xref: /linux-6.15/include/linux/fs.h (revision e67f88dd)
1 #ifndef _LINUX_FS_H
2 #define _LINUX_FS_H
3 
4 /*
5  * This file has definitions for some important file table
6  * structures etc.
7  */
8 
9 #include <linux/limits.h>
10 #include <linux/ioctl.h>
11 #include <linux/blk_types.h>
12 #include <linux/types.h>
13 
14 /*
15  * It's silly to have NR_OPEN bigger than NR_FILE, but you can change
16  * the file limit at runtime and only root can increase the per-process
17  * nr_file rlimit, so it's safe to set up a ridiculously high absolute
18  * upper limit on files-per-process.
19  *
20  * Some programs (notably those using select()) may have to be
21  * recompiled to take full advantage of the new limits..
22  */
23 
24 /* Fixed constants first: */
25 #undef NR_OPEN
26 #define INR_OPEN 1024		/* Initial setting for nfile rlimits */
27 
28 #define BLOCK_SIZE_BITS 10
29 #define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
30 
31 #define SEEK_SET	0	/* seek relative to beginning of file */
32 #define SEEK_CUR	1	/* seek relative to current file position */
33 #define SEEK_END	2	/* seek relative to end of file */
34 #define SEEK_MAX	SEEK_END
35 
36 struct fstrim_range {
37 	__u64 start;
38 	__u64 len;
39 	__u64 minlen;
40 };
41 
42 /* And dynamically-tunable limits and defaults: */
43 struct files_stat_struct {
44 	unsigned long nr_files;		/* read only */
45 	unsigned long nr_free_files;	/* read only */
46 	unsigned long max_files;		/* tunable */
47 };
48 
49 struct inodes_stat_t {
50 	int nr_inodes;
51 	int nr_unused;
52 	int dummy[5];		/* padding for sysctl ABI compatibility */
53 };
54 
55 
56 #define NR_FILE  8192	/* this can well be larger on a larger system */
57 
58 #define MAY_EXEC 1
59 #define MAY_WRITE 2
60 #define MAY_READ 4
61 #define MAY_APPEND 8
62 #define MAY_ACCESS 16
63 #define MAY_OPEN 32
64 #define MAY_CHDIR 64
65 
66 /*
67  * flags in file.f_mode.  Note that FMODE_READ and FMODE_WRITE must correspond
68  * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
69  */
70 
71 /* file is open for reading */
72 #define FMODE_READ		((__force fmode_t)0x1)
73 /* file is open for writing */
74 #define FMODE_WRITE		((__force fmode_t)0x2)
75 /* file is seekable */
76 #define FMODE_LSEEK		((__force fmode_t)0x4)
77 /* file can be accessed using pread */
78 #define FMODE_PREAD		((__force fmode_t)0x8)
79 /* file can be accessed using pwrite */
80 #define FMODE_PWRITE		((__force fmode_t)0x10)
81 /* File is opened for execution with sys_execve / sys_uselib */
82 #define FMODE_EXEC		((__force fmode_t)0x20)
83 /* File is opened with O_NDELAY (only set for block devices) */
84 #define FMODE_NDELAY		((__force fmode_t)0x40)
85 /* File is opened with O_EXCL (only set for block devices) */
86 #define FMODE_EXCL		((__force fmode_t)0x80)
87 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
88    (specialy hack for floppy.c) */
89 #define FMODE_WRITE_IOCTL	((__force fmode_t)0x100)
90 
91 /*
92  * Don't update ctime and mtime.
93  *
94  * Currently a special hack for the XFS open_by_handle ioctl, but we'll
95  * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
96  */
97 #define FMODE_NOCMTIME		((__force fmode_t)0x800)
98 
99 /* Expect random access pattern */
100 #define FMODE_RANDOM		((__force fmode_t)0x1000)
101 
102 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
103 #define FMODE_UNSIGNED_OFFSET	((__force fmode_t)0x2000)
104 
105 /* File is opened with O_PATH; almost nothing can be done with it */
106 #define FMODE_PATH		((__force fmode_t)0x4000)
107 
108 /* File was opened by fanotify and shouldn't generate fanotify events */
109 #define FMODE_NONOTIFY		((__force fmode_t)0x1000000)
110 
111 /*
112  * The below are the various read and write types that we support. Some of
113  * them include behavioral modifiers that send information down to the
114  * block layer and IO scheduler. Terminology:
115  *
116  *	The block layer uses device plugging to defer IO a little bit, in
117  *	the hope that we will see more IO very shortly. This increases
118  *	coalescing of adjacent IO and thus reduces the number of IOs we
119  *	have to send to the device. It also allows for better queuing,
120  *	if the IO isn't mergeable. If the caller is going to be waiting
121  *	for the IO, then he must ensure that the device is unplugged so
122  *	that the IO is dispatched to the driver.
123  *
124  *	All IO is handled async in Linux. This is fine for background
125  *	writes, but for reads or writes that someone waits for completion
126  *	on, we want to notify the block layer and IO scheduler so that they
127  *	know about it. That allows them to make better scheduling
128  *	decisions. So when the below references 'sync' and 'async', it
129  *	is referencing this priority hint.
130  *
131  * With that in mind, the available types are:
132  *
133  * READ			A normal read operation. Device will be plugged.
134  * READ_SYNC		A synchronous read. Device is not plugged, caller can
135  *			immediately wait on this read without caring about
136  *			unplugging.
137  * READA		Used for read-ahead operations. Lower priority, and the
138  *			block layer could (in theory) choose to ignore this
139  *			request if it runs into resource problems.
140  * WRITE		A normal async write. Device will be plugged.
141  * WRITE_SYNC		Synchronous write. Identical to WRITE, but passes down
142  *			the hint that someone will be waiting on this IO
143  *			shortly. The write equivalent of READ_SYNC.
144  * WRITE_ODIRECT	Special case write for O_DIRECT only.
145  * WRITE_FLUSH		Like WRITE_SYNC but with preceding cache flush.
146  * WRITE_FUA		Like WRITE_SYNC but data is guaranteed to be on
147  *			non-volatile media on completion.
148  * WRITE_FLUSH_FUA	Combination of WRITE_FLUSH and FUA. The IO is preceded
149  *			by a cache flush and data is guaranteed to be on
150  *			non-volatile media on completion.
151  *
152  */
153 #define RW_MASK			REQ_WRITE
154 #define RWA_MASK		REQ_RAHEAD
155 
156 #define READ			0
157 #define WRITE			RW_MASK
158 #define READA			RWA_MASK
159 
160 #define READ_SYNC		(READ | REQ_SYNC)
161 #define READ_META		(READ | REQ_META)
162 #define WRITE_SYNC		(WRITE | REQ_SYNC | REQ_NOIDLE)
163 #define WRITE_ODIRECT		(WRITE | REQ_SYNC)
164 #define WRITE_META		(WRITE | REQ_META)
165 #define WRITE_FLUSH		(WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
166 #define WRITE_FUA		(WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
167 #define WRITE_FLUSH_FUA		(WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
168 
169 #define SEL_IN		1
170 #define SEL_OUT		2
171 #define SEL_EX		4
172 
173 /* public flags for file_system_type */
174 #define FS_REQUIRES_DEV 1
175 #define FS_BINARY_MOUNTDATA 2
176 #define FS_HAS_SUBTYPE 4
177 #define FS_REVAL_DOT	16384	/* Check the paths ".", ".." for staleness */
178 #define FS_RENAME_DOES_D_MOVE	32768	/* FS will handle d_move()
179 					 * during rename() internally.
180 					 */
181 
182 /*
183  * These are the fs-independent mount-flags: up to 32 flags are supported
184  */
185 #define MS_RDONLY	 1	/* Mount read-only */
186 #define MS_NOSUID	 2	/* Ignore suid and sgid bits */
187 #define MS_NODEV	 4	/* Disallow access to device special files */
188 #define MS_NOEXEC	 8	/* Disallow program execution */
189 #define MS_SYNCHRONOUS	16	/* Writes are synced at once */
190 #define MS_REMOUNT	32	/* Alter flags of a mounted FS */
191 #define MS_MANDLOCK	64	/* Allow mandatory locks on an FS */
192 #define MS_DIRSYNC	128	/* Directory modifications are synchronous */
193 #define MS_NOATIME	1024	/* Do not update access times. */
194 #define MS_NODIRATIME	2048	/* Do not update directory access times */
195 #define MS_BIND		4096
196 #define MS_MOVE		8192
197 #define MS_REC		16384
198 #define MS_VERBOSE	32768	/* War is peace. Verbosity is silence.
199 				   MS_VERBOSE is deprecated. */
200 #define MS_SILENT	32768
201 #define MS_POSIXACL	(1<<16)	/* VFS does not apply the umask */
202 #define MS_UNBINDABLE	(1<<17)	/* change to unbindable */
203 #define MS_PRIVATE	(1<<18)	/* change to private */
204 #define MS_SLAVE	(1<<19)	/* change to slave */
205 #define MS_SHARED	(1<<20)	/* change to shared */
206 #define MS_RELATIME	(1<<21)	/* Update atime relative to mtime/ctime. */
207 #define MS_KERNMOUNT	(1<<22) /* this is a kern_mount call */
208 #define MS_I_VERSION	(1<<23) /* Update inode I_version field */
209 #define MS_STRICTATIME	(1<<24) /* Always perform atime updates */
210 #define MS_BORN		(1<<29)
211 #define MS_ACTIVE	(1<<30)
212 #define MS_NOUSER	(1<<31)
213 
214 /*
215  * Superblock flags that can be altered by MS_REMOUNT
216  */
217 #define MS_RMT_MASK	(MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_I_VERSION)
218 
219 /*
220  * Old magic mount flag and mask
221  */
222 #define MS_MGC_VAL 0xC0ED0000
223 #define MS_MGC_MSK 0xffff0000
224 
225 /* Inode flags - they have nothing to superblock flags now */
226 
227 #define S_SYNC		1	/* Writes are synced at once */
228 #define S_NOATIME	2	/* Do not update access times */
229 #define S_APPEND	4	/* Append-only file */
230 #define S_IMMUTABLE	8	/* Immutable file */
231 #define S_DEAD		16	/* removed, but still open directory */
232 #define S_NOQUOTA	32	/* Inode is not counted to quota */
233 #define S_DIRSYNC	64	/* Directory modifications are synchronous */
234 #define S_NOCMTIME	128	/* Do not update file c/mtime */
235 #define S_SWAPFILE	256	/* Do not truncate: swapon got its bmaps */
236 #define S_PRIVATE	512	/* Inode is fs-internal */
237 #define S_IMA		1024	/* Inode has an associated IMA struct */
238 #define S_AUTOMOUNT	2048	/* Automount/referral quasi-directory */
239 
240 /*
241  * Note that nosuid etc flags are inode-specific: setting some file-system
242  * flags just means all the inodes inherit those flags by default. It might be
243  * possible to override it selectively if you really wanted to with some
244  * ioctl() that is not currently implemented.
245  *
246  * Exception: MS_RDONLY is always applied to the entire file system.
