Merge branch 'pstore' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/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
96 /*
97 * Don't update ctime and mtime.
98 *
99 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
100 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
101 */
102 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
103
104 /* Expect random access pattern */
105 #define FMODE_RANDOM ((__force fmode_t)0x1000)
106
107 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
108 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
109
110 /* File is opened with O_PATH; almost nothing can be done with it */
111 #define FMODE_PATH ((__force fmode_t)0x4000)
112
113 /* File was opened by fanotify and shouldn't generate fanotify events */
114 #define FMODE_NONOTIFY ((__force fmode_t)0x1000000)
115
116 /*
117 * The below are the various read and write types that we support. Some of
118 * them include behavioral modifiers that send information down to the
119 * block layer and IO scheduler. Terminology:
120 *
121 * The block layer uses device plugging to defer IO a little bit, in
122 * the hope that we will see more IO very shortly. This increases
123 * coalescing of adjacent IO and thus reduces the number of IOs we
124 * have to send to the device. It also allows for better queuing,
125 * if the IO isn't mergeable. If the caller is going to be waiting
126 * for the IO, then he must ensure that the device is unplugged so
127 * that the IO is dispatched to the driver.
128 *
129 * All IO is handled async in Linux. This is fine for background
130 * writes, but for reads or writes that someone waits for completion
131 * on, we want to notify the block layer and IO scheduler so that they
132 * know about it. That allows them to make better scheduling
133 * decisions. So when the below references 'sync' and 'async', it
134 * is referencing this priority hint.
135 *
136 * With that in mind, the available types are:
137 *
138 * READ A normal read operation. Device will be plugged.
139 * READ_SYNC A synchronous read. Device is not plugged, caller can
140 * immediately wait on this read without caring about
141 * unplugging.
142 * READA Used for read-ahead operations. Lower priority, and the
143 * block layer could (in theory) choose to ignore this
144 * request if it runs into resource problems.
145 * WRITE A normal async write. Device will be plugged.
146 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
147 * the hint that someone will be waiting on this IO
148 * shortly. The write equivalent of READ_SYNC.
149 * WRITE_ODIRECT Special case write for O_DIRECT only.
150 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
151 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
152 * non-volatile media on completion.
153 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
154 * by a cache flush and data is guaranteed to be on
155 * non-volatile media on completion.
156 *
157 */
158 #define RW_MASK REQ_WRITE
159 #define RWA_MASK REQ_RAHEAD
160
161 #define READ 0
162 #define WRITE RW_MASK
163 #define READA RWA_MASK
164
165 #define READ_SYNC (READ | REQ_SYNC)
166 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
167 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
168 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
169 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
170 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
171
172 #define SEL_IN 1
173 #define SEL_OUT 2
174 #define SEL_EX 4
175
176 /* public flags for file_system_type */
177 #define FS_REQUIRES_DEV 1
178 #define FS_BINARY_MOUNTDATA 2
179 #define FS_HAS_SUBTYPE 4
180 #define FS_REVAL_DOT 16384 /* Check the paths ".", ".." for staleness */
181 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move()
182 * during rename() internally.
183 */
184
185 /*
186 * These are the fs-independent mount-flags: up to 32 flags are supported
187 */
188 #define MS_RDONLY 1 /* Mount read-only */
189 #define MS_NOSUID 2 /* Ignore suid and sgid bits */
190 #define MS_NODEV 4 /* Disallow access to device special files */
191 #define MS_NOEXEC 8 /* Disallow program execution */
192 #define MS_SYNCHRONOUS 16 /* Writes are synced at once */
193 #define MS_REMOUNT 32 /* Alter flags of a mounted FS */
194 #define MS_MANDLOCK 64 /* Allow mandatory locks on an FS */
195 #define MS_DIRSYNC 128 /* Directory modifications are synchronous */
196 #define MS_NOATIME 1024 /* Do not update access times. */
197 #define MS_NODIRATIME 2048 /* Do not update directory access times */
198 #define MS_BIND 4096
199 #define MS_MOVE 8192
200 #define MS_REC 16384
201 #define MS_VERBOSE 32768 /* War is peace. Verbosity is silence.
202 MS_VERBOSE is deprecated. */
203 #define MS_SILENT 32768
204 #define MS_POSIXACL (1<<16) /* VFS does not apply the umask */
205 #define MS_UNBINDABLE (1<<17) /* change to unbindable */
206 #define MS_PRIVATE (1<<18) /* change to private */
207 #define MS_SLAVE (1<<19) /* change to slave */
208 #define MS_SHARED (1<<20) /* change to shared */
209 #define MS_RELATIME (1<<21) /* Update atime relative to mtime/ctime. */
210 #define MS_KERNMOUNT (1<<22) /* this is a kern_mount call */
211 #define MS_I_VERSION (1<<23) /* Update inode I_version field */
212 #define MS_STRICTATIME (1<<24) /* Always perform atime updates */
213 #define MS_NOSEC (1<<28)
214 #define MS_BORN (1<<29)
215 #define MS_ACTIVE (1<<30)
216 #define MS_NOUSER (1<<31)
217
218 /*
219 * Superblock flags that can be altered by MS_REMOUNT
220 */
221 #define MS_RMT_MASK (MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_I_VERSION)
222
223 /*
224 * Old magic mount flag and mask
225 */
226 #define MS_MGC_VAL 0xC0ED0000
227 #define MS_MGC_MSK 0xffff0000
228
229 /* Inode flags - they have nothing to superblock flags now */
230
231 #define S_SYNC 1 /* Writes are synced at once */
232 #define S_NOATIME 2 /* Do not update access times */
233 #define S_APPEND 4 /* Append-only file */
234 #define S_IMMUTABLE 8 /* Immutable file */
235 #define S_DEAD 16 /* removed, but still open directory */
236 #define S_NOQUOTA 32 /* Inode is not counted to quota */
237 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
238 #define S_NOCMTIME 128 /* Do not update file c/mtime */
239 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
240 #define S_PRIVATE 512 /* Inode is fs-internal */
241 #define S_IMA 1024 /* Inode has an associated IMA struct */
242 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
243 #define S_NOSEC 4096 /* no suid or xattr security attributes */
244
245 /*
246 * Note that nosuid etc flags are inode-specific: setting some file-system
247 * flags just means all the inodes inherit those flags by default. It might be
248 * possible to override it selectively if you really wanted to with some
249 * ioctl() that is not currently implemented.
250 *
251 * Exception: MS_RDONLY is always applied to the entire file system.
252 *
253 * Unfortunately, it is possible to change a filesystems flags with it mounted
254 * with files in use. This means that all of the inodes will not have their
255 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
256 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
257 */
258 #define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
259
260 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
261 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
262 ((inode)->i_flags & S_SYNC))
263 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
264 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
265 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
266 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
267 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
268
269 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
270 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
271 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
272 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
273
274 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
275 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
276 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
277 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
278 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
279 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
280 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
281
282 /* the read-only stuff doesn't really belong here, but any other place is
283 probably as bad and I don't want to create yet another include file. */
284
285 #define BLKROSET _IO(0x12,93) /* set device read-only (0 = read-write) */
286 #define BLKROGET _IO(0x12,94) /* get read-only status (0 = read_write) */
287 #define BLKRRPART _IO(0x12,95) /* re-read partition table */
288 #define BLKGETSIZE _IO(0x12,96) /* return device size /512 (long *arg) */
289 #define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */
290 #define BLKRASET _IO(0x12,98) /* set read ahead for block device */
291 #define BLKRAGET _IO(0x12,99) /* get current read ahead setting */
292 #define BLKFRASET _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
293 #define BLKFRAGET _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
294 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
295 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
296 #define BLKSSZGET _IO(0x12,104)/* get block device sector size */
297 #if 0
298 #define BLKPG _IO(0x12,105)/* See blkpg.h */
299
300 /* Some people are morons. Do not use sizeof! */
301
302 #define BLKELVGET _IOR(0x12,106,size_t)/* elevator get */
303 #define BLKELVSET _IOW(0x12,107,size_t)/* elevator set */
304 /* This was here just to show that the number is taken -
305 probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
306 #endif
307 /* A jump here: 108-111 have been used for various private purposes. */
308 #define BLKBSZGET _IOR(0x12,112,size_t)
309 #define BLKBSZSET _IOW(0x12,113,size_t)
310 #define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
311 #define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
312 #define BLKTRACESTART _IO(0x12,116)
313 #define BLKTRACESTOP _IO(0x12,117)
314 #define BLKTRACETEARDOWN _IO(0x12,118)
315 #define BLKDISCARD _IO(0x12,119)
316 #define BLKIOMIN _IO(0x12,120)
317 #define BLKIOOPT _IO(0x12,121)
318 #define BLKALIGNOFF _IO(0x12,122)
319 #define BLKPBSZGET _IO(0x12,123)
320 #define BLKDISCARDZEROES _IO(0x12,124)
321 #define BLKSECDISCARD _IO(0x12,125)
322
323 #define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
324 #define FIBMAP _IO(0x00,1) /* bmap access */
325 #define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */
326 #define FIFREEZE _IOWR('X', 119, int) /* Freeze */
327 #define FITHAW _IOWR('X', 120, int) /* Thaw */
328 #define FITRIM _IOWR('X', 121, struct fstrim_range) /* Trim */
329
330 #define FS_IOC_GETFLAGS _IOR('f', 1, long)
331 #define FS_IOC_SETFLAGS _IOW('f', 2, long)
332 #define FS_IOC_GETVERSION _IOR('v', 1, long)
333 #define FS_IOC_SETVERSION _IOW('v', 2, long)
334 #define FS_IOC_FIEMAP _IOWR('f', 11, struct fiemap)
335 #define FS_IOC32_GETFLAGS _IOR('f', 1, int)
336 #define FS_IOC32_SETFLAGS _IOW('f', 2, int)
337 #define FS_IOC32_GETVERSION _IOR('v', 1, int)
338 #define FS_IOC32_SETVERSION _IOW('v', 2, int)
339
340 /*
341 * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS)
342 */
343 #define FS_SECRM_FL 0x00000001 /* Secure deletion */
344 #define FS_UNRM_FL 0x00000002 /* Undelete */
345 #define FS_COMPR_FL 0x00000004 /* Compress file */
346 #define FS_SYNC_FL 0x00000008 /* Synchronous updates */
347 #define FS_IMMUTABLE_FL 0x00000010 /* Immutable file */
348 #define FS_APPEND_FL 0x00000020 /* writes to file may only append */
349 #define FS_NODUMP_FL 0x00000040 /* do not dump file */
350 #define FS_NOATIME_FL 0x00000080 /* do not update atime */
351 /* Reserved for compression usage... */
352 #define FS_DIRTY_FL 0x00000100
353 #define FS_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
354 #define FS_NOCOMP_FL 0x00000400 /* Don't compress */
355 #define FS_ECOMPR_FL 0x00000800 /* Compression error */
356 /* End compression flags --- maybe not all used */
357 #define FS_BTREE_FL 0x00001000 /* btree format dir */
358 #define FS_INDEX_FL 0x00001000 /* hash-indexed directory */
359 #define FS_IMAGIC_FL 0x00002000 /* AFS directory */
360 #define FS_JOURNAL_DATA_FL 0x00004000 /* Reserved for ext3 */
361 #define FS_NOTAIL_FL 0x00008000 /* file tail should not be merged */
362 #define FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
363 #define FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
364 #define FS_EXTENT_FL 0x00080000 /* Extents */
365 #define FS_DIRECTIO_FL 0x00100000 /* Use direct i/o */
366 #define FS_NOCOW_FL 0x00800000 /* Do not cow file */
367 #define FS_RESERVED_FL 0x80000000 /* reserved for ext2 lib */
368
369 #define FS_FL_USER_VISIBLE 0x0003DFFF /* User visible flags */
370 #define FS_FL_USER_MODIFIABLE 0x000380FF /* User modifiable flags */
371
372
373 #define SYNC_FILE_RANGE_WAIT_BEFORE 1
374 #define SYNC_FILE_RANGE_WRITE 2
375 #define SYNC_FILE_RANGE_WAIT_AFTER 4
376
377 #ifdef __KERNEL__
378
379 #include <linux/linkage.h>
380 #include <linux/wait.h>
381 #include <linux/kdev_t.h>
382 #include <linux/dcache.h>
383 #include <linux/path.h>
384 #include <linux/stat.h>
385 #include <linux/cache.h>
386 #include <linux/list.h>
387 #include <linux/radix-tree.h>
388 #include <linux/prio_tree.h>
389 #include <linux/init.h>
390 #include <linux/pid.h>
391 #include <linux/mutex.h>
392 #include <linux/capability.h>
393 #include <linux/semaphore.h>
394 #include <linux/fiemap.h>
395 #include <linux/rculist_bl.h>
396 #include <linux/shrinker.h>
397 #include <linux/atomic.h>
398
399 #include <asm/byteorder.h>
400
401 struct export_operations;
402 struct hd_geometry;
403 struct iovec;
404 struct nameidata;
405 struct kiocb;
406 struct kobject;
407 struct pipe_inode_info;
408 struct poll_table_struct;
409 struct kstatfs;
410 struct vm_area_struct;
411 struct vfsmount;
412 struct cred;
413
414 extern void __init inode_init(void);
415 extern void __init inode_init_early(void);
416 extern void __init files_init(unsigned long);
417
418 extern struct files_stat_struct files_stat;
419 extern unsigned long get_max_files(void);
420 extern int sysctl_nr_open;
421 extern struct inodes_stat_t inodes_stat;
422 extern int leases_enable, lease_break_time;
423
424 struct buffer_head;
425 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
426 struct buffer_head *bh_result, int create);
427 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
428 ssize_t bytes, void *private, int ret,
429 bool is_async);
430
431 /*
432 * Attribute flags. These should be or-ed together to figure out what
433 * has been changed!
