1 #ifndef _LINUX_FS_H
2 #define _LINUX_FS_H
3 
4 /*
5  * This file has definitions for some important file table
6  * structures etc.
7  */
8 
9 #include <linux/limits.h>
10 #include <linux/ioctl.h>
11 #include <linux/blk_types.h>
12 #include <linux/types.h>
13 
14 /*
15  * It's silly to have NR_OPEN bigger than NR_FILE, but you can change
16  * the file limit at runtime and only root can increase the per-process
17  * nr_file rlimit, so it's safe to set up a ridiculously high absolute
18  * upper limit on files-per-process.
19  *
20  * Some programs (notably those using select()) may have to be
21  * recompiled to take full advantage of the new limits..
22  */
23 
24 /* Fixed constants first: */
25 #undef NR_OPEN
26 #define INR_OPEN_CUR 1024	/* Initial setting for nfile rlimits */
27 #define INR_OPEN_MAX 4096	/* Hard limit for nfile rlimits */
28 
29 #define BLOCK_SIZE_BITS 10
30 #define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
31 
32 #define SEEK_SET	0	/* seek relative to beginning of file */
33 #define SEEK_CUR	1	/* seek relative to current file position */
34 #define SEEK_END	2	/* seek relative to end of file */
35 #define SEEK_DATA	3	/* seek to the next data */
36 #define SEEK_HOLE	4	/* seek to the next hole */
37 #define SEEK_MAX	SEEK_HOLE
38 
39 struct fstrim_range {
40 	__u64 start;
41 	__u64 len;
42 	__u64 minlen;
43 };
44 
45 /* And dynamically-tunable limits and defaults: */
46 struct files_stat_struct {
47 	unsigned long nr_files;		/* read only */
48 	unsigned long nr_free_files;	/* read only */
49 	unsigned long max_files;		/* tunable */
50 };
51 
52 struct inodes_stat_t {
53 	int nr_inodes;
54 	int nr_unused;
55 	int dummy[5];		/* padding for sysctl ABI compatibility */
56 };
57 
58 
59 #define NR_FILE  8192	/* this can well be larger on a larger system */
60 
61 #define MAY_EXEC		0x00000001
62 #define MAY_WRITE		0x00000002
63 #define MAY_READ		0x00000004
64 #define MAY_APPEND		0x00000008
65 #define MAY_ACCESS		0x00000010
66 #define MAY_OPEN		0x00000020
67 #define MAY_CHDIR		0x00000040
68 /* called from RCU mode, don't block */
69 #define MAY_NOT_BLOCK		0x00000080
70 
71 /*
72  * flags in file.f_mode.  Note that FMODE_READ and FMODE_WRITE must correspond
73  * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
74  */
75 
76 /* file is open for reading */
77 #define FMODE_READ		((__force fmode_t)0x1)
78 /* file is open for writing */
79 #define FMODE_WRITE		((__force fmode_t)0x2)
80 /* file is seekable */
81 #define FMODE_LSEEK		((__force fmode_t)0x4)
82 /* file can be accessed using pread */
83 #define FMODE_PREAD		((__force fmode_t)0x8)
84 /* file can be accessed using pwrite */
85 #define FMODE_PWRITE		((__force fmode_t)0x10)
86 /* File is opened for execution with sys_execve / sys_uselib */
87 #define FMODE_EXEC		((__force fmode_t)0x20)
88 /* File is opened with O_NDELAY (only set for block devices) */
89 #define FMODE_NDELAY		((__force fmode_t)0x40)
90 /* File is opened with O_EXCL (only set for block devices) */
91 #define FMODE_EXCL		((__force fmode_t)0x80)
92 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
93    (specialy hack for floppy.c) */
94 #define FMODE_WRITE_IOCTL	((__force fmode_t)0x100)
95 /* 32bit hashes as llseek() offset (for directories) */
96 #define FMODE_32BITHASH         ((__force fmode_t)0x200)
97 /* 64bit hashes as llseek() offset (for directories) */
98 #define FMODE_64BITHASH         ((__force fmode_t)0x400)
99 
100 /*
101  * Don't update ctime and mtime.
102  *
103  * Currently a special hack for the XFS open_by_handle ioctl, but we'll
104  * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
105  */
106 #define FMODE_NOCMTIME		((__force fmode_t)0x800)
107 
108 /* Expect random access pattern */
109 #define FMODE_RANDOM		((__force fmode_t)0x1000)
110 
111 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
112 #define FMODE_UNSIGNED_OFFSET	((__force fmode_t)0x2000)
113 
114 /* File is opened with O_PATH; almost nothing can be done with it */
115 #define FMODE_PATH		((__force fmode_t)0x4000)
116 
117 /* File was opened by fanotify and shouldn't generate fanotify events */
118 #define FMODE_NONOTIFY		((__force fmode_t)0x1000000)
119 
120 /*
121  * The below are the various read and write types that we support. Some of
122  * them include behavioral modifiers that send information down to the
123  * block layer and IO scheduler. Terminology:
124  *
125  *	The block layer uses device plugging to defer IO a little bit, in
126  *	the hope that we will see more IO very shortly. This increases
127  *	coalescing of adjacent IO and thus reduces the number of IOs we
128  *	have to send to the device. It also allows for better queuing,
129  *	if the IO isn't mergeable. If the caller is going to be waiting
130  *	for the IO, then he must ensure that the device is unplugged so
131  *	that the IO is dispatched to the driver.
132  *
133  *	All IO is handled async in Linux. This is fine for background
134  *	writes, but for reads or writes that someone waits for completion
135  *	on, we want to notify the block layer and IO scheduler so that they
136  *	know about it. That allows them to make better scheduling
137  *	decisions. So when the below references 'sync' and 'async', it
138  *	is referencing this priority hint.
139  *
140  * With that in mind, the available types are:
141  *
142  * READ			A normal read operation. Device will be plugged.
143  * READ_SYNC		A synchronous read. Device is not plugged, caller can
144  *			immediately wait on this read without caring about
145  *			unplugging.
146  * READA		Used for read-ahead operations. Lower priority, and the
147  *			block layer could (in theory) choose to ignore this
148  *			request if it runs into resource problems.
149  * WRITE		A normal async write. Device will be plugged.
150  * WRITE_SYNC		Synchronous write. Identical to WRITE, but passes down
151  *			the hint that someone will be waiting on this IO
152  *			shortly. The write equivalent of READ_SYNC.
153  * WRITE_ODIRECT	Special case write for O_DIRECT only.
154  * WRITE_FLUSH		Like WRITE_SYNC but with preceding cache flush.
155  * WRITE_FUA		Like WRITE_SYNC but data is guaranteed to be on
156  *			non-volatile media on completion.
157  * WRITE_FLUSH_FUA	Combination of WRITE_FLUSH and FUA. The IO is preceded
158  *			by a cache flush and data is guaranteed to be on
159  *			non-volatile media on completion.
160  *
161  */
162 #define RW_MASK			REQ_WRITE
163 #define RWA_MASK		REQ_RAHEAD
164 
165 #define READ			0
166 #define WRITE			RW_MASK
167 #define READA			RWA_MASK
168 
169 #define READ_SYNC		(READ | REQ_SYNC)
170 #define WRITE_SYNC		(WRITE | REQ_SYNC | REQ_NOIDLE)
171 #define WRITE_ODIRECT		(WRITE | REQ_SYNC)
172 #define WRITE_FLUSH		(WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
173 #define WRITE_FUA		(WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
174 #define WRITE_FLUSH_FUA		(WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
175 
176 #define SEL_IN		1
177 #define SEL_OUT		2
178 #define SEL_EX		4
179 
180 /* public flags for file_system_type */
181 #define FS_REQUIRES_DEV 1
182 #define FS_BINARY_MOUNTDATA 2
183 #define FS_HAS_SUBTYPE 4
184 #define FS_REVAL_DOT	16384	/* Check the paths ".", ".." for staleness */
185 #define FS_RENAME_DOES_D_MOVE	32768	/* FS will handle d_move()
186 					 * during rename() internally.
187 					 */
188 
189 /*
190  * These are the fs-independent mount-flags: up to 32 flags are supported
191  */
192 #define MS_RDONLY	 1	/* Mount read-only */
193 #define MS_NOSUID	 2	/* Ignore suid and sgid bits */
194 #define MS_NODEV	 4	/* Disallow access to device special files */
195 #define MS_NOEXEC	 8	/* Disallow program execution */
196 #define MS_SYNCHRONOUS	16	/* Writes are synced at once */
197 #define MS_REMOUNT	32	/* Alter flags of a mounted FS */
198 #define MS_MANDLOCK	64	/* Allow mandatory locks on an FS */
199 #define MS_DIRSYNC	128	/* Directory modifications are synchronous */
200 #define MS_NOATIME	1024	/* Do not update access times. */
201 #define MS_NODIRATIME	2048	/* Do not update directory access times */
202 #define MS_BIND		4096
203 #define MS_MOVE		8192
204 #define MS_REC		16384
205 #define MS_VERBOSE	32768	/* War is peace. Verbosity is silence.
206 				   MS_VERBOSE is deprecated. */
207 #define MS_SILENT	32768
208 #define MS_POSIXACL	(1<<16)	/* VFS does not apply the umask */
209 #define MS_UNBINDABLE	(1<<17)	/* change to unbindable */
210 #define MS_PRIVATE	(1<<18)	/* change to private */
211 #define MS_SLAVE	(1<<19)	/* change to slave */
212 #define MS_SHARED	(1<<20)	/* change to shared */
213 #define MS_RELATIME	(1<<21)	/* Update atime relative to mtime/ctime. */
214 #define MS_KERNMOUNT	(1<<22) /* this is a kern_mount call */
215 #define MS_I_VERSION	(1<<23) /* Update inode I_version field */
216 #define MS_STRICTATIME	(1<<24) /* Always perform atime updates */
217 #define MS_NOSEC	(1<<28)
218 #define MS_BORN		(1<<29)
219 #define MS_ACTIVE	(1<<30)
220 #define MS_NOUSER	(1<<31)
221 
222 /*
223  * Superblock flags that can be altered by MS_REMOUNT
224  */
225 #define MS_RMT_MASK	(MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_I_VERSION)
226 
227 /*
228  * Old magic mount flag and mask
229  */
230 #define MS_MGC_VAL 0xC0ED0000
231 #define MS_MGC_MSK 0xffff0000
232 
233 /* Inode flags - they have nothing to superblock flags now */
234 
235 #define S_SYNC		1	/* Writes are synced at once */
236 #define S_NOATIME	2	/* Do not update access times */
237 #define S_APPEND	4	/* Append-only file */
238 #define S_IMMUTABLE	8	/* Immutable file */
239 #define S_DEAD		16	/* removed, but still open directory */
240 #define S_NOQUOTA	32	/* Inode is not counted to quota */
241 #define S_DIRSYNC	64	/* Directory modifications are synchronous */
242 #define S_NOCMTIME	128	/* Do not update file c/mtime */
243 #define S_SWAPFILE	256	/* Do not truncate: swapon got its bmaps */
244 #define S_PRIVATE	512	/* Inode is fs-internal */
245 #define S_IMA		1024	/* Inode has an associated IMA struct */
246 #define S_AUTOMOUNT	2048	/* Automount/referral quasi-directory */
247 #define S_NOSEC		4096	/* no suid or xattr security attributes */
248 
249 /*
250  * Note that nosuid etc flags are inode-specific: setting some file-system
251  * flags just means all the inodes inherit those flags by default. It might be
252  * possible to override it selectively if you really wanted to with some
253  * ioctl() that is not currently implemented.
254  *
255  * Exception: MS_RDONLY is always applied to the entire file system.
