xref: /linux-6.15/include/linux/fs.h (revision 82de9a0c)
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 1
62 #define MAY_WRITE 2
63 #define MAY_READ 4
64 #define MAY_APPEND 8
65 #define MAY_ACCESS 16
66 #define MAY_OPEN 32
67 #define MAY_CHDIR 64
68 #define MAY_NOT_BLOCK 128	/* called from RCU mode, don't block */
69 
70 /*
71  * flags in file.f_mode.  Note that FMODE_READ and FMODE_WRITE must correspond
72  * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
73  */
74 
75 /* file is open for reading */
76 #define FMODE_READ		((__force fmode_t)0x1)
77 /* file is open for writing */
78 #define FMODE_WRITE		((__force fmode_t)0x2)
79 /* file is seekable */
80 #define FMODE_LSEEK		((__force fmode_t)0x4)
81 /* file can be accessed using pread */
82 #define FMODE_PREAD		((__force fmode_t)0x8)
83 /* file can be accessed using pwrite */
84 #define FMODE_PWRITE		((__force fmode_t)0x10)
85 /* File is opened for execution with sys_execve / sys_uselib */
86 #define FMODE_EXEC		((__force fmode_t)0x20)
87 /* File is opened with O_NDELAY (only set for block devices) */
88 #define FMODE_NDELAY		((__force fmode_t)0x40)
89 /* File is opened with O_EXCL (only set for block devices) */
90 #define FMODE_EXCL		((__force fmode_t)0x80)
91 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
92    (specialy hack for floppy.c) */
93 #define FMODE_WRITE_IOCTL	((__force fmode_t)0x100)
94 
95 /*
96  * Don't update ctime and mtime.
97  *
98  * Currently a special hack for the XFS open_by_handle ioctl, but we'll
99  * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
100  */
101 #define FMODE_NOCMTIME		((__force fmode_t)0x800)
102 
103 /* Expect random access pattern */
104 #define FMODE_RANDOM		((__force fmode_t)0x1000)
105 
106 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
107 #define FMODE_UNSIGNED_OFFSET	((__force fmode_t)0x2000)
108 
109 /* File is opened with O_PATH; almost nothing can be done with it */
110 #define FMODE_PATH		((__force fmode_t)0x4000)
111 
112 /* File was opened by fanotify and shouldn't generate fanotify events */
113 #define FMODE_NONOTIFY		((__force fmode_t)0x1000000)
114 
115 /*
116  * The below are the various read and write types that we support. Some of
117  * them include behavioral modifiers that send information down to the
118  * block layer and IO scheduler. Terminology:
119  *
120  *	The block layer uses device plugging to defer IO a little bit, in
121  *	the hope that we will see more IO very shortly. This increases
122  *	coalescing of adjacent IO and thus reduces the number of IOs we
123  *	have to send to the device. It also allows for better queuing,
124  *	if the IO isn't mergeable. If the caller is going to be waiting
125  *	for the IO, then he must ensure that the device is unplugged so
126  *	that the IO is dispatched to the driver.
127  *
128  *	All IO is handled async in Linux. This is fine for background
129  *	writes, but for reads or writes that someone waits for completion
130  *	on, we want to notify the block layer and IO scheduler so that they
131  *	know about it. That allows them to make better scheduling
132  *	decisions. So when the below references 'sync' and 'async', it
133  *	is referencing this priority hint.
134  *
135  * With that in mind, the available types are:
136  *
137  * READ			A normal read operation. Device will be plugged.
138  * READ_SYNC		A synchronous read. Device is not plugged, caller can
139  *			immediately wait on this read without caring about
140  *			unplugging.
141  * READA		Used for read-ahead operations. Lower priority, and the
142  *			block layer could (in theory) choose to ignore this
143  *			request if it runs into resource problems.
144  * WRITE		A normal async write. Device will be plugged.
145  * WRITE_SYNC		Synchronous write. Identical to WRITE, but passes down
146  *			the hint that someone will be waiting on this IO
147  *			shortly. The write equivalent of READ_SYNC.
148  * WRITE_ODIRECT	Special case write for O_DIRECT only.
149  * WRITE_FLUSH		Like WRITE_SYNC but with preceding cache flush.
150  * WRITE_FUA		Like WRITE_SYNC but data is guaranteed to be on
151  *			non-volatile media on completion.
152  * WRITE_FLUSH_FUA	Combination of WRITE_FLUSH and FUA. The IO is preceded
153  *			by a cache flush and data is guaranteed to be on
154  *			non-volatile media on completion.
155  *
156  */
157 #define RW_MASK			REQ_WRITE
158 #define RWA_MASK		REQ_RAHEAD
159 
160 #define READ			0
161 #define WRITE			RW_MASK
162 #define READA			RWA_MASK
163 
164 #define READ_SYNC		(READ | REQ_SYNC)
165 #define READ_META		(READ | REQ_META)
166 #define WRITE_SYNC		(WRITE | REQ_SYNC | REQ_NOIDLE)
167 #define WRITE_ODIRECT		(WRITE | REQ_SYNC)
168 #define WRITE_META		(WRITE | REQ_META)
169 #define WRITE_FLUSH		(WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
170 #define WRITE_FUA		(WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
171 #define WRITE_FLUSH_FUA		(WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
172 
173 #define SEL_IN		1
174 #define SEL_OUT		2
175 #define SEL_EX		4
176 
177 /* public flags for file_system_type */
178 #define FS_REQUIRES_DEV 1
179 #define FS_BINARY_MOUNTDATA 2
180 #define FS_HAS_SUBTYPE 4
181 #define FS_REVAL_DOT	16384	/* Check the paths ".", ".." for staleness */
182 #define FS_RENAME_DOES_D_MOVE	32768	/* FS will handle d_move()
183 					 * during rename() internally.
184 					 */
185 
186 /*
187  * These are the fs-independent mount-flags: up to 32 flags are supported
188  */
189 #define MS_RDONLY	 1	/* Mount read-only */
190 #define MS_NOSUID	 2	/* Ignore suid and sgid bits */
191 #define MS_NODEV	 4	/* Disallow access to device special files */
192 #define MS_NOEXEC	 8	/* Disallow program execution */
193 #define MS_SYNCHRONOUS	16	/* Writes are synced at once */
194 #define MS_REMOUNT	32	/* Alter flags of a mounted FS */
195 #define MS_MANDLOCK	64	/* Allow mandatory locks on an FS */
196 #define MS_DIRSYNC	128	/* Directory modifications are synchronous */
197 #define MS_NOATIME	1024	/* Do not update access times. */
198 #define MS_NODIRATIME	2048	/* Do not update directory access times */
199 #define MS_BIND		4096
200 #define MS_MOVE		8192
201 #define MS_REC		16384
202 #define MS_VERBOSE	32768	/* War is peace. Verbosity is silence.
203 				   MS_VERBOSE is deprecated. */
204 #define MS_SILENT	32768
205 #define MS_POSIXACL	(1<<16)	/* VFS does not apply the umask */
206 #define MS_UNBINDABLE	(1<<17)	/* change to unbindable */
207 #define MS_PRIVATE	(1<<18)	/* change to private */
208 #define MS_SLAVE	(1<<19)	/* change to slave */
209 #define MS_SHARED	(1<<20)	/* change to shared */
210 #define MS_RELATIME	(1<<21)	/* Update atime relative to mtime/ctime. */
211 #define MS_KERNMOUNT	(1<<22) /* this is a kern_mount call */
212 #define MS_I_VERSION	(1<<23) /* Update inode I_version field */
213 #define MS_STRICTATIME	(1<<24) /* Always perform atime updates */
214 #define MS_NOSEC	(1<<28)
215 #define MS_BORN		(1<<29)
216 #define MS_ACTIVE	(1<<30)
217 #define MS_NOUSER	(1<<31)
218 
219 /*
220  * Superblock flags that can be altered by MS_REMOUNT
221  */
222 #define MS_RMT_MASK	(MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_I_VERSION)
223 
224 /*
225  * Old magic mount flag and mask
226  */
227 #define MS_MGC_VAL 0xC0ED0000
228 #define MS_MGC_MSK 0xffff0000
229 
230 /* Inode flags - they have nothing to superblock flags now */
231 
232 #define S_SYNC		1	/* Writes are synced at once */
233 #define S_NOATIME	2	/* Do not update access times */
234 #define S_APPEND	4	/* Append-only file */
235 #define S_IMMUTABLE	8	/* Immutable file */
236 #define S_DEAD		16	/* removed, but still open directory */
237 #define S_NOQUOTA	32	/* Inode is not counted to quota */
238 #define S_DIRSYNC	64	/* Directory modifications are synchronous */
239 #define S_NOCMTIME	128	/* Do not update file c/mtime */
240 #define S_SWAPFILE	256	/* Do not truncate: swapon got its bmaps */
241 #define S_PRIVATE	512	/* Inode is fs-internal */
242 #define S_IMA		1024	/* Inode has an associated IMA struct */
243 #define S_AUTOMOUNT	2048	/* Automount/referral quasi-directory */
244 #define S_NOSEC		4096	/* no suid or xattr security attributes */
245 
246 /*
247  * Note that nosuid etc flags are inode-specific: setting some file-system
248  * flags just means all the inodes inherit those flags by default. It might be
249  * possible to override it selectively if you really wanted to with some
250  * ioctl() that is not currently implemented.
251  *
252  * Exception: MS_RDONLY is always applied to the entire file system.