434 */
435 #define ATTR_MODE (1 << 0)
436 #define ATTR_UID (1 << 1)
437 #define ATTR_GID (1 << 2)
438 #define ATTR_SIZE (1 << 3)
439 #define ATTR_ATIME (1 << 4)
440 #define ATTR_MTIME (1 << 5)
441 #define ATTR_CTIME (1 << 6)
442 #define ATTR_ATIME_SET (1 << 7)
443 #define ATTR_MTIME_SET (1 << 8)
444 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
445 #define ATTR_ATTR_FLAG (1 << 10)
446 #define ATTR_KILL_SUID (1 << 11)
447 #define ATTR_KILL_SGID (1 << 12)
448 #define ATTR_FILE (1 << 13)
449 #define ATTR_KILL_PRIV (1 << 14)
450 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
451 #define ATTR_TIMES_SET (1 << 16)
452
453 /*
454 * This is the Inode Attributes structure, used for notify_change(). It
455 * uses the above definitions as flags, to know which values have changed.
456 * Also, in this manner, a Filesystem can look at only the values it cares
457 * about. Basically, these are the attributes that the VFS layer can
458 * request to change from the FS layer.
459 *
460 * Derek Atkins <warlord@MIT.EDU> 94-10-20
461 */
462 struct iattr {
463 unsigned int ia_valid;
464 umode_t ia_mode;
465 uid_t ia_uid;
466 gid_t ia_gid;
467 loff_t ia_size;
468 struct timespec ia_atime;
469 struct timespec ia_mtime;
470 struct timespec ia_ctime;
471
472 /*
473 * Not an attribute, but an auxiliary info for filesystems wanting to
474 * implement an ftruncate() like method. NOTE: filesystem should
475 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
476 */
477 struct file *ia_file;
478 };
479
480 /*
481 * Includes for diskquotas.
482 */
483 #include <linux/quota.h>
484
485 /**
486 * enum positive_aop_returns - aop return codes with specific semantics
487 *
488 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
489 * completed, that the page is still locked, and
490 * should be considered active. The VM uses this hint
491 * to return the page to the active list -- it won't
492 * be a candidate for writeback again in the near
493 * future. Other callers must be careful to unlock
494 * the page if they get this return. Returned by
495 * writepage();
496 *
497 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
498 * unlocked it and the page might have been truncated.
499 * The caller should back up to acquiring a new page and
500 * trying again. The aop will be taking reasonable
501 * precautions not to livelock. If the caller held a page
502 * reference, it should drop it before retrying. Returned
503 * by readpage().
504 *
505 * address_space_operation functions return these large constants to indicate
506 * special semantics to the caller. These are much larger than the bytes in a
507 * page to allow for functions that return the number of bytes operated on in a
508 * given page.
509 */
510
511 enum positive_aop_returns {
512 AOP_WRITEPAGE_ACTIVATE = 0x80000,
513 AOP_TRUNCATED_PAGE = 0x80001,
514 };
515
516 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
517 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
518 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
519 * helper code (eg buffer layer)
520 * to clear GFP_FS from alloc */
521
522 /*
523 * oh the beauties of C type declarations.
524 */
525 struct page;
526 struct address_space;
527 struct writeback_control;
528
529 struct iov_iter {
530 const struct iovec *iov;
531 unsigned long nr_segs;
532 size_t iov_offset;
533 size_t count;
534 };
535
536 size_t iov_iter_copy_from_user_atomic(struct page *page,
537 struct iov_iter *i, unsigned long offset, size_t bytes);
538 size_t iov_iter_copy_from_user(struct page *page,
539 struct iov_iter *i, unsigned long offset, size_t bytes);
540 void iov_iter_advance(struct iov_iter *i, size_t bytes);
541 int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
542 size_t iov_iter_single_seg_count(struct iov_iter *i);
543
544 static inline void iov_iter_init(struct iov_iter *i,
545 const struct iovec *iov, unsigned long nr_segs,
546 size_t count, size_t written)
547 {
548 i->iov = iov;
549 i->nr_segs = nr_segs;
550 i->iov_offset = 0;
551 i->count = count + written;
552
553 iov_iter_advance(i, written);
554 }
555
556 static inline size_t iov_iter_count(struct iov_iter *i)
557 {
558 return i->count;
559 }
560
561 /*
562 * "descriptor" for what we're up to with a read.
563 * This allows us to use the same read code yet
564 * have multiple different users of the data that
565 * we read from a file.
566 *
567 * The simplest case just copies the data to user
568 * mode.
569 */
570 typedef struct {
571 size_t written;
572 size_t count;
573 union {
574 char __user *buf;
575 void *data;
576 } arg;
577 int error;
578 } read_descriptor_t;
579
580 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
581 unsigned long, unsigned long);
582
583 struct address_space_operations {
584 int (*writepage)(struct page *page, struct writeback_control *wbc);
585 int (*readpage)(struct file *, struct page *);
586
587 /* Write back some dirty pages from this mapping. */
588 int (*writepages)(struct address_space *, struct writeback_control *);
589
590 /* Set a page dirty. Return true if this dirtied it */
591 int (*set_page_dirty)(struct page *page);
592
593 int (*readpages)(struct file *filp, struct address_space *mapping,
594 struct list_head *pages, unsigned nr_pages);
595
596 int (*write_begin)(struct file *, struct address_space *mapping,
597 loff_t pos, unsigned len, unsigned flags,
598 struct page **pagep, void **fsdata);
599 int (*write_end)(struct file *, struct address_space *mapping,
600 loff_t pos, unsigned len, unsigned copied,
601 struct page *page, void *fsdata);
602
603 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
604 sector_t (*bmap)(struct address_space *, sector_t);
605 void (*invalidatepage) (struct page *, unsigned long);
606 int (*releasepage) (struct page *, gfp_t);
607 void (*freepage)(struct page *);
608 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
609 loff_t offset, unsigned long nr_segs);
610 int (*get_xip_mem)(struct address_space *, pgoff_t, int,
611 void **, unsigned long *);
612 /* migrate the contents of a page to the specified target */
613 int (*migratepage) (struct address_space *,
614 struct page *, struct page *);
615 int (*launder_page) (struct page *);
616 int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
617 unsigned long);
618 int (*error_remove_page)(struct address_space *, struct page *);
619 };
620
621 extern const struct address_space_operations empty_aops;
622
623 /*
624 * pagecache_write_begin/pagecache_write_end must be used by general code
625 * to write into the pagecache.
626 */
627 int pagecache_write_begin(struct file *, struct address_space *mapping,
628 loff_t pos, unsigned len, unsigned flags,
629 struct page **pagep, void **fsdata);
630
631 int pagecache_write_end(struct file *, struct address_space *mapping,
632 loff_t pos, unsigned len, unsigned copied,
633 struct page *page, void *fsdata);
634
635 struct backing_dev_info;
636 struct address_space {
637 struct inode *host; /* owner: inode, block_device */
638 struct radix_tree_root page_tree; /* radix tree of all pages */
639 spinlock_t tree_lock; /* and lock protecting it */
640 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
641 struct prio_tree_root i_mmap; /* tree of private and shared mappings */
642 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
643 struct mutex i_mmap_mutex; /* protect tree, count, list */
644 /* Protected by tree_lock together with the radix tree */
645 unsigned long nrpages; /* number of total pages */
646 pgoff_t writeback_index;/* writeback starts here */
647 const struct address_space_operations *a_ops; /* methods */
648 unsigned long flags; /* error bits/gfp mask */
649 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
650 spinlock_t private_lock; /* for use by the address_space */
651 struct list_head private_list; /* ditto */
652 struct address_space *assoc_mapping; /* ditto */
653 } __attribute__((aligned(sizeof(long))));
654 /*
655 * On most architectures that alignment is already the case; but
656 * must be enforced here for CRIS, to let the least significant bit
657 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
658 */
659
660 struct block_device {
661 dev_t bd_dev; /* not a kdev_t - it's a search key */
662 int bd_openers;
663 struct inode * bd_inode; /* will die */
664 struct super_block * bd_super;
665 struct mutex bd_mutex; /* open/close mutex */
666 struct list_head bd_inodes;
667 void * bd_claiming;
668 void * bd_holder;
669 int bd_holders;
670 bool bd_write_holder;
671 #ifdef CONFIG_SYSFS
672 struct list_head bd_holder_disks;
673 #endif
674 struct block_device * bd_contains;
675 unsigned bd_block_size;
676 struct hd_struct * bd_part;
677 /* number of times partitions within this device have been opened. */
678 unsigned bd_part_count;
679 int bd_invalidated;
680 struct gendisk * bd_disk;
681 struct list_head bd_list;
682 /*
683 * Private data. You must have bd_claim'ed the block_device
684 * to use this. NOTE: bd_claim allows an owner to claim
685 * the same device multiple times, the owner must take special
686 * care to not mess up bd_private for that case.