256  *
257  * Unfortunately, it is possible to change a filesystems flags with it mounted
258  * with files in use.  This means that all of the inodes will not have their
259  * i_flags updated.  Hence, i_flags no longer inherit the superblock mount
260  * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
261  */
262 #define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
263 
264 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
265 #define IS_SYNC(inode)		(__IS_FLG(inode, MS_SYNCHRONOUS) || \
266 					((inode)->i_flags & S_SYNC))
267 #define IS_DIRSYNC(inode)	(__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
268 					((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
269 #define IS_MANDLOCK(inode)	__IS_FLG(inode, MS_MANDLOCK)
270 #define IS_NOATIME(inode)   __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
271 #define IS_I_VERSION(inode)   __IS_FLG(inode, MS_I_VERSION)
272 
273 #define IS_NOQUOTA(inode)	((inode)->i_flags & S_NOQUOTA)
274 #define IS_APPEND(inode)	((inode)->i_flags & S_APPEND)
275 #define IS_IMMUTABLE(inode)	((inode)->i_flags & S_IMMUTABLE)
276 #define IS_POSIXACL(inode)	__IS_FLG(inode, MS_POSIXACL)
277 
278 #define IS_DEADDIR(inode)	((inode)->i_flags & S_DEAD)
279 #define IS_NOCMTIME(inode)	((inode)->i_flags & S_NOCMTIME)
280 #define IS_SWAPFILE(inode)	((inode)->i_flags & S_SWAPFILE)
281 #define IS_PRIVATE(inode)	((inode)->i_flags & S_PRIVATE)
282 #define IS_IMA(inode)		((inode)->i_flags & S_IMA)
283 #define IS_AUTOMOUNT(inode)	((inode)->i_flags & S_AUTOMOUNT)
284 #define IS_NOSEC(inode)		((inode)->i_flags & S_NOSEC)
285 
286 /* the read-only stuff doesn't really belong here, but any other place is
287    probably as bad and I don't want to create yet another include file. */
288 
289 #define BLKROSET   _IO(0x12,93)	/* set device read-only (0 = read-write) */
290 #define BLKROGET   _IO(0x12,94)	/* get read-only status (0 = read_write) */
291 #define BLKRRPART  _IO(0x12,95)	/* re-read partition table */
292 #define BLKGETSIZE _IO(0x12,96)	/* return device size /512 (long *arg) */
293 #define BLKFLSBUF  _IO(0x12,97)	/* flush buffer cache */
294 #define BLKRASET   _IO(0x12,98)	/* set read ahead for block device */
295 #define BLKRAGET   _IO(0x12,99)	/* get current read ahead setting */
296 #define BLKFRASET  _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
297 #define BLKFRAGET  _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
298 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
299 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
300 #define BLKSSZGET  _IO(0x12,104)/* get block device sector size */
301 #if 0
302 #define BLKPG      _IO(0x12,105)/* See blkpg.h */
303 
304 /* Some people are morons.  Do not use sizeof! */
305 
306 #define BLKELVGET  _IOR(0x12,106,size_t)/* elevator get */
307 #define BLKELVSET  _IOW(0x12,107,size_t)/* elevator set */
308 /* This was here just to show that the number is taken -
309    probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
310 #endif
311 /* A jump here: 108-111 have been used for various private purposes. */
312 #define BLKBSZGET  _IOR(0x12,112,size_t)
313 #define BLKBSZSET  _IOW(0x12,113,size_t)
314 #define BLKGETSIZE64 _IOR(0x12,114,size_t)	/* return device size in bytes (u64 *arg) */
315 #define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
316 #define BLKTRACESTART _IO(0x12,116)
317 #define BLKTRACESTOP _IO(0x12,117)
318 #define BLKTRACETEARDOWN _IO(0x12,118)
319 #define BLKDISCARD _IO(0x12,119)
320 #define BLKIOMIN _IO(0x12,120)
321 #define BLKIOOPT _IO(0x12,121)
322 #define BLKALIGNOFF _IO(0x12,122)
323 #define BLKPBSZGET _IO(0x12,123)
324 #define BLKDISCARDZEROES _IO(0x12,124)
325 #define BLKSECDISCARD _IO(0x12,125)
326 #define BLKROTATIONAL _IO(0x12,126)
327 
328 #define BMAP_IOCTL 1		/* obsolete - kept for compatibility */
329 #define FIBMAP	   _IO(0x00,1)	/* bmap access */
330 #define FIGETBSZ   _IO(0x00,2)	/* get the block size used for bmap */
331 #define FIFREEZE	_IOWR('X', 119, int)	/* Freeze */
332 #define FITHAW		_IOWR('X', 120, int)	/* Thaw */
333 #define FITRIM		_IOWR('X', 121, struct fstrim_range)	/* Trim */
334 
335 #define	FS_IOC_GETFLAGS			_IOR('f', 1, long)
336 #define	FS_IOC_SETFLAGS			_IOW('f', 2, long)
337 #define	FS_IOC_GETVERSION		_IOR('v', 1, long)
338 #define	FS_IOC_SETVERSION		_IOW('v', 2, long)
339 #define FS_IOC_FIEMAP			_IOWR('f', 11, struct fiemap)
340 #define FS_IOC32_GETFLAGS		_IOR('f', 1, int)
341 #define FS_IOC32_SETFLAGS		_IOW('f', 2, int)
342 #define FS_IOC32_GETVERSION		_IOR('v', 1, int)
343 #define FS_IOC32_SETVERSION		_IOW('v', 2, int)
344 
345 /*
346  * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS)
347  */
348 #define	FS_SECRM_FL			0x00000001 /* Secure deletion */
349 #define	FS_UNRM_FL			0x00000002 /* Undelete */
350 #define	FS_COMPR_FL			0x00000004 /* Compress file */
351 #define FS_SYNC_FL			0x00000008 /* Synchronous updates */
352 #define FS_IMMUTABLE_FL			0x00000010 /* Immutable file */
353 #define FS_APPEND_FL			0x00000020 /* writes to file may only append */
354 #define FS_NODUMP_FL			0x00000040 /* do not dump file */
355 #define FS_NOATIME_FL			0x00000080 /* do not update atime */
356 /* Reserved for compression usage... */
357 #define FS_DIRTY_FL			0x00000100
358 #define FS_COMPRBLK_FL			0x00000200 /* One or more compressed clusters */
359 #define FS_NOCOMP_FL			0x00000400 /* Don't compress */
360 #define FS_ECOMPR_FL			0x00000800 /* Compression error */
361 /* End compression flags --- maybe not all used */
362 #define FS_BTREE_FL			0x00001000 /* btree format dir */
363 #define FS_INDEX_FL			0x00001000 /* hash-indexed directory */
364 #define FS_IMAGIC_FL			0x00002000 /* AFS directory */
365 #define FS_JOURNAL_DATA_FL		0x00004000 /* Reserved for ext3 */
366 #define FS_NOTAIL_FL			0x00008000 /* file tail should not be merged */
367 #define FS_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
368 #define FS_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
369 #define FS_EXTENT_FL			0x00080000 /* Extents */
370 #define FS_DIRECTIO_FL			0x00100000 /* Use direct i/o */
371 #define FS_NOCOW_FL			0x00800000 /* Do not cow file */
372 #define FS_RESERVED_FL			0x80000000 /* reserved for ext2 lib */
373 
374 #define FS_FL_USER_VISIBLE		0x0003DFFF /* User visible flags */
375 #define FS_FL_USER_MODIFIABLE		0x000380FF /* User modifiable flags */
376 
377 
378 #define SYNC_FILE_RANGE_WAIT_BEFORE	1
379 #define SYNC_FILE_RANGE_WRITE		2
380 #define SYNC_FILE_RANGE_WAIT_AFTER	4
381 
382 #ifdef __KERNEL__
383 
384 #include <linux/linkage.h>
385 #include <linux/wait.h>
386 #include <linux/kdev_t.h>
387 #include <linux/dcache.h>
388 #include <linux/path.h>
389 #include <linux/stat.h>
390 #include <linux/cache.h>
391 #include <linux/list.h>
392 #include <linux/radix-tree.h>
393 #include <linux/prio_tree.h>
394 #include <linux/init.h>
395 #include <linux/pid.h>
396 #include <linux/bug.h>
397 #include <linux/mutex.h>
398 #include <linux/capability.h>
399 #include <linux/semaphore.h>
400 #include <linux/fiemap.h>
401 #include <linux/rculist_bl.h>
402 #include <linux/atomic.h>
403 #include <linux/shrinker.h>
404 #include <linux/migrate_mode.h>
405 
406 #include <asm/byteorder.h>
407 
408 struct export_operations;
409 struct hd_geometry;
410 struct iovec;
411 struct nameidata;
412 struct kiocb;
413 struct kobject;
414 struct pipe_inode_info;
415 struct poll_table_struct;
416 struct kstatfs;
417 struct vm_area_struct;
418 struct vfsmount;
419 struct cred;
420 
421 extern void __init inode_init(void);
422 extern void __init inode_init_early(void);
423 extern void __init files_init(unsigned long);
424 
425 extern struct files_stat_struct files_stat;
426 extern unsigned long get_max_files(void);
427 extern int sysctl_nr_open;
428 extern struct inodes_stat_t inodes_stat;
429 extern int leases_enable, lease_break_time;
430 
431 struct buffer_head;
432 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
433 			struct buffer_head *bh_result, int create);
434 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
435 			ssize_t bytes, void *private, int ret,
436 			bool is_async);
437 
438 /*
439  * Attribute flags.  These should be or-ed together to figure out what
440  * has been changed!
441  */
442 #define ATTR_MODE	(1 << 0)
443 #define ATTR_UID	(1 << 1)
444 #define ATTR_GID	(1 << 2)
445 #define ATTR_SIZE	(1 << 3)
446 #define ATTR_ATIME	(1 << 4)
447 #define ATTR_MTIME	(1 << 5)
448 #define ATTR_CTIME	(1 << 6)
449 #define ATTR_ATIME_SET	(1 << 7)
450 #define ATTR_MTIME_SET	(1 << 8)
451 #define ATTR_FORCE	(1 << 9) /* Not a change, but a change it */
452 #define ATTR_ATTR_FLAG	(1 << 10)
453 #define ATTR_KILL_SUID	(1 << 11)
454 #define ATTR_KILL_SGID	(1 << 12)
455 #define ATTR_FILE	(1 << 13)
456 #define ATTR_KILL_PRIV	(1 << 14)
457 #define ATTR_OPEN	(1 << 15) /* Truncating from open(O_TRUNC) */
458 #define ATTR_TIMES_SET	(1 << 16)
459 
460 /*
461  * This is the Inode Attributes structure, used for notify_change().  It
462  * uses the above definitions as flags, to know which values have changed.
463  * Also, in this manner, a Filesystem can look at only the values it cares
464  * about.  Basically, these are the attributes that the VFS layer can
465  * request to change from the FS layer.
466  *
467  * Derek Atkins <warlord@MIT.EDU> 94-10-20
468  */
469 struct iattr {
470 	unsigned int	ia_valid;
471 	umode_t		ia_mode;
472 	uid_t		ia_uid;
473 	gid_t		ia_gid;
474 	loff_t		ia_size;
475 	struct timespec	ia_atime;
476 	struct timespec	ia_mtime;
477 	struct timespec	ia_ctime;
478 
479 	/*
480 	 * Not an attribute, but an auxiliary info for filesystems wanting to
481 	 * implement an ftruncate() like method.  NOTE: filesystem should
482 	 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
483 	 */
484 	struct file	*ia_file;
485 };
486 
487 /*
488  * Includes for diskquotas.
489  */
490 #include <linux/quota.h>
491 
492 /**
493  * enum positive_aop_returns - aop return codes with specific semantics
494  *
495  * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
496  * 			    completed, that the page is still locked, and
497  * 			    should be considered active.  The VM uses this hint
498  * 			    to return the page to the active list -- it won't
499  * 			    be a candidate for writeback again in the near
500  * 			    future.  Other callers must be careful to unlock
501  * 			    the page if they get this return.  Returned by
502  * 			    writepage();
503  *
504  * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
505  *  			unlocked it and the page might have been truncated.
506  *  			The caller should back up to acquiring a new page and
507  *  			trying again.  The aop will be taking reasonable
508  *  			precautions not to livelock.  If the caller held a page
509  *  			reference, it should drop it before retrying.  Returned
510  *  			by readpage().
511  *
512  * address_space_operation functions return these large constants to indicate
513  * special semantics to the caller.  These are much larger than the bytes in a
514  * page to allow for functions that return the number of bytes operated on in a
515  * given page.
516  */
517 
518 enum positive_aop_returns {
519 	AOP_WRITEPAGE_ACTIVATE	= 0x80000,
520 	AOP_TRUNCATED_PAGE	= 0x80001,
521 };
522 
523 #define AOP_FLAG_UNINTERRUPTIBLE	0x0001 /* will not do a short write */
524 #define AOP_FLAG_CONT_EXPAND		0x0002 /* called from cont_expand */
525 #define AOP_FLAG_NOFS			0x0004 /* used by filesystem to direct
526 						* helper code (eg buffer layer)
527 						* to clear GFP_FS from alloc */
528 
529 /*
530  * oh the beauties of C type declarations.
531  */
532 struct page;
533 struct address_space;
534 struct writeback_control;
535 
536 struct iov_iter {
537 	const struct iovec *iov;
538 	unsigned long nr_segs;
539 	size_t iov_offset;
540 	size_t count;
541 };
542 
543 size_t iov_iter_copy_from_user_atomic(struct page *page,
544 		struct iov_iter *i, unsigned long offset, size_t bytes);
545 size_t iov_iter_copy_from_user(struct page *page,
546 		struct iov_iter *i, unsigned long offset, size_t bytes);
547 void iov_iter_advance(struct iov_iter *i, size_t bytes);
548 int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
549 size_t iov_iter_single_seg_count(struct iov_iter *i);
550 
iov_iter_init(struct iov_iter * i,const struct iovec * iov,unsigned long nr_segs,size_t count,size_t written)551 static inline void iov_iter_init(struct iov_iter *i,
552 			const struct iovec *iov, unsigned long nr_segs,
553 			size_t count, size_t written)
554 {
555 	i->iov = iov;
556 	i->nr_segs = nr_segs;
557 	i->iov_offset = 0;
558 	i->count = count + written;
559 
560 	iov_iter_advance(i, written);
561 }
562 
iov_iter_count(struct iov_iter * i)563 static inline size_t iov_iter_count(struct iov_iter *i)
564 {
565 	return i->count;
566 }
567 
568 /*
569  * "descriptor" for what we're up to with a read.
570  * This allows us to use the same read code yet
571  * have multiple different users of the data that
572  * we read from a file.
573  *
574  * The simplest case just copies the data to user
575  * mode.
