Merge branch 'for-3.4' of git://linux-nfs.org/~bfields/linux
[GitHub/mt8127/android_kernel_alcatel_ttab.git] / include / linux / fs.h
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
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
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 */
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 */
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
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 */
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 */
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 write_seqcount_begin(&inode->i_size_seqcount);
919 inode->i_size = i_size;
920 write_seqcount_end(&inode->i_size_seqcount);
921 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
922 preempt_disable();
923 inode->i_size = i_size;
924 preempt_enable();
925 #else
926 inode->i_size = i_size;
927 #endif
928 }
929
930 static inline unsigned iminor(const struct inode *inode)
931 {
932 return MINOR(inode->i_rdev);
933 }
934
935 static inline unsigned imajor(const struct inode *inode)
936 {
937 return MAJOR(inode->i_rdev);
938 }
939
940 extern struct block_device *I_BDEV(struct inode *inode);
941
942 struct fown_struct {
943 rwlock_t lock; /* protects pid, uid, euid fields */
944 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
945 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
946 uid_t uid, euid; /* uid/euid of process setting the owner */
947 int signum; /* posix.1b rt signal to be delivered on IO */
948 };
949
950 /*
951 * Track a single file's readahead state
952 */
953 struct file_ra_state {
954 pgoff_t start; /* where readahead started */
955 unsigned int size; /* # of readahead pages */
956 unsigned int async_size; /* do asynchronous readahead when
957 there are only # of pages ahead */
958
959 unsigned int ra_pages; /* Maximum readahead window */
960 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
961 loff_t prev_pos; /* Cache last read() position */
962 };
963
964 /*
965 * Check if @index falls in the readahead windows.
966 */
967 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
968 {
969 return (index >= ra->start &&
970 index < ra->start + ra->size);
971 }
972
973 #define FILE_MNT_WRITE_TAKEN 1
974 #define FILE_MNT_WRITE_RELEASED 2
975
976 struct file {
977 /*
978 * fu_list becomes invalid after file_free is called and queued via
979 * fu_rcuhead for RCU freeing
980 */
981 union {
982 struct list_head fu_list;
983 struct rcu_head fu_rcuhead;
984 } f_u;
985 struct path f_path;
986 #define f_dentry f_path.dentry
987 #define f_vfsmnt f_path.mnt
988 const struct file_operations *f_op;
989
990 /*
991 * Protects f_ep_links, f_flags, f_pos vs i_size in lseek SEEK_CUR.
992 * Must not be taken from IRQ context.
993 */
994 spinlock_t f_lock;
995 #ifdef CONFIG_SMP
996 int f_sb_list_cpu;
997 #endif
998 atomic_long_t f_count;
999 unsigned int f_flags;
1000 fmode_t f_mode;
1001 loff_t f_pos;
1002 struct fown_struct f_owner;
1003 const struct cred *f_cred;
1004 struct file_ra_state f_ra;
1005
1006 u64 f_version;
1007 #ifdef CONFIG_SECURITY
1008 void *f_security;
1009 #endif
1010 /* needed for tty driver, and maybe others */
1011 void *private_data;
1012
1013 #ifdef CONFIG_EPOLL
1014 /* Used by fs/eventpoll.c to link all the hooks to this file */
1015 struct list_head f_ep_links;
1016 struct list_head f_tfile_llink;
1017 #endif /* #ifdef CONFIG_EPOLL */
1018 struct address_space *f_mapping;
1019 #ifdef CONFIG_DEBUG_WRITECOUNT
1020 unsigned long f_mnt_write_state;
1021 #endif
1022 };
1023
1024 struct file_handle {
1025 __u32 handle_bytes;
1026 int handle_type;
1027 /* file identifier */
1028 unsigned char f_handle[0];
1029 };
1030
1031 #define get_file(x) atomic_long_inc(&(x)->f_count)
1032 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
1033 #define file_count(x) atomic_long_read(&(x)->f_count)
1034
1035 #ifdef CONFIG_DEBUG_WRITECOUNT
1036 static inline void file_take_write(struct file *f)
1037 {
1038 WARN_ON(f->f_mnt_write_state != 0);
1039 f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
1040 }
1041 static inline void file_release_write(struct file *f)
1042 {
1043 f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
1044 }
1045 static inline void file_reset_write(struct file *f)
1046 {
1047 f->f_mnt_write_state = 0;
1048 }
1049 static inline void file_check_state(struct file *f)
1050 {
1051 /*
1052 * At this point, either both or neither of these bits
1053 * should be set.
1054 */
1055 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
1056 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
1057 }
1058 static inline int file_check_writeable(struct file *f)
1059 {
1060 if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
1061 return 0;
1062 printk(KERN_WARNING "writeable file with no "
1063 "mnt_want_write()\n");
1064 WARN_ON(1);
1065 return -EINVAL;
1066 }
1067 #else /* !CONFIG_DEBUG_WRITECOUNT */
1068 static inline void file_take_write(struct file *filp) {}
1069 static inline void file_release_write(struct file *filp) {}
1070 static inline void file_reset_write(struct file *filp) {}
1071 static inline void file_check_state(struct file *filp) {}
1072 static inline int file_check_writeable(struct file *filp)
1073 {
1074 return 0;
1075 }
1076 #endif /* CONFIG_DEBUG_WRITECOUNT */
1077
1078 #define MAX_NON_LFS ((1UL<<31) - 1)
1079
1080 /* Page cache limit. The filesystems should put that into their s_maxbytes
1081 limits, otherwise bad things can happen in VM. */
1082 #if BITS_PER_LONG==32
1083 #define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
1084 #elif BITS_PER_LONG==64
1085 #define MAX_LFS_FILESIZE 0x7fffffffffffffffUL
1086 #endif
1087
1088 #define FL_POSIX 1
1089 #define FL_FLOCK 2
1090 #define FL_ACCESS 8 /* not trying to lock, just looking */
1091 #define FL_EXISTS 16 /* when unlocking, test for existence */
1092 #define FL_LEASE 32 /* lease held on this file */
1093 #define FL_CLOSE 64 /* unlock on close */
1094 #define FL_SLEEP 128 /* A blocking lock */
1095 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
1096 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
1097
1098 /*
1099 * Special return value from posix_lock_file() and vfs_lock_file() for
1100 * asynchronous locking.
1101 */
1102 #define FILE_LOCK_DEFERRED 1
1103
1104 /*
1105 * The POSIX file lock owner is determined by
1106 * the "struct files_struct" in the thread group
1107 * (or NULL for no owner - BSD locks).
1108 *
1109 * Lockd stuffs a "host" pointer into this.