687 */
688 unsigned long bd_private;
689
690 /* The counter of freeze processes */
691 int bd_fsfreeze_count;
692 /* Mutex for freeze */
693 struct mutex bd_fsfreeze_mutex;
694 };
695
696 /*
697 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
698 * radix trees
699 */
700 #define PAGECACHE_TAG_DIRTY 0
701 #define PAGECACHE_TAG_WRITEBACK 1
702 #define PAGECACHE_TAG_TOWRITE 2
703
704 int mapping_tagged(struct address_space *mapping, int tag);
705
706 /*
707 * Might pages of this file be mapped into userspace?
708 */
709 static inline int mapping_mapped(struct address_space *mapping)
710 {
711 return !prio_tree_empty(&mapping->i_mmap) ||
712 !list_empty(&mapping->i_mmap_nonlinear);
713 }
714
715 /*
716 * Might pages of this file have been modified in userspace?
717 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
718 * marks vma as VM_SHARED if it is shared, and the file was opened for
719 * writing i.e. vma may be mprotected writable even if now readonly.
720 */
721 static inline int mapping_writably_mapped(struct address_space *mapping)
722 {
723 return mapping->i_mmap_writable != 0;
724 }
725
726 /*
727 * Use sequence counter to get consistent i_size on 32-bit processors.
728 */
729 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
730 #include <linux/seqlock.h>
731 #define __NEED_I_SIZE_ORDERED
732 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
733 #else
734 #define i_size_ordered_init(inode) do { } while (0)
735 #endif
736
737 struct posix_acl;
738 #define ACL_NOT_CACHED ((void *)(-1))
739
740 #define IOP_FASTPERM 0x0001
741 #define IOP_LOOKUP 0x0002
742 #define IOP_NOFOLLOW 0x0004
743
744 /*
745 * Keep mostly read-only and often accessed (especially for
746 * the RCU path lookup and 'stat' data) fields at the beginning
747 * of the 'struct inode'
748 */
749 struct inode {
750 umode_t i_mode;
751 unsigned short i_opflags;
752 uid_t i_uid;
753 gid_t i_gid;
754 unsigned int i_flags;
755
756 #ifdef CONFIG_FS_POSIX_ACL
757 struct posix_acl *i_acl;
758 struct posix_acl *i_default_acl;
759 #endif
760
761 const struct inode_operations *i_op;
762 struct super_block *i_sb;
763 struct address_space *i_mapping;
764
765 #ifdef CONFIG_SECURITY
766 void *i_security;
767 #endif
768
769 /* Stat data, not accessed from path walking */
770 unsigned long i_ino;
771 unsigned int i_nlink;
772 dev_t i_rdev;
773 loff_t i_size;
774 struct timespec i_atime;
775 struct timespec i_mtime;
776 struct timespec i_ctime;
777 unsigned int i_blkbits;
778 blkcnt_t i_blocks;
779
780 #ifdef __NEED_I_SIZE_ORDERED
781 seqcount_t i_size_seqcount;
782 #endif
783
784 /* Misc */
785 unsigned long i_state;
786 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
787 struct mutex i_mutex;
788
789 unsigned long dirtied_when; /* jiffies of first dirtying */
790
791 struct hlist_node i_hash;
792 struct list_head i_wb_list; /* backing dev IO list */
793 struct list_head i_lru; /* inode LRU list */
794 struct list_head i_sb_list;
795 union {
796 struct list_head i_dentry;
797 struct rcu_head i_rcu;
798 };
799 atomic_t i_count;
800 u64 i_version;
801 unsigned short i_bytes;
802 atomic_t i_dio_count;
803 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
804 struct file_lock *i_flock;
805 struct address_space i_data;
806 #ifdef CONFIG_QUOTA
807 struct dquot *i_dquot[MAXQUOTAS];
808 #endif
809 struct list_head i_devices;
810 union {
811 struct pipe_inode_info *i_pipe;
812 struct block_device *i_bdev;
813 struct cdev *i_cdev;
814 };
815
816 __u32 i_generation;
817
818 #ifdef CONFIG_FSNOTIFY
819 __u32 i_fsnotify_mask; /* all events this inode cares about */
820 struct hlist_head i_fsnotify_marks;
821 #endif
822
823 #ifdef CONFIG_IMA
824 atomic_t i_readcount; /* struct files open RO */
825 #endif
826 atomic_t i_writecount;
827 void *i_private; /* fs or device private pointer */
828 };
829
830 static inline int inode_unhashed(struct inode *inode)
831 {
832 return hlist_unhashed(&inode->i_hash);
833 }
834
835 /*
836 * inode->i_mutex nesting subclasses for the lock validator:
837 *
838 * 0: the object of the current VFS operation
839 * 1: parent
840 * 2: child/target
841 * 3: quota file
842 *
843 * The locking order between these classes is
844 * parent -> child -> normal -> xattr -> quota
845 */
846 enum inode_i_mutex_lock_class
847 {
848 I_MUTEX_NORMAL,
849 I_MUTEX_PARENT,
850 I_MUTEX_CHILD,
851 I_MUTEX_XATTR,
852 I_MUTEX_QUOTA
853 };
854
855 /*
856 * NOTE: in a 32bit arch with a preemptable kernel and
857 * an UP compile the i_size_read/write must be atomic
858 * with respect to the local cpu (unlike with preempt disabled),
859 * but they don't need to be atomic with respect to other cpus like in
860 * true SMP (so they need either to either locally disable irq around
861 * the read or for example on x86 they can be still implemented as a
862 * cmpxchg8b without the need of the lock prefix). For SMP compiles
863 * and 64bit archs it makes no difference if preempt is enabled or not.
864 */
865 static inline loff_t i_size_read(const struct inode *inode)
866 {
867 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
868 loff_t i_size;
869 unsigned int seq;
870
871 do {
872 seq = read_seqcount_begin(&inode->i_size_seqcount);
873 i_size = inode->i_size;
874 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
875 return i_size;
876 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
877 loff_t i_size;
878
879 preempt_disable();
880 i_size = inode->i_size;
881 preempt_enable();
882 return i_size;
883 #else
884 return inode->i_size;
885 #endif
886 }
887
888 /*
889 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
890 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
891 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
892 */
893 static inline void i_size_write(struct inode *inode, loff_t i_size)
894 {
895 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
896 write_seqcount_begin(&inode->i_size_seqcount);
897 inode->i_size = i_size;
898 write_seqcount_end(&inode->i_size_seqcount);
899 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
900 preempt_disable();
901 inode->i_size = i_size;
902 preempt_enable();
903 #else
904 inode->i_size = i_size;
905 #endif
906 }
907
908 static inline unsigned iminor(const struct inode *inode)
909 {
910 return MINOR(inode->i_rdev);
911 }
912
913 static inline unsigned imajor(const struct inode *inode)
914 {
915 return MAJOR(inode->i_rdev);
916 }
917
918 extern struct block_device *I_BDEV(struct inode *inode);
919
920 struct fown_struct {
921 rwlock_t lock; /* protects pid, uid, euid fields */
922 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
923 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
924 uid_t uid, euid; /* uid/euid of process setting the owner */
925 int signum; /* posix.1b rt signal to be delivered on IO */
926 };
927
928 /*
929 * Track a single file's readahead state
930 */
931 struct file_ra_state {
932 pgoff_t start; /* where readahead started */
933 unsigned int size; /* # of readahead pages */
934 unsigned int async_size; /* do asynchronous readahead when
935 there are only # of pages ahead */
936
937 unsigned int ra_pages; /* Maximum readahead window */
938 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
939 loff_t prev_pos; /* Cache last read() position */
940 };
941
942 /*
943 * Check if @index falls in the readahead windows.
944 */
945 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
946 {
947 return (index >= ra->start &&
948 index < ra->start + ra->size);
949 }
950
951 #define FILE_MNT_WRITE_TAKEN 1
952 #define FILE_MNT_WRITE_RELEASED 2
953
954 struct file {
955 /*
956 * fu_list becomes invalid after file_free is called and queued via
957 * fu_rcuhead for RCU freeing
958 */
959 union {
960 struct list_head fu_list;
961 struct rcu_head fu_rcuhead;
962 } f_u;
963 struct path f_path;
964 #define f_dentry f_path.dentry
965 #define f_vfsmnt f_path.mnt
966 const struct file_operations *f_op;
967
968 /*
969 * Protects f_ep_links, f_flags, f_pos vs i_size in lseek SEEK_CUR.
970 * Must not be taken from IRQ context.
971 */
972 spinlock_t f_lock;
973 #ifdef CONFIG_SMP
974 int f_sb_list_cpu;
975 #endif
976 atomic_long_t f_count;
977 unsigned int f_flags;
978 fmode_t f_mode;
979 loff_t f_pos;
980 struct fown_struct f_owner;
981 const struct cred *f_cred;
982 struct file_ra_state f_ra;
983
984 u64 f_version;
985 #ifdef CONFIG_SECURITY
986 void *f_security;
987 #endif
988 /* needed for tty driver, and maybe others */
989 void *private_data;
990
991 #ifdef CONFIG_EPOLL
992 /* Used by fs/eventpoll.c to link all the hooks to this file */
993 struct list_head f_ep_links;
994 #endif /* #ifdef CONFIG_EPOLL */
995 struct address_space *f_mapping;
996 #ifdef CONFIG_DEBUG_WRITECOUNT
997 unsigned long f_mnt_write_state;
998 #endif
999 };
1000
1001 struct file_handle {
1002 __u32 handle_bytes;
1003 int handle_type;
1004 /* file identifier */
1005 unsigned char f_handle[0];
1006 };
1007
1008 #define get_file(x) atomic_long_inc(&(x)->f_count)
1009 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
1010 #define file_count(x) atomic_long_read(&(x)->f_count)
1011
1012 #ifdef CONFIG_DEBUG_WRITECOUNT
1013 static inline void file_take_write(struct file *f)
1014 {
1015 WARN_ON(f->f_mnt_write_state != 0);
1016 f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
1017 }
1018 static inline void file_release_write(struct file *f)
1019 {
1020 f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
1021 }
1022 static inline void file_reset_write(struct file *f)
1023 {
1024 f->f_mnt_write_state = 0;
1025 }
1026 static inline void file_check_state(struct file *f)
1027 {
1028 /*
1029 * At this point, either both or neither of these bits
1030 * should be set.
