookup5_nowait(struct super_block *sb,
2319 		unsigned long hashval, int (*test)(struct inode *, void *),
2320 		void *data);
2321 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2322 		int (*test)(struct inode *, void *), void *data);
2323 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2324 
2325 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2326 extern struct inode * iget_locked(struct super_block *, unsigned long);
2327 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2328 extern int insert_inode_locked(struct inode *);
2329 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2330 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2331 #else
2332 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2333 #endif
2334 extern void unlock_new_inode(struct inode *);
2335 extern unsigned int get_next_ino(void);
2336 
2337 extern void __iget(struct inode * inode);
2338 extern void iget_failed(struct inode *);
2339 extern void end_writeback(struct inode *);
2340 extern void __destroy_inode(struct inode *);
2341 extern struct inode *new_inode_pseudo(struct super_block *sb);
2342 extern struct inode *new_inode(struct super_block *sb);
2343 extern void free_inode_nonrcu(struct inode *inode);
2344 extern int should_remove_suid(struct dentry *);
2345 extern int file_remove_suid(struct file *);
2346 
2347 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2348 static inline void insert_inode_hash(struct inode *inode)
2349 {
2350 	__insert_inode_hash(inode, inode->i_ino);
2351 }
2352 
2353 extern void __remove_inode_hash(struct inode *);
2354 static inline void remove_inode_hash(struct inode *inode)
2355 {
2356 	if (!inode_unhashed(inode))
2357 		__remove_inode_hash(inode);
2358 }
2359 
2360 extern void inode_sb_list_add(struct inode *inode);
2361 
2362 #ifdef CONFIG_BLOCK
2363 extern void submit_bio(int, struct bio *);
2364 extern int bdev_read_only(struct block_device *);
2365 #endif
2366 extern int set_blocksize(struct block_device *, int);
2367 extern int sb_set_blocksize(struct super_block *, int);
2368 extern int sb_min_blocksize(struct super_block *, int);
2369 
2370 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2371 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2372 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
2373 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2374 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2375 extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2376 		loff_t *);
2377 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2378 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2379 		unsigned long *, loff_t, loff_t *, size_t, size_t);
2380 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2381 		unsigned long, loff_t, loff_t *, size_t, ssize_t);
2382 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2383 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2384 extern int generic_segment_checks(const struct iovec *iov,
2385 		unsigned long *nr_segs, size_t *count, int access_flags);
2386 
2387 /* fs/block_dev.c */
2388 extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2389 				unsigned long nr_segs, loff_t pos);
2390 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2391 			int datasync);
2392 
2393 /* fs/splice.c */
2394 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2395 		struct pipe_inode_info *, size_t, unsigned int);
2396 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2397 		struct pipe_inode_info *, size_t, unsigned int);
2398 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
2399 		struct file *, loff_t *, size_t, unsigned int);
2400 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2401 		struct file *out, loff_t *, size_t len, unsigned int flags);
2402 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2403 		size_t len, unsigned int flags);
2404 
2405 extern void
2406 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2407 extern loff_t noop_llseek(struct file *file, loff_t offset, int origin);
2408 extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
2409 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
2410 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2411 		int origin, loff_t maxsize);
2412 extern int generic_file_open(struct inode * inode, struct file * filp);
2413 extern int nonseekable_open(struct inode * inode, struct file * filp);
2414 
2415 #ifdef CONFIG_FS_XIP
2416 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2417 			     loff_t *ppos);
2418 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2419 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2420 			      size_t len, loff_t *ppos);
2421 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2422 #else
2423 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2424 {
2425 	return 0;
2426 }
2427 #endif
2428 
2429 #ifdef CONFIG_BLOCK
2430 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2431 			    loff_t file_offset);
2432 
2433 enum {
2434 	/* need locking between buffered and direct access */
2435 	DIO_LOCKING	= 0x01,
2436 
2437 	/* filesystem does not support filling holes */
2438 	DIO_SKIP_HOLES	= 0x02,
2439 };
2440 
2441 void dio_end_io(struct bio *bio, int error);
2442 void inode_dio_wait(struct inode *inode);
2443 void inode_dio_done(struct inode *inode);
2444 
2445 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2446 	struct block_device *bdev, const struct iovec *iov, loff_t offset,
2447 	unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2448 	dio_submit_t submit_io,	int flags);
2449 
2450 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2451 		struct inode *inode, const struct iovec *iov, loff_t offset,
2452 		unsigned long nr_segs, get_block_t get_block)
2453 {
2454 	return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