1110 */
1111 typedef struct files_struct *fl_owner_t;
1112
1113 struct file_lock_operations {
1114 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1115 void (*fl_release_private)(struct file_lock *);
1116 };
1117
1118 struct lock_manager_operations {
1119 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
1120 void (*lm_notify)(struct file_lock *); /* unblock callback */
1121 int (*lm_grant)(struct file_lock *, struct file_lock *, int);
1122 void (*lm_release_private)(struct file_lock *);
1123 void (*lm_break)(struct file_lock *);
1124 int (*lm_change)(struct file_lock **, int);
1125 };
1126
1127 struct lock_manager {
1128 struct list_head list;
1129 };
1130
1131 void locks_start_grace(struct lock_manager *);
1132 void locks_end_grace(struct lock_manager *);
1133 int locks_in_grace(void);
1134
1135 /* that will die - we need it for nfs_lock_info */
1136 #include <linux/nfs_fs_i.h>
1137
1138 struct file_lock {
1139 struct file_lock *fl_next; /* singly linked list for this inode */
1140 struct list_head fl_link; /* doubly linked list of all locks */
1141 struct list_head fl_block; /* circular list of blocked processes */
1142 fl_owner_t fl_owner;
1143 unsigned int fl_flags;
1144 unsigned char fl_type;
1145 unsigned int fl_pid;
1146 struct pid *fl_nspid;
1147 wait_queue_head_t fl_wait;
1148 struct file *fl_file;
1149 loff_t fl_start;
1150 loff_t fl_end;
1151
1152 struct fasync_struct * fl_fasync; /* for lease break notifications */
1153 /* for lease breaks: */
1154 unsigned long fl_break_time;
1155 unsigned long fl_downgrade_time;
1156
1157 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1158 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1159 union {
1160 struct nfs_lock_info nfs_fl;
1161 struct nfs4_lock_info nfs4_fl;
1162 struct {
1163 struct list_head link; /* link in AFS vnode's pending_locks list */
1164 int state; /* state of grant or error if -ve */
1165 } afs;
1166 } fl_u;
1167 };
1168
1169 /* The following constant reflects the upper bound of the file/locking space */
1170 #ifndef OFFSET_MAX
1171 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1172 #define OFFSET_MAX INT_LIMIT(loff_t)
1173 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1174 #endif
1175
1176 #include <linux/fcntl.h>
1177
1178 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1179
1180 #ifdef CONFIG_FILE_LOCKING
1181 extern int fcntl_getlk(struct file *, struct flock __user *);
1182 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1183 struct flock __user *);
1184
1185 #if BITS_PER_LONG == 32
1186 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
1187 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1188 struct flock64 __user *);
1189 #endif
1190
1191 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1192 extern int fcntl_getlease(struct file *filp);
1193
1194 /* fs/locks.c */
1195 void locks_free_lock(struct file_lock *fl);
1196 extern void locks_init_lock(struct file_lock *);
1197 extern struct file_lock * locks_alloc_lock(void);
1198 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1199 extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
1200 extern void locks_remove_posix(struct file *, fl_owner_t);
1201 extern void locks_remove_flock(struct file *);
1202 extern void locks_release_private(struct file_lock *);
1203 extern void posix_test_lock(struct file *, struct file_lock *);
1204 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1205 extern int posix_lock_file_wait(struct file *, struct file_lock *);
1206 extern int posix_unblock_lock(struct file *, struct file_lock *);
1207 extern int vfs_test_lock(struct file *, struct file_lock *);
1208 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1209 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1210 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1211 extern int __break_lease(struct inode *inode, unsigned int flags);
1212 extern void lease_get_mtime(struct inode *, struct timespec *time);
1213 extern int generic_setlease(struct file *, long, struct file_lock **);
1214 extern int vfs_setlease(struct file *, long, struct file_lock **);
1215 extern int lease_modify(struct file_lock **, int);
1216 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1217 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1218 extern void lock_flocks(void);
1219 extern void unlock_flocks(void);
1220 #else /* !CONFIG_FILE_LOCKING */
1221 static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1222 {
1223 return -EINVAL;
1224 }
1225
1226 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1227 unsigned int cmd, struct flock __user *user)
1228 {
1229 return -EACCES;
1230 }
1231
1232 #if BITS_PER_LONG == 32
1233 static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1234 {
1235 return -EINVAL;
1236 }
1237
1238 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1239 unsigned int cmd, struct flock64 __user *user)
1240 {
1241 return -EACCES;
1242 }
1243 #endif
1244 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1245 {
1246 return 0;
1247 }
1248
1249 static inline int fcntl_getlease(struct file *filp)
1250 {
1251 return 0;
1252 }
1253
1254 static inline void locks_init_lock(struct file_lock *fl)
1255 {
1256 return;
1257 }
1258
1259 static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1260 {
1261 return;
1262 }
1263
1264 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1265 {
1266 return;
1267 }
1268
1269 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1270 {
1271 return;
1272 }
1273
1274 static inline void locks_remove_flock(struct file *filp)
1275 {
1276 return;
1277 }
1278
1279 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1280 {
1281 return;
1282 }
1283
1284 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1285 struct file_lock *conflock)
1286 {
1287 return -ENOLCK;
1288 }
1289
1290 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1291 {
1292 return -ENOLCK;
1293 }
1294
1295 static inline int posix_unblock_lock(struct file *filp,
1296 struct file_lock *waiter)
1297 {
1298 return -ENOENT;
1299 }
1300
1301 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1302 {
1303 return 0;
1304 }
1305
1306 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1307 struct file_lock *fl, struct file_lock *conf)
1308 {
1309 return -ENOLCK;
1310 }
1311
1312 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1313 {
1314 return 0;
1315 }
1316
1317 static inline int flock_lock_file_wait(struct file *filp,
1318 struct file_lock *request)
1319 {
1320 return -ENOLCK;
1321 }
1322
1323 static inline int __break_lease(struct inode *inode, unsigned int mode)
1324 {
1325 return 0;
1326 }
1327
1328 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1329 {
1330 return;
1331 }
1332
1333 static inline int generic_setlease(struct file *filp, long arg,
1334 struct file_lock **flp)
1335 {
1336 return -EINVAL;
1337 }
1338
1339 static inline int vfs_setlease(struct file *filp, long arg,
1340 struct file_lock **lease)
1341 {
1342 return -EINVAL;
1343 }
1344
1345 static inline int lease_modify(struct file_lock **before, int arg)
1346 {
1347 return -EINVAL;
1348 }
1349
1350 static inline int lock_may_read(struct inode *inode, loff_t start,
1351 unsigned long len)
1352 {
1353 return 1;
1354 }
1355
1356 static inline int lock_may_write(struct inode *inode, loff_t start,
1357 unsigned long len)
1358 {
1359 return 1;
1360 }
1361
1362 static inline void lock_flocks(void)
1363 {
1364 }
1365