247  *
248  * Unfortunately, it is possible to change a filesystems flags with it mounted
249  * with files in use.  This means that all of the inodes will not have their
250  * i_flags updated.  Hence, i_flags no longer inherit the superblock mount
251  * flags, so these have to be checked separately. -- [email protected]
252  */
253 #define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
254 
255 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
256 #define IS_SYNC(inode)		(__IS_FLG(inode, MS_SYNCHRONOUS) || \
257 					((inode)->i_flags & S_SYNC))
258 #define IS_DIRSYNC(inode)	(__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
259 					((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
260 #define IS_MANDLOCK(inode)	__IS_FLG(inode, MS_MANDLOCK)
261 #define IS_NOATIME(inode)   __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
262 #define IS_I_VERSION(inode)   __IS_FLG(inode, MS_I_VERSION)
263 
264 #define IS_NOQUOTA(inode)	((inode)->i_flags & S_NOQUOTA)
265 #define IS_APPEND(inode)	((inode)->i_flags & S_APPEND)
266 #define IS_IMMUTABLE(inode)	((inode)->i_flags & S_IMMUTABLE)
267 #define IS_POSIXACL(inode)	__IS_FLG(inode, MS_POSIXACL)
268 
269 #define IS_DEADDIR(inode)	((inode)->i_flags & S_DEAD)
270 #define IS_NOCMTIME(inode)	((inode)->i_flags & S_NOCMTIME)
271 #define IS_SWAPFILE(inode)	((inode)->i_flags & S_SWAPFILE)
272 #define IS_PRIVATE(inode)	((inode)->i_flags & S_PRIVATE)
273 #define IS_IMA(inode)		((inode)->i_flags & S_IMA)
274 #define IS_AUTOMOUNT(inode)	((inode)->i_flags & S_AUTOMOUNT)
275 
276 /* the read-only stuff doesn't really belong here, but any other place is
277    probably as bad and I don't want to create yet another include file. */
278 
279 #define BLKROSET   _IO(0x12,93)	/* set device read-only (0 = read-write) */
280 #define BLKROGET   _IO(0x12,94)	/* get read-only status (0 = read_write) */
281 #define BLKRRPART  _IO(0x12,95)	/* re-read partition table */
282 #define BLKGETSIZE _IO(0x12,96)	/* return device size /512 (long *arg) */
283 #define BLKFLSBUF  _IO(0x12,97)	/* flush buffer cache */
284 #define BLKRASET   _IO(0x12,98)	/* set read ahead for block device */
285 #define BLKRAGET   _IO(0x12,99)	/* get current read ahead setting */
286 #define BLKFRASET  _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
287 #define BLKFRAGET  _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
288 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
289 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
290 #define BLKSSZGET  _IO(0x12,104)/* get block device sector size */
291 #if 0
292 #define BLKPG      _IO(0x12,105)/* See blkpg.h */
293 
294 /* Some people are morons.  Do not use sizeof! */
295 
296 #define BLKELVGET  _IOR(0x12,106,size_t)/* elevator get */
297 #define BLKELVSET  _IOW(0x12,107,size_t)/* elevator set */
298 /* This was here just to show that the number is taken -
299    probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
300 #endif
301 /* A jump here: 108-111 have been used for various private purposes. */
302 #define BLKBSZGET  _IOR(0x12,112,size_t)
303 #define BLKBSZSET  _IOW(0x12,113,size_t)
304 #define BLKGETSIZE64 _IOR(0x12,114,size_t)	/* return device size in bytes (u64 *arg) */
305 #define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
306 #define BLKTRACESTART _IO(0x12,116)
307 #define BLKTRACESTOP _IO(0x12,117)
308 #define BLKTRACETEARDOWN _IO(0x12,118)
309 #define BLKDISCARD _IO(0x12,119)
310 #define BLKIOMIN _IO(0x12,120)
311 #define BLKIOOPT _IO(0x12,121)
312 #define BLKALIGNOFF _IO(0x12,122)
313 #define BLKPBSZGET _IO(0x12,123)
314 #define BLKDISCARDZEROES _IO(0x12,124)
315 #define BLKSECDISCARD _IO(0x12,125)
316 
317 #define BMAP_IOCTL 1		/* obsolete - kept for compatibility */
318 #define FIBMAP	   _IO(0x00,1)	/* bmap access */
319 #define FIGETBSZ   _IO(0x00,2)	/* get the block size used for bmap */
320 #define FIFREEZE	_IOWR('X', 119, int)	/* Freeze */
321 #define FITHAW		_IOWR('X', 120, int)	/* Thaw */
322 #define FITRIM		_IOWR('X', 121, struct fstrim_range)	/* Trim */
323 
324 #define	FS_IOC_GETFLAGS			_IOR('f', 1, long)
325 #define	FS_IOC_SETFLAGS			_IOW('f', 2, long)
326 #define	FS_IOC_GETVERSION		_IOR('v', 1, long)
327 #define	FS_IOC_SETVERSION		_IOW('v', 2, long)
328 #define FS_IOC_FIEMAP			_IOWR('f', 11, struct fiemap)
329 #define FS_IOC32_GETFLAGS		_IOR('f', 1, int)
330 #define FS_IOC32_SETFLAGS		_IOW('f', 2, int)
331 #define FS_IOC32_GETVERSION		_IOR('v', 1, int)
332 #define FS_IOC32_SETVERSION		_IOW('v', 2, int)
333 
334 /*
335  * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS)
336  */
337 #define	FS_SECRM_FL			0x00000001 /* Secure deletion */
338 #define	FS_UNRM_FL			0x00000002 /* Undelete */
339 #define	FS_COMPR_FL			0x00000004 /* Compress file */
340 #define FS_SYNC_FL			0x00000008 /* Synchronous updates */
341 #define FS_IMMUTABLE_FL			0x00000010 /* Immutable file */
342 #define FS_APPEND_FL			0x00000020 /* writes to file may only append */
343 #define FS_NODUMP_FL			0x00000040 /* do not dump file */
344 #define FS_NOATIME_FL			0x00000080 /* do not update atime */
345 /* Reserved for compression usage... */
346 #define FS_DIRTY_FL			0x00000100
347 #define FS_COMPRBLK_FL			0x00000200 /* One or more compressed clusters */
348 #define FS_NOCOMP_FL			0x00000400 /* Don't compress */
349 #define FS_ECOMPR_FL			0x00000800 /* Compression error */
350 /* End compression flags --- maybe not all used */
351 #define FS_BTREE_FL			0x00001000 /* btree format dir */
352 #define FS_INDEX_FL			0x00001000 /* hash-indexed directory */
353 #define FS_IMAGIC_FL			0x00002000 /* AFS directory */
354 #define FS_JOURNAL_DATA_FL		0x00004000 /* Reserved for ext3 */
355 #define FS_NOTAIL_FL			0x00008000 /* file tail should not be merged */
356 #define FS_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
357 #define FS_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
358 #define FS_EXTENT_FL			0x00080000 /* Extents */
359 #define FS_DIRECTIO_FL			0x00100000 /* Use direct i/o */
360 #define FS_NOCOW_FL			0x00800000 /* Do not cow file */
361 #define FS_COW_FL			0x02000000 /* Cow file */
362 #define FS_RESERVED_FL			0x80000000 /* reserved for ext2 lib */
363 
364 #define FS_FL_USER_VISIBLE		0x0003DFFF /* User visible flags */
365 #define FS_FL_USER_MODIFIABLE		0x000380FF /* User modifiable flags */
366 
367 
368 #define SYNC_FILE_RANGE_WAIT_BEFORE	1
369 #define SYNC_FILE_RANGE_WRITE		2
370 #define SYNC_FILE_RANGE_WAIT_AFTER	4
371 
372 #ifdef __KERNEL__
373 
374 #include <linux/linkage.h>
375 #include <linux/wait.h>
376 #include <linux/types.h>
377 #include <linux/kdev_t.h>
378 #include <linux/dcache.h>
379 #include <linux/path.h>
380 #include <linux/stat.h>
381 #include <linux/cache.h>
382 #include <linux/list.h>
383 #include <linux/radix-tree.h>
384 #include <linux/prio_tree.h>
385 #include <linux/init.h>
386 #include <linux/pid.h>
387 #include <linux/mutex.h>
388 #include <linux/capability.h>
389 #include <linux/semaphore.h>
390 #include <linux/fiemap.h>
391 #include <linux/rculist_bl.h>
392 
393 #include <asm/atomic.h>
394 #include <asm/byteorder.h>
395 
396 struct export_operations;
397 struct hd_geometry;
398 struct iovec;
399 struct nameidata;
400 struct kiocb;
401 struct kobject;
402 struct pipe_inode_info;
403 struct poll_table_struct;
404 struct kstatfs;
405 struct vm_area_struct;
406 struct vfsmount;
407 struct cred;
408 
409 extern void __init inode_init(void);
410 extern void __init inode_init_early(void);
411 extern void __init files_init(unsigned long);
412 
413 extern struct files_stat_struct files_stat;
414 extern unsigned long get_max_files(void);
415 extern int sysctl_nr_open;
416 extern struct inodes_stat_t inodes_stat;
417 extern int leases_enable, lease_break_time;
418 
419 struct buffer_head;
420 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
421 			struct buffer_head *bh_result, int create);
422 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
423 			ssize_t bytes, void *private, int ret,
424 			bool is_async);
425 
426 /*
427  * Attribute flags.  These should be or-ed together to figure out what
428  * has been changed!
429  */
430 #define ATTR_MODE	(1 << 0)
431 #define ATTR_UID	(1 << 1)
432 #define ATTR_GID	(1 << 2)
433 #define ATTR_SIZE	(1 << 3)
434 #define ATTR_ATIME	(1 << 4)
435 #define ATTR_MTIME	(1 << 5)
436 #define ATTR_CTIME	(1 << 6)
437 #define ATTR_ATIME_SET	(1 << 7)
438 #define ATTR_MTIME_SET	(1 << 8)
439 #define ATTR_FORCE	(1 << 9) /* Not a change, but a change it */
440 #define ATTR_ATTR_FLAG	(1 << 10)
441 #define ATTR_KILL_SUID	(1 << 11)
442 #define ATTR_KILL_SGID	(1 << 12)
443 #define ATTR_FILE	(1 << 13)
444 #define ATTR_KILL_PRIV	(1 << 14)
445 #define ATTR_OPEN	(1 << 15) /* Truncating from open(O_TRUNC) */
446 #define ATTR_TIMES_SET	(1 << 16)
447 
448 /*
449  * This is the Inode Attributes structure, used for notify_change().  It
450  * uses the above definitions as flags, to know which values have changed.
451  * Also, in this manner, a Filesystem can look at only the values it cares
452  * about.  Basically, these are the attributes that the VFS layer can
453  * request to change from the FS layer.
454  *
455  * Derek Atkins <[email protected]> 94-10-20
456  */
457 struct iattr {
458 	unsigned int	ia_valid;
459 	umode_t		ia_mode;
460 	uid_t		ia_uid;
461 	gid_t		ia_gid;
462 	loff_t		ia_size;
463 	struct timespec	ia_atime;
464 	struct timespec	ia_mtime;
465 	struct timespec	ia_ctime;
466 
467 	/*
468 	 * Not an attribute, but an auxiliary info for filesystems wanting to
469 	 * implement an ftruncate() like method.  NOTE: filesystem should
470 	 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
471 	 */
472 	struct file	*ia_file;
473 };
474 
475 /*
476  * Includes for diskquotas.
477  */
478 #include <linux/quota.h>
479 
480 /**
481  * enum positive_aop_returns - aop return codes with specific semantics
482  *
483  * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
484  * 			    completed, that the page is still locked, and
485  * 			    should be considered active.  The VM uses this hint
486  * 			    to return the page to the active list -- it won't
487  * 			    be a candidate for writeback again in the near
488  * 			    future.  Other callers must be careful to unlock
489  * 			    the page if they get this return.  Returned by
490  * 			    writepage();
491  *
492  * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
493  *  			unlocked it and the page might have been truncated.
494  *  			The caller should back up to acquiring a new page and
495  *  			trying again.  The aop will be taking reasonable
496  *  			precautions not to livelock.  If the caller held a page
497  *  			reference, it should drop it before retrying.  Returned
498  *  			by readpage().
499  *
500  * address_space_operation functions return these large constants to indicate
501  * special semantics to the caller.  These are much larger than the bytes in a
502  * page to allow for functions that return the number of bytes operated on in a
503  * given page.
504  */
505 
506 enum positive_aop_returns {
507 	AOP_WRITEPAGE_ACTIVATE	= 0x80000,
508 	AOP_TRUNCATED_PAGE	= 0x80001,
509 };
510 
511 #define AOP_FLAG_UNINTERRUPTIBLE	0x0001 /* will not do a short write */
512 #define AOP_FLAG_CONT_EXPAND		0x0002 /* called from cont_expand */
513 #define AOP_FLAG_NOFS			0x0004 /* used by filesystem to direct
514 						* helper code (eg buffer layer)
515 						* to clear GFP_FS from alloc */
516 
517 /*
518  * oh the beauties of C type declarations.