576  */
577 typedef struct {
578 	size_t written;
579 	size_t count;
580 	union {
581 		char __user *buf;
582 		void *data;
583 	} arg;
584 	int error;
585 } read_descriptor_t;
586 
587 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
588 		unsigned long, unsigned long);
589 
590 struct address_space_operations {
591 	int (*writepage)(struct page *page, struct writeback_control *wbc);
592 	int (*readpage)(struct file *, struct page *);
593 
594 	/* Write back some dirty pages from this mapping. */
595 	int (*writepages)(struct address_space *, struct writeback_control *);
596 
597 	/* Set a page dirty.  Return true if this dirtied it */
598 	int (*set_page_dirty)(struct page *page);
599 
600 	int (*readpages)(struct file *filp, struct address_space *mapping,
601 			struct list_head *pages, unsigned nr_pages);
602 
603 	int (*write_begin)(struct file *, struct address_space *mapping,
604 				loff_t pos, unsigned len, unsigned flags,
605 				struct page **pagep, void **fsdata);
606 	int (*write_end)(struct file *, struct address_space *mapping,
607 				loff_t pos, unsigned len, unsigned copied,
608 				struct page *page, void *fsdata);
609 
610 	/* Unfortunately this kludge is needed for FIBMAP. Don't use it */
611 	sector_t (*bmap)(struct address_space *, sector_t);
612 	void (*invalidatepage) (struct page *, unsigned long);
613 	int (*releasepage) (struct page *, gfp_t);
614 	void (*freepage)(struct page *);
615 	ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
616 			loff_t offset, unsigned long nr_segs);
617 	int (*get_xip_mem)(struct address_space *, pgoff_t, int,
618 						void **, unsigned long *);
619 	/*
620 	 * migrate the contents of a page to the specified target. If sync
621 	 * is false, it must not block.
622 	 */
623 	int (*migratepage) (struct address_space *,
624 			struct page *, struct page *, enum migrate_mode);
625 	int (*launder_page) (struct page *);
626 	int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
627 					unsigned long);
628 	int (*error_remove_page)(struct address_space *, struct page *);
629 };
630 
631 extern const struct address_space_operations empty_aops;
632 
633 /*
634  * pagecache_write_begin/pagecache_write_end must be used by general code
635  * to write into the pagecache.
636  */
637 int pagecache_write_begin(struct file *, struct address_space *mapping,
638 				loff_t pos, unsigned len, unsigned flags,
639 				struct page **pagep, void **fsdata);
640 
641 int pagecache_write_end(struct file *, struct address_space *mapping,
642 				loff_t pos, unsigned len, unsigned copied,
643 				struct page *page, void *fsdata);
644 
645 struct backing_dev_info;
646 struct address_space {
647 	struct inode		*host;		/* owner: inode, block_device */
648 	struct radix_tree_root	page_tree;	/* radix tree of all pages */
649 	spinlock_t		tree_lock;	/* and lock protecting it */
650 	unsigned int		i_mmap_writable;/* count VM_SHARED mappings */
651 	struct prio_tree_root	i_mmap;		/* tree of private and shared mappings */
652 	struct list_head	i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
653 	struct mutex		i_mmap_mutex;	/* protect tree, count, list */
654 	/* Protected by tree_lock together with the radix tree */
655 	unsigned long		nrpages;	/* number of total pages */
656 	pgoff_t			writeback_index;/* writeback starts here */
657 	const struct address_space_operations *a_ops;	/* methods */
658 	unsigned long		flags;		/* error bits/gfp mask */
659 	struct backing_dev_info *backing_dev_info; /* device readahead, etc */
660 	spinlock_t		private_lock;	/* for use by the address_space */
661 	struct list_head	private_list;	/* ditto */
662 	struct address_space	*assoc_mapping;	/* ditto */
663 } __attribute__((aligned(sizeof(long))));
664 	/*
665 	 * On most architectures that alignment is already the case; but
666 	 * must be enforced here for CRIS, to let the least significant bit
667 	 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
668 	 */
669 struct request_queue;
670 
671 struct block_device {
672 	dev_t			bd_dev;  /* not a kdev_t - it's a search key */
673 	int			bd_openers;
674 	struct inode *		bd_inode;	/* will die */
675 	struct super_block *	bd_super;
676 	struct mutex		bd_mutex;	/* open/close mutex */
677 	struct list_head	bd_inodes;
678 	void *			bd_claiming;
679 	void *			bd_holder;
680 	int			bd_holders;
681 	bool			bd_write_holder;
682 #ifdef CONFIG_SYSFS
683 	struct list_head	bd_holder_disks;
684 #endif
685 	struct block_device *	bd_contains;
686 	unsigned		bd_block_size;
687 	struct hd_struct *	bd_part;
688 	/* number of times partitions within this device have been opened. */
689 	unsigned		bd_part_count;
690 	int			bd_invalidated;
691 	struct gendisk *	bd_disk;
692 	struct request_queue *  bd_queue;
693 	struct list_head	bd_list;
694 	/*
695 	 * Private data.  You must have bd_claim'ed the block_device
696 	 * to use this.  NOTE:  bd_claim allows an owner to claim
697 	 * the same device multiple times, the owner must take special
698 	 * care to not mess up bd_private for that case.
699 	 */
700 	unsigned long		bd_private;
701 
702 	/* The counter of freeze processes */
703 	int			bd_fsfreeze_count;
704 	/* Mutex for freeze */
705 	struct mutex		bd_fsfreeze_mutex;
706 };
707 
708 /*
709  * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
710  * radix trees
711  */
712 #define PAGECACHE_TAG_DIRTY	0
713 #define PAGECACHE_TAG_WRITEBACK	1
714 #define PAGECACHE_TAG_TOWRITE	2
715 
716 int mapping_tagged(struct address_space *mapping, int tag);
717 
718 /*
719  * Might pages of this file be mapped into userspace?
720  */
mapping_mapped(struct address_space * mapping)721 static inline int mapping_mapped(struct address_space *mapping)
722 {
723 	return	!prio_tree_empty(&mapping->i_mmap) ||
724 		!list_empty(&mapping->i_mmap_nonlinear);
725 }
726 
727 /*
728  * Might pages of this file have been modified in userspace?
729  * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
730  * marks vma as VM_SHARED if it is shared, and the file was opened for
731  * writing i.e. vma may be mprotected writable even if now readonly.
732  */
mapping_writably_mapped(struct address_space * mapping)733 static inline int mapping_writably_mapped(struct address_space *mapping)
734 {
735 	return mapping->i_mmap_writable != 0;
736 }
737 
738 /*
739  * Use sequence counter to get consistent i_size on 32-bit processors.
740  */
741 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
742 #include <linux/seqlock.h>
743 #define __NEED_I_SIZE_ORDERED
744 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
745 #else
746 #define i_size_ordered_init(inode) do { } while (0)
747 #endif
748 
749 struct posix_acl;
750 #define ACL_NOT_CACHED ((void *)(-1))
751 
752 #define IOP_FASTPERM	0x0001
753 #define IOP_LOOKUP	0x0002
754 #define IOP_NOFOLLOW	0x0004
755 
756 /*
757  * Keep mostly read-only and often accessed (especially for
758  * the RCU path lookup and 'stat' data) fields at the beginning
759  * of the 'struct inode'
760  */
761 struct inode {
762 	umode_t			i_mode;
763 	unsigned short		i_opflags;
764 	uid_t			i_uid;
765 	gid_t			i_gid;
766 	unsigned int		i_flags;
767 
768 #ifdef CONFIG_FS_POSIX_ACL
769 	struct posix_acl	*i_acl;
770 	struct posix_acl	*i_default_acl;
771 #endif
772 
773 	const struct inode_operations	*i_op;
774 	struct super_block	*i_sb;
775 	struct address_space	*i_mapping;
776 
777 #ifdef CONFIG_SECURITY
778 	void			*i_security;
779 #endif
780 
781 	/* Stat data, not accessed from path walking */
782 	unsigned long		i_ino;
783 	/*
784 	 * Filesystems may only read i_nlink directly.  They shall use the
785 	 * following functions for modification:
786 	 *
787 	 *    (set|clear|inc|drop)_nlink
788 	 *    inode_(inc|dec)_link_count
789 	 */
790 	union {
791 		const unsigned int i_nlink;
792 		unsigned int __i_nlink;
793 	};
794 	dev_t			i_rdev;
795 	struct timespec		i_atime;
796 	struct timespec		i_mtime;
797 	struct timespec		i_ctime;
798 	spinlock_t		i_lock;	/* i_blocks, i_bytes, maybe i_size */
799 	unsigned short          i_bytes;
800 	blkcnt_t		i_blocks;
801 	loff_t			i_size;
802 
803 #ifdef __NEED_I_SIZE_ORDERED
804 	seqcount_t		i_size_seqcount;
805 #endif
806 
807 	/* Misc */
808 	unsigned long		i_state;
809 	struct mutex		i_mutex;
810 
811 	unsigned long		dirtied_when;	/* jiffies of first dirtying */
812 
813 	struct hlist_node	i_hash;
814 	struct list_head	i_wb_list;	/* backing dev IO list */
815 	struct list_head	i_lru;		/* inode LRU list */
816 	struct list_head	i_sb_list;
817 	union {
818 		struct list_head	i_dentry;
819 		struct rcu_head		i_rcu;
820 	};
821 	atomic_t		i_count;
822 	unsigned int		i_blkbits;
823 	u64			i_version;
824 	atomic_t		i_dio_count;
825 	atomic_t		i_writecount;
826 	const struct file_operations	*i_fop;	/* former ->i_op->default_file_ops */
827 	struct file_lock	*i_flock;
828 	struct address_space	i_data;
829 #ifdef CONFIG_QUOTA
830 	struct dquot		*i_dquot[MAXQUOTAS];
831 #endif
832 	struct list_head	i_devices;
833 	union {
834 		struct pipe_inode_info	*i_pipe;
835 		struct block_device	*i_bdev;
836 		struct cdev		*i_cdev;
837 	};
838 
839 	__u32			i_generation;
840 
841 #ifdef CONFIG_FSNOTIFY
842 	__u32			i_fsnotify_mask; /* all events this inode cares about */
843 	struct hlist_head	i_fsnotify_marks;
844 #endif
845 
846 #ifdef CONFIG_IMA
847 	atomic_t		i_readcount; /* struct files open RO */
848 #endif
849 	void			*i_private; /* fs or device private pointer */
850 };
851 
inode_unhashed(struct inode * inode)852 static inline int inode_unhashed(struct inode *inode)
853 {
854 	return hlist_unhashed(&inode->i_hash);
855 }
856 
857 /*
858  * inode->i_mutex nesting subclasses for the lock validator:
859  *
860  * 0: the object of the current VFS operation
861  * 1: parent
862  * 2: child/target
863  * 3: quota file
864  *
865  * The locking order between these classes is
866  * parent -> child -> normal -> xattr -> quota
867  */
868 enum inode_i_mutex_lock_class
869 {
870 	I_MUTEX_NORMAL,
871 	I_MUTEX_PARENT,
872 	I_MUTEX_CHILD,
873 	I_MUTEX_XATTR,
874 	I_MUTEX_QUOTA
875 };
876 
877 /*
878  * NOTE: in a 32bit arch with a preemptable kernel and
879  * an UP compile the i_size_read/write must be atomic
880  * with respect to the local cpu (unlike with preempt disabled),
881  * but they don't need to be atomic with respect to other cpus like in
882  * true SMP (so they need either to either locally disable irq around
883  * the read or for example on x86 they can be still implemented as a
884  * cmpxchg8b without the need of the lock prefix). For SMP compiles
885  * and 64bit archs it makes no difference if preempt is enabled or not.
886  */
i_size_read(const struct inode * inode)887 static inline loff_t i_size_read(const struct inode *inode)
888 {
889 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
890 	loff_t i_size;
891 	unsigned int seq;
892 
893 	do {
894 		seq = read_seqcount_begin(&inode->i_size_seqcount);
895 		i_size = inode->i_size;
896 	} while (read_seqcount_retry(&inode->i_size_seqcount, seq));
897 	return i_size;
898 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
899 	loff_t i_size;
900 
901 	preempt_disable();
902 	i_size = inode->i_size;
903 	preempt_enable();
904 	return i_size;
905 #else
906 	return inode->i_size;
907 #endif
908 }
909 
910 /*
911  * NOTE: unlike i_size_read(), i_size_write() does need locking around it
912  * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
913  * can be lost, resulting in subsequent i_size_read() calls spinning forever.
914  */
i_size_write(struct inode * inode,loff_t i_size)915 static inline void i_size_write(struct inode *inode, loff_t i_size)
916 {
917 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
918 	preempt_disable();
919 	write_seqcount_begin(&inode->i_size_seqcount);
920 	inode->i_size = i_size;
921 	write_seqcount_end(&inode->i_size_seqcount);
922 	preempt_enable();
923 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
924 	preempt_disable();
925 	inode->i_size = i_size;
926 	preempt_enable();
927 #else
928 	inode->i_size = i_size;
929 #endif
930 }
931 
iminor(const struct inode * inode)932 static inline unsigned iminor(const struct inode *inode)
933 {
934 	return MINOR(inode->i_rdev);
935 }
936 
imajor(const struct inode * inode)937 static inline unsigned imajor(const struct inode *inode)
938 {
939 	return MAJOR(inode->i_rdev);
940 }
941 
942 extern struct block_device *I_BDEV(struct inode *inode);
943 
944 struct fown_struct {
945 	rwlock_t lock;          /* protects pid, uid, euid fields */
946 	struct pid *pid;	/* pid or -pgrp where SIGIO should be sent */
947 	enum pid_type pid_type;	/* Kind of process group SIGIO should be sent to */
948 	uid_t uid, euid;	/* uid/euid of process setting the owner */
949 	int signum;		/* posix.1b rt signal to be delivered on IO */
950 };
951 
952 /*
953  * Track a single file's readahead state
954  */
955 struct file_ra_state {
956 	pgoff_t start;			/* where readahead started */
957 	unsigned int size;		/* # of readahead pages */
958 	unsigned int async_size;	/* do asynchronous readahead when
959 					   there are only # of pages ahead */
960 
961 	unsigned int ra_pages;		/* Maximum readahead window */
962 	unsigned int mmap_miss;		/* Cache miss stat for mmap accesses */
963 	loff_t prev_pos;		/* Cache last read() position */
964 };
965 
966 /*
967  * Check if @index falls in the readahead windows.