253  *
254  * Unfortunately, it is possible to change a filesystems flags with it mounted
255  * with files in use.  This means that all of the inodes will not have their
256  * i_flags updated.  Hence, i_flags no longer inherit the superblock mount
257  * flags, so these have to be checked separately. -- [email protected]
258  */
259 #define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
260 
261 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
262 #define IS_SYNC(inode)		(__IS_FLG(inode, MS_SYNCHRONOUS) || \
263 					((inode)->i_flags & S_SYNC))
264 #define IS_DIRSYNC(inode)	(__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
265 					((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
266 #define IS_MANDLOCK(inode)	__IS_FLG(inode, MS_MANDLOCK)
267 #define IS_NOATIME(inode)   __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
268 #define IS_I_VERSION(inode)   __IS_FLG(inode, MS_I_VERSION)
269 
270 #define IS_NOQUOTA(inode)	((inode)->i_flags & S_NOQUOTA)
271 #define IS_APPEND(inode)	((inode)->i_flags & S_APPEND)
272 #define IS_IMMUTABLE(inode)	((inode)->i_flags & S_IMMUTABLE)
273 #define IS_POSIXACL(inode)	__IS_FLG(inode, MS_POSIXACL)
274 
275 #define IS_DEADDIR(inode)	((inode)->i_flags & S_DEAD)
276 #define IS_NOCMTIME(inode)	((inode)->i_flags & S_NOCMTIME)
277 #define IS_SWAPFILE(inode)	((inode)->i_flags & S_SWAPFILE)
278 #define IS_PRIVATE(inode)	((inode)->i_flags & S_PRIVATE)
279 #define IS_IMA(inode)		((inode)->i_flags & S_IMA)
280 #define IS_AUTOMOUNT(inode)	((inode)->i_flags & S_AUTOMOUNT)
281 #define IS_NOSEC(inode)		((inode)->i_flags & S_NOSEC)
282 
283 /* the read-only stuff doesn't really belong here, but any other place is
284    probably as bad and I don't want to create yet another include file. */
285 
286 #define BLKROSET   _IO(0x12,93)	/* set device read-only (0 = read-write) */
287 #define BLKROGET   _IO(0x12,94)	/* get read-only status (0 = read_write) */
288 #define BLKRRPART  _IO(0x12,95)	/* re-read partition table */
289 #define BLKGETSIZE _IO(0x12,96)	/* return device size /512 (long *arg) */
290 #define BLKFLSBUF  _IO(0x12,97)	/* flush buffer cache */
291 #define BLKRASET   _IO(0x12,98)	/* set read ahead for block device */
292 #define BLKRAGET   _IO(0x12,99)	/* get current read ahead setting */
293 #define BLKFRASET  _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
294 #define BLKFRAGET  _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
295 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
296 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
297 #define BLKSSZGET  _IO(0x12,104)/* get block device sector size */
298 #if 0
299 #define BLKPG      _IO(0x12,105)/* See blkpg.h */
300 
301 /* Some people are morons.  Do not use sizeof! */
302 
303 #define BLKELVGET  _IOR(0x12,106,size_t)/* elevator get */
304 #define BLKELVSET  _IOW(0x12,107,size_t)/* elevator set */
305 /* This was here just to show that the number is taken -
306    probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
307 #endif
308 /* A jump here: 108-111 have been used for various private purposes. */
309 #define BLKBSZGET  _IOR(0x12,112,size_t)
310 #define BLKBSZSET  _IOW(0x12,113,size_t)
311 #define BLKGETSIZE64 _IOR(0x12,114,size_t)	/* return device size in bytes (u64 *arg) */
312 #define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
313 #define BLKTRACESTART _IO(0x12,116)
314 #define BLKTRACESTOP _IO(0x12,117)
315 #define BLKTRACETEARDOWN _IO(0x12,118)
316 #define BLKDISCARD _IO(0x12,119)
317 #define BLKIOMIN _IO(0x12,120)
318 #define BLKIOOPT _IO(0x12,121)
319 #define BLKALIGNOFF _IO(0x12,122)
320 #define BLKPBSZGET _IO(0x12,123)
321 #define BLKDISCARDZEROES _IO(0x12,124)
322 #define BLKSECDISCARD _IO(0x12,125)
323 
324 #define BMAP_IOCTL 1		/* obsolete - kept for compatibility */
325 #define FIBMAP	   _IO(0x00,1)	/* bmap access */
326 #define FIGETBSZ   _IO(0x00,2)	/* get the block size used for bmap */
327 #define FIFREEZE	_IOWR('X', 119, int)	/* Freeze */
328 #define FITHAW		_IOWR('X', 120, int)	/* Thaw */
329 #define FITRIM		_IOWR('X', 121, struct fstrim_range)	/* Trim */
330 
331 #define	FS_IOC_GETFLAGS			_IOR('f', 1, long)
332 #define	FS_IOC_SETFLAGS			_IOW('f', 2, long)
333 #define	FS_IOC_GETVERSION		_IOR('v', 1, long)
334 #define	FS_IOC_SETVERSION		_IOW('v', 2, long)
335 #define FS_IOC_FIEMAP			_IOWR('f', 11, struct fiemap)
336 #define FS_IOC32_GETFLAGS		_IOR('f', 1, int)
337 #define FS_IOC32_SETFLAGS		_IOW('f', 2, int)
338 #define FS_IOC32_GETVERSION		_IOR('v', 1, int)
339 #define FS_IOC32_SETVERSION		_IOW('v', 2, int)
340 
341 /*
342  * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS)
343  */
344 #define	FS_SECRM_FL			0x00000001 /* Secure deletion */
345 #define	FS_UNRM_FL			0x00000002 /* Undelete */
346 #define	FS_COMPR_FL			0x00000004 /* Compress file */
347 #define FS_SYNC_FL			0x00000008 /* Synchronous updates */
348 #define FS_IMMUTABLE_FL			0x00000010 /* Immutable file */
349 #define FS_APPEND_FL			0x00000020 /* writes to file may only append */
350 #define FS_NODUMP_FL			0x00000040 /* do not dump file */
351 #define FS_NOATIME_FL			0x00000080 /* do not update atime */
352 /* Reserved for compression usage... */
353 #define FS_DIRTY_FL			0x00000100
354 #define FS_COMPRBLK_FL			0x00000200 /* One or more compressed clusters */
355 #define FS_NOCOMP_FL			0x00000400 /* Don't compress */
356 #define FS_ECOMPR_FL			0x00000800 /* Compression error */
357 /* End compression flags --- maybe not all used */
358 #define FS_BTREE_FL			0x00001000 /* btree format dir */
359 #define FS_INDEX_FL			0x00001000 /* hash-indexed directory */
360 #define FS_IMAGIC_FL			0x00002000 /* AFS directory */
361 #define FS_JOURNAL_DATA_FL		0x00004000 /* Reserved for ext3 */
362 #define FS_NOTAIL_FL			0x00008000 /* file tail should not be merged */
363 #define FS_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
364 #define FS_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
365 #define FS_EXTENT_FL			0x00080000 /* Extents */
366 #define FS_DIRECTIO_FL			0x00100000 /* Use direct i/o */
367 #define FS_NOCOW_FL			0x00800000 /* Do not cow file */
368 #define FS_RESERVED_FL			0x80000000 /* reserved for ext2 lib */
369 
370 #define FS_FL_USER_VISIBLE		0x0003DFFF /* User visible flags */
371 #define FS_FL_USER_MODIFIABLE		0x000380FF /* User modifiable flags */
372 
373 
374 #define SYNC_FILE_RANGE_WAIT_BEFORE	1
375 #define SYNC_FILE_RANGE_WRITE		2
376 #define SYNC_FILE_RANGE_WAIT_AFTER	4
377 
378 #ifdef __KERNEL__
379 
380 #include <linux/linkage.h>
381 #include <linux/wait.h>
382 #include <linux/kdev_t.h>
383 #include <linux/dcache.h>
384 #include <linux/path.h>
385 #include <linux/stat.h>
386 #include <linux/cache.h>
387 #include <linux/list.h>
388 #include <linux/radix-tree.h>
389 #include <linux/prio_tree.h>
390 #include <linux/init.h>
391 #include <linux/pid.h>
392 #include <linux/mutex.h>
393 #include <linux/capability.h>
394 #include <linux/semaphore.h>
395 #include <linux/fiemap.h>
396 #include <linux/rculist_bl.h>
397 #include <linux/shrinker.h>
398 #include <linux/atomic.h>
399 
400 #include <asm/byteorder.h>
401 
402 struct export_operations;
403 struct hd_geometry;
404 struct iovec;
405 struct nameidata;
406 struct kiocb;
407 struct kobject;
408 struct pipe_inode_info;
409 struct poll_table_struct;
410 struct kstatfs;
411 struct vm_area_struct;
412 struct vfsmount;
413 struct cred;
414 
415 extern void __init inode_init(void);
416 extern void __init inode_init_early(void);
417 extern void __init files_init(unsigned long);
418 
419 extern struct files_stat_struct files_stat;
420 extern unsigned long get_max_files(void);
421 extern int sysctl_nr_open;
422 extern struct inodes_stat_t inodes_stat;
423 extern int leases_enable, lease_break_time;
424 
425 struct buffer_head;
426 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
427 			struct buffer_head *bh_result, int create);
428 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
429 			ssize_t bytes, void *private, int ret,
430 			bool is_async);
431 
432 /*
433  * Attribute flags.  These should be or-ed together to figure out what
434  * has been changed!
435  */
436 #define ATTR_MODE	(1 << 0)
437 #define ATTR_UID	(1 << 1)
438 #define ATTR_GID	(1 << 2)
439 #define ATTR_SIZE	(1 << 3)
440 #define ATTR_ATIME	(1 << 4)
441 #define ATTR_MTIME	(1 << 5)
442 #define ATTR_CTIME	(1 << 6)
443 #define ATTR_ATIME_SET	(1 << 7)
444 #define ATTR_MTIME_SET	(1 << 8)
445 #define ATTR_FORCE	(1 << 9) /* Not a change, but a change it */
446 #define ATTR_ATTR_FLAG	(1 << 10)
447 #define ATTR_KILL_SUID	(1 << 11)
448 #define ATTR_KILL_SGID	(1 << 12)
449 #define ATTR_FILE	(1 << 13)
450 #define ATTR_KILL_PRIV	(1 << 14)
451 #define ATTR_OPEN	(1 << 15) /* Truncating from open(O_TRUNC) */
452 #define ATTR_TIMES_SET	(1 << 16)
453 
454 /*
455  * This is the Inode Attributes structure, used for notify_change().  It
456  * uses the above definitions as flags, to know which values have changed.
457  * Also, in this manner, a Filesystem can look at only the values it cares
458  * about.  Basically, these are the attributes that the VFS layer can
459  * request to change from the FS layer.
460  *
461  * Derek Atkins <[email protected]> 94-10-20
462  */
463 struct iattr {
464 	unsigned int	ia_valid;
465 	umode_t		ia_mode;
466 	uid_t		ia_uid;
467 	gid_t		ia_gid;
468 	loff_t		ia_size;
469 	struct timespec	ia_atime;
470 	struct timespec	ia_mtime;
471 	struct timespec	ia_ctime;
472 
473 	/*
474 	 * Not an attribute, but an auxiliary info for filesystems wanting to
475 	 * implement an ftruncate() like method.  NOTE: filesystem should
476 	 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
477 	 */
478 	struct file	*ia_file;
479 };
480 
481 /*
482  * Includes for diskquotas.
483  */
484 #include <linux/quota.h>
485 
486 /**
487  * enum positive_aop_returns - aop return codes with specific semantics
488  *
489  * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
490  * 			    completed, that the page is still locked, and
491  * 			    should be considered active.  The VM uses this hint
492  * 			    to return the page to the active list -- it won't
493  * 			    be a candidate for writeback again in the near
494  * 			    future.  Other callers must be careful to unlock
495  * 			    the page if they get this return.  Returned by
496  * 			    writepage();
497  *
498  * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
499  *  			unlocked it and the page might have been truncated.
500  *  			The caller should back up to acquiring a new page and
501  *  			trying again.  The aop will be taking reasonable
502  *  			precautions not to livelock.  If the caller held a page
503  *  			reference, it should drop it before retrying.  Returned
504  *  			by readpage().
505  *
506  * address_space_operation functions return these large constants to indicate
507  * special semantics to the caller.  These are much larger than the bytes in a
508  * page to allow for functions that return the number of bytes operated on in a
509  * given page.
510  */
511 
512 enum positive_aop_returns {
513 	AOP_WRITEPAGE_ACTIVATE	= 0x80000,
514 	AOP_TRUNCATED_PAGE	= 0x80001,
515 };
516 
517 #define AOP_FLAG_UNINTERRUPTIBLE	0x0001 /* will not do a short write */
518 #define AOP_FLAG_CONT_EXPAND		0x0002 /* called from cont_expand */
519 #define AOP_FLAG_NOFS			0x0004 /* used by filesystem to direct
520 						* helper code (eg buffer layer)
521 						* to clear GFP_FS from alloc */
522 
523 /*
524  * oh the beauties of C type declarations.
525  */
526 struct page;
527 struct address_space;
528 struct writeback_control;
529 
530 struct iov_iter {
531 	const struct iovec *iov;
532 	unsigned long nr_segs;
533 	size_t iov_offset;
534 	size_t count;
535 };
536 
537 size_t iov_iter_copy_from_user_atomic(struct page *page,
538 		struct iov_iter *i, unsigned long offset, size_t bytes);
539 size_t iov_iter_copy_from_user(struct page *page,
540 		struct iov_iter *i, unsigned long offset, size_t bytes);
541 void iov_iter_advance(struct iov_iter *i, size_t bytes);
542 int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
543 size_t iov_iter_single_seg_count(struct iov_iter *i);
544 
545 static inline void iov_iter_init(struct iov_iter *i,
546 			const struct iovec *iov, unsigned long nr_segs,
547 			size_t count, size_t written)
548 {
549 	i->iov = iov;
550 	i->nr_segs = nr_segs;
551 	i->iov_offset = 0;
552 	i->count = count + written;
553 
554 	iov_iter_advance(i, written);
555 }
556 
557 static inline size_t iov_iter_count(struct iov_iter *i)
558 {
559 	return i->count;
560 }
561 
562 /*
563  * "descriptor" for what we're up to with a read.
564  * This allows us to use the same read code yet
565  * have multiple different users of the data that
566  * we read from a file.
567  *
568  * The simplest case just copies the data to user
569  * mode.