1366 static inline void unlock_flocks(void)
1367 {
1368 }
1369
1370 #endif /* !CONFIG_FILE_LOCKING */
1371
1372
1373 struct fasync_struct {
1374 spinlock_t fa_lock;
1375 int magic;
1376 int fa_fd;
1377 struct fasync_struct *fa_next; /* singly linked list */
1378 struct file *fa_file;
1379 struct rcu_head fa_rcu;
1380 };
1381
1382 #define FASYNC_MAGIC 0x4601
1383
1384 /* SMP safe fasync helpers: */
1385 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1386 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1387 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1388 extern struct fasync_struct *fasync_alloc(void);
1389 extern void fasync_free(struct fasync_struct *);
1390
1391 /* can be called from interrupts */
1392 extern void kill_fasync(struct fasync_struct **, int, int);
1393
1394 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1395 extern int f_setown(struct file *filp, unsigned long arg, int force);
1396 extern void f_delown(struct file *filp);
1397 extern pid_t f_getown(struct file *filp);
1398 extern int send_sigurg(struct fown_struct *fown);
1399
1400 /*
1401 * Umount options
1402 */
1403
1404 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1405 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1406 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1407 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1408 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1409
1410 extern struct list_head super_blocks;
1411 extern spinlock_t sb_lock;
1412
1413 struct super_block {
1414 struct list_head s_list; /* Keep this first */
1415 dev_t s_dev; /* search index; _not_ kdev_t */
1416 unsigned char s_dirt;
1417 unsigned char s_blocksize_bits;
1418 unsigned long s_blocksize;
1419 loff_t s_maxbytes; /* Max file size */
1420 struct file_system_type *s_type;
1421 const struct super_operations *s_op;
1422 const struct dquot_operations *dq_op;
1423 const struct quotactl_ops *s_qcop;
1424 const struct export_operations *s_export_op;
1425 unsigned long s_flags;
1426 unsigned long s_magic;
1427 struct dentry *s_root;
1428 struct rw_semaphore s_umount;
1429 struct mutex s_lock;
1430 int s_count;
1431 atomic_t s_active;
1432 #ifdef CONFIG_SECURITY
1433 void *s_security;
1434 #endif
1435 const struct xattr_handler **s_xattr;
1436
1437 struct list_head s_inodes; /* all inodes */
1438 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1439 #ifdef CONFIG_SMP
1440 struct list_head __percpu *s_files;
1441 #else
1442 struct list_head s_files;
1443 #endif
1444 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1445 /* s_dentry_lru, s_nr_dentry_unused protected by dcache.c lru locks */
1446 struct list_head s_dentry_lru; /* unused dentry lru */
1447 int s_nr_dentry_unused; /* # of dentry on lru */
1448
1449 /* s_inode_lru_lock protects s_inode_lru and s_nr_inodes_unused */
1450 spinlock_t s_inode_lru_lock ____cacheline_aligned_in_smp;
1451 struct list_head s_inode_lru; /* unused inode lru */
1452 int s_nr_inodes_unused; /* # of inodes on lru */
1453
1454 struct block_device *s_bdev;
1455 struct backing_dev_info *s_bdi;
1456 struct mtd_info *s_mtd;
1457 struct hlist_node s_instances;
1458 struct quota_info s_dquot; /* Diskquota specific options */
1459
1460 int s_frozen;
1461 wait_queue_head_t s_wait_unfrozen;
1462
1463 char s_id[32]; /* Informational name */
1464 u8 s_uuid[16]; /* UUID */
1465
1466 void *s_fs_info; /* Filesystem private info */
1467 unsigned int s_max_links;
1468 fmode_t s_mode;
1469
1470 /* Granularity of c/m/atime in ns.
1471 Cannot be worse than a second */
1472 u32 s_time_gran;
1473
1474 /*
1475 * The next field is for VFS *only*. No filesystems have any business
1476 * even looking at it. You had been warned.
1477 */
1478 struct mutex s_vfs_rename_mutex; /* Kludge */
1479
1480 /*
1481 * Filesystem subtype. If non-empty the filesystem type field
1482 * in /proc/mounts will be "type.subtype"
1483 */
1484 char *s_subtype;
1485
1486 /*
1487 * Saved mount options for lazy filesystems using
1488 * generic_show_options()
1489 */
1490 char __rcu *s_options;
1491 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1492
1493 /*
1494 * Saved pool identifier for cleancache (-1 means none)
1495 */
1496 int cleancache_poolid;
1497
1498 struct shrinker s_shrink; /* per-sb shrinker handle */
1499
1500 /* Number of inodes with nlink == 0 but still referenced */
1501 atomic_long_t s_remove_count;
1502
1503 /* Being remounted read-only */
1504 int s_readonly_remount;
1505 };
1506
1507 /* superblock cache pruning functions */
1508 extern void prune_icache_sb(struct super_block *sb, int nr_to_scan);
1509 extern void prune_dcache_sb(struct super_block *sb, int nr_to_scan);
1510
1511 extern struct timespec current_fs_time(struct super_block *sb);
1512
1513 /*
1514 * Snapshotting support.
1515 */
1516 enum {
1517 SB_UNFROZEN = 0,
1518 SB_FREEZE_WRITE = 1,
1519 SB_FREEZE_TRANS = 2,
1520 };
1521
1522 #define vfs_check_frozen(sb, level) \
1523 wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
1524
1525 /*
1526 * until VFS tracks user namespaces for inodes, just make all files
1527 * belong to init_user_ns
1528 */
1529 extern struct user_namespace init_user_ns;
1530 #define inode_userns(inode) (&init_user_ns)
1531 extern bool inode_owner_or_capable(const struct inode *inode);
1532
1533 /* not quite ready to be deprecated, but... */
1534 extern void lock_super(struct super_block *);
1535 extern void unlock_super(struct super_block *);
1536
1537 /*
1538 * VFS helper functions..
1539 */
1540 extern int vfs_create(struct inode *, struct dentry *, umode_t, struct nameidata *);
1541 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1542 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1543 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1544 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1545 extern int vfs_rmdir(struct inode *, struct dentry *);
1546 extern int vfs_unlink(struct inode *, struct dentry *);
1547 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1548
1549 /*
1550 * VFS dentry helper functions.
1551 */
1552 extern void dentry_unhash(struct dentry *dentry);
1553
1554 /*
1555 * VFS file helper functions.
1556 */
1557 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1558 umode_t mode);
1559 /*
1560 * VFS FS_IOC_FIEMAP helper definitions.
1561 */
1562 struct fiemap_extent_info {
1563 unsigned int fi_flags; /* Flags as passed from user */
1564 unsigned int fi_extents_mapped; /* Number of mapped extents */
1565 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1566 struct fiemap_extent __user *fi_extents_start; /* Start of
1567 fiemap_extent array */
1568 };
1569 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1570 u64 phys, u64 len, u32 flags);
1571 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1572
1573 /*
1574 * File types
1575 *
1576 * NOTE! These match bits 12..15 of stat.st_mode
1577 * (ie "(i_mode >> 12) & 15").
1578 */
1579 #define DT_UNKNOWN 0
1580 #define DT_FIFO 1
1581 #define DT_CHR 2
1582 #define DT_DIR 4
1583 #define DT_BLK 6
1584 #define DT_REG 8
1585 #define DT_LNK 10
1586 #define DT_SOCK 12
1587 #define DT_WHT 14
1588
1589 /*
1590 * This is the "filldir" function type, used by readdir() to let
1591 * the kernel specify what kind of dirent layout it wants to have.
1592 * This allows the kernel to read directories into kernel space or
1593 * to have different dirent layouts depending on the binary type.