519  */
520 struct page;
521 struct address_space;
522 struct writeback_control;
523 
524 struct iov_iter {
525 	const struct iovec *iov;
526 	unsigned long nr_segs;
527 	size_t iov_offset;
528 	size_t count;
529 };
530 
531 size_t iov_iter_copy_from_user_atomic(struct page *page,
532 		struct iov_iter *i, unsigned long offset, size_t bytes);
533 size_t iov_iter_copy_from_user(struct page *page,
534 		struct iov_iter *i, unsigned long offset, size_t bytes);
535 void iov_iter_advance(struct iov_iter *i, size_t bytes);
536 int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
537 size_t iov_iter_single_seg_count(struct iov_iter *i);
538 
539 static inline void iov_iter_init(struct iov_iter *i,
540 			const struct iovec *iov, unsigned long nr_segs,
541 			size_t count, size_t written)
542 {
543 	i->iov = iov;
544 	i->nr_segs = nr_segs;
545 	i->iov_offset = 0;
546 	i->count = count + written;
547 
548 	iov_iter_advance(i, written);
549 }
550 
551 static inline size_t iov_iter_count(struct iov_iter *i)
552 {
553 	return i->count;
554 }
555 
556 /*
557  * "descriptor" for what we're up to with a read.
558  * This allows us to use the same read code yet
559  * have multiple different users of the data that
560  * we read from a file.
561  *
562  * The simplest case just copies the data to user
563  * mode.
564  */
565 typedef struct {
566 	size_t written;
567 	size_t count;
568 	union {
569 		char __user *buf;
570 		void *data;
571 	} arg;
572 	int error;
573 } read_descriptor_t;
574 
575 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
576 		unsigned long, unsigned long);
577 
578 struct address_space_operations {
579 	int (*writepage)(struct page *page, struct writeback_control *wbc);
580 	int (*readpage)(struct file *, struct page *);
581 
582 	/* Write back some dirty pages from this mapping. */
583 	int (*writepages)(struct address_space *, struct writeback_control *);
584 
585 	/* Set a page dirty.  Return true if this dirtied it */
586 	int (*set_page_dirty)(struct page *page);
587 
588 	int (*readpages)(struct file *filp, struct address_space *mapping,
589 			struct list_head *pages, unsigned nr_pages);
590 
591 	int (*write_begin)(struct file *, struct address_space *mapping,
592 				loff_t pos, unsigned len, unsigned flags,
593 				struct page **pagep, void **fsdata);
594 	int (*write_end)(struct file *, struct address_space *mapping,
595 				loff_t pos, unsigned len, unsigned copied,
596 				struct page *page, void *fsdata);
597 
598 	/* Unfortunately this kludge is needed for FIBMAP. Don't use it */
599 	sector_t (*bmap)(struct address_space *, sector_t);
600 	void (*invalidatepage) (struct page *, unsigned long);
601 	int (*releasepage) (struct page *, gfp_t);
602 	void (*freepage)(struct page *);
603 	ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
604 			loff_t offset, unsigned long nr_segs);
605 	int (*get_xip_mem)(struct address_space *, pgoff_t, int,
606 						void **, unsigned long *);
607 	/* migrate the contents of a page to the specified target */
608 	int (*migratepage) (struct address_space *,
609 			struct page *, struct page *);
610 	int (*launder_page) (struct page *);
611 	int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
612 					unsigned long);
613 	int (*error_remove_page)(struct address_space *, struct page *);
614 };
615 
616 extern const struct address_space_operations empty_aops;
617 
618 /*
619  * pagecache_write_begin/pagecache_write_end must be used by general code
620  * to write into the pagecache.
621  */
622 int pagecache_write_begin(struct file *, struct address_space *mapping,
623 				loff_t pos, unsigned len, unsigned flags,
624 				struct page **pagep, void **fsdata);
625 
626 int pagecache_write_end(struct file *, struct address_space *mapping,
627 				loff_t pos, unsigned len, unsigned copied,
628 				struct page *page, void *fsdata);
629 
630 struct backing_dev_info;
631 struct address_space {
632 	struct inode		*host;		/* owner: inode, block_device */
633 	struct radix_tree_root	page_tree;	/* radix tree of all pages */
634 	spinlock_t		tree_lock;	/* and lock protecting it */
635 	unsigned int		i_mmap_writable;/* count VM_SHARED mappings */
636 	struct prio_tree_root	i_mmap;		/* tree of private and shared mappings */
637 	struct list_head	i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
638 	spinlock_t		i_mmap_lock;	/* protect tree, count, list */
639 	unsigned int		truncate_count;	/* Cover race condition with truncate */
640 	unsigned long		nrpages;	/* number of total pages */
641 	pgoff_t			writeback_index;/* writeback starts here */
642 	const struct address_space_operations *a_ops;	/* methods */
643 	unsigned long		flags;		/* error bits/gfp mask */
644 	struct backing_dev_info *backing_dev_info; /* device readahead, etc */
645 	spinlock_t		private_lock;	/* for use by the address_space */
646 	struct list_head	private_list;	/* ditto */
647 	struct address_space	*assoc_mapping;	/* ditto */
648 	struct mutex		unmap_mutex;    /* to protect unmapping */
649 } __attribute__((aligned(sizeof(long))));
650 	/*
651 	 * On most architectures that alignment is already the case; but
652 	 * must be enforced here for CRIS, to let the least significant bit
653 	 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
654 	 */
655 
656 struct block_device {
657 	dev_t			bd_dev;  /* not a kdev_t - it's a search key */
658 	int			bd_openers;
659 	struct inode *		bd_inode;	/* will die */
660 	struct super_block *	bd_super;
661 	struct mutex		bd_mutex;	/* open/close mutex */
662 	struct list_head	bd_inodes;
663 	void *			bd_claiming;
664 	void *			bd_holder;
665 	int			bd_holders;
666 	bool			bd_write_holder;
667 #ifdef CONFIG_SYSFS
668 	struct list_head	bd_holder_disks;
669 #endif
670 	struct block_device *	bd_contains;
671 	unsigned		bd_block_size;
672 	struct hd_struct *	bd_part;
673 	/* number of times partitions within this device have been opened. */
674 	unsigned		bd_part_count;
675 	int			bd_invalidated;
676 	struct gendisk *	bd_disk;
677 	struct list_head	bd_list;
678 	/*
679 	 * Private data.  You must have bd_claim'ed the block_device
680 	 * to use this.  NOTE:  bd_claim allows an owner to claim
681 	 * the same device multiple times, the owner must take special
682 	 * care to not mess up bd_private for that case.
683 	 */
684 	unsigned long		bd_private;
685 
686 	/* The counter of freeze processes */
687 	int			bd_fsfreeze_count;
688 	/* Mutex for freeze */
689 	struct mutex		bd_fsfreeze_mutex;
690 };
691 
692 /*
693  * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
694  * radix trees
695  */
696 #define PAGECACHE_TAG_DIRTY	0
697 #define PAGECACHE_TAG_WRITEBACK	1
698 #define PAGECACHE_TAG_TOWRITE	2
699 
700 int mapping_tagged(struct address_space *mapping, int tag);
701 
702 /*
703  * Might pages of this file be mapped into userspace?
704  */
705 static inline int mapping_mapped(struct address_space *mapping)
706 {
707 	return	!prio_tree_empty(&mapping->i_mmap) ||
708 		!list_empty(&mapping->i_mmap_nonlinear);
709 }
710 
711 /*
712  * Might pages of this file have been modified in userspace?
713  * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
714  * marks vma as VM_SHARED if it is shared, and the file was opened for
715  * writing i.e. vma may be mprotected writable even if now readonly.
716  */
717 static inline int mapping_writably_mapped(struct address_space *mapping)
718 {
719 	return mapping->i_mmap_writable != 0;
720 }
721 
722 /*
723  * Use sequence counter to get consistent i_size on 32-bit processors.
724  */
725 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
726 #include <linux/seqlock.h>
727 #define __NEED_I_SIZE_ORDERED
728 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
729 #else
730 #define i_size_ordered_init(inode) do { } while (0)
731 #endif
732 
733 struct posix_acl;
734 #define ACL_NOT_CACHED ((void *)(-1))
735 
736 struct inode {
737 	/* RCU path lookup touches following: */
738 	umode_t			i_mode;
739 	uid_t			i_uid;
740 	gid_t			i_gid;
741 	const struct inode_operations	*i_op;
742 	struct super_block	*i_sb;
743 
744 	spinlock_t		i_lock;	/* i_blocks, i_bytes, maybe i_size */
745 	unsigned int		i_flags;
746 	struct mutex		i_mutex;
747 
748 	unsigned long		i_state;
749 	unsigned long		dirtied_when;	/* jiffies of first dirtying */
750 
751 	struct hlist_node	i_hash;
752 	struct list_head	i_wb_list;	/* backing dev IO list */
753 	struct list_head	i_lru;		/* inode LRU list */
754 	struct list_head	i_sb_list;
755 	union {
756 		struct list_head	i_dentry;
757 		struct rcu_head		i_rcu;
758 	};
759 	unsigned long		i_ino;
760 	atomic_t		i_count;
761 	unsigned int		i_nlink;
762 	dev_t			i_rdev;
763 	unsigned int		i_blkbits;
764 	u64			i_version;
765 	loff_t			i_size;
766 #ifdef __NEED_I_SIZE_ORDERED
767 	seqcount_t		i_size_seqcount;
768 #endif
769 	struct timespec		i_atime;
770 	struct timespec		i_mtime;
771 	struct timespec		i_ctime;
772 	blkcnt_t		i_blocks;
773 	unsigned short          i_bytes;
774 	struct rw_semaphore	i_alloc_sem;
775 	const struct file_operations	*i_fop;	/* former ->i_op->default_file_ops */
776 	struct file_lock	*i_flock;
777 	struct address_space	*i_mapping;
778 	struct address_space	i_data;
779 #ifdef CONFIG_QUOTA
780 	struct dquot		*i_dquot[MAXQUOTAS];
781 #endif
782 	struct list_head	i_devices;
783 	union {
784 		struct pipe_inode_info	*i_pipe;
785 		struct block_device	*i_bdev;
786 		struct cdev		*i_cdev;
787 	};
788 
789 	__u32			i_generation;
790 
791 #ifdef CONFIG_FSNOTIFY
792 	__u32			i_fsnotify_mask; /* all events this inode cares about */
793 	struct hlist_head	i_fsnotify_marks;
794 #endif
795 
796 #ifdef CONFIG_IMA
797 	atomic_t		i_readcount; /* struct files open RO */
798 #endif
799 	atomic_t		i_writecount;
800 #ifdef CONFIG_SECURITY
801 	void			*i_security;
802 #endif
803 #ifdef CONFIG_FS_POSIX_ACL
804 	struct posix_acl	*i_acl;
805 	struct posix_acl	*i_default_acl;
806 #endif
807 	void			*i_private; /* fs or device private pointer */
808 };
809 
810 static inline int inode_unhashed(struct inode *inode)
811 {
812 	return hlist_unhashed(&inode->i_hash);
813 }
814 
815 /*
816  * inode->i_mutex nesting subclasses for the lock validator:
817  *
818  * 0: the object of the current VFS operation
819  * 1: parent
820  * 2: child/target
821  * 3: quota file
822  *
823  * The locking order between these classes is
824  * parent -> child -> normal -> xattr -> quota
825  */
826 enum inode_i_mutex_lock_class
827 {
828 	I_MUTEX_NORMAL,
829 	I_MUTEX_PARENT,
830 	I_MUTEX_CHILD,
831 	I_MUTEX_XATTR,
832 	I_MUTEX_QUOTA
833 };
834 
835 /*
836  * NOTE: in a 32bit arch with a preemptable kernel and
837  * an UP compile the i_size_read/write must be atomic
838  * with respect to the local cpu (unlike with preempt disabled),
839  * but they don't need to be atomic with respect to other cpus like in
840  * true SMP (so they need either to either locally disable irq around
841  * the read or for example on x86 they can be still implemented as a
842  * cmpxchg8b without the need of the lock prefix). For SMP compiles
843  * and 64bit archs it makes no difference if preempt is enabled or not.
844  */
845 static inline loff_t i_size_read(const struct inode *inode)
846 {
847 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
848 	loff_t i_size;
849 	unsigned int seq;
850 
851 	do {
852 		seq = read_seqcount_begin(&inode->i_size_seqcount);
853 		i_size = inode->i_size;
854 	} while (read_seqcount_retry(&inode->i_size_seqcount, seq));
855 	return i_size;
856 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
857 	loff_t i_size;
858 
859 	preempt_disable();
860 	i_size = inode->i_size;
861 	preempt_enable();
862 	return i_size;
863 #else
864 	return inode->i_size;
865 #endif
866 }
867 
868 /*
869  * NOTE: unlike i_size_read(), i_size_write() does need locking around it
870  * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
871  * can be lost, resulting in subsequent i_size_read() calls spinning forever.