1031 */
1032 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
1033 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
1034 }
1035 static inline int file_check_writeable(struct file *f)
1036 {
1037 if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
1038 return 0;
1039 printk(KERN_WARNING "writeable file with no "
1040 "mnt_want_write()\n");
1041 WARN_ON(1);
1042 return -EINVAL;
1043 }
1044 #else /* !CONFIG_DEBUG_WRITECOUNT */
1045 static inline void file_take_write(struct file *filp) {}
1046 static inline void file_release_write(struct file *filp) {}
1047 static inline void file_reset_write(struct file *filp) {}
1048 static inline void file_check_state(struct file *filp) {}
1049 static inline int file_check_writeable(struct file *filp)
1050 {
1051 return 0;
1052 }
1053 #endif /* CONFIG_DEBUG_WRITECOUNT */
1054
1055 #define MAX_NON_LFS ((1UL<<31) - 1)
1056
1057 /* Page cache limit. The filesystems should put that into their s_maxbytes
1058 limits, otherwise bad things can happen in VM. */
1059 #if BITS_PER_LONG==32
1060 #define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
1061 #elif BITS_PER_LONG==64
1062 #define MAX_LFS_FILESIZE 0x7fffffffffffffffUL
1063 #endif
1064
1065 #define FL_POSIX 1
1066 #define FL_FLOCK 2
1067 #define FL_ACCESS 8 /* not trying to lock, just looking */
1068 #define FL_EXISTS 16 /* when unlocking, test for existence */
1069 #define FL_LEASE 32 /* lease held on this file */
1070 #define FL_CLOSE 64 /* unlock on close */
1071 #define FL_SLEEP 128 /* A blocking lock */
1072 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
1073 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
1074
1075 /*
1076 * Special return value from posix_lock_file() and vfs_lock_file() for
1077 * asynchronous locking.
1078 */
1079 #define FILE_LOCK_DEFERRED 1
1080
1081 /*
1082 * The POSIX file lock owner is determined by
1083 * the "struct files_struct" in the thread group
1084 * (or NULL for no owner - BSD locks).
1085 *
1086 * Lockd stuffs a "host" pointer into this.
1087 */
1088 typedef struct files_struct *fl_owner_t;
1089
1090 struct file_lock_operations {
1091 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1092 void (*fl_release_private)(struct file_lock *);
1093 };
1094
1095 struct lock_manager_operations {
1096 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
1097 void (*lm_notify)(struct file_lock *); /* unblock callback */
1098 int (*lm_grant)(struct file_lock *, struct file_lock *, int);
1099 void (*lm_release_private)(struct file_lock *);
1100 void (*lm_break)(struct file_lock *);
1101 int (*lm_change)(struct file_lock **, int);
1102 };
1103
1104 struct lock_manager {
1105 struct list_head list;
1106 };
1107
1108 void locks_start_grace(struct lock_manager *);
1109 void locks_end_grace(struct lock_manager *);
1110 int locks_in_grace(void);
1111
1112 /* that will die - we need it for nfs_lock_info */
1113 #include <linux/nfs_fs_i.h>
1114
1115 struct file_lock {
1116 struct file_lock *fl_next; /* singly linked list for this inode */
1117 struct list_head fl_link; /* doubly linked list of all locks */
1118 struct list_head fl_block; /* circular list of blocked processes */
1119 fl_owner_t fl_owner;
1120 unsigned int fl_flags;
1121 unsigned char fl_type;
1122 unsigned int fl_pid;
1123 struct pid *fl_nspid;
1124 wait_queue_head_t fl_wait;
1125 struct file *fl_file;
1126 loff_t fl_start;
1127 loff_t fl_end;
1128
1129 struct fasync_struct * fl_fasync; /* for lease break notifications */
1130 /* for lease breaks: */
1131 unsigned long fl_break_time;
1132 unsigned long fl_downgrade_time;
1133
1134 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1135 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1136 union {
1137 struct nfs_lock_info nfs_fl;
1138 struct nfs4_lock_info nfs4_fl;
1139 struct {
1140 struct list_head link; /* link in AFS vnode's pending_locks list */
1141 int state; /* state of grant or error if -ve */
1142 } afs;
1143 } fl_u;
1144 };
1145
1146 /* The following constant reflects the upper bound of the file/locking space */
1147 #ifndef OFFSET_MAX
1148 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1149 #define OFFSET_MAX INT_LIMIT(loff_t)
1150 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1151 #endif
1152
1153 #include <linux/fcntl.h>
1154
1155 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1156
1157 #ifdef CONFIG_FILE_LOCKING
1158 extern int fcntl_getlk(struct file *, struct flock __user *);
1159 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1160 struct flock __user *);
1161
1162 #if BITS_PER_LONG == 32
1163 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
1164 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1165 struct flock64 __user *);
1166 #endif
1167
1168 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1169 extern int fcntl_getlease(struct file *filp);
1170
1171 /* fs/locks.c */
1172 void locks_free_lock(struct file_lock *fl);
1173 extern void locks_init_lock(struct file_lock *);
1174 extern struct file_lock * locks_alloc_lock(void);
1175 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1176 extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
1177 extern void locks_remove_posix(struct file *, fl_owner_t);
1178 extern void locks_remove_flock(struct file *);
1179 extern void locks_release_private(struct file_lock *);
1180 extern void posix_test_lock(struct file *, struct file_lock *);
1181 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1182 extern int posix_lock_file_wait(struct file *, struct file_lock *);
1183 extern int posix_unblock_lock(struct file *, struct file_lock *);
1184 extern int vfs_test_lock(struct file *, struct file_lock *);
1185 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1186 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1187 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1188 extern int __break_lease(struct inode *inode, unsigned int flags);
1189 extern void lease_get_mtime(struct inode *, struct timespec *time);
1190 extern int generic_setlease(struct file *, long, struct file_lock **);
1191 extern int vfs_setlease(struct file *, long, struct file_lock **);
1192 extern int lease_modify(struct file_lock **, int);
1193 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1194 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1195 extern void lock_flocks(void);
1196 extern void unlock_flocks(void);
1197 #else /* !CONFIG_FILE_LOCKING */
1198 static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1199 {
1200 return -EINVAL;
1201 }
1202
1203 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1204 unsigned int cmd, struct flock __user *user)
1205 {
1206 return -EACCES;
1207 }
1208
1209 #if BITS_PER_LONG == 32
1210 static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1211 {
1212 return -EINVAL;
1213 }
1214
1215 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1216 unsigned int cmd, struct flock64 __user *user)
1217 {
1218 return -EACCES;
1219 }
1220 #endif
1221 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1222 {
1223 return 0;
1224 }
1225
1226 static inline int fcntl_getlease(struct file *filp)
1227 {
1228 return 0;
1229 }
1230
1231 static inline void locks_init_lock(struct file_lock *fl)
1232 {
1233 return;
1234 }
1235
1236 static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1237 {
1238 return;
1239 }
1240
1241 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1242 {
1243 return;
1244 }
1245
1246 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1247 {
1248 return;
1249 }
1250
1251 static inline void locks_remove_flock(struct file *filp)
1252 {
1253 return;
1254 }
1255
1256 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1257 {
1258 return;
1259 }
1260
1261 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1262 struct file_lock *conflock)
1263 {
1264 return -ENOLCK;
1265 }
1266
1267 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1268 {
1269 return -ENOLCK;
1270 }
1271
1272 static inline int posix_unblock_lock(struct file *filp,
1273 struct file_lock *waiter)
1274 {
1275 return -ENOENT;
1276 }
1277
1278 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1279 {
1280 return 0;
1281 }
1282
1283 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1284 struct file_lock *fl, struct file_lock *conf)
1285 {
1286 return -ENOLCK;
1287 }
1288
1289 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1290 {
1291 return 0;
1292 }
1293
1294 static inline int flock_lock_file_wait(struct file *filp,
1295 struct file_lock *request)
1296 {
1297 return -ENOLCK;
1298 }
1299
1300 static inline int __break_lease(struct inode *inode, unsigned int mode)
1301 {
1302 return 0;
1303 }
1304
1305 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1306 {
1307 return;
1308 }
1309
1310 static inline int generic_setlease(struct file *filp, long arg,
1311 struct file_lock **flp)
1312 {
1313 return -EINVAL;
1314 }
1315
1316 static inline int vfs_setlease(struct file *filp, long arg,
1317 struct file_lock **lease)
1318 {
1319 return -EINVAL;
1320 }
1321
1322 static inline int lease_modify(struct file_lock **before, int arg)
1323 {
1324 return -EINVAL;
1325 }
1326
1327 static inline int lock_may_read(struct inode *inode, loff_t start,
1328 unsigned long len)
1329 {
1330 return 1;
1331 }
1332
1333 static inline int lock_may_write(struct inode *inode, loff_t start,
1334 unsigned long len)
1335 {
1336 return 1;
1337 }
1338
1339 static inline void lock_flocks(void)
1340 {
1341 }
1342
1343 static inline void unlock_flocks(void)
1344 {
1345 }
1346
1347 #endif /* !CONFIG_FILE_LOCKING */
1348
1349
1350 struct fasync_struct {
1351 spinlock_t fa_lock;
1352 int magic;
1353 int fa_fd;
1354 struct fasync_struct *fa_next; /* singly linked list */
1355 struct file *fa_file;
1356 struct rcu_head fa_rcu;
1357 };
1358
1359 #define FASYNC_MAGIC 0x4601
1360
1361 /* SMP safe fasync helpers: */
1362 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1363 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1364 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1365 extern struct fasync_struct *fasync_alloc(void);
1366 extern void fasync_free(struct fasync_struct *);
1367
1368 /* can be called from interrupts */
1369 extern void kill_fasync(struct fasync_struct **, int, int);
1370
1371 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1372 extern int f_setown(struct file *filp, unsigned long arg, int force);
1373 extern void f_delown(struct file *filp);
1374 extern pid_t f_getown(struct file *filp);
1375 extern int send_sigurg(struct fown_struct *fown);
1376
1377 /*
1378 * Umount options
1379 */
1380
1381 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1382 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1383 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1384 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1385 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1386
1387 extern struct list_head super_blocks;
1388 extern spinlock_t sb_lock;
1389
1390 struct super_block {
1391 struct list_head s_list; /* Keep this first */
1392 dev_t s_dev; /* search index; _not_ kdev_t */
1393 unsigned char s_dirt;
1394 unsigned char s_blocksize_bits;
1395 unsigned long s_blocksize;
1396 loff_t s_maxbytes; /* Max file size */
1397 struct file_system_type *s_type;
1398 const struct super_operations *s_op;
1399 const struct dquot_operations *dq_op;
1400 const struct quotactl_ops *s_qcop;
1401 const struct export_operations *s_export_op;
1402 unsigned long s_flags;
1403 unsigned long s_magic;
1404 struct dentry *s_root;
1405 struct rw_semaphore s_umount;
1406 struct mutex s_lock;
1407 int s_count;
1408 atomic_t s_active;
1409 #ifdef CONFIG_SECURITY
1410 void *s_security;
1411 #endif
1412 const struct xattr_handler **s_xattr;
1413
1414 struct list_head s_inodes; /* all inodes */
1415 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1416 #ifdef CONFIG_SMP
1417 struct list_head __percpu *s_files;
1418 #else
1419 struct list_head s_files;
1420 #endif
1421 /* s_dentry_lru, s_nr_dentry_unused protected by dcache.c lru locks */
1422 struct list_head s_dentry_lru; /* unused dentry lru */
1423 int s_nr_dentry_unused; /* # of dentry on lru */
1424
1425 /* s_inode_lru_lock protects s_inode_lru and s_nr_inodes_unused */
1426 spinlock_t s_inode_lru_lock ____cacheline_aligned_in_smp;
1427 struct list_head s_inode_lru; /* unused inode lru */
1428 int s_nr_inodes_unused; /* # of inodes on lru */
1429
1430 struct block_device *s_bdev;
1431 struct backing_dev_info *s_bdi;
1432 struct mtd_info *s_mtd;
1433 struct list_head s_instances;
1434 struct quota_info s_dquot; /* Diskquota specific options */
1435
1436 int s_frozen;
1437 wait_queue_head_t s_wait_unfrozen;
1438
1439 char s_id[32]; /* Informational name */
1440 u8 s_uuid[16]; /* UUID */
1441
1442 void *s_fs_info; /* Filesystem private info */
1443 fmode_t s_mode;
1444
1445 /* Granularity of c/m/atime in ns.