1594 */
1595 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1596 struct block_device_operations;
1597
1598 /* These macros are for out of kernel modules to test that
1599 * the kernel supports the unlocked_ioctl and compat_ioctl
1600 * fields in struct file_operations. */
1601 #define HAVE_COMPAT_IOCTL 1
1602 #define HAVE_UNLOCKED_IOCTL 1
1603
1604 struct file_operations {
1605 struct module *owner;
1606 loff_t (*llseek) (struct file *, loff_t, int);
1607 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1608 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1609 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1610 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1611 int (*readdir) (struct file *, void *, filldir_t);
1612 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1613 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1614 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1615 int (*mmap) (struct file *, struct vm_area_struct *);
1616 int (*open) (struct inode *, struct file *);
1617 int (*flush) (struct file *, fl_owner_t id);
1618 int (*release) (struct inode *, struct file *);
1619 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1620 int (*aio_fsync) (struct kiocb *, int datasync);
1621 int (*fasync) (int, struct file *, int);
1622 int (*lock) (struct file *, int, struct file_lock *);
1623 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1624 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1625 int (*check_flags)(int);
1626 int (*flock) (struct file *, int, struct file_lock *);
1627 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1628 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1629 int (*setlease)(struct file *, long, struct file_lock **);
1630 long (*fallocate)(struct file *file, int mode, loff_t offset,
1631 loff_t len);
1632 };
1633
1634 struct inode_operations {
1635 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1636 void * (*follow_link) (struct dentry *, struct nameidata *);
1637 int (*permission) (struct inode *, int);
1638 struct posix_acl * (*get_acl)(struct inode *, int);
1639
1640 int (*readlink) (struct dentry *, char __user *,int);
1641 void (*put_link) (struct dentry *, struct nameidata *, void *);
1642
1643 int (*create) (struct inode *,struct dentry *,umode_t,struct nameidata *);
1644 int (*link) (struct dentry *,struct inode *,struct dentry *);
1645 int (*unlink) (struct inode *,struct dentry *);
1646 int (*symlink) (struct inode *,struct dentry *,const char *);
1647 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1648 int (*rmdir) (struct inode *,struct dentry *);
1649 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1650 int (*rename) (struct inode *, struct dentry *,
1651 struct inode *, struct dentry *);
1652 void (*truncate) (struct inode *);
1653 int (*setattr) (struct dentry *, struct iattr *);
1654 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1655 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1656 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1657 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1658 int (*removexattr) (struct dentry *, const char *);
1659 void (*truncate_range)(struct inode *, loff_t, loff_t);
1660 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1661 u64 len);
1662 } ____cacheline_aligned;
1663
1664 struct seq_file;
1665
1666 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1667 unsigned long nr_segs, unsigned long fast_segs,
1668 struct iovec *fast_pointer,
1669 struct iovec **ret_pointer,
1670 int check_access);
1671
1672 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1673 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1674 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1675 unsigned long, loff_t *);
1676 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1677 unsigned long, loff_t *);
1678
1679 struct super_operations {
1680 struct inode *(*alloc_inode)(struct super_block *sb);
1681 void (*destroy_inode)(struct inode *);
1682
1683 void (*dirty_inode) (struct inode *, int flags);
1684 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1685 int (*drop_inode) (struct inode *);
1686 void (*evict_inode) (struct inode *);
1687 void (*put_super) (struct super_block *);
1688 void (*write_super) (struct super_block *);
1689 int (*sync_fs)(struct super_block *sb, int wait);
1690 int (*freeze_fs) (struct super_block *);
1691 int (*unfreeze_fs) (struct super_block *);
1692 int (*statfs) (struct dentry *, struct kstatfs *);
1693 int (*remount_fs) (struct super_block *, int *, char *);
1694 void (*umount_begin) (struct super_block *);
1695
1696 int (*show_options)(struct seq_file *, struct dentry *);
1697 int (*show_devname)(struct seq_file *, struct dentry *);
1698 int (*show_path)(struct seq_file *, struct dentry *);
1699 int (*show_stats)(struct seq_file *, struct dentry *);
1700 #ifdef CONFIG_QUOTA
1701 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1702 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1703 #endif
1704 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1705 int (*nr_cached_objects)(struct super_block *);
1706 void (*free_cached_objects)(struct super_block *, int);
1707 };
1708
1709 /*
1710 * Inode state bits. Protected by inode->i_lock
1711 *
1712 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1713 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1714 *
1715 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1716 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1717 * various stages of removing an inode.
1718 *
1719 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1720 *
1721 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1722 * fdatasync(). i_atime is the usual cause.
1723 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1724 * these changes separately from I_DIRTY_SYNC so that we
1725 * don't have to write inode on fdatasync() when only
1726 * mtime has changed in it.
1727 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1728 * I_NEW Serves as both a mutex and completion notification.
1729 * New inodes set I_NEW. If two processes both create
1730 * the same inode, one of them will release its inode and
1731 * wait for I_NEW to be released before returning.
1732 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1733 * also cause waiting on I_NEW, without I_NEW actually
1734 * being set. find_inode() uses this to prevent returning
1735 * nearly-dead inodes.
1736 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1737 * is zero. I_FREEING must be set when I_WILL_FREE is
1738 * cleared.
1739 * I_FREEING Set when inode is about to be freed but still has dirty
1740 * pages or buffers attached or the inode itself is still
1741 * dirty.
1742 * I_CLEAR Added by end_writeback(). In this state the inode is clean
1743 * and can be destroyed. Inode keeps I_FREEING.
1744 *
1745 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1746 * prohibited for many purposes. iget() must wait for
1747 * the inode to be completely released, then create it
1748 * anew. Other functions will just ignore such inodes,
1749 * if appropriate. I_NEW is used for waiting.
1750 *
1751 * I_SYNC Synchonized write of dirty inode data. The bits is
1752 * set during data writeback, and cleared with a wakeup
1753 * on the bit address once it is done.
1754 *
1755 * I_REFERENCED Marks the inode as recently references on the LRU list.
1756 *
1757 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1758 *
1759 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1760 */
1761 #define I_DIRTY_SYNC (1 << 0)
1762 #define I_DIRTY_DATASYNC (1 << 1)
1763 #define I_DIRTY_PAGES (1 << 2)
1764 #define __I_NEW 3
1765 #define I_NEW (1 << __I_NEW)
1766 #define I_WILL_FREE (1 << 4)
1767 #define I_FREEING (1 << 5)
1768 #define I_CLEAR (1 << 6)
1769 #define __I_SYNC 7
1770 #define I_SYNC (1 << __I_SYNC)
1771 #define I_REFERENCED (1 << 8)
1772 #define __I_DIO_WAKEUP 9
1773 #define I_DIO_WAKEUP (1 << I_DIO_WAKEUP)
1774
1775 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1776
1777 extern void __mark_inode_dirty(struct inode *, int);
1778 static inline void mark_inode_dirty(struct inode *inode)
1779 {
1780 __mark_inode_dirty(inode, I_DIRTY);
1781 }
1782
1783 static inline void mark_inode_dirty_sync(struct inode *inode)
1784 {
1785 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1786 }
1787
1788 extern void inc_nlink(struct inode *inode);
1789 extern void drop_nlink(struct inode *inode);
1790 extern void clear_nlink(struct inode *inode);
1791 extern void set_nlink(struct inode *inode, unsigned int nlink);
1792
1793 static inline void inode_inc_link_count(struct inode *inode)
1794 {
1795 inc_nlink(inode);
1796 mark_inode_dirty(inode);
1797 }
1798
1799 static inline void inode_dec_link_count(struct inode *inode)
1800 {
1801 drop_nlink(inode);
1802 mark_inode_dirty(inode);
1803 }
1804
1805 /**
1806 * inode_inc_iversion - increments i_version
1807 * @inode: inode that need to be updated
1808 *
1809 * Every time the inode is modified, the i_version field will be incremented.