872  */
873 static inline void i_size_write(struct inode *inode, loff_t i_size)
874 {
875 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
876 	write_seqcount_begin(&inode->i_size_seqcount);
877 	inode->i_size = i_size;
878 	write_seqcount_end(&inode->i_size_seqcount);
879 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
880 	preempt_disable();
881 	inode->i_size = i_size;
882 	preempt_enable();
883 #else
884 	inode->i_size = i_size;
885 #endif
886 }
887 
888 static inline unsigned iminor(const struct inode *inode)
889 {
890 	return MINOR(inode->i_rdev);
891 }
892 
893 static inline unsigned imajor(const struct inode *inode)
894 {
895 	return MAJOR(inode->i_rdev);
896 }
897 
898 extern struct block_device *I_BDEV(struct inode *inode);
899 
900 struct fown_struct {
901 	rwlock_t lock;          /* protects pid, uid, euid fields */
902 	struct pid *pid;	/* pid or -pgrp where SIGIO should be sent */
903 	enum pid_type pid_type;	/* Kind of process group SIGIO should be sent to */
904 	uid_t uid, euid;	/* uid/euid of process setting the owner */
905 	int signum;		/* posix.1b rt signal to be delivered on IO */
906 };
907 
908 /*
909  * Track a single file's readahead state
910  */
911 struct file_ra_state {
912 	pgoff_t start;			/* where readahead started */
913 	unsigned int size;		/* # of readahead pages */
914 	unsigned int async_size;	/* do asynchronous readahead when
915 					   there are only # of pages ahead */
916 
917 	unsigned int ra_pages;		/* Maximum readahead window */
918 	unsigned int mmap_miss;		/* Cache miss stat for mmap accesses */
919 	loff_t prev_pos;		/* Cache last read() position */
920 };
921 
922 /*
923  * Check if @index falls in the readahead windows.
924  */
925 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
926 {
927 	return (index >= ra->start &&
928 		index <  ra->start + ra->size);
929 }
930 
931 #define FILE_MNT_WRITE_TAKEN	1
932 #define FILE_MNT_WRITE_RELEASED	2
933 
934 struct file {
935 	/*
936 	 * fu_list becomes invalid after file_free is called and queued via
937 	 * fu_rcuhead for RCU freeing
938 	 */
939 	union {
940 		struct list_head	fu_list;
941 		struct rcu_head 	fu_rcuhead;
942 	} f_u;
943 	struct path		f_path;
944 #define f_dentry	f_path.dentry
945 #define f_vfsmnt	f_path.mnt
946 	const struct file_operations	*f_op;
947 	spinlock_t		f_lock;  /* f_ep_links, f_flags, no IRQ */
948 #ifdef CONFIG_SMP
949 	int			f_sb_list_cpu;
950 #endif
951 	atomic_long_t		f_count;
952 	unsigned int 		f_flags;
953 	fmode_t			f_mode;
954 	loff_t			f_pos;
955 	struct fown_struct	f_owner;
956 	const struct cred	*f_cred;
957 	struct file_ra_state	f_ra;
958 
959 	u64			f_version;
960 #ifdef CONFIG_SECURITY
961 	void			*f_security;
962 #endif
963 	/* needed for tty driver, and maybe others */
964 	void			*private_data;
965 
966 #ifdef CONFIG_EPOLL
967 	/* Used by fs/eventpoll.c to link all the hooks to this file */
968 	struct list_head	f_ep_links;
969 #endif /* #ifdef CONFIG_EPOLL */
970 	struct address_space	*f_mapping;
971 #ifdef CONFIG_DEBUG_WRITECOUNT
972 	unsigned long f_mnt_write_state;
973 #endif
974 };
975 
976 struct file_handle {
977 	__u32 handle_bytes;
978 	int handle_type;
979 	/* file identifier */
980 	unsigned char f_handle[0];
981 };
982 
983 #define get_file(x)	atomic_long_inc(&(x)->f_count)
984 #define fput_atomic(x)	atomic_long_add_unless(&(x)->f_count, -1, 1)
985 #define file_count(x)	atomic_long_read(&(x)->f_count)
986 
987 #ifdef CONFIG_DEBUG_WRITECOUNT
988 static inline void file_take_write(struct file *f)
989 {
990 	WARN_ON(f->f_mnt_write_state != 0);
991 	f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
992 }
993 static inline void file_release_write(struct file *f)
994 {
995 	f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
996 }
997 static inline void file_reset_write(struct file *f)
998 {
999 	f->f_mnt_write_state = 0;
1000 }
1001 static inline void file_check_state(struct file *f)
1002 {
1003 	/*
1004 	 * At this point, either both or neither of these bits
1005 	 * should be set.
1006 	 */
1007 	WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
1008 	WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
1009 }
1010 static inline int file_check_writeable(struct file *f)
1011 {
1012 	if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
1013 		return 0;
1014 	printk(KERN_WARNING "writeable file with no "
1015 			    "mnt_want_write()\n");
1016 	WARN_ON(1);
1017 	return -EINVAL;
1018 }
1019 #else /* !CONFIG_DEBUG_WRITECOUNT */
1020 static inline void file_take_write(struct file *filp) {}
1021 static inline void file_release_write(struct file *filp) {}
1022 static inline void file_reset_write(struct file *filp) {}
1023 static inline void file_check_state(struct file *filp) {}
1024 static inline int file_check_writeable(struct file *filp)
1025 {
1026 	return 0;
1027 }
1028 #endif /* CONFIG_DEBUG_WRITECOUNT */
1029 
1030 #define	MAX_NON_LFS	((1UL<<31) - 1)
1031 
1032 /* Page cache limit. The filesystems should put that into their s_maxbytes
1033    limits, otherwise bad things can happen in VM. */
1034 #if BITS_PER_LONG==32
1035 #define MAX_LFS_FILESIZE	(((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
1036 #elif BITS_PER_LONG==64
1037 #define MAX_LFS_FILESIZE 	0x7fffffffffffffffUL
1038 #endif
1039 
1040 #define FL_POSIX	1
1041 #define FL_FLOCK	2
1042 #define FL_ACCESS	8	/* not trying to lock, just looking */
1043 #define FL_EXISTS	16	/* when unlocking, test for existence */
1044 #define FL_LEASE	32	/* lease held on this file */
1045 #define FL_CLOSE	64	/* unlock on close */
1046 #define FL_SLEEP	128	/* A blocking lock */
1047 
1048 /*
1049  * Special return value from posix_lock_file() and vfs_lock_file() for
1050  * asynchronous locking.
1051  */
1052 #define FILE_LOCK_DEFERRED 1
1053 
1054 /*
1055  * The POSIX file lock owner is determined by
1056  * the "struct files_struct" in the thread group
1057  * (or NULL for no owner - BSD locks).
1058  *
1059  * Lockd stuffs a "host" pointer into this.
1060  */
1061 typedef struct files_struct *fl_owner_t;
1062 
1063 struct file_lock_operations {
1064 	void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1065 	void (*fl_release_private)(struct file_lock *);
1066 };
1067 
1068 struct lock_manager_operations {
1069 	int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
1070 	void (*fl_notify)(struct file_lock *);	/* unblock callback */
1071 	int (*fl_grant)(struct file_lock *, struct file_lock *, int);
1072 	void (*fl_release_private)(struct file_lock *);
1073 	void (*fl_break)(struct file_lock *);
1074 	int (*fl_change)(struct file_lock **, int);
1075 };
1076 
1077 struct lock_manager {
1078 	struct list_head list;
1079 };
1080 
1081 void locks_start_grace(struct lock_manager *);
1082 void locks_end_grace(struct lock_manager *);
1083 int locks_in_grace(void);
1084 
1085 /* that will die - we need it for nfs_lock_info */
1086 #include <linux/nfs_fs_i.h>
1087 
1088 struct file_lock {
1089 	struct file_lock *fl_next;	/* singly linked list for this inode  */
1090 	struct list_head fl_link;	/* doubly linked list of all locks */
1091 	struct list_head fl_block;	/* circular list of blocked processes */
1092 	fl_owner_t fl_owner;
1093 	unsigned char fl_flags;
1094 	unsigned char fl_type;
1095 	unsigned int fl_pid;
1096 	struct pid *fl_nspid;
1097 	wait_queue_head_t fl_wait;
1098 	struct file *fl_file;
1099 	loff_t fl_start;
1100 	loff_t fl_end;
1101 
1102 	struct fasync_struct *	fl_fasync; /* for lease break notifications */
1103 	unsigned long fl_break_time;	/* for nonblocking lease breaks */
1104 
1105 	const struct file_lock_operations *fl_ops;	/* Callbacks for filesystems */
1106 	const struct lock_manager_operations *fl_lmops;	/* Callbacks for lockmanagers */
1107 	union {
1108 		struct nfs_lock_info	nfs_fl;
1109 		struct nfs4_lock_info	nfs4_fl;
1110 		struct {
1111 			struct list_head link;	/* link in AFS vnode's pending_locks list */
1112 			int state;		/* state of grant or error if -ve */
1113 		} afs;
1114 	} fl_u;
1115 };
1116 
1117 /* The following constant reflects the upper bound of the file/locking space */
1118 #ifndef OFFSET_MAX
1119 #define INT_LIMIT(x)	(~((x)1 << (sizeof(x)*8 - 1)))
1120 #define OFFSET_MAX	INT_LIMIT(loff_t)
1121 #define OFFT_OFFSET_MAX	INT_LIMIT(off_t)
1122 #endif
1123 
1124 #include <linux/fcntl.h>
1125 
1126 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1127 
1128 #ifdef CONFIG_FILE_LOCKING
1129 extern int fcntl_getlk(struct file *, struct flock __user *);
1130 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1131 			struct flock __user *);
1132 
1133 #if BITS_PER_LONG == 32
1134 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
1135 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1136 			struct flock64 __user *);
1137 #endif
1138 
1139 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1140 extern int fcntl_getlease(struct file *filp);
1141 
1142 /* fs/locks.c */
1143 void locks_free_lock(struct file_lock *fl);
1144 extern void locks_init_lock(struct file_lock *);
1145 extern struct file_lock * locks_alloc_lock(void);
1146 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1147 extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
1148 extern void locks_remove_posix(struct file *, fl_owner_t);
1149 extern void locks_remove_flock(struct file *);
1150 extern void locks_release_private(struct file_lock *);
1151 extern void posix_test_lock(struct file *, struct file_lock *);
1152 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1153 extern int posix_lock_file_wait(struct file *, struct file_lock *);
1154 extern int posix_unblock_lock(struct file *, struct file_lock *);
1155 extern int vfs_test_lock(struct file *, struct file_lock *);
1156 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1157 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1158 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1159 extern int __break_lease(struct inode *inode, unsigned int flags);
1160 extern void lease_get_mtime(struct inode *, struct timespec *time);
1161 extern int generic_setlease(struct file *, long, struct file_lock **);
1162 extern int vfs_setlease(struct file *, long, struct file_lock **);
1163 extern int lease_modify(struct file_lock **, int);
1164 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1165 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1166 extern void lock_flocks(void);
1167 extern void unlock_flocks(void);
1168 #else /* !CONFIG_FILE_LOCKING */
1169 static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1170 {
1171 	return -EINVAL;
1172 }
1173 