570  */
571 typedef struct {
572 	size_t written;
573 	size_t count;
574 	union {
575 		char __user *buf;
576 		void *data;
577 	} arg;
578 	int error;
579 } read_descriptor_t;
580 
581 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
582 		unsigned long, unsigned long);
583 
584 struct address_space_operations {
585 	int (*writepage)(struct page *page, struct writeback_control *wbc);
586 	int (*readpage)(struct file *, struct page *);
587 
588 	/* Write back some dirty pages from this mapping. */
589 	int (*writepages)(struct address_space *, struct writeback_control *);
590 
591 	/* Set a page dirty.  Return true if this dirtied it */
592 	int (*set_page_dirty)(struct page *page);
593 
594 	int (*readpages)(struct file *filp, struct address_space *mapping,
595 			struct list_head *pages, unsigned nr_pages);
596 
597 	int (*write_begin)(struct file *, struct address_space *mapping,
598 				loff_t pos, unsigned len, unsigned flags,
599 				struct page **pagep, void **fsdata);
600 	int (*write_end)(struct file *, struct address_space *mapping,
601 				loff_t pos, unsigned len, unsigned copied,
602 				struct page *page, void *fsdata);
603 
604 	/* Unfortunately this kludge is needed for FIBMAP. Don't use it */
605 	sector_t (*bmap)(struct address_space *, sector_t);
606 	void (*invalidatepage) (struct page *, unsigned long);
607 	int (*releasepage) (struct page *, gfp_t);
608 	void (*freepage)(struct page *);
609 	ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
610 			loff_t offset, unsigned long nr_segs);
611 	int (*get_xip_mem)(struct address_space *, pgoff_t, int,
612 						void **, unsigned long *);
613 	/* migrate the contents of a page to the specified target */
614 	int (*migratepage) (struct address_space *,
615 			struct page *, struct page *);
616 	int (*launder_page) (struct page *);
617 	int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
618 					unsigned long);
619 	int (*error_remove_page)(struct address_space *, struct page *);
620 };
621 
622 extern const struct address_space_operations empty_aops;
623 
624 /*
625  * pagecache_write_begin/pagecache_write_end must be used by general code
626  * to write into the pagecache.
627  */
628 int pagecache_write_begin(struct file *, struct address_space *mapping,
629 				loff_t pos, unsigned len, unsigned flags,
630 				struct page **pagep, void **fsdata);
631 
632 int pagecache_write_end(struct file *, struct address_space *mapping,
633 				loff_t pos, unsigned len, unsigned copied,
634 				struct page *page, void *fsdata);
635 
636 struct backing_dev_info;
637 struct address_space {
638 	struct inode		*host;		/* owner: inode, block_device */
639 	struct radix_tree_root	page_tree;	/* radix tree of all pages */
640 	spinlock_t		tree_lock;	/* and lock protecting it */
641 	unsigned int		i_mmap_writable;/* count VM_SHARED mappings */
642 	struct prio_tree_root	i_mmap;		/* tree of private and shared mappings */
643 	struct list_head	i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
644 	struct mutex		i_mmap_mutex;	/* protect tree, count, list */
645 	/* Protected by tree_lock together with the radix tree */
646 	unsigned long		nrpages;	/* number of total pages */
647 	pgoff_t			writeback_index;/* writeback starts here */
648 	const struct address_space_operations *a_ops;	/* methods */
649 	unsigned long		flags;		/* error bits/gfp mask */
650 	struct backing_dev_info *backing_dev_info; /* device readahead, etc */
651 	spinlock_t		private_lock;	/* for use by the address_space */
652 	struct list_head	private_list;	/* ditto */
653 	struct address_space	*assoc_mapping;	/* ditto */
654 } __attribute__((aligned(sizeof(long))));
655 	/*
656 	 * On most architectures that alignment is already the case; but
657 	 * must be enforced here for CRIS, to let the least significant bit
658 	 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
659 	 */
660 
661 struct block_device {
662 	dev_t			bd_dev;  /* not a kdev_t - it's a search key */
663 	int			bd_openers;
664 	struct inode *		bd_inode;	/* will die */
665 	struct super_block *	bd_super;
666 	struct mutex		bd_mutex;	/* open/close mutex */
667 	struct list_head	bd_inodes;
668 	void *			bd_claiming;
669 	void *			bd_holder;
670 	int			bd_holders;
671 	bool			bd_write_holder;
672 #ifdef CONFIG_SYSFS
673 	struct list_head	bd_holder_disks;
674 #endif
675 	struct block_device *	bd_contains;
676 	unsigned		bd_block_size;
677 	struct hd_struct *	bd_part;
678 	/* number of times partitions within this device have been opened. */
679 	unsigned		bd_part_count;
680 	int			bd_invalidated;
681 	struct gendisk *	bd_disk;
682 	struct list_head	bd_list;
683 	/*
684 	 * Private data.  You must have bd_claim'ed the block_device
685 	 * to use this.  NOTE:  bd_claim allows an owner to claim
686 	 * the same device multiple times, the owner must take special
687 	 * care to not mess up bd_private for that case.
688 	 */
689 	unsigned long		bd_private;
690 
691 	/* The counter of freeze processes */
692 	int			bd_fsfreeze_count;
693 	/* Mutex for freeze */
694 	struct mutex		bd_fsfreeze_mutex;
695 };
696 
697 /*
698  * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
699  * radix trees
700  */
701 #define PAGECACHE_TAG_DIRTY	0
702 #define PAGECACHE_TAG_WRITEBACK	1
703 #define PAGECACHE_TAG_TOWRITE	2
704 
705 int mapping_tagged(struct address_space *mapping, int tag);
706 
707 /*
708  * Might pages of this file be mapped into userspace?
709  */
710 static inline int mapping_mapped(struct address_space *mapping)
711 {
712 	return	!prio_tree_empty(&mapping->i_mmap) ||
713 		!list_empty(&mapping->i_mmap_nonlinear);
714 }
715 
716 /*
717  * Might pages of this file have been modified in userspace?
718  * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
719  * marks vma as VM_SHARED if it is shared, and the file was opened for
720  * writing i.e. vma may be mprotected writable even if now readonly.
721  */
722 static inline int mapping_writably_mapped(struct address_space *mapping)
723 {
724 	return mapping->i_mmap_writable != 0;
725 }
726 
727 /*
728  * Use sequence counter to get consistent i_size on 32-bit processors.
729  */
730 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
731 #include <linux/seqlock.h>
732 #define __NEED_I_SIZE_ORDERED
733 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
734 #else
735 #define i_size_ordered_init(inode) do { } while (0)
736 #endif
737 
738 struct posix_acl;
739 #define ACL_NOT_CACHED ((void *)(-1))
740 
741 struct inode {
742 	/* RCU path lookup touches following: */
743 	umode_t			i_mode;
744 	uid_t			i_uid;
745 	gid_t			i_gid;
746 	const struct inode_operations	*i_op;
747 	struct super_block	*i_sb;
748 
749 	spinlock_t		i_lock;	/* i_blocks, i_bytes, maybe i_size */
750 	unsigned int		i_flags;
751 	unsigned long		i_state;
752 #ifdef CONFIG_SECURITY
753 	void			*i_security;
754 #endif
755 	struct mutex		i_mutex;
756 
757 
758 	unsigned long		dirtied_when;	/* jiffies of first dirtying */
759 
760 	struct hlist_node	i_hash;
761 	struct list_head	i_wb_list;	/* backing dev IO list */
762 	struct list_head	i_lru;		/* inode LRU list */
763 	struct list_head	i_sb_list;
764 	union {
765 		struct list_head	i_dentry;
766 		struct rcu_head		i_rcu;
767 	};
768 	unsigned long		i_ino;
769 	atomic_t		i_count;
770 	unsigned int		i_nlink;
771 	dev_t			i_rdev;
772 	unsigned int		i_blkbits;
773 	u64			i_version;
774 	loff_t			i_size;
775 #ifdef __NEED_I_SIZE_ORDERED
776 	seqcount_t		i_size_seqcount;
777 #endif
778 	struct timespec		i_atime;
779 	struct timespec		i_mtime;
780 	struct timespec		i_ctime;
781 	blkcnt_t		i_blocks;
782 	unsigned short          i_bytes;
783 	atomic_t		i_dio_count;
784 	const struct file_operations	*i_fop;	/* former ->i_op->default_file_ops */
785 	struct file_lock	*i_flock;
786 	struct address_space	*i_mapping;
787 	struct address_space	i_data;
788 #ifdef CONFIG_QUOTA
789 	struct dquot		*i_dquot[MAXQUOTAS];
790 #endif
791 	struct list_head	i_devices;
792 	union {
793 		struct pipe_inode_info	*i_pipe;
794 		struct block_device	*i_bdev;
795 		struct cdev		*i_cdev;
796 	};
797 
798 	__u32			i_generation;
799 
800 #ifdef CONFIG_FSNOTIFY
801 	__u32			i_fsnotify_mask; /* all events this inode cares about */
802 	struct hlist_head	i_fsnotify_marks;
803 #endif
804 
805 #ifdef CONFIG_IMA
806 	atomic_t		i_readcount; /* struct files open RO */
807 #endif
808 	atomic_t		i_writecount;
809 #ifdef CONFIG_FS_POSIX_ACL
810 	struct posix_acl	*i_acl;
811 	struct posix_acl	*i_default_acl;
812 #endif
813 	void			*i_private; /* fs or device private pointer */
814 };
815 
816 static inline int inode_unhashed(struct inode *inode)
817 {
818 	return hlist_unhashed(&inode->i_hash);
819 }
820 
821 /*
822  * inode->i_mutex nesting subclasses for the lock validator:
823  *
824  * 0: the object of the current VFS operation
825  * 1: parent
826  * 2: child/target
827  * 3: quota file
828  *
829  * The locking order between these classes is
830  * parent -> child -> normal -> xattr -> quota
831  */
832 enum inode_i_mutex_lock_class
833 {
834 	I_MUTEX_NORMAL,
835 	I_MUTEX_PARENT,
836 	I_MUTEX_CHILD,
837 	I_MUTEX_XATTR,
838 	I_MUTEX_QUOTA
839 };
840 
841 /*
842  * NOTE: in a 32bit arch with a preemptable kernel and
843  * an UP compile the i_size_read/write must be atomic
844  * with respect to the local cpu (unlike with preempt disabled),
845  * but they don't need to be atomic with respect to other cpus like in
846  * true SMP (so they need either to either locally disable irq around
847  * the read or for example on x86 they can be still implemented as a
848  * cmpxchg8b without the need of the lock prefix). For SMP compiles
849  * and 64bit archs it makes no difference if preempt is enabled or not.
850  */
851 static inline loff_t i_size_read(const struct inode *inode)
852 {
853 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
854 	loff_t i_size;
855 	unsigned int seq;
856 
857 	do {
858 		seq = read_seqcount_begin(&inode->i_size_seqcount);
859 		i_size = inode->i_size;
860 	} while (read_seqcount_retry(&inode->i_size_seqcount, seq));
861 	return i_size;
862 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
863 	loff_t i_size;
864 
865 	preempt_disable();
866 	i_size = inode->i_size;
867 	preempt_enable();
868 	return i_size;
869 #else
870 	return inode->i_size;
871 #endif
872 }
873 
874 /*
875  * NOTE: unlike i_size_read(), i_size_write() does need locking around it
876  * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
877  * can be lost, resulting in subsequent i_size_read() calls spinning forever.
878  */
879 static inline void i_size_write(struct inode *inode, loff_t i_size)
880 {
881 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
882 	write_seqcount_begin(&inode->i_size_seqcount);
883 	inode->i_size = i_size;
884 	write_seqcount_end(&inode->i_size_seqcount);
885 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
886 	preempt_disable();
887 	inode->i_size = i_size;
888 	preempt_enable();
889 #else
890 	inode->i_size = i_size;
891 #endif
892 }
893 
894 static inline unsigned iminor(const struct inode *inode)
895 {
896 	return MINOR(inode->i_rdev);
897 }
898 
899 static inline unsigned imajor(const struct inode *inode)
900 {
901 	return MAJOR(inode->i_rdev);
902 }
903 
904 extern struct block_device *I_BDEV(struct inode *inode);
905 
906 struct fown_struct {
907 	rwlock_t lock;          /* protects pid, uid, euid fields */
908 	struct pid *pid;	/* pid or -pgrp where SIGIO should be sent */
909 	enum pid_type pid_type;	/* Kind of process group SIGIO should be sent to */
910 	uid_t uid, euid;	/* uid/euid of process setting the owner */
911 	int signum;		/* posix.1b rt signal to be delivered on IO */
912 };
913 
914 /*
915  * Track a single file's readahead state
916  */
917 struct file_ra_state {
918 	pgoff_t start;			/* where readahead started */
919 	unsigned int size;		/* # of readahead pages */
920 	unsigned int async_size;	/* do asynchronous readahead when
921 					   there are only # of pages ahead */
922 
923 	unsigned int ra_pages;		/* Maximum readahead window */
924 	unsigned int mmap_miss;		/* Cache miss stat for mmap accesses */
925 	loff_t prev_pos;		/* Cache last read() position */
926 };
927 
928 /*
929  * Check if @index falls in the readahead windows.