968  */
ra_has_index(struct file_ra_state * ra,pgoff_t index)969 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
970 {
971 	return (index >= ra->start &&
972 		index <  ra->start + ra->size);
973 }
974 
975 #define FILE_MNT_WRITE_TAKEN	1
976 #define FILE_MNT_WRITE_RELEASED	2
977 
978 struct file {
979 	/*
980 	 * fu_list becomes invalid after file_free is called and queued via
981 	 * fu_rcuhead for RCU freeing
982 	 */
983 	union {
984 		struct list_head	fu_list;
985 		struct rcu_head 	fu_rcuhead;
986 	} f_u;
987 	struct path		f_path;
988 #define f_dentry	f_path.dentry
989 #define f_vfsmnt	f_path.mnt
990 	const struct file_operations	*f_op;
991 
992 	/*
993 	 * Protects f_ep_links, f_flags, f_pos vs i_size in lseek SEEK_CUR.
994 	 * Must not be taken from IRQ context.
995 	 */
996 	spinlock_t		f_lock;
997 #ifdef CONFIG_SMP
998 	int			f_sb_list_cpu;
999 #endif
1000 	atomic_long_t		f_count;
1001 	unsigned int 		f_flags;
1002 	fmode_t			f_mode;
1003 	loff_t			f_pos;
1004 	struct fown_struct	f_owner;
1005 	const struct cred	*f_cred;
1006 	struct file_ra_state	f_ra;
1007 
1008 	u64			f_version;
1009 #ifdef CONFIG_SECURITY
1010 	void			*f_security;
1011 #endif
1012 	/* needed for tty driver, and maybe others */
1013 	void			*private_data;
1014 
1015 #ifdef CONFIG_EPOLL
1016 	/* Used by fs/eventpoll.c to link all the hooks to this file */
1017 	struct list_head	f_ep_links;
1018 	struct list_head	f_tfile_llink;
1019 #endif /* #ifdef CONFIG_EPOLL */
1020 	struct address_space	*f_mapping;
1021 #ifdef CONFIG_DEBUG_WRITECOUNT
1022 	unsigned long f_mnt_write_state;
1023 #endif
1024 };
1025 
1026 struct file_handle {
1027 	__u32 handle_bytes;
1028 	int handle_type;
1029 	/* file identifier */
1030 	unsigned char f_handle[0];
1031 };
1032 
1033 #define get_file(x)	atomic_long_inc(&(x)->f_count)
1034 #define fput_atomic(x)	atomic_long_add_unless(&(x)->f_count, -1, 1)
1035 #define file_count(x)	atomic_long_read(&(x)->f_count)
1036 
1037 #ifdef CONFIG_DEBUG_WRITECOUNT
file_take_write(struct file * f)1038 static inline void file_take_write(struct file *f)
1039 {
1040 	WARN_ON(f->f_mnt_write_state != 0);
1041 	f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
1042 }
file_release_write(struct file * f)1043 static inline void file_release_write(struct file *f)
1044 {
1045 	f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
1046 }
file_reset_write(struct file * f)1047 static inline void file_reset_write(struct file *f)
1048 {
1049 	f->f_mnt_write_state = 0;
1050 }
file_check_state(struct file * f)1051 static inline void file_check_state(struct file *f)
1052 {
1053 	/*
1054 	 * At this point, either both or neither of these bits
1055 	 * should be set.
1056 	 */
1057 	WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
1058 	WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
1059 }
file_check_writeable(struct file * f)1060 static inline int file_check_writeable(struct file *f)
1061 {
1062 	if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
1063 		return 0;
1064 	printk(KERN_WARNING "writeable file with no "
1065 			    "mnt_want_write()\n");
1066 	WARN_ON(1);
1067 	return -EINVAL;
1068 }
1069 #else /* !CONFIG_DEBUG_WRITECOUNT */
file_take_write(struct file * filp)1070 static inline void file_take_write(struct file *filp) {}
file_release_write(struct file * filp)1071 static inline void file_release_write(struct file *filp) {}
file_reset_write(struct file * filp)1072 static inline void file_reset_write(struct file *filp) {}
file_check_state(struct file * filp)1073 static inline void file_check_state(struct file *filp) {}
file_check_writeable(struct file * filp)1074 static inline int file_check_writeable(struct file *filp)
1075 {
1076 	return 0;
1077 }
1078 #endif /* CONFIG_DEBUG_WRITECOUNT */
1079 
1080 #define	MAX_NON_LFS	((1UL<<31) - 1)
1081 
1082 /* Page cache limit. The filesystems should put that into their s_maxbytes
1083    limits, otherwise bad things can happen in VM. */
1084 #if BITS_PER_LONG==32
1085 #define MAX_LFS_FILESIZE	(((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
1086 #elif BITS_PER_LONG==64
1087 #define MAX_LFS_FILESIZE 	0x7fffffffffffffffUL
1088 #endif
1089 
1090 #define FL_POSIX	1
1091 #define FL_FLOCK	2
1092 #define FL_ACCESS	8	/* not trying to lock, just looking */
1093 #define FL_EXISTS	16	/* when unlocking, test for existence */
1094 #define FL_LEASE	32	/* lease held on this file */
1095 #define FL_CLOSE	64	/* unlock on close */
1096 #define FL_SLEEP	128	/* A blocking lock */
1097 #define FL_DOWNGRADE_PENDING	256 /* Lease is being downgraded */
1098 #define FL_UNLOCK_PENDING	512 /* Lease is being broken */
1099 
1100 /*
1101  * Special return value from posix_lock_file() and vfs_lock_file() for
1102  * asynchronous locking.
1103  */
1104 #define FILE_LOCK_DEFERRED 1
1105 
1106 /*
1107  * The POSIX file lock owner is determined by
1108  * the "struct files_struct" in the thread group
1109  * (or NULL for no owner - BSD locks).
1110  *
1111  * Lockd stuffs a "host" pointer into this.
1112  */
1113 typedef struct files_struct *fl_owner_t;
1114 
1115 struct file_lock_operations {
1116 	void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1117 	void (*fl_release_private)(struct file_lock *);
1118 };
1119 
1120 struct lock_manager_operations {
1121 	int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
1122 	void (*lm_notify)(struct file_lock *);	/* unblock callback */
1123 	int (*lm_grant)(struct file_lock *, struct file_lock *, int);
1124 	void (*lm_release_private)(struct file_lock *);
1125 	void (*lm_break)(struct file_lock *);
1126 	int (*lm_change)(struct file_lock **, int);
1127 };
1128 
1129 struct lock_manager {
1130 	struct list_head list;
1131 };
1132 
1133 void locks_start_grace(struct lock_manager *);
1134 void locks_end_grace(struct lock_manager *);
1135 int locks_in_grace(void);
1136 
1137 /* that will die - we need it for nfs_lock_info */
1138 #include <linux/nfs_fs_i.h>
1139 
1140 struct file_lock {
1141 	struct file_lock *fl_next;	/* singly linked list for this inode  */
1142 	struct list_head fl_link;	/* doubly linked list of all locks */
1143 	struct list_head fl_block;	/* circular list of blocked processes */
1144 	fl_owner_t fl_owner;
1145 	unsigned int fl_flags;
1146 	unsigned char fl_type;
1147 	unsigned int fl_pid;
1148 	struct pid *fl_nspid;
1149 	wait_queue_head_t fl_wait;
1150 	struct file *fl_file;
1151 	loff_t fl_start;
1152 	loff_t fl_end;
1153 
1154 	struct fasync_struct *	fl_fasync; /* for lease break notifications */
1155 	/* for lease breaks: */
1156 	unsigned long fl_break_time;
1157 	unsigned long fl_downgrade_time;
1158 
1159 	const struct file_lock_operations *fl_ops;	/* Callbacks for filesystems */
1160 	const struct lock_manager_operations *fl_lmops;	/* Callbacks for lockmanagers */
1161 	union {
1162 		struct nfs_lock_info	nfs_fl;
1163 		struct nfs4_lock_info	nfs4_fl;
1164 		struct {
1165 			struct list_head link;	/* link in AFS vnode's pending_locks list */
1166 			int state;		/* state of grant or error if -ve */
1167 		} afs;
1168 	} fl_u;
1169 };
1170 
1171 /* The following constant reflects the upper bound of the file/locking space */
1172 #ifndef OFFSET_MAX
1173 #define INT_LIMIT(x)	(~((x)1 << (sizeof(x)*8 - 1)))
1174 #define OFFSET_MAX	INT_LIMIT(loff_t)
1175 #define OFFT_OFFSET_MAX	INT_LIMIT(off_t)
1176 #endif
1177 
1178 #include <linux/fcntl.h>
1179 
1180 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1181 
1182 #ifdef CONFIG_FILE_LOCKING
1183 extern int fcntl_getlk(struct file *, struct flock __user *);
1184 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1185 			struct flock __user *);
1186 
1187 #if BITS_PER_LONG == 32
1188 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
1189 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1190 			struct flock64 __user *);
1191 #endif
1192 
1193 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1194 extern int fcntl_getlease(struct file *filp);
1195 
1196 /* fs/locks.c */
1197 void locks_free_lock(struct file_lock *fl);
1198 extern void locks_init_lock(struct file_lock *);
1199 extern struct file_lock * locks_alloc_lock(void);
1200 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1201 extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
1202 extern void locks_remove_posix(struct file *, fl_owner_t);
1203 extern void locks_remove_flock(struct file *);
1204 extern void locks_release_private(struct file_lock *);
1205 extern void posix_test_lock(struct file *, struct file_lock *);
1206 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1207 extern int posix_lock_file_wait(struct file *, struct file_lock *);
1208 extern int posix_unblock_lock(struct file *, struct file_lock *);
1209 extern int vfs_test_lock(struct file *, struct file_lock *);
1210 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1211 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1212 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1213 extern int __break_lease(struct inode *inode, unsigned int flags);
1214 extern void lease_get_mtime(struct inode *, struct timespec *time);
1215 extern int generic_setlease(struct file *, long, struct file_lock **);
1216 extern int vfs_setlease(struct file *, long, struct file_lock **);
1217 extern int lease_modify(struct file_lock **, int);
1218 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1219 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1220 extern void locks_delete_block(struct file_lock *waiter);
1221 extern void lock_flocks(void);
1222 extern void unlock_flocks(void);
1223 #else /* !CONFIG_FILE_LOCKING */
fcntl_getlk(struct file * file,struct flock __user * user)1224 static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1225 {
1226 	return -EINVAL;
1227 }
1228 
fcntl_setlk(unsigned int fd,struct file * file,unsigned int cmd,struct flock __user * user)1229 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1230 			      unsigned int cmd, struct flock __user *user)
1231 {
1232 	return -EACCES;
1233 }
1234 
1235 #if BITS_PER_LONG == 32
fcntl_getlk64(struct file * file,struct flock64 __user * user)1236 static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1237 {
1238 	return -EINVAL;
1239 }
1240 
fcntl_setlk64(unsigned int fd,struct file * file,unsigned int cmd,struct flock64 __user * user)1241 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1242 				unsigned int cmd, struct flock64 __user *user)
1243 {
1244 	return -EACCES;
1245 }
1246 #endif
fcntl_setlease(unsigned int fd,struct file * filp,long arg)1247 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1248 {
1249 	return 0;
1250 }
1251 
fcntl_getlease(struct file * filp)1252 static inline int fcntl_getlease(struct file *filp)
1253 {
1254 	return 0;
1255 }
1256 
locks_init_lock(struct file_lock * fl)1257 static inline void locks_init_lock(struct file_lock *fl)
1258 {
1259 	return;
1260 }
1261 
__locks_copy_lock(struct file_lock * new,struct file_lock * fl)1262 static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1263 {
1264 	return;
1265 }
1266 
locks_copy_lock(struct file_lock * new,struct file_lock * fl)1267 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1268 {