1174 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1175 			      unsigned int cmd, struct flock __user *user)
1176 {
1177 	return -EACCES;
1178 }
1179 
1180 #if BITS_PER_LONG == 32
1181 static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1182 {
1183 	return -EINVAL;
1184 }
1185 
1186 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1187 				unsigned int cmd, struct flock64 __user *user)
1188 {
1189 	return -EACCES;
1190 }
1191 #endif
1192 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1193 {
1194 	return 0;
1195 }
1196 
1197 static inline int fcntl_getlease(struct file *filp)
1198 {
1199 	return 0;
1200 }
1201 
1202 static inline void locks_init_lock(struct file_lock *fl)
1203 {
1204 	return;
1205 }
1206 
1207 static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1208 {
1209 	return;
1210 }
1211 
1212 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1213 {
1214 	return;
1215 }
1216 
1217 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1218 {
1219 	return;
1220 }
1221 
1222 static inline void locks_remove_flock(struct file *filp)
1223 {
1224 	return;
1225 }
1226 
1227 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1228 {
1229 	return;
1230 }
1231 
1232 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1233 				  struct file_lock *conflock)
1234 {
1235 	return -ENOLCK;
1236 }
1237 
1238 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1239 {
1240 	return -ENOLCK;
1241 }
1242 
1243 static inline int posix_unblock_lock(struct file *filp,
1244 				     struct file_lock *waiter)
1245 {
1246 	return -ENOENT;
1247 }
1248 
1249 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1250 {
1251 	return 0;
1252 }
1253 
1254 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1255 				struct file_lock *fl, struct file_lock *conf)
1256 {
1257 	return -ENOLCK;
1258 }
1259 
1260 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1261 {
1262 	return 0;
1263 }
1264 
1265 static inline int flock_lock_file_wait(struct file *filp,
1266 				       struct file_lock *request)
1267 {
1268 	return -ENOLCK;
1269 }
1270 
1271 static inline int __break_lease(struct inode *inode, unsigned int mode)
1272 {
1273 	return 0;
1274 }
1275 
1276 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1277 {
1278 	return;
1279 }
1280 
1281 static inline int generic_setlease(struct file *filp, long arg,
1282 				    struct file_lock **flp)
1283 {
1284 	return -EINVAL;
1285 }
1286 
1287 static inline int vfs_setlease(struct file *filp, long arg,
1288 			       struct file_lock **lease)
1289 {
1290 	return -EINVAL;
1291 }
1292 
1293 static inline int lease_modify(struct file_lock **before, int arg)
1294 {
1295 	return -EINVAL;
1296 }
1297 
1298 static inline int lock_may_read(struct inode *inode, loff_t start,
1299 				unsigned long len)
1300 {
1301 	return 1;
1302 }
1303 
1304 static inline int lock_may_write(struct inode *inode, loff_t start,
1305 				 unsigned long len)
1306 {
1307 	return 1;
1308 }
1309 
1310 static inline void lock_flocks(void)
1311 {
1312 }
1313 
1314 static inline void unlock_flocks(void)
1315 {
1316 }
1317 
1318 #endif /* !CONFIG_FILE_LOCKING */
1319 
1320 
1321 struct fasync_struct {
1322 	spinlock_t		fa_lock;
1323 	int			magic;
1324 	int			fa_fd;
1325 	struct fasync_struct	*fa_next; /* singly linked list */
1326 	struct file		*fa_file;
1327 	struct rcu_head		fa_rcu;
1328 };
1329 
1330 #define FASYNC_MAGIC 0x4601
1331 
1332 /* SMP safe fasync helpers: */
1333 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1334 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1335 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1336 extern struct fasync_struct *fasync_alloc(void);
1337 extern void fasync_free(struct fasync_struct *);
1338 
1339 /* can be called from interrupts */
1340 extern void kill_fasync(struct fasync_struct **, int, int);
1341 
1342 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1343 extern int f_setown(struct file *filp, unsigned long arg, int force);
1344 extern void f_delown(struct file *filp);
1345 extern pid_t f_getown(struct file *filp);
1346 extern int send_sigurg(struct fown_struct *fown);
1347 
1348 /*
1349  *	Umount options
1350  */
1351 
1352 #define MNT_FORCE	0x00000001	/* Attempt to forcibily umount */
1353 #define MNT_DETACH	0x00000002	/* Just detach from the tree */
1354 #define MNT_EXPIRE	0x00000004	/* Mark for expiry */
1355 #define UMOUNT_NOFOLLOW	0x00000008	/* Don't follow symlink on umount */
1356 #define UMOUNT_UNUSED	0x80000000	/* Flag guaranteed to be unused */
1357 
1358 extern struct list_head super_blocks;
1359 extern spinlock_t sb_lock;
1360 
1361 struct super_block {
1362 	struct list_head	s_list;		/* Keep this first */
1363 	dev_t			s_dev;		/* search index; _not_ kdev_t */
1364 	unsigned char		s_dirt;
1365 	unsigned char		s_blocksize_bits;
1366 	unsigned long		s_blocksize;
1367 	loff_t			s_maxbytes;	/* Max file size */
1368 	struct file_system_type	*s_type;
1369 	const struct super_operations	*s_op;
1370 	const struct dquot_operations	*dq_op;
1371 	const struct quotactl_ops	*s_qcop;
1372 	const struct export_operations *s_export_op;
1373 	unsigned long		s_flags;
1374 	unsigned long		s_magic;
1375 	struct dentry		*s_root;
1376 	struct rw_semaphore	s_umount;
1377 	struct mutex		s_lock;
1378 	int			s_count;
1379 	atomic_t		s_active;
1380 #ifdef CONFIG_SECURITY
1381 	void                    *s_security;
1382 #endif
1383 	const struct xattr_handler **s_xattr;
1384 
1385 	struct list_head	s_inodes;	/* all inodes */
1386 	struct hlist_bl_head	s_anon;		/* anonymous dentries for (nfs) exporting */
1387 #ifdef CONFIG_SMP
1388 	struct list_head __percpu *s_files;
1389 #else
1390 	struct list_head	s_files;
1391 #endif
1392 	/* s_dentry_lru, s_nr_dentry_unused protected by dcache.c lru locks */
1393 	struct list_head	s_dentry_lru;	/* unused dentry lru */
1394 	int			s_nr_dentry_unused;	/* # of dentry on lru */
1395 
1396 	struct block_device	*s_bdev;
1397 	struct backing_dev_info *s_bdi;
1398 	struct mtd_info		*s_mtd;
1399 	struct list_head	s_instances;
1400 	struct quota_info	s_dquot;	/* Diskquota specific options */
1401 
1402 	int			s_frozen;
1403 	wait_queue_head_t	s_wait_unfrozen;
1404 
1405 	char s_id[32];				/* Informational name */
1406 	u8 s_uuid[16];				/* UUID */
1407 
1408 	void 			*s_fs_info;	/* Filesystem private info */
1409 	fmode_t			s_mode;
1410 
1411 	/* Granularity of c/m/atime in ns.
1412 	   Cannot be worse than a second */
1413 	u32		   s_time_gran;
1414 
1415 	/*
1416 	 * The next field is for VFS *only*. No filesystems have any business
1417 	 * even looking at it. You had been warned.
1418 	 */
1419 	struct mutex s_vfs_rename_mutex;	/* Kludge */
1420 
1421 	/*
1422 	 * Filesystem subtype.  If non-empty the filesystem type field
1423 	 * in /proc/mounts will be "type.subtype"
1424 	 */
1425 	char *s_subtype;
1426 
1427 	/*
1428 	 * Saved mount options for lazy filesystems using
1429 	 * generic_show_options()
1430 	 */
1431 	char __rcu *s_options;
1432 	const struct dentry_operations *s_d_op; /* default d_op for dentries */
1433 };
1434 
1435 extern struct timespec current_fs_time(struct super_block *sb);
1436 
1437 /*
1438  * Snapshotting support.
1439  */
1440 enum {
1441 	SB_UNFROZEN = 0,
1442 	SB_FREEZE_WRITE	= 1,
1443 	SB_FREEZE_TRANS = 2,
1444 };
1445 
1446 #define vfs_check_frozen(sb, level) \
1447 	wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
1448 
1449 #define get_fs_excl() atomic_inc(&current->fs_excl)
1450 #define put_fs_excl() atomic_dec(&current->fs_excl)
1451 #define has_fs_excl() atomic_read(&current->fs_excl)
1452 
1453 /*
1454  * until VFS tracks user namespaces for inodes, just make all files
1455  * belong to init_user_ns
1456  */
1457 extern struct user_namespace init_user_ns;
1458 #define inode_userns(inode) (&init_user_ns)
1459 extern bool inode_owner_or_capable(const struct inode *inode);
1460 
1461 /* not quite ready to be deprecated, but... */
1462 extern void lock_super(struct super_block *);
1463 extern void unlock_super(struct super_block *);
1464 
1465 /*
1466  * VFS helper functions..
1467  */
1468 extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
1469 extern int vfs_mkdir(struct inode *, struct dentry *, int);
1470 extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
1471 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1472 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1473 extern int vfs_rmdir(struct inode *, struct dentry *);
1474 extern int vfs_unlink(struct inode *, struct dentry *);
1475 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1476 
1477 /*
1478  * VFS dentry helper functions.
1479  */
1480 extern void dentry_unhash(struct dentry *dentry);
1481 
1482 /*
1483  * VFS file helper functions.
1484  */
1485 extern int file_permission(struct file *, int);
1486 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1487 			mode_t mode);
1488 /*
1489  * VFS FS_IOC_FIEMAP helper definitions.
1490  */
1491 struct fiemap_extent_info {
1492 	unsigned int fi_flags;		/* Flags as passed from user */
1493 	unsigned int fi_extents_mapped;	/* Number of mapped extents */
1494 	unsigned int fi_extents_max;	/* Size of fiemap_extent array */
1495 	struct fiemap_extent __user *fi_extents_start; /* Start of
1496 							fiemap_extent array */
1497 };
1498 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1499 			    u64 phys, u64 len, u32 flags);
1500 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1501 
1502 /*
1503  * File types
1504  *
1505  * NOTE! These match bits 12..15 of stat.st_mode
1506  * (ie "(i_mode >> 12) & 15").
1507  */
1508 #define DT_UNKNOWN	0
1509 #define DT_FIFO		1
1510 #define DT_CHR		2
1511 #define DT_DIR		4
1512 #define DT_BLK		6
1513 #define DT_REG		8
1514 #define DT_LNK		10
1515 #define DT_SOCK		12
1516 #define DT_WHT		14
1517 
1518 /*
1519  * This is the "filldir" function type, used by readdir() to let
1520  * the kernel specify what kind of dirent layout it wants to have.
1521  * This allows the kernel to read directories into kernel space or
1522  * to have different dirent layouts depending on the binary type.
1523  */
1524 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1525 struct block_device_operations;
1526 
1527 /* These macros are for out of kernel modules to test that
1528  * the kernel supports the unlocked_ioctl and compat_ioctl
1529  * fields in struct file_operations. */
1530 #define HAVE_COMPAT_IOCTL 1
1531 #define HAVE_UNLOCKED_IOCTL 1
1532 
1533 /*
1534  * NOTE:
1535  * all file operations except setlease can be called without
1536  * the big kernel lock held in all filesystems.