930  */
931 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
932 {
933 	return (index >= ra->start &&
934 		index <  ra->start + ra->size);
935 }
936 
937 #define FILE_MNT_WRITE_TAKEN	1
938 #define FILE_MNT_WRITE_RELEASED	2
939 
940 struct file {
941 	/*
942 	 * fu_list becomes invalid after file_free is called and queued via
943 	 * fu_rcuhead for RCU freeing
944 	 */
945 	union {
946 		struct list_head	fu_list;
947 		struct rcu_head 	fu_rcuhead;
948 	} f_u;
949 	struct path		f_path;
950 #define f_dentry	f_path.dentry
951 #define f_vfsmnt	f_path.mnt
952 	const struct file_operations	*f_op;
953 	spinlock_t		f_lock;  /* f_ep_links, f_flags, no IRQ */
954 #ifdef CONFIG_SMP
955 	int			f_sb_list_cpu;
956 #endif
957 	atomic_long_t		f_count;
958 	unsigned int 		f_flags;
959 	fmode_t			f_mode;
960 	loff_t			f_pos;
961 	struct fown_struct	f_owner;
962 	const struct cred	*f_cred;
963 	struct file_ra_state	f_ra;
964 
965 	u64			f_version;
966 #ifdef CONFIG_SECURITY
967 	void			*f_security;
968 #endif
969 	/* needed for tty driver, and maybe others */
970 	void			*private_data;
971 
972 #ifdef CONFIG_EPOLL
973 	/* Used by fs/eventpoll.c to link all the hooks to this file */
974 	struct list_head	f_ep_links;
975 #endif /* #ifdef CONFIG_EPOLL */
976 	struct address_space	*f_mapping;
977 #ifdef CONFIG_DEBUG_WRITECOUNT
978 	unsigned long f_mnt_write_state;
979 #endif
980 };
981 
982 struct file_handle {
983 	__u32 handle_bytes;
984 	int handle_type;
985 	/* file identifier */
986 	unsigned char f_handle[0];
987 };
988 
989 #define get_file(x)	atomic_long_inc(&(x)->f_count)
990 #define fput_atomic(x)	atomic_long_add_unless(&(x)->f_count, -1, 1)
991 #define file_count(x)	atomic_long_read(&(x)->f_count)
992 
993 #ifdef CONFIG_DEBUG_WRITECOUNT
994 static inline void file_take_write(struct file *f)
995 {
996 	WARN_ON(f->f_mnt_write_state != 0);
997 	f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
998 }
999 static inline void file_release_write(struct file *f)
1000 {
1001 	f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
1002 }
1003 static inline void file_reset_write(struct file *f)
1004 {
1005 	f->f_mnt_write_state = 0;
1006 }
1007 static inline void file_check_state(struct file *f)
1008 {
1009 	/*
1010 	 * At this point, either both or neither of these bits
1011 	 * should be set.
1012 	 */
1013 	WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
1014 	WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
1015 }
1016 static inline int file_check_writeable(struct file *f)
1017 {
1018 	if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
1019 		return 0;
1020 	printk(KERN_WARNING "writeable file with no "
1021 			    "mnt_want_write()\n");
1022 	WARN_ON(1);
1023 	return -EINVAL;
1024 }
1025 #else /* !CONFIG_DEBUG_WRITECOUNT */
1026 static inline void file_take_write(struct file *filp) {}
1027 static inline void file_release_write(struct file *filp) {}
1028 static inline void file_reset_write(struct file *filp) {}
1029 static inline void file_check_state(struct file *filp) {}
1030 static inline int file_check_writeable(struct file *filp)
1031 {
1032 	return 0;
1033 }
1034 #endif /* CONFIG_DEBUG_WRITECOUNT */
1035 
1036 #define	MAX_NON_LFS	((1UL<<31) - 1)
1037 
1038 /* Page cache limit. The filesystems should put that into their s_maxbytes
1039    limits, otherwise bad things can happen in VM. */
1040 #if BITS_PER_LONG==32
1041 #define MAX_LFS_FILESIZE	(((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
1042 #elif BITS_PER_LONG==64
1043 #define MAX_LFS_FILESIZE 	0x7fffffffffffffffUL
1044 #endif
1045 
1046 #define FL_POSIX	1
1047 #define FL_FLOCK	2
1048 #define FL_ACCESS	8	/* not trying to lock, just looking */
1049 #define FL_EXISTS	16	/* when unlocking, test for existence */
1050 #define FL_LEASE	32	/* lease held on this file */
1051 #define FL_CLOSE	64	/* unlock on close */
1052 #define FL_SLEEP	128	/* A blocking lock */
1053 
1054 /*
1055  * Special return value from posix_lock_file() and vfs_lock_file() for
1056  * asynchronous locking.
1057  */
1058 #define FILE_LOCK_DEFERRED 1
1059 
1060 /*
1061  * The POSIX file lock owner is determined by
1062  * the "struct files_struct" in the thread group
1063  * (or NULL for no owner - BSD locks).
1064  *
1065  * Lockd stuffs a "host" pointer into this.
1066  */
1067 typedef struct files_struct *fl_owner_t;
1068 
1069 struct file_lock_operations {
1070 	void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1071 	void (*fl_release_private)(struct file_lock *);
1072 };
1073 
1074 struct lock_manager_operations {
1075 	int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
1076 	void (*lm_notify)(struct file_lock *);	/* unblock callback */
1077 	int (*lm_grant)(struct file_lock *, struct file_lock *, int);
1078 	void (*lm_release_private)(struct file_lock *);
1079 	void (*lm_break)(struct file_lock *);
1080 	int (*lm_change)(struct file_lock **, int);
1081 };
1082 
1083 struct lock_manager {
1084 	struct list_head list;
1085 };
1086 
1087 void locks_start_grace(struct lock_manager *);
1088 void locks_end_grace(struct lock_manager *);
1089 int locks_in_grace(void);
1090 
1091 /* that will die - we need it for nfs_lock_info */
1092 #include <linux/nfs_fs_i.h>
1093 
1094 struct file_lock {
1095 	struct file_lock *fl_next;	/* singly linked list for this inode  */
1096 	struct list_head fl_link;	/* doubly linked list of all locks */
1097 	struct list_head fl_block;	/* circular list of blocked processes */
1098 	fl_owner_t fl_owner;
1099 	unsigned char fl_flags;
1100 	unsigned char fl_type;
1101 	unsigned int fl_pid;
1102 	struct pid *fl_nspid;
1103 	wait_queue_head_t fl_wait;
1104 	struct file *fl_file;
1105 	loff_t fl_start;
1106 	loff_t fl_end;
1107 
1108 	struct fasync_struct *	fl_fasync; /* for lease break notifications */
1109 	unsigned long fl_break_time;	/* for nonblocking lease breaks */
1110 
1111 	const struct file_lock_operations *fl_ops;	/* Callbacks for filesystems */
1112 	const struct lock_manager_operations *fl_lmops;	/* Callbacks for lockmanagers */
1113 	union {
1114 		struct nfs_lock_info	nfs_fl;
1115 		struct nfs4_lock_info	nfs4_fl;
1116 		struct {
1117 			struct list_head link;	/* link in AFS vnode's pending_locks list */
1118 			int state;		/* state of grant or error if -ve */
1119 		} afs;
1120 	} fl_u;
1121 };
1122 
1123 /* The following constant reflects the upper bound of the file/locking space */
1124 #ifndef OFFSET_MAX
1125 #define INT_LIMIT(x)	(~((x)1 << (sizeof(x)*8 - 1)))
1126 #define OFFSET_MAX	INT_LIMIT(loff_t)
1127 #define OFFT_OFFSET_MAX	INT_LIMIT(off_t)
1128 #endif
1129 
1130 #include <linux/fcntl.h>
1131 
1132 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1133 
1134 #ifdef CONFIG_FILE_LOCKING
1135 extern int fcntl_getlk(struct file *, struct flock __user *);
1136 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1137 			struct flock __user *);
1138 
1139 #if BITS_PER_LONG == 32
1140 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
1141 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1142 			struct flock64 __user *);
1143 #endif
1144 
1145 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1146 extern int fcntl_getlease(struct file *filp);
1147 
1148 /* fs/locks.c */
1149 void locks_free_lock(struct file_lock *fl);
1150 extern void locks_init_lock(struct file_lock *);
1151 extern struct file_lock * locks_alloc_lock(void);
1152 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1153 extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
1154 extern void locks_remove_posix(struct file *, fl_owner_t);
1155 extern void locks_remove_flock(struct file *);
1156 extern void locks_release_private(struct file_lock *);
1157 extern void posix_test_lock(struct file *, struct file_lock *);
1158 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1159 extern int posix_lock_file_wait(struct file *, struct file_lock *);
1160 extern int posix_unblock_lock(struct file *, struct file_lock *);
1161 extern int vfs_test_lock(struct file *, struct file_lock *);
1162 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1163 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1164 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1165 extern int __break_lease(struct inode *inode, unsigned int flags);
1166 extern void lease_get_mtime(struct inode *, struct timespec *time);
1167 extern int generic_setlease(struct file *, long, struct file_lock **);
1168 extern int vfs_setlease(struct file *, long, struct file_lock **);
1169 extern int lease_modify(struct file_lock **, int);
1170 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1171 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1172 extern void lock_flocks(void);
1173 extern void unlock_flocks(void);
1174 #else /* !CONFIG_FILE_LOCKING */
1175 static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1176 {
1177 	return -EINVAL;
1178 }
1179 
1180 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1181 			      unsigned int cmd, struct flock __user *user)
1182 {
1183 	return -EACCES;
1184 }
1185 
1186 #if BITS_PER_LONG == 32
1187 static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1188 {
1189 	return -EINVAL;
1190 }
1191 
1192 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1193 				unsigned int cmd, struct flock64 __user *user)
1194 {
1195 	return -EACCES;
1196 }
1197 #endif
1198 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1199 {
1200 	return 0;
1201 }
1202 
1203 static inline int fcntl_getlease(struct file *filp)
1204 {
1205 	return 0;
1206 }
1207 
1208 static inline void locks_init_lock(struct file_lock *fl)
1209 {
1210 	return;
1211 }
1212 
1213 static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1214 {
1215 	return;
1216 }
1217 
1218 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1219 {
1220 	return;
1221 }
1222 
1223 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1224 {
1225 	return;
1226 }
1227 
1228 static inline void locks_remove_flock(struct file *filp)
1229 {
1230 	return;
1231 }
1232 
1233 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1234 {
1235 	return;
1236 }
1237 
1238 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1239 				  struct file_lock *conflock)
1240 {
1241 	return -ENOLCK;
1242 }
1243 
1244 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1245 {
1246 	return -ENOLCK;
1247 }
1248 
1249 static inline int posix_unblock_lock(struct file *filp,
1250 				     struct file_lock *waiter)
1251 {
1252 	return -ENOENT;
1253 }
1254 
1255 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1256 {
1257 	return 0;
1258 }
1259 
1260 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1261 				struct file_lock *fl, struct file_lock *conf)
1262 {
1263 	return -ENOLCK;
1264 }
1265 
1266 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1267 {
1268 	return 0;
1269 }
1270 
1271 static inline int flock_lock_file_wait(struct file *filp,
1272 				       struct file_lock *request)
1273 {
1274 	return -ENOLCK;
1275 }
1276 
1277 static inline int __break_lease(struct inode *inode, unsigned int mode)
1278 {
1279 	return 0;
1280 }
1281 
1282 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1283 {
1284 	return;
1285 }
1286 
1287 static inline int generic_setlease(struct file *filp, long arg,
1288 				    struct file_lock **flp)
1289 {
1290 	return -EINVAL;
1291 }
1292 
1293 static inline int vfs_setlease(struct file *filp, long arg,
1294 			       struct file_lock **lease)
1295 {
1296 	return -EINVAL;
1297 }
1298 
1299 static inline int lease_modify(struct file_lock **before, int arg)
1300 {
1301 	return -EINVAL;
1302 }
1303 
1304 static inline int lock_may_read(struct inode *inode, loff_t start,
1305 				unsigned long len)
1306 {
1307 	return 1;
1308 }
1309 
1310 static inline int lock_may_write(struct inode *inode, loff_t start,
1311 				 unsigned long len)
1312 {
1313 	return 1;
1314 }
1315 
1316 static inline void lock_flocks(void)
1317 {
1318 }
1319 
1320 static inline void unlock_flocks(void)
1321 {
1322 }
1323 
1324 #endif /* !CONFIG_FILE_LOCKING */
1325 
1326 
1327 struct fasync_struct {
1328 	spinlock_t		fa_lock;
1329 	int			magic;
1330 	int			fa_fd;
1331 	struct fasync_struct	*fa_next; /* singly linked list */
1332 	struct file		*fa_file;
1333 	struct rcu_head		fa_rcu;
1334 };
1335 
1336 #define FASYNC_MAGIC 0x4601
1337 
1338 /* SMP safe fasync helpers: */
1339 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1340 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1341 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1342 extern struct fasync_struct *fasync_alloc(void);