1446 Cannot be worse than a second */
1447 u32 s_time_gran;
1448
1449 /*
1450 * The next field is for VFS *only*. No filesystems have any business
1451 * even looking at it. You had been warned.
1452 */
1453 struct mutex s_vfs_rename_mutex; /* Kludge */
1454
1455 /*
1456 * Filesystem subtype. If non-empty the filesystem type field
1457 * in /proc/mounts will be "type.subtype"
1458 */
1459 char *s_subtype;
1460
1461 /*
1462 * Saved mount options for lazy filesystems using
1463 * generic_show_options()
1464 */
1465 char __rcu *s_options;
1466 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1467
1468 /*
1469 * Saved pool identifier for cleancache (-1 means none)
1470 */
1471 int cleancache_poolid;
1472
1473 struct shrinker s_shrink; /* per-sb shrinker handle */
1474 };
1475
1476 /* superblock cache pruning functions */
1477 extern void prune_icache_sb(struct super_block *sb, int nr_to_scan);
1478 extern void prune_dcache_sb(struct super_block *sb, int nr_to_scan);
1479
1480 extern struct timespec current_fs_time(struct super_block *sb);
1481
1482 /*
1483 * Snapshotting support.
1484 */
1485 enum {
1486 SB_UNFROZEN = 0,
1487 SB_FREEZE_WRITE = 1,
1488 SB_FREEZE_TRANS = 2,
1489 };
1490
1491 #define vfs_check_frozen(sb, level) \
1492 wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
1493
1494 /*
1495 * until VFS tracks user namespaces for inodes, just make all files
1496 * belong to init_user_ns
1497 */
1498 extern struct user_namespace init_user_ns;
1499 #define inode_userns(inode) (&init_user_ns)
1500 extern bool inode_owner_or_capable(const struct inode *inode);
1501
1502 /* not quite ready to be deprecated, but... */
1503 extern void lock_super(struct super_block *);
1504 extern void unlock_super(struct super_block *);
1505
1506 /*
1507 * VFS helper functions..
1508 */
1509 extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
1510 extern int vfs_mkdir(struct inode *, struct dentry *, int);
1511 extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
1512 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1513 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1514 extern int vfs_rmdir(struct inode *, struct dentry *);
1515 extern int vfs_unlink(struct inode *, struct dentry *);
1516 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1517
1518 /*
1519 * VFS dentry helper functions.
1520 */
1521 extern void dentry_unhash(struct dentry *dentry);
1522
1523 /*
1524 * VFS file helper functions.
1525 */
1526 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1527 mode_t mode);
1528 /*
1529 * VFS FS_IOC_FIEMAP helper definitions.
1530 */
1531 struct fiemap_extent_info {
1532 unsigned int fi_flags; /* Flags as passed from user */
1533 unsigned int fi_extents_mapped; /* Number of mapped extents */
1534 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1535 struct fiemap_extent __user *fi_extents_start; /* Start of
1536 fiemap_extent array */
1537 };
1538 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1539 u64 phys, u64 len, u32 flags);
1540 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1541
1542 /*
1543 * File types
1544 *
1545 * NOTE! These match bits 12..15 of stat.st_mode
1546 * (ie "(i_mode >> 12) & 15").
1547 */
1548 #define DT_UNKNOWN 0
1549 #define DT_FIFO 1
1550 #define DT_CHR 2
1551 #define DT_DIR 4
1552 #define DT_BLK 6
1553 #define DT_REG 8
1554 #define DT_LNK 10
1555 #define DT_SOCK 12
1556 #define DT_WHT 14
1557
1558 /*
1559 * This is the "filldir" function type, used by readdir() to let
1560 * the kernel specify what kind of dirent layout it wants to have.
1561 * This allows the kernel to read directories into kernel space or
1562 * to have different dirent layouts depending on the binary type.
1563 */
1564 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1565 struct block_device_operations;
1566
1567 /* These macros are for out of kernel modules to test that
1568 * the kernel supports the unlocked_ioctl and compat_ioctl
1569 * fields in struct file_operations. */
1570 #define HAVE_COMPAT_IOCTL 1
1571 #define HAVE_UNLOCKED_IOCTL 1
1572
1573 struct file_operations {
1574 struct module *owner;
1575 loff_t (*llseek) (struct file *, loff_t, int);
1576 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1577 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1578 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1579 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1580 int (*readdir) (struct file *, void *, filldir_t);
1581 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1582 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1583 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1584 int (*mmap) (struct file *, struct vm_area_struct *);
1585 int (*open) (struct inode *, struct file *);
1586 int (*flush) (struct file *, fl_owner_t id);
1587 int (*release) (struct inode *, struct file *);
1588 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1589 int (*aio_fsync) (struct kiocb *, int datasync);
1590 int (*fasync) (int, struct file *, int);
1591 int (*lock) (struct file *, int, struct file_lock *);
1592 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1593 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1594 int (*check_flags)(int);
1595 int (*flock) (struct file *, int, struct file_lock *);
1596 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1597 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1598 int (*setlease)(struct file *, long, struct file_lock **);
1599 long (*fallocate)(struct file *file, int mode, loff_t offset,
1600 loff_t len);
1601 };
1602
1603 struct inode_operations {
1604 struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
1605 void * (*follow_link) (struct dentry *, struct nameidata *);
1606 int (*permission) (struct inode *, int);
1607 struct posix_acl * (*get_acl)(struct inode *, int);
1608
1609 int (*readlink) (struct dentry *, char __user *,int);
1610 void (*put_link) (struct dentry *, struct nameidata *, void *);
1611
1612 int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
1613 int (*link) (struct dentry *,struct inode *,struct dentry *);
1614 int (*unlink) (struct inode *,struct dentry *);
1615 int (*symlink) (struct inode *,struct dentry *,const char *);
1616 int (*mkdir) (struct inode *,struct dentry *,int);
1617 int (*rmdir) (struct inode *,struct dentry *);
1618 int (*mknod) (struct inode *,struct dentry *,int,dev_t);
1619 int (*rename) (struct inode *, struct dentry *,
1620 struct inode *, struct dentry *);
1621 void (*truncate) (struct inode *);
1622 int (*setattr) (struct dentry *, struct iattr *);
1623 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1624 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1625 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1626 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1627 int (*removexattr) (struct dentry *, const char *);
1628 void (*truncate_range)(struct inode *, loff_t, loff_t);
1629 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1630 u64 len);
1631 } ____cacheline_aligned;
1632
1633 struct seq_file;
1634
1635 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1636 unsigned long nr_segs, unsigned long fast_segs,
1637 struct iovec *fast_pointer,
1638 struct iovec **ret_pointer,
1639 int check_access);
1640
1641 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1642 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1643 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1644 unsigned long, loff_t *);
1645 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1646 unsigned long, loff_t *);
1647
1648 struct super_operations {
1649 struct inode *(*alloc_inode)(struct super_block *sb);
1650 void (*destroy_inode)(struct inode *);
1651
1652 void (*dirty_inode) (struct inode *, int flags);
1653 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1654 int (*drop_inode) (struct inode *);
1655 void (*evict_inode) (struct inode *);
1656 void (*put_super) (struct super_block *);
1657 void (*write_super) (struct super_block *);
1658 int (*sync_fs)(struct super_block *sb, int wait);
1659 int (*freeze_fs) (struct super_block *);
1660 int (*unfreeze_fs) (struct super_block *);
1661 int (*statfs) (struct dentry *, struct kstatfs *);
1662 int (*remount_fs) (struct super_block *, int *, char *);
1663 void (*umount_begin) (struct super_block *);
1664
1665 int (*show_options)(struct seq_file *, struct vfsmount *);
1666 int (*show_devname)(struct seq_file *, struct vfsmount *);
1667 int (*show_path)(struct seq_file *, struct vfsmount *);
1668 int (*show_stats)(struct seq_file *, struct vfsmount *);
1669 #ifdef CONFIG_QUOTA
1670 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1671 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1672 #endif
1673 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1674 int (*nr_cached_objects)(struct super_block *);
1675 void (*free_cached_objects)(struct super_block *, int);
1676 };
1677
1678 /*
1679 * Inode state bits. Protected by inode->i_lock
1680 *
1681 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1682 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1683 *
1684 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1685 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1686 * various stages of removing an inode.
1687 *
1688 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1689 *
1690 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1691 * fdatasync(). i_atime is the usual cause.
1692 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1693 * these changes separately from I_DIRTY_SYNC so that we
1694 * don't have to write inode on fdatasync() when only
1695 * mtime has changed in it.
1696 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1697 * I_NEW Serves as both a mutex and completion notification.
1698 * New inodes set I_NEW. If two processes both create
1699 * the same inode, one of them will release its inode and
1700 * wait for I_NEW to be released before returning.
1701 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1702 * also cause waiting on I_NEW, without I_NEW actually
1703 * being set. find_inode() uses this to prevent returning
1704 * nearly-dead inodes.
1705 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1706 * is zero. I_FREEING must be set when I_WILL_FREE is
1707 * cleared.
1708 * I_FREEING Set when inode is about to be freed but still has dirty
1709 * pages or buffers attached or the inode itself is still
1710 * dirty.
1711 * I_CLEAR Added by end_writeback(). In this state the inode is clean
1712 * and can be destroyed. Inode keeps I_FREEING.