1537  */
1538 struct file_operations {
1539 	struct module *owner;
1540 	loff_t (*llseek) (struct file *, loff_t, int);
1541 	ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1542 	ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1543 	ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1544 	ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1545 	int (*readdir) (struct file *, void *, filldir_t);
1546 	unsigned int (*poll) (struct file *, struct poll_table_struct *);
1547 	long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1548 	long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1549 	int (*mmap) (struct file *, struct vm_area_struct *);
1550 	int (*open) (struct inode *, struct file *);
1551 	int (*flush) (struct file *, fl_owner_t id);
1552 	int (*release) (struct inode *, struct file *);
1553 	int (*fsync) (struct file *, int datasync);
1554 	int (*aio_fsync) (struct kiocb *, int datasync);
1555 	int (*fasync) (int, struct file *, int);
1556 	int (*lock) (struct file *, int, struct file_lock *);
1557 	ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1558 	unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1559 	int (*check_flags)(int);
1560 	int (*flock) (struct file *, int, struct file_lock *);
1561 	ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1562 	ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1563 	int (*setlease)(struct file *, long, struct file_lock **);
1564 	long (*fallocate)(struct file *file, int mode, loff_t offset,
1565 			  loff_t len);
1566 };
1567 
1568 #define IPERM_FLAG_RCU	0x0001
1569 
1570 struct inode_operations {
1571 	struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1572 	void * (*follow_link) (struct dentry *, struct nameidata *);
1573 	int (*permission) (struct inode *, int, unsigned int);
1574 	int (*check_acl)(struct inode *, int, unsigned int);
1575 
1576 	int (*readlink) (struct dentry *, char __user *,int);
1577 	void (*put_link) (struct dentry *, struct nameidata *, void *);
1578 
1579 	int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1580 	int (*link) (struct dentry *,struct inode *,struct dentry *);
1581 	int (*unlink) (struct inode *,struct dentry *);
1582 	int (*symlink) (struct inode *,struct dentry *,const char *);
1583 	int (*mkdir) (struct inode *,struct dentry *,int);
1584 	int (*rmdir) (struct inode *,struct dentry *);
1585 	int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1586 	int (*rename) (struct inode *, struct dentry *,
1587 			struct inode *, struct dentry *);
1588 	void (*truncate) (struct inode *);
1589 	int (*setattr) (struct dentry *, struct iattr *);
1590 	int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1591 	int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1592 	ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1593 	ssize_t (*listxattr) (struct dentry *, char *, size_t);
1594 	int (*removexattr) (struct dentry *, const char *);
1595 	void (*truncate_range)(struct inode *, loff_t, loff_t);
1596 	int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1597 		      u64 len);
1598 } ____cacheline_aligned;
1599 
1600 struct seq_file;
1601 
1602 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1603 				unsigned long nr_segs, unsigned long fast_segs,
1604 				struct iovec *fast_pointer,
1605 				struct iovec **ret_pointer);
1606 
1607 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1608 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1609 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1610 		unsigned long, loff_t *);
1611 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1612 		unsigned long, loff_t *);
1613 
1614 struct super_operations {
1615    	struct inode *(*alloc_inode)(struct super_block *sb);
1616 	void (*destroy_inode)(struct inode *);
1617 
1618    	void (*dirty_inode) (struct inode *);
1619 	int (*write_inode) (struct inode *, struct writeback_control *wbc);
1620 	int (*drop_inode) (struct inode *);
1621 	void (*evict_inode) (struct inode *);
1622 	void (*put_super) (struct super_block *);
1623 	void (*write_super) (struct super_block *);
1624 	int (*sync_fs)(struct super_block *sb, int wait);
1625 	int (*freeze_fs) (struct super_block *);
1626 	int (*unfreeze_fs) (struct super_block *);
1627 	int (*statfs) (struct dentry *, struct kstatfs *);
1628 	int (*remount_fs) (struct super_block *, int *, char *);
1629 	void (*umount_begin) (struct super_block *);
1630 
1631 	int (*show_options)(struct seq_file *, struct vfsmount *);
1632 	int (*show_devname)(struct seq_file *, struct vfsmount *);
1633 	int (*show_path)(struct seq_file *, struct vfsmount *);
1634 	int (*show_stats)(struct seq_file *, struct vfsmount *);
1635 #ifdef CONFIG_QUOTA
1636 	ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1637 	ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1638 #endif
1639 	int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1640 };
1641 
1642 /*
1643  * Inode state bits.  Protected by inode->i_lock
1644  *
1645  * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1646  * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1647  *
1648  * Four bits define the lifetime of an inode.  Initially, inodes are I_NEW,
1649  * until that flag is cleared.  I_WILL_FREE, I_FREEING and I_CLEAR are set at
1650  * various stages of removing an inode.
1651  *
1652  * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1653  *
1654  * I_DIRTY_SYNC		Inode is dirty, but doesn't have to be written on
1655  *			fdatasync().  i_atime is the usual cause.
1656  * I_DIRTY_DATASYNC	Data-related inode changes pending. We keep track of
1657  *			these changes separately from I_DIRTY_SYNC so that we
1658  *			don't have to write inode on fdatasync() when only
1659  *			mtime has changed in it.
1660  * I_DIRTY_PAGES	Inode has dirty pages.  Inode itself may be clean.
1661  * I_NEW		Serves as both a mutex and completion notification.
1662  *			New inodes set I_NEW.  If two processes both create
1663  *			the same inode, one of them will release its inode and
1664  *			wait for I_NEW to be released before returning.
1665  *			Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1666  *			also cause waiting on I_NEW, without I_NEW actually
1667  *			being set.  find_inode() uses this to prevent returning
1668  *			nearly-dead inodes.
1669  * I_WILL_FREE		Must be set when calling write_inode_now() if i_count
1670  *			is zero.  I_FREEING must be set when I_WILL_FREE is
1671  *			cleared.
1672  * I_FREEING		Set when inode is about to be freed but still has dirty
1673  *			pages or buffers attached or the inode itself is still
1674  *			dirty.
1675  * I_CLEAR		Added by end_writeback().  In this state the inode is clean
1676  *			and can be destroyed.  Inode keeps I_FREEING.
1677  *
1678  *			Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1679  *			prohibited for many purposes.  iget() must wait for
1680  *			the inode to be completely released, then create it
1681  *			anew.  Other functions will just ignore such inodes,
1682  *			if appropriate.  I_NEW is used for waiting.
1683  *
1684  * I_SYNC		Synchonized write of dirty inode data.  The bits is
1685  *			set during data writeback, and cleared with a wakeup
1686  *			on the bit address once it is done.
1687  *
1688  * Q: What is the difference between I_WILL_FREE and I_FREEING?
1689  */
1690 #define I_DIRTY_SYNC		(1 << 0)
1691 #define I_DIRTY_DATASYNC	(1 << 1)
1692 #define I_DIRTY_PAGES		(1 << 2)
1693 #define __I_NEW			3
1694 #define I_NEW			(1 << __I_NEW)
1695 #define I_WILL_FREE		(1 << 4)
1696 #define I_FREEING		(1 << 5)
1697 #define I_CLEAR			(1 << 6)
1698 #define __I_SYNC		7
1699 #define I_SYNC			(1 << __I_SYNC)
1700 #define I_REFERENCED		(1 << 8)
1701 
1702 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1703 
1704 extern void __mark_inode_dirty(struct inode *, int);
1705 static inline void mark_inode_dirty(struct inode *inode)
1706 {
1707 	__mark_inode_dirty(inode, I_DIRTY);
1708 }
1709 
1710 static inline void mark_inode_dirty_sync(struct inode *inode)
1711 {
1712 	__mark_inode_dirty(inode, I_DIRTY_SYNC);
1713 }
1714 
1715 /**
1716  * inc_nlink - directly increment an inode's link count
1717  * @inode: inode
1718  *
1719  * This is a low-level filesystem helper to replace any
1720  * direct filesystem manipulation of i_nlink.  Currently,
1721  * it is only here for parity with dec_nlink().
1722  */
1723 static inline void inc_nlink(struct inode *inode)
1724 {
1725 	inode->i_nlink++;
1726 }
1727 
1728 static inline void inode_inc_link_count(struct inode *inode)
1729 {
1730 	inc_nlink(inode);
1731 	mark_inode_dirty(inode);
1732 }
1733 
1734 /**
1735  * drop_nlink - directly drop an inode's link count
1736  * @inode: inode
1737  *
1738  * This is a low-level filesystem helper to replace any
1739  * direct filesystem manipulation of i_nlink.  In cases
1740  * where we are attempting to track writes to the
1741  * filesystem, a decrement to zero means an imminent
1742  * write when the file is truncated and actually unlinked
1743  * on the filesystem.
1744  */
1745 static inline void drop_nlink(struct inode *inode)
1746 {
1747 	inode->i_nlink--;
1748 }
1749 
1750 /**
1751  * clear_nlink - directly zero an inode's link count
1752  * @inode: inode
1753  *
1754  * This is a low-level filesystem helper to replace any
1755  * direct filesystem manipulation of i_nlink.  See
1756  * drop_nlink() for why we care about i_nlink hitting zero.
1757  */
1758 static inline void clear_nlink(struct inode *inode)
1759 {
1760 	inode->i_nlink = 0;
1761 }
1762 
1763 static inline void inode_dec_link_count(struct inode *inode)
1764 {
1765 	drop_nlink(inode);
1766 	mark_inode_dirty(inode);
1767 }
1768 
1769 /**
1770  * inode_inc_iversion - increments i_version
1771  * @inode: inode that need to be updated
1772  *
1773  * Every time the inode is modified, the i_version field will be incremented.
1774  * The filesystem has to be mounted with i_version flag
1775  */
1776 
1777 static inline void inode_inc_iversion(struct inode *inode)
1778 {
1779        spin_lock(&inode->i_lock);
1780        inode->i_version++;
1781        spin_unlock(&inode->i_lock);
1782 }
1783 
1784 extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
1785 static inline void file_accessed(struct file *file)
1786 {
1787 	if (!(file->f_flags & O_NOATIME))
1788 		touch_atime(file->f_path.mnt, file->f_path.dentry);
1789 }
1790 
1791 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1792 int sync_inode_metadata(struct inode *inode, int wait);
1793 
1794 struct file_system_type {
1795 	const char *name;
1796 	int fs_flags;
1797 	struct dentry *(*mount) (struct file_system_type *, int,
1798 		       const char *, void *);
1799 	void (*kill_sb) (struct super_block *);
1800 	struct module *owner;
1801 	struct file_system_type * next;
1802 	struct list_head fs_supers;
1803 
1804 	struct lock_class_key s_lock_key;
1805 	struct lock_class_key s_umount_key;
1806 	struct lock_class_key s_vfs_rename_key;
1807 
1808 	struct lock_class_key i_lock_key;
1809 	struct lock_class_key i_mutex_key;
1810 	struct lock_class_key i_mutex_dir_key;
1811 	struct lock_class_key i_alloc_sem_key;
1812 };
1813 
1814 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1815 	void *data, int (*fill_super)(struct super_block *, void *, int));
1816 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1817 	int flags, const char *dev_name, void *data,
1818 	int (*fill_super)(struct super_block *, void *, int));
1819 extern struct dentry *mount_single(struct file_system_type *fs_type,
1820 	int flags, void *data,
1821 	int (*fill_super)(struct super_block *, void *, int));
1822 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1823 	int flags, void *data,
1824 	int (*fill_super)(struct super_block *, void *, int));
1825 void generic_shutdown_super(struct super_block *sb);
1826 void kill_block_super(struct super_block *sb);
1827 void kill_anon_super(struct super_block *sb);
1828 void kill_litter_super(struct super_block *sb);
1829 void deactivate_super(struct super_block *sb);
1830 void deactivate_locked_super(struct super_block *sb);
1831 int set_anon_super(struct super_block *s, void *data);
1832 struct super_block *sget(struct file_system_type *type,
1833 			int (*test)(struct super_block *,void *),
1834 			int (*set)(struct super_block *,void *),
1835 			void *data);
1836 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1837 	const struct super_operations *ops,
1838 	const struct dentry_operations *dops,
1839 	unsigned long);
1840 
1841 static inline void sb_mark_dirty(struct super_block *sb)
1842 {
1843 	sb->s_dirt = 1;
1844 }
1845 static inline void sb_mark_clean(struct super_block *sb)
1846 {
1847 	sb->s_dirt = 0;
1848 }
1849 static inline int sb_is_dirty(struct super_block *sb)
1850 {
1851 	return sb->s_dirt;
1852 }
1853 
1854 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1855 #define fops_get(fops) \
1856 	(((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1857 #define fops_put(fops) \
1858 	do { if (fops) module_put((fops)->owner); } while(0)
1859 
1860 extern int register_filesystem(struct file_system_type *);
1861 extern int unregister_filesystem(struct file_system_type *);
1862 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1863 #define kern_mount(type) kern_mount_data(type, NULL)
1864 extern int may_umount_tree(struct vfsmount *);
1865 extern int may_umount(struct vfsmount *);
1866 extern long do_mount(char *, char *, char *, unsigned long, void *);
1867 extern struct vfsmount *collect_mounts(struct path *);
1868 extern void drop_collected_mounts(struct vfsmount *);
1869 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1870 			  struct vfsmount *);
1871 extern int vfs_statfs(struct path *, struct kstatfs *);
1872 extern int user_statfs(const char __user *, struct kstatfs *);
1873 extern int fd_statfs(int, struct kstatfs *);
1874 extern int statfs_by_dentry(struct dentry *, struct kstatfs *);
1875 extern int freeze_super(struct super_block *super);
1876 extern int thaw_super(struct super_block *super);
1877 
1878 extern int current_umask(void);
1879 
1880 /* /sys/fs */
1881 extern struct kobject *fs_kobj;
1882 
1883 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1884 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1885 
1886 #define FLOCK_VERIFY_READ  1
1887 #define FLOCK_VERIFY_WRITE 2
1888 
1889 #ifdef CONFIG_FILE_LOCKING
1890 extern int locks_mandatory_locked(struct inode *);
1891 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1892 
1893 /*
1894  * Candidates for mandatory locking have the setgid bit set
1895  * but no group execute bit -  an otherwise meaningless combination.