1269 	return;
1270 }
1271 
locks_remove_posix(struct file * filp,fl_owner_t owner)1272 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1273 {
1274 	return;
1275 }
1276 
locks_remove_flock(struct file * filp)1277 static inline void locks_remove_flock(struct file *filp)
1278 {
1279 	return;
1280 }
1281 
posix_test_lock(struct file * filp,struct file_lock * fl)1282 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1283 {
1284 	return;
1285 }
1286 
posix_lock_file(struct file * filp,struct file_lock * fl,struct file_lock * conflock)1287 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1288 				  struct file_lock *conflock)
1289 {
1290 	return -ENOLCK;
1291 }
1292 
posix_lock_file_wait(struct file * filp,struct file_lock * fl)1293 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1294 {
1295 	return -ENOLCK;
1296 }
1297 
posix_unblock_lock(struct file * filp,struct file_lock * waiter)1298 static inline int posix_unblock_lock(struct file *filp,
1299 				     struct file_lock *waiter)
1300 {
1301 	return -ENOENT;
1302 }
1303 
vfs_test_lock(struct file * filp,struct file_lock * fl)1304 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1305 {
1306 	return 0;
1307 }
1308 
vfs_lock_file(struct file * filp,unsigned int cmd,struct file_lock * fl,struct file_lock * conf)1309 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1310 				struct file_lock *fl, struct file_lock *conf)
1311 {
1312 	return -ENOLCK;
1313 }
1314 
vfs_cancel_lock(struct file * filp,struct file_lock * fl)1315 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1316 {
1317 	return 0;
1318 }
1319 
flock_lock_file_wait(struct file * filp,struct file_lock * request)1320 static inline int flock_lock_file_wait(struct file *filp,
1321 				       struct file_lock *request)
1322 {
1323 	return -ENOLCK;
1324 }
1325 
__break_lease(struct inode * inode,unsigned int mode)1326 static inline int __break_lease(struct inode *inode, unsigned int mode)
1327 {
1328 	return 0;
1329 }
1330 
lease_get_mtime(struct inode * inode,struct timespec * time)1331 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1332 {
1333 	return;
1334 }
1335 
generic_setlease(struct file * filp,long arg,struct file_lock ** flp)1336 static inline int generic_setlease(struct file *filp, long arg,
1337 				    struct file_lock **flp)
1338 {
1339 	return -EINVAL;
1340 }
1341 
vfs_setlease(struct file * filp,long arg,struct file_lock ** lease)1342 static inline int vfs_setlease(struct file *filp, long arg,
1343 			       struct file_lock **lease)
1344 {
1345 	return -EINVAL;
1346 }
1347 
lease_modify(struct file_lock ** before,int arg)1348 static inline int lease_modify(struct file_lock **before, int arg)
1349 {
1350 	return -EINVAL;
1351 }
1352 
lock_may_read(struct inode * inode,loff_t start,unsigned long len)1353 static inline int lock_may_read(struct inode *inode, loff_t start,
1354 				unsigned long len)
1355 {
1356 	return 1;
1357 }
1358 
lock_may_write(struct inode * inode,loff_t start,unsigned long len)1359 static inline int lock_may_write(struct inode *inode, loff_t start,
1360 				 unsigned long len)
1361 {
1362 	return 1;
1363 }
1364 
locks_delete_block(struct file_lock * waiter)1365 static inline void locks_delete_block(struct file_lock *waiter)
1366 {
1367 }
1368 
lock_flocks(void)1369 static inline void lock_flocks(void)
1370 {
1371 }
1372 
unlock_flocks(void)1373 static inline void unlock_flocks(void)
1374 {
1375 }
1376 
1377 #endif /* !CONFIG_FILE_LOCKING */
1378 
1379 
1380 struct fasync_struct {
1381 	spinlock_t		fa_lock;
1382 	int			magic;
1383 	int			fa_fd;
1384 	struct fasync_struct	*fa_next; /* singly linked list */
1385 	struct file		*fa_file;
1386 	struct rcu_head		fa_rcu;
1387 };
1388 
1389 #define FASYNC_MAGIC 0x4601
1390 
1391 /* SMP safe fasync helpers: */
1392 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1393 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1394 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1395 extern struct fasync_struct *fasync_alloc(void);
1396 extern void fasync_free(struct fasync_struct *);
1397 
1398 /* can be called from interrupts */
1399 extern void kill_fasync(struct fasync_struct **, int, int);
1400 
1401 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1402 extern int f_setown(struct file *filp, unsigned long arg, int force);
1403 extern void f_delown(struct file *filp);
1404 extern pid_t f_getown(struct file *filp);
1405 extern int send_sigurg(struct fown_struct *fown);
1406 
1407 /*
1408  *	Umount options
1409  */
1410 
1411 #define MNT_FORCE	0x00000001	/* Attempt to forcibily umount */
1412 #define MNT_DETACH	0x00000002	/* Just detach from the tree */
1413 #define MNT_EXPIRE	0x00000004	/* Mark for expiry */
1414 #define UMOUNT_NOFOLLOW	0x00000008	/* Don't follow symlink on umount */
1415 #define UMOUNT_UNUSED	0x80000000	/* Flag guaranteed to be unused */
1416 
1417 extern struct list_head super_blocks;
1418 extern spinlock_t sb_lock;
1419 
1420 struct super_block {
1421 	struct list_head	s_list;		/* Keep this first */
1422 	dev_t			s_dev;		/* search index; _not_ kdev_t */
1423 	unsigned char		s_dirt;
1424 	unsigned char		s_blocksize_bits;
1425 	unsigned long		s_blocksize;
1426 	loff_t			s_maxbytes;	/* Max file size */
1427 	struct file_system_type	*s_type;
1428 	const struct super_operations	*s_op;
1429 	const struct dquot_operations	*dq_op;
1430 	const struct quotactl_ops	*s_qcop;
1431 	const struct export_operations *s_export_op;
1432 	unsigned long		s_flags;
1433 	unsigned long		s_magic;
1434 	struct dentry		*s_root;
1435 	struct rw_semaphore	s_umount;
1436 	struct mutex		s_lock;
1437 	int			s_count;
1438 	atomic_t		s_active;
1439 #ifdef CONFIG_SECURITY
1440 	void                    *s_security;
1441 #endif
1442 	const struct xattr_handler **s_xattr;
1443 
1444 	struct list_head	s_inodes;	/* all inodes */
1445 	struct hlist_bl_head	s_anon;		/* anonymous dentries for (nfs) exporting */
1446 #ifdef CONFIG_SMP
1447 	struct list_head __percpu *s_files;
1448 #else
1449 	struct list_head	s_files;
1450 #endif
1451 	struct list_head	s_mounts;	/* list of mounts; _not_ for fs use */
1452 	/* s_dentry_lru, s_nr_dentry_unused protected by dcache.c lru locks */
1453 	struct list_head	s_dentry_lru;	/* unused dentry lru */
1454 	int			s_nr_dentry_unused;	/* # of dentry on lru */
1455 
1456 	/* s_inode_lru_lock protects s_inode_lru and s_nr_inodes_unused */
1457 	spinlock_t		s_inode_lru_lock ____cacheline_aligned_in_smp;
1458 	struct list_head	s_inode_lru;		/* unused inode lru */
1459 	int			s_nr_inodes_unused;	/* # of inodes on lru */
1460 
1461 	struct block_device	*s_bdev;
1462 	struct backing_dev_info *s_bdi;
1463 	struct mtd_info		*s_mtd;
1464 	struct hlist_node	s_instances;
1465 	struct quota_info	s_dquot;	/* Diskquota specific options */
1466 
1467 	int			s_frozen;
1468 	wait_queue_head_t	s_wait_unfrozen;
1469 
1470 	char s_id[32];				/* Informational name */
1471 	u8 s_uuid[16];				/* UUID */
1472 
1473 	void 			*s_fs_info;	/* Filesystem private info */
1474 	unsigned int		s_max_links;
1475 	fmode_t			s_mode;
1476 
1477 	/* Granularity of c/m/atime in ns.
1478 	   Cannot be worse than a second */
1479 	u32		   s_time_gran;
1480 
1481 	/*
1482 	 * The next field is for VFS *only*. No filesystems have any business
1483 	 * even looking at it. You had been warned.
1484 	 */
1485 	struct mutex s_vfs_rename_mutex;	/* Kludge */
1486 
1487 	/*
1488 	 * Filesystem subtype.  If non-empty the filesystem type field
1489 	 * in /proc/mounts will be "type.subtype"
1490 	 */
1491 	char *s_subtype;
1492 
1493 	/*
1494 	 * Saved mount options for lazy filesystems using
1495 	 * generic_show_options()
1496 	 */
1497 	char __rcu *s_options;
1498 	const struct dentry_operations *s_d_op; /* default d_op for dentries */
1499 
1500 	/*
1501 	 * Saved pool identifier for cleancache (-1 means none)
1502 	 */
1503 	int cleancache_poolid;
1504 
1505 	struct shrinker s_shrink;	/* per-sb shrinker handle */
1506 
1507 	/* Number of inodes with nlink == 0 but still referenced */
1508 	atomic_long_t s_remove_count;
1509 
1510 	/* Being remounted read-only */
1511 	int s_readonly_remount;
1512 };
1513 
1514 /* superblock cache pruning functions */
1515 extern void prune_icache_sb(struct super_block *sb, int nr_to_scan);
1516 extern void prune_dcache_sb(struct super_block *sb, int nr_to_scan);
1517 
1518 extern struct timespec current_fs_time(struct super_block *sb);
1519 
1520 /*
1521  * Snapshotting support.
1522  */
1523 enum {
1524 	SB_UNFROZEN = 0,
1525 	SB_FREEZE_WRITE	= 1,
1526 	SB_FREEZE_TRANS = 2,
1527 };
1528 
1529 #define vfs_check_frozen(sb, level) \
1530 	wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
1531 
1532 /*
1533  * until VFS tracks user namespaces for inodes, just make all files
1534  * belong to init_user_ns
1535  */
1536 extern struct user_namespace init_user_ns;
1537 #define inode_userns(inode) (&init_user_ns)
1538 extern bool inode_owner_or_capable(const struct inode *inode);
1539 
1540 /* not quite ready to be deprecated, but... */
1541 extern void lock_super(struct super_block *);
1542 extern void unlock_super(struct super_block *);
1543 
1544 /*
1545  * VFS helper functions..
1546  */
1547 extern int vfs_create(struct inode *, struct dentry *, umode_t, struct nameidata *);
1548 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1549 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1550 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1551 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1552 extern int vfs_rmdir(struct inode *, struct dentry *);
1553 extern int vfs_unlink(struct inode *, struct dentry *);
1554 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1555 
1556 /*
1557  * VFS dentry helper functions.
1558  */
1559 extern void dentry_unhash(struct dentry *dentry);
1560 
1561 /*
1562  * VFS file helper functions.
1563  */
1564 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1565 			umode_t mode);
1566 /*
1567  * VFS FS_IOC_FIEMAP helper definitions.
1568  */
1569 struct fiemap_extent_info {
1570 	unsigned int fi_flags;		/* Flags as passed from user */
1571 	unsigned int fi_extents_mapped;	/* Number of mapped extents */
1572 	unsigned int fi_extents_max;	/* Size of fiemap_extent array */
1573 	struct fiemap_extent __user *fi_extents_start; /* Start of
1574 							fiemap_extent array */
1575 };
1576 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1577 			    u64 phys, u64 len, u32 flags);
1578 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1579 
1580 /*
1581  * File types
1582  *
1583  * NOTE! These match bits 12..15 of stat.st_mode
1584  * (ie "(i_mode >> 12) & 15").
1585  */
1586 #define DT_UNKNOWN	0
1587 #define DT_FIFO		1
1588 #define DT_CHR		2
1589 #define DT_DIR		4
1590 #define DT_BLK		6
1591 #define DT_REG		8
1592 #define DT_LNK		10
1593 #define DT_SOCK		12
1594 #define DT_WHT		14
1595 
1596 /*
1597  * This is the "filldir" function type, used by readdir() to let
1598  * the kernel specify what kind of dirent layout it wants to have.
1599  * This allows the kernel to read directories into kernel space or
1600  * to have different dirent layouts depending on the binary type.