2455 				    offset, nr_segs, get_block, NULL, NULL,
2456 				    DIO_LOCKING | DIO_SKIP_HOLES);
2457 }
2458 #else
2459 static inline void inode_dio_wait(struct inode *inode)
2460 {
2461 }
2462 #endif
2463 
2464 extern const struct file_operations generic_ro_fops;
2465 
2466 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2467 
2468 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2469 extern int vfs_follow_link(struct nameidata *, const char *);
2470 extern int page_readlink(struct dentry *, char __user *, int);
2471 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2472 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2473 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2474 		int nofs);
2475 extern int page_symlink(struct inode *inode, const char *symname, int len);
2476 extern const struct inode_operations page_symlink_inode_operations;
2477 extern int generic_readlink(struct dentry *, char __user *, int);
2478 extern void generic_fillattr(struct inode *, struct kstat *);
2479 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2480 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2481 void inode_add_bytes(struct inode *inode, loff_t bytes);
2482 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2483 loff_t inode_get_bytes(struct inode *inode);
2484 void inode_set_bytes(struct inode *inode, loff_t bytes);
2485 
2486 extern int vfs_readdir(struct file *, filldir_t, void *);
2487 
2488 extern int vfs_stat(const char __user *, struct kstat *);
2489 extern int vfs_lstat(const char __user *, struct kstat *);
2490 extern int vfs_fstat(unsigned int, struct kstat *);
2491 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2492 
2493 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2494 		    unsigned long arg);
2495 extern int __generic_block_fiemap(struct inode *inode,
2496 				  struct fiemap_extent_info *fieinfo,
2497 				  loff_t start, loff_t len,
2498 				  get_block_t *get_block);
2499 extern int generic_block_fiemap(struct inode *inode,
2500 				struct fiemap_extent_info *fieinfo, u64 start,
2501 				u64 len, get_block_t *get_block);
2502 
2503 extern void get_filesystem(struct file_system_type *fs);
2504 extern void put_filesystem(struct file_system_type *fs);
2505 extern struct file_system_type *get_fs_type(const char *name);
2506 extern struct super_block *get_super(struct block_device *);
2507 extern struct super_block *get_active_super(struct block_device *bdev);
2508 extern struct super_block *user_get_super(dev_t);
2509 extern void drop_super(struct super_block *sb);
2510 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2511 extern void iterate_supers_type(struct file_system_type *,
2512 			        void (*)(struct super_block *, void *), void *);
2513 
2514 extern int dcache_dir_open(struct inode *, struct file *);
2515 extern int dcache_dir_close(struct inode *, struct file *);
2516 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2517 extern int dcache_readdir(struct file *, void *, filldir_t);
2518 extern int simple_setattr(struct dentry *, struct iattr *);
2519 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2520 extern int simple_statfs(struct dentry *, struct kstatfs *);
2521 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2522 extern int simple_unlink(struct inode *, struct dentry *);
2523 extern int simple_rmdir(struct inode *, struct dentry *);
2524 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2525 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2526 extern int simple_empty(struct dentry *);
2527 extern int simple_readpage(struct file *file, struct page *page);
2528 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2529 			loff_t pos, unsigned len, unsigned flags,
2530 			struct page **pagep, void **fsdata);
2531 extern int simple_write_end(struct file *file, struct address_space *mapping,
2532 			loff_t pos, unsigned len, unsigned copied,
2533 			struct page *page, void *fsdata);
2534 
2535 extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
2536 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2537 extern const struct file_operations simple_dir_operations;
2538 extern const struct inode_operations simple_dir_inode_operations;
2539 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2540 struct dentry *d_alloc_name(struct dentry *, const char *);
2541 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2542 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2543 extern void simple_release_fs(struct vfsmount **mount, int *count);
2544 
2545 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2546 			loff_t *ppos, const void *from, size_t available);
2547 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2548 		const void __user *from, size_t count);
2549 
2550 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2551 
2552 extern int generic_check_addressable(unsigned, u64);
2553 
2554 #ifdef CONFIG_MIGRATION
2555 extern int buffer_migrate_page(struct address_space *,
2556 				struct page *, struct page *);
2557 #else
2558 #define buffer_migrate_page NULL
2559 #endif
2560 
2561 extern int inode_change_ok(const struct inode *, struct iattr *);
2562 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2563 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2564 
2565 extern void file_update_time(struct file *file);
2566 
2567 extern int generic_show_options(struct seq_file *m, struct vfsmount *mnt);
2568 extern void save_mount_options(struct super_block *sb, char *options);
2569 extern void replace_mount_options(struct super_block *sb, char *options);
2570 
2571 static inline ino_t parent_ino(struct dentry *dentry)
2572 {
2573 	ino_t res;
2574 
2575 	/*
2576 	 * Don't strictly need d_lock here? If the parent ino could change
2577 	 * then surely we'd have a deeper race in the caller?