1896  */
1897 
1898 static inline int __mandatory_lock(struct inode *ino)
1899 {
1900 	return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1901 }
1902 
1903 /*
1904  * ... and these candidates should be on MS_MANDLOCK mounted fs,
1905  * otherwise these will be advisory locks
1906  */
1907 
1908 static inline int mandatory_lock(struct inode *ino)
1909 {
1910 	return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1911 }
1912 
1913 static inline int locks_verify_locked(struct inode *inode)
1914 {
1915 	if (mandatory_lock(inode))
1916 		return locks_mandatory_locked(inode);
1917 	return 0;
1918 }
1919 
1920 static inline int locks_verify_truncate(struct inode *inode,
1921 				    struct file *filp,
1922 				    loff_t size)
1923 {
1924 	if (inode->i_flock && mandatory_lock(inode))
1925 		return locks_mandatory_area(
1926 			FLOCK_VERIFY_WRITE, inode, filp,
1927 			size < inode->i_size ? size : inode->i_size,
1928 			(size < inode->i_size ? inode->i_size - size
1929 			 : size - inode->i_size)
1930 		);
1931 	return 0;
1932 }
1933 
1934 static inline int break_lease(struct inode *inode, unsigned int mode)
1935 {
1936 	if (inode->i_flock)
1937 		return __break_lease(inode, mode);
1938 	return 0;
1939 }
1940 #else /* !CONFIG_FILE_LOCKING */
1941 static inline int locks_mandatory_locked(struct inode *inode)
1942 {
1943 	return 0;
1944 }
1945 
1946 static inline int locks_mandatory_area(int rw, struct inode *inode,
1947 				       struct file *filp, loff_t offset,
1948 				       size_t count)
1949 {
1950 	return 0;
1951 }
1952 
1953 static inline int __mandatory_lock(struct inode *inode)
1954 {
1955 	return 0;
1956 }
1957 
1958 static inline int mandatory_lock(struct inode *inode)
1959 {
1960 	return 0;
1961 }
1962 
1963 static inline int locks_verify_locked(struct inode *inode)
1964 {
1965 	return 0;
1966 }
1967 
1968 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
1969 					size_t size)
1970 {
1971 	return 0;
1972 }
1973 
1974 static inline int break_lease(struct inode *inode, unsigned int mode)
1975 {
1976 	return 0;
1977 }
1978 
1979 #endif /* CONFIG_FILE_LOCKING */
1980 
1981 /* fs/open.c */
1982 
1983 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
1984 		       struct file *filp);
1985 extern int do_fallocate(struct file *file, int mode, loff_t offset,
1986 			loff_t len);
1987 extern long do_sys_open(int dfd, const char __user *filename, int flags,
1988 			int mode);
1989 extern struct file *filp_open(const char *, int, int);
1990 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
1991 				   const char *, int);
1992 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int,
1993 				 const struct cred *);
1994 extern int filp_close(struct file *, fl_owner_t id);
1995 extern char * getname(const char __user *);
1996 
1997 /* fs/ioctl.c */
1998 
1999 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2000 
2001 /* fs/dcache.c */
2002 extern void __init vfs_caches_init_early(void);
2003 extern void __init vfs_caches_init(unsigned long);
2004 
2005 extern struct kmem_cache *names_cachep;
2006 
2007 #define __getname_gfp(gfp)	kmem_cache_alloc(names_cachep, (gfp))
2008 #define __getname()		__getname_gfp(GFP_KERNEL)
2009 #define __putname(name)		kmem_cache_free(names_cachep, (void *)(name))
2010 #ifndef CONFIG_AUDITSYSCALL
2011 #define putname(name)   __putname(name)
2012 #else
2013 extern void putname(const char *name);
2014 #endif
2015 
2016 #ifdef CONFIG_BLOCK
2017 extern int register_blkdev(unsigned int, const char *);
2018 extern void unregister_blkdev(unsigned int, const char *);
2019 extern struct block_device *bdget(dev_t);
2020 extern struct block_device *bdgrab(struct block_device *bdev);
2021 extern void bd_set_size(struct block_device *, loff_t size);
2022 extern void bd_forget(struct inode *inode);
2023 extern void bdput(struct block_device *);
2024 extern void invalidate_bdev(struct block_device *);
2025 extern int sync_blockdev(struct block_device *bdev);
2026 extern struct super_block *freeze_bdev(struct block_device *);
2027 extern void emergency_thaw_all(void);
2028 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2029 extern int fsync_bdev(struct block_device *);
2030 #else
2031 static inline void bd_forget(struct inode *inode) {}
2032 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2033 static inline void invalidate_bdev(struct block_device *bdev) {}
2034 
2035 static inline struct super_block *freeze_bdev(struct block_device *sb)
2036 {
2037 	return NULL;
2038 }
2039 
2040 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2041 {
2042 	return 0;
2043 }
2044 #endif
2045 extern int sync_filesystem(struct super_block *);
2046 extern const struct file_operations def_blk_fops;
2047 extern const struct file_operations def_chr_fops;
2048 extern const struct file_operations bad_sock_fops;
2049 extern const struct file_operations def_fifo_fops;
2050 #ifdef CONFIG_BLOCK
2051 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2052 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2053 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2054 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2055 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2056 					       void *holder);
2057 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2058 					      void *holder);
2059 extern int blkdev_put(struct block_device *bdev, fmode_t mode);
2060 #ifdef CONFIG_SYSFS
2061 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2062 extern void bd_unlink_disk_holder(struct block_device *bdev,
2063 				  struct gendisk *disk);
2064 #else
2065 static inline int bd_link_disk_holder(struct block_device *bdev,
2066 				      struct gendisk *disk)
2067 {
2068 	return 0;
2069 }
2070 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2071 					 struct gendisk *disk)
2072 {
2073 }
2074 #endif
2075 #endif
2076 
2077 /* fs/char_dev.c */
2078 #define CHRDEV_MAJOR_HASH_SIZE	255
2079 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2080 extern int register_chrdev_region(dev_t, unsigned, const char *);
2081 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2082 			     unsigned int count, const char *name,
2083 			     const struct file_operations *fops);
2084 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2085 				unsigned int count, const char *name);
2086 extern void unregister_chrdev_region(dev_t, unsigned);
2087 extern void chrdev_show(struct seq_file *,off_t);
2088 
2089 static inline int register_chrdev(unsigned int major, const char *name,
2090 				  const struct file_operations *fops)
2091 {
2092 	return __register_chrdev(major, 0, 256, name, fops);
2093 }
2094 
2095 static inline void unregister_chrdev(unsigned int major, const char *name)
2096 {
2097 	__unregister_chrdev(major, 0, 256, name);
2098 }
2099 
2100 /* fs/block_dev.c */
2101 #define BDEVNAME_SIZE	32	/* Largest string for a blockdev identifier */
2102 #define BDEVT_SIZE	10	/* Largest string for MAJ:MIN for blkdev */
2103 
2104 #ifdef CONFIG_BLOCK
2105 #define BLKDEV_MAJOR_HASH_SIZE	255
2106 extern const char *__bdevname(dev_t, char *buffer);
2107 extern const char *bdevname(struct block_device *bdev, char *buffer);
2108 extern struct block_device *lookup_bdev(const char *);
2109 extern void blkdev_show(struct seq_file *,off_t);
2110 
2111 #else
2112 #define BLKDEV_MAJOR_HASH_SIZE	0
2113 #endif
2114 
2115 extern void init_special_inode(struct inode *, umode_t, dev_t);
2116 
2117 /* Invalid inode operations -- fs/bad_inode.c */
2118 extern void make_bad_inode(struct inode *);
2119 extern int is_bad_inode(struct inode *);
2120 
2121 extern const struct file_operations read_pipefifo_fops;
2122 extern const struct file_operations write_pipefifo_fops;
2123 extern const struct file_operations rdwr_pipefifo_fops;
2124 
2125 extern int fs_may_remount_ro(struct super_block *);
2126 
2127 #ifdef CONFIG_BLOCK
2128 /*
2129  * return READ, READA, or WRITE
2130  */
2131 #define bio_rw(bio)		((bio)->bi_rw & (RW_MASK | RWA_MASK))
2132 
2133 /*
2134  * return data direction, READ or WRITE
2135  */
2136 #define bio_data_dir(bio)	((bio)->bi_rw & 1)
2137 
2138 extern void check_disk_size_change(struct gendisk *disk,
2139 				   struct block_device *bdev);
2140 extern int revalidate_disk(struct gendisk *);
2141 extern int check_disk_change(struct block_device *);
2142 extern int __invalidate_device(struct block_device *, bool);
2143 extern int invalidate_partition(struct gendisk *, int);
2144 #endif
2145 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2146 					pgoff_t start, pgoff_t end);
2147 
2148 static inline void invalidate_remote_inode(struct inode *inode)
2149 {
2150 	if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2151 	    S_ISLNK(inode->i_mode))
2152 		invalidate_mapping_pages(inode->i_mapping, 0, -1);
2153 }
2154 extern int invalidate_inode_pages2(struct address_space *mapping);
2155 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2156 					 pgoff_t start, pgoff_t end);
2157 extern int write_inode_now(struct inode *, int);
2158 extern int filemap_fdatawrite(struct address_space *);
2159 extern int filemap_flush(struct address_space *);
2160 extern int filemap_fdatawait(struct address_space *);
2161 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2162 				   loff_t lend);
2163 extern int filemap_write_and_wait(struct address_space *mapping);
2164 extern int filemap_write_and_wait_range(struct address_space *mapping,
2165 				        loff_t lstart, loff_t lend);
2166 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2167 				loff_t start, loff_t end, int sync_mode);
2168 extern int filemap_fdatawrite_range(struct address_space *mapping,
2169 				loff_t start, loff_t end);
2170 
2171 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2172 			   int datasync);
2173 extern int vfs_fsync(struct file *file, int datasync);
2174 extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
2175 extern void sync_supers(void);
2176 extern void emergency_sync(void);
2177 extern void emergency_remount(void);
2178 #ifdef CONFIG_BLOCK
2179 extern sector_t bmap(struct inode *, sector_t);
2180 #endif
2181 extern int notify_change(struct dentry *, struct iattr *);
2182 extern int inode_permission(struct inode *, int);
2183 extern int generic_permission(struct inode *, int, unsigned int,
2184 		int (*check_acl)(struct inode *, int, unsigned int));
2185 
2186 static inline bool execute_ok(struct inode *inode)
2187 {
2188 	return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2189 }
2190 
2191 extern int get_write_access(struct inode *);
2192 extern int deny_write_access(struct file *);
2193 static inline void put_write_access(struct inode * inode)
2194 {
2195 	atomic_dec(&inode->i_writecount);
2196 }
2197 static inline void allow_write_access(struct file *file)
2198 {
2199 	if (file)
2200 		atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
2201 }
2202 #ifdef CONFIG_IMA
2203 static inline void i_readcount_dec(struct inode *inode)
2204 {
2205 	BUG_ON(!atomic_read(&inode->i_readcount));
2206 	atomic_dec(&inode->i_readcount);
2207 }
2208 static inline void i_readcount_inc(struct inode *inode)
2209 {
2210 	atomic_inc(&inode->i_readcount);
2211 }
2212 #else
2213 static inline void i_readcount_dec(struct inode *inode)
2214 {
2215 	return;
2216 }
2217 static inline void i_readcount_inc(struct inode *inode)
2218 {
2219 	return;
2220 }
2221 #endif
2222 extern int do_pipe_flags(int *, int);
2223 extern struct file *create_read_pipe(struct file *f, int flags);
2224 extern struct file *create_write_pipe(int flags);
2225 extern void free_write_pipe(struct file *);
2226 
2227 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2228 extern struct file * open_exec(const char *);
2229 
2230 /* fs/dcache.c -- generic fs support functions */
2231 extern int is_subdir(struct dentry *, struct dentry *);
2232 extern int path_is_under(struct path *, struct path *);
2233 extern ino_t find_inode_number(struct dentry *, struct qstr *);
2234 
2235 #include <linux/err.h>
2236 
2237 /* needed for stackable file system support */
2238 extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
2239 
2240 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
2241 
2242 extern int inode_init_always(struct super_block *, struct inode *);
2243 extern void inode_init_once(struct inode *);
2244 extern void address_space_init_once(struct address_space *mapping);
2245 extern void ihold(struct inode * inode);
2246 extern void iput(struct inode *);
2247 extern struct inode * igrab(struct inode *);
2248 extern ino_t iunique(struct super_block *, ino_t);
2249 extern int inode_needs_sync(struct inode *inode);
2250 extern int generic_delete_inode(struct inode *inode);
2251 extern int generic_drop_inode(struct inode *inode);
2252 
2253 extern struct inode *ilookup5_nowait(struct super_block *sb,
2254 		unsigned long hashval, int (*test)(struct inode *, void *),
2255 		void *data);
2256 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2257 		int (*test)(struct inode *, void *), void *data);
2258 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2259 
2260 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2261 extern struct inode * iget_locked(struct super_block *, unsigned long);
2262 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2263 extern int insert_inode_locked(struct inode *);
2264 extern void unlock_new_inode(struct inode *);
2265 extern unsigned int get_next_ino(void);
2266 
2267 extern void __iget(struct inode * inode);
2268 extern void iget_failed(struct inode *);
2269 extern void end_writeback(struct inode *);