1810 * The filesystem has to be mounted with i_version flag
1811 */
1812
1813 static inline void inode_inc_iversion(struct inode *inode)
1814 {
1815 spin_lock(&inode->i_lock);
1816 inode->i_version++;
1817 spin_unlock(&inode->i_lock);
1818 }
1819
1820 extern void touch_atime(struct path *);
1821 static inline void file_accessed(struct file *file)
1822 {
1823 if (!(file->f_flags & O_NOATIME))
1824 touch_atime(&file->f_path);
1825 }
1826
1827 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1828 int sync_inode_metadata(struct inode *inode, int wait);
1829
1830 struct file_system_type {
1831 const char *name;
1832 int fs_flags;
1833 struct dentry *(*mount) (struct file_system_type *, int,
1834 const char *, void *);
1835 void (*kill_sb) (struct super_block *);
1836 struct module *owner;
1837 struct file_system_type * next;
1838 struct hlist_head fs_supers;
1839
1840 struct lock_class_key s_lock_key;
1841 struct lock_class_key s_umount_key;
1842 struct lock_class_key s_vfs_rename_key;
1843
1844 struct lock_class_key i_lock_key;
1845 struct lock_class_key i_mutex_key;
1846 struct lock_class_key i_mutex_dir_key;
1847 };
1848
1849 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1850 void *data, int (*fill_super)(struct super_block *, void *, int));
1851 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1852 int flags, const char *dev_name, void *data,
1853 int (*fill_super)(struct super_block *, void *, int));
1854 extern struct dentry *mount_single(struct file_system_type *fs_type,
1855 int flags, void *data,
1856 int (*fill_super)(struct super_block *, void *, int));
1857 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1858 int flags, void *data,
1859 int (*fill_super)(struct super_block *, void *, int));
1860 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
1861 void generic_shutdown_super(struct super_block *sb);
1862 void kill_block_super(struct super_block *sb);
1863 void kill_anon_super(struct super_block *sb);
1864 void kill_litter_super(struct super_block *sb);
1865 void deactivate_super(struct super_block *sb);
1866 void deactivate_locked_super(struct super_block *sb);
1867 int set_anon_super(struct super_block *s, void *data);
1868 int get_anon_bdev(dev_t *);
1869 void free_anon_bdev(dev_t);
1870 struct super_block *sget(struct file_system_type *type,
1871 int (*test)(struct super_block *,void *),
1872 int (*set)(struct super_block *,void *),
1873 void *data);
1874 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1875 const struct super_operations *ops,
1876 const struct dentry_operations *dops,
1877 unsigned long);
1878
1879 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1880 #define fops_get(fops) \
1881 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1882 #define fops_put(fops) \
1883 do { if (fops) module_put((fops)->owner); } while(0)
1884
1885 extern int register_filesystem(struct file_system_type *);
1886 extern int unregister_filesystem(struct file_system_type *);
1887 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1888 #define kern_mount(type) kern_mount_data(type, NULL)
1889 extern void kern_unmount(struct vfsmount *mnt);
1890 extern int may_umount_tree(struct vfsmount *);
1891 extern int may_umount(struct vfsmount *);
1892 extern long do_mount(char *, char *, char *, unsigned long, void *);
1893 extern struct vfsmount *collect_mounts(struct path *);
1894 extern void drop_collected_mounts(struct vfsmount *);
1895 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1896 struct vfsmount *);
1897 extern int vfs_statfs(struct path *, struct kstatfs *);
1898 extern int user_statfs(const char __user *, struct kstatfs *);
1899 extern int fd_statfs(int, struct kstatfs *);
1900 extern int vfs_ustat(dev_t, struct kstatfs *);
1901 extern int freeze_super(struct super_block *super);
1902 extern int thaw_super(struct super_block *super);
1903 extern bool our_mnt(struct vfsmount *mnt);
1904
1905 extern int current_umask(void);
1906
1907 /* /sys/fs */
1908 extern struct kobject *fs_kobj;
1909
1910 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1911 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1912
1913 #define FLOCK_VERIFY_READ 1
1914 #define FLOCK_VERIFY_WRITE 2
1915
1916 #ifdef CONFIG_FILE_LOCKING
1917 extern int locks_mandatory_locked(struct inode *);
1918 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1919
1920 /*
1921 * Candidates for mandatory locking have the setgid bit set
1922 * but no group execute bit - an otherwise meaningless combination.
1923 */
1924
1925 static inline int __mandatory_lock(struct inode *ino)
1926 {
1927 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1928 }
1929
1930 /*
1931 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1932 * otherwise these will be advisory locks
1933 */
1934
1935 static inline int mandatory_lock(struct inode *ino)
1936 {
1937 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1938 }
1939
1940 static inline int locks_verify_locked(struct inode *inode)
1941 {
1942 if (mandatory_lock(inode))
1943 return locks_mandatory_locked(inode);
1944 return 0;
1945 }
1946
1947 static inline int locks_verify_truncate(struct inode *inode,
1948 struct file *filp,
1949 loff_t size)
1950 {
1951 if (inode->i_flock && mandatory_lock(inode))
1952 return locks_mandatory_area(
1953 FLOCK_VERIFY_WRITE, inode, filp,
1954 size < inode->i_size ? size : inode->i_size,
1955 (size < inode->i_size ? inode->i_size - size
1956 : size - inode->i_size)
1957 );
1958 return 0;
1959 }
1960
1961 static inline int break_lease(struct inode *inode, unsigned int mode)
1962 {
1963 if (inode->i_flock)
1964 return __break_lease(inode, mode);
1965 return 0;
1966 }
1967 #else /* !CONFIG_FILE_LOCKING */
1968 static inline int locks_mandatory_locked(struct inode *inode)
1969 {
1970 return 0;
1971 }
1972
1973 static inline int locks_mandatory_area(int rw, struct inode *inode,
1974 struct file *filp, loff_t offset,
1975 size_t count)
1976 {
1977 return 0;
1978 }
1979
1980 static inline int __mandatory_lock(struct inode *inode)
1981 {
1982 return 0;
1983 }
1984
1985 static inline int mandatory_lock(struct inode *inode)
1986 {
1987 return 0;
1988 }
1989
1990 static inline int locks_verify_locked(struct inode *inode)
1991 {
1992 return 0;
1993 }
1994
1995 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
1996 size_t size)
1997 {
1998 return 0;
1999 }
2000
2001 static inline int break_lease(struct inode *inode, unsigned int mode)
2002 {
2003 return 0;
2004 }
2005
2006 #endif /* CONFIG_FILE_LOCKING */
2007
2008 /* fs/open.c */
2009
2010 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2011 struct file *filp);
2012 extern int do_fallocate(struct file *file, int mode, loff_t offset,
2013 loff_t len);
2014 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2015 umode_t mode);
2016 extern struct file *filp_open(const char *, int, umode_t);
2017 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2018 const char *, int);
2019 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int,
2020 const struct cred *);
2021 extern int filp_close(struct file *, fl_owner_t id);
2022 extern char * getname(const char __user *);
2023
2024 /* fs/ioctl.c */
2025
2026 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2027
2028 /* fs/dcache.c */
2029 extern void __init vfs_caches_init_early(void);
2030 extern void __init vfs_caches_init(unsigned long);
2031
2032 extern struct kmem_cache *names_cachep;
2033
2034 #define __getname_gfp(gfp) kmem_cache_alloc(names_cachep, (gfp))
2035 #define __getname() __getname_gfp(GFP_KERNEL)
2036 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2037 #ifndef CONFIG_AUDITSYSCALL
2038 #define putname(name) __putname(name)
2039 #else