1601  */
1602 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1603 struct block_device_operations;
1604 
1605 /* These macros are for out of kernel modules to test that
1606  * the kernel supports the unlocked_ioctl and compat_ioctl
1607  * fields in struct file_operations. */
1608 #define HAVE_COMPAT_IOCTL 1
1609 #define HAVE_UNLOCKED_IOCTL 1
1610 
1611 struct file_operations {
1612 	struct module *owner;
1613 	loff_t (*llseek) (struct file *, loff_t, int);
1614 	ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1615 	ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1616 	ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1617 	ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1618 	int (*readdir) (struct file *, void *, filldir_t);
1619 	unsigned int (*poll) (struct file *, struct poll_table_struct *);
1620 	long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1621 	long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1622 	int (*mmap) (struct file *, struct vm_area_struct *);
1623 	int (*open) (struct inode *, struct file *);
1624 	int (*flush) (struct file *, fl_owner_t id);
1625 	int (*release) (struct inode *, struct file *);
1626 	int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1627 	int (*aio_fsync) (struct kiocb *, int datasync);
1628 	int (*fasync) (int, struct file *, int);
1629 	int (*lock) (struct file *, int, struct file_lock *);
1630 	ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1631 	unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1632 	int (*check_flags)(int);
1633 	int (*flock) (struct file *, int, struct file_lock *);
1634 	ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1635 	ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1636 	int (*setlease)(struct file *, long, struct file_lock **);
1637 	long (*fallocate)(struct file *file, int mode, loff_t offset,
1638 			  loff_t len);
1639 };
1640 
1641 struct inode_operations {
1642 	struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1643 	void * (*follow_link) (struct dentry *, struct nameidata *);
1644 	int (*permission) (struct inode *, int);
1645 	struct posix_acl * (*get_acl)(struct inode *, int);
1646 
1647 	int (*readlink) (struct dentry *, char __user *,int);
1648 	void (*put_link) (struct dentry *, struct nameidata *, void *);
1649 
1650 	int (*create) (struct inode *,struct dentry *,umode_t,struct nameidata *);
1651 	int (*link) (struct dentry *,struct inode *,struct dentry *);
1652 	int (*unlink) (struct inode *,struct dentry *);
1653 	int (*symlink) (struct inode *,struct dentry *,const char *);
1654 	int (*mkdir) (struct inode *,struct dentry *,umode_t);
1655 	int (*rmdir) (struct inode *,struct dentry *);
1656 	int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1657 	int (*rename) (struct inode *, struct dentry *,
1658 			struct inode *, struct dentry *);
1659 	void (*truncate) (struct inode *);
1660 	int (*setattr) (struct dentry *, struct iattr *);
1661 	int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1662 	int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1663 	ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1664 	ssize_t (*listxattr) (struct dentry *, char *, size_t);
1665 	int (*removexattr) (struct dentry *, const char *);
1666 	void (*truncate_range)(struct inode *, loff_t, loff_t);
1667 	int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1668 		      u64 len);
1669 } ____cacheline_aligned;
1670 
1671 struct seq_file;
1672 
1673 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1674 			      unsigned long nr_segs, unsigned long fast_segs,
1675 			      struct iovec *fast_pointer,
1676 			      struct iovec **ret_pointer,
1677 			      int check_access);
1678 
1679 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1680 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1681 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1682 		unsigned long, loff_t *);
1683 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1684 		unsigned long, loff_t *);
1685 
1686 struct super_operations {
1687    	struct inode *(*alloc_inode)(struct super_block *sb);
1688 	void (*destroy_inode)(struct inode *);
1689 
1690    	void (*dirty_inode) (struct inode *, int flags);
1691 	int (*write_inode) (struct inode *, struct writeback_control *wbc);
1692 	int (*drop_inode) (struct inode *);
1693 	void (*evict_inode) (struct inode *);
1694 	void (*put_super) (struct super_block *);
1695 	void (*write_super) (struct super_block *);
1696 	int (*sync_fs)(struct super_block *sb, int wait);
1697 	int (*freeze_fs) (struct super_block *);
1698 	int (*unfreeze_fs) (struct super_block *);
1699 	int (*statfs) (struct dentry *, struct kstatfs *);
1700 	int (*remount_fs) (struct super_block *, int *, char *);
1701 	void (*umount_begin) (struct super_block *);
1702 
1703 	int (*show_options)(struct seq_file *, struct dentry *);
1704 	int (*show_devname)(struct seq_file *, struct dentry *);
1705 	int (*show_path)(struct seq_file *, struct dentry *);
1706 	int (*show_stats)(struct seq_file *, struct dentry *);
1707 #ifdef CONFIG_QUOTA
1708 	ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1709 	ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1710 #endif
1711 	int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1712 	int (*nr_cached_objects)(struct super_block *);
1713 	void (*free_cached_objects)(struct super_block *, int);
1714 };
1715 
1716 /*
1717  * Inode state bits.  Protected by inode->i_lock
1718  *
1719  * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1720  * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1721  *
1722  * Four bits define the lifetime of an inode.  Initially, inodes are I_NEW,
1723  * until that flag is cleared.  I_WILL_FREE, I_FREEING and I_CLEAR are set at
1724  * various stages of removing an inode.
1725  *
1726  * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1727  *
1728  * I_DIRTY_SYNC		Inode is dirty, but doesn't have to be written on
1729  *			fdatasync().  i_atime is the usual cause.
1730  * I_DIRTY_DATASYNC	Data-related inode changes pending. We keep track of
1731  *			these changes separately from I_DIRTY_SYNC so that we
1732  *			don't have to write inode on fdatasync() when only
1733  *			mtime has changed in it.
1734  * I_DIRTY_PAGES	Inode has dirty pages.  Inode itself may be clean.
1735  * I_NEW		Serves as both a mutex and completion notification.
1736  *			New inodes set I_NEW.  If two processes both create
1737  *			the same inode, one of them will release its inode and
1738  *			wait for I_NEW to be released before returning.
1739  *			Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1740  *			also cause waiting on I_NEW, without I_NEW actually
1741  *			being set.  find_inode() uses this to prevent returning
1742  *			nearly-dead inodes.
1743  * I_WILL_FREE		Must be set when calling write_inode_now() if i_count
1744  *			is zero.  I_FREEING must be set when I_WILL_FREE is
1745  *			cleared.
1746  * I_FREEING		Set when inode is about to be freed but still has dirty
1747  *			pages or buffers attached or the inode itself is still
1748  *			dirty.
1749  * I_CLEAR		Added by end_writeback().  In this state the inode is clean
1750  *			and can be destroyed.  Inode keeps I_FREEING.
1751  *
1752  *			Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1753  *			prohibited for many purposes.  iget() must wait for
1754  *			the inode to be completely released, then create it
1755  *			anew.  Other functions will just ignore such inodes,
1756  *			if appropriate.  I_NEW is used for waiting.
1757  *
1758  * I_SYNC		Synchonized write of dirty inode data.  The bits is
1759  *			set during data writeback, and cleared with a wakeup
1760  *			on the bit address once it is done.
1761  *
1762  * I_REFERENCED		Marks the inode as recently references on the LRU list.
1763  *
1764  * I_DIO_WAKEUP		Never set.  Only used as a key for wait_on_bit().
1765  *
1766  * Q: What is the difference between I_WILL_FREE and I_FREEING?
1767  */
1768 #define I_DIRTY_SYNC		(1 << 0)
1769 #define I_DIRTY_DATASYNC	(1 << 1)
1770 #define I_DIRTY_PAGES		(1 << 2)
1771 #define __I_NEW			3
1772 #define I_NEW			(1 << __I_NEW)
1773 #define I_WILL_FREE		(1 << 4)
1774 #define I_FREEING		(1 << 5)
1775 #define I_CLEAR			(1 << 6)
1776 #define __I_SYNC		7
1777 #define I_SYNC			(1 << __I_SYNC)
1778 #define I_REFERENCED		(1 << 8)
1779 #define __I_DIO_WAKEUP		9
1780 #define I_DIO_WAKEUP		(1 << I_DIO_WAKEUP)
1781 
1782 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1783 
1784 extern void __mark_inode_dirty(struct inode *, int);
mark_inode_dirty(struct inode * inode)1785 static inline void mark_inode_dirty(struct inode *inode)
1786 {
1787 	__mark_inode_dirty(inode, I_DIRTY);
1788 }
1789 
mark_inode_dirty_sync(struct inode * inode)1790 static inline void mark_inode_dirty_sync(struct inode *inode)
1791 {
1792 	__mark_inode_dirty(inode, I_DIRTY_SYNC);
1793 }
1794 
1795 extern void inc_nlink(struct inode *inode);
1796 extern void drop_nlink(struct inode *inode);
1797 extern void clear_nlink(struct inode *inode);
1798 extern void set_nlink(struct inode *inode, unsigned int nlink);
1799 
inode_inc_link_count(struct inode * inode)1800 static inline void inode_inc_link_count(struct inode *inode)
1801 {
1802 	inc_nlink(inode);
1803 	mark_inode_dirty(inode);
1804 }
1805 
inode_dec_link_count(struct inode * inode)1806 static inline void inode_dec_link_count(struct inode *inode)
1807 {
1808 	drop_nlink(inode);
1809 	mark_inode_dirty(inode);
1810 }
1811 
1812 /**
1813  * inode_inc_iversion - increments i_version
1814  * @inode: inode that need to be updated
1815  *
1816  * Every time the inode is modified, the i_version field will be incremented.
1817  * The filesystem has to be mounted with i_version flag
1818  */
1819 
inode_inc_iversion(struct inode * inode)1820 static inline void inode_inc_iversion(struct inode *inode)
1821 {
1822        spin_lock(&inode->i_lock);
1823        inode->i_version++;
1824        spin_unlock(&inode->i_lock);
1825 }
1826 
1827 extern void touch_atime(struct path *);
file_accessed(struct file * file)1828 static inline void file_accessed(struct file *file)
1829 {
1830 	if (!(file->f_flags & O_NOATIME))
1831 		touch_atime(&file->f_path);
1832 }
1833 
1834 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1835 int sync_inode_metadata(struct inode *inode, int wait);
1836 
1837 struct file_system_type {
1838 	const char *name;
1839 	int fs_flags;
1840 	struct dentry *(*mount) (struct file_system_type *, int,
1841 		       const char *, void *);
1842 	void (*kill_sb) (struct super_block *);
1843 	struct module *owner;
1844 	struct file_system_type * next;
1845 	struct hlist_head fs_supers;
1846 
1847 	struct lock_class_key s_lock_key;
1848 	struct lock_class_key s_umount_key;
1849 	struct lock_class_key s_vfs_rename_key;
1850 
1851 	struct lock_class_key i_lock_key;
1852 	struct lock_class_key i_mutex_key;
1853 	struct lock_class_key i_mutex_dir_key;
1854 };
1855 
1856 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1857 	void *data, int (*fill_super)(struct super_block *, void *, int));
1858 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1859 	int flags, const char *dev_name, void *data,
1860 	int (*fill_super)(struct super_block *, void *, int));
1861 extern struct dentry *mount_single(struct file_system_type *fs_type,
1862 	int flags, void *data,
1863 	int (*fill_super)(struct super_block *, void *, int));
1864 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1865 	int flags, void *data,
1866 	int (*fill_super)(struct super_block *, void *, int));
1867 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
1868 void generic_shutdown_super(struct super_block *sb);
1869 void kill_block_super(struct super_block *sb);
1870 void kill_anon_super(struct super_block *sb);
1871 void kill_litter_super(struct super_block *sb);
1872 void deactivate_super(struct super_block *sb);
1873 void deactivate_locked_super(struct super_block *sb);
1874 int set_anon_super(struct super_block *s, void *data);
1875 int get_anon_bdev(dev_t *);
1876 void free_anon_bdev(dev_t);
1877 struct super_block *sget(struct file_system_type *type,
1878 			int (*test)(struct super_block *,void *),
1879 			int (*set)(struct super_block *,void *),
1880 			void *data);
1881 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1882 	const struct super_operations *ops,
1883 	const struct dentry_operations *dops,
1884 	unsigned long);
1885 
1886 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1887 #define fops_get(fops) \
1888 	(((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1889 #define fops_put(fops) \
1890 	do { if (fops) module_put((fops)->owner); } while(0)
1891 
1892 extern int register_filesystem(struct file_system_type *);
1893 extern int unregister_filesystem(struct file_system_type *);
1894 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1895 #define kern_mount(type) kern_mount_data(type, NULL)
1896 extern void kern_unmount(struct vfsmount *mnt);
1897 extern int may_umount_tree(struct vfsmount *);
1898 extern int may_umount(struct vfsmount *);
1899 extern long do_mount(char *, char *, char *, unsigned long, void *);
1900 extern struct vfsmount *collect_mounts(struct path *);
1901 extern void drop_collected_mounts(struct vfsmount *);
1902 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1903 			  struct vfsmount *);
1904 extern int vfs_statfs(struct path *, struct kstatfs *);
1905 extern int user_statfs(const char __user *, struct kstatfs *);
1906 extern int fd_statfs(int, struct kstatfs *);
1907 extern int vfs_ustat(dev_t, struct kstatfs *);
1908 extern int freeze_super(struct super_block *super);
1909 extern int thaw_super(struct super_block *super);
1910 extern bool our_mnt(struct vfsmount *mnt);
1911 
1912 extern int current_umask(void);
1913 
1914 /* /sys/fs */
1915 extern struct kobject *fs_kobj;
1916 
1917 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1918 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1919 
1920 #define FLOCK_VERIFY_READ  1
1921 #define FLOCK_VERIFY_WRITE 2
1922 
1923 #ifdef CONFIG_FILE_LOCKING
1924 extern int locks_mandatory_locked(struct inode *);
1925 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1926 
1927 /*
1928  * Candidates for mandatory locking have the setgid bit set
1929  * but no group execute bit -  an otherwise meaningless combination.