1343 extern void fasync_free(struct fasync_struct *);
1344 
1345 /* can be called from interrupts */
1346 extern void kill_fasync(struct fasync_struct **, int, int);
1347 
1348 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1349 extern int f_setown(struct file *filp, unsigned long arg, int force);
1350 extern void f_delown(struct file *filp);
1351 extern pid_t f_getown(struct file *filp);
1352 extern int send_sigurg(struct fown_struct *fown);
1353 
1354 /*
1355  *	Umount options
1356  */
1357 
1358 #define MNT_FORCE	0x00000001	/* Attempt to forcibily umount */
1359 #define MNT_DETACH	0x00000002	/* Just detach from the tree */
1360 #define MNT_EXPIRE	0x00000004	/* Mark for expiry */
1361 #define UMOUNT_NOFOLLOW	0x00000008	/* Don't follow symlink on umount */
1362 #define UMOUNT_UNUSED	0x80000000	/* Flag guaranteed to be unused */
1363 
1364 extern struct list_head super_blocks;
1365 extern spinlock_t sb_lock;
1366 
1367 struct super_block {
1368 	struct list_head	s_list;		/* Keep this first */
1369 	dev_t			s_dev;		/* search index; _not_ kdev_t */
1370 	unsigned char		s_dirt;
1371 	unsigned char		s_blocksize_bits;
1372 	unsigned long		s_blocksize;
1373 	loff_t			s_maxbytes;	/* Max file size */
1374 	struct file_system_type	*s_type;
1375 	const struct super_operations	*s_op;
1376 	const struct dquot_operations	*dq_op;
1377 	const struct quotactl_ops	*s_qcop;
1378 	const struct export_operations *s_export_op;
1379 	unsigned long		s_flags;
1380 	unsigned long		s_magic;
1381 	struct dentry		*s_root;
1382 	struct rw_semaphore	s_umount;
1383 	struct mutex		s_lock;
1384 	int			s_count;
1385 	atomic_t		s_active;
1386 #ifdef CONFIG_SECURITY
1387 	void                    *s_security;
1388 #endif
1389 	const struct xattr_handler **s_xattr;
1390 
1391 	struct list_head	s_inodes;	/* all inodes */
1392 	struct hlist_bl_head	s_anon;		/* anonymous dentries for (nfs) exporting */
1393 #ifdef CONFIG_SMP
1394 	struct list_head __percpu *s_files;
1395 #else
1396 	struct list_head	s_files;
1397 #endif
1398 	/* s_dentry_lru, s_nr_dentry_unused protected by dcache.c lru locks */
1399 	struct list_head	s_dentry_lru;	/* unused dentry lru */
1400 	int			s_nr_dentry_unused;	/* # of dentry on lru */
1401 
1402 	/* s_inode_lru_lock protects s_inode_lru and s_nr_inodes_unused */
1403 	spinlock_t		s_inode_lru_lock ____cacheline_aligned_in_smp;
1404 	struct list_head	s_inode_lru;		/* unused inode lru */
1405 	int			s_nr_inodes_unused;	/* # of inodes on lru */
1406 
1407 	struct block_device	*s_bdev;
1408 	struct backing_dev_info *s_bdi;
1409 	struct mtd_info		*s_mtd;
1410 	struct list_head	s_instances;
1411 	struct quota_info	s_dquot;	/* Diskquota specific options */
1412 
1413 	int			s_frozen;
1414 	wait_queue_head_t	s_wait_unfrozen;
1415 
1416 	char s_id[32];				/* Informational name */
1417 	u8 s_uuid[16];				/* UUID */
1418 
1419 	void 			*s_fs_info;	/* Filesystem private info */
1420 	fmode_t			s_mode;
1421 
1422 	/* Granularity of c/m/atime in ns.
1423 	   Cannot be worse than a second */
1424 	u32		   s_time_gran;
1425 
1426 	/*
1427 	 * The next field is for VFS *only*. No filesystems have any business
1428 	 * even looking at it. You had been warned.
1429 	 */
1430 	struct mutex s_vfs_rename_mutex;	/* Kludge */
1431 
1432 	/*
1433 	 * Filesystem subtype.  If non-empty the filesystem type field
1434 	 * in /proc/mounts will be "type.subtype"
1435 	 */
1436 	char *s_subtype;
1437 
1438 	/*
1439 	 * Saved mount options for lazy filesystems using
1440 	 * generic_show_options()
1441 	 */
1442 	char __rcu *s_options;
1443 	const struct dentry_operations *s_d_op; /* default d_op for dentries */
1444 
1445 	/*
1446 	 * Saved pool identifier for cleancache (-1 means none)
1447 	 */
1448 	int cleancache_poolid;
1449 
1450 	struct shrinker s_shrink;	/* per-sb shrinker handle */
1451 };
1452 
1453 /* superblock cache pruning functions */
1454 extern void prune_icache_sb(struct super_block *sb, int nr_to_scan);
1455 extern void prune_dcache_sb(struct super_block *sb, int nr_to_scan);
1456 
1457 extern struct timespec current_fs_time(struct super_block *sb);
1458 
1459 /*
1460  * Snapshotting support.
1461  */
1462 enum {
1463 	SB_UNFROZEN = 0,
1464 	SB_FREEZE_WRITE	= 1,
1465 	SB_FREEZE_TRANS = 2,
1466 };
1467 
1468 #define vfs_check_frozen(sb, level) \
1469 	wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
1470 
1471 /*
1472  * until VFS tracks user namespaces for inodes, just make all files
1473  * belong to init_user_ns
1474  */
1475 extern struct user_namespace init_user_ns;
1476 #define inode_userns(inode) (&init_user_ns)
1477 extern bool inode_owner_or_capable(const struct inode *inode);
1478 
1479 /* not quite ready to be deprecated, but... */
1480 extern void lock_super(struct super_block *);
1481 extern void unlock_super(struct super_block *);
1482 
1483 /*
1484  * VFS helper functions..
1485  */
1486 extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
1487 extern int vfs_mkdir(struct inode *, struct dentry *, int);
1488 extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
1489 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1490 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1491 extern int vfs_rmdir(struct inode *, struct dentry *);
1492 extern int vfs_unlink(struct inode *, struct dentry *);
1493 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1494 
1495 /*
1496  * VFS dentry helper functions.
1497  */
1498 extern void dentry_unhash(struct dentry *dentry);
1499 
1500 /*
1501  * VFS file helper functions.
1502  */
1503 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1504 			mode_t mode);
1505 /*
1506  * VFS FS_IOC_FIEMAP helper definitions.
1507  */
1508 struct fiemap_extent_info {
1509 	unsigned int fi_flags;		/* Flags as passed from user */
1510 	unsigned int fi_extents_mapped;	/* Number of mapped extents */
1511 	unsigned int fi_extents_max;	/* Size of fiemap_extent array */
1512 	struct fiemap_extent __user *fi_extents_start; /* Start of
1513 							fiemap_extent array */
1514 };
1515 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1516 			    u64 phys, u64 len, u32 flags);
1517 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1518 
1519 /*
1520  * File types
1521  *
1522  * NOTE! These match bits 12..15 of stat.st_mode
1523  * (ie "(i_mode >> 12) & 15").
1524  */
1525 #define DT_UNKNOWN	0
1526 #define DT_FIFO		1
1527 #define DT_CHR		2
1528 #define DT_DIR		4
1529 #define DT_BLK		6
1530 #define DT_REG		8
1531 #define DT_LNK		10
1532 #define DT_SOCK		12
1533 #define DT_WHT		14
1534 
1535 /*
1536  * This is the "filldir" function type, used by readdir() to let
1537  * the kernel specify what kind of dirent layout it wants to have.
1538  * This allows the kernel to read directories into kernel space or
1539  * to have different dirent layouts depending on the binary type.
1540  */
1541 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1542 struct block_device_operations;
1543 
1544 /* These macros are for out of kernel modules to test that
1545  * the kernel supports the unlocked_ioctl and compat_ioctl
1546  * fields in struct file_operations. */
1547 #define HAVE_COMPAT_IOCTL 1
1548 #define HAVE_UNLOCKED_IOCTL 1
1549 
1550 struct file_operations {
1551 	struct module *owner;
1552 	loff_t (*llseek) (struct file *, loff_t, int);
1553 	ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1554 	ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1555 	ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1556 	ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1557 	int (*readdir) (struct file *, void *, filldir_t);
1558 	unsigned int (*poll) (struct file *, struct poll_table_struct *);
1559 	long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1560 	long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1561 	int (*mmap) (struct file *, struct vm_area_struct *);
1562 	int (*open) (struct inode *, struct file *);
1563 	int (*flush) (struct file *, fl_owner_t id);
1564 	int (*release) (struct inode *, struct file *);
1565 	int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1566 	int (*aio_fsync) (struct kiocb *, int datasync);
1567 	int (*fasync) (int, struct file *, int);
1568 	int (*lock) (struct file *, int, struct file_lock *);
1569 	ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1570 	unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1571 	int (*check_flags)(int);
1572 	int (*flock) (struct file *, int, struct file_lock *);
1573 	ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1574 	ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1575 	int (*setlease)(struct file *, long, struct file_lock **);
1576 	long (*fallocate)(struct file *file, int mode, loff_t offset,
1577 			  loff_t len);
1578 };
1579 
1580 struct inode_operations {
1581 	struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1582 	void * (*follow_link) (struct dentry *, struct nameidata *);
1583 	int (*permission) (struct inode *, int);
1584 	struct posix_acl * (*get_acl)(struct inode *, int);
1585 
1586 	int (*readlink) (struct dentry *, char __user *,int);
1587 	void (*put_link) (struct dentry *, struct nameidata *, void *);
1588 
1589 	int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1590 	int (*link) (struct dentry *,struct inode *,struct dentry *);
1591 	int (*unlink) (struct inode *,struct dentry *);
1592 	int (*symlink) (struct inode *,struct dentry *,const char *);
1593 	int (*mkdir) (struct inode *,struct dentry *,int);
1594 	int (*rmdir) (struct inode *,struct dentry *);
1595 	int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1596 	int (*rename) (struct inode *, struct dentry *,
1597 			struct inode *, struct dentry *);
1598 	void (*truncate) (struct inode *);
1599 	int (*setattr) (struct dentry *, struct iattr *);
1600 	int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1601 	int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1602 	ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1603 	ssize_t (*listxattr) (struct dentry *, char *, size_t);
1604 	int (*removexattr) (struct dentry *, const char *);
1605 	void (*truncate_range)(struct inode *, loff_t, loff_t);
1606 	int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1607 		      u64 len);
1608 } ____cacheline_aligned;
1609 
1610 struct seq_file;
1611 
1612 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1613 				unsigned long nr_segs, unsigned long fast_segs,
1614 				struct iovec *fast_pointer,
1615 				struct iovec **ret_pointer);
1616 
1617 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1618 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1619 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1620 		unsigned long, loff_t *);
1621 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1622 		unsigned long, loff_t *);
1623 
1624 struct super_operations {
1625    	struct inode *(*alloc_inode)(struct super_block *sb);
1626 	void (*destroy_inode)(struct inode *);
1627 
1628    	void (*dirty_inode) (struct inode *, int flags);
1629 	int (*write_inode) (struct inode *, struct writeback_control *wbc);
1630 	int (*drop_inode) (struct inode *);
1631 	void (*evict_inode) (struct inode *);
1632 	void (*put_super) (struct super_block *);
1633 	void (*write_super) (struct super_block *);
1634 	int (*sync_fs)(struct super_block *sb, int wait);
1635 	int (*freeze_fs) (struct super_block *);
1636 	int (*unfreeze_fs) (struct super_block *);
1637 	int (*statfs) (struct dentry *, struct kstatfs *);
1638 	int (*remount_fs) (struct super_block *, int *, char *);
1639 	void (*umount_begin) (struct super_block *);
1640 
1641 	int (*show_options)(struct seq_file *, struct vfsmount *);
1642 	int (*show_devname)(struct seq_file *, struct vfsmount *);
1643 	int (*show_path)(struct seq_file *, struct vfsmount *);
1644 	int (*show_stats)(struct seq_file *, struct vfsmount *);
1645 #ifdef CONFIG_QUOTA
1646 	ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1647 	ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1648 #endif
1649 	int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1650 	int (*nr_cached_objects)(struct super_block *);
1651 	void (*free_cached_objects)(struct super_block *, int);
1652 };
1653 
1654 /*
1655  * Inode state bits.  Protected by inode->i_lock
1656  *
1657  * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1658  * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1659  *
1660  * Four bits define the lifetime of an inode.  Initially, inodes are I_NEW,
1661  * until that flag is cleared.  I_WILL_FREE, I_FREEING and I_CLEAR are set at
1662  * various stages of removing an inode.