2040 extern void putname(const char *name);
2041 #endif
2042
2043 #ifdef CONFIG_BLOCK
2044 extern int register_blkdev(unsigned int, const char *);
2045 extern void unregister_blkdev(unsigned int, const char *);
2046 extern struct block_device *bdget(dev_t);
2047 extern struct block_device *bdgrab(struct block_device *bdev);
2048 extern void bd_set_size(struct block_device *, loff_t size);
2049 extern void bd_forget(struct inode *inode);
2050 extern void bdput(struct block_device *);
2051 extern void invalidate_bdev(struct block_device *);
2052 extern int sync_blockdev(struct block_device *bdev);
2053 extern void kill_bdev(struct block_device *);
2054 extern struct super_block *freeze_bdev(struct block_device *);
2055 extern void emergency_thaw_all(void);
2056 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2057 extern int fsync_bdev(struct block_device *);
2058 #else
2059 static inline void bd_forget(struct inode *inode) {}
2060 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2061 static inline void kill_bdev(struct block_device *bdev) {}
2062 static inline void invalidate_bdev(struct block_device *bdev) {}
2063
2064 static inline struct super_block *freeze_bdev(struct block_device *sb)
2065 {
2066 return NULL;
2067 }
2068
2069 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2070 {
2071 return 0;
2072 }
2073 #endif
2074 extern int sync_filesystem(struct super_block *);
2075 extern const struct file_operations def_blk_fops;
2076 extern const struct file_operations def_chr_fops;
2077 extern const struct file_operations bad_sock_fops;
2078 extern const struct file_operations def_fifo_fops;
2079 #ifdef CONFIG_BLOCK
2080 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2081 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2082 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2083 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2084 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2085 void *holder);
2086 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2087 void *holder);
2088 extern int blkdev_put(struct block_device *bdev, fmode_t mode);
2089 #ifdef CONFIG_SYSFS
2090 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2091 extern void bd_unlink_disk_holder(struct block_device *bdev,
2092 struct gendisk *disk);
2093 #else
2094 static inline int bd_link_disk_holder(struct block_device *bdev,
2095 struct gendisk *disk)
2096 {
2097 return 0;
2098 }
2099 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2100 struct gendisk *disk)
2101 {
2102 }
2103 #endif
2104 #endif
2105
2106 /* fs/char_dev.c */
2107 #define CHRDEV_MAJOR_HASH_SIZE 255
2108 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2109 extern int register_chrdev_region(dev_t, unsigned, const char *);
2110 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2111 unsigned int count, const char *name,
2112 const struct file_operations *fops);
2113 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2114 unsigned int count, const char *name);
2115 extern void unregister_chrdev_region(dev_t, unsigned);
2116 extern void chrdev_show(struct seq_file *,off_t);
2117
2118 static inline int register_chrdev(unsigned int major, const char *name,
2119 const struct file_operations *fops)
2120 {
2121 return __register_chrdev(major, 0, 256, name, fops);
2122 }
2123
2124 static inline void unregister_chrdev(unsigned int major, const char *name)
2125 {
2126 __unregister_chrdev(major, 0, 256, name);
2127 }
2128
2129 /* fs/block_dev.c */
2130 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2131 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2132
2133 #ifdef CONFIG_BLOCK
2134 #define BLKDEV_MAJOR_HASH_SIZE 255
2135 extern const char *__bdevname(dev_t, char *buffer);
2136 extern const char *bdevname(struct block_device *bdev, char *buffer);
2137 extern struct block_device *lookup_bdev(const char *);
2138 extern void blkdev_show(struct seq_file *,off_t);
2139
2140 #else
2141 #define BLKDEV_MAJOR_HASH_SIZE 0
2142 #endif
2143
2144 extern void init_special_inode(struct inode *, umode_t, dev_t);
2145
2146 /* Invalid inode operations -- fs/bad_inode.c */
2147 extern void make_bad_inode(struct inode *);
2148 extern int is_bad_inode(struct inode *);
2149
2150 extern const struct file_operations read_pipefifo_fops;
2151 extern const struct file_operations write_pipefifo_fops;
2152 extern const struct file_operations rdwr_pipefifo_fops;
2153
2154 #ifdef CONFIG_BLOCK
2155 /*
2156 * return READ, READA, or WRITE
2157 */
2158 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2159
2160 /*
2161 * return data direction, READ or WRITE
2162 */
2163 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2164
2165 extern void check_disk_size_change(struct gendisk *disk,
2166 struct block_device *bdev);
2167 extern int revalidate_disk(struct gendisk *);
2168 extern int check_disk_change(struct block_device *);
2169 extern int __invalidate_device(struct block_device *, bool);
2170 extern int invalidate_partition(struct gendisk *, int);
2171 #endif
2172 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2173 pgoff_t start, pgoff_t end);
2174
2175 static inline void invalidate_remote_inode(struct inode *inode)
2176 {
2177 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2178 S_ISLNK(inode->i_mode))
2179 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2180 }
2181 extern int invalidate_inode_pages2(struct address_space *mapping);
2182 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2183 pgoff_t start, pgoff_t end);
2184 extern int write_inode_now(struct inode *, int);
2185 extern int filemap_fdatawrite(struct address_space *);
2186 extern int filemap_flush(struct address_space *);
2187 extern int filemap_fdatawait(struct address_space *);
2188 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2189 loff_t lend);
2190 extern int filemap_write_and_wait(struct address_space *mapping);
2191 extern int filemap_write_and_wait_range(struct address_space *mapping,
2192 loff_t lstart, loff_t lend);
2193 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2194 loff_t start, loff_t end, int sync_mode);
2195 extern int filemap_fdatawrite_range(struct address_space *mapping,
2196 loff_t start, loff_t end);
2197
2198 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2199 int datasync);
2200 extern int vfs_fsync(struct file *file, int datasync);
2201 extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
2202 extern void sync_supers(void);
2203 extern void emergency_sync(void);
2204 extern void emergency_remount(void);
2205 #ifdef CONFIG_BLOCK
2206 extern sector_t bmap(struct inode *, sector_t);
2207 #endif
2208 extern int notify_change(struct dentry *, struct iattr *);
2209 extern int inode_permission(struct inode *, int);
2210 extern int generic_permission(struct inode *, int);
2211
2212 static inline bool execute_ok(struct inode *inode)
2213 {
2214 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2215 }
2216
2217 /*
2218 * get_write_access() gets write permission for a file.
2219 * put_write_access() releases this write permission.
2220 * This is used for regular files.
2221 * We cannot support write (and maybe mmap read-write shared) accesses and
2222 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2223 * can have the following values:
2224 * 0: no writers, no VM_DENYWRITE mappings
2225 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2226 * > 0: (i_writecount) users are writing to the file.
2227 *
2228 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2229 * except for the cases where we don't hold i_writecount yet. Then we need to
2230 * use {get,deny}_write_access() - these functions check the sign and refuse
2231 * to do the change if sign is wrong.