1713 *
1714 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1715 * prohibited for many purposes. iget() must wait for
1716 * the inode to be completely released, then create it
1717 * anew. Other functions will just ignore such inodes,
1718 * if appropriate. I_NEW is used for waiting.
1719 *
1720 * I_SYNC Synchonized write of dirty inode data. The bits is
1721 * set during data writeback, and cleared with a wakeup
1722 * on the bit address once it is done.
1723 *
1724 * I_REFERENCED Marks the inode as recently references on the LRU list.
1725 *
1726 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1727 *
1728 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1729 */
1730 #define I_DIRTY_SYNC (1 << 0)
1731 #define I_DIRTY_DATASYNC (1 << 1)
1732 #define I_DIRTY_PAGES (1 << 2)
1733 #define __I_NEW 3
1734 #define I_NEW (1 << __I_NEW)
1735 #define I_WILL_FREE (1 << 4)
1736 #define I_FREEING (1 << 5)
1737 #define I_CLEAR (1 << 6)
1738 #define __I_SYNC 7
1739 #define I_SYNC (1 << __I_SYNC)
1740 #define I_REFERENCED (1 << 8)
1741 #define __I_DIO_WAKEUP 9
1742 #define I_DIO_WAKEUP (1 << I_DIO_WAKEUP)
1743
1744 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1745
1746 extern void __mark_inode_dirty(struct inode *, int);
1747 static inline void mark_inode_dirty(struct inode *inode)
1748 {
1749 __mark_inode_dirty(inode, I_DIRTY);
1750 }
1751
1752 static inline void mark_inode_dirty_sync(struct inode *inode)
1753 {
1754 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1755 }
1756
1757 /**
1758 * inc_nlink - directly increment an inode's link count
1759 * @inode: inode
1760 *
1761 * This is a low-level filesystem helper to replace any
1762 * direct filesystem manipulation of i_nlink. Currently,
1763 * it is only here for parity with dec_nlink().
1764 */
1765 static inline void inc_nlink(struct inode *inode)
1766 {
1767 inode->i_nlink++;
1768 }
1769
1770 static inline void inode_inc_link_count(struct inode *inode)
1771 {
1772 inc_nlink(inode);
1773 mark_inode_dirty(inode);
1774 }
1775
1776 /**
1777 * drop_nlink - directly drop an inode's link count
1778 * @inode: inode
1779 *
1780 * This is a low-level filesystem helper to replace any
1781 * direct filesystem manipulation of i_nlink. In cases
1782 * where we are attempting to track writes to the
1783 * filesystem, a decrement to zero means an imminent
1784 * write when the file is truncated and actually unlinked
1785 * on the filesystem.
1786 */
1787 static inline void drop_nlink(struct inode *inode)
1788 {
1789 inode->i_nlink--;
1790 }
1791
1792 /**
1793 * clear_nlink - directly zero an inode's link count
1794 * @inode: inode
1795 *
1796 * This is a low-level filesystem helper to replace any
1797 * direct filesystem manipulation of i_nlink. See
1798 * drop_nlink() for why we care about i_nlink hitting zero.
1799 */
1800 static inline void clear_nlink(struct inode *inode)
1801 {
1802 inode->i_nlink = 0;
1803 }
1804
1805 static inline void inode_dec_link_count(struct inode *inode)
1806 {
1807 drop_nlink(inode);
1808 mark_inode_dirty(inode);
1809 }
1810
1811 /**
1812 * inode_inc_iversion - increments i_version
1813 * @inode: inode that need to be updated
1814 *
1815 * Every time the inode is modified, the i_version field will be incremented.
1816 * The filesystem has to be mounted with i_version flag
1817 */
1818
1819 static inline void inode_inc_iversion(struct inode *inode)
1820 {
1821 spin_lock(&inode->i_lock);
1822 inode->i_version++;
1823 spin_unlock(&inode->i_lock);
1824 }
1825
1826 extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
1827 static inline void file_accessed(struct file *file)
1828 {
1829 if (!(file->f_flags & O_NOATIME))
1830 touch_atime(file->f_path.mnt, file->f_path.dentry);
1831 }
1832
1833 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1834 int sync_inode_metadata(struct inode *inode, int wait);
1835
1836 struct file_system_type {
1837 const char *name;
1838 int fs_flags;
1839 struct dentry *(*mount) (struct file_system_type *, int,
1840 const char *, void *);
1841 void (*kill_sb) (struct super_block *);
1842 struct module *owner;
1843 struct file_system_type * next;
1844 struct list_head fs_supers;
1845
1846 struct lock_class_key s_lock_key;
1847 struct lock_class_key s_umount_key;
1848 struct lock_class_key s_vfs_rename_key;
1849
1850 struct lock_class_key i_lock_key;
1851 struct lock_class_key i_mutex_key;
1852 struct lock_class_key i_mutex_dir_key;
1853 };
1854
1855 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1856 void *data, int (*fill_super)(struct super_block *, void *, int));
1857 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1858 int flags, const char *dev_name, void *data,
1859 int (*fill_super)(struct super_block *, void *, int));
1860 extern struct dentry *mount_single(struct file_system_type *fs_type,
1861 int flags, void *data,
1862 int (*fill_super)(struct super_block *, void *, int));
1863 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1864 int flags, void *data,
1865 int (*fill_super)(struct super_block *, void *, int));
1866 void generic_shutdown_super(struct super_block *sb);
1867 void kill_block_super(struct super_block *sb);
1868 void kill_anon_super(struct super_block *sb);
1869 void kill_litter_super(struct super_block *sb);
1870 void deactivate_super(struct super_block *sb);
1871 void deactivate_locked_super(struct super_block *sb);
1872 int set_anon_super(struct super_block *s, void *data);
1873 int get_anon_bdev(dev_t *);
1874 void free_anon_bdev(dev_t);
1875 struct super_block *sget(struct file_system_type *type,
1876 int (*test)(struct super_block *,void *),
1877 int (*set)(struct super_block *,void *),
1878 void *data);
1879 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1880 const struct super_operations *ops,
1881 const struct dentry_operations *dops,
1882 unsigned long);
1883
1884 static inline void sb_mark_dirty(struct super_block *sb)
1885 {
1886 sb->s_dirt = 1;
1887 }
1888 static inline void sb_mark_clean(struct super_block *sb)
1889 {
1890 sb->s_dirt = 0;
1891 }
1892 static inline int sb_is_dirty(struct super_block *sb)
1893 {
1894 return sb->s_dirt;
1895 }
1896
1897 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1898 #define fops_get(fops) \
1899 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1900 #define fops_put(fops) \
1901 do { if (fops) module_put((fops)->owner); } while(0)
1902
1903 extern int register_filesystem(struct file_system_type *);
1904 extern int unregister_filesystem(struct file_system_type *);
1905 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1906 #define kern_mount(type) kern_mount_data(type, NULL)
1907 extern void kern_unmount(struct vfsmount *mnt);
1908 extern int may_umount_tree(struct vfsmount *);
1909 extern int may_umount(struct vfsmount *);
1910 extern long do_mount(char *, char *, char *, unsigned long, void *);
1911 extern struct vfsmount *collect_mounts(struct path *);
1912 extern void drop_collected_mounts(struct vfsmount *);
1913 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1914 struct vfsmount *);
1915 extern int vfs_statfs(struct path *, struct kstatfs *);
1916 extern int user_statfs(const char __user *, struct kstatfs *);
1917 extern int fd_statfs(int, struct kstatfs *);
1918 extern int statfs_by_dentry(struct dentry *, struct kstatfs *);
1919 extern int freeze_super(struct super_block *super);
1920 extern int thaw_super(struct super_block *super);
1921
1922 extern int current_umask(void);
1923
1924 /* /sys/fs */
1925 extern struct kobject *fs_kobj;
1926
1927 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1928 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1929
1930 #define FLOCK_VERIFY_READ 1
1931 #define FLOCK_VERIFY_WRITE 2
1932
1933 #ifdef CONFIG_FILE_LOCKING
1934 extern int locks_mandatory_locked(struct inode *);
1935 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1936
1937 /*
1938 * Candidates for mandatory locking have the setgid bit set
1939 * but no group execute bit - an otherwise meaningless combination.