1663  *
1664  * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1665  *
1666  * I_DIRTY_SYNC		Inode is dirty, but doesn't have to be written on
1667  *			fdatasync().  i_atime is the usual cause.
1668  * I_DIRTY_DATASYNC	Data-related inode changes pending. We keep track of
1669  *			these changes separately from I_DIRTY_SYNC so that we
1670  *			don't have to write inode on fdatasync() when only
1671  *			mtime has changed in it.
1672  * I_DIRTY_PAGES	Inode has dirty pages.  Inode itself may be clean.
1673  * I_NEW		Serves as both a mutex and completion notification.
1674  *			New inodes set I_NEW.  If two processes both create
1675  *			the same inode, one of them will release its inode and
1676  *			wait for I_NEW to be released before returning.
1677  *			Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1678  *			also cause waiting on I_NEW, without I_NEW actually
1679  *			being set.  find_inode() uses this to prevent returning
1680  *			nearly-dead inodes.
1681  * I_WILL_FREE		Must be set when calling write_inode_now() if i_count
1682  *			is zero.  I_FREEING must be set when I_WILL_FREE is
1683  *			cleared.
1684  * I_FREEING		Set when inode is about to be freed but still has dirty
1685  *			pages or buffers attached or the inode itself is still
1686  *			dirty.
1687  * I_CLEAR		Added by end_writeback().  In this state the inode is clean
1688  *			and can be destroyed.  Inode keeps I_FREEING.
1689  *
1690  *			Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1691  *			prohibited for many purposes.  iget() must wait for
1692  *			the inode to be completely released, then create it
1693  *			anew.  Other functions will just ignore such inodes,
1694  *			if appropriate.  I_NEW is used for waiting.
1695  *
1696  * I_SYNC		Synchonized write of dirty inode data.  The bits is
1697  *			set during data writeback, and cleared with a wakeup
1698  *			on the bit address once it is done.
1699  *
1700  * I_REFERENCED		Marks the inode as recently references on the LRU list.
1701  *
1702  * I_DIO_WAKEUP		Never set.  Only used as a key for wait_on_bit().
1703  *
1704  * Q: What is the difference between I_WILL_FREE and I_FREEING?
1705  */
1706 #define I_DIRTY_SYNC		(1 << 0)
1707 #define I_DIRTY_DATASYNC	(1 << 1)
1708 #define I_DIRTY_PAGES		(1 << 2)
1709 #define __I_NEW			3
1710 #define I_NEW			(1 << __I_NEW)
1711 #define I_WILL_FREE		(1 << 4)
1712 #define I_FREEING		(1 << 5)
1713 #define I_CLEAR			(1 << 6)
1714 #define __I_SYNC		7
1715 #define I_SYNC			(1 << __I_SYNC)
1716 #define I_REFERENCED		(1 << 8)
1717 #define __I_DIO_WAKEUP		9
1718 #define I_DIO_WAKEUP		(1 << I_DIO_WAKEUP)
1719 
1720 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1721 
1722 extern void __mark_inode_dirty(struct inode *, int);
1723 static inline void mark_inode_dirty(struct inode *inode)
1724 {
1725 	__mark_inode_dirty(inode, I_DIRTY);
1726 }
1727 
1728 static inline void mark_inode_dirty_sync(struct inode *inode)
1729 {
1730 	__mark_inode_dirty(inode, I_DIRTY_SYNC);
1731 }
1732 
1733 /**
1734  * inc_nlink - directly increment an inode's link count
1735  * @inode: inode
1736  *
1737  * This is a low-level filesystem helper to replace any
1738  * direct filesystem manipulation of i_nlink.  Currently,
1739  * it is only here for parity with dec_nlink().
1740  */
1741 static inline void inc_nlink(struct inode *inode)
1742 {
1743 	inode->i_nlink++;
1744 }
1745 
1746 static inline void inode_inc_link_count(struct inode *inode)
1747 {
1748 	inc_nlink(inode);
1749 	mark_inode_dirty(inode);
1750 }
1751 
1752 /**
1753  * drop_nlink - directly drop an inode's link count
1754  * @inode: inode
1755  *
1756  * This is a low-level filesystem helper to replace any
1757  * direct filesystem manipulation of i_nlink.  In cases
1758  * where we are attempting to track writes to the
1759  * filesystem, a decrement to zero means an imminent
1760  * write when the file is truncated and actually unlinked
1761  * on the filesystem.
1762  */
1763 static inline void drop_nlink(struct inode *inode)
1764 {
1765 	inode->i_nlink--;
1766 }
1767 
1768 /**
1769  * clear_nlink - directly zero an inode's link count
1770  * @inode: inode
1771  *
1772  * This is a low-level filesystem helper to replace any
1773  * direct filesystem manipulation of i_nlink.  See
1774  * drop_nlink() for why we care about i_nlink hitting zero.
1775  */
1776 static inline void clear_nlink(struct inode *inode)
1777 {
1778 	inode->i_nlink = 0;
1779 }
1780 
1781 static inline void inode_dec_link_count(struct inode *inode)
1782 {
1783 	drop_nlink(inode);
1784 	mark_inode_dirty(inode);
1785 }
1786 
1787 /**
1788  * inode_inc_iversion - increments i_version
1789  * @inode: inode that need to be updated
1790  *
1791  * Every time the inode is modified, the i_version field will be incremented.
1792  * The filesystem has to be mounted with i_version flag
1793  */
1794 
1795 static inline void inode_inc_iversion(struct inode *inode)
1796 {
1797        spin_lock(&inode->i_lock);
1798        inode->i_version++;
1799        spin_unlock(&inode->i_lock);
1800 }
1801 
1802 extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
1803 static inline void file_accessed(struct file *file)
1804 {
1805 	if (!(file->f_flags & O_NOATIME))
1806 		touch_atime(file->f_path.mnt, file->f_path.dentry);
1807 }
1808 
1809 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1810 int sync_inode_metadata(struct inode *inode, int wait);
1811 
1812 struct file_system_type {
1813 	const char *name;
1814 	int fs_flags;
1815 	struct dentry *(*mount) (struct file_system_type *, int,
1816 		       const char *, void *);
1817 	void (*kill_sb) (struct super_block *);
1818 	struct module *owner;
1819 	struct file_system_type * next;
1820 	struct list_head fs_supers;
1821 
1822 	struct lock_class_key s_lock_key;
1823 	struct lock_class_key s_umount_key;
1824 	struct lock_class_key s_vfs_rename_key;
1825 
1826 	struct lock_class_key i_lock_key;
1827 	struct lock_class_key i_mutex_key;
1828 	struct lock_class_key i_mutex_dir_key;
1829 };
1830 
1831 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1832 	void *data, int (*fill_super)(struct super_block *, void *, int));
1833 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1834 	int flags, const char *dev_name, void *data,
1835 	int (*fill_super)(struct super_block *, void *, int));
1836 extern struct dentry *mount_single(struct file_system_type *fs_type,
1837 	int flags, void *data,
1838 	int (*fill_super)(struct super_block *, void *, int));
1839 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1840 	int flags, void *data,
1841 	int (*fill_super)(struct super_block *, void *, int));
1842 void generic_shutdown_super(struct super_block *sb);
1843 void kill_block_super(struct super_block *sb);
1844 void kill_anon_super(struct super_block *sb);
1845 void kill_litter_super(struct super_block *sb);
1846 void deactivate_super(struct super_block *sb);
1847 void deactivate_locked_super(struct super_block *sb);
1848 int set_anon_super(struct super_block *s, void *data);
1849 int get_anon_bdev(dev_t *);
1850 void free_anon_bdev(dev_t);
1851 struct super_block *sget(struct file_system_type *type,
1852 			int (*test)(struct super_block *,void *),
1853 			int (*set)(struct super_block *,void *),
1854 			void *data);
1855 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1856 	const struct super_operations *ops,
1857 	const struct dentry_operations *dops,
1858 	unsigned long);
1859 
1860 static inline void sb_mark_dirty(struct super_block *sb)
1861 {
1862 	sb->s_dirt = 1;
1863 }
1864 static inline void sb_mark_clean(struct super_block *sb)
1865 {
1866 	sb->s_dirt = 0;
1867 }
1868 static inline int sb_is_dirty(struct super_block *sb)
1869 {
1870 	return sb->s_dirt;
1871 }
1872 
1873 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1874 #define fops_get(fops) \
1875 	(((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1876 #define fops_put(fops) \
1877 	do { if (fops) module_put((fops)->owner); } while(0)
1878 
1879 extern int register_filesystem(struct file_system_type *);
1880 extern int unregister_filesystem(struct file_system_type *);
1881 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1882 #define kern_mount(type) kern_mount_data(type, NULL)
1883 extern void kern_unmount(struct vfsmount *mnt);
1884 extern int may_umount_tree(struct vfsmount *);
1885 extern int may_umount(struct vfsmount *);
1886 extern long do_mount(char *, char *, char *, unsigned long, void *);
1887 extern struct vfsmount *collect_mounts(struct path *);
1888 extern void drop_collected_mounts(struct vfsmount *);
1889 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1890 			  struct vfsmount *);
1891 extern int vfs_statfs(struct path *, struct kstatfs *);
1892 extern int user_statfs(const char __user *, struct kstatfs *);
1893 extern int fd_statfs(int, struct kstatfs *);
1894 extern int statfs_by_dentry(struct dentry *, struct kstatfs *);
1895 extern int freeze_super(struct super_block *super);
1896 extern int thaw_super(struct super_block *super);
1897 
1898 extern int current_umask(void);
1899 
1900 /* /sys/fs */
1901 extern struct kobject *fs_kobj;
1902 
1903 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1904 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1905 
1906 #define FLOCK_VERIFY_READ  1
1907 #define FLOCK_VERIFY_WRITE 2
1908 
1909 #ifdef CONFIG_FILE_LOCKING
1910 extern int locks_mandatory_locked(struct inode *);
1911 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1912 
1913 /*
1914  * Candidates for mandatory locking have the setgid bit set
1915  * but no group execute bit -  an otherwise meaningless combination.