2232 */
2233 static inline int get_write_access(struct inode *inode)
2234 {
2235 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2236 }
2237 static inline int deny_write_access(struct file *file)
2238 {
2239 struct inode *inode = file->f_path.dentry->d_inode;
2240 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2241 }
2242 static inline void put_write_access(struct inode * inode)
2243 {
2244 atomic_dec(&inode->i_writecount);
2245 }
2246 static inline void allow_write_access(struct file *file)
2247 {
2248 if (file)
2249 atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
2250 }
2251 #ifdef CONFIG_IMA
2252 static inline void i_readcount_dec(struct inode *inode)
2253 {
2254 BUG_ON(!atomic_read(&inode->i_readcount));
2255 atomic_dec(&inode->i_readcount);
2256 }
2257 static inline void i_readcount_inc(struct inode *inode)
2258 {
2259 atomic_inc(&inode->i_readcount);
2260 }
2261 #else
2262 static inline void i_readcount_dec(struct inode *inode)
2263 {
2264 return;
2265 }
2266 static inline void i_readcount_inc(struct inode *inode)
2267 {
2268 return;
2269 }
2270 #endif
2271 extern int do_pipe_flags(int *, int);
2272 extern struct file *create_read_pipe(struct file *f, int flags);
2273 extern struct file *create_write_pipe(int flags);
2274 extern void free_write_pipe(struct file *);
2275
2276 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2277 extern struct file * open_exec(const char *);
2278
2279 /* fs/dcache.c -- generic fs support functions */
2280 extern int is_subdir(struct dentry *, struct dentry *);
2281 extern int path_is_under(struct path *, struct path *);
2282 extern ino_t find_inode_number(struct dentry *, struct qstr *);
2283
2284 #include <linux/err.h>
2285
2286 /* needed for stackable file system support */
2287 extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
2288
2289 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
2290
2291 extern int inode_init_always(struct super_block *, struct inode *);
2292 extern void inode_init_once(struct inode *);
2293 extern void address_space_init_once(struct address_space *mapping);
2294 extern void ihold(struct inode * inode);
2295 extern void iput(struct inode *);
2296 extern struct inode * igrab(struct inode *);
2297 extern ino_t iunique(struct super_block *, ino_t);
2298 extern int inode_needs_sync(struct inode *inode);
2299 extern int generic_delete_inode(struct inode *inode);
2300 static inline int generic_drop_inode(struct inode *inode)
2301 {
2302 return !inode->i_nlink || inode_unhashed(inode);
2303 }
2304
2305 extern struct inode *ilookup5_nowait(struct super_block *sb,
2306 unsigned long hashval, int (*test)(struct inode *, void *),
2307 void *data);
2308 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2309 int (*test)(struct inode *, void *), void *data);
2310 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2311
2312 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2313 extern struct inode * iget_locked(struct super_block *, unsigned long);
2314 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2315 extern int insert_inode_locked(struct inode *);
2316 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2317 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2318 #else
2319 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2320 #endif
2321 extern void unlock_new_inode(struct inode *);
2322 extern unsigned int get_next_ino(void);
2323
2324 extern void __iget(struct inode * inode);
2325 extern void iget_failed(struct inode *);
2326 extern void end_writeback(struct inode *);
2327 extern void __destroy_inode(struct inode *);
2328 extern struct inode *new_inode_pseudo(struct super_block *sb);
2329 extern struct inode *new_inode(struct super_block *sb);
2330 extern void free_inode_nonrcu(struct inode *inode);
2331 extern int should_remove_suid(struct dentry *);
2332 extern int file_remove_suid(struct file *);
2333
2334 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2335 static inline void insert_inode_hash(struct inode *inode)
2336 {
2337 __insert_inode_hash(inode, inode->i_ino);
2338 }
2339
2340 extern void __remove_inode_hash(struct inode *);
2341 static inline void remove_inode_hash(struct inode *inode)
2342 {
2343 if (!inode_unhashed(inode))
2344 __remove_inode_hash(inode);
2345 }
2346
2347 extern void inode_sb_list_add(struct inode *inode);
2348
2349 #ifdef CONFIG_BLOCK
2350 extern void submit_bio(int, struct bio *);
2351 extern int bdev_read_only(struct block_device *);
2352 #endif
2353 extern int set_blocksize(struct block_device *, int);
2354 extern int sb_set_blocksize(struct super_block *, int);
2355 extern int sb_min_blocksize(struct super_block *, int);
2356
2357 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2358 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2359 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
2360 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2361 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2362 extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2363 loff_t *);
2364 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2365 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2366 unsigned long *, loff_t, loff_t *, size_t, size_t);
2367 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2368 unsigned long, loff_t, loff_t *, size_t, ssize_t);
2369 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2370 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2371 extern int generic_segment_checks(const struct iovec *iov,
2372 unsigned long *nr_segs, size_t *count, int access_flags);
2373
2374 /* fs/block_dev.c */
2375 extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2376 unsigned long nr_segs, loff_t pos);
2377 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2378 int datasync);
2379 extern void block_sync_page(struct page *page);
2380
2381 /* fs/splice.c */
2382 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2383 struct pipe_inode_info *, size_t, unsigned int);
2384 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2385 struct pipe_inode_info *, size_t, unsigned int);
2386 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
2387 struct file *, loff_t *, size_t, unsigned int);
2388 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2389 struct file *out, loff_t *, size_t len, unsigned int flags);
2390 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2391 size_t len, unsigned int flags);
2392
2393 extern void
2394 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2395 extern loff_t noop_llseek(struct file *file, loff_t offset, int origin);
2396 extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
2397 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
2398 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2399 int origin, loff_t maxsize);
2400 extern int generic_file_open(struct inode * inode, struct file * filp);
2401 extern int nonseekable_open(struct inode * inode, struct file * filp);
2402
2403 #ifdef CONFIG_FS_XIP
2404 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2405 loff_t *ppos);
2406 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2407 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2408 size_t len, loff_t *ppos);
2409 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2410 #else
2411 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2412 {
2413 return 0;
2414 }
2415 #endif
2416
2417 #ifdef CONFIG_BLOCK
2418 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2419 loff_t file_offset);
2420
2421 enum {
2422 /* need locking between buffered and direct access */
2423 DIO_LOCKING = 0x01,
2424
2425 /* filesystem does not support filling holes */
2426 DIO_SKIP_HOLES = 0x02,
2427 };
2428
2429 void dio_end_io(struct bio *bio, int error);
2430 void inode_dio_wait(struct inode *inode);
2431 void inode_dio_done(struct inode *inode);
2432
2433 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2434 struct block_device *bdev, const struct iovec *iov, loff_t offset,
2435 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2436 dio_submit_t submit_io, int flags);
2437
2438 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2439 struct inode *inode, const struct iovec *iov, loff_t offset,
2440 unsigned long nr_segs, get_block_t get_block)
2441 {
2442 return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
2443 offset, nr_segs, get_block, NULL, NULL,
2444 DIO_LOCKING | DIO_SKIP_HOLES);
2445 }
2446 #else
2447 static inline void inode_dio_wait(struct inode *inode)
2448 {
2449 }
2450 #endif
2451
2452 extern const struct file_operations generic_ro_fops;
2453
2454 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2455
2456 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2457 extern int vfs_follow_link(struct nameidata *, const char *);
2458 extern int page_readlink(struct dentry *, char __user *, int);