1940 */
1941
1942 static inline int __mandatory_lock(struct inode *ino)
1943 {
1944 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1945 }
1946
1947 /*
1948 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1949 * otherwise these will be advisory locks
1950 */
1951
1952 static inline int mandatory_lock(struct inode *ino)
1953 {
1954 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1955 }
1956
1957 static inline int locks_verify_locked(struct inode *inode)
1958 {
1959 if (mandatory_lock(inode))
1960 return locks_mandatory_locked(inode);
1961 return 0;
1962 }
1963
1964 static inline int locks_verify_truncate(struct inode *inode,
1965 struct file *filp,
1966 loff_t size)
1967 {
1968 if (inode->i_flock && mandatory_lock(inode))
1969 return locks_mandatory_area(
1970 FLOCK_VERIFY_WRITE, inode, filp,
1971 size < inode->i_size ? size : inode->i_size,
1972 (size < inode->i_size ? inode->i_size - size
1973 : size - inode->i_size)
1974 );
1975 return 0;
1976 }
1977
1978 static inline int break_lease(struct inode *inode, unsigned int mode)
1979 {
1980 if (inode->i_flock)
1981 return __break_lease(inode, mode);
1982 return 0;
1983 }
1984 #else /* !CONFIG_FILE_LOCKING */
1985 static inline int locks_mandatory_locked(struct inode *inode)
1986 {
1987 return 0;
1988 }
1989
1990 static inline int locks_mandatory_area(int rw, struct inode *inode,
1991 struct file *filp, loff_t offset,
1992 size_t count)
1993 {
1994 return 0;
1995 }
1996
1997 static inline int __mandatory_lock(struct inode *inode)
1998 {
1999 return 0;
2000 }
2001
2002 static inline int mandatory_lock(struct inode *inode)
2003 {
2004 return 0;
2005 }
2006
2007 static inline int locks_verify_locked(struct inode *inode)
2008 {
2009 return 0;
2010 }
2011
2012 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2013 size_t size)
2014 {
2015 return 0;
2016 }
2017
2018 static inline int break_lease(struct inode *inode, unsigned int mode)
2019 {
2020 return 0;
2021 }
2022
2023 #endif /* CONFIG_FILE_LOCKING */
2024
2025 /* fs/open.c */
2026
2027 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2028 struct file *filp);
2029 extern int do_fallocate(struct file *file, int mode, loff_t offset,
2030 loff_t len);
2031 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2032 int mode);
2033 extern struct file *filp_open(const char *, int, int);
2034 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2035 const char *, int);
2036 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int,
2037 const struct cred *);
2038 extern int filp_close(struct file *, fl_owner_t id);
2039 extern char * getname(const char __user *);
2040
2041 /* fs/ioctl.c */
2042
2043 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2044
2045 /* fs/dcache.c */
2046 extern void __init vfs_caches_init_early(void);
2047 extern void __init vfs_caches_init(unsigned long);
2048
2049 extern struct kmem_cache *names_cachep;
2050
2051 #define __getname_gfp(gfp) kmem_cache_alloc(names_cachep, (gfp))
2052 #define __getname() __getname_gfp(GFP_KERNEL)
2053 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2054 #ifndef CONFIG_AUDITSYSCALL
2055 #define putname(name) __putname(name)
2056 #else
2057 extern void putname(const char *name);
2058 #endif
2059
2060 #ifdef CONFIG_BLOCK
2061 extern int register_blkdev(unsigned int, const char *);
2062 extern void unregister_blkdev(unsigned int, const char *);
2063 extern struct block_device *bdget(dev_t);
2064 extern struct block_device *bdgrab(struct block_device *bdev);
2065 extern void bd_set_size(struct block_device *, loff_t size);
2066 extern void bd_forget(struct inode *inode);
2067 extern void bdput(struct block_device *);
2068 extern void invalidate_bdev(struct block_device *);
2069 extern int sync_blockdev(struct block_device *bdev);
2070 extern struct super_block *freeze_bdev(struct block_device *);
2071 extern void emergency_thaw_all(void);
2072 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2073 extern int fsync_bdev(struct block_device *);
2074 #else
2075 static inline void bd_forget(struct inode *inode) {}
2076 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2077 static inline void invalidate_bdev(struct block_device *bdev) {}
2078
2079 static inline struct super_block *freeze_bdev(struct block_device *sb)
2080 {
2081 return NULL;
2082 }
2083
2084 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2085 {
2086 return 0;
2087 }
2088 #endif
2089 extern int sync_filesystem(struct super_block *);
2090 extern const struct file_operations def_blk_fops;
2091 extern const struct file_operations def_chr_fops;
2092 extern const struct file_operations bad_sock_fops;
2093 extern const struct file_operations def_fifo_fops;
2094 #ifdef CONFIG_BLOCK
2095 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2096 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2097 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2098 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2099 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2100 void *holder);
2101 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2102 void *holder);
2103 extern int blkdev_put(struct block_device *bdev, fmode_t mode);
2104 #ifdef CONFIG_SYSFS
2105 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2106 extern void bd_unlink_disk_holder(struct block_device *bdev,
2107 struct gendisk *disk);
2108 #else
2109 static inline int bd_link_disk_holder(struct block_device *bdev,
2110 struct gendisk *disk)
2111 {
2112 return 0;
2113 }
2114 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2115 struct gendisk *disk)
2116 {
2117 }
2118 #endif
2119 #endif
2120
2121 /* fs/char_dev.c */
2122 #define CHRDEV_MAJOR_HASH_SIZE 255
2123 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2124 extern int register_chrdev_region(dev_t, unsigned, const char *);
2125 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2126 unsigned int count, const char *name,
2127 const struct file_operations *fops);
2128 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2129 unsigned int count, const char *name);
2130 extern void unregister_chrdev_region(dev_t, unsigned);
2131 extern void chrdev_show(struct seq_file *,off_t);
2132
2133 static inline int register_chrdev(unsigned int major, const char *name,
2134 const struct file_operations *fops)
2135 {
2136 return __register_chrdev(major, 0, 256, name, fops);
2137 }
2138
2139 static inline void unregister_chrdev(unsigned int major, const char *name)
2140 {
2141 __unregister_chrdev(major, 0, 256, name);
2142 }
2143
2144 /* fs/block_dev.c */
2145 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2146 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2147
2148 #ifdef CONFIG_BLOCK
2149 #define BLKDEV_MAJOR_HASH_SIZE 255
2150 extern const char *__bdevname(dev_t, char *buffer);
2151 extern const char *bdevname(struct block_device *bdev, char *buffer);
2152 extern struct block_device *lookup_bdev(const char *);
2153 extern void blkdev_show(struct seq_file *,off_t);
2154
2155 #else
2156 #define BLKDEV_MAJOR_HASH_SIZE 0
2157 #endif
2158
2159 extern void init_special_inode(struct inode *, umode_t, dev_t);
2160
2161 /* Invalid inode operations -- fs/bad_inode.c */
2162 extern void make_bad_inode(struct inode *);
2163 extern int is_bad_inode(struct inode *);
2164
2165 extern const struct file_operations read_pipefifo_fops;
2166 extern const struct file_operations write_pipefifo_fops;
2167 extern const struct file_operations rdwr_pipefifo_fops;
2168
2169 extern int fs_may_remount_ro(struct super_block *);
2170
2171 #ifdef CONFIG_BLOCK
2172 /*
2173 * return READ, READA, or WRITE
2174 */
2175 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2176
2177 /*
2178 * return data direction, READ or WRITE
2179 */
2180 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2181
2182 extern void check_disk_size_change(struct gendisk *disk,
2183 struct block_device *bdev);
2184 extern int revalidate_disk(struct gendisk *);
2185 extern int check_disk_change(struct block_device *);
2186 extern int __invalidate_device(struct block_device *, bool);
2187 extern int invalidate_partition(struct gendisk *, int);
2188 #endif
2189 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2190 pgoff_t start, pgoff_t end);
2191
2192 static inline void invalidate_remote_inode(struct inode *inode)
2193 {
2194 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2195 S_ISLNK(inode->i_mode))
2196 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2197 }
2198 extern int invalidate_inode_pages2(struct address_space *mapping);
2199 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2200 pgoff_t start, pgoff_t end);
2201 extern int write_inode_now(struct inode *, int);
2202 extern int filemap_fdatawrite(struct address_space *);
2203 extern int filemap_flush(struct address_space *);
2204 extern int filemap_fdatawait(struct address_space *);
2205 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2206 loff_t lend);
2207 extern int filemap_write_and_wait(struct address_space *mapping);
2208 extern int filemap_write_and_wait_range(struct address_space *mapping,
2209 loff_t lstart, loff_t lend);
2210 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2211 loff_t start, loff_t end, int sync_mode);
2212 extern int filemap_fdatawrite_range(struct address_space *mapping,
2213 loff_t start, loff_t end);
2214
2215 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2216 int datasync);
2217 extern int vfs_fsync(struct file *file, int datasync);
2218 extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
2219 extern void sync_supers(void);
2220 extern void emergency_sync(void);
2221 extern void emergency_remount(void);
2222 #ifdef CONFIG_BLOCK
2223 extern sector_t bmap(struct inode *, sector_t);
2224 #endif
2225 extern int notify_change(struct dentry *, struct iattr *);
2226 extern int inode_permission(struct inode *, int);
2227 extern int generic_permission(struct inode *, int);
2228
2229 static inline bool execute_ok(struct inode *inode)
2230 {
2231 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2232 }
2233
2234 /*
2235 * get_write_access() gets write permission for a file.
2236 * put_write_access() releases this write permission.
2237 * This is used for regular files.
2238 * We cannot support write (and maybe mmap read-write shared) accesses and
2239 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2240 * can have the following values:
2241 * 0: no writers, no VM_DENYWRITE mappings
2242 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2243 * > 0: (i_writecount) users are writing to the file.
2244 *
2245 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2246 * except for the cases where we don't hold i_writecount yet. Then we need to
2247 * use {get,deny}_write_access() - these functions check the sign and refuse
2248 * to do the change if sign is wrong.
2249 */
2250 static inline int get_write_access(struct inode *inode)
2251 {
2252 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2253 }
2254 static inline int deny_write_access(struct file *file)
2255 {
2256 struct inode *inode = file->f_path.dentry->d_inode;
2257 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2258 }
2259 static inline void put_write_access(struct inode * inode)
2260 {
2261 atomic_dec(&inode->i_writecount);
2262 }
2263 static inline void allow_write_access(struct file *file)
2264 {
2265 if (file)
2266 atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
2267 }
2268 #ifdef CONFIG_IMA
2269 static inline void i_readcount_dec(struct inode *inode)
2270 {
2271 BUG_ON(!atomic_read(&inode->i_readcount));
2272 atomic_dec(&inode->i_readcount);
2273 }
2274 static inline void i_readcount_inc(struct inode *inode)
2275 {
2276 atomic_inc(&inode->i_readcount);
2277 }
2278 #else
2279 static inline void i_readcount_dec(struct inode *inode)
2280 {
2281 return;
2282 }
2283 static inline void i_readcount_inc(struct inode *inode)
2284 {
2285 return;
2286 }
2287 #endif
2288 extern int do_pipe_flags(int *, int);
2289 extern struct file *create_read_pipe(struct file *f, int flags);
2290 extern struct file *create_write_pipe(int flags);
2291 extern void free_write_pipe(struct file *);
2292
2293 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2294 extern struct file * open_exec(const char *);
2295
2296 /* fs/dcache.c -- generic fs support functions */
2297 extern int is_subdir(struct dentry *, struct dentry *);
2298 extern int path_is_under(struct path *, struct path *);
2299 extern ino_t find_inode_number(struct dentry *, struct qstr *);
2300
2301 #include <linux/err.h>
2302
2303 /* needed for stackable file system support */
2304 extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
2305
2306 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
2307
2308 extern int inode_init_always(struct super_block *, struct inode *);
2309 extern void inode_init_once(struct inode *);
2310 extern void address_space_init_once(struct address_space *mapping);
2311 extern void ihold(struct inode * inode);
2312 extern void iput(struct inode *);
2313 extern struct inode * igrab(struct inode *);
2314 extern ino_t iunique(struct super_block *, ino_t);
2315 extern int inode_needs_sync(struct inode *inode);
2316 extern int generic_delete_inode(struct inode *inode);
2317 extern int generic_drop_inode(struct inode *inode);
2318
2319 extern struct inode *ilookup5_nowait(struct super_block *sb,
2320 unsigned long hashval, int (*test)(struct inode *, void *),
2321 void *data);
2322 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2323 int (*test)(struct inode *, void *), void *data);
2324 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2325
2326 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2327 extern struct inode * iget_locked(struct super_block *, unsigned long);
2328 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2329 extern int insert_inode_locked(struct inode *);
2330 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2331 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2332 #else
2333 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2334 #endif
2335 extern void unlock_new_inode(struct inode *);
2336 extern unsigned int get_next_ino(void);
2337