1916  */
1917 
1918 static inline int __mandatory_lock(struct inode *ino)
1919 {
1920 	return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1921 }
1922 
1923 /*
1924  * ... and these candidates should be on MS_MANDLOCK mounted fs,
1925  * otherwise these will be advisory locks
1926  */
1927 
1928 static inline int mandatory_lock(struct inode *ino)
1929 {
1930 	return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1931 }
1932 
1933 static inline int locks_verify_locked(struct inode *inode)
1934 {
1935 	if (mandatory_lock(inode))
1936 		return locks_mandatory_locked(inode);
1937 	return 0;
1938 }
1939 
1940 static inline int locks_verify_truncate(struct inode *inode,
1941 				    struct file *filp,
1942 				    loff_t size)
1943 {
1944 	if (inode->i_flock && mandatory_lock(inode))
1945 		return locks_mandatory_area(
1946 			FLOCK_VERIFY_WRITE, inode, filp,
1947 			size < inode->i_size ? size : inode->i_size,
1948 			(size < inode->i_size ? inode->i_size - size
1949 			 : size - inode->i_size)
1950 		);
1951 	return 0;
1952 }
1953 
1954 static inline int break_lease(struct inode *inode, unsigned int mode)
1955 {
1956 	if (inode->i_flock)
1957 		return __break_lease(inode, mode);
1958 	return 0;
1959 }
1960 #else /* !CONFIG_FILE_LOCKING */
1961 static inline int locks_mandatory_locked(struct inode *inode)
1962 {
1963 	return 0;
1964 }
1965 
1966 static inline int locks_mandatory_area(int rw, struct inode *inode,
1967 				       struct file *filp, loff_t offset,
1968 				       size_t count)
1969 {
1970 	return 0;
1971 }
1972 
1973 static inline int __mandatory_lock(struct inode *inode)
1974 {
1975 	return 0;
1976 }
1977 
1978 static inline int mandatory_lock(struct inode *inode)
1979 {
1980 	return 0;
1981 }
1982 
1983 static inline int locks_verify_locked(struct inode *inode)
1984 {
1985 	return 0;
1986 }
1987 
1988 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
1989 					size_t size)
1990 {
1991 	return 0;
1992 }
1993 
1994 static inline int break_lease(struct inode *inode, unsigned int mode)
1995 {
1996 	return 0;
1997 }
1998 
1999 #endif /* CONFIG_FILE_LOCKING */
2000 
2001 /* fs/open.c */
2002 
2003 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2004 		       struct file *filp);
2005 extern int do_fallocate(struct file *file, int mode, loff_t offset,
2006 			loff_t len);
2007 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2008 			int mode);
2009 extern struct file *filp_open(const char *, int, int);
2010 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2011 				   const char *, int);
2012 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int,
2013 				 const struct cred *);
2014 extern int filp_close(struct file *, fl_owner_t id);
2015 extern char * getname(const char __user *);
2016 
2017 /* fs/ioctl.c */
2018 
2019 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2020 
2021 /* fs/dcache.c */
2022 extern void __init vfs_caches_init_early(void);
2023 extern void __init vfs_caches_init(unsigned long);
2024 
2025 extern struct kmem_cache *names_cachep;
2026 
2027 #define __getname_gfp(gfp)	kmem_cache_alloc(names_cachep, (gfp))
2028 #define __getname()		__getname_gfp(GFP_KERNEL)
2029 #define __putname(name)		kmem_cache_free(names_cachep, (void *)(name))
2030 #ifndef CONFIG_AUDITSYSCALL
2031 #define putname(name)   __putname(name)
2032 #else
2033 extern void putname(const char *name);
2034 #endif
2035 
2036 #ifdef CONFIG_BLOCK
2037 extern int register_blkdev(unsigned int, const char *);
2038 extern void unregister_blkdev(unsigned int, const char *);
2039 extern struct block_device *bdget(dev_t);
2040 extern struct block_device *bdgrab(struct block_device *bdev);
2041 extern void bd_set_size(struct block_device *, loff_t size);
2042 extern void bd_forget(struct inode *inode);
2043 extern void bdput(struct block_device *);
2044 extern void invalidate_bdev(struct block_device *);
2045 extern int sync_blockdev(struct block_device *bdev);
2046 extern struct super_block *freeze_bdev(struct block_device *);
2047 extern void emergency_thaw_all(void);
2048 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2049 extern int fsync_bdev(struct block_device *);
2050 #else
2051 static inline void bd_forget(struct inode *inode) {}
2052 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2053 static inline void invalidate_bdev(struct block_device *bdev) {}
2054 
2055 static inline struct super_block *freeze_bdev(struct block_device *sb)
2056 {
2057 	return NULL;
2058 }
2059 
2060 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2061 {
2062 	return 0;
2063 }
2064 #endif
2065 extern int sync_filesystem(struct super_block *);
2066 extern const struct file_operations def_blk_fops;
2067 extern const struct file_operations def_chr_fops;
2068 extern const struct file_operations bad_sock_fops;
2069 extern const struct file_operations def_fifo_fops;
2070 #ifdef CONFIG_BLOCK
2071 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2072 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2073 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2074 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2075 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2076 					       void *holder);
2077 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2078 					      void *holder);
2079 extern int blkdev_put(struct block_device *bdev, fmode_t mode);
2080 #ifdef CONFIG_SYSFS
2081 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2082 extern void bd_unlink_disk_holder(struct block_device *bdev,
2083 				  struct gendisk *disk);
2084 #else
2085 static inline int bd_link_disk_holder(struct block_device *bdev,
2086 				      struct gendisk *disk)
2087 {
2088 	return 0;
2089 }
2090 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2091 					 struct gendisk *disk)
2092 {
2093 }
2094 #endif
2095 #endif
2096 
2097 /* fs/char_dev.c */
2098 #define CHRDEV_MAJOR_HASH_SIZE	255
2099 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2100 extern int register_chrdev_region(dev_t, unsigned, const char *);
2101 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2102 			     unsigned int count, const char *name,
2103 			     const struct file_operations *fops);
2104 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2105 				unsigned int count, const char *name);
2106 extern void unregister_chrdev_region(dev_t, unsigned);
2107 extern void chrdev_show(struct seq_file *,off_t);
2108 
2109 static inline int register_chrdev(unsigned int major, const char *name,
2110 				  const struct file_operations *fops)
2111 {
2112 	return __register_chrdev(major, 0, 256, name, fops);
2113 }
2114 
2115 static inline void unregister_chrdev(unsigned int major, const char *name)
2116 {
2117 	__unregister_chrdev(major, 0, 256, name);
2118 }
2119 
2120 /* fs/block_dev.c */
2121 #define BDEVNAME_SIZE	32	/* Largest string for a blockdev identifier */
2122 #define BDEVT_SIZE	10	/* Largest string for MAJ:MIN for blkdev */
2123 
2124 #ifdef CONFIG_BLOCK
2125 #define BLKDEV_MAJOR_HASH_SIZE	255
2126 extern const char *__bdevname(dev_t, char *buffer);
2127 extern const char *bdevname(struct block_device *bdev, char *buffer);
2128 extern struct block_device *lookup_bdev(const char *);
2129 extern void blkdev_show(struct seq_file *,off_t);
2130 
2131 #else
2132 #define BLKDEV_MAJOR_HASH_SIZE	0
2133 #endif
2134 
2135 extern void init_special_inode(struct inode *, umode_t, dev_t);
2136 
2137 /* Invalid inode operations -- fs/bad_inode.c */
2138 extern void make_bad_inode(struct inode *);
2139 extern int is_bad_inode(struct inode *);
2140 
2141 extern const struct file_operations read_pipefifo_fops;
2142 extern const struct file_operations write_pipefifo_fops;
2143 extern const struct file_operations rdwr_pipefifo_fops;
2144 
2145 extern int fs_may_remount_ro(struct super_block *);
2146 
2147 #ifdef CONFIG_BLOCK
2148 /*
2149  * return READ, READA, or WRITE
2150  */
2151 #define bio_rw(bio)		((bio)->bi_rw & (RW_MASK | RWA_MASK))
2152 
2153 /*
2154  * return data direction, READ or WRITE
2155  */
2156 #define bio_data_dir(bio)	((bio)->bi_rw & 1)
2157 
2158 extern void check_disk_size_change(struct gendisk *disk,
2159 				   struct block_device *bdev);
2160 extern int revalidate_disk(struct gendisk *);
2161 extern int check_disk_change(struct block_device *);
2162 extern int __invalidate_device(struct block_device *, bool);
2163 extern int invalidate_partition(struct gendisk *, int);
2164 #endif
2165 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2166 					pgoff_t start, pgoff_t end);
2167 
2168 static inline void invalidate_remote_inode(struct inode *inode)
2169 {
2170 	if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2171 	    S_ISLNK(inode->i_mode))
2172 		invalidate_mapping_pages(inode->i_mapping, 0, -1);
2173 }
2174 extern int invalidate_inode_pages2(struct address_space *mapping);
2175 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2176 					 pgoff_t start, pgoff_t end);
2177 extern int write_inode_now(struct inode *, int);
2178 extern int filemap_fdatawrite(struct address_space *);
2179 extern int filemap_flush(struct address_space *);
2180 extern int filemap_fdatawait(struct address_space *);
2181 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2182 				   loff_t lend);
2183 extern int filemap_write_and_wait(struct address_space *mapping);
2184 extern int filemap_write_and_wait_range(struct address_space *mapping,
2185 				        loff_t lstart, loff_t lend);
2186 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2187 				loff_t start, loff_t end, int sync_mode);
2188 extern int filemap_fdatawrite_range(struct address_space *mapping,
2189 				loff_t start, loff_t end);
2190 
2191 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2192 			   int datasync);
2193 extern int vfs_fsync(struct file *file, int datasync);
2194 extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
2195 extern void sync_supers(void);
2196 extern void emergency_sync(void);
2197 extern void emergency_remount(void);
2198 #ifdef CONFIG_BLOCK
2199 extern sector_t bmap(struct inode *, sector_t);
2200 #endif
2201 extern int notify_change(struct dentry *, struct iattr *);
2202 extern int inode_permission(struct inode *, int);
2203 extern int generic_permission(struct inode *, int);
2204 
2205 static inline bool execute_ok(struct inode *inode)
2206 {
2207 	return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2208 }
2209 
2210 /*
2211  * get_write_access() gets write permission for a file.
2212  * put_write_access() releases this write permission.
2213  * This is used for regular files.
2214  * We cannot support write (and maybe mmap read-write shared) accesses and
2215  * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2216  * can have the following values:
2217  * 0: no writers, no VM_DENYWRITE mappings
2218  * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2219  * > 0: (i_writecount) users are writing to the file.
2220  *
2221  * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2222  * except for the cases where we don't hold i_writecount yet. Then we need to
2223  * use {get,deny}_write_access() - these functions check the sign and refuse
2224  * to do the change if sign is wrong.