1930  */
1931 
__mandatory_lock(struct inode * ino)1932 static inline int __mandatory_lock(struct inode *ino)
1933 {
1934 	return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1935 }
1936 
1937 /*
1938  * ... and these candidates should be on MS_MANDLOCK mounted fs,
1939  * otherwise these will be advisory locks
1940  */
1941 
mandatory_lock(struct inode * ino)1942 static inline int mandatory_lock(struct inode *ino)
1943 {
1944 	return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1945 }
1946 
locks_verify_locked(struct inode * inode)1947 static inline int locks_verify_locked(struct inode *inode)
1948 {
1949 	if (mandatory_lock(inode))
1950 		return locks_mandatory_locked(inode);
1951 	return 0;
1952 }
1953 
locks_verify_truncate(struct inode * inode,struct file * filp,loff_t size)1954 static inline int locks_verify_truncate(struct inode *inode,
1955 				    struct file *filp,
1956 				    loff_t size)
1957 {
1958 	if (inode->i_flock && mandatory_lock(inode))
1959 		return locks_mandatory_area(
1960 			FLOCK_VERIFY_WRITE, inode, filp,
1961 			size < inode->i_size ? size : inode->i_size,
1962 			(size < inode->i_size ? inode->i_size - size
1963 			 : size - inode->i_size)
1964 		);
1965 	return 0;
1966 }
1967 
break_lease(struct inode * inode,unsigned int mode)1968 static inline int break_lease(struct inode *inode, unsigned int mode)
1969 {
1970 	if (inode->i_flock)
1971 		return __break_lease(inode, mode);
1972 	return 0;
1973 }
1974 #else /* !CONFIG_FILE_LOCKING */
locks_mandatory_locked(struct inode * inode)1975 static inline int locks_mandatory_locked(struct inode *inode)
1976 {
1977 	return 0;
1978 }
1979 
locks_mandatory_area(int rw,struct inode * inode,struct file * filp,loff_t offset,size_t count)1980 static inline int locks_mandatory_area(int rw, struct inode *inode,
1981 				       struct file *filp, loff_t offset,
1982 				       size_t count)
1983 {
1984 	return 0;
1985 }
1986 
__mandatory_lock(struct inode * inode)1987 static inline int __mandatory_lock(struct inode *inode)
1988 {
1989 	return 0;
1990 }
1991 
mandatory_lock(struct inode * inode)1992 static inline int mandatory_lock(struct inode *inode)
1993 {
1994 	return 0;
1995 }
1996 
locks_verify_locked(struct inode * inode)1997 static inline int locks_verify_locked(struct inode *inode)
1998 {
1999 	return 0;
2000 }
2001 
locks_verify_truncate(struct inode * inode,struct file * filp,size_t size)2002 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2003 					size_t size)
2004 {
2005 	return 0;
2006 }
2007 
break_lease(struct inode * inode,unsigned int mode)2008 static inline int break_lease(struct inode *inode, unsigned int mode)
2009 {
2010 	return 0;
2011 }
2012 
2013 #endif /* CONFIG_FILE_LOCKING */
2014 
2015 /* fs/open.c */
2016 
2017 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2018 		       struct file *filp);
2019 extern int do_fallocate(struct file *file, int mode, loff_t offset,
2020 			loff_t len);
2021 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2022 			umode_t mode);
2023 extern struct file *filp_open(const char *, int, umode_t);
2024 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2025 				   const char *, int);
2026 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int,
2027 				 const struct cred *);
2028 extern int filp_close(struct file *, fl_owner_t id);
2029 extern char * getname(const char __user *);
2030 
2031 /* fs/ioctl.c */
2032 
2033 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2034 
2035 /* fs/dcache.c */
2036 extern void __init vfs_caches_init_early(void);
2037 extern void __init vfs_caches_init(unsigned long);
2038 
2039 extern struct kmem_cache *names_cachep;
2040 
2041 #define __getname_gfp(gfp)	kmem_cache_alloc(names_cachep, (gfp))
2042 #define __getname()		__getname_gfp(GFP_KERNEL)
2043 #define __putname(name)		kmem_cache_free(names_cachep, (void *)(name))
2044 #ifndef CONFIG_AUDITSYSCALL
2045 #define putname(name)   __putname(name)
2046 #else
2047 extern void putname(const char *name);
2048 #endif
2049 
2050 #ifdef CONFIG_BLOCK
2051 extern int register_blkdev(unsigned int, const char *);
2052 extern void unregister_blkdev(unsigned int, const char *);
2053 extern struct block_device *bdget(dev_t);
2054 extern struct block_device *bdgrab(struct block_device *bdev);
2055 extern void bd_set_size(struct block_device *, loff_t size);
2056 extern sector_t blkdev_max_block(struct block_device *bdev);
2057 extern void bd_forget(struct inode *inode);
2058 extern void bdput(struct block_device *);
2059 extern void invalidate_bdev(struct block_device *);
2060 extern int sync_blockdev(struct block_device *bdev);
2061 extern void kill_bdev(struct block_device *);
2062 extern struct super_block *freeze_bdev(struct block_device *);
2063 extern void emergency_thaw_all(void);
2064 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2065 extern int fsync_bdev(struct block_device *);
2066 #else
bd_forget(struct inode * inode)2067 static inline void bd_forget(struct inode *inode) {}
sync_blockdev(struct block_device * bdev)2068 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
kill_bdev(struct block_device * bdev)2069 static inline void kill_bdev(struct block_device *bdev) {}
invalidate_bdev(struct block_device * bdev)2070 static inline void invalidate_bdev(struct block_device *bdev) {}
2071 
freeze_bdev(struct block_device * sb)2072 static inline struct super_block *freeze_bdev(struct block_device *sb)
2073 {
2074 	return NULL;
2075 }
2076 
thaw_bdev(struct block_device * bdev,struct super_block * sb)2077 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2078 {
2079 	return 0;
2080 }
2081 #endif
2082 extern int sync_filesystem(struct super_block *);
2083 extern const struct file_operations def_blk_fops;
2084 extern const struct file_operations def_chr_fops;
2085 extern const struct file_operations bad_sock_fops;
2086 extern const struct file_operations def_fifo_fops;
2087 #ifdef CONFIG_BLOCK
2088 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2089 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2090 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2091 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2092 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2093 					       void *holder);
2094 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2095 					      void *holder);
2096 extern int blkdev_put(struct block_device *bdev, fmode_t mode);
2097 #ifdef CONFIG_SYSFS
2098 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2099 extern void bd_unlink_disk_holder(struct block_device *bdev,
2100 				  struct gendisk *disk);
2101 #else
bd_link_disk_holder(struct block_device * bdev,struct gendisk * disk)2102 static inline int bd_link_disk_holder(struct block_device *bdev,
2103 				      struct gendisk *disk)
2104 {
2105 	return 0;
2106 }
bd_unlink_disk_holder(struct block_device * bdev,struct gendisk * disk)2107 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2108 					 struct gendisk *disk)
2109 {
2110 }
2111 #endif
2112 #endif
2113 
2114 /* fs/char_dev.c */
2115 #define CHRDEV_MAJOR_HASH_SIZE	255
2116 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2117 extern int register_chrdev_region(dev_t, unsigned, const char *);
2118 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2119 			     unsigned int count, const char *name,
2120 			     const struct file_operations *fops);
2121 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2122 				unsigned int count, const char *name);
2123 extern void unregister_chrdev_region(dev_t, unsigned);
2124 extern void chrdev_show(struct seq_file *,off_t);
2125 
register_chrdev(unsigned int major,const char * name,const struct file_operations * fops)2126 static inline int register_chrdev(unsigned int major, const char *name,
2127 				  const struct file_operations *fops)
2128 {
2129 	return __register_chrdev(major, 0, 256, name, fops);
2130 }
2131 
unregister_chrdev(unsigned int major,const char * name)2132 static inline void unregister_chrdev(unsigned int major, const char *name)
2133 {
2134 	__unregister_chrdev(major, 0, 256, name);
2135 }
2136 
2137 /* fs/block_dev.c */
2138 #define BDEVNAME_SIZE	32	/* Largest string for a blockdev identifier */
2139 #define BDEVT_SIZE	10	/* Largest string for MAJ:MIN for blkdev */
2140 
2141 #ifdef CONFIG_BLOCK
2142 #define BLKDEV_MAJOR_HASH_SIZE	255
2143 extern const char *__bdevname(dev_t, char *buffer);
2144 extern const char *bdevname(struct block_device *bdev, char *buffer);
2145 extern struct block_device *lookup_bdev(const char *);
2146 extern void blkdev_show(struct seq_file *,off_t);
2147 
2148 #else
2149 #define BLKDEV_MAJOR_HASH_SIZE	0
2150 #endif
2151 
2152 extern void init_special_inode(struct inode *, umode_t, dev_t);
2153 
2154 /* Invalid inode operations -- fs/bad_inode.c */
2155 extern void make_bad_inode(struct inode *);
2156 extern int is_bad_inode(struct inode *);
2157 
2158 extern const struct file_operations read_pipefifo_fops;
2159 extern const struct file_operations write_pipefifo_fops;
2160 extern const struct file_operations rdwr_pipefifo_fops;
2161 
2162 #ifdef CONFIG_BLOCK
2163 /*
2164  * return READ, READA, or WRITE
2165  */
2166 #define bio_rw(bio)		((bio)->bi_rw & (RW_MASK | RWA_MASK))
2167 
2168 /*
2169  * return data direction, READ or WRITE
2170  */
2171 #define bio_data_dir(bio)	((bio)->bi_rw & 1)
2172 
2173 extern void check_disk_size_change(struct gendisk *disk,
2174 				   struct block_device *bdev);
2175 extern int revalidate_disk(struct gendisk *);
2176 extern int check_disk_change(struct block_device *);
2177 extern int __invalidate_device(struct block_device *, bool);
2178 extern int invalidate_partition(struct gendisk *, int);
2179 #endif
2180 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2181 					pgoff_t start, pgoff_t end);
2182 
invalidate_remote_inode(struct inode * inode)2183 static inline void invalidate_remote_inode(struct inode *inode)
2184 {
2185 	if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2186 	    S_ISLNK(inode->i_mode))
2187 		invalidate_mapping_pages(inode->i_mapping, 0, -1);
2188 }
2189 extern int invalidate_inode_pages2(struct address_space *mapping);
2190 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2191 					 pgoff_t start, pgoff_t end);
2192 extern int write_inode_now(struct inode *, int);
2193 extern int filemap_fdatawrite(struct address_space *);
2194 extern int filemap_flush(struct address_space *);
2195 extern int filemap_fdatawait(struct address_space *);
2196 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2197 				   loff_t lend);
2198 extern int filemap_write_and_wait(struct address_space *mapping);
2199 extern int filemap_write_and_wait_range(struct address_space *mapping,
2200 				        loff_t lstart, loff_t lend);
2201 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2202 				loff_t start, loff_t end, int sync_mode);
2203 extern int filemap_fdatawrite_range(struct address_space *mapping,
2204 				loff_t start, loff_t end);
2205 
2206 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2207 			   int datasync);
2208 extern int vfs_fsync(struct file *file, int datasync);
2209 extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
2210 extern void sync_supers(void);
2211 extern void emergency_sync(void);
2212 extern void emergency_remount(void);
2213 #ifdef CONFIG_BLOCK
2214 extern sector_t bmap(struct inode *, sector_t);
2215 #endif
2216 extern int notify_change(struct dentry *, struct iattr *);
2217 extern int inode_permission(struct inode *, int);
2218 extern int generic_permission(struct inode *, int);
2219 
execute_ok(struct inode * inode)2220 static inline bool execute_ok(struct inode *inode)
2221 {
2222 	return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2223 }
2224 
2225 /*
2226  * get_write_access() gets write permission for a file.
2227  * put_write_access() releases this write permission.
2228  * This is used for regular files.
2229  * We cannot support write (and maybe mmap read-write shared) accesses and
2230  * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2231  * can have the following values:
2232  * 0: no writers, no VM_DENYWRITE mappings
2233  * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2234  * > 0: (i_writecount) users are writing to the file.
2235  *
2236  * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2237  * except for the cases where we don't hold i_writecount yet. Then we need to
2238  * use {get,deny}_write_access() - these functions check the sign and refuse
2239  * to do the change if sign is wrong.