2338 extern void __iget(struct inode * inode);
2339 extern void iget_failed(struct inode *);
2340 extern void end_writeback(struct inode *);
2341 extern void __destroy_inode(struct inode *);
2342 extern struct inode *new_inode_pseudo(struct super_block *sb);
2343 extern struct inode *new_inode(struct super_block *sb);
2344 extern void free_inode_nonrcu(struct inode *inode);
2345 extern int should_remove_suid(struct dentry *);
2346 extern int file_remove_suid(struct file *);
2347
2348 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2349 static inline void insert_inode_hash(struct inode *inode)
2350 {
2351 __insert_inode_hash(inode, inode->i_ino);
2352 }
2353
2354 extern void __remove_inode_hash(struct inode *);
2355 static inline void remove_inode_hash(struct inode *inode)
2356 {
2357 if (!inode_unhashed(inode))
2358 __remove_inode_hash(inode);
2359 }
2360
2361 extern void inode_sb_list_add(struct inode *inode);
2362
2363 #ifdef CONFIG_BLOCK
2364 extern void submit_bio(int, struct bio *);
2365 extern int bdev_read_only(struct block_device *);
2366 #endif
2367 extern int set_blocksize(struct block_device *, int);
2368 extern int sb_set_blocksize(struct super_block *, int);
2369 extern int sb_min_blocksize(struct super_block *, int);
2370
2371 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2372 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2373 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
2374 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2375 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2376 extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2377 loff_t *);
2378 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2379 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2380 unsigned long *, loff_t, loff_t *, size_t, size_t);
2381 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2382 unsigned long, loff_t, loff_t *, size_t, ssize_t);
2383 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2384 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2385 extern int generic_segment_checks(const struct iovec *iov,
2386 unsigned long *nr_segs, size_t *count, int access_flags);
2387
2388 /* fs/block_dev.c */
2389 extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2390 unsigned long nr_segs, loff_t pos);
2391 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2392 int datasync);
2393
2394 /* fs/splice.c */
2395 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2396 struct pipe_inode_info *, size_t, unsigned int);
2397 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2398 struct pipe_inode_info *, size_t, unsigned int);
2399 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
2400 struct file *, loff_t *, size_t, unsigned int);
2401 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2402 struct file *out, loff_t *, size_t len, unsigned int flags);
2403 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2404 size_t len, unsigned int flags);
2405
2406 extern void
2407 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2408 extern loff_t noop_llseek(struct file *file, loff_t offset, int origin);
2409 extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
2410 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
2411 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2412 int origin, loff_t maxsize);
2413 extern int generic_file_open(struct inode * inode, struct file * filp);
2414 extern int nonseekable_open(struct inode * inode, struct file * filp);
2415
2416 #ifdef CONFIG_FS_XIP
2417 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2418 loff_t *ppos);
2419 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2420 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2421 size_t len, loff_t *ppos);
2422 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2423 #else
2424 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2425 {
2426 return 0;
2427 }
2428 #endif
2429
2430 #ifdef CONFIG_BLOCK
2431 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2432 loff_t file_offset);
2433
2434 enum {
2435 /* need locking between buffered and direct access */
2436 DIO_LOCKING = 0x01,
2437
2438 /* filesystem does not support filling holes */
2439 DIO_SKIP_HOLES = 0x02,
2440 };
2441
2442 void dio_end_io(struct bio *bio, int error);
2443 void inode_dio_wait(struct inode *inode);
2444 void inode_dio_done(struct inode *inode);
2445
2446 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2447 struct block_device *bdev, const struct iovec *iov, loff_t offset,
2448 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2449 dio_submit_t submit_io, int flags);
2450
2451 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2452 struct inode *inode, const struct iovec *iov, loff_t offset,
2453 unsigned long nr_segs, get_block_t get_block)
2454 {
2455 return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
2456 offset, nr_segs, get_block, NULL, NULL,
2457 DIO_LOCKING | DIO_SKIP_HOLES);
2458 }
2459 #else
2460 static inline void inode_dio_wait(struct inode *inode)
2461 {
2462 }
2463 #endif
2464
2465 extern const struct file_operations generic_ro_fops;
2466
2467 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2468
2469 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2470 extern int vfs_follow_link(struct nameidata *, const char *);
2471 extern int page_readlink(struct dentry *, char __user *, int);
2472 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2473 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2474 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2475 int nofs);
2476 extern int page_symlink(struct inode *inode, const char *symname, int len);
2477 extern const struct inode_operations page_symlink_inode_operations;
2478 extern int generic_readlink(struct dentry *, char __user *, int);
2479 extern void generic_fillattr(struct inode *, struct kstat *);
2480 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2481 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2482 void inode_add_bytes(struct inode *inode, loff_t bytes);
2483 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2484 loff_t inode_get_bytes(struct inode *inode);
2485 void inode_set_bytes(struct inode *inode, loff_t bytes);
2486
2487 extern int vfs_readdir(struct file *, filldir_t, void *);
2488
2489 extern int vfs_stat(const char __user *, struct kstat *);
2490 extern int vfs_lstat(const char __user *, struct kstat *);
2491 extern int vfs_fstat(unsigned int, struct kstat *);
2492 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2493
2494 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2495 unsigned long arg);
2496 extern int __generic_block_fiemap(struct inode *inode,
2497 struct fiemap_extent_info *fieinfo,
2498 loff_t start, loff_t len,
2499 get_block_t *get_block);
2500 extern int generic_block_fiemap(struct inode *inode,
2501 struct fiemap_extent_info *fieinfo, u64 start,
2502 u64 len, get_block_t *get_block);
2503
2504 extern void get_filesystem(struct file_system_type *fs);
2505 extern void put_filesystem(struct file_system_type *fs);
2506 extern struct file_system_type *get_fs_type(const char *name);
2507 extern struct super_block *get_super(struct block_device *);
2508 extern struct super_block *get_active_super(struct block_device *bdev);
2509 extern struct super_block *user_get_super(dev_t);
2510 extern void drop_super(struct super_block *sb);
2511 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2512 extern void iterate_supers_type(struct file_system_type *,
2513 void (*)(struct super_block *, void *), void *);
2514
2515 extern int dcache_dir_open(struct inode *, struct file *);
2516 extern int dcache_dir_close(struct inode *, struct file *);
2517 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2518 extern int dcache_readdir(struct file *, void *, filldir_t);
2519 extern int simple_setattr(struct dentry *, struct iattr *);
2520 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2521 extern int simple_statfs(struct dentry *, struct kstatfs *);
2522 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2523 extern int simple_unlink(struct inode *, struct dentry *);
2524 extern int simple_rmdir(struct inode *, struct dentry *);
2525 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2526 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2527 extern int simple_empty(struct dentry *);
2528 extern int simple_readpage(struct file *file, struct page *page);
2529 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2530 loff_t pos, unsigned len, unsigned flags,
2531 struct page **pagep, void **fsdata);
2532 extern int simple_write_end(struct file *file, struct address_space *mapping,
2533 loff_t pos, unsigned len, unsigned copied,
2534 struct page *page, void *fsdata);
2535
2536 extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
2537 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2538 extern const struct file_operations simple_dir_operations;
2539 extern const struct inode_operations simple_dir_inode_operations;
2540 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2541 struct dentry *d_alloc_name(struct dentry *, const char *);
2542 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2543 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2544 extern void simple_release_fs(struct vfsmount **mount, int *count);
2545
2546 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2547 loff_t *ppos, const void *from, size_t available);
2548 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2549 const void __user *from, size_t count);
2550
2551 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2552
2553 extern int generic_check_addressable(unsigned, u64);
2554
2555 #ifdef CONFIG_MIGRATION
2556 extern int buffer_migrate_page(struct address_space *,
2557 struct page *, struct page *);
2558 #else
2559 #define buffer_migrate_page NULL
2560 #endif
2561
2562 extern int inode_change_ok(const struct inode *, struct iattr *);
2563 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2564 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2565
2566 extern void file_update_time(struct file *file);
2567
2568 extern int generic_show_options(struct seq_file *m, struct vfsmount *mnt);
2569 extern void save_mount_options(struct super_block *sb, char *options);
2570 extern void replace_mount_options(struct super_block *sb, char *options);
2571
2572 static inline ino_t parent_ino(struct dentry *dentry)
2573 {
2574 ino_t res;
2575
2576 /*
2577 * Don't strictly need d_lock here? If the parent ino could change
2578 * then surely we'd have a deeper race in the caller?
2579 */
2580 spin_lock(&dentry->d_lock);
2581 res = dentry->d_parent->d_inode->i_ino;
2582 spin_unlock(&dentry->d_lock);
2583 return res;
2584 }
2585
2586 /* Transaction based IO helpers */
2587
2588 /*
2589 * An argresp is stored in an allocated page and holds the
2590 * size of the argument or response, along with its content
2591 */
2592 struct simple_transaction_argresp {
2593 ssize_t size;
2594 char data[0];
2595 };
2596
2597 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2598
2599 char *simple_transaction_get(struct file *file, const char __user *buf,
2600 size_t size);
2601 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2602 size_t size, loff_t *pos);
2603 int simple_transaction_release(struct inode *inode, struct file *file);
2604
2605 void simple_transaction_set(struct file *file, size_t n);
2606
2607 /*
2608 * simple attribute files
2609 *
2610 * These attributes behave similar to those in sysfs:
2611 *
2612 * Writing to an attribute immediately sets a value, an open file can be
2613 * written to multiple times.
2614 *
2615 * Reading from an attribute creates a buffer from the value that might get
2616 * read with multiple read calls. When the attribute has been read
2617 * completely, no further read calls are possible until the file is opened
2618 * again.
2619 *
2620 * All attributes contain a text representation of a numeric value
2621 * that are accessed with the get() and set() functions.
2622 */
2623 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2624 static int __fops ## _open(struct inode *inode, struct file *file) \
2625 { \
2626 __simple_attr_check_format(__fmt, 0ull); \
2627 return simple_attr_open(inode, file, __get, __set, __fmt); \
2628 } \
2629 static const struct file_operations __fops = { \
2630 .owner = THIS_MODULE, \
2631 .open = __fops ## _open, \
2632 .release = simple_attr_release, \
2633 .read = simple_attr_read, \
2634 .write = simple_attr_write, \
2635 .llseek = generic_file_llseek, \
2636 };
2637
2638 static inline __printf(1, 2)
2639 void __simple_attr_check_format(const char *fmt, ...)
2640 {
2641 /* don't do anything, just let the compiler check the arguments; */
2642 }
2643
2644 int simple_attr_open(struct inode *inode, struct file *file,
2645 int (*get)(void *, u64 *), int (*set)(void *, u64),
2646 const char *fmt);
2647 int simple_attr_release(struct inode *inode, struct file *file);
2648 ssize_t simple_attr_read(struct file *file, char __user *buf,
2649 size_t len, loff_t *ppos);
2650 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2651 size_t len, loff_t *ppos);
2652
2653 struct ctl_table;
2654 int proc_nr_files(struct ctl_table *table, int write,
2655 void __user *buffer, size_t *lenp, loff_t *ppos);
2656 int proc_nr_dentry(struct ctl_table *table, int write,
2657 void __user *buffer, size_t *lenp, loff_t *ppos);
2658 int proc_nr_inodes(struct ctl_table *table, int write,
2659 void __user *buffer, size_t *lenp, loff_t *ppos);
2660 int __init get_filesystem_list(char *buf);
2661
2662 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2663 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2664
2665 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2666 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2667 (flag & __FMODE_NONOTIFY)))
2668
2669 static inline int is_sxid(mode_t mode)
2670 {
2671 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2672 }
2673
2674 static inline void inode_has_no_xattr(struct inode *inode)
2675 {
2676 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2677 inode->i_flags |= S_NOSEC;
2678 }
2679
2680 #endif /* __KERNEL__ */
2681 #endif /* _LINUX_FS_H */