2270 extern void __destroy_inode(struct inode *);
2271 extern struct inode *new_inode(struct super_block *);
2272 extern void free_inode_nonrcu(struct inode *inode);
2273 extern int should_remove_suid(struct dentry *);
2274 extern int file_remove_suid(struct file *);
2275 
2276 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2277 extern void remove_inode_hash(struct inode *);
2278 static inline void insert_inode_hash(struct inode *inode)
2279 {
2280 	__insert_inode_hash(inode, inode->i_ino);
2281 }
2282 extern void inode_sb_list_add(struct inode *inode);
2283 
2284 #ifdef CONFIG_BLOCK
2285 extern void submit_bio(int, struct bio *);
2286 extern int bdev_read_only(struct block_device *);
2287 #endif
2288 extern int set_blocksize(struct block_device *, int);
2289 extern int sb_set_blocksize(struct super_block *, int);
2290 extern int sb_min_blocksize(struct super_block *, int);
2291 
2292 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2293 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2294 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
2295 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2296 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2297 extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2298 		loff_t *);
2299 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2300 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2301 		unsigned long *, loff_t, loff_t *, size_t, size_t);
2302 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2303 		unsigned long, loff_t, loff_t *, size_t, ssize_t);
2304 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2305 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2306 extern int generic_segment_checks(const struct iovec *iov,
2307 		unsigned long *nr_segs, size_t *count, int access_flags);
2308 
2309 /* fs/block_dev.c */
2310 extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2311 				unsigned long nr_segs, loff_t pos);
2312 extern int blkdev_fsync(struct file *filp, int datasync);
2313 
2314 /* fs/splice.c */
2315 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2316 		struct pipe_inode_info *, size_t, unsigned int);
2317 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2318 		struct pipe_inode_info *, size_t, unsigned int);
2319 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
2320 		struct file *, loff_t *, size_t, unsigned int);
2321 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2322 		struct file *out, loff_t *, size_t len, unsigned int flags);
2323 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2324 		size_t len, unsigned int flags);
2325 
2326 extern void
2327 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2328 extern loff_t noop_llseek(struct file *file, loff_t offset, int origin);
2329 extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
2330 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
2331 extern loff_t generic_file_llseek_unlocked(struct file *file, loff_t offset,
2332 			int origin);
2333 extern int generic_file_open(struct inode * inode, struct file * filp);
2334 extern int nonseekable_open(struct inode * inode, struct file * filp);
2335 
2336 #ifdef CONFIG_FS_XIP
2337 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2338 			     loff_t *ppos);
2339 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2340 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2341 			      size_t len, loff_t *ppos);
2342 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2343 #else
2344 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2345 {
2346 	return 0;
2347 }
2348 #endif
2349 
2350 #ifdef CONFIG_BLOCK
2351 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2352 			    loff_t file_offset);
2353 
2354 enum {
2355 	/* need locking between buffered and direct access */
2356 	DIO_LOCKING	= 0x01,
2357 
2358 	/* filesystem does not support filling holes */
2359 	DIO_SKIP_HOLES	= 0x02,
2360 };
2361 
2362 void dio_end_io(struct bio *bio, int error);
2363 
2364 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2365 	struct block_device *bdev, const struct iovec *iov, loff_t offset,
2366 	unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2367 	dio_submit_t submit_io,	int flags);
2368 
2369 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2370 	struct inode *inode, struct block_device *bdev, const struct iovec *iov,
2371 	loff_t offset, unsigned long nr_segs, get_block_t get_block,
2372 	dio_iodone_t end_io)
2373 {
2374 	return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
2375 				    nr_segs, get_block, end_io, NULL,
2376 				    DIO_LOCKING | DIO_SKIP_HOLES);
2377 }
2378 #endif
2379 
2380 extern const struct file_operations generic_ro_fops;
2381 
2382 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2383 
2384 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2385 extern int vfs_follow_link(struct nameidata *, const char *);
2386 extern int page_readlink(struct dentry *, char __user *, int);
2387 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2388 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2389 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2390 		int nofs);
2391 extern int page_symlink(struct inode *inode, const char *symname, int len);
2392 extern const struct inode_operations page_symlink_inode_operations;
2393 extern int generic_readlink(struct dentry *, char __user *, int);
2394 extern void generic_fillattr(struct inode *, struct kstat *);
2395 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2396 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2397 void inode_add_bytes(struct inode *inode, loff_t bytes);
2398 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2399 loff_t inode_get_bytes(struct inode *inode);
2400 void inode_set_bytes(struct inode *inode, loff_t bytes);
2401 
2402 extern int vfs_readdir(struct file *, filldir_t, void *);
2403 
2404 extern int vfs_stat(const char __user *, struct kstat *);
2405 extern int vfs_lstat(const char __user *, struct kstat *);
2406 extern int vfs_fstat(unsigned int, struct kstat *);
2407 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2408 
2409 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2410 		    unsigned long arg);
2411 extern int __generic_block_fiemap(struct inode *inode,
2412 				  struct fiemap_extent_info *fieinfo,
2413 				  loff_t start, loff_t len,
2414 				  get_block_t *get_block);
2415 extern int generic_block_fiemap(struct inode *inode,
2416 				struct fiemap_extent_info *fieinfo, u64 start,
2417 				u64 len, get_block_t *get_block);
2418 
2419 extern void get_filesystem(struct file_system_type *fs);
2420 extern void put_filesystem(struct file_system_type *fs);
2421 extern struct file_system_type *get_fs_type(const char *name);
2422 extern struct super_block *get_super(struct block_device *);
2423 extern struct super_block *get_active_super(struct block_device *bdev);
2424 extern struct super_block *user_get_super(dev_t);
2425 extern void drop_super(struct super_block *sb);
2426 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2427 
2428 extern int dcache_dir_open(struct inode *, struct file *);
2429 extern int dcache_dir_close(struct inode *, struct file *);
2430 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2431 extern int dcache_readdir(struct file *, void *, filldir_t);
2432 extern int simple_setattr(struct dentry *, struct iattr *);
2433 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2434 extern int simple_statfs(struct dentry *, struct kstatfs *);
2435 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2436 extern int simple_unlink(struct inode *, struct dentry *);
2437 extern int simple_rmdir(struct inode *, struct dentry *);
2438 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2439 extern int noop_fsync(struct file *, int);
2440 extern int simple_empty(struct dentry *);
2441 extern int simple_readpage(struct file *file, struct page *page);
2442 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2443 			loff_t pos, unsigned len, unsigned flags,
2444 			struct page **pagep, void **fsdata);
2445 extern int simple_write_end(struct file *file, struct address_space *mapping,
2446 			loff_t pos, unsigned len, unsigned copied,
2447 			struct page *page, void *fsdata);
2448 
2449 extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
2450 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2451 extern const struct file_operations simple_dir_operations;
2452 extern const struct inode_operations simple_dir_inode_operations;
2453 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2454 struct dentry *d_alloc_name(struct dentry *, const char *);
2455 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2456 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2457 extern void simple_release_fs(struct vfsmount **mount, int *count);
2458 
2459 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2460 			loff_t *ppos, const void *from, size_t available);
2461 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2462 		const void __user *from, size_t count);
2463 
2464 extern int generic_file_fsync(struct file *, int);
2465 
2466 extern int generic_check_addressable(unsigned, u64);
2467 
2468 #ifdef CONFIG_MIGRATION
2469 extern int buffer_migrate_page(struct address_space *,
2470 				struct page *, struct page *);
2471 #else
2472 #define buffer_migrate_page NULL
2473 #endif
2474 
2475 extern int inode_change_ok(const struct inode *, struct iattr *);
2476 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2477 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2478 
2479 extern void file_update_time(struct file *file);
2480 
2481 extern int generic_show_options(struct seq_file *m, struct vfsmount *mnt);
2482 extern void save_mount_options(struct super_block *sb, char *options);
2483 extern void replace_mount_options(struct super_block *sb, char *options);
2484 
2485 static inline ino_t parent_ino(struct dentry *dentry)
2486 {
2487 	ino_t res;
2488 
2489 	/*
2490 	 * Don't strictly need d_lock here? If the parent ino could change
2491 	 * then surely we'd have a deeper race in the caller?
2492 	 */
2493 	spin_lock(&dentry->d_lock);
2494 	res = dentry->d_parent->d_inode->i_ino;
2495 	spin_unlock(&dentry->d_lock);
2496 	return res;
2497 }
2498 
2499 /* Transaction based IO helpers */
2500 
2501 /*
2502  * An argresp is stored in an allocated page and holds the
2503  * size of the argument or response, along with its content
2504  */
2505 struct simple_transaction_argresp {
2506 	ssize_t size;
2507 	char data[0];
2508 };
2509 
2510 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2511 
2512 char *simple_transaction_get(struct file *file, const char __user *buf,
2513 				size_t size);
2514 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2515 				size_t size, loff_t *pos);
2516 int simple_transaction_release(struct inode *inode, struct file *file);
2517 
2518 void simple_transaction_set(struct file *file, size_t n);
2519 
2520 /*
2521  * simple attribute files
2522  *
2523  * These attributes behave similar to those in sysfs:
2524  *
2525  * Writing to an attribute immediately sets a value, an open file can be
2526  * written to multiple times.
2527  *
2528  * Reading from an attribute creates a buffer from the value that might get
2529  * read with multiple read calls. When the attribute has been read
2530  * completely, no further read calls are possible until the file is opened
2531  * again.
2532  *
2533  * All attributes contain a text representation of a numeric value
2534  * that are accessed with the get() and set() functions.
2535  */
2536 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt)		\
2537 static int __fops ## _open(struct inode *inode, struct file *file)	\
2538 {									\
2539 	__simple_attr_check_format(__fmt, 0ull);			\
2540 	return simple_attr_open(inode, file, __get, __set, __fmt);	\
2541 }									\
2542 static const struct file_operations __fops = {				\
2543 	.owner	 = THIS_MODULE,						\
2544 	.open	 = __fops ## _open,					\
2545 	.release = simple_attr_release,					\
2546 	.read	 = simple_attr_read,					\
2547 	.write	 = simple_attr_write,					\
2548 	.llseek	 = generic_file_llseek,					\
2549 };
2550 
2551 static inline void __attribute__((format(printf, 1, 2)))
2552 __simple_attr_check_format(const char *fmt, ...)
2553 {
2554 	/* don't do anything, just let the compiler check the arguments; */
2555 }
2556 
2557 int simple_attr_open(struct inode *inode, struct file *file,
2558 		     int (*get)(void *, u64 *), int (*set)(void *, u64),
2559 		     const char *fmt);
2560 int simple_attr_release(struct inode *inode, struct file *file);
2561 ssize_t simple_attr_read(struct file *file, char __user *buf,
2562 			 size_t len, loff_t *ppos);
2563 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2564 			  size_t len, loff_t *ppos);
2565 
2566 struct ctl_table;
2567 int proc_nr_files(struct ctl_table *table, int write,
2568 		  void __user *buffer, size_t *lenp, loff_t *ppos);
2569 int proc_nr_dentry(struct ctl_table *table, int write,
2570 		  void __user *buffer, size_t *lenp, loff_t *ppos);
2571 int proc_nr_inodes(struct ctl_table *table, int write,
2572 		   void __user *buffer, size_t *lenp, loff_t *ppos);
2573 int __init get_filesystem_list(char *buf);
2574 
2575 #define __FMODE_EXEC		((__force int) FMODE_EXEC)
2576 #define __FMODE_NONOTIFY	((__force int) FMODE_NONOTIFY)
2577 
2578 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2579 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2580 					    (flag & __FMODE_NONOTIFY)))
2581 
2582 #endif /* __KERNEL__ */
2583 #endif /* _LINUX_FS_H */
2584