2225  */
2226 static inline int get_write_access(struct inode *inode)
2227 {
2228 	return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2229 }
2230 static inline int deny_write_access(struct file *file)
2231 {
2232 	struct inode *inode = file->f_path.dentry->d_inode;
2233 	return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2234 }
2235 static inline void put_write_access(struct inode * inode)
2236 {
2237 	atomic_dec(&inode->i_writecount);
2238 }
2239 static inline void allow_write_access(struct file *file)
2240 {
2241 	if (file)
2242 		atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
2243 }
2244 #ifdef CONFIG_IMA
2245 static inline void i_readcount_dec(struct inode *inode)
2246 {
2247 	BUG_ON(!atomic_read(&inode->i_readcount));
2248 	atomic_dec(&inode->i_readcount);
2249 }
2250 static inline void i_readcount_inc(struct inode *inode)
2251 {
2252 	atomic_inc(&inode->i_readcount);
2253 }
2254 #else
2255 static inline void i_readcount_dec(struct inode *inode)
2256 {
2257 	return;
2258 }
2259 static inline void i_readcount_inc(struct inode *inode)
2260 {
2261 	return;
2262 }
2263 #endif
2264 extern int do_pipe_flags(int *, int);
2265 extern struct file *create_read_pipe(struct file *f, int flags);
2266 extern struct file *create_write_pipe(int flags);
2267 extern void free_write_pipe(struct file *);
2268 
2269 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2270 extern struct file * open_exec(const char *);
2271 
2272 /* fs/dcache.c -- generic fs support functions */
2273 extern int is_subdir(struct dentry *, struct dentry *);
2274 extern int path_is_under(struct path *, struct path *);
2275 extern ino_t find_inode_number(struct dentry *, struct qstr *);
2276 
2277 #include <linux/err.h>
2278 
2279 /* needed for stackable file system support */
2280 extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
2281 
2282 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
2283 
2284 extern int inode_init_always(struct super_block *, struct inode *);
2285 extern void inode_init_once(struct inode *);
2286 extern void address_space_init_once(struct address_space *mapping);
2287 extern void ihold(struct inode * inode);
2288 extern void iput(struct inode *);
2289 extern struct inode * igrab(struct inode *);
2290 extern ino_t iunique(struct super_block *, ino_t);
2291 extern int inode_needs_sync(struct inode *inode);
2292 extern int generic_delete_inode(struct inode *inode);
2293 extern int generic_drop_inode(struct inode *inode);
2294 
2295 extern struct inode *ilookup5_nowait(struct super_block *sb,
2296 		unsigned long hashval, int (*test)(struct inode *, void *),
2297 		void *data);
2298 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2299 		int (*test)(struct inode *, void *), void *data);
2300 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2301 
2302 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2303 extern struct inode * iget_locked(struct super_block *, unsigned long);
2304 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2305 extern int insert_inode_locked(struct inode *);
2306 extern void unlock_new_inode(struct inode *);
2307 extern unsigned int get_next_ino(void);
2308 
2309 extern void __iget(struct inode * inode);
2310 extern void iget_failed(struct inode *);
2311 extern void end_writeback(struct inode *);
2312 extern void __destroy_inode(struct inode *);
2313 extern struct inode *new_inode_pseudo(struct super_block *sb);
2314 extern struct inode *new_inode(struct super_block *sb);
2315 extern void free_inode_nonrcu(struct inode *inode);
2316 extern int should_remove_suid(struct dentry *);
2317 extern int file_remove_suid(struct file *);
2318 
2319 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2320 static inline void insert_inode_hash(struct inode *inode)
2321 {
2322 	__insert_inode_hash(inode, inode->i_ino);
2323 }
2324 
2325 extern void __remove_inode_hash(struct inode *);
2326 static inline void remove_inode_hash(struct inode *inode)
2327 {
2328 	if (!inode_unhashed(inode))
2329 		__remove_inode_hash(inode);
2330 }
2331 
2332 extern void inode_sb_list_add(struct inode *inode);
2333 
2334 #ifdef CONFIG_BLOCK
2335 extern void submit_bio(int, struct bio *);
2336 extern int bdev_read_only(struct block_device *);
2337 #endif
2338 extern int set_blocksize(struct block_device *, int);
2339 extern int sb_set_blocksize(struct super_block *, int);
2340 extern int sb_min_blocksize(struct super_block *, int);
2341 
2342 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2343 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2344 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
2345 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2346 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2347 extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2348 		loff_t *);
2349 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2350 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2351 		unsigned long *, loff_t, loff_t *, size_t, size_t);
2352 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2353 		unsigned long, loff_t, loff_t *, size_t, ssize_t);
2354 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2355 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2356 extern int generic_segment_checks(const struct iovec *iov,
2357 		unsigned long *nr_segs, size_t *count, int access_flags);
2358 
2359 /* fs/block_dev.c */
2360 extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2361 				unsigned long nr_segs, loff_t pos);
2362 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2363 			int datasync);
2364 
2365 /* fs/splice.c */
2366 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2367 		struct pipe_inode_info *, size_t, unsigned int);
2368 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2369 		struct pipe_inode_info *, size_t, unsigned int);
2370 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
2371 		struct file *, loff_t *, size_t, unsigned int);
2372 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2373 		struct file *out, loff_t *, size_t len, unsigned int flags);
2374 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2375 		size_t len, unsigned int flags);
2376 
2377 extern void
2378 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2379 extern loff_t noop_llseek(struct file *file, loff_t offset, int origin);
2380 extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
2381 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
2382 extern loff_t generic_file_llseek_unlocked(struct file *file, loff_t offset,
2383 			int origin);
2384 extern int generic_file_open(struct inode * inode, struct file * filp);
2385 extern int nonseekable_open(struct inode * inode, struct file * filp);
2386 
2387 #ifdef CONFIG_FS_XIP
2388 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2389 			     loff_t *ppos);
2390 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2391 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2392 			      size_t len, loff_t *ppos);
2393 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2394 #else
2395 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2396 {
2397 	return 0;
2398 }
2399 #endif
2400 
2401 #ifdef CONFIG_BLOCK
2402 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2403 			    loff_t file_offset);
2404 
2405 enum {
2406 	/* need locking between buffered and direct access */
2407 	DIO_LOCKING	= 0x01,
2408 
2409 	/* filesystem does not support filling holes */
2410 	DIO_SKIP_HOLES	= 0x02,
2411 };
2412 
2413 void dio_end_io(struct bio *bio, int error);
2414 void inode_dio_wait(struct inode *inode);
2415 void inode_dio_done(struct inode *inode);
2416 
2417 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2418 	struct block_device *bdev, const struct iovec *iov, loff_t offset,
2419 	unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2420 	dio_submit_t submit_io,	int flags);
2421 
2422 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2423 		struct inode *inode, const struct iovec *iov, loff_t offset,
2424 		unsigned long nr_segs, get_block_t get_block)
2425 {
2426 	return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
2427 				    offset, nr_segs, get_block, NULL, NULL,
2428 				    DIO_LOCKING | DIO_SKIP_HOLES);
2429 }
2430 #else
2431 static inline void inode_dio_wait(struct inode *inode)
2432 {
2433 }
2434 #endif
2435 
2436 extern const struct file_operations generic_ro_fops;
2437 
2438 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2439 
2440 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2441 extern int vfs_follow_link(struct nameidata *, const char *);
2442 extern int page_readlink(struct dentry *, char __user *, int);
2443 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2444 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2445 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2446 		int nofs);
2447 extern int page_symlink(struct inode *inode, const char *symname, int len);
2448 extern const struct inode_operations page_symlink_inode_operations;
2449 extern int generic_readlink(struct dentry *, char __user *, int);
2450 extern void generic_fillattr(struct inode *, struct kstat *);
2451 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2452 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2453 void inode_add_bytes(struct inode *inode, loff_t bytes);
2454 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2455 loff_t inode_get_bytes(struct inode *inode);
2456 void inode_set_bytes(struct inode *inode, loff_t bytes);
2457 
2458 extern int vfs_readdir(struct file *, filldir_t, void *);
2459 
2460 extern int vfs_stat(const char __user *, struct kstat *);
2461 extern int vfs_lstat(const char __user *, struct kstat *);
2462 extern int vfs_fstat(unsigned int, struct kstat *);
2463 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2464 
2465 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2466 		    unsigned long arg);
2467 extern int __generic_block_fiemap(struct inode *inode,
2468 				  struct fiemap_extent_info *fieinfo,
2469 				  loff_t start, loff_t len,
2470 				  get_block_t *get_block);
2471 extern int generic_block_fiemap(struct inode *inode,
2472 				struct fiemap_extent_info *fieinfo, u64 start,
2473 				u64 len, get_block_t *get_block);
2474 
2475 extern void get_filesystem(struct file_system_type *fs);
2476 extern void put_filesystem(struct file_system_type *fs);
2477 extern struct file_system_type *get_fs_type(const char *name);
2478 extern struct super_block *get_super(struct block_device *);
2479 extern struct super_block *get_active_super(struct block_device *bdev);
2480 extern struct super_block *user_get_super(dev_t);
2481 extern void drop_super(struct super_block *sb);
2482 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2483 extern void iterate_supers_type(struct file_system_type *,
2484 			        void (*)(struct super_block *, void *), void *);
2485 
2486 extern int dcache_dir_open(struct inode *, struct file *);
2487 extern int dcache_dir_close(struct inode *, struct file *);
2488 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2489 extern int dcache_readdir(struct file *, void *, filldir_t);
2490 extern int simple_setattr(struct dentry *, struct iattr *);
2491 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2492 extern int simple_statfs(struct dentry *, struct kstatfs *);
2493 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2494 extern int simple_unlink(struct inode *, struct dentry *);
2495 extern int simple_rmdir(struct inode *, struct dentry *);
2496 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2497 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2498 extern int simple_empty(struct dentry *);
2499 extern int simple_readpage(struct file *file, struct page *page);
2500 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2501 			loff_t pos, unsigned len, unsigned flags,
2502 			struct page **pagep, void **fsdata);
2503 extern int simple_write_end(struct file *file, struct address_space *mapping,
2504 			loff_t pos, unsigned len, unsigned copied,
2505 			struct page *page, void *fsdata);
2506 
2507 extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
2508 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2509 extern const struct file_operations simple_dir_operations;
2510 extern const struct inode_operations simple_dir_inode_operations;
2511 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2512 struct dentry *d_alloc_name(struct dentry *, const char *);
2513 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2514 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2515 extern void simple_release_fs(struct vfsmount **mount, int *count);
2516 
2517 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2518 			loff_t *ppos, const void *from, size_t available);
2519 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2520 		const void __user *from, size_t count);
2521 
2522 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2523 
2524 extern int generic_check_addressable(unsigned, u64);
2525 
2526 #ifdef CONFIG_MIGRATION
2527 extern int buffer_migrate_page(struct address_space *,
2528 				struct page *, struct page *);
2529 #else
2530 #define buffer_migrate_page NULL
2531 #endif
2532 
2533 extern int inode_change_ok(const struct inode *, struct iattr *);
2534 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2535 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2536 
2537 extern void file_update_time(struct file *file);
2538 
2539 extern int generic_show_options(struct seq_file *m, struct vfsmount *mnt);
2540 extern void save_mount_options(struct super_block *sb, char *options);
2541 extern void replace_mount_options(struct super_block *sb, char *options);
2542 
2543 static inline ino_t parent_ino(struct dentry *dentry)
2544 {
2545 	ino_t res;
2546 
2547 	/*
2548 	 * Don't strictly need d_lock here? If the parent ino could change
2549 	 * then surely we'd have a deeper race in the caller?
2550 	 */
2551 	spin_lock(&dentry->d_lock);
2552 	res = dentry->d_parent->d_inode->i_ino;
2553 	spin_unlock(&dentry->d_lock);
2554 	return res;
2555 }
2556 
2557 /* Transaction based IO helpers */
2558 
2559 /*
2560  * An argresp is stored in an allocated page and holds the
2561  * size of the argument or response, along with its content
2562  */
2563 struct simple_transaction_argresp {
2564 	ssize_t size;
2565 	char data[0];
2566 };
2567 
2568 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2569 
2570 char *simple_transaction_get(struct file *file, const char __user *buf,
2571 				size_t size);
2572 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2573 				size_t size, loff_t *pos);
2574 int simple_transaction_release(struct inode *inode, struct file *file);
2575 
2576 void simple_transaction_set(struct file *file, size_t n);
2577 
2578 /*
2579  * simple attribute files
2580  *
2581  * These attributes behave similar to those in sysfs:
2582  *
2583  * Writing to an attribute immediately sets a value, an open file can be
2584  * written to multiple times.
2585  *
2586  * Reading from an attribute creates a buffer from the value that might get
2587  * read with multiple read calls. When the attribute has been read
2588  * completely, no further read calls are possible until the file is opened
2589  * again.
2590  *
2591  * All attributes contain a text representation of a numeric value
2592  * that are accessed with the get() and set() functions.
2593  */
2594 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt)		\
2595 static int __fops ## _open(struct inode *inode, struct file *file)	\
2596 {									\
2597 	__simple_attr_check_format(__fmt, 0ull);			\
2598 	return simple_attr_open(inode, file, __get, __set, __fmt);	\
2599 }									\
2600 static const struct file_operations __fops = {				\
2601 	.owner	 = THIS_MODULE,						\
2602 	.open	 = __fops ## _open,					\
2603 	.release = simple_attr_release,					\
2604 	.read	 = simple_attr_read,					\
2605 	.write	 = simple_attr_write,					\
2606 	.llseek	 = generic_file_llseek,					\
2607 };
2608 
2609 static inline void __attribute__((format(printf, 1, 2)))
2610 __simple_attr_check_format(const char *fmt, ...)
2611 {
2612 	/* don't do anything, just let the compiler check the arguments; */
2613 }
2614 
2615 int simple_attr_open(struct inode *inode, struct file *file,
2616 		     int (*get)(void *, u64 *), int (*set)(void *, u64),
2617 		     const char *fmt);
2618 int simple_attr_release(struct inode *inode, struct file *file);
2619 ssize_t simple_attr_read(struct file *file, char __user *buf,
2620 			 size_t len, loff_t *ppos);
2621 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2622 			  size_t len, loff_t *ppos);
2623 
2624 struct ctl_table;
2625 int proc_nr_files(struct ctl_table *table, int write,
2626 		  void __user *buffer, size_t *lenp, loff_t *ppos);
2627 int proc_nr_dentry(struct ctl_table *table, int write,
2628 		  void __user *buffer, size_t *lenp, loff_t *ppos);
2629 int proc_nr_inodes(struct ctl_table *table, int write,
2630 		   void __user *buffer, size_t *lenp, loff_t *ppos);
2631 int __init get_filesystem_list(char *buf);
2632 
2633 #define __FMODE_EXEC		((__force int) FMODE_EXEC)
2634 #define __FMODE_NONOTIFY	((__force int) FMODE_NONOTIFY)
2635 
2636 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2637 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2638 					    (flag & __FMODE_NONOTIFY)))
2639 
2640 static inline int is_sxid(mode_t mode)
2641 {
2642 	return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2643 }
2644 
2645 static inline void inode_has_no_xattr(struct inode *inode)
2646 {
2647 	if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2648 		inode->i_flags |= S_NOSEC;
2649 }
2650 
2651 #endif /* __KERNEL__ */
2652 #endif /* _LINUX_FS_H */
2653