2240  */
get_write_access(struct inode * inode)2241 static inline int get_write_access(struct inode *inode)
2242 {
2243 	return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2244 }
deny_write_access(struct file * file)2245 static inline int deny_write_access(struct file *file)
2246 {
2247 	struct inode *inode = file->f_path.dentry->d_inode;
2248 	return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2249 }
put_write_access(struct inode * inode)2250 static inline void put_write_access(struct inode * inode)
2251 {
2252 	atomic_dec(&inode->i_writecount);
2253 }
allow_write_access(struct file * file)2254 static inline void allow_write_access(struct file *file)
2255 {
2256 	if (file)
2257 		atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
2258 }
2259 #ifdef CONFIG_IMA
i_readcount_dec(struct inode * inode)2260 static inline void i_readcount_dec(struct inode *inode)
2261 {
2262 	BUG_ON(!atomic_read(&inode->i_readcount));
2263 	atomic_dec(&inode->i_readcount);
2264 }
i_readcount_inc(struct inode * inode)2265 static inline void i_readcount_inc(struct inode *inode)
2266 {
2267 	atomic_inc(&inode->i_readcount);
2268 }
2269 #else
i_readcount_dec(struct inode * inode)2270 static inline void i_readcount_dec(struct inode *inode)
2271 {
2272 	return;
2273 }
i_readcount_inc(struct inode * inode)2274 static inline void i_readcount_inc(struct inode *inode)
2275 {
2276 	return;
2277 }
2278 #endif
2279 extern int do_pipe_flags(int *, int);
2280 extern struct file *create_read_pipe(struct file *f, int flags);
2281 extern struct file *create_write_pipe(int flags);
2282 extern void free_write_pipe(struct file *);
2283 
2284 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2285 extern struct file * open_exec(const char *);
2286 
2287 /* fs/dcache.c -- generic fs support functions */
2288 extern int is_subdir(struct dentry *, struct dentry *);
2289 extern int path_is_under(struct path *, struct path *);
2290 extern ino_t find_inode_number(struct dentry *, struct qstr *);
2291 
2292 #include <linux/err.h>
2293 
2294 /* needed for stackable file system support */
2295 extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
2296 
2297 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
2298 
2299 extern int inode_init_always(struct super_block *, struct inode *);
2300 extern void inode_init_once(struct inode *);
2301 extern void address_space_init_once(struct address_space *mapping);
2302 extern void ihold(struct inode * inode);
2303 extern void iput(struct inode *);
2304 extern struct inode * igrab(struct inode *);
2305 extern ino_t iunique(struct super_block *, ino_t);
2306 extern int inode_needs_sync(struct inode *inode);
2307 extern int generic_delete_inode(struct inode *inode);
generic_drop_inode(struct inode * inode)2308 static inline int generic_drop_inode(struct inode *inode)
2309 {
2310 	return !inode->i_nlink || inode_unhashed(inode);
2311 }
2312 
2313 extern struct inode *ilookup5_nowait(struct super_block *sb,
2314 		unsigned long hashval, int (*test)(struct inode *, void *),
2315 		void *data);
2316 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2317 		int (*test)(struct inode *, void *), void *data);
2318 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2319 
2320 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2321 extern struct inode * iget_locked(struct super_block *, unsigned long);
2322 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2323 extern int insert_inode_locked(struct inode *);
2324 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2325 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2326 #else
lockdep_annotate_inode_mutex_key(struct inode * inode)2327 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2328 #endif
2329 extern void unlock_new_inode(struct inode *);
2330 extern unsigned int get_next_ino(void);
2331 
2332 extern void __iget(struct inode * inode);
2333 extern void iget_failed(struct inode *);
2334 extern void end_writeback(struct inode *);
2335 extern void __destroy_inode(struct inode *);
2336 extern struct inode *new_inode_pseudo(struct super_block *sb);
2337 extern struct inode *new_inode(struct super_block *sb);
2338 extern void free_inode_nonrcu(struct inode *inode);
2339 extern int should_remove_suid(struct dentry *);
2340 extern int file_remove_suid(struct file *);
2341 
2342 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
insert_inode_hash(struct inode * inode)2343 static inline void insert_inode_hash(struct inode *inode)
2344 {
2345 	__insert_inode_hash(inode, inode->i_ino);
2346 }
2347 
2348 extern void __remove_inode_hash(struct inode *);
remove_inode_hash(struct inode * inode)2349 static inline void remove_inode_hash(struct inode *inode)
2350 {
2351 	if (!inode_unhashed(inode))
2352 		__remove_inode_hash(inode);
2353 }
2354 
2355 extern void inode_sb_list_add(struct inode *inode);
2356 
2357 #ifdef CONFIG_BLOCK
2358 extern void submit_bio(int, struct bio *);
2359 extern int bdev_read_only(struct block_device *);
2360 #endif
2361 extern int set_blocksize(struct block_device *, int);
2362 extern int sb_set_blocksize(struct super_block *, int);
2363 extern int sb_min_blocksize(struct super_block *, int);
2364 
2365 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2366 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2367 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
2368 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2369 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2370 extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2371 		loff_t *);
2372 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2373 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2374 		unsigned long *, loff_t, loff_t *, size_t, size_t);
2375 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2376 		unsigned long, loff_t, loff_t *, size_t, ssize_t);
2377 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2378 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2379 extern int generic_segment_checks(const struct iovec *iov,
2380 		unsigned long *nr_segs, size_t *count, int access_flags);
2381 
2382 /* fs/block_dev.c */
2383 extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2384 				unsigned long nr_segs, loff_t pos);
2385 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2386 			int datasync);
2387 extern void block_sync_page(struct page *page);
2388 
2389 /* fs/splice.c */
2390 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2391 		struct pipe_inode_info *, size_t, unsigned int);
2392 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2393 		struct pipe_inode_info *, size_t, unsigned int);
2394 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
2395 		struct file *, loff_t *, size_t, unsigned int);
2396 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2397 		struct file *out, loff_t *, size_t len, unsigned int flags);
2398 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2399 		size_t len, unsigned int flags);
2400 
2401 extern void
2402 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2403 extern loff_t noop_llseek(struct file *file, loff_t offset, int origin);
2404 extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
2405 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
2406 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2407 		int origin, loff_t maxsize);
2408 extern int generic_file_open(struct inode * inode, struct file * filp);
2409 extern int nonseekable_open(struct inode * inode, struct file * filp);
2410 
2411 #ifdef CONFIG_FS_XIP
2412 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2413 			     loff_t *ppos);
2414 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2415 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2416 			      size_t len, loff_t *ppos);
2417 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2418 #else
xip_truncate_page(struct address_space * mapping,loff_t from)2419 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2420 {
2421 	return 0;
2422 }
2423 #endif
2424 
2425 #ifdef CONFIG_BLOCK
2426 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2427 			    loff_t file_offset);
2428 
2429 enum {
2430 	/* need locking between buffered and direct access */
2431 	DIO_LOCKING	= 0x01,
2432 
2433 	/* filesystem does not support filling holes */
2434 	DIO_SKIP_HOLES	= 0x02,
2435 };
2436 
2437 void dio_end_io(struct bio *bio, int error);
2438 void inode_dio_wait(struct inode *inode);
2439 void inode_dio_done(struct inode *inode);
2440 
2441 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2442 	struct block_device *bdev, const struct iovec *iov, loff_t offset,
2443 	unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2444 	dio_submit_t submit_io,	int flags);
2445 
blockdev_direct_IO(int rw,struct kiocb * iocb,struct inode * inode,const struct iovec * iov,loff_t offset,unsigned long nr_segs,get_block_t get_block)2446 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2447 		struct inode *inode, const struct iovec *iov, loff_t offset,
2448 		unsigned long nr_segs, get_block_t get_block)
2449 {
2450 	return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
2451 				    offset, nr_segs, get_block, NULL, NULL,
2452 				    DIO_LOCKING | DIO_SKIP_HOLES);
2453 }
2454 #else
inode_dio_wait(struct inode * inode)2455 static inline void inode_dio_wait(struct inode *inode)
2456 {
2457 }
2458 #endif
2459 
2460 extern const struct file_operations generic_ro_fops;
2461 
2462 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2463 
2464 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2465 extern int vfs_follow_link(struct nameidata *, const char *);
2466 extern int page_readlink(struct dentry *, char __user *, int);
2467 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2468 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2469 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2470 		int nofs);
2471 extern int page_symlink(struct inode *inode, const char *symname, int len);
2472 extern const struct inode_operations page_symlink_inode_operations;
2473 extern int generic_readlink(struct dentry *, char __user *, int);
2474 extern void generic_fillattr(struct inode *, struct kstat *);
2475 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2476 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2477 void inode_add_bytes(struct inode *inode, loff_t bytes);
2478 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2479 loff_t inode_get_bytes(struct inode *inode);
2480 void inode_set_bytes(struct inode *inode, loff_t bytes);
2481 
2482 extern int vfs_readdir(struct file *, filldir_t, void *);
2483 
2484 extern int vfs_stat(const char __user *, struct kstat *);
2485 extern int vfs_lstat(const char __user *, struct kstat *);
2486 extern int vfs_fstat(unsigned int, struct kstat *);
2487 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2488 
2489 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2490 		    unsigned long arg);
2491 extern int __generic_block_fiemap(struct inode *inode,
2492 				  struct fiemap_extent_info *fieinfo,
2493 				  loff_t start, loff_t len,
2494 				  get_block_t *get_block);
2495 extern int generic_block_fiemap(struct inode *inode,
2496 				struct fiemap_extent_info *fieinfo, u64 start,
2497 				u64 len, get_block_t *get_block);
2498 
2499 extern void get_filesystem(struct file_system_type *fs);
2500 extern void put_filesystem(struct file_system_type *fs);
2501 extern struct file_system_type *get_fs_type(const char *name);
2502 extern struct super_block *get_super(struct block_device *);
2503 extern struct super_block *get_super_thawed(struct block_device *);
2504 extern struct super_block *get_active_super(struct block_device *bdev);
2505 extern void drop_super(struct super_block *sb);
2506 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2507 extern void iterate_supers_type(struct file_system_type *,
2508 			        void (*)(struct super_block *, void *), void *);
2509 
2510 extern int dcache_dir_open(struct inode *, struct file *);
2511 extern int dcache_dir_close(struct inode *, struct file *);
2512 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2513 extern int dcache_readdir(struct file *, void *, filldir_t);
2514 extern int simple_setattr(struct dentry *, struct iattr *);
2515 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2516 extern int simple_statfs(struct dentry *, struct kstatfs *);
2517 extern int simple_open(struct inode *inode, struct file *file);
2518 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2519 extern int simple_unlink(struct inode *, struct dentry *);
2520 extern int simple_rmdir(struct inode *, struct dentry *);
2521 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2522 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2523 extern int simple_empty(struct dentry *);
2524 extern int simple_readpage(struct file *file, struct page *page);
2525 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2526 			loff_t pos, unsigned len, unsigned flags,
2527 			struct page **pagep, void **fsdata);
2528 extern int simple_write_end(struct file *file, struct address_space *mapping,
2529 			loff_t pos, unsigned len, unsigned copied,
2530 			struct page *page, void *fsdata);
2531 
2532 extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
2533 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2534 extern const struct file_operations simple_dir_operations;
2535 extern const struct inode_operations simple_dir_inode_operations;
2536 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2537 struct dentry *d_alloc_name(struct dentry *, const char *);
2538 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2539 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2540 extern void simple_release_fs(struct vfsmount **mount, int *count);
2541 
2542 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2543 			loff_t *ppos, const void *from, size_t available);
2544 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2545 		const void __user *from, size_t count);
2546 
2547 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2548 
2549 extern int generic_check_addressable(unsigned, u64);
2550 
2551 #ifdef CONFIG_MIGRATION
2552 extern int buffer_migrate_page(struct address_space *,
2553 				struct page *, struct page *,
2554 				enum migrate_mode);
2555 #else
2556 #define buffer_migrate_page NULL
2557 #endif
2558 
2559 extern int inode_change_ok(const struct inode *, struct iattr *);
2560 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2561 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2562 
2563 extern void file_update_time(struct file *file);
2564 
2565 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2566 extern void save_mount_options(struct super_block *sb, char *options);
2567 extern void replace_mount_options(struct super_block *sb, char *options);
2568 
parent_ino(struct dentry * dentry)2569 static inline ino_t parent_ino(struct dentry *dentry)
2570 {
2571 	ino_t res;
2572 
2573 	/*
2574 	 * Don't strictly need d_lock here? If the parent ino could change
2575 	 * then surely we'd have a deeper race in the caller?
2576 	 */
2577 	spin_lock(&dentry->d_lock);
2578 	res = dentry->d_parent->d_inode->i_ino;
2579 	spin_unlock(&dentry->d_lock);
2580 	return res;
2581 }
2582 
2583 /* Transaction based IO helpers */
2584 
2585 /*
2586  * An argresp is stored in an allocated page and holds the
2587  * size of the argument or response, along with its content
2588  */
2589 struct simple_transaction_argresp {
2590 	ssize_t size;
2591 	char data[0];
2592 };
2593 
2594 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2595 
2596 char *simple_transaction_get(struct file *file, const char __user *buf,
2597 				size_t size);
2598 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2599 				size_t size, loff_t *pos);
2600 int simple_transaction_release(struct inode *inode, struct file *file);
2601 
2602 void simple_transaction_set(struct file *file, size_t n);
2603 
2604 /*
2605  * simple attribute files
2606  *
2607  * These attributes behave similar to those in sysfs:
2608  *
2609  * Writing to an attribute immediately sets a value, an open file can be
2610  * written to multiple times.
2611  *
2612  * Reading from an attribute creates a buffer from the value that might get
2613  * read with multiple read calls. When the attribute has been read
2614  * completely, no further read calls are possible until the file is opened
2615  * again.
2616  *
2617  * All attributes contain a text representation of a numeric value
2618  * that are accessed with the get() and set() functions.
2619  */
2620 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt)		\
2621 static int __fops ## _open(struct inode *inode, struct file *file)	\
2622 {									\
2623 	__simple_attr_check_format(__fmt, 0ull);			\
2624 	return simple_attr_open(inode, file, __get, __set, __fmt);	\
2625 }									\
2626 static const struct file_operations __fops = {				\
2627 	.owner	 = THIS_MODULE,						\
2628 	.open	 = __fops ## _open,					\
2629 	.release = simple_attr_release,					\
2630 	.read	 = simple_attr_read,					\
2631 	.write	 = simple_attr_write,					\
2632 	.llseek	 = generic_file_llseek,					\
2633 };
2634 
2635 static inline __printf(1, 2)
__simple_attr_check_format(const char * fmt,...)2636 void __simple_attr_check_format(const char *fmt, ...)
2637 {
2638 	/* don't do anything, just let the compiler check the arguments; */
2639 }
2640 
2641 int simple_attr_open(struct inode *inode, struct file *file,
2642 		     int (*get)(void *, u64 *), int (*set)(void *, u64),
2643 		     const char *fmt);
2644 int simple_attr_release(struct inode *inode, struct file *file);
2645 ssize_t simple_attr_read(struct file *file, char __user *buf,
2646 			 size_t len, loff_t *ppos);
2647 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2648 			  size_t len, loff_t *ppos);
2649 
2650 struct ctl_table;
2651 int proc_nr_files(struct ctl_table *table, int write,
2652 		  void __user *buffer, size_t *lenp, loff_t *ppos);
2653 int proc_nr_dentry(struct ctl_table *table, int write,
2654 		  void __user *buffer, size_t *lenp, loff_t *ppos);
2655 int proc_nr_inodes(struct ctl_table *table, int write,
2656 		   void __user *buffer, size_t *lenp, loff_t *ppos);
2657 int __init get_filesystem_list(char *buf);
2658 
2659 #define __FMODE_EXEC		((__force int) FMODE_EXEC)
2660 #define __FMODE_NONOTIFY	((__force int) FMODE_NONOTIFY)
2661 
2662 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2663 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2664 					    (flag & __FMODE_NONOTIFY)))
2665 
is_sxid(umode_t mode)2666 static inline int is_sxid(umode_t mode)
2667 {
2668 	return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2669 }
2670 
inode_has_no_xattr(struct inode * inode)2671 static inline void inode_has_no_xattr(struct inode *inode)
2672 {
2673 	if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2674 		inode->i_flags |= S_NOSEC;
2675 }
2676 
2677 #endif /* __KERNEL__ */
2678 #endif /* _LINUX_FS_H */
2679