2578 	 */
2579 	spin_lock(&dentry->d_lock);
2580 	res = dentry->d_parent->d_inode->i_ino;
2581 	spin_unlock(&dentry->d_lock);
2582 	return res;
2583 }
2584 
2585 /* Transaction based IO helpers */
2586 
2587 /*
2588  * An argresp is stored in an allocated page and holds the
2589  * size of the argument or response, along with its content
2590  */
2591 struct simple_transaction_argresp {
2592 	ssize_t size;
2593 	char data[0];
2594 };
2595 
2596 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2597 
2598 char *simple_transaction_get(struct file *file, const char __user *buf,
2599 				size_t size);
2600 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2601 				size_t size, loff_t *pos);
2602 int simple_transaction_release(struct inode *inode, struct file *file);
2603 
2604 void simple_transaction_set(struct file *file, size_t n);
2605 
2606 /*
2607  * simple attribute files
2608  *
2609  * These attributes behave similar to those in sysfs:
2610  *
2611  * Writing to an attribute immediately sets a value, an open file can be
2612  * written to multiple times.
2613  *
2614  * Reading from an attribute creates a buffer from the value that might get
2615  * read with multiple read calls. When the attribute has been read
2616  * completely, no further read calls are possible until the file is opened
2617  * again.
2618  *
2619  * All attributes contain a text representation of a numeric value
2620  * that are accessed with the get() and set() functions.
2621  */
2622 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt)		\
2623 static int __fops ## _open(struct inode *inode, struct file *file)	\
2624 {									\
2625 	__simple_attr_check_format(__fmt, 0ull);			\
2626 	return simple_attr_open(inode, file, __get, __set, __fmt);	\
2627 }									\
2628 static const struct file_operations __fops = {				\
2629 	.owner	 = THIS_MODULE,						\
2630 	.open	 = __fops ## _open,					\
2631 	.release = simple_attr_release,					\
2632 	.read	 = simple_attr_read,					\
2633 	.write	 = simple_attr_write,					\
2634 	.llseek	 = generic_file_llseek,					\
2635 };
2636 
2637 static inline void __attribute__((format(printf, 1, 2)))
2638 __simple_attr_check_format(const char *fmt, ...)
2639 {
2640 	/* don't do anything, just let the compiler check the arguments; */
2641 }
2642 
2643 int simple_attr_open(struct inode *inode, struct file *file,
2644 		     int (*get)(void *, u64 *), int (*set)(void *, u64),
2645 		     const char *fmt);
2646 int simple_attr_release(struct inode *inode, struct file *file);
2647 ssize_t simple_attr_read(struct file *file, char __user *buf,
2648 			 size_t len, loff_t *ppos);
2649 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2650 			  size_t len, loff_t *ppos);
2651 
2652 struct ctl_table;
2653 int proc_nr_files(struct ctl_table *table, int write,
2654 		  void __user *buffer, size_t *lenp, loff_t *ppos);
2655 int proc_nr_dentry(struct ctl_table *table, int write,
2656 		  void __user *buffer, size_t *lenp, loff_t *ppos);
2657 int proc_nr_inodes(struct ctl_table *table, int write,
2658 		   void __user *buffer, size_t *lenp, loff_t *ppos);
2659 int __init get_filesystem_list(char *buf);
2660 
2661 #define __FMODE_EXEC		((__force int) FMODE_EXEC)
2662 #define __FMODE_NONOTIFY	((__force int) FMODE_NONOTIFY)
2663 
2664 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2665 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2666 					    (flag & __FMODE_NONOTIFY)))
2667 
2668 static inline int is_sxid(mode_t mode)
2669 {
2670 	return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2671 }
2672 
2673 static inline void inode_has_no_xattr(struct inode *inode)
2674 {
2675 	if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2676 		inode->i_flags |= S_NOSEC;
2677 }
2678 
2679 #endif /* __KERNEL__ */
2680 #endif /* _LINUX_FS_H */
2681