2459 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2460 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2461 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2462 int nofs);
2463 extern int page_symlink(struct inode *inode, const char *symname, int len);
2464 extern const struct inode_operations page_symlink_inode_operations;
2465 extern int generic_readlink(struct dentry *, char __user *, int);
2466 extern void generic_fillattr(struct inode *, struct kstat *);
2467 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2468 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2469 void inode_add_bytes(struct inode *inode, loff_t bytes);
2470 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2471 loff_t inode_get_bytes(struct inode *inode);
2472 void inode_set_bytes(struct inode *inode, loff_t bytes);
2473
2474 extern int vfs_readdir(struct file *, filldir_t, void *);
2475
2476 extern int vfs_stat(const char __user *, struct kstat *);
2477 extern int vfs_lstat(const char __user *, struct kstat *);
2478 extern int vfs_fstat(unsigned int, struct kstat *);
2479 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2480
2481 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2482 unsigned long arg);
2483 extern int __generic_block_fiemap(struct inode *inode,
2484 struct fiemap_extent_info *fieinfo,
2485 loff_t start, loff_t len,
2486 get_block_t *get_block);
2487 extern int generic_block_fiemap(struct inode *inode,
2488 struct fiemap_extent_info *fieinfo, u64 start,
2489 u64 len, get_block_t *get_block);
2490
2491 extern void get_filesystem(struct file_system_type *fs);
2492 extern void put_filesystem(struct file_system_type *fs);
2493 extern struct file_system_type *get_fs_type(const char *name);
2494 extern struct super_block *get_super(struct block_device *);
2495 extern struct super_block *get_super_thawed(struct block_device *);
2496 extern struct super_block *get_active_super(struct block_device *bdev);
2497 extern void drop_super(struct super_block *sb);
2498 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2499 extern void iterate_supers_type(struct file_system_type *,
2500 void (*)(struct super_block *, void *), void *);
2501
2502 extern int dcache_dir_open(struct inode *, struct file *);
2503 extern int dcache_dir_close(struct inode *, struct file *);
2504 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2505 extern int dcache_readdir(struct file *, void *, filldir_t);
2506 extern int simple_setattr(struct dentry *, struct iattr *);
2507 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2508 extern int simple_statfs(struct dentry *, struct kstatfs *);
2509 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2510 extern int simple_unlink(struct inode *, struct dentry *);
2511 extern int simple_rmdir(struct inode *, struct dentry *);
2512 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2513 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2514 extern int simple_empty(struct dentry *);
2515 extern int simple_readpage(struct file *file, struct page *page);
2516 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2517 loff_t pos, unsigned len, unsigned flags,
2518 struct page **pagep, void **fsdata);
2519 extern int simple_write_end(struct file *file, struct address_space *mapping,
2520 loff_t pos, unsigned len, unsigned copied,
2521 struct page *page, void *fsdata);
2522
2523 extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
2524 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2525 extern const struct file_operations simple_dir_operations;
2526 extern const struct inode_operations simple_dir_inode_operations;
2527 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2528 struct dentry *d_alloc_name(struct dentry *, const char *);
2529 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2530 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2531 extern void simple_release_fs(struct vfsmount **mount, int *count);
2532
2533 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2534 loff_t *ppos, const void *from, size_t available);
2535 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2536 const void __user *from, size_t count);
2537
2538 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2539
2540 extern int generic_check_addressable(unsigned, u64);
2541
2542 #ifdef CONFIG_MIGRATION
2543 extern int buffer_migrate_page(struct address_space *,
2544 struct page *, struct page *,
2545 enum migrate_mode);
2546 #else
2547 #define buffer_migrate_page NULL
2548 #endif
2549
2550 extern int inode_change_ok(const struct inode *, struct iattr *);
2551 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2552 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2553
2554 extern void file_update_time(struct file *file);
2555
2556 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2557 extern void save_mount_options(struct super_block *sb, char *options);
2558 extern void replace_mount_options(struct super_block *sb, char *options);
2559
2560 static inline ino_t parent_ino(struct dentry *dentry)
2561 {
2562 ino_t res;
2563
2564 /*
2565 * Don't strictly need d_lock here? If the parent ino could change
2566 * then surely we'd have a deeper race in the caller?
2567 */
2568 spin_lock(&dentry->d_lock);
2569 res = dentry->d_parent->d_inode->i_ino;
2570 spin_unlock(&dentry->d_lock);
2571 return res;
2572 }
2573
2574 /* Transaction based IO helpers */
2575
2576 /*
2577 * An argresp is stored in an allocated page and holds the
2578 * size of the argument or response, along with its content
2579 */
2580 struct simple_transaction_argresp {
2581 ssize_t size;
2582 char data[0];
2583 };
2584
2585 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2586
2587 char *simple_transaction_get(struct file *file, const char __user *buf,
2588 size_t size);
2589 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2590 size_t size, loff_t *pos);
2591 int simple_transaction_release(struct inode *inode, struct file *file);
2592
2593 void simple_transaction_set(struct file *file, size_t n);
2594
2595 /*
2596 * simple attribute files
2597 *
2598 * These attributes behave similar to those in sysfs:
2599 *
2600 * Writing to an attribute immediately sets a value, an open file can be
2601 * written to multiple times.
2602 *
2603 * Reading from an attribute creates a buffer from the value that might get
2604 * read with multiple read calls. When the attribute has been read
2605 * completely, no further read calls are possible until the file is opened
2606 * again.
2607 *
2608 * All attributes contain a text representation of a numeric value
2609 * that are accessed with the get() and set() functions.
2610 */
2611 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2612 static int __fops ## _open(struct inode *inode, struct file *file) \
2613 { \
2614 __simple_attr_check_format(__fmt, 0ull); \
2615 return simple_attr_open(inode, file, __get, __set, __fmt); \
2616 } \
2617 static const struct file_operations __fops = { \
2618 .owner = THIS_MODULE, \
2619 .open = __fops ## _open, \
2620 .release = simple_attr_release, \
2621 .read = simple_attr_read, \
2622 .write = simple_attr_write, \
2623 .llseek = generic_file_llseek, \
2624 };
2625
2626 static inline __printf(1, 2)
2627 void __simple_attr_check_format(const char *fmt, ...)
2628 {
2629 /* don't do anything, just let the compiler check the arguments; */
2630 }
2631
2632 int simple_attr_open(struct inode *inode, struct file *file,
2633 int (*get)(void *, u64 *), int (*set)(void *, u64),
2634 const char *fmt);
2635 int simple_attr_release(struct inode *inode, struct file *file);
2636 ssize_t simple_attr_read(struct file *file, char __user *buf,
2637 size_t len, loff_t *ppos);
2638 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2639 size_t len, loff_t *ppos);
2640
2641 struct ctl_table;
2642 int proc_nr_files(struct ctl_table *table, int write,
2643 void __user *buffer, size_t *lenp, loff_t *ppos);
2644 int proc_nr_dentry(struct ctl_table *table, int write,
2645 void __user *buffer, size_t *lenp, loff_t *ppos);
2646 int proc_nr_inodes(struct ctl_table *table, int write,
2647 void __user *buffer, size_t *lenp, loff_t *ppos);
2648 int __init get_filesystem_list(char *buf);
2649
2650 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2651 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2652
2653 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2654 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2655 (flag & __FMODE_NONOTIFY)))
2656
2657 static inline int is_sxid(umode_t mode)
2658 {
2659 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2660 }
2661
2662 static inline void inode_has_no_xattr(struct inode *inode)
2663 {
2664 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2665 inode->i_flags |= S_NOSEC;
2666 }
2667
2668 #endif /* __KERNEL__ */
2669 #endif /* _LINUX_FS_H */